University of Southern California Dissertations and Theses

An investigation into the regional segmentation of the commercial real estate market in the United States

(USC Thesis Other)

An investigation into the regional segmentation of the commercial real estate market in the United States

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content A N IN V ESTIG A TIO N IN TO T H E R EG IO N A L SE G M E N T A T IO N O F T H E C O M M E R C IA L R E A L ESTA TE M A R K E T IN T H E U N ITED STA TES by Ross Stanley Selvidge A Dissertation Presented to the FA C U L T Y O F T H E G R A D U A T E S C H O O L U N IV ERSITY O F SO U T H E R N CA LIFO RN IA In Partial Fulfillment of the Requirements for the Degree D O C T O R O F PH ILO SO PH Y (Business Administration) December 1985 UMI Number: DP22628 All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. Dissertation Publishing UMI DP22628 Published by ProQuest LLC (2014). Copyright in the Dissertation held by the Author. Microform Edition © ProQuest LLC. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106-1346 U N IV E R S IT Y O F S O U T H E R N C A L IF O R N IA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES, CALIFORNIA 90089 Com S V -6 ? 3! f y This dissertation, written by R O SS ST A N L E Y SELV ID G E under the direction of hx.$........ Dissertation Committee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillm ent of re quirements for the degree of D O C TO R OF PH ILOSOPH Y Dean of G raduate Studies Y)d fg Dec emb e r # 5# , # # 19 8 5 DISSERTATION COMMITTEE C hairperson To C SB 1 1 j A C K N O W L E D G E M E N T S j 1 I It is with sincere gratitude that I acknowledge the . | invaluable assistance and support of the members of my . - I ; committee, Rocky Tarantello, Victor A . Canto, and William j , i I C . Baer, in the completion of this undertaking. I am j also indebted to a number of additional faculty members | and sta ff of the Graduate School of Business. In I , \ particular I would like to express m y deep appreciation to ; Richard V . Eastin, David Dale-Johnson, Merlin C . Findlay, ! , Terence C . Langetieg, Paul A . Gruendemann, and Dean A . . Baroni. In addition, I would like to thank the members of my family and most particularly my sister, Judith, who i provided both encouragement and support when it was needed | i i i most. ! Finally, this study would not have been possible I without the cooperation of the Building Owners and | I I I ; Managers Association. The data to which they so ; j graciously permitted access were indispenable. j TABLE OF CONTENTS Page i D ED IC A TIO N 1 1 A C K N O W L E D G E M E N T S iii LIST O F T A B L E S vi LIST O F FIG U RES vii i i i i ! C H A PT E R I Real Estate Asset Attributes and Markets 1 1.1 Introduction 1 1.2 Distinguishing Real Estate Characteristics 6 1.3 Summary 20 II Traditional Financial Pricing Models 21 II. 1 Major Pricing Models 21 II. 2 Model Assumptions 22 II. 3 Additional C A P M Assumptions 27 II.4 C A P M Theory 30 II. 5 Arbitrage Pricing Theory 33 II.6 Results of Empirical Studies 35 II. 7 Summary 37 III Impediments to Real Estate Asset Analysis 38 III. 1 Introduction 38 III. 2 Divisibility 38 III. 3 Information 44 III.4 Transaction Costs 54 III. 5 Trading in All Assets 58 III. 6 Return and Risk Measurement 59 III. 7 Summary 61 IV Real Estate and Factor Price Equalization 65 IV. 1 Introduction 65 IV. 2 Trade Theory 66 IV. 3 Trade Theory in Regional Analysis 70 x v C H A PT E R Page IV.4 Real Estate Application 74 IV.5 Summary 79 V The Model 81 V.l Introduction 81 V.2 Basic Relationships 81 V.3 Net Returns 83 V.4 Constrained Mobility 86 V.5 Incorporation of Transportation Costs 90 V.6 Equilibrium and Local Fiscal Policies 92 V.7 Relationships for Empirical Analysis 95 V.8 Expected Values 98 VI Data and Analytic Procedures 107 VI.1 Data Sources 107 VI.2 Data Formats 108 VI.3 City and Regional Designations 110 VI.4 Pooling of Data Across Cities 111 VI.5 Autocorrelation of Data 113 VI.6 Two-Step Full Transform Analysis 115 VI.7 Joint Probability of Outcomes 116 VI.8 Summary 118 VII Results € 123 VII.1 Preliminary Data Analysis 123 VII.2 Gross Rental Equilibrium 125 VII.3 Net Rental Equilibrium 143 VIII Summary and Conclusions 175 VIII.1 Summary 175 VIII.2 Conclusions 181 R EFER EN C ES 185 106 120 160 161 164 166 167 168 171 173 174 L IST OF TABLES Expected Values of Variables' Coefficients Regions and Cities with. Number Codes Data Pooling F-Test Results Equation (V-6) Results Equation (V-6'); Cities Results Equation (V-61); Regions Results Gross Rental Equilibrium Results Sum mary Equation (V-7) Results Equation (V-71); Cities Results Equation (V-71); Regions Results Net Rental Equilibrium Results Summary L IST OF FIGURES FIG U RE Page VI-1 B O M A Data Report Format 121 VI-2 M ap of B O M A Regions 122 C H A PT E R I R E A L ESTA TE A SSET A TTRIBU TES A N D M A R K E T S i i 1.1 IN TR O D U C TIO N ; i i From time to time in popular real estate and i business publications i t is not uncommon to find discussions of the rela tiv e strength, quality, or , | desirability of different regional real estate markets. I 1 ; Such discussions often take the form of developing a ! ! j 1 ranking or categorization of the different regions for I ; the purpose of establishing investment preferences. These ; rankings which imply the existance of extraordinary , , investment opportunities also im p licitly imply a i 1 significant interregional inefficiency in the investment real estate market. Two possible explanations for such i opportunities, if they do in fact exist, could be: a) ; semi-strong market inefficiency, or b) market segmentation I i resulting from severe factor immobility and trading i barriers. i A n example of the former would be the existance of a regional market offering extraordinary investment returns ! i which is distinguished from other markets by its relative j \ i I strength or future promise based on assorted economic ! ; factors. Indeed, who has not heard of preferences on the | i 1 part of many investors for "sun belt" regions based on j their comparative advantages as opposed to markets in the "frost belt." Similarly, it is also quite common to hear expressions of a preference for regions whose economic ; well being seems to be positively correlated with "high tech" industrial development rather than those tied to the prosperity of "smoke stack" industries. S It seems, therefore, that in the area of real estate investment there is a view with a substantial following t which claims an investor can experience superior performance by investing in regions which have revealed themselves, according to widely available economic j information, to be those areas with prospects for the j greatest relative growth. The exact nature of this i "superior performance" and the rationale behind the expectations of "greater relative growth," however, are usually rather vague or ill-defined. i ! Such a view in itia lly may appear to be eminently i :reasonable. What could be more logical than to expect J , . , investors who acquire assets m regions that experience |relatively high economic growth and properity to earn, in |general, relatively high returns? The question that ought ! ;to arise, though, is whether or not higher returns can be expected as a result of investing in a preferred or favored area if that investment is undertaken after a i !region has been widely publicized as being desirable or ■ 2 ; a fte r i t has exhibited the ch aracteristics which a consensus believes will lead to relatively high growth. 1 In terms of tra d itio n a l financial theoretic 1 literature, then, the question which should be asked is ’ whether or not, with respect to real estate investment, widely publicized, readily available or easily compiled i information about different regions' economic prospects jcan be incorporated into one's investment selection I process with the expectation of earning superior returns, i If i t can be demonstrated that an investor could !consistently earn superior returns based on such an i investment scheme, one would have evidence refuting the i applicability of the familiar concept of semi-strong i market efficiency in the case of interregional real estate investment. | An example of market segmentation resulting from ;factor immobility and restrained trade would be one in 1 which both factors of production and the goods produced are priced exclusively on endogenous criteria. The body of 'financial and economic literature in the area of trade i !theory has explored the extent to which factors employed ^and goods produced in different markets will tend (or not) i to be priced uniformly. For the most part, this issue I jreduces down to the question of the degree of mobility of j factors of production and the ability of goods to be ; traded across markets. Barriers to such mobility and free j 1 i ! trade can be either natural or man-made. Testing for the ' presence or absence of factor price or return equalization i then is to a great extent a test of the scope of the ; | ; effectiveness of such barriers. This dissertation examines whether theory and data ; I i exist to support either or these possible explanations of ; ;market inefficiency. In preparation for properly ? I addressing these possible explanations of market I inefficiency in real estate it will be necessary to * ! examine the degree of comparability between real estate !and the other classes of assets upon which so much of i .existing financial theory (including market efficiency) ! . . . . I has been derived. Therefore, we w ill begin with a ' thorough discussion of the extent to which real estate and < real estate investment are characterized by attributes | I ! ; common to, and recognized by, generally accepted financial ; » ,theories. After a long period of comparative neglect i t is I i relatively recently that real estate has begun to come I . . . ;under rigorous scrutiny m the financial literature. This is probably due to a number of factors. Certainly among \ those could be the seemingly extraordinary returns which | ' I jmany real estate investors in the mid and latter 1970's t iare perceived to have earned. In addition, a continuing 4 I process of deregulation is widely expected to open the ] I door to increased real estate investing on the part of | i institutions which in the past have been either excluded 1 from or severely limited in their holding of real estate assets. ; Ironically, at the same time, an increase in i I regulation in the form of recently enacted statues may I i ! j also have the effect of widening the holding of real : estate and therefore increasing academic interest in it as i an investment vehicle. These statues are of course those j which seek to provide greater incentives for the prudent ! investment of pension fund assets by establishing personal ; j liability on the part of fund managers. Increasingly they j I j must now undertake demonstratively sound investment « < practices. Also there may be on the part of real estate j I investment advisors a need or desire for widely accepted j ; theoretic justification for the recommendations they make i 1 : to a clientele which, along with society in general, may 1 ! I have become more litigious, j Finally, i t is possible that we are presently j observing a general broadening of investment perspectives I j which is a consequence of improvements in communication i | and in the collection and dessemination of information , f j about, among other possibilities, real estate investment i i opportunities. Once again, along with such a broadening 1 of interest there also would come increased concerned for 1 I a theoretical rationale upon which to base investment decisions. i A s has been the case with other assets, the ultimate ! i ! objective of theoretical research into real estate is to model its behavior as an investment (i.e., to develop ! valuation models and establish investment policy ^ guidelines). Alas, despite the interest which is now being directed its way, the development of a sound, widely accepted real estate theory has not progressed very far. ' This lach of progress could be due to the relatively brief ' period of time during which i t has been subjected to I ! serious examination. Another possible explanation, i 1 ; however, could be the extent to which real estate is f different in nature from other types of assets that have I | been the subject of earlier and longer inquiry and around j which most of existing financial theory has been built. This second explanation opens a very broad area of | inquiry. It is with this issue of the comparability of ( [ real estate with other assets that this investigation shall begin. i ! 1.2 DISTINGUISHING R E A L ESTA TE CH A RA CTERISTICS Several features of real estate which may give rise to analytic difficulties are quite commonly used to | i distinguish i t from other asset types. These features ■ include the presence of consumption as well as investment motivation in some transactions, the heterogeneous nature of the goods involved, lack of liquidity, and degree of market segmentation. Each of these will be discussed in ; , i turn with a focus on the extent to which they may or may j not present analytic obstacles. I i I J 1.2.1 Consumption and Investment | ! W hen an attempt is being made to analyze investment ! ! behavior i t can be confounding to have consumption in 1 addition to investment as part of the motivation for ' I i , transactions. The most common real estate transaction of j j all, that involving owner occupied housing exhibits these ■ 1 two elements. A consumption aspect is unlikely to be ! I significant in transactions involving real estate which j does not fall into the owner occupied housing category, j : i t With respect to owner occupied housing, however, a : t consumption aspect in a tra n sa c tio n should not ; ! automatically preclude analysis of that exchange from its : i : investment standpoint. If such a consumption aspect does I exist, it is a matter of whether or not, through analysis, I the effects of both factors can be separated. > i j For owner occupied housing this question has been addressed before. There is a significant body of 7 ' literature which covers this question which has centered | around the issue of what factors influence tenure choice ! (the decision to rent or owner occupy one's housing), j While no consensus has evolved, a number of analysts feel i , s they have made substantial progress in identifying and ' separating some of the investm ent aspects and i I I considerations of what ought to be, in an uncomplicated | ! | form, a simple matter of housing consumption (Henderson j and Ioannides, 1983). ' Factors such as wealth or lif e cycle income, ! | , transaction costs (involved in moving from one home to ; t i I another), taxation and finally risk avoidance with respect : to rent levels, inflation and housing price levels are j those most commonly identified (Bossons, 1978; Ioannides, 1 ! I 1979; Rosen, 1979; Shelton, 1968). Yet, despite these i having been ascertained, home ownership has only in the | 1 , most general way been incorporated into financial analysis f | of individual consumption and investment functions. Without a more definitive understanding and specification |of owner occupied housing's investment aspects, i t will < ' not be possible to combine the tenure choice decision j ■ properly with other specific investment possibilities. In particular, when one looks at the question of real i estate investment from the standpoint of choosing among a number of alternative regional markets, the investment ; component of the tenure decision would have an I I : additional complicating aspect. The region within which , one lives is usually not determined by the consideration 1 of where one's owner occupied housing investment is maximized. Thus, an individual is unlikely to be : confronted with a decision involving choice between j 1 investment in owner occupied housing in one region versus ! I another. Rather, what is most often the case is a j I decision between renting versus buying housing in a ! i . ! j predetermined region. For this decision to be properly j * weighted, however, i t would be necessary to consider as ; I ; i i ; well the possibilities of investment m real estate or , ( other assets both in that predetermined region and | ; l | elsewhere. Without a completely compatible specification j I , j of owner occupied housing's investment component, combining the consideration of tenure choice and the holding of other assets becomes infeasible. Such a : completely compatable specification does not now exist. I , At the same time, this complication of the investment I . aspect of owner occupied housing is only of significance 1 1 : to the extent that individuals who could be owner * t t i | occupants are participants in the market at large. If the j market is segmented with respect to classes of investors J for various types of properties there may not be a problem i across all types of real estate. That is, if institutional I I ! investors, which can not owner occupy housing, were to j i dominate a segment of the market the issue of there being < both investment and consumption features of homeownership 1 would be moot for that segment. i i i ; I : I i 1.2.2 Heterogeneity of Real Estate Assets I i . j ; The heterogeneity of real estate from a physical j ■ standpoint is also very commonly cited as an impediment to ; systematic analysis. To be sure, every piece of real , I estate is unique in a physical sense. Each parcel of land is a combination of characteristics such as topography, t | zoning and access or proximity to other locations which is ! ' i different from all others. Likewise, the improvements on j any piece of land (if not in itially , at least later in ' their life) differ in some degree from all others with respect to age, condition, size or functional design. But \ | does this make real estate noncomparable with other ; I i investment possibilities for purposes of analysis or , infeasible for such analysis from a practical point of j view? j The position that real estate does present such a problem may be argued as follows. W hen a real estate j i J transaction takes place i t is one in which the asset , changing hands is precisely like no other asset which has I just recently traded in the market place (except of course 10 I ---------------------------------------------- ^ ^ - -J ■ in that small minority of situations m which the very same asset was traded such as in the case of a double . escrow during which there is no change whatsoever in the • k property's status). Being a unique transaction then the I . ! question arises how can a predictable pricing mechanism be 1 i ' constructed. ■ I Perhaps a modeling based on breaking down properties ! . into component characteristics, each of which could be i I I ' priced, would be a reasonable approach. In following such | I I I a procedure, one would take properties, each of which is a , | i unique aglomeration of many characteristics, and express J ; I ; their overall values as a function of the separate values ! , of their constituent components. Hedonic price modeling I is the term given to this process. There have been many ' j | effo rts at analysis of th is sort m real estate ; p articu larly in the case of resid en tial properties ! (Lancaster, 1966; Rosen, 1974). A very great percentage i ! of the short terra price variation observed for seemingly ! heterogeneous residential properties has been explained ■ ' i with some of these models. However, they are not without j | difficulties. Am ong those difficulties are the selection of appropriate property attributes, linearity of the value ! relationships, multicolinearity among the attributes j i ! selected, and possible segmentation within the real estate market. 11 ; Perhaps a better approach to take is one which 1 I ' focuses not on the physical aspects of the property but ■ rather on that aspect for which it is being considered as ; ^ an investment by its potential purchaser—its return. To , concentrate on the prospects for returns is what is done, ; for instance, in the case of bonds. Each bond issue is ; unique from the standpoint of all the factors by which it i I ! ! might be described (the issuer, amount, interest rate, j call provisions, convertability, etc.). And, for many i 1 B j issues there is a very thin market to say the least. ' | This, however, does not prompt criticisms on the basis of ! I I : that heterogeneity because of the widespread use of bond ratings—a criteria that incorporates consideration of ! risk and return classes. \ : i i ! | Should this not be a legitimate approach for real i { i 1 estate? In other words, in spite of the physical | : I ; uniqueness of each real estate asset, why can not common attributes or characteristics of the returns from those j ‘assets be the basis for some analytic or systematic ^ I pricing scheme? From this point of view the issue of ! ' heterogeneity becomes one of both information and its | i proper use. That is to say, the question is one of what j I I | knowledge is available about what assets and how can it j most appropriately and usefully be presented to potential ! i ' I J investors. ' i i 1 i I 12 ! I 1.2.3 Real Estate Liquidity' Perhaps the distinguishing characteristic of real estate which comes most often and quickly to the minds of , investors is that of illiquidity. Liquidity, though, is a ' somewhat illusive concept. In a rudimentary way i t j involves the ability to exchange an asset for cash. That J ! I j "ability" involves two different facets according some theoretical views of the concept. j ) First, there is the idea of marketability. What is j meant by this is the ability to transact at a reasonable ! volume without appreciable delay and at an price close to ! the underlying equilibrium price. In other words, the , ! 1 J ability to buy or sell a significant volume of assets I within a short period of time without a significant price ! concession. Put a slightly different way, marketability ; can also be thought of as the proportion of the market ; < i value which one can expect to receive at various time j intervals after an asset has been placed on the market. i Inherent in these views are some assumptions about the I i separability of the effect on transaction prices of ; i , temporary imbalances in buy and sell orders and about j 1 i shifts in the underlying price equilibrium resulting from j i changes in basic supply and demand conditions (Tinic and , West, 1979). It is the former with which marketability is j I concerned. Marketability then is largely a function of the volume of transactions and the organization of the j ! market in which an asset is traded. ! The other aspect which has come to be thought of in . ; the context of liquidity is capital certainty. Price , ! i ! stability or predictability over time is what is meant by j i I this term. At this point consideration is given to the i I I nature of the underlying price equilibrium caused by the basic supply and demand conditions mentioned earlier. A ! subjective estimation of the dispersion of possible market ■ values in the future is the essence of this concept * (Moore, 1968). j \ W hen broken down in this fashion the nebulous quality 1 of liquidity is much less daunting from an analytic j , standpoint. This is not to imply that a single ! i i operationally measurable standard for liquidity exists. ; One does not. However, taken se p ara te ly , both ! 1 marketability and capital certainty can be dealt with in I real estate just as with any other type of asset. What is probably meant most often by investors when J they speak of the illiquidity of real estate is in fact f I ; its lack of marketability. In other words, concern for | ! the issue of capital certainty is probably far outweighed J , by a concern which is focused on the extent to which a price consession may be involved in a transaction as a function of length of market exposure. Real assets such I as real estate are generally considered to be much less marketable than financial assets. This is in a large part attributable to their heterogeneity and complementarity as well as relatively high transaction costs. 1.2.4 Market Segmentation F in ally , the m atter of rea l e s ta te market segmentation could be a complicating if not distinguishing factor in the area of investment analysis. One can think of segmentation on two different levels. First, real estate could be thought of as traded in a market separate from that in which the exchanges of other assets take place. Such a segmentation could be the re su lt of statutory limitations on the investment options of various classes of potential investors, agency considerations, market constraining limitations resulting from economies of scale, or other factors. If segmentation of that nature does exist it would be futile to apply pricing models developed for other types of assets to the real estate market without a great deal of studied alteration. A second level of segmentation is that which may exist within the market for all real estate assets. Consider the ways in which an investor might characterize various real estate opportunities if he were asked to do so. If he were looking at a wide variety of properties, surely among the characterizations would be found the following: excellent (or poor) tax shelter benefits, minor (or excessive) management burdens, and large (or small) capital requirements. These aspects of real estate investments provide the basis for a brief inquiry into the nature of possible segmentation. The type of property under consideration can have a profound effect on the expectations for an investment. One such effect is the resulting tax treatment which an investor would enjoy. The holding of investment real estate has long been considered as an excellent tax reduction strategy. Even as tax laws have undergone revisions to eliminate "tax loopholes," real estate usually has emerged in a more favorable position relative to other types of assets. The extent to which a property is afforded tax advantages is a function of both the use to which i t is put and, if i t is renovated upon acquisition, also its age. The best known tax reduction features of real estate are its generous depreciation allowances. These allowances are based on the use of each property, with residential usage receiving the most liberal treatment. A more recent tax provision, not formerly available to real estate investors, is the tax credit which may be obtained upon the renovation of older existing properties. These 1 features along with a lack of "at risk" provisions put , real estate into an extraordinarily attractive position : and, among different classes of real estate, produce a wide range of effects. With tax benefits varying greatly among d ifferen t real estate c la ssific a tio n s the segmentation question arises in the form of whether or not : the supply of investments in the different classes and the ; number of investors in different tax circumstances are t such that those investment classes are dominated by distinctly different investor clienteles. Another difference between real estate investment 1 ; classes is the extent to which the investor may need to be i involved in the management of the asset and the level of effort which that entails. This would be a function of I the configuration and use of the property as well, ' perhaps, as the geographic location. For instance, a single occupant warehouse under a triple-net lease to a i “blue chip" tenant would involve vastly lower levels of | effort and responsibilities on the part of the owner than ; the holding of a ten-unit residential property. In I addition, ownership at a distance can increase the ! ! necessary levels of effo rt. To some extent that ! U i ' consideration of geographic proximity may have the effect of limiting the demand for some types of real estate properties to the geographic area where they are located. 17 In these cases the segmentation question would be one of whether or not classes which differ widely with respect to owner involvement also appeal only to divergent investor groups. The th ird candidate for being a b asis for segmentation is that of the varying capital requirements which pertain to real estate investments. The range of real estate values across the market is vast. Even if one were to assume that any investment undertaken would be "mortgaged to the hilt," the top end of the range of investment capital required would s till be high in an absolute sense by almost any standard. So, the question arises, do we again have the possibility that investors of one description align themselves with one portion of the market while others participate in a different portion. If segmentation based on capital requirements exists, one would of course expect i t to be such that investors with large resources participated in the high end of the market and those with fewer resources predominated in the lower end. These three types of possible real estate market segmentation are not ignored in financial theory. For instance, the tax clientele effect theory has similarities to what was described as a possible segmentation of the real estate market along the lines of tax benefit 18 d iffe re n tia ls (Elton and Gruber, 1970; M iller and Modigliani, 1961; Pettit, 1977). The notion of trading ; upon human capital may help explain any apparent segmentation which seems to be dependent upon the i magnitude of the involvement of the asset holder in the management of the asset. In addition, considerations of | agency relationships are implicit in the ideas of market ^ segmentation based on both the degree of management burden and capital requirements. And finally, there are aspects of divisibility, intermediation, and agency questions involved in the capital requirements issue as well. | So far in this presentation no position has been | taken as to the existance or not of segmentation in the i real estate market. What has been put forth is a very ; b r ie f discu ssio n of some of the commonly used ! > characterizations of different real estate investments, the ram ifications that they might hold for market : segmentation and the sim ilarity any such segmentation ; might have with widely recognized financial theoretic i | concepts. The significance of the question about I j segmentation is that if the real estate market does in i I fact operate in distinct segments, then account must be t I ! taken of this when an empirical analysis of investment theories is performed. If this segmentation does exist and researchers do not pay heed, the outcome of any 19 studies which they might conduct would be quite suspect. , | Validations of any models under those circumstances would ; be fallacious, and the failure of any tests would as : easily be attributable to the presence of segmentation as ; i to a defect in the model's specification. j i i j I .3 S U M M A R Y I I i I This discussion of the purported consumption aspect, ; heterogeneity, illiquidity and segmentation of the real estate market has been b rie f and general. It was ; provided to introduce the difficulties and complications ; I that might be encountered in the theoretic analysis of J I J j real estate investment. What follows is a closer ! ! examination of the investment attributes of real estate i and the assumptions underlying major asset pricing models. ! I | That review will shed additional light on the prospects, j I 1 ‘ using one or another of those models, for successfully : i ; i investigating either of those two possible explanations : for interregional inefficiency in the real estate market. I j I I I ! i j I L 20 C H A PTE R II TRA D ITIO N A L FINANCIAL PRICING M O D E L S II. 1 M A JO R PRICING M O D E L S Presently there are two major asset pricing models which dominate the financial literature. Those are the Capital Asset Pricing Model (CA PM ) and the Arbitrage Pricing Theory (APT). The C A PM was originally developed nearly twenty years ago while the A PT was first formulated in 1976 (Markowitz, 1959; Ross, 1976; Sharpe, 1970). They dominate in that, to a greater extent than any other a sse t p ricin g models, they have been subjected (excruciatingly so in the case of the CA PM ) to extensive theoretical and empirical investigation. Despite the fact that neither has been conclusively demonstrated through those studies to offer a completely comprehensive and correct description of asset market behavior, they do have distinct theoretical appeal and significant empirical robustness. If one were to look for particular theories into which the incorporation of real estate might be attempted, there are no more plausable candidates than the C A PM and the APT. In order to examine real estate investment's compatability with those models, this discussion will firs t state their underlying assumptions. The models themselves will then briefly be described, followed by an enumeration of the aspects of both real estate and the models which might hamper the models' use. I I I .2 M O D E L A SSU M PTIO N S There are a number of basic assumptions which are I | common to both models. Included in those are the I ' following: 1. Investors are price takers in a frictionless market j 2. Investors prefer more to less and are risk averse 3. Investors have homogenous expectations with I | respect to asset returns ( i 4. All assets are marketable and tradeable Each of these assumptions is a necessary condition for the | CA PM or APT to hold in its purest form. A b rief i i ■ explanation is in order. I I .2.1 Frictionless Markets The concept of investors being price takers in a i frictionless market has its e lf five basic underlying : components. These are: 1. The presence of many buyers and sellers 2. Infinite divisibility of assets 22 3. Costless information equally available to all 1 participants 4. N o transaction costs 5. N o taxes f The presence of many buyers and sellers in the f j ; market place, also sometimes called atomistic competition, 1 is in effect the contention that no buyer or seller j I can, on his own, by his presence or absence in the | I market, have a noticeable influence on the basic I demand-supply equilibrium. In other words, no ; i ' individual seller of goods is offering so great an : !amount that market prices must fall for all to be i I , absorbed nor does any buyer require so much of a good that ' i prices rise to meet his demand. Thus, only through ; i aggregate actions of many buyers or sellers will price j ' changes be observed. ' I Infinite divisibility of assets implies that holders ' i : :of those assets are unconstrained with respect to how much i or l it t le of a good an investor may transact in the market place. Such a lack of re s tric tio n on the size of ; ' I transactions means that an investor has complete , i |flex ib ility from the standpoint of portfolio revision. ! That is, as the need may arise to increase or reduce the amount of an asset which he holds, the investor is not limited in the range of possible adjustments by any I i 23 : minimum size requirement for a transaction. Whatever the required revision amount, however small, the transaction would be permitted by the market. In addition, there i would be no price concession or premium called for in i transactions involving one amount versus another. ! Specifying that information is both costless and simultaneously available to all participants in the market ‘ is a means of eliminating what could otherwise be a i j complicating factor. If information is available in that fashion, then all participants trade on the basis of the i same knowledge. If this were not the case i t might be ' justifiable to say that observed behavior in asset markets I i was the re su lt not only of the a ttrib u te s of any j particular asset but also of the constraints which may i i attend the dissemination and utilization of information. That is, asset behavior could in that case also be a i | function of the different information sets which were held i by the various market participants. Assuming only one information set upon which all investors base their j investment decisions equalizes all parties from that , standpoint; none would enjoy a comparative advantage over ! ! any others in that regard. This assumption makes it I j possible to focus attention on the assets themselves rather than their holders. I I 1 Finally, transaction costs and taxes are in itia lly : assumed away for reasons similar to fhe rationale for : standardized information. Both aspects are basically impediments to allocational efficiency. They are impediments in that the presence of either may cause an ! . . . , ! I asset holder who has made an investment decision (to buy , i i ■ or to sell) in an otherwise perfect market to deviate from ; that course on the basis of market mechanics (transaction ’ J costs) or government policy (taxation). Different levels : I i of transaction costs can cause different investment i , decision outcomes irrespective of constant underlying i i . | I factors for the asset involved. Likewise, variations m ; i : ' tax rules or their differential effect on different j investors can also produce disparate investment decisions | * despite constancy among an asset's basic attributes of ! I ■ value. In summary then, in a frictionless market in which i I all are price takers, no participant individually affects I the market equilibrium and all asset holders are able to j I costlessly move among all assets at any volume desired ! without informational disadvantage or other impediments i ; compared to other investors. j J ‘ II.2.2 Rationality and Risk Aversion j That investors would be assumed to prefer more of any good to less (ceteris paribus) requires no further comment j | but to say that this is an operational definition of j ; rationality. Risk aversion on the other hand bears further discussion. The essence of risk aversion can be captured by considering the possible investor attitudes 1 toward a choice between a certain outcome of a particular value and an uncertain outcome (a gamble) which has an actuarial value of the same amount. Given equal expected I values then, one with certain ty and the other with i uncertainty (i.e., the chance to receive a greater or I lesser amount than the expected value), an investor is t I considered to be a risk seeker if he prefers the uncertain ; choice, risk neutral if he is indifferent between the two, i : and a risk averter if he prefers the certain choice. If one thinks of th is in terms of a u tility of wealth I function plotted with wealth on the horizontal axis and ; utility of wealth plotted on the vertical axis, a rational J risk averse investor's u tility function would have a positive f ir s t derivative and a negative second i derivative. ; I I .2.3 Homogeneous Expectations Homogeneous expectations being assumed is an extension on the assumption of all information being costlessly and simultaneously available to all market participants. Specifically, it holds j not only have the same information but I 26 that all investors that they interpret ; those identical information sets in the same way.1 In other words, no participants in the market are fooled and any conclusion drawn from market data is that which all . others would draw. If homogeneous expectations were not assumed one might approach the information and ; expectations questions differently. In such a case, incomplete or uneven dissemination of information might [ ju st as plausably be the cause of a d isp arity of ! expectations as could a lack of homogeneous expectations i concerning information sets which are identical. Furthermore, any theory which does not assume homogeneous 1 expectations would probably be so ungainly as to present j intractable problems from the empirical standpoint. | II.2.4 Trade in All Assets Finally, the ability to trade all assets in the ; market place is esential if investors are to be able to 1 tailor their portfolios to their changing needs. What is meant by this is that whatever thing of value an investor has or can hope to gain in the future can be traded to i other investors at market prices. Thus there would be no ; asset in any portfolio which would be immune from either augmentation or reduction through market transactions. I I .3 A D D ITIO N A L C A P M A SSU M PTIO N S In addition to the preceding, factors which apply to 27 Fboth the C A P M and A PT, there are three assumptions which 1 apply to the C A PM but are not requirements of the APT. 1 These are: 1. Return and risk are expressed in terms of mean I j and variance 2. Jointly normal return distributions or quadratic , utility functions j 3. The availability of an asset earning a riskless ' rate of return at which one can borrow or lend in unlimited amounts Expressing return and risk quantitatively requires ; some knowledge of the possible outcomes with which investors in a particular asset would be faced. Also, one must look at the nature of that distribution or character I i of the investor's u tility function if a two parameter model is used. Finally, weight should be given to computational convenience. Given these conditions, mean and variance (being among the easiest to calculate and ! most widely recognized attributes of a distribution) have genuine appeal. Their legitimate use as the parameters of ! an investors choice however has a further implication. J j That implication is that either the distribution of j j returns must be jointly normal or the investor's u tility ! function must be quadratic in form. ! ! ! 28 i If'the distribution involved' happens to be jointly ! normal then the use of mean and variance is not only easy, ■ ! ; it completely specifies all of the pertinent properties. For distributions which are not jointly normal, mean and , variance offer an incomplete specification. However, the j ■ weight of evidence in the financial literature at this I i time indicates that for the asset types tested, the ; I * I jo in tly normal d istrib u tio n assumption, while not ; i I : precisely correct, is at least a reasonable approximation 1 I 1 ( Faina, 1965). i Alternatively, if the distribution is known not to be 1 jointly normal or if the assumption of normality is not | I i i I j justified, i t can s t i ll be legitimate to use mean and | I i variance as parameters of choice. This is the case if the , I j 1 investor's utility function can be assumed to be quadratic ! in form. In that event i t would not matter i f the | ! distribution of returns required more than mean and i I , variance to provide a complete description. Only the mean i and variance parameters, and no others, are significant | ; to a quadratic utility function. j Finally, the assumption concerning riskless borrowing ! and lending is an outgrowth of the concept of frictionless markets (no transaction costs) and the existance of an j asset which has constant returns in all states of nature. j ' I In that i t adheres to all of the APT's basic assumptions, 1 Htliese additional limitations in effect can be considered : i . to make the C A PM what amounts to a constrained, special , i case of the APT. ; XI.4 C A P M T H E O R Y I I i | Since the C A P M theory which has been under the longest ! , and most extensive investigation, it will be described | i I I first. Briefly it is the contention of the C A P M that the j equilibrium return on an asset will be equal to the risk ; I free rate of return plus a premium which is a linear ; function of the assets covariance with the returns to a I ' ; portfolio of a ll risky assets. This is based on i ; th eo retical extensions of the previously mentioned ; I ' i assumptions. I j The reasoning, roughly, is as follows. Given a j i propensity on the part of investors for investment ■ preference based on rationality and risk aversion they . will select investments at any particular level of return I j for which risk (variance) is lowest. In addition, since | I ; the returns to risky assets are not a ll perfectly ; i ( ! positively correlated, i t is possible by combining many ' risky assets to form portfolios which, when considered > , from the risk-return standpoint, would be preferable to the holding of an individual asset. Of all the possible portfolio combinations, those which would be preferred J | would be those which at any given level of return have the least variance. Portfolios of that description are termed to be Markowitz efficient (Markowitz, 1959). ; Now, if one combines the idea of holding Markowitz efficient portfolios with the assumptions that borrowing , and lending at the risk free rate is possible and that [ expectations are homogeneous, what can be derived is a t linear plot of investment opportunities (in mean-variance space) which would meet the rational risk averse investment criteria for any potential investor regardless I of his particular risk-return trade off preference. ! In other words, the plot of combinations of some j amount of a unique Markowitz efficient portfolio with different amounts of borrowing and lending at the risk I j free rate will satisfy any investor's needs. This plot is 1 known as the Capital Market Line (CML). Further, given I i that all investors have homogeneous expectations, that t each is holding varying proportions of his portfolio in a . combination of risky assets which is within its e lf I | proportioned identically for all investors among those | risky assets, and that all assets must be held by someone, it follows that the portfolio of risky assets is in fact a | market portfolio in which all assets are represented in i the proportion which their value bears to the value of the market as a whole. Having developed a theoretic cbnstruct for explaining ' i the behavior of investors from a risk-return perspective, the question now arises what ramifications does i t have I j for the pricing of individual assets? Considering that each investor will hold a well diversified portfolio of 1 ! I risky assets in combination with a riskless asset, i t is j ! . immediately apparent that the risk of any asset which t ■ » i would be examined on an individual basis would only be of j interest to the extent i t affected the returns to the ; portfolio into which it would be incorporated. That is, j | only risk which can not be diversified away will be of any ! ■ consequence to an investor and thus that is the only risk : I | component for which the market would require a premium. j I That risk component—the extent to which the asset*s i i returns vary relative to the market portfolio as a whole— j I j is known as systematic risk. It can not be diversified ; away and is measured and referred to in financial literature as the asset's Beta. That portion of an ; individual asset's variance which can effectively be I eliminated through the process of combining it with other I 3 i assets in a portfolio is known as the asset's unsystematic I t i risk. Since it can, by definition, be entirely eliminated j | t through diversification it is of no consequence and would j not be expected to bear a premium in the market place. ! 32 The plot of the trade-off between the Beta for any i ' asset (the extent to which its returns vary relative to I ! the returns to the market as a whole) and its return is : known as the Security Market Line (SML). This is the | return risk relationship determined by the market place ' i and by which all assets would be priced. This, then, is ; I } . the basic concept involved in the CA PM ; that assets are : • f priced according to a linear relationship between return ! : i to an asset and the extent to which its return co-varies • i with the market returns as a whole. I | I I .5 A R B ITR A G E PRICING T H E O R Y j The Arbitrage Pricing Theory, as indicated earlier, i 1 ; can be considered a general version of a multifactor model i I * < I of which the C A PM is a rather constrained type. The ; essence of the APT is that when, by constructing a * portfolio in a particular fashion, an investor achieves a position in which he has no net investment and no risk he | ! ! ■ must in equilibrium earn a zero expected return (Ross, j | I 1 1976). A portfolio which has been constructed in this way j | ! is called an arbitrage portfolio. The returns on the : assets which make up these arbitrage portfolios are i | themselves considered to be a function of some unspecified i number of factors common to all assets. Through empirical j examination (by means of factor analysis) of arbitrage portfolios of varying make-ups, it'i s possible to arrive at a breakdown of vectors or factors which correlate with observed returns. Through this technique one would also generate correlations between the returns of the assets in I 1 the portfolios and the several factors which appear to be ! ! ; i I the determinants of return. j i i \ Obviously, the A PT is more elegant and less encumbered i i I 1 by constraining assumptions about asset markets than the | ! CAPM . Specifically, the APT makes no assumptions regarding the nature of the distributions of assets' 1 : i returns or the shape of investors' u tility functions I (other than rationality and risk aversion). The market j I I portfolio plays no part in the development of the theory j j and, unlike the CA PM , there is no need to assume one is J ! dealing with all tradeable assets when carrying out | I ' ■ empirical analysis. Finally, there is no requirement that i J returns be a function of only one factor; rather, any ; number are possible. A disadvantage of the APT is that while factor analysis will isolate the significant factors which are the determinants of asset returns, i t does not I i i identify what those factors are. Analysis suggests that ; \ | there seem to be perhaps five factors at work in the j ! pricing process but, alas, they have not been identified j (Roll and Ross, 1980). ! I 34 HTI.6 R ESU LTS O F EM PIRICAL STUDIES I Empirical findings in the many studies of the C A P M have been intriguing and illuminating. While ideally it is necessary for returns on assets to be normally distributed for the C A PM to hold, indications are that even though this could not be the case (an investor’s down I I side risk would be limited to -100% for instance) the j ! distributions of returns on most common stocks are, over a , ' wide range, insignificantly different from being normally i j distributed (Fama, 1965). When looking into the question | I 1 of the linearity of the relationship between return and 1 systematic risk the case for linearity is almost always . i . supported (i.e., not rejected). I I i The inquiry into the issue of a riskless asset, i i I | : however, has turned out quite differently. In most studies the intercept values for return at a level of no j ; risk have been too high to be legitimately associated with ; returns to risk free assets. An alternative to the ! combination of a risk free asset and the market portfolio ; has been proposed in the form of the zero Beta portfolio ! 1 which, as it designation suggests, would be uncorrelated : with the market portfolio (Fama and Macbeth, 1973). ! i 1 I Another difficulty with the theory and empirical , I i results has to do with the slope of the SM L. Time and | again results have shown that the slope is not as steep as ! I — ' ' . 1 .would be called for in the model. This would indicate , I ' that low Beta assets are earning too much and high Beta ; assets too l i t t l e . Finally, when viewed in the aggregate, systematic risk does prove to be by far the ‘ I ’ most complete explanation for differentials in the level ; of returns. I ! A serious theoretical objection has been raised about I I 1 the C A PM which brings into question the validity of I J virtually all empirical work which has been done on the ; I model. This objection rests on the fact th at i f a ! i I I , portfolio against which returns are being regressed is | , I , mean variance efficient then the relationship between the j ! ; 1 asset's return and covariance with that portfolio will by I | ■ : mathematical necessity be a linear one (Roll, 1977). 1 I ! Thus, any test of the C A P M is a test of both the linearity | ) ' I | of Beta and the mean variance efficiency of the market j i 1 portfolio. In order to separate those two hypotheses i t | i I would be necessary to establish that the genuine market j ! portfolio of all assets is itself mean variance efficient, j This, of course, is an impossibility given the number of I assets of which i t would be constituted. This is not to J suggest though that the C A PM is invalid but rather that , one must approach empirical analysis of the model with great care. > I I .7 S U M M A R Y There are a whole range of assumptions involved in ; the C A PM and APT which may seem to be unrealistic and which potentially could bring the usefulness of those models into question. This has long been recognized and ' many investigations have been made into the robustness of the models in the face of relaxation of one or another of I those assumptions. Results of those studies have been mixed but the models seem to be rather more robust than I one might i n iti a l l y expect. The models are not I I discredited and appear to be at least of some use in I , explaining the behavior of common stock pricing. I I I I 37 C H A PT E R III IM PED IM EN TS T O R E A L ESTA TE A SSET A N A LY SIS t ; III.l IN TR O D U C TIO N 1 From the discussion in the preceeding chapter it is apparent that while the C A PM and APT are not perfect theoretical constructs from all points of view they both do have aspects which recommend them for serious consideration as models which generally describe asset | behavior and into which one might initially attempt to fit I real estate investment. What follows here is a discussion i of the aspects of real estate as an investment which could be expected to cause p articu lar d iffic u lty in its | incorporation into the C A PM or A PT constructs. i III.2 DIVISIBILITY The need for the market within which asset are traded I i , to be one in which investors are price takers and which | operates frictionlessly presents significant problems in ■ the area of real estate. Consider firs t the requirement ' that assets be infinitely divisible. Real estate is I ! traditionally thought of as being a "lumpy" asset. The i i I 38 market place does not have provisions for breaking down real estate assets into small components which are easily traded. Some limited approaches to divisibility however are described below. I I III.2.1 Tenancy in C om m on 1 i I i A n asset need not to be physically divisible to satisfy this requirement. Tenancy in common whereby a single piece of real estate is owned d ire c tly and • I concurrently by two or more persons is recognized | j throughout the United States. With a tenancy in common ! each part owner owns an undivided, but not necessarily equal, interest in the property. This type of ownership | is much more common for non-investment as opposed to j i ! investment type holdings. The number of owners is usually j small and there very often is an association between them, ! such as other business activities or family relationship, , i i in addition to the holding of the real estate. By law ' such an owner can freely dispose of any part or all of his holdings? from a practical standpoint however, there are , often difficulties. For instance, sale of an interest to ( I one of the other co-owners would only be possible if one j ! of those others were simultaneously looking for just such j i an acquisition. O n the other hand, sale of an interest to an "outsider" would involve that outsider becoming a co- ■ owner with a group of individuals usually having other i relationships among themselves and on a basis which establishes no formal goals or management procedures with | respect to the property. Thus, practical considerations , having primarily to do with marketability mitigate against tenancy in common forms of real estate ownership as an I approach to achieving unrestrained divisibility. i j III.2.2 Corporate Ownership I Other than direct forms of multiple ownership are i ' possible through several alternate means. Those involve i multiple ownership of various entities which themselves , hold t itle to the real estate. Ownership of real estate through a corporate entity is of course one of those alternatives. In fact, a very significant proportion of ' all corporate holdings consists of real estate. However, in the vast majority of cases in which real estate is held : in corporate hands, there are also other classes of assets held at the same time and usually other real estate j properties as well. Further, the holding of real estate in a corporate mode has two disadvantages centered on tax i treatment relative to other indirect holding methods. | Those disadvantages are, first, the taxation of investment income at both the corporate and individual levels and, second, the inability to charge off real i I i i 40 estate generated tax losses against income from other sources or assets. Consequently, while corporate holding of real estate is commonplace and corporate ownership i ; interests are readily divisible, other considerations such as the propensity of most corporations to hold a variety of assets and the relative tax disadvantages seem to have : prevented this holding method from becoming the answer to divisibility of individual real estate interests. Holding real estate through a real estate investment trust (REIT) provides for easy divisibility of ownership 1 interests just as does the corporate mode but with the i added advantage that the taxation of income at two levels ! is avoided. However, except in the very rare instance of ! a specialty trust, REITs hold portfolios of real estate (augmented often in small proportions by other assets) ; rather than single properties. i I III.2.3 Partnerships Partnerships holding real estate can avoid the tax J disadvantages inherent in corporate and to a lesser extent I REIT ownership while at the same time providing for i |d iv isib ility of interests. It is common to divide these i I real estate partnerships into two classifications: Public J and Private. Public partnerships are those whose i | offerings have been reviewed and approved for completeness 41 ! "by governmental securities registration agencies. They | are for the most part large undertakings in which many I limited partners participate and which hold a number of different properties. The interests in partnerships are ; limited in that holders do not engage in the management of : I J the partnership as a whole; this is the responsibility of | | : | designated general partners who are usually real estate | | professionals often with ties to major institutional organizations. j i The in te re sts are available in a variety of I (denominations but the lower limit is usually rather more I ' i than the minimum amount of a portfolio adjustment which an I ! i t i I individual may wish to make. W hen the public partnership | is established the interests are widely advertised and ' 1 I ! frequently marketed through major brokerage houses or j other institutional organizations. However, upon entering J t * | into such a partnership partners must recognize that j ! resale opportunities are quite limited as a market for I secondary transactions usually does not exist. Within , these parameters, public real estate partnership activity i ; has grown enormously in recent years with billions of i I dollars of annual growth in placement activity. j | Private partnerships are those which are limited in | j size to the extent they need not register for processing i with governmental agencies. This limitation in size has 42 { to do with the number of partners who may participate. As a result of the smaller number of partners (as few as two ; are possible), these naturally can be expected to have holdings at a much smaller scale than public partnerships. In fact, private partnerships for the purpose of holding modest, individual properties are very common. In ! aggregate, because of th e ir vastly greater number, : holdings of all private partnerships far exceed those of the public variety. By reason of the small number of u ltim ate participants in each partnership the initial marketing is modest by comparison with that for public offerings and I ) again, once in such an investment, the lack of a secondary I market severely lim its one's ability to exit before i ! liquidation of the organization. In many respects, therefore, the private partnership is rather like a tenancy in common with the added feature of a formal I organizational structure which establishes goals and ; operating procedures to be followed from inception to ! i completion. In summary, as conventions now have it, real estate 1 presents serious problems from the d iv is ib ility : standpoint. For the most part real estate ex ists, physically, in components which are far larger than any individual may find convenient for the purposes of : portfolio adjustment. Attempts at a physical division or partition to pare an interest down to a desired size would I I be extremely ill-advised in the vast majority of cases. Bearing that in mind, division of ownership interests rather than the physical property has great appeal. As discussed above, however, the present modes of I multiple, concurrent ownership have developed in such a ! way as to p resen t one or another of two major difficulties. In the case of divisible real estate ownership interests which are relatively marketable one is 1 almost always presented with a situation in which what is ; held is not a single property but rather a portfolio of a i | variety of properties. O n the other hand, if an ownership I i j interest involves only a single property, then i t is ; almost certainly one which is, compared to many other types of assets, quite constrained with respect to marketability. Thus, real estate is a far cry from in fin ite or even easy d iv is ib ility and therefore applications of theoretical constructs for which that feature is an important requisite must be approached with > great caution to say the very least. I I | III. 3 IN FO R M A TIO N ♦ The issue of inform ation probably p resen ts : difficulties of equal magnitude for the application of I 44 ; popular financial theories. As with other types of assets, real estate is certainly subject to the influence ; of both national and regional events. However, due to real estate's fixity in a physical sense and the great I extent to which i t offers its services to what is quite often a very geographically localized market i t is ' ■ probably much more sensitive, re la tiv e ly , to local economic, political and social factors than many other j types of assets. If this is accepted then an argument can foe made that t in the case of real estate investment a prudent investor | is faced with potential information sets for various ' ■ properties which would seem inordinately large and I ; possibly less easily discernible than for investment possibilities within other asset classes. If for real estate investment i t is true that information is more I critic a l or that there is more of i t which is relevant, \ then disparities between theoretical assumptions and observed market conditions are lik ely to be more i significant and troublesome than for other classes of i i i assets. t I | III.3.1 Information Sources i As implied above, real estate tends to be offered, in ; many cases, in what is in effect a very local market. 45 ! This is not to say, however, that the transactions take ; place within a formally organized exchange. In the vast i . majority of cases they do not. There is rarely a single clearing-house through which all potential buyers and sellers make their positions known, and there is usually very fragmentary or incomplete reporting of transactions. 1 What information collection and dissemination occurs is ! usually performed by real estate brokers as a portion of ! their brokerage services or by real estate investment j I counselors who also often are involved in brokerage or I | syndication activities. i 1 i I 1 The evolution of the market to this point is not * ! ! unexpected given the substantial information requirements | i ; for considered investment and the relatively infrequent ; ; forays into the market place which many participants make. | | 1 ! There xs a development, however, whxch may brxng about a i j trend toward less of a local market for a greater i ■ • ; proportion of real estate investment opportunities. That j : is the increased role which is being played by i i > | syndications of both the public and private variety. As | J . f more and more of real estate is held by these entities it | I should be expected that standardization and other efforts j |a t enhancing m arketability w ill lead to individual | J properties effectively coming under consideration by a ; f very much larger and more geographically diverse j I I » , assortment of potential investors. i III. 3. 2 Information Categories I Information which is of in te re s t to potential 1 investors is often thought of as fallin g into two categories. The first, general information, would include | knowledge of particular circumstances or events which | would have an impact on the class of assets being ! examined. The second, market information, would be that which has to do with knowledge of prices bid and asked, identification of prospective buyers and sellers, and I | determination of the quality or distinguishing attributes of the assets being exchanged (Hirshleifer, 1971). A n example of general information might involve anything from I I i the recent trend of the weekly national M l money supply [ figures to an unannounced decision by a governmental agency to impose a building moratorium based on newly projected public utility and service inadequacies. Market information can be illustrated by the example of the I ; knowledge that Ajax Partnership is currently offering for i ! sale its centrally located and completely rented-up | 100,000 square foot Commerce Office Tower at a price of $25,000,000. General information, by reason of its very nature, might be available from any of a wide range of sources. Public announcements or reports in various forms are obvious examples and probably the source of most information upon which investors base their decisions. Non-publicly available information, however, can also be of great importance and value and may be obtained by an investor through almost any means ranging from the illegal i to the most innocent happenstance. Acquisition of general i information would not be a function of the structure of the market for any particular type of asset. Obtaining what has been defined earlier as market I information, though, would be highly dependent on the ■ characteristics of the market for the asset involved. | Prices change as buyers and sellers enter and leave : markets with varying frequency. In the absence of a i completely centralized market a prospective trader can not ; know all the prices (bid and asked) or all the other .p o ten tial trading partners at any given time. In a d d itio n , inform ation about re c e n tly completed transactions would not reliably be available. These short :comings in markets which are not completely centralized !cause investors to seek out of information concerning j prices and participants. i I j III.3.3 Information Acquisition and Utilization | Search activity will be carried-out to a greater or ‘ lesser extent, based on several considerations. Among | those considerations are the investor's estimation of the bid-ask distribution in the market and the proportion of ' his wealth which will be involved in any transaction i (Stigler, 1961). Basically, as with other economically determined activities, the optimal amount of search to be J carried-out should be that amount which equates the i ! marginal cost of the searching with the expected marginal ; gain from that activity. With unique goods or, as real estate might be considered, with heterogeneous packages of an asset, the cost of search is very high because of the I small number of prospects existing in the large population which must be searched. i This expense is a powerful inducement to localize I transations for the purpose of identifying potential ! buyers and sellers. Localization of transactions can take place in several ways. One way is to establish dealer markets in which a dealer stands ready to consumate ' ■ a transaction on short notice, and earns a return based on I I this provision of marketability. In real estate markets i i 1 the localization of transactions tends to be accomplished , by means of brokers providing marketing services such as I i advertising and ofher measures to ensure adequate i exposure. In this way a broker, a multiple listing I service, or classified advertisements can take on the i I role of a consensual central exchange. Market information concerning the quality of j attributes of differentiable assets summons up two , problems. The f i r s t has to do with the matter of specifying the level of quality or meaningfully ' quantifying attributes (Hirshleifer, 1973). In this area one faces the hazards both of su b jectiv ity and of agreement on how to specify most appropriately asset j j c h a r a c te r is tic s . The other challenge involves I establishing the authenticity of representations made , about assets which are being offered for sale by sellers. I ] Im p lic it in th is perceived need to provide ■ authentication is the assumption that an asymmetry of I 5 information exists between trading partners. In other words, in the common case there are things known by the seller about the asset he is selling which i t is in his ; interest to disclose or reveal to prospective buyers. The ; difficulty arises in this being done in a way to give the recipient of the information confidence in its reliability , despite the obvious potential for a conflict of interest I on the part of the provider. There are a number of |approaches to a solution. The seller of an asset may j engage in informative advertising where verifiable facts J are provided, guarantees which convincingly bind the j seller may be offered, or signaling may take place. With L_ 50 I I many types of goods it is possible' through brand names to ; provide a measure of authentication to claims made in the market place. I In the case of real estate, i t is rarely possible to I provide brand name assurance on individual properties. i However, a similar effect could be achieved through , i recognition of the reputation of agents (such as brokers , ; or real estate investment counselors) with whom an | I 1 j investor deals or perhaps the initiators of syndicates who i i ' ; I * j offer interests in one or more properties. If real estate j ^ comes to be held more and more by syndications which are I i put together by persons widely experienced in the industry j it can be expected that authentication will be more easily | i ! achieved than would be the case with less initiated investors operating on their own. i If one comes into possession of information which is | not known by other participants in the market place it is | j legitimate to ask to what use that information should be j ! put in order to yield the most value to its holder. In J ‘ t addition to using information about some asset to actually ! trade in that asset, the holder should also consider the p o ssib ility of trading in the information its e lf , j i < I Transactions of that sort in information also have to take ! into consideration the need for authentication as well as i ! the protection of the information from resale. The latter 51 1 concern would be particularly d ifficu lt and might be t ( ; enough of a problem to convince the holder of exclusive : information to forgo the trading in it and to settle upon trading in the asset to which it relates. If one decides to trade in an asset its e lf oh the j basis of information which is known to be unavailable to j i one's trading partners there exists a situation much like j , that in which an insider is present. In that situation \ | i t is in the interest of the insider that his status and I I information not become known. If it does not become known he can expect to earn returns on it. If, however, it does i j become known that some insider is present or if the actual ! id en tity of the insider is revealed, then gaming j strategies can be expected to be implemented by the non- j i : insiders with a cost and commensurate reduction in returns ; / | to the insider. | In a market in which real estate is traded and where i | the identity Of traders becomes well known to such other i participants as brokers and agents, the likelihood is i j quite small that one who frequently attempts to reap large I |g a in s from exclusive inform ation would remain I | inconspicious. Other traders could be expected to learn J ! ! j of his presence and to alter their conduct accordingly. , III.3.4 Optimal Production of Information I If it is assumed that information in a marketplace or exchange is valuable, i t can be expected that there will be those who will seek out that information with the * intention of trading on it themselves or selling it to , I , others. If the production of information is considered to ; « j be an activity which consumes significant resources the . , < question arises as to how much information production is 1 , ; appropriate or optimal. The view is held by some that ; i i when information production is individually optimal it is j ' at the same time socially sub-optimal (Draper and Findlay, I 1982). The basis for this opinion is the difference between purely redistributive and productive allocational J | effects. That is, the private redistributive value of new i 1 information is believed to be sufficient to motivate ■ i : ; efforts beyond those required for merely the improvement : ' of production allocation. While there is not unanimity on i this point, holding that an optimal amount of information | I will be produced may be too optimistic. I ( A number of factors have been presented which could | influence the way in which information should be treated i j in the modeling and analysis of the real estate market, j ! Legitimate questions remain as to what degree of operational and allocational efficiency exists with respect to this informational aspect. Inefficiencies in I i I ' 53 [ this area, however, should not be taken as precluding or automatically defeating any analysis. Rather, they could be taken to indicate that more imaginative and innovative i approaches are needed to provide greater insights into the 1 behavior of real estate assets. ' III.4 TR A N SA C TIO N C O STS I ( Transaction costs are another area which could present considerable problems in the study of real estate investment through traditional financial models. Real I ! estate historically has been recognized as an asset for which, compared to other classes of assets, the costs of transaction are quite high. Transacting in any asset I j involves a number of common steps: comparing alternatives, ; communicating offers, negotiating terms, carrying out and | verifying execution of those terms, and ensuring against i ■ non-performance. Expenses for those activities are ; incurred in almost every exchange in which an investor ! I moves from one asset to another. In the simplest sense, i I j the magnitude of a trade's transaction cost is the net ; difference between the proceeds which are received by the seller of the asset and the total of all costs borne by i ^ the buyer. j The costs of using the usual market structure to i I | execute a trade are often thought of as falling into two 1 different categories, marketability services and exchange , ' services (Trinic and West, 1979). W hen dealing with a non standard good (or one which does not transact frequently) the seller or prospective buyer may be willing to make a price concession in order to avoid delay. In such t : situations, for some types of assets dealer markets have j evolved with the purpose of providing marketability ! ! I i services. Here the dealer takes on the role of an ! | ' j intermediary who specializes in offsetting temporary ! i imbalances in the market. In effect, two markets emerge, t i i , One develops for sell orders and another for buy orders. j I j The bid-ask spread between those two markets should equal , i : the dealer's marginal cost of holding the asset. The more I » frequent the trades the lower the holding cost will be. j 1 This sort of solution is seldom seen in the real estate : 1 markets. ; i ! The other c la s s ific a tio n of transaction costs, exchange services, will often involve the compensation of various parties for particular services in the areas of j search for traders, fa c ilita tio n of negotiation or t 1 bargaining, investment counselling, provision of other ; market information, and sundry other a c tiv itie s . ! Information, which has been covered m the previous ( section and is often treated separately by others, will be ! , < i excluded from this discussion of transaction costs. In the [ ! case of real estate, brokers perform many of these various exchange services as part of their normal function. For : their efforts, the brokers are usually compensated on the basis of a commission tied to the value of the property involved. Others who also frequently participate in real i ' estate transactions are escrow officers, title insurance ; agents, notaries, accountants, and attorneys. In addition to expenses for those professional services, loan initiation fees upon acquisition and occasionally loan i j prepayment charges upon sale can assume significant ! proportions on properties which are financed in part 1 by debt. Of all of these assorted costs, brokerage fees are usually the highest. They are ordinarily set based ■ both on the value and type of property being sold. I I Commissions on vacant land sales for instance will usually i be substantially higher than for an office building and the percent of property value charged as a commission on a $2,000,000 property will be significantly lower than for a i $100,000 property. Complicated transaction cost situations do not i necessarily mean that inefficiencies will be present in i | the marketplace. Opportunities for inefficiencies are I |created but, to the extent the price charged for > ' transaction services is equal to the cost of providing the services, the market will be functioning efficiently 56 (Demsetz, 1968). The presence of transaction costs will have an impact on the frequency with which trades are made for any asset. The higher the cost of trading the fewer trades will be made. The operative rule which an investor must follow is: when considering a move from one asset to another the marginal benefit of taking a position in the new asset must exceed the cost of the transaction required to make the move. Using a common model to analyze different classes of assets for which transaction costs differ greatly is fraught with danger. To the extent transaction cost are added to the exchange process additional demand in terrelatio n sh ip s are introduced which can not automatically be assumed tractable in the context of models or theories which hold that exchange is costless. The transaction cost question perhaps could most easily be answered by maintaining that in consideration of the magnitude and uncertainty of those costs any model dealing with real estate specify not just a point estimate of value but a range instead. This begs the question though; how much of a range is appropriate and on what basis? Further, the question might also be asked as to how useful a model is if it produces only a wide range within which any value would be legitimate. 57 I I I .5 TR A D IN G IN A LL A SSETS ] I i Lack of marketability of all assets is a rather less , obvious issue when applying traditional financial models to real estate but one which merits brief examination. I Such lack of marketability or trading of assets to which j significant returns should be attributed can thwart analysis of pricing mechanisms. As explained earlier in I ! the case of market segmentation, i t would be futile to conduct empirical validations of financial models using j i data from assets which trade in a variety of (i.e., | i segmented) markets. Assets traded in separate markets ! I , i : could not be assumed to be governed by the same models, ; ' and thus it would be inappropriate to attempt to explain ! f I , the behavior of any asset except by means of the model ; ! ! i which pertains to its market. The mixing of models and | i i asset classes could easily render invalid any resulting | 1 i r : empirical findings or conclusions. Similarly, there could ; I I also be severe problems if , rather than improperly j I matching assets and models, one were to attempt to explain j 1 t observed returns by means of a model which entirely 1 ignores one or another sets of assets from which a portion of those returns is derived. i When one considers many aspects of real estate investment, particularly information, it should not be unexpected that the role of human capital will arise. As ; with other assets, human capital is employed by its holders in the furtherance of investment activities. Depending on human capital's relevance to the returns in any particular market, an investor will enjoy relatively : greater or lesser advantages based on his endowment of, , ; I , ability to maintain and capacity to augment his human capital. Given for instance the information constraints i i j in the real estate markets and the need to monitor agents ■ j i | activities, real estate is a viable candidate for being a j ' market in which human capital could be very significant. j I To the extent human capital does play a part in i 1 . i | information gathering or the monitoring of agents, it does j ! so as a return not to the marketable real estate asset but ! rather to the nontradable human cap ital. By being i nontradable, however, human capital can not be accounted for in a conventional sense in any model which focuses i ' : strictly on behavior within a transaction context. Hence, : 1 j ; if a nontradable asset such as human capital does play an ! important role in any portfolio, the explanation of the 1 : behavior of the other (tradeable) assets by means of a 1 , model which recognizes only their contributions will be seriously flawed from the start. i \ I II.6 R E T U R N A N D RISK M E A S U R E M E N T i . I j S till another problem one might encounter while j ! ! 59 1 adapting real estate to generally accepted financial 1 | theories is the very measurement of return and- risk ' (Findlay, Mesner, and Tarantello, 1979). Most real estate assets tend to have a holding period (and thus an opportunity to observe a market price) which extends over j what for many other assets would be several (at least) trading intervals. Information on actual trades and price 1 I ! levels in those cases is available for many classes of j : j assets on a much more frequent basis than is generally the ; I . i | case for real estate. Very simply then, the question ; j ! : should be asked as to whether or not there is an inherent 1 problem of lack of comparability when observations for determining the parameters of the returns to real estate j I ' I are made only very infrequently compared to other assets, j j Consider a case in which two common stocks listed on j i the N ew York Stock Exchange are subjected to examination over a period of some years. One is studied using its readily available daily transaction data while the other j is viewed only on the basis of transactions every other { I ; year or so. This incongruence is in effect what one is j i ! ] faced with the case of most data gathering and analysis i I i i which includes real estate along with other much more I frequently traded assets. i For a mix of asset types paired observations over concurrent time frames are not often present. Real estate t I t 60 i , i j returns are usually expressed on an annualized basis in terms of IRR or FM RR while an asset such as common stock ; ' usually has its return expressed as an annual holding period return. This annualized versus annual return incompatability will also render comparison of variances , inconclusive. Any model which depends on consistent and ; | compatable return and risk measures among all the assets ] I . • to which i t is applied will fail if those conditions are j i not present. In most cases they will not be present if ' i | real estate is combined with other assets such as common j I 1 stocks. J I I i III.7 S U M M A R Y ; In this chapter real estate has been shown to differ j I ■ from common stock and other financial investment j ' i 1 instruments in a number of respects which potentially 1 : present significant problems in utilizing models such as ■ the C A P M or A PT to explain pricing behavior. Those most J ’ significant differences include assumptions regarding : I i frictionless markets, tradability of assets, and measures ! j l , of rate of return and risk. ! . I ; ! j First, with respect to frictionless markets, real * estate seems to fa ll far short in the areas of | divisibility, information, and transaction costs. While | I i real estate is not easily divided physically into j ' # . i , conveniently sized trading units, the division of ; ownership interests is possible. That however is hardly a ! solution in that i t raises the additional problems of , marketability of interests in closely held properties, i * ownership interests being in a portfolio of a number of j properties as opposed to an individual property, and the ' substantial amount of management and professional services i . i j being jo in tly purchased xn many such transactxons. j 1 j Moreover, since gathering information may be a greater ' i ; t factor in determxning returns to real estate than for many j : other assets, its consideration may require more complex i t treatm ent than is offered by models which assume i j i fric tio n le s s markets. Transaction costs, i t is | | i J occasionally suggested, can be accommodated by relying on | j models specifying a range within which a price should fall 1 as opposed to a single value. The nature of the costs ; encountered in real estate transactions, however, may be such that the specification of a range would have to be so wide as to eliminate the usefulness of the model. Second, the ability (or not) to trade in or market I assets can be a serious concern. Severe difficulties are j sure to be experienced if one is trying to explain price I j I \ behavior which is a function of returns to several assets > i . ! by examxnxng only one of them. Dealxng wxth xnvestment , real estate, which may well be dependent on non-tradable , ; human capital for a significant proportion of its returns, ■ is an instance where the threat of such d ifficu lties is quite real. Finally, in dealing with a portfolio of mixed assets or when making comparisons among assets using a particular model it is essential that return parameters be expressed in a consistant fashion. In the case of real estate and I assets such as common stock this consistancy is clearly | not present. For real estate, returns (and also risk) are I I i expressed as annualized figures because actual annual I return information is rarely available. Figures presented for stocks on the other hand are usually true annual , return based on trades taking place over that interval. i : Mismatching these measures of the essence of those I ; investments dooms any attempt at genuine comparison. Accordingly, it would not seem propitious to embark , upon an analysis of semi-strong real estate market | efficiency using a variation of either the C A PM or APT. I , Further, i t would s t ill be extraordinarily d ifficu lt to i : carry out an empirical study even if these significant d ev ia tio n s of re a l e s ta te from the id eal a sse t characteristics embodied in those models were to be , elim inated. This conclusion is for the most part I ! attributable to the dearth of market data on real estate | transactions and valuation. This lack of adequate data to incorporate into empirical study or investment decision models "has not escaped notice. There are several trade organizations which, in the last few years, have begun compiling market transaction and valuation data on real estate in various , markets. Several are building data bases for eventual f marketing to as yet unspecified users. At least one is I j being compiled on a cooperative basis by and for an | association made up of institutional investors and pension fund fiduciaries. However, because of the brief time period covered by those data bases and the initially small i [ sample coverages, none is suitable at this time for use in a rigorous analysis. i I 64 I i i C H A PTER IV R E A L ESTA TE A N D FA C TO R PRICE EQ U A LIZA TIO N IV.1 IN TR O D U C TIO N The discussions and analyses presented in Chapters II and III dispense with the idea of conducting an inquiry [ into semi-strong market efficiency as an explanation for supposed persistent differential regional real estate 1 investment returns. Attention, therefore, will now be i turned to the question of regional market segmentation ■ ' based on factor immobility and trade barriers. To address this issue the factor price equalization theorem will be utilized. Factor price equalization is a consequence of an equilibrating process in a market or markets operating ; under certain assumptions of efficiency and competition. ] Trade theory speaks to this issue and has been widely i studied in a context of international trade and regional ! I analysis. ; First, consider one market and the prices and returns ; which could be expected to be observed. The following conditions will be said to hold: a) homogeneous products I b) perfect information c) free entry and exit by buyers J ! 1 and sellers d) no transportation costs within the market e) no transaction costs f) numerous buyers and sellers. One would expect under these conditions to observe that | any particular good or commodity would trade throughout j : the entire market area at one price. If for some reason 1 I ' this was not in itia lly the case, that is if some good sold ; : for a different (relatively higher or lower) price at one ; i 1 location than at others within market, arbitragers would be motivated to engage in trade. Through that trading J ' activity price differentials would be driven to zero. ; ! Tbus, price would be driven to an equilibrium for all ! * i i catagories of goods throughout the market area, and a one , j price law market would prevail. j j j If those assumptions of a perfectly competitive I , j : market are maintained, an analogous conclusion can be i : reached with respect to the pricing of factors of i production. Price d iffe re n tia ls for a p articu lar classification of factor would be driven out and an equilibrium reached in which the price of each factor is ! I I equal to the value of its marginal product. j : I i I i | IV.2 T R A D E T H E O R Y i Classical trade theory holds that costless mobility I 1 ! i I of commodities and factors of production between different ; I markets would tend to equalize prices of those commodities and factors in the several marketsi The process by which | price would equilibrate would be the same as in the one ; market case. In fact, the designation of several markets » with such free movement of both commodities and factors ! \ 3 1 begs the question of their designation as other than one • I market. When lim itations with respect to factors and i , ; commodity mobility are eliminated then the designation of ; I several regions as different or separate markets is no j t longer appropriate. j i At this point the theoretical question arises: To what extent do limitations on factor mobility or commodity I ;trading between markets effect the prices of those factors j j and commodities in the different markets? Great emphasis j I ' has been given to th is question in the context of | international trade. The view has been put forth through ;successive theoretical analyses that commodity mobility | I , coupled with factor immobility, or commodity immobility coupled with factor mobility (as well as combinations of j jthe two) can lead to equilibrium adjustments resulting in price equalization in the different markets (Greenwood, |1976). i I Taking the case of commodity mobility and factor I immobility, i t has been demonstrated with certain 1 i 'simplifying assumptions that pricing outcomes will be the , ; same as with perfect factor mobility (Samuelson, 1948 and 1949). While the assumptions are recognized as being I ' sig n ific a n tly more re s tr ic tiv e than actual market ; conditions and caution should be exercised in interpreting i or drawing conclusions based on the model, such elementary ; examples often offer useful insight to understanding I complex problems. The main assumptions from which that I ! outcome followed are: 1) two countries 2) two factors of production 3) two commodities being produced 4) production functions are identical in each country and | homogeneous of degree one 5) commodities move costlessly i I between the two countries. In this example which demonstrates the validity of | the hypothesis the two factors of production are land and I I I labor and the two goods or commodities are food and , clo th in g . A fter e s ta b lis h in g the fa c to r p ric e , equalization under those circumstances the theorem is then I shown to be extendable to cases in which there are any number of goods and any number factors as long as the . number of goods plus mobile factors equals or exceeds the total number of factors. If the number of factors should : exceed the number of goods, mechanisms do not exist which i would cause factor prices to equalize in the different markets. i I I I A later study takes the factor price equalization ! theorem analysis a step further and develops a proof for the equalization of both factor and commodity prices in a . set of markets in which factors of production are perfectly mobile but with trade in commodities so constrained that no inter-market commodity trade exists ! (Mundell, 1957). While the case of free commodity trade I ] and factor immobility may seem to be closer to the actual i I market conditions which exist between countries, the opposite can be of use in examining some international trade issues. i ! Consider then what is implied by the two means by I which factor and commodity prices are driven to equality , in d iffe re n t markets. On one hand an increase in t ! impediments to trade in commodities stimulates factor ; movement between markets and on the other hand an increase . in restrictions to factor mobility stimulates inter-market I trade in commodities. The most apparent application of . the former to an international trade issue is in the area ; of ta riffs. It would follow from the above that as ! ta riffs or other man-made barriers are erected against inter-market commodity trading an impetus will be provided to stim ulate inter-m arket movement by factors of production. Another example of a change in the ease of ! commodity trading would be a change in transportation ! costs. Once again, as with the immobile factor and free I ' trade model, this mobile factor and constrained trading model includes assumptions which are blatantly violated in . actual markets. This should not, however, prevent one i ! | from gaining useful insights into the general responses which are likely to be induced by the various stimulae I ; that are applied to the model. I ! For the interested reader, both in tu itiv e and I I ' mathematical proofs of the factor price equalization theorem under conditions of free commodity trade and factor of production immobility are offered in Samuelson (1949) and Laffer and Miles (1982). | IV.3 T R A D E T H E O R Y IN R EG IO N A L A N A LY SIS i The assumptions underlying trade models which have been widely used to consider international trade questions, as mentioned, are rather restrictive in their ' eliminating consideration of potentially significant aspects of such a market's structure. Absolute free trade ■ in commodities coupled with factor immobility or vice I j versa are unlikely to be strictly true in an international ‘ market. The actual condition would surely lie somewhere I 1 between those extremes. In the case of regional markets within one country actual market characteristics are also | probably short of those assumptions but arguments may be | made that the regional case more closely conforms than the international one. I The question then may arises is the factor price i equalization model more appropriate or valid for regional I ' analysis than for studying international trade? In other words, should the apparent closer f it between the model and actual market conditions inspire significantly more | confidence in the model? Would regional equilibrium be f j attained more quickly than in the case of an international I system? Or, what might limit or prevent the attainment of | factor price equalization? If the model does fail or is i s t i ll open to criticism i t would probably be in the area j of the assumptions being too unrealistic or because of the natural lim itation of trying to apply a static analytic j procedure to a dynamic situation. Considering some of the model's assumptions we can look first to the issue of free trade in commodities. In dealing with a set of regional markets in the United | States the question of ta riffs becomes moot. Such man- | made b a r r ie r s are c o n s titu tio n a lly p ro h ib ite d , j Transportation of goods between regions becomes a : legitimate issue, however, because of the very real j physical distances involved and the positive costs of moving goods any unit of distance. Likewise, with factors ; of production such as labor or physical capital the cost ! of movement (physically) must be considered. Regional economics is steeped in variables associated with distance. It "has been demonstrated that in a competitive equilibrium not only are factors priced according to their marginal product but they will differ in price between regions by the marginal cost of transporting the factor from one region to another (Lefeber, 1958). Another constraint on cap ital m obility may be imperfect knowledge or information. In order for a decision to be made with respect to migration from one i j region to another the risk and return characteristics of i the investments in the alternative regions must be known. i ; That information may not be available. Further, rigidity j in the capital markets through institutionalization or , regulation can have an adverse effect on mobility. And, finally, differential regional tax structures may also be an impediment. ! The assumption of production functions yielding constant returns to scale (homogeneous of degree one) is also somewhat troublesome. In regional economics and : location theory in particular a great deal of attention is given to the question of why certain economic activities i | tend to group together or agglomerate at particular locations. Those observed tendancies on the part of some activities is not easily explained without relying to some 1 extent on production functions which exhibit positive , economies of scale. The preceding identifies questions and presents brief ■ observations concerning assumptions underlying the factor price equalization model. It does not in any way purport |to make a convincing argument one way or the other ! concerning the extent to which those deviations of actual ; markets from the model's ideal will affect its usefulness I ! in explaining market activity. However, beyond the issue of the conformity between the model assumptions and actual j market circumstances is the question of the extent to which a model based on static analysis will apply in a ; situation which may be characterized by very dynamic I | effects. I j Two aspects which might produce confounding dynamic I ; effects are innovation and d iffe re n tia l impacts of technological change. Diffusion of innovation among regions may not be simultaneous because of the different rates at which some regions accept changes (Richardson, ; 1969). W hen technological changes occur they tend to have i | differential effects on different industries and thus, | to a certain extent, to different regions. Both of these 1 elements can impart disturbances and upset equilibria. If I I I these events occur frequently enough what one would ; observe in a set of regional markets with minor : imperfections would be most likely not an equilibrium with a particular factor price equilization but rather constant > movement toward (but never actually reaching) equilibrium. Empirical tests of factor price equilibrium on a , regional basis have shown that while the ability of factors to migrate and goods to trade freely do tend to , lessen differentials, some differentials persist. In i stu d ie s focusing on th a t reg io n al a p p lic a tio n , j d i f f e r e n t i a l s have been a ttr ib u te d to several possibilities. Among those possible explanations which have been put forward are price level differentials ! (Coelho and Ghali, 1971), differences in level of ; technology (Batra and Scully, 1972), factor migration , costs (Greenwood, 1975) and state fiscal policies (Canto ‘ and W ebb, 1983). IV.4 R E A L ESTA TE APPLICATION | Having discussed the factor price equalization : theorem in general terms, let us now turn to the matter of 1 its usefulness in investigating regional real estate market efficiency. How should real estate be viewed as a ! constituient part of an economy made up of several regions i ; in which several factors (both mobile and immobile) i | produce goods which are traded throughout all regions? I < i : 74 I First, real estate is a factor of production which ! provides shelter and security for persons and equipment which together form various production functions producing . a variety of goods and services. Additionally, it is not : a pure factor hut rather a combination of land and capital in a wide range of proportions. The land component by its very nature is, of course, immobile. The amount of real estate associated with a given amount of land can be increased or decreased by adding or reducing the amount of captial in combination. The ! capital component, prior to being incorporated is quite mobile. However, once in place on the land, i t is from a practical and physical standpoint virtually immobile. It | is certainly so interregionally and almost always immobile I I intraregionally. To be sure, it is possible to delete or i remove capital which has been incorporated with a piece of * ; real estate but that removal in almost all cases results in a distruction of the capital which is being removed. The other factors of production, labor and non-real ! estate capital, are by comparison very much more mobile j and able to migrate between regions. Within the United 1 States mobility of labor is not impeded by any legal i constraints. However, other limitations may cause labor j migration to be less than perfectly free. When one is | considering the question of migrating, not only the wage I ! i i 75 ' differential between the two regions but also the cost of travel and household relocation must be weighed. Further, such difficult to quantify factors as the breaking of many socio-cultural ties, general environmental conditions (i.e., amenities and climate), and even the accuracy of information and uncertainties about distant labor markets | will have a bearing on the labor migration decision 1 (Emerson and Lamphear, 1975). As with labor, cap ital faces no absolute legal j limitation on mobility among regions in the United States j but as mentioned earlier i t could be less than perfectly | mobile because of institutional and other constraints. Transportation costs would of course impede the migration of any physical capital. And again, uncertainty and lack t i of information flow between regional markets may be the ; biggest culprit. To the extent the characteristics of the i investment opportunities are not known (or are less well I known) capital migration from other regions could be hindered. To the extent capital is funneled to investment opportunities through institutional intermediaries located in financial centers i t might not be incorrect to expect I that access to that factor will be in part a function of ; an investment opportunity's distance from that center (Richardson, 1969). | The goods and services provided by the various , combinations of real estate and other factors of production in the numerous production functions have all manner of degrees of mobility. The physical goods which I are produced face no t a r i f f s but w ill encounter inevitable, and sometimes large, transportation costs. O n the other hand, depending on their nature, services may or I may not face significant cost differences in being sold in j i a region other than the one in which its factors of ! t production are located. j i i Given the preceding, it appears reasonable to propose | I ' viewing real estate in the context of the factor price ; : equalization theorem. In that context, real estate could i i j be said to f it reasonably well in the role of a factor of , | I j production which has severely constrained mobility i i l ' t I between regions. Labor and capital would fit the role of I i i ! factors with a high, although im perfect, degree of | I mobility between regions. The goods and services produced ! by those factors would constitute commodities with high j ; interregional free trade characteristics. It should be | expected then, that despite its interregionally immobile 1 nature, through the functioning of the factor price | equalization theorem real estate could well be priced to produce equal returns throughout all regions. i ! 77 I (— - r " — * ' ■ “ [ I Given that real estate can reasonably be viewed in that theoretical construct, what data then would be required for an empirical analysis of the extent of its i fit? Basically, the information required would need to be real estate pricing data, over time and in various . separate regional markets. As was mentioned in earlier j i chapters, real estate is not traded in a centralized j 1 market. Data on individual trades are scarce at worst and ; extraordinarly decentralized and inconveniently organized ; I at best. This stymies any effort to apply pricing models ! J I . . . ! based on agglomeration of data on individual trades into ' I complete portfolios of all assets. j A body of data however which has potential is that I i ! j which has been collected by The Building Owners and ; i I j Managers Association (B O M A ) beginning in 1920. Briefly, j i these data consist of average rental and expense figures j : for different classes of real property averaged for the I United States as a whole and also broken down into ' individual city and regional averages. The data are ’ expressed in dollars per rentable square foot of the 1 I i i i several property types. If the return to real estate 1 assets over time can be expressed or measured by the ! j i annual percent change in its income (rents) then the B O M A ; 1 i I data could be used to compare the returns in any j I ; particular region or city with that of the United States | ! i i 78 ; as a whole. If the factor price equalization theorem 1 drives prices of immobile factors, such as real estate, to the point that returns are equalized across markets then an analysis should reveal that individual region's or i . city's returns are very highly correlated with average returns for the country as a whole. This is precisely the analysis undertaken in this research. ! Regional real estate market segmentation is most j often characterized in terms of rate of return to I i investors. However, as will be seen in the next chapter, the formulation of a market segmentation test in the | context of the factor price equalization model will be in j terms of change in returns (rent per unit of real estate). ! j The lack of sale price data in addition to rental data 1 precludes the rate of return approach. Results which : support factor price equalization and its underlying assumptions such as factor mobility and interregional trade would be consistent with a market which also equates ! rates of return. i i j IV. 5 S U M M A R Y Through a series of theoretical analyses it has been shown th at even in the case of immobile factors of 1 production, factor prices will be equalized across markets if there is a sufficient level of free trade in goods among those m arkets. This is the fa c to r p rice equalization theorem. Empirical analyses have tested this and, for the most part, have confirmed the tendencies implied by the model. Those results however fall short of exactly replicating the theorem's ideal. Real estate can be adapted to this model and generally fits the role of an immobile factor of production. With that background the possibility of testing for regional real estate market segmentation is raised. As with any empirical analysis in real estate, collecting relevant data is a challenge to say the least. In this instance the data available through B O M A were selected to construct a model and empirically test for the presence or absence of factor price equalization and thus have an element of evidence which comments on regional market segmentation. 80 C H A PTER V TH E M O D E L V.l IN TR O D U C TIO N The question before us is one of whether the commercial real estate market in the United States is geographically segmented or integrated on the basis of rate of return. If not perfectly integrated, some indication of the degree to which i t is would be of in te re st. A model w ill be proposed to te s t for differences based on an available body of data for real estate in geographically separate markets. V.2 BASIC RELATIONSHIPS A testable construct can be developed if we firs t step back to some rudimentary relationships and then procede forward on the basis of the assumptions in the preceding sections. W e can begin with the following pricing relationship P S = a A PA + a B PB + a C PC (V-l) where Pg denotes the price of a good or service S, and P^, Pg, and P q denote the prices for factors of production A , B, and C respectively. The coefficients a^, aB, and a q ; are based on the production function by which factors A , B and C produce output S. Thus, we have the price of a good or service expressed as a function of the prices of the i factor inputs present in its production function. For any individual market one can determine the price ; of the products produced therein if the prevailing factor prices and coefficients present in equation (V-l) are known. Likewise if one knows that basic relationship, the i | price of the good, and the prices of two of the three i | factors then the price of the third factor can be determined. If different markets operate in complete isolation i t is not unreasonable to expect that across : markets one would find d iffe re n t values for the I coefficients and different prices for the factors and I i goods produced. With the introduction of trade in goods and services and the mobility of factors of production one is able to ! make further assumptions about the behavior of prices. If I | the trade in goods and services between markets is unrestrained (i.e., costless) and all factors are able to ; freely migrate between markets then the several markets do ; in fact act like one competitive market. Trade and i i j migration will occur up the point that no arbitrage I opportunities exist. Through this operation of the factor 82 j equalization theorem one would find equalization of prices and returns for all goods and factors across a ll markets even if some number of factors (within a theoretical limit) are fixed or immobile. Thus, with an efficiency among markets spawned by ; free trade in goods and imperfect mobility on the part of j | factors of production, the price for any particular good in all markets would be the same and the rents paid to each category of factors in all markets would be equal, i This pre-supposes that along with free trade and factor I mobility technological levels in the different markets , equilibrate to bring about identical production functions. i i | V.3 N ET R E T U R N S I i Equality of gross prices across markets however does not imply th at net prices for a ll factors w ill be equalized. In the arbitrage process by which gross prices equalize, mobile consumers of factor services survey the ; markets and select and employ factors on the basis that ! they are available at no lower gross prices. If any mobile factor found its e lf in a market in which the ( i 1 expenses to operate were higher than in any other market ' , ... xt would have an xncentxve and, by defxnxtion, the abxlity j to move to another market. A choice which a factor in I i [ such a position would not have would be the raising of its j i ; price. In order then not to suffer a! lower net price I : received than could be obtained elsewhere, the mobile factor would move to the market with lower operating j expenses. There it would receive the same gross price as i 1 in its previous location and enjoy a higher net price. Through an arbitrage-like process it is apparent that net ; prices as well as gross prices paid to mobile factors will i | equilibrate across all markets. I 1 In the case of immobile factors no such equality of net prices received needs to prevail. To be sure, just as I | with any factor, the gross price paid for immobile factor ! services will be equal across a ll markets. But if, for ' instance, the costs of operating in any particular : market were to rise causing a concurrent fall in net I I income the immobile factor would be powerless to recoup 1 the loss. It has no market power to effect a compensating i t rise in gross price received and i t can not flee to another market. The cost can not be passed on nor can i t ! be escaped through mobility. The immobile factor will i I ■ bear the full burden. The lo c a lly applied tax is the most e a s ily ] ; appreciated example of the different capacities of mobile , and immobile factors to shed or avoid the burdens of ' locally imposed cost differentials. Suppose firs t that i across all markets no taxes on the returns to factors of 84 production exist. If one market then imposes a tax thus ■ making operation in that market relatively less rewarding (i.e., reduced net income) than operating in the other markets the mobile factors will migrate to those other : i locations. But, the immobile factor has no such option. | It can not migrate and i t can not raise the gross price i t ' i receives which is set at an equal level across a ll | I markets. The locally imposed tax will be fully absorbed ; i by the immobile factor through a reduction in its net : price received. ! It can not escape notice that in the United States j there are both federal taxes which apply equally in the markets and a variety of local taxes. This must be I taken into consideration when attempting to view local | taxes as a cost or burden which may be inescapable for certain factors of production. Inasmuch as federal taxes apply equally across all markets and if the possibilities 1 of international trade and mobility are put aside, those federal taxes will not be able to be avoided by any factor, mobile or immobile. At the same time the possibility exists that to the extent the demand function for those factors is not perfectly elastic a portion of that burden can be passed on to the purchaser of the j factor services. i With local taxes a similar view must be taken in j respect to the average burden which exists across all • markets. Each local jurisdiction has autonomy to set what it feels is the optimal tax rate. Those rates differ | between jurisdictions. However, just as a uniform federal ' , tax rate can be viewed as an inescapable levy, so i t is j ; appropriate to view the average local tax rate prevailing ; ( i ! across all markets as the zero benchmark from which to ; i measure any relative impact which may be attributable to those taxes. If an immobile factor is located in a market ! which increases its local tax rate to a level above the | J prevailing average across all markets that immobile factor i | has no recourse but to "stand s till and be robbed." j I By the same token that a mobile factor can not be i expected to tolerate a windfall loss through locally j imposed costs which could be avoided by migration. It is also the case that mobile factors will not be able to enjoy a windfall gain based on some local condition. It ' • is the obverse of the process by which burdens are shed I that mobile factor migration will eliminate windfall gains i through arbitrage between markets. j V.4 C O N STR A IN ED M OBILITY Up to this point in this chapter two models of pricing have been discussed. One involves the pricing of 86 I factors and goods in separate markets with no intermarket j trade in goods or factor migration. In that model price differentials for factors and goods across markets is not , surprising. The other model is one in which intermarket i ; trade in goods and factor migration bring about efficiency through the operation of the factor price equalization theorem. In that model's most pure form the prices i j observed for different classes of goods and factors are i j uniform across all markets. I These two different views of price behavior are rather extreme given the seemingly real, although I ! imperfect, degree of intermarket trade and factor mobility I | which one observes in most cases. The question arises at i ■ this point as to whether or not market efficiency can prevail if mobility of factors or trade in goods is constrained but not entirely blocked. An ancillary question is whether observed differences in the price i levels across markets are evidence enough to reject the hypothesis that the markets are efficient. If differences in price levels are not accompanied by interm arket differences in the production of the goods then the . opportunity for arbitrage profits exists. If however the I I price differentials just offset the cost differentials then an efficient equilibrium can exist. 87 ■ The impediment most often cited which would s t i l l ; permit this is transportation costs. These expenses involved in moving goods and services from the market in which their factors of production are located to another market where they are sold may re su lt in price differentials. While this lack of free flow violates one I ; of the assumptions on which the development of the factor i f ; 1 price equalization theorem was based it is possible for an : : efficient equilibrium to evolve in which the differences ! i in prices observed are exactly equal to the costs of ; transporting goods or relocating mobile factors. The resulting equilibrium would not be the same in the face of transportation costs as i t would without, j Prices for a good in a market which is a net exporter of ! that good would be lower than in the absence of ; transportation costs. In a market in which a particular I : i j good was a net import the price for that good would be : higher than in the absence of transportation costs. O n ; the whole, with transportation costs less intermarket f ! trade will take place than in the case of an equilibrium j j which evolves from zero transportation costs. j j If we ignore transportation costs for the moment and j ! return to the question of developing a model and 1 I i I ; hypothesis for testing real estate market segmentation on i t i geographic grounds we can procede as follows. In the , i 88 j ; context of the factor price equalization theorem it can be j assumed, with respect to regional markets in the United States, that real estate represents an immobile factor of ? production, that both labor and capital are free to move ■ ! between markets to seek the highest returns, and that all ; I the classes of goods produced are freely traded across all i markets. Under such circumstances one is assured that ■ through arbitrage the prices of the goods within the ! i | different classes will be equal in all markets and further I that the prices of the mobile factors, labor and capital, will equilibrate across markets. But, i t should also follow through the operation of the factor price j : j equalization theorem that the prices of the immobile | t ; factor, real estate, will equilibrate across all markets. { j ■ In other words, even for the immobile (real estate) , 1 component of the production process, prices paid will move i j to a level such that in no region will it be possible to ; I earn a rate of return higher than the average rate of ; I return for all regions combined. j The specification of equation (V-l) could be altered 1 to i I i i i i i L PS ~ 01LPL + aK PK + aREPR E (V- 2 ) 8 9 J ! This will accommodate our model's explicit dealing with j ! only three factors of production: mobile labor (subscript ' ! L) and capital (subscript K), and immobile real estate (subscript RE). With common technology the values for the coefficients a L, a K , and would be equal across | ; all markets. Through factor price equalization the values , i I 1 i . for Pg, PL, PK, and PR E would be equal across all markets ; i also. | I 1 I I V.5 IN CO RPO RA TIO N O F TR A N SPO R TA TIO N C O STS ! i The idea that a unit of downtown office real estate I I rents for the same price across all markets in the United , States is patently incorrect. The explanation for this I ; could be one of market inefficiency. O n the other hand it j f I could be a m anifestation of an e ffic ie n t market i I ! incorporating the cost difference in producing and selling * , a service in the same market compared to producing i t in one market and selling it in another. For the case in which the differential is a market clearing difference of ■ the latter type, a simple two market example will follow. [ ; Equation (V-2) can be altered to produce I ; i i i I ! I J I - ! I PX " *yx ~ aLPL + a K PK + aREPR E (V-3) i y : where px denotes the price of the service in market X and : tyX denotes the cost of transporting the service from : ; market Y where i t is produced to market X where i t is ' sold. It will be assumed that t for any good or service , , is propdrtional to the distance from the market in which I i t is produced to any market in which i t is sold. Also, j since capital and labor were assumed to be completely free ■ to move between markets, PE and PR are the same across j ! all markets. PR E denotes the gross real estate rent in , Y i j market Y . j I W e now introduce an equation representing the factor j | i ' pricing if the good is both produced and sold in market.: | PX ~ aLPL + aK PK + aREPR E (V-4) : x ' j where PL and PR are again present at their market wide values and PR E denotes the gross real estate rent in x market X . At this point equations (V-3) and (V-4) can be , combined to produce the following relationship. PR E - PR E - - - tV -5) i Y X “ r e ! I i I Thus, it has been demonstrated that with costly movement , of goods and services the equilibrium returns to immobile I ; factors will differ across markets as a function of those transportation costs. As long as aR E 9reater than zero the rental rates in markets distant from where the ; goods and services are consumed w ill decline as transportation costs increase. As labor and capital are added to the lis t of factors which are costly to move the returns to those factors will i also diverge across markets. Finally, when all factors t and goods are faced with transportation costs which are prohibitive (an operational definition of immobility) the I markets will have become completely segmented. Without the requisite flow of goods and factors, prices of goods : and returns to factors across markets will no longer i i necessarily be related. I i : V . 6 EQUILIBRIUM A N D L O C A L FISCAL POLICIES I An important contribution to the more complete ; understanding of market equilibrium with factor mobility ; has been made by Canto and Webb (1981). They have f ! developed a model which demonstrates the equilibrium relationship between the percent change in the production ' of market goods in the United States and the expenditure and tax rate p o licies of both federal and sta te governments. The essence of their model is that state and f i | local fiscal policy, in addition to that of the federal I I 92 1 __ ; government/ is of great significance in the determination : of national market equilibrium. Their development turns on th& key insight that in an integrated economy the equilibrium condition can be viewed : in more than one context. One is from the point of view I of the equilibrium within an individual market. This , would involve the allocation of goods and factors of i j production within that market and may also involve the j trade in goods and migration of factors among markets. | i The other involves an overall equilibrium which aggregates the supply and demand of goods and factors of production i at a national level. i For example, if in any market in an integrated national economy there is an in itia l imbalance between supply and demand for a good or factor of production there will be pressure for a relative price change. But, if . goods are traded freely and some factors of production are 1 mobile, the opportunity for arbitrage will be seized. If the market is small compared to the national economy, through this arbitrage the deviation of its relative prices from the national level would be eliminated. Thus, i • relative excesses of supply and demand between markets i j would also be eliminated. Such intermarket trading and I J mxgratxon however wxll not necessarily brxng about a full I ! equilibrium which eliminates nation-wide excess supply or | demand. After any relative excesses among local markets ! are eliminated one may s till find nation-wide excesses in supply or demand. If so, there w ill need to be an : adjustment in relative prices across all markets to bring ! about the full equilibrium. With prices then being set on the basis of an equilibrium which extends across all 1 markets, the changes which occur will have proportionate I | effects in each. I I Thus, events which occur in an individual market are 1 ! j integral elements of a comprehensxve equilibrium i I specification. As suggested above, an example could be ! the establishment of one or another alternative fiscal ■ policies by a local government. Through its spending and I tax programs a local government can have a significant ■ impact on the level of aggregate demand and the after-tax 1 return to immobile factors of production. To the extent i ' ■ all fiscal policy affects aggregate demand, it will affect ■ equilibrium conditions and cause real estate rent levels to move proportionately across all markets. If fiscal ; policy enhances the effective supply (e.g. utilization) of a factor in a particular market, i t will have a local I effect which should increase the return to the factor in ! that market. One point which must be kept in mind is that j a local government's fiscal policies are local in nature ; only to the extent they differ from the average of all I I 9 4 I other local governments' policies. In summary, i t is possible to determine national [ 1 relative price levels from the summation of the individual t supply and demand relationships for goods and factors in individual markets. However, as Canto and Webb (1981) ; I I explain relative prices derived in that fashion are only a • necessary but not a sufficient condition to assure a full 1 j equilibrium. For equilibrium to be present in individual ‘ markets it is necessary that quantity adjustments in goods ! and mobile factors be possible between markets. Through i ■ fiscal policy, local governments have significant power to influence such adjustments. ! ; i ; i : r ! V.7 RELATIONSHIPS FO R EM PIRICAL A N A LY SIS ! j ; Several relationships have been established to this j I i point. First, i t was shown that through factor price . equalization changes in real estate rent levels in ! I different markets will be proportionate. Second, through : the Canto and Webb model i t can be infered th at a I I relationship is implicit between changes in the returns to ; I i j real estate and changes in local government policies j ' I i toward taxes and expenditures. These relationships I ] i provide the framework by which the interregional ! ; I 1 integration of the commercial real estate market are | I examined in this research. I This framework takes the form of a model with both local and national equilibrium conditions. The firs t deals with the national equilibrium involving changes in gross returns to real estate in different markets. The second deals with the local equilibrium of net returns to real estate in individual markets. Those equilibria are specified as follows: (V—6) (V— 7) in which the variables Vl^ to V9^ are defined as: Annual percent change in average square foot rent in city or region i V 2 Annual percent change in average square foot rent in the United States. Annual percent change in average square foot net income in city or region i . V4 Annual percent change in average square foot net income in the United States Annual change in difference between average square foot property tax in city or region i and in the United States. 96 V 6 i Annual changes in difference between average state and local non-property tax rate in city i and in the United States. V7± Annual change in difference between per ca p ita s ta te and lo cal government expenditures (net of capital expenditures and transfer payments) in city i and the United States average. V 8 i Annual change in difference between per capita state and local government capital expenditures in city i and the United States average. V9± Differences between annual percent change in per capita income in city i and in the United States. Data lim itations which are described in the data section compelled the paring down of equations (V-6 ) and (V-7) to the forms given below. Equations (V-6 ) and (V-7) will be used for the time period 1959-1980. Equations (V— 6 ') and (V-7') will cover the time period 1948 to 1980. vli = 3 q + P 2 V 2 + 35v5i (V-6 ') V3i = 3 o + 34 V 4 + (V-7') 97 V. 8 EX PEC TED V A L U E S Having specified the model it is appropriate at this point to review the implications of each of the terms in equations (V-6 ), (V-6 1), (V-7), and (V-71). Intuitive insights which should be of assistance in interpreting the results of the empirical analysis will be included. Also, specification of the expected values of the coefficients under an hypothesis of fully integrated markets will be given. The value for the constant term would be expected to be zero if the various cities and regions are functioning as an integrated economy. To the extent there are adjustment costs the value could be positive or negative. Therefore, our hypothesis with integrated markets will be A g rea t deal of d isc u ssio n has taken place surrounding the relatio n sh ip between the local and national average gross returns to immobile factors. This will be addressed through equations (V-6 ) and (V-6 ‘) which have the annual percent change in gross rent in city i as their dependent variable. It should be clear that under factor price equalization the expected value of the coefficient for the annual percent change in the national average gross rent should be unity. Hence our hypothesis for that variable under nationally integrated markets 9 8 will be H q:B2 = 1 * 0 * O n the other hand, if markets are operating in an environment of no intermarket trade or factor migration and if shifts in local aggregate demand are uncorrelated between markets then the expected coefficients for the annual percent change in the national average gross rent would be zero. It has been explained previously that under factor price equalization gross returns to factors will be the same across a ll markets. However, because of certain expenses which are peculiar to different localities, net returns to immobile factors need not be the same. In fact, net returns to the immobile factors should differ across markets by the full amount of those burdens which are stric tly local (but not national) in nature. The deviation of a city's property tax from the national average has already been cited as such a burden. Another category of expenses which can differ widely across markets are those attributed to local environmental conditions. Two examples of such conditions are climate and the latitude at which a particular market lies. Climatic differences between markets have an enormous impact on the re la tiv e expenditures for energy, u t i l i t i e s , and maintenance. In more severe climates, costs for heating, ventilating and air conditioning far excede those in more 99 moderate areas. Also, along with relatively severe weather conditions one usually finds a greater need for maintenance. The latitude at which a property is located can also have an effect on its operating costs relative to other areas. As properties are located in more northern latitudes the expenses incurred for increased lighting during winter months are not necessarily off set by reduced requirements in summer months. The longer daylight hours during summer months may come when no lighting is required. Thus, the winter deficit would not be compensated. These expenses are a substantial component of real estate's operating expenses. Thus, with at least local taxes, u tilitie s , energy, and maintenance having "local” components i t is not d iffic u lt to argue that net incomes at a local level at not necessarily correlated across markets. This relationship is captured in equations (V-7) and (V-71) through the utilization of the percent change in net income in a particular city as the dependent variable and the percent change in the national average net income, variable V4, as one of the independent variables. The hypothesis for its coefficient with integrated markets will be H q: 34 = 0. Variable V5, captures the deviation between a particular city's property tax rates and the national 100 average of local property taxes. In equations (V-6 ) and (V-6 * ) this variable is regressed against gross rent levels in individual cities or regions. This should give some indication of the extent to which a stric tly local burden can be shifted. If real estate is the only immobile factor and if consumers of real estate services are mobile then a shift of the burden by means of a rent level increase should not be possible. This would call for an hypothesis of H Q: 3 5 = 0. However, if consumers of real estate are not entirely free to move to other markets then it would be possible to shift the burden through an increase in the gross rent. In that event variable V5's coefficient would be positive. In equations (V-7) and (V-71) the opportunity is present for variable V 5 to demonstrate a local tax impact on returns to an immobile factor which are net of taxes and other expenses. If the burden is absorbed the net income will fall as local taxes rise. That effect of local taxes on immobile factors in integrated markets calls for an hypothesis of H Q:3g<0. The less the burden is borne, the less negative would be the coefficient. If the coefficient is zero the suggestion is that a shift of the burden has left after-tax income unchanged. It would be wise here to remember that property taxes are not the only strictly local burdens. Nor are they the 101 main burden of all operating expenses. Other components of those expenses are also moving in a fashion uncorrelated with counterparts in other markets. Therefore, net incomes are fluctuating on the basis of factors which may dominate the impact of property tax effects. In that case the property tax effects may be quite difficult to reliably d istill in empirical analysis. This could present a significant obstacle to determining the incidence of the taxes. Another opportunity to detect the incidence of local taxes is present with the annual change in the differences between a city's local level of non-property taxes and the national average of all local non-property taxes, variable V 6 . Recall that a differential of a local tax above the national average of local taxes should not be able to be levied except on consumers or factors which are immobile or of limited mobility. But with integrated markets the gross returns paid to factors and the prices of the goods consumed are set at the national level. Thus, in the case of equation (V-6 ) which deals with gross rent changes, this variable would yield an integrated market hypothesis of H q: 3 6 = 0 . Just as with property taxes, when dealing with net returns to immobile factors one would expect other local taxes to be fully absorbed and result in a reduction of 102 net returns that net. To the extent the taxes represented by variable V 6 are incedent on real estate they are not likely to be shifted. Therefore, in equation (V-7) this would call for a hypothesis in which H Q: 3g<0. Variable V 7 captures the effect of local government's expenditures as they deviate from the average expenditures of all local governments combined. Expenditures on capital goods and transfer payments have been excluded. This prim arily measures education, police, fire protection, sanitation, administration and recreation expenditure differentials. A n essential issue with this variable is the extent to which i t affects aggregate demand in the city under examination. As aggregate demand in a city rises or falls, corresponding pressure on relative prices could be expected. However, if the city is small relative to the rest of the cities combined and if the city markets are integrated, trade in goods and factor migration will occur and quash any movement away from a nationally imposed regime of relative prices. Thus, gross and net returns to real estate would be uncorrelated with respect to this variable. As a result, for both equations (V-6 ) and (V-7) the hypothesis will be H Q: 3y = 0. This matter of aggregate demand being affected by expenditures begs the question of private sector valuation 103 of government, services. Suppose, at the margin, the private-sector valued the services provided at an amount equal to the government's cost to provide them. If this were the case then there would be a neutral effect on aggregate demand. Again, no correlation with gross or net returns to real estate could be inferred. Variable V 8 measures the change in the difference between a particular city's capital expenditures and the national average of all cities. The preceding discussion covering non-capital government expenditures would also apply to this variable. One additional comment should be made however. The nature of this class of goods (e.g., heavily weighted toward transportation facilities) raises the possibility that there could be the opportunity for strictly local effects which raise the utilization rate of immobile factors such as real estate. If that were the case the returns could rise. In the absence of any specific information on the likelihood of such an effect, this variable will carry a hypothesis of H 0: = 0 . By measuring the difference between a city and the national average percent change in per capita income, variable V 9 approximates a measure of relative change in local overall economic activity. This is intended to be a surrogate for relative change in aggregate demand. If local aggregate demand rises one would expect to observe 104 upward pressure on relative prices of the immobile factor. The part a well integrated market would play at this point can not be overlooked. As with the government expenditures variables, if trade in goods and services is well established and if some factors are mobile, the effect on the individual market by the other markets could overwhelm the lo cally induced excess demand. No persistent condition perm itting arbitrage would be tolerated by the integrated market. Once again, for equations (V 6 ) and (V7) the hypothesis will be H Q:gg = 0 . A summary of the expected values for the coefficients of these independent variables under conditions of fully integrated markets and mobile tenants is presented in Table V-l. 105 TABLE V-l EX PEC TED V A L U E S O F VARIABLES' COEFFICIENTS Coefficient for Variable V 2 V 4 V 5 (equations V - 6 and V - 6 ' ) V 5 (equations V-7 and V-7 ' ) V 6 (equations V - 6 and V-6 ' ) V 6 (equations V-7 and V-7 ' ) V 7 V 8 V 9 106 Expected Value 1.0 0 0 <0 0 <0 0 0 0 C H A PTER VI D A T A A N D A N A LY TIC PR O C E D U R E S VI.1 D A T A SO U R C E S The data which will be used to conduct the analysis have come from three sources. The data on downtown office property income and expenses have been obtained from the Building Owners and Managers Association, the data on population and local government finances have been obtained from the U.S. Department of Commerce, Bureau of the Census, and the Consumer Price Index data have been obtained from the Department of Labor, Bureau of Labor Statistics. As has been mentioned, the compilation of data for the purpose of em pirical analysis of real estate investment is quite difficult. To a large extent, this is because of the lack of a central market place or exchange where transactions would take place and be recorded. Fortunately, for decades, B O M A has been compiling, in a consistent and system atic way, extensive data on a particular catagory of investment real estate in the United States. The Building Owners and Managers Association was organized in 1908 with the objective of 107 providing mutual benefit to members by the exchange of information on building designs, construction methods, products, marketing techniques and other matters of mutual interest to owners and operators of investment real estate. It now totals approximately 5,000 members who own or manage nearly 2 billion square feet of properties. Since 1920 B O M A has collected and published income and expense figures for office buildings in various cities through out the United States. This annual report is the recognized industry standard for catagorization of income and expenses for office properties. It is also the standard for the correct sp ecificatio n of property attributes such as gross and net rentable area, etc. Those data are solicted for voluntary submission from B O M A members and others known to own or operate investment real estate. The motivation for submission of data is an incentive on the part of individual members to make the organization work more effectively as a trade group. VI.2 D A T A FO R M A T S The B O M A data are published annually in their Office Building Experience Exchange Report. The data are presented as consolidated average figures for the United States as a whole, by seven regional breakdowns, and by the city within which the reporting properties are 108 located. In 1980 the report was based on information from nearly 1 , 0 0 0 buildings which totaled approximately 250,000,000 square feet of office space. The buildings were located in 55 different cities. The data published are unaudited. Therefore, certain errors may be present which can not be determined. Before implementing the analysis of any of the equations specified in Chapter V, the B O M A data were plotted as a time series of rents and net income for the various cities and regions. N o striking anomolies were apparent. It is d if f ic u lt to imagine su ffic ie n t motivation on any individual contributor's part to submit intentionally falsified data. If errors are present they are most likely to be unintentional. In addition, if any exist and are consistent from year to year then the validity of the time series analysis may not be seriously degraded. For each city the report states the number of buildings included in the sample and the square footage involved. A n average vacancy factor is also given. The average income for all buildings in the sample is given in terms of dollars per square foot. All expenses (except for debt service and taxes on building income) are also given in dollars per square foot. They are broken down into standard catagories thus providing separate figures for items such as maintenance, repairs, utilities, general 109 administration, insurance, and real property taxes. An example of the reporting format is provided in Figure VI- 1 . In 1947, 1974 and 1981 modifications were made to the reporting format and catagories. The 1974 modifications were minor and through small manipulations it is possible to make the data for the 1947 to 1973 time frame completely compatible with those of the 1974 to 1980 time frame. It is not practical to manipulate either the pre- 1947 or post-1980 data to achieve compatibility with that of the 1947 to 1980 period, Therefore, because of the need to maintain consistency in the data to be used in this analysis, the time period will extend from 1947 through 1980. VI.3 CITY A N D R EG IO N A L D ESIG N A TIO N S Throughout the years these data have been published, cities have been added or deleted. All of the cities for which information was reported from 1947 through 1980 were tabulated and included in the analysis. Cities which few observations or no report for a number of years were set aside. A total of 29 cities were selected to form the sample of geographically segmented markets. Of those 30 cities, 15 had full reporting for the entire 34 year time frame and none had less than 28 years. Since the data 110 were also available on a regional basis those were also obtained for a separate analysis using only the regional and national figures. A listing of the cities and regions which will be used along with their number codes is provided in Table VI-1. A map diagraming the lim its of the regional divisions is provided in Figure VI-2. Data needed to provide values for variables V 6 , V7, V 8 , and V9 are available on a basis which is compatable (geographically and temporally) with the B O M A data for only part of the 1947 to 1980 time frame. Prior to 1959 the Bureau of the Census does not have published figures for s ta te s and lo c a l governmental revenues and expenditures on an annual basis. Those data are not available elsewhere in a compatible form. It will be necessary therefore to limit the use of variables V 6 , V7, V 8 , and V 9 to analyses beginning with 1959. VI.4 PO O LIN G O F D A T A A C R O SS CITIES Based on the models, hypotheses, and data presented this analysis will deal with four equations, two each over two different time frames. The procedures used for each equation will be nearly identical. They will differ only in that we will implement regional in addition to city breakdowns when dealing with equations (V-6 1) and (V- 7'). N o regional based data will be incorporated in 111 equations (V-6 ) and (V-7) which include state and local government expenditure variables. Both time series and cross-section analysis will be undertaken. The firs t step will be a pooled analysis combining a ll cities [and regions with equations (V-6 1) and (V-71)] using an ordinary least squares procedure. Next, an ordinary least squares procedure, with Durbin-Watson statistics requested, will be run on each city [and region with equations (V-6 1) and (V-7')] individually. The question arises at this point as to whether or not i t is legitimate from a sta tistic a l standpoint to combine the data from all those separate cities and regions into one large data base for a single analysis as was described. Analysis of variance enables one to answer this # question through an F-test procedure. This is a test of homogeneity which gives information as to whether or not the relationships suggested by the pooled results are reasonable given the results obtained from the separate (by city or region) analyses. The formula for this test is (S-, + S3 )/K(p-1) F = __ .___ ._________ <vi-n (S4 /(pmc-pK) where (S -^ + S3) denotes the pooled sum of squared errors, 112 k denotes the number of independent variables plus one, p denotes the number of groups in the pool of data, S4 denotes the sum of squared errors of all groups, and pmc d en o te s the t o t a l number of o b s e r v a tio n s . The degrees of freedom equal k(p-l) and pmc - pk. If the caculated F-ratio falls outside the limits prescribed by the particular degrees of freedom, the individual groups may not legitimately be combined for meaningful statistical analysis (Johnston, 1972). VI.5 A U TO C O R R ELA TIO N O F D A T A After the question of pooling has been resolved, attention needs to turn to the question of autocorrelation among the variables in the city or regional groupings. Classical least squares models assume that disturbance terms obtained are not serially correlated. It is quite common however when dealing with economic time series data to encounter sig n ific an t se ria l correlation among consecutive observations of variables. The presence of such an effect renders less valid the coefficients and* other output from the least squares procedure. Durbin- Watson statistics can reveal the presence and magnitude of autocorrelation in a body of data. This involves calculating the following D W statistic based on the error terms, e, generated in each time period. 113 n < et “ et - l >2 t =2 D W (VI-2) n 2 t=l That s ta tistic is then compared with lim its which are calculated based on the sample size and number of independent variables to determine whether the null hypothesis of an autocorrelation, p , equal to zero can or cannot be rejected (Hu, 1982). If evidence of significant autocorrelation is found in any of the individual city or regional ordinary least squares procedures, the data for that city or region can be transformed through a Cochran-Orcutt iterative process. Briefly, this involves utilizing the following formulae to alter the values of the data observations. Where X denotes the raw data observation, TX denotes the transformed value of X, and p denotes the correlation coefficient. for t=l (VI-3) TX t = xt - P xt-l for t = 2 to n (VI-4) 114 That transformed data should then be re-analyzed by an ordinary least squares procedure. The results of that analysis should exhibit less autocorrelation than the process applied to the raw data. A word of caution here is th a t the Co ch r a n - 0 r c u 11 process is only an approximation and may lead to only minor or negligible improvement (Johnston, 1972). VI. 6 TW O -STEP FU LL T R A N SFO R M A N A LY SIS With the transformed data, ordinary least squares analysis will again be carried out both on a pooled basis and by individual city or region. To determine the extent to which the Cochran-Orcutt transformation ameliorated the autocorrelation situation, Durbin-Watson statistics again w ill be obtained for the c ity and regional analyses. As an alternative to the Cochran-Orcutt method of handling autocorrelation, a two-step full transform method will be used. This is equivalent to a generalized least squares process using residuals from the ordinary least squares process to estim ate the covariances across observations (SA S User Guide, 1980). Results are similar in format to a pure ordinary least squares procedure except that the autocorrelation coefficients are given and the values of coefficients of the independent variables 115 have been compensated for autocorrelative effects. This usually yields results superior to the Cochran-Orcutt process. VI.7 JOINT PROBABILITY O F O U T C O M E S In the analytic steps described up to this point separate sets of data from th irty cities are tested to determine if particular values for the several variables can be rejected in those individual cities. In such testing some probability exists that erroneous indications of reject or fail to reject will' occur. By means of a calculation which utilizes the binomial distribution it is possible to calculate the probability that any particular number of erroneous indications of reject or fail to reject will occur. Consider a test as described earlier in which sets of data from thirty different cities are examined using a 90% confidence level. And, assume that the value of the coefficient for which rejection is being tested is in fact the value of that coefficient for all thirty cities. The most likely result which would be obtained would be twenty-seven fail to reject outcomes (90% of the total) and three reject outcomes (10% of the total). The twenty- seven fail to reject outcomes would give indications which correctly indicate the coefficients' values while the 116 three reject outcomes would give erroneous indications. Thus, a certain number of erroneous outcomes given a confidence level of less than 1 0 0 % should not be r surprising. B y calculating the probability of any particular set ; of outcomes giving erroneous indications and having occurred by chance i t is possible to obtain some idea of j the joint significance of the outcomes for any individual j variable across a ll cities. In other words, one may 1 ' determine what the likelihood is that a particular number ‘ of cities will erroneously indicate rejection of some value of the coefficient of a variable under examination. Similarly, one can also determine the likelihood that ! erroneous indications of failure to reject will occur. i Assume n equals the number of observations (cities in th is research), S equals the number of erroneous indications of reject or fail to reject, and q equals one minus the confidence interval. The probability, A . , that . S erroneous indications will occur can be calculated by t j means of the following equation. I I ■ nl ,n-2 s (VI-5) ! X = sT'Cn-2) -! --- (1"c ^) < 3 I An example of the u tiliz a tio n of such probability I i ' calculations would be as follows. Suppose data for a variable from thirty cities were examined for rejection of a hypothetical value using a 90% confidence level. If : results from three out of the thirty cities indicated : rejection the case for rejection in any city would be weak. A calculation of probabilities demonstrates that in ; i such a test, if the hypothetical value is correct for all , : i , cities, the most likely outcome would include three ; erroneous reject indications. On the other hand, if ; ! i results from fourteen of the thirty cities indicated j I rejection the case for rejection in most of those fourteen ' ! would be quite strong. A calculation of the probability of all fourteen rejection indications being erroneous reveals . such an occurrance to be very unlikely. I I ] VI. 8 S U M M A R Y ; I For each of the four equations, each in its own time period, the following steps will be undertaken: | i 1) Ordinary least squares analysis of all data pooled 2) Ordinary least squares analysis of individual i cities or regions 1 i ; 3) F-test to determine pooling legitimacy of 1) j i 4) Check for autocorrelation in 2) and Cochran- j | ! Orcutt transform data if necessary j 5) Ordinary le a s t squares a n a ly sis of a l l transformed data pooled 6 ) Ordinary least squares analysis of individual cities and regions using transformed data 7) Two-step transform analysis of raw data 8 ) Calculation of probability of outcomes utilizing binomial distribution 119 TABLE V I- 1 R EG IO N S A N D CITIES 34 N O R T H C EN TR A L 01 Akron 06 Chicago 07 Cincinnati 08 Cleveland 12 Detroit 14 Indianapolis 35 M ID W E ST N O R T H E R N 11 Des Moines 15 Kansas City 17 Milwaukee 18 Minneapolis 21 Omaha 25 St. Louis W ITH N U M B E R C O D E S 37 PACIFIC SO U T H W E ST 16 Los Angeles 26 San Francisco 38 SO U T H E R N 02 Atlanta 04 Birmingham 05 Carolinas/Virginia 39 SO U T H W E ST 09 Dallas 10 Denver 13 Houston 20 Oklahoma City 29 Tulsa 36 PACIFIC SO U T H W E ST 24 Portland 27 Seattle 28 Spokane 40 M ID D LE A TLA N TIC 03 Baltimore 19 N ew York City 22 Philadelphia 23 Pittsburgh 30 Washington D.C. 120 FIGURE V I - 1 B O M A D A T A R EPO R T F O R M A T 1980 LOS ANGELEs N um ber o f B uildin gs Total R entable S a . Ft. 18 7 , 2 8 2 , 5 4 8 18 6 , 4 1 0 , 3 9 3 i a 6 , 2 2 9 , 4 0 9 ACCOUNT BLOG.TOTAL OFC. TOTAL OFC. RENTED 11 CLZ1MIHG 1 2 E LECTE1C AL STSTEHS 13C COSBIMED 8 TIC 15 E LET ATOE S 16 1 G E B EE ll BO ILEX KG COSTS 16B ADBISISTEATITE COSTS 1? E1ERGT TOTAL OPEHATIBG *1' 1 O n . 5 (18) 1 9 . 7 (17) 3 4 . 1 (18) 2 6 . 4 ( 1 8 ) 7 1. 3 ( 18) 27 . 1(16) 1 5 4 . 3 ( 1 8 ) 4 3 5 . 6 (18) 1 0 9 . 8 ( 1 8 ) 2 1 . 3 ( 1 7 ) 3 4 . 5 (18) 27. 1 ( 18) 70 . 3 (18) 2 7 . 3 ( 1 6 ) 153 . 9 (18) 4 4 2 . 2 (18) 113. 0 2 2 . 0 35. 5 2 7 . 9 7 2 . 4 2 6 . 1 158 . 4 4 5 5 . 1 B1 ILTEBATICBS-TEBAMT 1BE1 B3 DECORATING- TESIST ABEA TOTIL COBSTBOCTICB *8 ' 2 2 . 0 (9) I t . 1 (3) 2 1 . 5 ( 1 0 ) 22. 2 (9) 2 6 . 9 (3) 24. 0 (10) 2 3 . 4 2 7 . 4 25. 3 TOTAL OPEB 1TIMG E I P . 1 AID B 4 1 5 . 7 ( 1 0 ) ’ 4 3 0 . 7 ( 1 0 ) 4 5 3 . 2 C1 ISSOBANCE C2 A BEAL ESTATE T AI C2B PEBSCI1L P B 0 P . T 1 I ETC. # TOTAL I I I E D CHABGBS ’ C* 1 2 . 6 (16) 7 3 . 5( 18) 2 - 2 ( 1 3 ) 8 7 . 0 ( 1 8 ) 1 2 . 5 (16) 7 4 . 5( 18) 2. 1 (13) 8 7 . 8 ( 1 8 ) 12. 9 7 6 . 7 2. 2 9 0 . 3 t TOTIL CPEBATIBG EIP.A*B*C 5 0 4 . 3 (10) 5 2 0 . 6 (10) 5 4 7 . 7 « BET IICOBE - G U I - LOSS 6 1 1 . 1 ( 1 1 ) 0. 0 (0) 7 5 3 . 5 ( 11) 0. 0 (0) 7 7 7 . 2 0 . 0 LEASE EXPI9SE AB0BT1ZED TEBABT ALTERATIONS * t TOTAL OPEB ATIHG EXP.A»B*C 6 . 3 ( 9 ) 5 1 . 0 (7) 5 8 4 . 0 (7) 6 . 4 ( 9 ) 48. 9 ( 7 ) 601 .7 (7) 6 . 7 4 9 . S 6 5 7 . 3 * • BET ZBCCBE - GAIB - LOSS 4 5 0 . 6 (9) 0. 0 (0) 6 0 0 . 3 (9) 0. 0 (0) 6 2 5. e 0. 0 BEBTAL IBCOBE - OEEICE ABE1 - STOBE A SEA - SZCEAGE ABEA - SFECIAL ABEA TOTAL BEBTAL IBCCSI ELECTBICAL IBCOSE BISCELLABZCOS IBCOBE TOTAL 0PEBAT1IG IBCOSE 1 0 0 9 . 9 (9) 3 0 7 . 7 ( 8 ) 2 4 3 . 1 (5) 1 0 3 7 . 7 (10) 7 - 7 (4) 8 7 . 4 ( 10) 1 1 2 0 . 8 ( 1 1 ) 1 0 9 6 . 5 ( 1 0 ) 1 1 3 2 . 0 LABCS COST OPEBATIBG 1ATX0 % aiBAGESEBt BATIO * ATG OEEICE TACABCT % ATC OEEICE OCCOPABCT 1 AVG OEEICE TEBABT SQ. ET. ATC STOBE TEBABT SC. ET. ABC. s o . ET. PEB EEPLOTEE AtG SO. PT. PEB PEBSO* 4 7 . 7 (12) 4 8 . 9 (1 1) 3 7 . 6 (11) 2 . 8 (18) 9 7 . 2 ( 1 8 ) 224 9 9 . 7 (17) 3 0 6 6 . 9 (9) 4 6 2 0 5 . 5 (1 1) 2 5 7 . 4 (14) 2 3 5 . 7 ( 14) 2 2 8 . 7 121 FIGURE VI-2 M A P O F B O M A R EG IO N S PACIFIC NORTHWEST /M ID D L E ATLANTIC! MIDWEST NORTHERN NORTH CENTRAL PACIFIC SOUTHWEST SOUTHERN SOUTHWEST MIDDLE ATLANTIC— S ta te s of Maine, Vermont. New H am pshire. M assach u setts. C o n n ecticu t, R hode Island. New York. P ennsylvania. M aryland. D elaw are. New Je rse y , District of C olum bia. MIDWEST NORTHERN— S ta te s of North D akota. South D akota, N eb rask a. K ansas, M issouri. Iowa. M innesota. W isconsin. NORTH CENTRAL— S ta te s of M ichigan. Illinois. Indiana, Ohio PACIFIC NORTHWEST— S la te s of W ashington, O regon. M ontana, Idaho. W yoming. A laska. PACIFIC SOUTHWEST— S ta te s of California, N evada, Utah. Arizona. Hawaii SOUTHERN— S tates of Kentucky, W est Virginia, Virginia. T e n n e sse e . North C arolina, South C arolina. G eorgia. A labam a. M ississippi. Louisiana. Florida. SOUTHW EST— S tates of A rkansas. O klahom a. T exas. New Mexico. C olorado. 122 C H A PTER VII RESU LTS VII.1 PRELIM IN A RY D A T A A N A LY SIS As explained in Chapter VI, a preliminary analysis of the data is necessary to determine the appropriateness of pooling data across cities and to compensate for any autocorrelation which may be present. The results of those steps will be presented first. VII.1.1 Pooling of City Data Ordinary least squares analysis of equations (V-6 ), (V-6 '), (V-7), and (V-7') was carried out on an individual city or regional basis and also with all of the cities' data pooled. To determine the legitimacy of such a pooling of the individual city data into one body of data an analysis of variance through an examination of the homogeneity of coefficients across all the equations was conducted by means of the earlier described F-test. With each of the four equations the F-statistic significantly exceeded the c ritic a l value of F with the applicable degrees of freedom at a 5% level. Consequently, homogeneity of coefficients across all cities must be rejected. Therefore, the pooling of all the individual city data is inappropriate. All analysis will have to be 123 performed for each city separately. The results of the F- tests are provided in Table VII-I. VII.1.2 Autocorrelation The Durbin-Watson s ta tistic s calculated for each individual city and region were then compared with the critical upper and lower limits determined by each city's particular sample size to resolve whether the assumption of a lack of autocorrelation among error terms must be rejected. Significant autocorrelation was found to be present in from 14 to 17 cities and in 4 of 5 of the > regions depending on the equation. To eliminate the adverse effects which the presence of autocorrelation has on the validity of the coefficients calculated in the analysis the data was transformed using the Cochran-Orcutt iterative process. The outcome of the F-tests using the transformed data again indicated that pooling of the data would be inappropriate. Up to th is point, the re su lts of the analysis procedure outlined in the previous section have established the following: a) the data from individual cities may not be meaningfully combined into a single pool for analysis, and b) significant autocorrelation is present in the raw data of approximately half the cities. 124_I VII.1.3 Two-Step Full Transform Compensation Being limited at this point to proceeding with time series analysis on the basis of individual city and regional' data, an alternative procedure for compensating for the presence of autocorrelation was used. This procedure known as a two-step full transform method was implemented by means of the statistical analysis program package which was used for all of the analyses. Output does not consist of a body of transformed data. Rather it consists of the autocorrelation coefficients which were found and the coefficients of the independent variables with compensation for the autocorrelation effects. Empirical conclusions drawn in this study are limited to the interpretation of the results of this two-step full transform procedure. Unless otherwise noted, a confidence level of 90% to determine statistical significance is used throughout. VII.2 G R O SS R EN TA L EQUILIBRIUM The change in local gross rental income per square foot (Vl^) as a function of variables V 2 and V 5 through V 9 was analyzed to determine support for the hypothesis of regionally integrated markets with mobile tenants. Two equations and two time periods were involved. Tables VII- 2, VII-3, and VII-4 present the numerical results of this 125 analysis. Table VII-5 presents a summary showing for each variable the number of cities and regions which rejected or failed to reject that hypothesis. VII.2.1 The Constant Term Consistent across the analyses is the outcome of the constant term. In every city and region the value of this term was not significantly different from zero. This is consistent with the hypothesis of fully integrated markets with mobile tenants. If a constant term had been found for which a value of zero could be rejected i t could indicate an adjustment process underway in that particular market (city or region). Or, i t could suggest different levels of gross rental income change over the long term. VII.2.2 1959 to 1980? Equation (V-6 ) Data on twenty-nine cities were available for this portion of the study. The results for the six dependent variables in this equation will be discussed in turn, relative to their support for the hypothesis of regionally integrated markets with mobile tenants. 126 VII.2.2.1 Change in Average National Gross Rental Income (V2) For this variable a fully integrated market with mobile tenants calls for a coefficient of unity. Under complete market segmentation the coefficient would be zero. The results for the various cities were as follows. A . Different from zero and not different from unity: P h ilad e lp h ia , P ittsb u rg h , Chicago, Cleveland, Minneapolis, St. Louis, Dallas, Houston, Portland, Spokane For these cities the hypothesis is not rejected. B. Not different from zero or unity: Baltimore, New York City, Washington D.C., Akron, C in c in a tti, D e tro it, Indianapolis, Des Moines, Kansas City, Omaha, Atlanta, Birmingham, Denver, Tulsa, Seattle, Los Angeles, San Francisco For these cities the results are inconclusive with respect to support for the hypothesis. C. Different from unity: Oklahoma City, Milwaukee For these cities the hypothesis is rejected. 127 Thus, out of twenty-nine cities, ten failed to reject the hypothesis of regionality integrated markets with mobile tenants, two rejected the hypothesis, and the remaining seventeen were inconclusive. Both of the cities drawn from the Pacific Southwest region, Los Angeles and San Francisco, had inconclusive results. Other than that geographic association, no pattern was apparent in the incidence of c ity 's co efficien ts rejectin g or failin g to re je c t the hypothesis. Of the two cities whose results reject a value of unity for their coefficients, one is positive and one is negative. Through the process described in Chapter VI the probability of two erroneous rejections occurring by chance can be calculated. In th is case th at probability is 0.24 and is, in fact, the most likely outcome. Therefore, two rejections occurring should not be taken as a serious argument against the hypothesis. Further, the probability that the ten failures to reject — 7 are erroneous is only 7.4 x 10 . V II.2.2.2 Change in Local Property Tax Differential (V5) For this variable, the hypothesis of regionally intergrated markets with mobile tenants would call for a coefficient of zero. O n the other hand, if tenants are 128 immobile a sharing of taxes would produce a positive coefficient. The results were as follows. Out of twenty-nine cities, nineteen had coefficients which were not significantly different from zero. Of the remaining ten, eight were positive and two were negative. Those ten cities for which the data reject the hypothesis are listed below. A. Positive: Philadelphia, Washington D.C., Cleveland, Minneapolis, Denver, Seattle, Spokane, Los Angeles B. Negative: Indianapolis, Atlanta Other than being a random occurrence, an intuitive explanation for any city having a negative coefficient does not come to mind. The positive outcomes could be interpreted as indications of a local tax burden shift from the property to an immobile tenant. With the possible exception of Spokane, all those cities with positive coefficients are large industrial or service oriented cities. They are geographically dispersed and have the full range of environmental conditions. The same, however, could be said of the cities which had coefficients which were not significantly different from zero. W hy that particular section of eight cities might 129 have less mobile tenants, a greater need for expensive public services, or a propensity to provide other amenities is not obvious. A puzzling aspect of the significantly positive coefficients is the magnitude of the implied tax burden shift. In those cases a one dollar change in the property tax was associated with a two to three dollar change in gross rents. The s t a ti s t i c a l significance of th is calculated shift is unknown. If the magnitude of that s h if t is s ig n if ic a n t then the p o s s i b il i ty of a misspecification of the variable is present. In summary, nineteen cities produced coefficients which support the hyothesis of an integrated market and mobile tenants. Ten c itie s do not support that hypothesis. The probability that ten cities would erro neously reject a value of zero by chance is 2.7 x 10”^. VII.2.2.3 Change in Local Non-Property Tax Differential (V 6 ) As with the local property tax, this burden will be incident upon immobile factors of production and consumers. To the extent tenants are mobile, local non property taxes will be of no consequence in determining the gross rents which they w ill be w illing to pay. Therefore, an hypothesis of tenant mobility within an 130 integrated market will call for a coefficient of zero. Twenty-five of tTie twenty-nine cities produced coefficients which were not significantly different from zero. In those cases the hypothesis of integrated markets with mobile tenants was not rejected. Of the four cities which had coefficients which were significantly different from zero, two were positive and two were negative. They were as follows. A . Positive: Atlanta, Portland B. Negative: Cincinnati, Los Angeles Only one of those four cities had local property tax coefficient different from zero. It is Los Angeles and the property tax coefficient was of the opposite sign. The probability th at four c itie s would erroneously indicate rejection of a zero coefficient purely by chance is 0.17. This number of rejections is only one greater than the number of erroneous rejections that is most likely to occur. VII.2.2.4 Change in Local Non-Capital Expenditures Differential (V7) Highly integrated markets in which gross rent levels are established on nation-wide criteria would be expected 131 to toe insensitive to changes in local arrgregate demand where the gross rent level is concerned. Under that hypothesis a coefficient of zero would be called for. If markets are segmented, the sign of this coefficient could indicate the extent to which the private sector values the goods and services being provided toy the government. Twenty-four of the cities produced coefficients which were not significantly different from zero. That outcome' is consistent with an hypothesis of integrated markets with trade in goods and services, and factor migration. Of the five c itie s not supporting the hypothesis which had coefficients significantly different from zero, two were positive and one was negative. They were as follows. A. Positive: Atlanta, Portland B. Negative: Washington D.C., St. Louis, Los Angeles The probability that five results would erroneously indicate rejection of the hypothesis purely by chance is — 9 . . . . 9.5 x 10 . While this is a small probability the number of reject results involved is only two more than the number which would be most likely to be observed. 132 VII.2.2.5 Change in Local Expenditure Differential (V 8 ) This variable would be expected to operate in the same fashion as the non-capital expenditure variable. Thus, with regionally integrated markets and mobile tenants a coefficient of zero would be expected. This expectation was fulfilled for nearly all of the cities. Specifically, twenty-seven cities produced coefficients which were not significantly different from zero. These support an integrated market hypothesis. Both of the cities which had coefficients significantly different from zero produced negative values. They were St. Louis and Atlanta. The probability that the two cities would produce results which erroneously reject the hypothesis by chance is 0.24. Such an outcome of two erroneous rejections is in fact the most likely which would occur. This is hardly a strong argument against the hypothesis. VII.2.2. 6 Change in Local Per Capita Income Differential (V9) In fully integrated markets with trade in goods and factor m obility th is surrogate for change in local aggregate demand should have no correlation with gross ______________133 rental levels. O n the other hand, if markets are segmented a positive relationship between this variable and gross rental levels would be expected. Twenty-six cities produced coefficients which were not sig n ific an tly different from zero. These support the hypothesis of an integrated market with mobile tenants. Of the three which had coefficients which were significantly different from zero, one was positive and two were negative. They were as follows. A. Positive: Washington D.C. B. Negative: Chicago, Cincinnati Just as with the p robability of two erroneous rejections, the probability of three erroneous rejections occurring is also 0.24. Again, this is the number of erroneous rejections which is most likely to occur and should not be considered to be significant evidence against counter to the hypothesis. VII.2.2.7 1959 to 1980 City Summary To the extent conclusive results were produced with respect to the average national gross rent level, the preponderance of those outcomes supported the hypothesis of an integrated market with mobile tenants. Only two 134 cities rejected the hypothesis while ten conclusively supported i t . In the case of the property tax variable, nineteen cities supported the integrated market hypothesis with mobile tenants. At the same time, however, eight cities appeared to suggest a degree of tenant immobility through the indication of a local property tax shift. In alm ost a l l cases, the outcomes for the coefficients of the variables measuring local fiscal policy and per cap ital income fa il to re je c t the hypothesis of market integration with mobile tenants. The number of outcomes which indicated rejection were not far from what could be attributed to chance. Several cities such as Washington D.C., St. Louis, Atlanta, and Los Angeles appeared to reject the hypothesis on several occasions. A clear pattern based on region was not apparent. Neither was there a suggestion of a logical relationship based on the signs of the coefficients for any individual city which, on more than one occasion, had values that were significantly different from zero. VII.2.3 1948 to 1980; Equation (V-6 '); Cities During this longer time frame data on th irty cities were available. The re su lts for the two dependent variables in this equation will be discussed relative to 135 their support for the hypothesis of regionally integrated markets with mobile tenants. V II.2.3.1 Change in Average National Gross Rental Income (V2) As with the shorter time period, the integrated market hypothesis would call for a coefficient of unity for this variable. Under complete market segmentation the coefficient would be zero. The results for the various cities were as follows. A . Different from zero and not different from unity: Philadelphia, Pittsburgh, Akron, Chicago, Cleveland, Indianapolis, Milwaukee, Minneapolis, Omaha, Atlanta, Dallas, Denver, Houston, Seattle, Spokane For these cities the hypothesis is not rejected. B. Not different from zero or unity: Baltimore, New York City, Washington D.C., Cincinnati, Des Moines, Kansas City, St. Louis, Tulsa, Portland For these cities the results are inconclusive with respect to support for the hypothesis. C. Different from unit: Detroit, Birmingham, Carolinas/Virginia, 136 Oklahoma City, Los Angeles For these cities the hypothesis is rejected. Thus, out of th irty cities, fifteen support the integrated market hypothesis, five reject the hypothesis and the remaining ten are inconclusive. There is a high correlation between the results of this time frame and the shorter one. All but one of those c itie s which conclusively supported the integrated market hypothesis did so again in this time period. The one exception was St. Louis which moved to the inconclusive group. Also, as was the case earlier, there does not seem to be any pattern of regional representation in or between those three categories of outcomes. The probability that those five outcomes which reject unity are erroneous and would — 9 appear by chance xs 1.2 x 10 . The probabxlxty that the fifteen outcomes which reject zero are erroneous and would appear by chance is 1.7 x 10-1^. Therefore, i t is in the realm of possibility tht the five rejections are erroneous indications. The fifteen fail to reject outcomes are extremely unlikely to be erroneous indications which have occurred by chance. VII2.3.2 Change in Local Property Tax Differential (V5) Once again, the expected value for this variable's 137 coefficient would be zero in a fully integrated market with mobile tenants. With immobile tenants the coefficient would be positive to the extent the local tax burden is shared. Out of the th irty c itie s , th irte e n produced coefficients which were not significantly different from zero. These support the hypothesis of an integrated market with mobile tenants. Of the remaining seventeen, fourteen were positive and three were negative. They were as follows. A. Positive: New York City, Philadelphia, Washington D.C., Akron, Chicago, Cleveland, Minneapolis, Carolinas/virginia, Dallas, Denver, Tulsa, S eattle, Spokane, Los Angeles B. Negative: Indianapolis, Atlanta, Birmingham It should be noted th at each city which had a coefficient significantly different from zero in the shorter time frame again appeared on that list. One sign of a regional pattern of results is present here. The cities in the North Central region seem to be inclined to demonstrate a tax burden shift. Of the total of fourteen cities which demonstrate an apparent tax burden shift 1 38 nearly a l l again dem onstrated a m u ltip lic a tiv e relationship between property tax change and gross rental income change. The probability that seventeen cities would erroneously reject a coefficient value of zero purely by — 1 0 chance is 3.0 x 12 . Thus, if this variable is properly specified we have a significant indication of a lack of tenant mobility. VII.2.3.3 1948 to 1980 City Summary As the time frame was lengthened, the results in the area of the average national gross rent level became more conclusive. The number of cities which supported the hypothesis of market integration with mobile tenants rose to fifteen. At the same time the number which rejected the hypothesis also rose (to five). The number of cities with inconclusive results fell from seventeen to ten. The number of cities which indicate that property tax burdens are shifted rose from eight to fourteen as the time frame was extended. O n the other hand, ‘thirteen cities s t i l l supported the hypothesis of integrated markets with mobile tenants. Cities which supported intergrated markets in the shorter time period and cities which indicated a tax burden shift did so again in the extended time frame. L ittle in the way of any pattern in regional outcomes 139 was manifest. VII.2.4 1948 to 1980; Equation (V-6 1); Regions Data from seven regions were utilized for this time frame. The results for the two dependent variables in this equation and their implications for the hypothesis of integrated markets with mobile tenants will be discussed. VII.2.4.1 Change in Average National Gross Rental Income (V2) When applied to the regions, just as with the city analysis, the hypothesis calls for a value of unity for th is v ariab le's co efficien t. With complete market segmentation a coefficient of zero would be expected. The results for the regions were as follows. A . Different from zero but not different from unity: Middle Atlantic, North Central, Midwest, Northern, Southwest, Pacific Northwest These regions fail to reject the hypothesis of integrated markets with tenant mobility. B. Different from zero and unity: Pacific Southwest This region rejects both the integrated and fully segmented hypotheses. 140 C. Different from unity and not different from zero: Southern This region rejects the integrated market hypothesis but supports complete segmentation. Thus, on a regional basis there appears to be less than full integration of markets with tenant mobility. Two of the three cities in our sample which are located in the Southern region might have suggested that region's outcome. Only Los Angeles in the Pacific Southwest region would suggest that region's outcome. The probability that two regions would produce erroneous reject outcomes is 0.04. It would be wise then to attach some significance to the rejection of the hypothesis by those regions. V II.2.4.2 Change in Local Property Tax Differential (V5) The hypothesis of integrated markets and mobile tenants calls for a value of zero for the coefficient of this variable. With immobile tenants a positive value would be expected. The results were as follows. A . Not different from zero: Middle A tlantic, Southern, Southwest, Pacific Southwest. These regions do not reject the hypothesis. 141 B. Different from zero and positive: North Central, Midwest Northern, Pacific Northwest These regions reject the hypothesis. In both the Pacific Northwest and North Central regions there was a distinct correlation between the observed outcomes of the cities and that of the overall region. The probability that three out of the seven regions would produce, purely by chance, erroneous results — O that reject the hypothesis is 3.6 x 10 . This is quite small and suggests the rejections of the hypothesis by these regions are quite unlikely to be erroneous indications which have occurred by chance. VII.2.5 Gross Rental Equilibrium Summary W hen the results of all the local gross rental income relationships are taken together several generalizations can be made. First, in very few instances in either time frame can the hypothesis of integrated markets with mobile tenants be rejected at the city level on the basis of local and national gross rent relationships. A number of the outcomes are inconclusive. That number falls however as the time frame is extended. At the regional level this research found that two regions rejected the integrated market hypothesis. One of those failed to reject complete 142 segmentation. Thus at the regional level it appears that a less convincing case can be made for market integration and tenant mobility. In the case of the property tax variable there were indications that in some cities tenants appear to be mobile while in others they bear part of the property tax burden as if immobile. In the longer time frame more cities exhibit a tax shift. The extent to which the burden is shared is not clear. The integrated market hypothesis could not be rejected with a reasonable degree of confidence on the basis of the outcomes of the local non-property tax or expenditure variables. The number of rejections of an integrated market hypothesis which were observed could easily be attributed to chance. Finally, the lack of a relationship between the relative local per capita income change and local gross rent levels suggests that local excesses of demand are probably being overwhelmed by influences from the national market. This is a strong indication of integration across markets. VII.3 N ET R EN TA L EQUILIBRIUM The change in local net rental income per square foot (V3^) as a function of variables V4 through V9 was 143 analyzed to determine support for the hypothesis of regionally integrated markets with mobile tenants. Two equations and two time periods were involved. Tabls VIII- 6 , VII-7, and VII- 8 present the numerical results of this analysis. Table VII-9 presents a summary showing for each variable the number of cities or regions which rejected or failed to reject that hypothesis. VII.3.1 The Constant Term Just as with the re su lts of the gross rental equilibrium, in no instance in this portion of the study did the constant term ever yield a result for which a value of zero could be rejected. This is consistent with the hypothesis of regionally integrated markets with mobile tenants. It is inconsistent with a condition of regional markets experiencing separate adjustments or long term differential rates of change in net rental income. VII.3.2 1959 to 1980; Equation (V-7) Data from twenty-nine cities were utilized for this time frame and equation. The re su lts for the six dependent variables and their indicated level of support for the hypothesis of regionally integrated markets with mobile tenants are presented in the following sections. 144 VII.3.2.1 Change in Average National Net Rental Income (V) For this variable the hypothesis would call for a coefficient of zero. This would be attributed to the effects of there being a number of operating expenses which are stric tly local in nature and not correlated between markets. Those uncorrelated variations would introduce a bias toward zero. The results for the cities were as follows. Out of the twenty-nine c itie s , twenty-one had coefficients which were not significantly different from zero, thus supporting the hypothesis of integrated markets. Of the eight which had coefficient significantly different from zero, six were positive and two were negative. The cities in those categories were as follows. A. Positive: Chicago, Dalis, Des Moines, Houston, Minneapolis, Pittsburgh B. Negative: Baltimore, Washington D.C. An implication for those cities with significantly positive coefficients could be the passing on to immobile tenants some share of the stric tly local variations in operating expenses. Only three of those cities (Chicago, Minneapolis, and Dallas) previously showed such a capacity 145 to shift property tax burdens in the gross rental income equations. The probability that eight cities would produce erroneous indications rejecting the hypothesis purely on O the basis of chance is only 4.7 x 10 . Thus, whxle there can be l i t t l e doubt that the majority of the fail to reject results are true indications, i t is also probably the case that the eight rejections reflect some true cases that are inconsistent with the hypothesis. VII.3.2.2 Change in Local Property Tax Differential (V5) For th is variable an hypothesis of regionally integrated markets with mobile tenants would call for a coefficient with a negative value. This would reflect the immobile property, not the tenant, absorbing that strictly local burden. Operating expenses other than local property taxes also vary at the local level. It is possible that the effects of those, if of great enough magnitude, could mask the effects of this variable. Out of twenty-nine c itie s , tw enty-three had coefficients which were not significantly different from less than zero. These cities, therefore, fail to reject the hypothesis. The six cities which did reject the hypothesis were the following. 146 Cleveland, Indianapolis, Des Moines, Omaha, Oklahoma City, Spokane For those cities which suggest the presence of a property tax burden shift there is no apparent pattern. Only three of the six demonstrated a shift in the gross rent equations. The probability that six cities would produce erroneous indications rejecting the hypothesis — 9 purely on the basis of chance xs only 4.2 x 10 . Therefore, while the twenty-three fail to reject results lend strong support to the hypothesis, i t is likely that several of the reject results are true indications of an inconsistency with regionally integrated markets with mobile tenants. VII.3.2.3 Change in Local Non-Property Tax Differential (V 6 ) Just as with the property tax, this local burden will be incident upon immobile factors of production and consumers. To the extent tenants are mobile they will not bear the burden of the stric tly local aspects of these taxes. Thus, the integrated market hypothesis with mobile tenants would call for a coefficient less than or equal to zero. Twenty-four of the twenty-nine cities in the sample have coefficients which do not reject values less than or 147 equal to zero. Those lend support to the integrated market hypothesis with mobile tenants. The five cities which do not support that hypothesis were the following. Washington D.C., Des Moines, St. Louis, Atlanta, Portland Of these five cities only Des Moines also rejected the hypothesis in the case of the property tax variable. N o other pattern in these rejections is apparent. The probability that five erroneous reject indications would _ O result is 9.5 x 10 . While the twenty-four fail to reject results lend strong support to the hypothesis, i t is also likely that of the five cities that reject i t several reflect true inconsistencies with regionally integrated markets with mobile tenants. VII.3.2.4 Change in Local Non-Capital Expenditure Differential (V7) Just as is the case with the gross rental income equations, this variable should be uncorrelated with the local net income if markets are regionally integrated and tenants are mobile. Hence, the coefficient called for would be equal to zero. Twenty-five cities did in fact produce coefficients which did not reject a value of zero. They thereby lent support to the integrated market hypothesis. Those cities 148 which had coefficients which were significantly different from zero included one which was positive and three which were negative. Those cities were as follows. A. Positive: Cleveland B. Negative: Washington D.C., Chicago, San Francisco This is the second time Washington D.C. has appeared to have a negative c o e ffic ien t for the non-capital expenditure variable. The other instance was of course in the gross rental equation. An implication of two such outcomes could be that local government expenditures are depressing aggregate demand in the city. This could also be the case for San Francisco which appears for the second time as well. The probability that four cities would produce erroneous rejections of the hypothesis is 0.17. This number of rejections is only one greater than the three erroneous rejections which is the most likely outcome which would be expected to be observed. This incidence of four rejections should not be taken as a strong indication contrary to the hypothesis of regionally integrated markets with mobile tenants. 149 VII.3.2.5 Change in Local C a p ita l Expenditure Differential (V 8 ) This variable would be expected to operate in the same fashion as the non-capital expenditure variable. Thus, under the hypothesis of regionally integrated markets with mobile tenants a coefficient of zero would be expected. Twenty-six cities produced coefficients which did not reject a value of zero. They thereby lent support to the hypothesis. All three of the cities whose coefficients rejected a value of zero had outcomes which were negative. Those cities were the following: Chicago, St. Louis, Atlanta Both St. Louis and Atlanta previously showed negative coefficients for capital expenditures. This could be an indication of local demand being adversely affected by suboptimal government expenditures on capital goods. In addition Chicago now has demonstrated a negative coefficient for both capital and non-capital goods. The suggestion here could be th at the local government services are being provided at a cost in excess of their value to the private sector. The probability that three erroneous reject outcomes would occur is 0.24. This is the very number of erroneous reject outcomes which should be expected to be observed 150 if, in fact, the true values of all the coefficients conform to the hypothesis. Thus, these few instances of cities which seem to reject the hypothesis of regionally integrated markets with mobile tenants should not be considered serious arguments against the hypothesis. VII.3.2.6 Change in Local Per Capita Income Differential (V9) Once again, as with the gross rental income equation, the hypothesis of regionally integrated market with mobile tenants would call for a coefficient of zero for this variable. To the extent trade in goods and migration of some factors takes places, local changes in aggregate demand would be overwhelmed. Twenty-six cities produced coefficients which did not reject a value of zero. This lends strong support to the hypothesis. Of the three cities which had coefficients which did reject a value of zero, two were positive and one was negative. Those cities were the following. A. Positive: Washington D.C., Oklahoma City B. Negative: Chicago Washington D .C. has again repeated a result from the gross rental income equation. A n interpretation of this 151 could be that with the rise in gross rents which accompanies a change in local aggregate demand, the net income also rises. Chicago on the other hand has repeated its opposite indication. Just as with the three outcomes described in the previous section (VII.3.2.5) which indicated a rejection of the hypothesis, these three rejectio n re su lts here should not be considered significant evidence to refute the existence of regionally integrated markets with mobile tenants. VII.3.2.7 1959 to 1980 Summary For the most part, the results of this equation and time frame demonstrate support for the hypothesis of regionally integrated markets with mobile tenants. In the case of the national average net rental income and property tax variables however, there are indications that in some cities a sharing of local burdens between the property and tenants is taking place. This suggests a degree of immobility on the part of tenants or constraints on trade in goods and services in some cities. The outcomes for the variables representing the non property taxes (both c la ssific a tio n s of government expenditures) and per capita income revealed rates of rejection of the integrated market hypothesis which were hardly different from the number of erroneous rejections 152 which might be expected purely on the basis of chance. VII.3.3 1948 to 1980; Equation (V-7'); Cities During this longer time frame data on thirty cities were available. The re su lts for the two dependent variables in this equation will be discussed in turn relative to their support for the hypothesis of regionally integrated markets with mobile tenants. VII.3.3.1 Changes in Average National Net Rental Income (V4) Just as with the shorter time frame, the hypothesis calls for a coefficient of zero for this variable. The results produced by the thirty cities were quite similar to those of the earlier time frame. Twenty-two cities had coefficients which did not reject a value of zero. Those cities support the integrated market hypothesis. Of the eight cities which produced coefficients which did reject a value of zero, six were positive and two were negative. Those cities were the following. A. Positive: P ittsb u rg h , Chicago, Des Moines, Minneapolis, Dallas, Houston B. Negative Baltimore, Cincinnati 153 Except for Cincinnati having replaced Washington D.C., this is the identical group of cities which was generated in the shorter time frame. And once again, the im plication is that for these c itie s with positive coefficients some sharing of local burdens by tenants may be taking place. The probability that eight cities would produce coefficients significantly different from zero purely by chance is 5.8 x 1 0 - ^. Thus, i t is rather unlikely that all eight of these cities outcomes are erroneous rejections of the hypothesis which occurred by chance. V II.3.3.2 Change in Local Property Tax Differential (V5) A coefficient with a negative value for this variable is called for by the hypothesis of regionally integrated markets with mobile tenants. If a positive value were found the implication would be that a sharing of the property tax burden was taking place. Twenty-two cities produced coefficients which did not reject a value of zero. Thus, those cities lend support to the hypothesis. The eight cities which had coefficients which rejected a value of zero were the following. New York City, Cleveland, Indianapolis, Des Moines, Omaha, Carolinas/Virginia, Denver, Spokane 154 This is the same list of exceptions from the shorter time frame with the addition of New York City and Carolinas/Virginia. Thus, the implied shift of property tax burdens seems to have expanded to two additional locations. This is not a large proportional decrease in the number of cities which fail to reject the hypothesis of regionally integrated markets with mobile tenants. The probability that eight cities would indicate erroneous rejections of the hypothesis purely by chance is only 5.8 x 10"^. Therefore, some of those outcomes are quite likely to reflect a true refutation of the hypothesis. VII.3.3.3 1948 to 1980 City Summary Compared to the results of the shorter time frame, these findings only slightly increase the indication that in some cities there is a sharing of local burdens between the immobile property and tenants. In each l i s t of exceptions to the hypothesis of regionally integrated markets with mobile tenants most of the cities which were present in the shorter time frame again appeared. The incidence of rejection of the hypothesis was only very slightly increased. Cases which failed to reject the hypothesis were still the great majority of the outcomes. 155 VII.3.4 1948 to 1980; Equation (V-7'); Regions Data from seven regions were utilized for this time frame. The results for the two dependent variables in this equation and their implications for the hypothesis of regionally integrated markets with mobile tenants will be discussed. VII.3.4.1 Average National Net Rental Income (V 4 ) Once again, the hypothesis calls for a coefficient of zero for this variable. In a departure from the results obtained in the city analysis, a majority of the regions produced coefficients which rejected a value of zero. Each was positive. The four regions which made up that group were the following. Middle Atlantic, Midwest Northern, Southwest, Pacific Northwest Thus only three out of the seven regions had results which were consistent with the hypothesis of regionally integrated markets with mobile tenants. This outcome might not be inexplicable. Consider that the zero coefficient hypotheis was based on many cities having various local and uncorrelated operating expenses. As cities are aggregated into regions the effect may be to moderate the variance of the changes in the new regionally 156 aggregated net income. If this occurs as many cities are aggregated into a few regions the significant correlation observed between the regions and the national average would not be so unexpected. At some level of aggregation of cities the correlation with the national average will approach a value of unity. The probability that those four results indicating a rejection of the hypothesis would occur purely by chance is 1.9 x 10“^. Therefore, those outcomes are very unlikely to entirely erroneous indications. VII.3.4.2 Change in Local Property Tax Differential (V5) As before, according to the hypothesis, the expected value of the coefficient for this variable is less than zero. The results produced six regions with coefficients for which a negative value was not rejected. Thus, there would be no indication of a property tax shift in those regions. The one region which proved to be an exception was the North Central. The probability that one out of seven regions would produce an erroneous indication rejecting the hypothesis of regionally integrated markets with mobile tenants is 0.37. That one case of inconsistency with the hypothesis should not cause great concern. 157 VII.3.5 Net Rental Equilibrium Summary Talcing these analyses based on net rental income as a whole, several pertinent observations can be made. First, based on the cities' analysis of the local and national net rental income relationships there is little evidence which would suggest that the net incomes are highly correlated between markets. O n the contrary they seem for the most part to be uncorrelated. This could be the result of the several local operating cost components which are uncorrelated between cities acting on a gross rental level which is established by the national market. This is consistent with regionally integrated markets with mobile tenants. As was explained earlier, the aggregation of data to the regional level may have the effect of substantially reducing the effective variance between the regions' net rental income and the national averages. This may explain the apparent lack of support for the hypothesis at the regional level. Second, the property tax variable results were consistent with the hypothesis of regionally integrated markets with mobile tenants for the majority of the cities. Further if other than a zero correlation were to be supported, i t could be attributable to nationally correlated components of income and expenses having an 158 overwhelming effect. Third, as was the case with the gross rental income equations, the variables capturing the effects of local fiscal policy and local aggregate demand rarely failed to support the hypothesis. The number of outcomes which failed to provide support was very close to the number of erroneous indications of rejection which would be expected to occur purely by chance. Conversely, the likelihood that the high number of consistent results occurred by chance is negligible. 159 D A T A T A B L E VII-1 PO O LIN G F-TEST R ESU LTS Equation D F F Critical < l c * 196,363 2. 90 1.34 V - 6 ' 84,479 7.09 1.39 V-7 196,363 5. 31 1.34 l > 84,479 8.84 1.39 160 TABLE VII-2 EQUATION (V-6) RESULTS C ity / legion C onstant V2 V5 V6 V7 1 -0.00922 0.725* 0.000273 0.00152 -0.00144 (0.0298) (1.019) (0.000721) (0.00259) 0.00279 2 -0.000849 0.292* -0.000448* 0.00345* -0.00485* (0.0129) (0.530) ( 0 . 000?15) (0.00197) (0.00169) 3 0.00493 0.860* -0.000161 0.00217 0.00131 (0.0371) (1.811) (0.000159) (0.00526) (0.00407) 4 -0.00617 -0.944* -0.00136 0.00180 -0.00118 (0.0336) (1.260) (0.000816) (0.00577) (0.00447) 6 0.000754 1.321 0.000103 -0.000981 -0.00113 (0.00697) (0.363) (0.000102) (0.00104) (0.000785) 7 0.0571 -0.434* 0.000182 0.0138* 0.00619 (0.0732) (3.019) (0.000795) (0.00827) (0.00928) 8 0.0238 1.780 0.00105* 0.00225 -0.00156 (0.0206) (0.751) (0.000323) (0.00195) (0.00300) 9 0.0119 2.950 0.000489 0.000481 0.000922 (0.0402) (1.537) (0.000305) (0.00540) (0.00434) 10 0.0196 0.963* 0.00100* -0.00131 -0.00143 (0.0154) (0.948) (0.000336) (0.00106) (0.000947) • I n d i c a t e s i n c o n s i s t e n t w ith h y p o th esis of r e g i o n a l l y i n t e g r a t e d m arkets with mobile te n a n t s V8 V9 R2 P t -0.00886 -0.214 0.53 -0.0687 (0.00883) (0.429) (-0 .2 4 ) 0.0123* 1.426 0.55 0.400 (0.00507) (0.925) (2.04) 0.120 0.876 0.20 0.247 (0.0181) (2.648) (1.19) -0.000441 -0.557 0.41 -0.115 (0.00172) (0.738) (-0 .4 8 ) -0.00280 -1.506* 0.60 0.371 (0.00283) (0.711) (1.87) -0.0516 -2.730* 0.58 -0.118 (0.0297) (1.557) (-0 .4 3 ) -0.00391 0.0649 0.62 -0.0103 (0.00915) (0.472) (-0 .0 4 ) 0.00329 0.756 0.31 0.368 (0.00586) (2.615) (1.86) -0.00137 -0.242 0.46 0.466 (0.00690) (0.912) (2.47) SE 0.0141 0.0486 0.535 0.0974 0.0234 0.176 0.0954 0.390 0.116 162 C i t y , Reg ion Constant V2 V5 TABLE VII- V6 11 0.00834 0.415* 0.000254 0.0000314 (0.0319) (1.590) (0.000344) (0.00210) 12 -0.0142 0.272* -0.000295 0.00188 (0.0210) (1.289) (0.000331) (0.00341) 13 0.00759 1.734 0.000364 0.000727 (0.0200) (0.739) (0.000329) (0.00307) 14 -0.0106 0.564* -0.000611* 0.000236 (0.0213) (0.834) (0.000236) (0.00226) 15 -0.00416 0.375* 0.000639 -0.00382 (0.0287 ) (1.093) (0.000499) (0.00363) 16 0.0413 0.231* 0.000524* 0.00137* (0.0496) (1.790) (0.000239) (0.00340) 17 0.0543 4.740* 0.000184 -0.000360 (0.0375) (2.100) (0.000497) (0.00367) 18 0.00542 1.182 0.000243* 0.00124 (0.00787) (0.423) (0.000114 ) (0.000951) 19 0.0106 1.0623* 0.000193 0.000326 (0.0219) (0.860) (0.000143) (0.00144) 20 -0.0142 -2.400* 0.00074 4 0.000669 (0.0483) (1.512) (0.00113) (0.00559) ! (co n tin u atio n ) V7 V8 V9 R2 !> t -0.00256 0.0211 -0.767 0. 20 0.284 (0.00252) (0.0169) (0.925) (1.39) 0.00359 -0.00982 0.795 0.21 0. 358 (0.00412) (0.00931) 10.981) (1.79) -0.00120 0.000151 -0.238 0.34 0. 450 (0.00269) (0.00277) (1.336) (2.36) -0.00134 -0.00549 1.357 0.42 0. 054 (0.00285) (0.00592) (0.977) (0.25) -0.00261 0.0121 0.596 0.21 -0.033 (0.00315) (0.0122) (1.598) (-0.15) -0.000725 -0.0192 1.225 0.78 -0.201 (0.00805) (0.0109) (4.544) (-0.71) -0.00314 -0.0129 -1.680) 0.38 0.374 (0.00598) (0.0122) (2.309) (1.89) 0.00112 0.000728 0.287 0.42 0.485 (0.00109) (0.00267) (0.461) (2.60) -0.000269 -0.00809 0.175 0. 37 0.103 (0.00157) (0.00896) (1.077) (0.48) 0.000976 -0.0125 0.615 0.54 -0.0483 (0.00215) (0.0202) (2.636) (-0.18) SE 0.298 0.167 0.0971 0.122 0.159 0.0465 0.586 0.0299 0.122 0.0771 163 TABLE VII : i t y / legion Constant V2 V5 V6 21 0.00682 0.902* -0.0000516 -0.0000735 (0.0421) (1.747) (0.000361) (0.00369) 22 -0.000504 0.909 0.000635* -0.00152 (0.00956) (0.449) (0.000141) (0.00132) 23 -0.00822 1.602 -0.0000928 -0.00172 (0.0119) (0.419) (0.000225) (0.00146) 24 0.0384 2.551 -0.000292* -0.00299 (0.0242) (1.303) (0.000454) (0.00218) 25 -0.00589 0.776 0.000341 -0.00633 (0.00676) (0.425) (0.000289) (0.00294) 26 0.00872 0.431* 0.000204 -0.00161 (0.0142) (0.486) (0.000167) (0.000526) 27 0.0114 0.694* 0.000526* -0.000737 (0.0182) (0.754) (0.000289) (0.00125) 28 0.0129 1.644 0.00196* 0.00363 (0.0204) (0.909) (0.000550) (0.00193) 29 0.0290 1.223* -0.000178 0.00584 (0.113) (1.964) (0.000892) (0.0135) 30 0.0307 0.284* 0.000664* -0.000701 (0.0200) (0.558) (0.000372) (0.000412) I i 2 ( c o n tin u a tio n ) j I ! V7 V8 V9 R2 t SE 0.00131 0.00327 -0.479 0.0635 0.06 0.551 (0.00252) (0.0111) (1.371) (0.29) -0.00174 0.00362 0.239 0.448 0.69 0.0385 (0.00121) (0.00492) (1.501) (2.35) -0.00112 -0.00643 -0.869 -0.0254 0.60 0.0327 (0.00156) (0.00530) (1.108) (-0 .1 1 ) 0.00161 0.0231 0.676 0.316 0.39 0.219 (0.00168) (0.0116) (1.560) (1.56) -0.00648* 0.00167* 0.206 0.475 0.96 0.00112 (0.00137) (0.00268) (0.316) (1.79) 0.000157* -0.000879 0.750 -0.0834 0.49 0.0278 (0.000940) (0.00296) (1.170) (-0 .3 9 ) -0.00150 0.00464 0.325 0.284 0.38 0.110 (0.00137) (0.00773) (0.955) (1.39) 0.00152* -0.00470 -0.834 0.498 0.64 0.162 (0.00169) (0*00866) (0.963) (2.69) -0.00292 0.0264 -3.332 -0.294 0.79 0.0456 (0.00431) (0.0715) (7.002) (-1 .0 2 ) -0.000801* 0.00126 0.740* -0.188 0.60 0.0137 (0.000927) (0.00460) (0.417) (-0 .7 4 ) TABLE VII-3 EQUATION (V-6'); CITIES RESULTS ‘ I n d i c a t e s i n c o n s i s t e n t w i t h h y p o t h e s i s o f r e g i o n a l l y i n t e g r a t e d m a r k e ts w i t h m o b ile t e n a n t s C i t y / Region C o n sta n t V2 V5 R2 0 t SE 1 0 .0 0 1 3 3 ( 0 .0 0 8 3 1 ) 1 .1 9 6 ( 0 . 3 4 5 ) 0 . 0 0 0 8 6 2 * ( 0 .0 0 0 2 7 8 ) 0 . 5 0 0 . 303 ( 1 . 5 6 ) 0 .0 3 4 7 2 - 0 . 0 0 4 4 0 ( 0 . 0 0 8 4 6 ) 0 . 7 2 0 ( 0 . 3 5 9 ) - 0 . 0 0 0 3 6 5 * ( 0 .0 0 0 2 1 2 ) 0 . 1 9 0 . 2 8 8 ( 1 . 7 2 ) 0 . 1 1 0 3 0 .0 0 1 6 7 ( 0 .0 2 8 3 ) 0 . 5 9 4 * ( 1 .3 9 1 ) - 0 . 0 0 0 1 5 0 ( 0 .0 0 0 1 1 8 ) 0 . 1 0 0 . 2 0 6 ( 1 . 0 7 ) 0. 645 4 - 0 . 0 1 0 4 ( 0 .0 1 4 4 ) - 0 . 4 3 4 * ( 0 . 5 6 6 ) - 0 . 0 0 0 8 0 7 * ( 0 .0 0 0 4 0 6 ) 0 . 1 4 0 . 0 6 8 ( 0 . 3 6 ) 0 . 1 4 9 5 0 .0 3 0 9 ( 0 .0 2 6 9 ) - 0 . 5 9 0 4 * ( 0 .8 0 1 ) 0 .0 0 0 9 9 7 * ( 0 .0 0 0 2 1 2 ) 0 . 6 2 - 0 . 3 5 6 ( - 1 . 6 1 ) 0 . 0 6 9 8 6 - 0 . 0 0 1 9 9 ( 0 .0 0 4 5 2 ) 0 . 9 4 8 ( 0 .2 0 1 ) 0 . 0 0 0 2 0 5 * ( 0 .0 0 0 0 7 2 7 ) 0 . 5 1 0 . 4 2 7 ( 2 . 7 1 ) 0 . 0 3 7 7 7 0 .0 0 6 8 9 ( 0 .0 2 7 6 ) 0 .4 1 0 * ( 1 .1 9 8 ) - 0 . 0 0 0 1 4 7 ( 0 .0 0 0 4 3 5 ) 0 . 0 1 0 . 2 0 4 ( 1 . 0 2 ) 0 . 4 0 7 8 0 .0 1 0 0 ( 0 .0 1 1 4 ) 1. 289 ( 0 .4 0 4 ) 0 . 0 0 0 9 1 9 ( 0 .0 0 0 2 2 8 ) 0 . 52 0 . 0 5 4 ( 0 . 3 1 ) 0 . 1 2 4 9 - 0 . 0 0 1 1 0 ( 0 .0 1 8 6 ) 1 .7 6 3 ( 0 .9 0 1 ) 0 .0 0 0 4 4 6 * ( 0 .0 0 0 2 1 2 ) 0 . 2 4 0 . 340 ( 1 . 9 5 ) 0 .4 4 1 10 0 .0 1 8 0 ( 0 .0 0 9 3 0 ) 0 . 7 9 8 ( 0 .4 3 4 ) 0 .0 0 0 8 7 9 * ( 0 .0 0 0 2 0 6 ) 0 . 3 9 0 . 4 5 7 ( 2 . 9 5 ) 0 . 1 6 9 11 - 0 . 0 0 2 0 1 ( 0 .0 1 7 3 ) - 0 . 0 7 8 1 * ( 0 .8 3 3 ) - 0 . 0 0 0 0 4 5 4 ( 0 .0 0 0 2 1 4 ) 0 . 0 1 0 . 268 ( 1 . 5 5 ) 0 . 3 9 7 12 - 0 . 0 0 2 5 3 . ( 0 . 0 1 0 4 ) 0 .1 6 8 * ( 0 .4 8 1 ) 0 .0 0 0 0 8 3 3 ( 0 .0 0 0 1 7 4 ) 0 . 0 1 0 . 4 5 0 ( 2 . 8 9 ) 0 . 210 13 - 0 . 0 0 2 3 6 ( 0 .0 0 8 8 4 ) 1 .3 9 2 ( 0 .4 4 1 ) 0 .0 0 0 2 6 5 ( 0 .0 0 0 2 1 9 ) 0 .3 3 0 . 4 7 2 ( 2 . 8 8 ) 0 . 1 1 3 14 - 0 . 0 0 0 1 1 8 ( 0 .0 1 2 4 ) 0 . 8 9 7 ( 0 .4 5 9 ) - 0 . 0 0 0 5 2 2 * ( 0 .0 0 0 1 7 8 ) 0 . 3 0 0 . 0 7 2 8 ( 0 . 4 1 ) 0 . 1 6 6 15 0 . 0 1 0 9 ( 0 .0 1 8 9 ) 0 .8 1 9 * ( 0 .8 5 6 ) 0 .0 0 0 6 3 3 ( 0 .0 0 0 3 8 5 ) 0 . 1 1 0 . 0 5 5 5 ( 0 . 2 8 ) 0 . 2 1 0 164 i t } egi 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 TABLE VII-3 (continuation) C on stant V2 V5 R2 0 t - 0 .0 0 1 2 6 0 .1 2 6 * 0 .0 0 1 0 6 * 0 .4 8 0. 113 ( 0 .0 1 0 2 ) ( 0 .4 4 2 ) ( 0 .0 0 0 3 2 3 ) ( 0 . 4 9 ) 0 .0 2 0 3 3 . 194 0 .0 0 0 1 3 7 0. 21 0 . 296 ( 0 .0 2 7 5 ) ( 1 .4 0 9 ) ( 0 .0 0 0 3 3 6 ) ( 1 .6 1 ) 0 . 0 0 5 4 6 1. 008 0 .0 0 0 2 2 2 * 0 .4 7 0 . 524 ( 0 .0 0 4 8 3 ) ( 0 .2 2 4 ) ( 0 .0 0 0 0 8 2 3 ) ( 3 - 5 3 ) 0 .0 0 0 4 4 6 0 .9 6 4 * 0 .0 0 0 2 3 2 * 0. 18 0 .0 1 0 7 (0 .0 1 7 4 ) ( 0 .6 0 5 ) ( 0 .0 0 0 1 3 5 ) ( 0 .0 6 ) 0 .0 0 6 9 8 - 0 .3 6 4 * 0 .0 0 0 1 2 4 0 .0 4 0. 192 ( 0 .0 1 6 9 ) ( 0 .7 1 8 ) ( 0 .0 0 0 1 6 0 ) ( 0 .9 7 ) 0 .0 0 8 2 0 1 .995 0 .0 0 0 1 0 4 0 .1 2 0. 160 ( 0 .0 2 5 8 ) ( 1 .0 2 8 ) ( 0 .0 0 0 2 8 5 ) ( 0 .9 3 ) 0 .0 0 5 3 0 1. 219 0 .0 0 0 5 9 4 * 0. 59 0 .4 0 2 ( 0 .0 0 5 9 5 ) ( 0 .2 6 6 ) ( 0 .0 0 0 1 1 6 ) ( 2 .5 2 ) - 0 .0 0 9 4 2 1 .3 8 2 - 0 .0 0 0 1 3 8 0 . 4 4 - 0 .0 9 4 7 ( 0 .0 0 9 0 2 ) ( 0 .2 8 5 ) (0 .0 0 0 1 4 1 ) ( - 9 - 5 4 ) - 0 . 0 0 0 0 3 1 8 0 .8 8 3 * - 0 .0 0 0 1 0 7 0 . 0 8 0 . 597 (0 .0 1 1 8 ) ( 0 .5 5 3 ) ( 0 .0 0 0 2 2 7 ) ( 4 .2 8 ) - 0 . 0 1 0 9 1-2 8 4 * - 0 . 0 0 0 1 2 5 0 .1 3 0 . 4 6 4 (0 .0 2 2 8 ) ( 0 .7 8 2 ) ( 0 .0 0 1 0 3 ) ( 2 .4 5 ) 0 .0 1 1 2 0 .9 6 9 0 .0 0 0 1 9 0 0 .0 8 6 0. 353 (0 .0 1 5 8 ) ( 0 .6 8 8 ) ( 0 .0 0 0 3 2 4 ) ( 2 . 1 6 ) 0 .0 0 3 3 9 0. 744 0 .0 0 0 6 1 6 * 0 . 2 8 0 .3 5 5 ( 0 .0 0 9 5 4 ) ( 0 .4 1 5 ) ( 0 .0 0 0 2 1 8 ) ( 2 .1 8 ) 0 .0 1 3 6 1 .5 6 2 0 .0 0 1 3 8 * 0. 34 0 . 4 2 7 ( 0 .0 1 3 0 ) ( 0 .5 8 6 ) ( 0 .0 0 0 4 3 2 ) ( 2 .7 1 ) 0 .0 0 0 7 3 3 0 .8 1 4 * 0 .0 0 0 6 4 0 * 0. 31 0 . 4 5 7 ( 0 .0 1 5 8 ) ( 0 .7 0 0 ) (0 .0 0 0 2 8 8 ) ( 2 .3 5 ) 0 .0 0 3 9 6 0 .2 0 0 * 0 .0 0 0 6 7 5 * 0 .2 3 - 0 . 0 2 4 4 ( 0 .0 1 3 8 ) (0 .5 7 7 ) ( 0 .0 0 0 3 5 1 ) ( - 0 . 0 9 7 7 ) SE 0 .0 2 5 2 0. 769 0 .0 5 0 0 0. 273 0. 180 0. 837 0 . 0 6 4 0 .0 6 3 6 0. 322 0 . 1 5 8 0 .4 0 2 0 .1 5 2 0. 311 0.110 0 .0 2 8 9 165 TABLE VII-4 EQUATION (V-6 ' ) ; REGIONS RESULTS • I n d i c a t e s i n c o n s i s t e n t w it h h y p o t h e s i s o f r e g i o n a l l y i n t e g r a t e d m arkets w ith m o b ile t e n a n t s C i t y / Reg ion C o n sta n t V2 V5 R2 P t SE 34 - 0 .0 0 1 2 9 ( 0 .0 0 3 2 2 ) 0. 928 ( 0 .1 4 3 ) 0 .0 0 0 3 6 3 * ( 0 .0 0 0 0 6 0 5 ) 0 .7 3 0 . 386 ( 2 .4 1 ) 0 .0 1 8 5 35 0 . 0 0 4 7 0 ( 0 .0 0 5 4 5 ) 1 .1 4 1 ( 0 .2 3 8 ) 0 .0 0 0 7 6 0 * ( 0 .0 0 0 1 9 9 ) 0 . 5 9 0 . 342 ( 2 .0 9 ) 0 .0 4 8 9 36 0 . 0 0 1 8 7 ( 0 .0 0 5 2 8 ) 0 .9 9 0 ( 0 .2 3 9 ) 0 .0 0 0 6 5 5 * ( 0 .0 0 0 1 7 2 ) 0 .5 2 0 . 4 9 2 ( 3 .2 5 ) 0 .0 5 5 9 37 0 . 0 0 4 7 6 ( 0 .0 0 4 1 0 ) 0 .4 7 5 * ( 0 .1 9 7 ) 0 .0 0 0 0 8 2 5 ( 0 .0 0 0 1 0 6 ) 0 .2 2 0 .6 4 2 ( 4 . 8 1 ) 0 .0 3 8 2 38 - 0 .0 0 1 7 1 ( 0 .0 0 7 0 5 ) 0 .3 9 1 * ( 0 .2 4 9 ) 0 .0 0 0 0 1 2 5 ( 0 .0 0 0 2 0 2 ) 0 . 0 8 - 0 . 0 0 6 5 9 ( - 0 . 0 3 7 ) 0 .0 4 5 9 39 - 0 .0 0 0 0 0 9 3 1 ( 0 .0 0 6 4 6 ) 0 .9 0 5 ( 0 .2 8 1 ) - 0 . 0 0 0 2 36 ( 0 .0 0 0 2 0 4 ) 0 .2 7 0 .3 7 5 ( 2 . 3 2 ) 0 .0 7 0 2 40 - 0 . 0 0 1 8 7 ( 0 .0 0 4 2 9 ) 1 .2 2 0 (0 .1 8 0 ) 0 .0 0 0 2 3 0 ( 0 .0 0 0 1 4 8 ) 0 .6 1 0. 255 ( 1 .5 1 ) 0 .0 2 5 5 166 167 Equations V-6 ( C i t i e s , 1959-1980) V-6 ' ( C i ti e s , 1948-1980) V-6' (Regions, 1948-1980) TABLE VII-5 GROSS RENTAL EQUILIBRIUM RESULTS SUMMARY Number of outcomes c o n s i s t e n t or i n c o n s i s t e n t w ith h y p o th e s is of r e g i o n a l l y i n t e g r a t e d m arkets w ith mobile t e n a n t s Constant V2 V5 V a ria b le s V6 V7 V8 V9 29 C o n siste n t 10 C o n siste n t 17 In co n clu siv e 2 I n c o n s i s t e n t 19 C o n s is te n t 25 C o n sis te n t 24 C o n sis te n t 27 C o n s is te n t 26 C o n sis te n t 10 I n c o n s i s t e n t 4 I n c o n s i s t e n t 5 I n c o n s i s t e n t 2 I n c o n s i s t e n t 3 I n c o n s i s t e n t 30 C o n sis te n t 15 C o n siste n t 10 In c o n clu siv e 5 I n c o n s i s t e n t 13 C o n siste n t 17 I n c o n s i s t e n t 30 C o n sis te n t 5 C o n siste n t 1 In co n clu siv e 1 I n c o n s i s t e n t 4 C o n siste n t 3 I n c o n s i s t e n t J 168 TABLE VII-6 EQUATION (V-7) RESULTS C ity / legion Constant V4 V5 V6 V7 1 -0.0501 0.340 -0.00198 0.0107 -0.00572 (0.112) (0.780) (0.00146) (0.0233) (0.00650) 2 0.00652 0.357 -0.00205 0.0171* 0.00818 (0.0373) (0.665) (0.000645) (0.0147) (0.00599) 3 -0.272 -5.871* -0.000923 0.0613 -0.00880 (0.324) (2.78) (0.000490! (0.0744) (0.0171) 4 0.0747 -1.340 -0.00204 -0.00636 0.00783 (0.107) (1.738) (0.00278) (0.00652) (0.0203) 6 0.0176 1.225* -0.000637 -0.00946 -0.00913* (0.0287) (0.555) (0.000383) (0.0109) (0.00385) 7 0.258 -4.983 -0.00260 -0.127 0.0411 (0.258) (4.572) (0.00293) (0.104) (0.0309) 8 0.156 -0.309 0.00330* -0.101 0.0265* (0.142) (2.542) (0.00229) (0.0664) (0.0151) 9 0.270 5.537* 0.000526 0.0216 0.00951 (0.199) (2.543) (0.00118) (0.0247) (0.0179) 10 0.0920 1.128 0.00115 0.0143 -0.00269 (0.0609) (1.307) (0.000950) (0.0245) (0.00426) ‘ I n d i c a te s i n c o n s i s t e n t w ith h y p o th esis of r e g io n a ll y i n t e g r a t e d m arkets w ith mobile t e n a n ts V8 V9 R2 P t -0.000392 1.380 0.45 -0.583 (0.00736) (1.321) (-2 .5 9 ) -0.00887* 3.908 0.51 0.411 (0.00504) (2.705) (2.12) 0.0227 -14.312 0.33 -0.452 (0.0145) (13.245) (-2 .3 8 ) -0.0143 -2.840 0.34 0.0294 (0.0150) (2.709) (0.12) -0.00660* -7.819* 0.64 0.196 (0.00296) (2.455) (0.94) 0.00384 -6.108 0.60 -0.00628 (0.0383) (5.694) ( -0 .0 2 ) -0.0184 -4.391 0.32 0.137 (0.0211) (3.461) (0.64) 0.00497 -3.900 0.37 -0.0950 (0.0186) (11.155) (-0 .4 4 ) -0.00331 -0.399 0.23 0.233 (0.00370) (3.356) (1.12) SE 0.114 0.425 10.266 1.318 0.316 2.339 4.936 6.504 1.455 TABLE VII-6 (continuation) City/ Region 11 12 13 14 15 16 17 18 19 20 Constant V4 V5 V6 V7 V8 V9 r2 P t 0.532 45.185* 0.0259* 0.973* -0.00843 0.0252 -27.211 0.57 -0.0728 (1.070) (15.584) (0.00928) (0.408) (0.0521) (0.0743 ) (25.450) (-0 .3 4 ) -0.357 1.398 0.00172 0.0823 0.0147 -0.0348 12.049 0.26 -0.326 (0.334) (4.236) (0.00225) (0.0568) (0.0254 ) (0.0329) (9.327) (-1 .6 1 ) 0.0174 1.395* -0.000195 -0.0127 0.00429 0.00328 -0.195 0.50 0.441 (0.0346) (0.581) (0.000587) (0.00486) (0.00539) (0.00483) (2.450) (2.30) -0.845 -22.261 0.0470* 0.482 0.0422 -0.137 -72.971 0.58 -0.152 (1.231) (18.793) (0.0112) (0.323) (0.114) (0.122) (50.082) (-0.72) -0.0351 -1.317 -0.0000952 0.0169 -0.00792 -0.00642 -1.989 0.19 -0.111 (0.0925) (1.223) (0.00132) (0.0359) (0.0107) (0.00921) (4.818) (-0.525) 0.307 3.718 0.00157 -0.0854 0.0111 -0.0331 22.373 0.57 -0.207 (0.262) (4.134) (0.00132) (0.0604) (0.0156) (0.0420) (23.849) (-0.735) -2.738 -32.522 -0.00522 -0.080 0.112 -0.154 -19.078 0.21 -0.112 (1.884) (25.406) (0.0215) (0.391) (0.133) (0.253) (89.286) (-0 .5 2 ) 0.0414 1.280* 0.000105 -0.00703 0.00352 0.00155 1.741 0.39 0.441 (0.0287) (0.580) (0.000391) (0.00943) (0.00323) (0.00369) (1.587) (2.30) 0.148 0.329 -0.0000939 -0.0575 0.000255 0.00129 0.372 0.14 0.139 (0.123) (2.074) (0.000845) (0.0474) (0.00823) (0.00880) (5.939) (0.65) 0.166 -1.642 0.00339* 0.0636 -0.000469 0.00828 16.862* 0.66 -0.0711 0.110 (1.132) (0.00241) (0.0475) (0.0128) (0.00490) (6.154) (-0 .2 6 ) SE 252.460 14.339 0.305 282.194 1.399 1.401 799.301 0.360 3.925 0.406 170 TABLE VII-6 (continuation) City/ Region 2i 22 23 24 25 26 27 28 29 30 C onstant V4 V5 V6 V7 V8 V9 R2 t -0.230 -1.742 0.00277* 0.00352 0.0202 -0.00189 3.750 0.28 -0.353 (0.368) (3.626) (0.00182) (0.0588) (0.0206) (0.0125) (6.795) (-1 .7 7 ) 0.0114 0.844 0.000131 0. UU<!98 -0.00437 -0.0111 -3.237 0.40 0.0695 (0.0590) (1.0217) (0.000743) (0.0262) (0.00716) (0.00701) (7.407) (0.32) -0.00238 2.142* -0.00166 -0.0282 -0.00506 -0.00489 -7.787 0.50 0.213 (0.0434) (0.867) (0.000949) (0.0198) (0.00619) (0.00642) (4.918) (1.02) 0.199 0.623 -0.00172 0.0407* -0.0128 -0.00364 8.753 0.28 0.329 (0.104) (2.289) (0.00175) (0.0472) (0.00926) (0.00682) (6.087) (1.63) 0.0152 0.681 0.000835 -0.00135* 0.00248 -0.0153* 1.828 0.93 0.0713 (0.0243) (0.403) (0.000837) (0.00982) (0.00793) (0.00341 (1.183) (0.23) 0.0203 0.3071 -0.0000190 0.00170 -0.00373* 0.000376 1.314 0.38 -0.129 (0.0306) (0.449) (0.000363) (0.00649) (0.00140) (0.00205) (2.251) (-0 .6 1 ) 0.0575 0.292 0.000599 0.0164 -0.00346 -0.00448 0.793 0.34 0.301 (0.0463) (0.871) (0.000761) (0.0202) (0.00324) (0.00363) (2.463) (1.48) 0.0186 -0.156 0.00323* -0.0213 0.00389 0.00591 -1.599 0.50 0.499 (0.0512) (1.0500) (0.00144) (0.0221) (0.00491) (0.00426) (2.452) (2.70) 0.729 -2.598 -0.0040 0.477 -0.0436 0.00279 24.398 0.76 -0.288 (0.616) (4.918) (0.00498) (0.404) (0.0827) (0.0238) (42.720) (-1 .0 0 ) 0.0639 -0.619* 0.000454 0.0147* -0.00181* -0.00189 3.292* 0.82 0.0375 (0.0269) (0.333) (0.000622) (0.00734) (0.000654) (0.00141) (0.677) (0.14) SE 17.757 0.884 0.574 3.595 0.0125 0.135 0.740 1.036 1.072 0.0358 j TABLE V I I -7 EQUATION ( V - 7 ‘ ); CITIES RESULTS ‘ I n d i c a t e s i n c o n s i s t e n t w i t h h y p o t h e s i s o f r e g i o n a l l y i n t e g r a t e d m a rkets w it h m o b ile t e n a n t s C i t y / Region C o n sta n t V4 V5 R t SE 1 0 .0 3 7 3 ( 0 .0 4 1 0 ) 0 . 3 2 6 ( 0 .6 1 5 ) 0 . 0 0 0 4 9 4 ( 0 .0 0 1 0 4 ) 0 .0 1 - 0 . 1 5 5 ( - 0 . 7 6 ) 0 .5 3 2 2 0 .0 0 4 7 2 ( 0 .0 2 0 1 ) 0 . 4 2 5 ( 0 .4 5 6 ) - 0 . 0 0 1 8 1 ( 0 .0 0 0 5 3 6 ) 0 . 2 8 0 . 348 ( 2 . 1 3 ) 0 . 6 9 2 3 - 0 . 2 2 4 ( 0 . 2 3 2 ) - 4 . 2 7 3 * ( 2 .2 8 2 ) - 0 . 0 0 0 4 3 9 ( 0 .0 0 0 3 8 5 ) 0 . 1 5 - 0 . 3 6 2 ( - 1 . 9 8 ) 1 3 .0 5 1 4 0 .0 1 7 7 ( 0 .0 4 7 3 ) - 1 . 3 7 0 ( 0 .9 7 3 ) - 0 . 0 0 0 8 9 9 ( 0 .0 0 1 3 3 ) 0 . 6 8 0 . 1 7 3 ( 0 . 9 3 ) 1 .9 2 1 5 0 . 0 7 2 5 ( 0 .0 7 0 3 ) 0 . 2 4 8 ( 1 .9 8 8 ) 0 .0 0 3 1 9 * ( 0 .0 0 1 0 4 ) 0 . 4 0 0 . 1 2 3 ( 0 . 5 2 ) 1. 335 6 0 . 0 1 0 6 ( 0 .0 2 0 0 ) 0 .8 5 8 * ( 0 .4 7 0 ) - 0 . 0 0 0 2 79* ( 0 .0 0 0 3 3 0 ) 0 . 1 4 0 . 4 1 9 ( 2 . 6 5 ) 0 . 748 7 0 . 1 2 9 ( 0 .0 7 5 8 ) - 3 . 7 0 8 * ( 1 .9 2 2 ) - 0 . 0 0 2 7 3 ( 0 .0 0 1 3 8 ) 0 . 2 9 0 . 356 ( 1 . 8 6 ) 4. 261 8 0 . 0 9 9 9 ( 0 .0 6 4 8 ) - 0 . 0 5 8 4 ( 1 .3 7 6 ) 0 .0 0 3 0 3 * ( 0 .0 0 1 5 6 ) 0 . 1 1 0 . 349 ( 2 . 1 4 ) 6 .4 0 2 9 0 . 1 6 4 ( 0 . 1 0 8 ) 4 . 3 6 3 * ( 1 .8 4 9 ) 0 .0 0 0 5 7 0 ( 0 .0 0 0 8 7 6 ) 0 . 2 0 - 0 . 0 1 4 6 ( - 0 . 0 7 ) 8 .1 1 7 10 0 .0 6 4 5 ( 0 .0 3 7 3 ) 0 . 2 8 4 ( 0 .8 1 6 ) 0 .0 0 1 2 7 * ( 0 .0 0 0 7 3 0 ) 0 . 1 0 0 . 297 ( 1 . 7 8 ) 2 .2 0 2 11 - 0 . 8 1 3 ( 0 .6 4 3 ) 3 8 .9 2 0 * ( 1 2 .8 5 8 ) 0 .0 1 3 3 * ( 0 .0 0 6 6 9 ) 0 . 3 1 0 . 100 ( 0 . 5 6 ) 4 0 7 .8 0 4 12 - 0 . 2 2 6 ( 0 . 2 3 0 ) 1 .7 4 3 ( 2 .4 3 1 ) 0 .0 0 0 2 8 5 ( 0 .0 0 1 5 3 ) 0 . 0 1 - 0 . 3 9 0 ( - 2 . 4 3 ) 1 9 .5 5 7 13 0 .0 1 2 3 ( 0 .0 2 2 7 ) 1 .4 6 4 * ( 0 .5 2 9 ) 0 .0 0 0 2 6 2 ( 0 .0 0 0 5 4 1 ) 0. 24 0 . 364 ( 2 . 1 0 ) 0 . 6 3 7 14 - 0 . 5 9 0 ( 0 .8 0 1 ) 3. 189 ( 1 0 .9 0 9 ) 0 .0 3 4 1 * ( 0 .0 0 7 9 1 ) 0 . 4 0 - 0 . 1 8 8 ( - 1 . 1 0 ) 4 0 8 .3 9 7 15 0 .0 1 8 0 ( 0 .0 6 1 4 ) - 1 . 1 3 5 ( 0 .9 9 6 ) 0 . 0 0 0 3 8 4 ( 0 .0 0 1 0 1 ) 0 . 0 8 - 0 . 0 6 2 8 ( - 0 . 3 2 ) 1 .7 7 7 171 1 I 1 TABLE VI1 -7 ( c o n t i n u a t i o n ) C i t y / R egion C o n sta n t V2 V5 R2 P t SE 16 - 0 . 0 1 6 1 ( 0 .0 1 9 3 ) - 0 . 2 7 3 ( 0 .5 9 5 ) 0 .0 0 0 3 0 7 ( 0 .0 0 0 7 1 0 ) 0.01 0 . 4 2 3 ( 2 . 0 3 ) 0 . 1 4 9 17 - 2 . 0 3 5 ( 1 .4 1 9 ) - 2 6 . 2 3 0 ( 2 0 .7 3 0 ) - 0 . 0 0 6 2 6 ( 0 .0 1 2 3 ) 0 . 0 8 - 0 . 1 2 6 ( - 0 . 6 6 ) 9 4 6 .1 3 3 18 0 . 0 4 0 7 ( 0 .0 1 7 5 ) 1 .4 0 6 * ( 0 .4 1 5 ) 0 .0 0 0 1 0 1 ( 0 .0 0 0 2 9 5 ) 0 . 2 8 0 . 4 6 0 ( 2 . 9 8 ) 0 . 6 1 7 19 - 0 . 0 1 3 4 ( 0 .0 7 3 2 ) - 1 . 1 8 0 ( 1 .5 7 8 ) 0 .0 0 0 9 9 1 * ( 0 .0 0 0 7 4 8 ) 0 . 0 6 0 . 2 5 6 ( 1 - 5 2 ) 7 .9 0 3 20 0 . 0 2 1 8 ( 0 .0 4 0 1 ) - 0 . 4 5 2 ( 0 .8 7 6 ) - 0 . 0 0 0 3 1 5 ( 0 .0 0 0 4 3 5 ) 0 . 0 3 0 . 278 ( 1 . 4 5 ) 1 .2 7 5 21 - 0 . 1 0 7 ( 0 . 2 0 9 ) 0 . 9 0 6 ( 2 .4 5 3 ) 0 .0 0 2 8 4 * ( 0 .0 0 1 2 0 ) 0 . 1 6 - 0 . 2 8 6 ( - 1 . 7 1 ) 2 1 .8 0 5 22 0 .0 0 9 6 2 ( 0 .0 2 3 4 ) 0. 582 ( 0 .5 8 6 ) 0 .0 0 0 1 9 9 ( 0 .0 0 0 4 9 8 ) 0 . 0 3 0 . 4 8 6 ( 3 . 1 9 ) 1 .1 2 2 23 0 . 0 0 7 8 9 ( 0 .0 3 2 4 ) 1 .7 1 9 * ( 0 .6 2 0 ) - 0 . 0 0 1 1 4 ( 0 .0 0 0 6 1 4 ) 0 . 2 7 0 . 1 1 5 ( 0 . 6 6 ) 1. 242 24 0 .0 7 6 4 ( 0 .0 5 2 7 ) - 0 . 7 6 3 ( 1 .2 2 5 ) - 0 . 0 0 1 4 5 ( 0 .0 0 0 9 6 2 ) 0 . 0 9 0 . 387 ( 2 . 4 1 ) 4. 985 25 0 . 0 2 8 6 ( 0 .0 9 2 5 ) 1 .0 5 4 7 ( 1 .7 3 7 ) - 0 .0 0 0 1 9 2 ( 0 .0 0 4 3 7 ) 0 .0 2 1 0 . 382 ( 1 . 9 4 ) 2 . 5 7 0 26 0 . 140 ( 0 .1 2 9 ) 1 . 6 5 1 ( 2 .6 6 6 ) 0 .0 0 0 2 4 2 ( 0 .0 0 2 6 0 ) 0 .0 1 0 .1 7 4 ( 1 .0 1 7 ) 2 1 .1 0 4 27 0 .0 2 9 3 ( 0 .0 2 5 8 ) 0 . 7 6 0 ( 0 .5 7 7 ) 0 . 0 0 0 6 2 5 ( 0 .0 0 0 5 8 8 ) 0 . 1 0 0 . 3 3 2 ( 2 . 0 2 ) 1 .0 9 8 28 0 . 0 5 5 0 ( 0 .0 3 0 5 ) - 0 . 3 6 6 ( 0 .7 3 5 ) 0 .0 0 2 3 2 * ( 0 .0 0 1 0 6 ) 0 . 1 7 0 . 4 7 2 ( 3 . 0 8 ) 1 . 7 9 9 29 0 . 0 9 4 7 ( 0 .0 6 9 4 ) - 1 . 6 1 4 ( 1 .8 9 3 ) 0 .0 0 0 1 6 5 ( 0 .0 0 1 4 4 ) 0 . 0 4 0 . 4 6 6 ( 2 . 4 1 ) 2 .6 7 1 30 - 0 . 0 0 2 3 6 ( 0 .0 3 7 9 ) - 0 . 4 6 3 ( 0 .4 8 4 ) 0 .0 0 0 3 4 3 ( 0 .0 0 0 8 8 6 ) 0 . 0 9 - 0 . 0 8 3 6 ( - 0 . 3 3 ) 0 . 1 9 5 172 EQUATION (V - 7 1); REGIONS RESULTS • I n d i c a t e s i n c o n s i s t e n t w ith h y p o t h e s i s o f r e g i o n a l l y i n t e g r a t e d m arkets w ith m o b ile t e n a n t s C i t y / Region C o n sta n t V4 V5 R2 P t SE 34 0 .0 0 9 7 7 (0 -0 2 2 5 ) 0 . 321 ( 0 .4 8 3 ) - 0 .0 0 0 8 2 8 * ( 0 .0 0 0 3 8 9 ) 0 . 1 6 0 . 246 (1 - 4 6 ) 0 .7 4 3 • 35 0 .0 1 6 9 ( 0 .0 1 4 9 ) 1 .0 2 6 * ( 0 .3 8 0 ) - 0 .0 0 0 1 4 7 ( 0 .0 0 0 6 4 8 ) 0. 21 0 . 4 5 6 ( 2. 94 ) 0 .4 4 1 36 0 .0 1 8 8 ( 0 .0 1 5 5 ) 0 .6 4 2 * ( 0 .3 6 2 ) 0 .0 0 0 8 5 5 ( 0 .0 0 0 5 0 3 ) 0 . 1 8 0 .4 5 5 ( 2 .9 4 ) 0 . 4 6 7 37 0 .0 1 4 7 ( 0 .0 1 1 5 ) 0 . 139 ( 0 .3 1 2 ) - 0 .0 0 0 2 1 3 ( 0 .0 0 0 3 2 5 ) 0.0 1 0. 571 ( 3 .9 9 ) 0 . 2 9 0 38 - 0 .0 0 0 8 3 2 ( 0 .0 1 4 8 ) 0. 364 ( 0 .3 1 1 ) - 0 .0 0 0 3 2 8 (0 .0 0 0 5 0 4 ) 0 .0 7 0 . 213 ( 1 .2 5 ) 0. 297 39 0 . 0 0 3 3 6 ( 0 .0 1 6 1 ) 1 .0 5 1 * ( 0 .3 3 5 ) - 0 . 0 0 0 2 1 0 (0 .0 0 0 4 6 8 ) 0 . 2 5 0. 234 ( 1 .3 8 ) 0 .3 6 2 40 0 .0 0 5 3 6 (0 .0 1 7 5 ) 0 .8 0 3 * ( 0 .4 0 5 ) - 0 . 0 0 0 8 5 7 (0 .0 0 0 6 6 0 ) 0. 21 0 . 337 ( 2 .0 6 ) 0 .4 9 6 TABLE VII-9 NET RENTAL EQUILIBRIUM RESULTS SUMMARY Number o f outcomes c o n s i s t e n t o r i n c o n s i s t e n t w ith h y p o th esis of r e g i o n a l l y i n t e g r a t e d m arkets w ith mobile t e n a n t s Equations C onstant V a r ia b le s V4 V5 V6 V7 V8 V-7 ( C i ti e s , 1959-1980) 29 C o n siste n t 21 C o n sis te n t 23 C o n s is te n t 24 C o n sis te n t 25 C o n sis te n t 26 C o n s is te n t 26 8 I n c o n s i s t e n t 6 I n c o n s i s t e n t 5 I n c o n s i s t e n t 4 I n c o n s i s t e n t 3 I n c o n s i s t e n t 3 V-7' ( C i ti e s , 1948-1980) 30 C o n sis te n t 22 C o n sis te n t 22 C o n siste n t 8 I n c o n s i s t e n t 8 I n c o n s i s t e n t V-7' (Regions, 1948-1980) 30 C o n sis te n t 3 C o n sis te n t 6 C o n siste n t 4 I n c o n s i s t e n t 1 I n c o n s i s t e n t V9 C o n s is te n t I n c o n s i s t e n t C H A PTER VIII S U M M A R Y A N D C O N C LU SIO N S VIII.1 S U M M A R Y As was mentioned in the introduction to this study, both general business and real estate trade literature frequently contain discussions of the relative strength or desirability of different real estate markets. Quite often this market differentiation is on the basis of geographic location. It is frequently implied if not explicitly stated that the performance of investment real estate in the region under discussion will be superior to th a t in other regions. The ex isten ce of such opportunities presented in this way assumes a significant inefficiency and segmentation of the investment real estate market across the United States. The existence, or not, of such inefficiencies or segmentation in the market for any type of asset goes right to the heart of many critic a l investment issues. Among them are risk, holding period, and diversification. As real estate ownership in the United States becomes more institutionalized the need rises for investment policy guidelines. Without such guidelines it is impossible to evaluate adequately the performance of the managers and fiduciaries who will play an increasingly important role 175 in investment decisions. Of great interest to those establishing guidelines would be accurate information concerning the presence of inefficiencies or segmentation in the overall market. I t has been the o b je c tiv e of th is study to investigate the question of the existence of geographic segmentation of a particular sector of the real estate market in the United States. The sector upon which attention has been focused is that of downtown office properties in major cities across all reigons of the country. Availability of data was one determining factor in selecting that property type. However, another factor was the extent to which that sector of the market, above all others, would be expected to be integrated nationally. If the evidence refutes this assertion, i t would be d ifficu lt to contend that other segments of the real estate market are likely to be integrated. The model used in the empirical analysis was derived from the factor price equalization theorem. It incorporates the issues of intermarket trade in goods and services, factor mobility, and local government fiscal policy. In this context downtown office properties are assumed to play the role of an immobile factor of production. If geographically distinct markets in the United States are in fact integrated into one national market, several patterns should be observed. First, gross rents to real estate would be expected to be correlated nationally. That is, in any particular geographic market the suppliers of real estate services would be price takers on a gross rent basis. Those gross : re n ts would be e s ta b lis h e d on n a tio n a l market considerations. Second, net rents received need not be | correlated across the d iffe re n t markets. This is attributable to real estate's inability to flee from or j ! pass onto mobile tenants the strictly localized aspects of ! operating costs such as property taxes, energy, and maintenance. Hence, the model dealt with two separate equilibrium I conditions. The national equilibrium condition was based on local and national gross rent relationships. The local j I equilibrium condition was based on net rental income j relationships. The integrated market's hypothesis calls ' for specific outcomes in each. To the extent they are i I confirmed or refuted by empirical findings, so then would | our confidence in the national integration of regional I real estate markets be strengthened or weakened . 1 i For the most part, as presented in Chapter VII, j the results of the tests tend not to provide support for ; rejection of the hypothesis of regionally integrated | I markets with mobile tenants. Most consistent in this | • indication was the outcome for the constant term in all equations. In no case in either time frame or in any city was this term significantly different from zero. The next strongest indications consistent with ' regionally integrated markets were the outcomes of the ;variables dealing with local fiscal policy (V 6 , V7, and V 8 ) and local aggregate demand (V9). For each of the four ’variables in both the gross and net rental equations I rejection of market integration was never indicated in t Jmore than five c itie s . This g reatest incidence of rejection is only two more than the number of erroneous reject outcomes which would be expected to occur purely by jchance. Results of the test of the relationship between change in gross rental income for an individual city or region (VI) and change in average gross rental income in j the United States (V 2) also lent support to the integrated market hypothesis. Support became stronger in the longer time frame. A number of cities and one region yielded I inconclusive results. In those cases values generated for I the co efficien ts did not allow for rejecting the jhypothesized values for either the fully integrated or (com pletely segmented market. That incidence of ■ inconclusive results could be attributable to inadequacies j in the data. As was mentioned in Chapter VI, the data was 1 I i I 178 ' compiled on the basis of voluntary submissions by property owners and managers. It was not gathered specifically for the purpose of this study. No particular geographic tendancy or pattern in the overall results was apparent other than the fact that several of the cities which had inconclusive indications or rejected market integration outright were situated in the two regions (Southern and |Pacific Southwest) which also had that outcome, j Somewhat less unqualified support was lent to the l hypothesis of integrated markets with mobile tenants by ; the results of the examination of the relationship between change in net rental income for individual cities or regions (V3) and the change in the average net rental :income in the United States (V4). Eight of the thirty cities and four of the seven regions indicated that the 'hypothesized zero correlation between local and national averages could be rejected. One implication of this outcome could be a lack of mobility on the part of tenants !which renders them subject to the burden of strictly local expenses which would otherwise (with tenant mobility) be escaped. S till, there is another explanation for such a I [result which is not inconsistent with market integration. i 'it is possible that the effects of changes in the strictly local component of operating expenses may be swamped by variations in gross rental incomes and components of I operating expenses which are common across all cities and > regions. If that were the case a rejection of zero i , correlation between changes in local and national average net incomes might merely be a reflection of the difficulty . , in capturing the effect of a relatively small factor in the presence of confounding effects of other factors of ' t J greater magnitude. In case of the relationship between changes in j local property tax differentials (V5) and changes in gross | rental income in individual c itie s or regions (VI) i ! I rejection of the hypothesis of integrated markets with j j i mobile tenants is called for in nearly half of the cities , and regions. That level of rejection is unlikely to be > I ■ ! attributable to pure chance. The appearance is that increases in the local property tax differential are | accompanied by increases in the local gross rental income. \ I : Such an occurrence could be interpreted as demonstrating a | I ■ shift of the property tax burden to an immobile tenant by i means of a gross rent increase. However/ an alternative ; interpretation of this outcome which is consistent with ! , i ! integrated markets and mobile tenants is possible. ; ; i ; Perhaps the relationship observed is a reflection of tax j I 1 \ revenues (not necessarily tax rates) increasing as a ; I jconsequence of an increase in property value implied by J ; I gross rental increases. \ 1 Results of the test of the relationship between changes in local property tax differentials (V5) and changes in net rental income for individual cities and regions (V3) lent substantial support to the hypothesis of market integration and tenant mobility. Of the seven regions, only one indicated rejection. That is, in fact, the very number of erroneous rejections which one would expect to observe on the basis of chances given the confidence level involved. In the case of the cities the [number of rejectio n s is six and eight for the two I different time frames. The likelihood of that many rejections being erroneous indications and having occurred purely by chance is rather small. Thus, there seems again to be some evidence that in a few of the cities the tenants are constrained from the standpoint of mobility and may be subject to at least a portion of the burden of changes in the stric tly local components of property itaxes. ''VI11. 2 C O N C L U S IO N S I ■ The great majority of conclusive outcomes in this jresearch are consistent with regionally integrated markets I jwith mobile tenants. In the gross rental income analysis a substantial number of inconclusive 1 ■obtained. The inconclusive nature of those j I 181 re su lts were findings could ; be attributable to problems with the data. The results of the tests of the property tax and net rental income variables produced more rejections of the integrated market hypothesis than would be expected to occur solely on the basis of chances. However, as mentioned in the , preceding paragraphs, there are in several of those cases possible explanations for such outcomes which are s t ill i ® consistent with integrated markets and mobile tenants. j Apart from widespread consistency with the I ' integrated market hypothesis, patterns of outcomes were not apparent. Among the different variables within an ‘individual city or region there were no meaningful patterns of rejection of the values hypothesized for an ;integrated market with tenant mobility. Sim ilarly, I [geographic patterns of rejection were absent except in the case of gross rental income in the Southern and Pacific ;Southwest regions. In those particular regions as a whole, and in several of the cities located therein, the results indicate that over the long term gross rental |rates have changed at a lower rate than the United States average. i j By the preponderance of the evidence developed in I ithis research, a regionally integrated market with mobile 'tenants can not be rejected. The d irec t operative [implication of this result would be that, over the long I i i 182 term, prospective tenants of downtown office space in I ■ cities in the United States would not be at any relative advantage or disadvantage from the standpoint of (gross) i rent a levels to take space in any p a rtic u la r city. Exceptions may have been revealed for three cities in the Southern and Pacific Southwest regions. That number of exceptions, however, is precisely the number of erroneous outcomes which is most lik ely to occur given the i j confidence level used in the empirical analysis. » t The outcome of this research does not provide any direct conclusive indication of the rates of return which ■ investors in real estate may experience in different |c itie s or regional markets. Investigation of that question requires data on transaction prices not just rental income and expense figures. Such transaction or |valuation data is not presently available and is unlikely to be so (in a form suitable for rigorous empirical analysis) for some years to come. Investors and others interested in that issue, however, should take note of the following. The hypothesis of regionally integrated jmarkets with which this study's findings are consistent I I linvolves assumptions such as mobility of capital and interregional trade in goods and services which are consistent with assumptions that might underlie a market ! which, is fully integrated with. respect to rate of return I : opportunites. I BIBLIO G RA PH Y Batra, R., & Scully, G . (1972). Technical Progress, economic growth and the North-South wage differential. Journal of Regional Science, 12, 375- 386. : Bossons, John D . (1978). Housing demand and household wealth. In L.S. Bourne and J.R. Hitchcock (Eds.), , Urban Housing Markets. Toronto: University of | Toronto Press. i ' iCanto, Victor A., & Webb, Robert I. (1981, January). Persistent Growth Rate Differentials Among States in a National Economy with Factor Mobility. Paper I presented at the Center for the Study of Private Enterprise of the University of Southern California Conference on Taxation of Income From Capital, Los i Angeles, C A . ; I ; Coelho, P., & Ghali, M . (1971). The end of the North- 1 South wage differential. American Economic Review, £1(2), 231-253. 1 Demsetz, Harold (1968). The cost of transacting. Quarterly Journal of Economics, 8^(1), 33-53. i jDraper, Dennis W .,& Findlay, M . Chapman (1982). Capital asset pricing and real estate valuation. A REU EA I Journal, _10(2), 152-183. Elton, E.J., & Gruber, M.J. ( 1970 ). Marginal stockholders' tax rates and the clientele effect. Review of Economics and Statistics, _52(1), 68-74. j Emerson, M . Jarvin, & Lamphear, F. Charles (1975). Urban i and Regional Economics. Boston: Allyn and Bacon I | Inc. — ! I |Fama, E.F. (1965). The behavior of stock market prices. Journal of Business, £8(1), 34-105. j j_ _ _ _ _ _ _ _ _, & MacBeth, J. ( 1 973 ). Risk, return and I i equilibrium: empirical test. Journal of Political ■ ! Economy, 8£(3), 607-636. ! f jFindlay, M.C., Messner, S.D., & Tarantello, R.A. (1979, | August). Risk Analysis in Real Estate. Paper i presented at the A R E U E A Annual Meeting, Chicago, IL. i j i I 185 Greenwood, M . (1973). Research on internal migration in the United States: a survey. Journal of Econonmic Literature, 13, 397-433. Henderson, J.V., & Ioannides, Y.M . (1983). A model of housing tenure choice. American Economic Review, 73(1), 98-113. ” Hirshleifer, J. (1973). Where are we in the theory of information? American Economic Review, 63^(2), 32-39. Hu, T. (1982). Econometrics: An Introductory Analysis (2nd Ed.). Baltimore: University Park Press. ; Ioannides, Y.M . (1979). Temporal risks and the tenure decision in housing markets. Economic Letters, 4, 293-297. I : Johnston, J. (1963). Econometric Methods (2nd Ed.). New York: McGraw-Hill. Laffer, A.B., & Miles, M .A. (1982). International Economics in an Integrated World. Glenview, IL: Scott, Foreman & Co. Lancester, K . (1966). A new approach to consumer theory. Journal of Political Economy, 74^(2), 132-157. Lefeber, L. (1964). Allocation in space. In P.N. Rosenstein-Rodan (Ed.), Pricing and Fiscal Policy. N ew York: Praeger. Markowitz, H .M . (1959). Portfolio Selection. New Haven: Yale University Press. Miller, M ., & Modigliani, F. (1961). Dividend policy growth and the valuation of shares. Journal of Business, 34_(4), 411-433. Moore, Basil J. (1968). An Introduction to the Theory of Finance. N ew York: The Free Press. Mundell, R.A. (1957). International Trade and Factor Mobility. American Economic Review, _57(3), 321-335. Pettit, R.R. (1977). Taxes, transaction costs and clientele effects of dividends. Journal of Financial Economics, 5(3), 419-436. 186 Richardson, Harry W . (1969). Regional Economics. New York: Praeger. Roll, R . (1977). A critique of the asset pricing theory's tests. Journal of Financial Economics, 4(2), 129- 176. _ _ _ _ _ _ _ _ _, & Ross, S. (1980). A n Empirical investigation of the arbitrage pricing theory. Journal of Finance, 35(5), 1073-1103. Rosen, S. (1974). Hedonic prices and implicit markets: product differentiation in pure completion. Journal of Political Economy, 81^(1), 32-55. Rosen, H.S. (1979). Housing decisions and the U.S. income tax: An econometric analysis. Journal of Public Economics, 11, 1-24. Ross, S. (1976). The arbitrage theory of capital asset pricing. Journal of Economic Theory, 1J3(2), 343-362. Samuelson, P.A. (1948). International trade and the equalization of factor prices. Economic Journal, 58,, 163-184. _ _ _ _ _ _ _ _ _ (1949). International factor-price equalization once again. Economic Journal, 59, 183-197. SA S User Guide: Econometrics. (1980). Raleigh, N.C.: SA S Institute Inc. Sharpe, William (1970). Portfolio Theory and Capital Markets. N ew York: McGraw-Hill. Shelton, John P. (1968). The cost of renting vs. owning a home. Land Economics, 44(1), 59-72. Stigler, George J. (1961). The economics of information. Journal of Political Economy, 69(3), 213-225. Tinic, Seha M ., & West, Richard R . (1979). Investment in Securities: An Efficient Markets Approach. Reading, M A : Adison-Wesley. 187

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Variable-rate versus fixed-rate mortgage instruments: A theoretic approach

PDF

The output and employment effects of fiscal policy in a classical model

PDF

A phenomenological inquiry into the experience of time pressure: The homeowner in foreclosure

PDF

Cross-sectional tests of tax and regulatory costs of a change in depreciation methods in the railroad industry

PDF

An investigation of the impact of congruent managerial values on appraisal behavior

PDF

An investigation of market fragmentation and the specialist's quotation strategy

PDF

Balance of trade, economic activity and public policy

PDF

A study of the factors affecting software testing performance and computer program reliability growth

PDF

A grounded theory account of strategic decision-making as a key mediating mechanism between organization and environment

PDF

Communications medium and product class: Their effect on negotiation outcome

PDF

An interactive cross-impact scenario generator for long-range forecasting

PDF

Credit markets under imperfect information

PDF

Cardiopulmonary coupling in the control of breathing during exercise

PDF

A study of the relationship of negotiator skill and power as determinants of negotiation outcome

PDF

An investigation of information requirements determination and analogical problem solving

PDF

Causal-integrative model: A management tool for the analysis of cost overruns in major systems acquisition

PDF

Communication and the implementation of information systems: The case of word processing in state and local governments

PDF

An analysis of United States overseas English language policy, 1938-1990

PDF

A contingent claims analysis of the equity interest in property-casualty insurance companies

PDF

A study on the content and determinants of close exchange relationships in channels of distribution

Asset Metadata

Creator Selvidge, Ross Stanley (author)

Core Title An investigation into the regional segmentation of the commercial real estate market in the United States

Degree Doctor of Philosophy

Degree Program Business Administration

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag business administration, general,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Canto, Victor A. (committee chair), Baer, William C. (committee member), Tarantello, Rocky (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-750639

Unique identifier UC11345314

Identifier DP22628.pdf (filename),usctheses-c17-750639 (legacy record id)

Legacy Identifier DP22628.pdf

Dmrecord 750639

Document Type Dissertation

Rights Selvidge, Ross Stanley

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA