University of Southern California Dissertations and Theses

Regulation through information: Capital markets and the environment

(USC Thesis Other)

Regulation through information: Capital markets and the environment

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content REGULATION THROUGH INFORMATION: CAPITAL MARKETS AND THE ENVIRONMENT Copyright 2002 by Tisha Lin Nakao Emerson A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (ECONOMICS) December 2002 Tisha Lin Nakao Emerson R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. UMI Number: 3093756 UMI UMI Microform 3093756 Copyright 2003 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES, CALIFORNIA 90007 This dissertation, written by Tisha Lin Nakao Emerson under the direction of h&$>.... Dissertation Committee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of re quirements for the degree of DOCTOR OF PHILOSOPHY m of Graduate Studies Date December 18, 2002 DISSERTATION COMMITTEE Chairperson R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Dedication This work is dedicated to my parents, Russell and Valerie Nakao, my grandparents, Robert and Kammy Jervis, and my husband, Peter Emerson. Thank you for always believing in me. You knew that I could do this even when I didn’t. Your love and support helped me every step of the way. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Acknowledgements I gratefully acknowledge the financial support of the University o f Southern California and of Baylor University. I would also like to thank my dissertation committee chairperson, Peter Rosendorff, for his helpful comments and support. Finally, I am most grateful to my mentor, friend, and advisor, Linwood Pendleton. Through many discussions, Linwood has provided invaluable support and guidance without which I would have been lost. My appreciation goes beyond words. Thank you. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Table of Contents Dedication............................................................................................................................................................ ii Acknowledgements............................................................................................................................................iii List of Tables......................................................................................................................................................vi List of Figures................................................................................................................................................... vii Abstract............................................................................................................................................................ viii Chapter 1: Introduction...................................................................................................................................... 1 Overview.....................................................................................................................................................2 Capital Markets Inefficiencies: Signaling Through Trade Policy, A Second Best Solution 4 Income, Disamenity, and Toxic Releases in the United States......................................................... 8 References................................................................................................................................................ 11 Chapter 2: Capital Markets Inefficiencies: Signaling Through Trade Policy, A Second Best Solution................................................................................................................................... 14 Chapter Abstract..................................................................................................................................... 15 Introduction.............................................................................................................................................. 16 Illustration of the Capital Market Inefficiency...................................................................................21 Structure o f the Economy.............................................................................................................21 Preliminaries o f the Model: Firm Quantity and Profit............................................................24 Investment Gam e........................................................................................................................... 26 Affect o f an Import Tariff on the Capital Market Inefficiency....................................................... 31 Affect o f Lobbying on the Capital Market Inefficiency.................................................................35 Signal of Exogenously Determined C ost...................................................................................36 Special C ases..................................................................................................................................43 Signal of Endogenously Determined C ost............................................................................... 45 Concluding Remarks.............................................................................................................................. 49 References................................................................................................................................................ 51 Chapter 3: Income, Disamenity, and Toxic Releases in the United States............................................. 55 Chapter Abstract..................................................................................................................................... 56 Introduction and Literature R eview .....................................................................................................57 Theoretical Framework and Hypothesis Construction......................................................................63 Producers, Consumers, and Government..................................................................................63 The Release Level Determination Mechanism and Hypotheses........................................... 66 Technological Constraints............................................................................................................68 Control Variables..........................................................................................................................68 D ata........................................................................................................................................................... 69 Toxic Release D ata........................................................................................................................69 Socioeconomic D ata......................................................................................................................74 Results.......................................................................................................................................................77 Determination of Releases: Biochemical characteristics....................................................... 77 Determinants of Releases: Technological constraints.............................................................80 Control Variables.......................................................................................................................... 81 Determinants of Releases: T im e..................................................................................................83 iv R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Concluding Remarks.............................................................................................................................. 83 References................................................................................................................................................ 85 Chapter 4: Conclusion..................................................................................................................................... 88 Summary...................................................................................................................................................89 Future Research....................................................................................................................................... 89 Trade................................................................................................................................................ 89 Environment...................................................................................................................................90 References................................................................................................................................................ 92 Bibliography......................................................................................................................................................93 Appendices.................................................................................................................................................... 100 Appendix to Chapter 2 ....................................................................................................................... 101 Appendix to Chapter 3 ....................................................................................................................... 103 v R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. List of Tables Table 3-1 Chemicals............................................................................................................................ 71 Table 3-2 Socioeconomic Variables....................................................................................................76 Table 3-3 Fixed Effects, Feasible Generalized Least Squares Model of Toxic Releases (correcting for panel heteroscedasticity) 1987-1996 ...................................................... 78 Table 3-4 Income Turning Points........................................................................................................ 80 Table 3-5 Release Estimates over T im e.............................................................................................. 82 Table 3-3 Fixed Effects, Feasible Generalized Least Squares Model of Toxic Releases (correcting for panel heteroscedasticity) 1989-1996..................................................... 104 vi R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. List of Figures Figure 2-1 Game T ree............................................................................................................................ 24 Figure 2-2 Game Tree with Import Tariff........................................................................................... 32 Figure 2-3 Investment Decision facing Industry Manager............................................................... 34 Figure 2-3a Effect o f Import Tariff on Investment D ecision...........................................................35 Figure 2-4 Game Tree with Exogenous Lobbying C ost.................................................................. 37 Figure 2-5 Game Tree with Endogenous Lobbying C ost................................................................46 Figure 2-6 Sketch of f { K ) ................................................................................................................ 101 vii R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Abstract In this set o f papers we consider the use of information as a regulatory tool. The simplest economic models often assume an environment with perfect information. Relaxing this restrictive and unrealistic assumption can significantly change the results o f our models, making them more realistic, but often also illustrating the inefficiencies that may arise in the presence of informational imperfections. Here we investigate government’s potential role in information provision and its effect on market outcomes. We consider an industry facing intensified competition from imports that needs to retool its production process to restore its international competitiveness. If this investment project has a positive net present value (NPV), then efficient capital markets will provide financing and retooling occurs. When managers have more information about market conditions than shareholders, however, prudent investment may not occur. We show that a financially sound investment may not occur at equilibrium in a two country, perfectly competitive model if an asymmetry exists between the information sets o f shareholders and managers. In such a case, the act o f petitioning for protection by the industry (and protection itself) can facilitate investments in technological upgrading that would not otherwise occur. Petitioning for protection signals the quality o f the investment project while the tariff shifts rents to the domestic industry making the investment payoffs more favorable. Finally, we show that the equilibrium strategy o f the domestic government allows lobbying (signal), but responds with free trade policy. Building on the literature concerning the Environmental Kuznet’s curve, we develop and test a simple economic model o f toxic releases for counties within the United States. We use a twofold measure o f environmental disamenity associated with toxic releases: the level and toxicity of releases. We find that in the United States, a concave relationship between income and release levels exists for many chemicals reported in the US EPA’s Toxic Release Inventory and that the sensitivity of this relationship (i.e. the concavity) depends importantly on the toxicity o f the release. v iii R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Finally, we more fully specify the political and economic factors determining toxic releases. We find that ceteris paribus, more capital-intensive counties have higher releases and that county characteristics influencing voting behavior affect release levels. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Chapter 1 Introduction R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Overview In this set o f papers we consider the use of information as a regulatory tool. The simplest economic models often assume an environment with perfect information. Relaxing this restrictive and highly unrealistic assumption significantly changes the results o f many o f our models, making them more realistic, but often also illustrating the inefficiencies that may arise in the presence o f informational imperfections. In our first paper, we investigate how government regulation can reduce inefficiencies in capital markets that stem from informational asymmetries. Then, in the second paper, we use the data from the Toxic Release Inventory to investigate the effect that toxicity has on the income-pollution relationship. The Toxic Release Inventory is a database whose roots lie in the United States government’s attempt to provide consumers with more information about their neighborhood environmental quality. In both cases, governments, serving as an information provider, can affect economic and environmental outcomes. More specifically, in “Capital Market Inefficiencies: Signaling Through Trade Policy, A Second Best Solution” we address the role free trade policy “exceptions” can play in reducing inefficiencies in the capital market. While this chapter is largely theoretical, there are clear connections between the trade policies studied and trade policies currently in effect in the U.S. and other WTO countries. We suggest a novel role for industry import protection petitions within an overarching free trade policy, where petitions for protection act as a signal that may reduce the incidence of failures in the working of the capital market. Signaling games are plentiful in the economics literature where signals often bridge an informational asymmetry between agents. In the trade literature specifically, there are a number o f studies that examine the role for trade policy in improving welfare in the presence o f informational asymmetries (see, for example, Donnenfeld, Weber, and Ben-Zion, 1985; Mayer 1984, Grossman and Horn 1988; Bagwell and Staiger 1989; Raff and Kim 1999). Representative o f this literature, Bagwell and Staiger (1989) show that a role for government arises when a firm producing an 2 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. ‘experience’ good attempts to enter a foreign market. In the face o f incumbent firm reputations, entry and signaling of a firm’s product type may be facilitated with an appropriate export tax or subsidy policy. The trade literature also investigates the optimal trade policy in the presence of other market inefficiencies (see Newbery and Stiglitz, 1984; Eaton and Grossman, 1985; Dixit 1987, 1988, 1989). In an uncertain environment, insurance market inefficiencies may be reduced through trade policy. Through our model, we extend the trade literature to include a role for trade policy in reducing capital market imperfections arising from asymmetric information. Then, in “Income, Disamenity, and Toxic Chemical Releases in the United States” we investigate the relationship between per capita income and environmental quality (as measured by toxic chemical releases). We extend the environmental Kuznets curve (EKC) literature by studying the effect o f toxicity on the relationship between income and pollution. We begin by developing a simple utility theoretical model of the interaction between consumers, producers, and the government. Consumers respond both to their experiences o f environmental degradation and to information provided by the government on toxic releases. We illustrate that in response to disutility suffered from toxic chemical releases, consumers appeal to government officials for stricter environmental regulations. Government officials then set environmental regulations in order to maximize social welfare. These regulations are then part o f the producers’ production decisions. Our model estimation results demonstrate that the EKC holds for a variety of chemicals (as reported by in Toxic Release Inventory or TRI) and that the concavity of the EKC increases with the toxicity of the releases. Consumer activism in response to environmental quality changes (that they, at least in part, are made cognizant o f by government reports) may well drive the observed decline in toxic release levels. Our study is part o f a larger body o f environmental literature that focuses on the regulatory effect o f information provision. ‘Information as regulation’ is the latest in a line o f market-based mechanisms focused on environmental regulation. Information regarding the environment can affect 3 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. consumer and producer behavior in a number of ways. Green consumers prefer environmentally friendly products, but need credible evidence that products are relatively ‘clean’ before they will pay premiums for such products (Cason and Gangadharan, 2002). Also, environmentally friendly producers rely on the preferences o f green consumers and thus need a credible mechanism to differentiate themselves from other producers. Alternatively, ‘dirty’ producers may experience losses if their environmentally irresponsible behavior comes to light. Evidence suggests that producers identified as relatively dirty by the TRI experience negative abnormal stock returns (Hamilton, 1995). In response to these negative abnormal returns, Konar and Cohen (1997) demonstrate that ‘poor’ environmental performers improve significantly more than their industry peers. These results suggest that the provision o f information through the TRI may significantly affect environmental quality. In capital markets and the environment, information can serve a key regulatory role, improving welfare and reducing the incidence o f market inefficiencies. The remainder o f this chapter provides more detailed introductions to the dissertation contents. Capital Market Inefficiencies: Signaling Through Trade Policy, A Second Best Solution While trade throughout the world has become arguably “freer,” most trade agreements continue to contain provisions that allow member countries to impose trade barriers. The World Trade Organization (and the General Agreement on Tariffs and Trade before it), the European Union, the North American Free Trade Agreement, and numerous other trade agreements have succeeded in significantly lowering tariffs and other trade barriers. While contributing to the development o f an increasingly global economy, many o f these trade agreements also have built-in clauses that allow member countries to protect domestic industries in certain circumstances. As a result, member 4 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. countries can be in M l compliance with the trade agreements, but still engage in protectionist activities. The continued use o f tariffs and other trade barriers in the face of the welfare gains associated with free trade suggests that policies allowing trade barriers serve some additional purpose. “Capital Market Inefficiencies: Signaling Through Trade Policy, A Second Best Solution” studies the effect of trade policy on capital market inefficiencies. While capital market inefficiencies arise out o f a variety of sources, we focus on a problem arising from asymmetric information between participating parties. As is often the case, firm managers have greater knowledge regarding the true value o f the firm and potential firm investments than does the rest o f the market. The disparity in information cannot be easily overcome due to confidentiality, fiduciary obligations, and perhaps most importantly, credibility issues (i.e. firms’ best interests are always served by conveying that they and their investments are highly profitable). In this environment, we demonstrate that firm managers bypass some profitable investments (see also Myers and Majluf, 1984) - an inefficiency. We show that in the event o f a capital market inefficiency that the government may reduce the incidence o f the inefficiency by implementing two types o f trade policies. First, the government could reduce the inefficiency in the capital markets by allowing domestic industries to petition for trade protection (e.g. the “safeguard” clause o f the WTO). Second, the inefficiency could also be reduced if the government imposes tariffs on competing imports. Many countries including the United States have legislated trade policies similar to those outlined in chapter 2. In the U.S., domestic industries may petition the government for trade protection when they face intensified import competition. Section 201 o f the U.S. Trade Act o f 1974 allows protection for industries while they adjust their production process to improve their international competitiveness. Updating equipment and other retooling, however, requires considerable financial resources that industries may not have. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. The efficient markets hypothesis states that at any given time, security prices fully reflect all available information (see Fama, 1970 for a review o f the theoretical and empirical evidence on the efficient markets hypothesis). Thus, if a firm issues new stock to finance a profitable project (positive net present value, NPV), then the capital market will provide the required funding and the firm will retool. The price o f the firm’s stock will reflect the firm’s value based on the information available to the market. When the firm managers and the market have the same information, all positive NPV projects are undertaken. We demonstrate that a problem arises when markets and industry managers have asymmetric information sets and uncertainty exists regarding the exact magnitude o f positive returns. Managers know the true value o f the firm, but the market is only able to calculate its expected value and thus securities prices reflect the firm’s expected value. When the true firm value exceeds its expected value, managers reject the project to avoid a negative impact on securities prices (relative to their true values). As a result, some positive NPV projects are rejected. This rejection represents a failure (inefficiency) in the working of the capital market. Since the capital market inefficiency is not directly remediable, we consider a second best solution where industries signal the true level of positive returns. That is, we analyze policies that involve the firm undertaking a costly action that only the highly profitable firms would accept. When such policies are in place and firms signal their true level of profitability, we show that this reduces the incidence o f the inefficiency by separating the highly profitable investments from the less profitable ones. When the market is able to distinguish the profitability o f the investments, securities prices more accurately reflect the true value o f the firm. As a result, managers are less likely to reject profitable investments in an attempt to avoid negative stock price effects. We demonstrate that policies like those legislated in section 201 provide industries with the opportunity to credibly signal the level o f returns to the capital market. This signaling mechanism reduces the incidence of the inefficiency. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Protective tariffs also increase returns to the project and further reduce the incidence o f the inefficiency. By increasing the value o f the retooling project (by shifting profits of foreign firms to domestic firms), tariffs increase the value assigned to the firm’s stock by the capital market. The higher securities prices reduce the probability that the firm managers will reject the project. Tariffs, however, cause other market distortions (e.g. changing otherwise efficient consumption and production choices) making them a less preferred policy. We show that a domestic government that maximizes domestic social welfare will construct petitioning (i.e. signaling) opportunities for firms, but will respond to lobbying with a policy o f free trade. We further demonstrate that even if firms know that the government will never grant protection, they will still petition in order to receive the signaling value of the petition. Between 1975 and 1987, sixty petitions for protection under section 201 were brought before the U.S. International Trade Commission (ITC). Only 20 petitions, 33 percent, resulted in trade protection. In 47 percent o f cases when the ITC recommended protection, the President eventually denied or modified the request on the grounds that protection would not serve the “national economic interest.” With respect to section 201, it would appear that the government largely followed a policy o f free trade with infrequent exceptions. During this period, industries still petitioned for protection under section 201 although chances o f obtaining relief were small, suggesting other reasons for appealing to 201 (i.e. petitioning as a signal). The desire behind protection petitions to shift profits from foreign to domestic firms (commonly known as the rent-shifting motive) is also very strong. In recent years, industries have realized that the burden o f proof for other statues is considerably lower than those for section 201. Antidumping and countervailing duties petitions are far more common than 201 petitions due to their relative ease of proof and greater probability of resulting in relief. Section 201 petitions have dropped from 45 in the six-year period between 1975 and 1981, to only 10 from 1995-2001. Those selectively appealing to 201, however, are receiving relief at higher rates. Thus, while more 7 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. selectively, section 201 is still currently used and provides industries with the opportunity to signal their value and time to retool to effectively meet import competition. Chapter two demonstrates that section 201 (and corresponding clauses in trade agreements) may serve an economic purpose other than the obvious protectionist function. The model provides an economic rationale for clauses that allow countries to temporarily escape from commitments in their free trade agreements. We also pose an alternative, efficient view o f lobbying behavior. While our trade model does not address every case, it does pose a new explanation for the presence o f (temporary) exceptions to free trade and the value of lobbying. Income, Disamenity, and Toxic Chemical Releases in the United States Another issue related to the free trade policies embodied in the WTO and other trade agreements is the effect o f free trade on environmental quality. The issue o f trade and its effect on the environment has been hotly debated and has led, in part, to the large demonstrations against the WTO (e.g. Seattle, 1999). Trade theories based on resource endowments hypothesize that countries will specialize in the production of and export the good whose most intensive factor they have in relative abundance. Given the international differences in environmental regulations, such a theory suggests that “dirty” production (pollution-intensive) will locate in developing countries that are relatively abundant in the environmental input. To date no empirical support has been found for this “pollution haven” hypothesis. Studies, however, have found evidence o f a positive relationship between the scale o f economic activity and pollution levels. That is, as production levels rise so does the level o f pollution, holding all else constant. This grim relationship, however, is only part of the Environmental Kuznet’s Curve theory (EKC). The EKC suggests that while increased production increases pollution, this effect only holds up to some income level. Then as increased production continues to increase income, the demand for environmental quality eventually results in 8 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. regulations and cleaner technologies that lower pollution in spite o f continued growth. That is, countries can grow out o f pollution. The EKC has been estimated both at international (Grossman and Krueger 1993, 1995) and domestic levels (Brooks and Sethi 1997, Arora and Cason 1998) suggesting that to the extent that trade spurs economic growth this effect can occur at various levels o f aggregation. While the simple relationship embodied in the EKC is commonly accepted (and is demonstrated in a significant empirical literature), the exact form o f the EKC varies by aggregation level and pollutant (Cavlovic, Baker, Berrens, and Gawande, 2000). In “Income, Disamenity, and Toxic Chemical Releases in the United States” we further investigate the role o f the pollutant in determining the shape o f the EKC examining the effect of toxicity on the income-pollution relationship. The empirical analysis in this chapter is based upon a simple model o f releases. We demonstrate that toxicity significantly affects the level of releases both directly and indirectly through its affect on the income-pollution relationship. Theoretical explanations for the EKC phenomenon have largely focused on production-based explanations where economies grow their way out o f pollution. That is, initial increases in the scale o f production increase the level o f pollution. As the scale o f economic activity grows, so does income. Thus, like all normal goods, increases in income result in increased demand for environmental quality through cleaner production techniques and increased regulation that puts downward pressure on pollution levels that eventually dominate the scale effect. This “technique effect” stems from consumers’ desire to avoid the disutility they experience as a result o f pollution. We explore the role of disutility by examining the relationship between a chemical’s toxicity and the release levels o f the chemical. To investigate the effect of toxicity on the EKC, we use the Toxic Release Inventory (TRI) complied by the U.S. Environmental Protection Agency. The TRI provides release data for U.S. counties for 1987-1996. Chemicals reported in the TRI vary significantly in their toxicity. We use 21 chemicals from the full range o f toxicity levels contained in the TRI to 9 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. construct our dependent variable. The dependent variable is a simple transformation (ln(l+releases)) o f the total annual air releases o f each chemical in each U.S. county. In addition to income and a number o f control variables, we also include variables that we believe influence the disutility o f toxins (e.g. toxicity, exposure, and physical characteristics). We demonstrate that the toxicity, exposure, and physical characteristics o f chemicals significantly impact their release levels. Chemicals with higher toxicity have more concave Environmental Kuznet’s Curves (EKC). Carcinogens have lower releases than other chemicals. The exposure level o f the surrounding population also decreases releases. In earlier work on the relationship between income and pollution, Antweiler, Copeland, and Taylor (2001) show that freer international trade has led to lower ambient levels o f sulfur dioxide internationally. The reduction in SO2 levels results from the dominance o f the technique effect over the scale effect, where the scale and technique effects are similar to those embodied in the EKC. Whether freer trade affects other environmental quality measures similarly to S 0 2 is not known and there is reason to believe that it may not. Grossman and Krueger (1993, 1995) estimate the relationship between a variety of measures of environmental quality and income. While the relationships all exhibit the EKC, the income turning points (the point at which income growth leads to declining levels o f pollution) and concavity o f the EKCs vary. These results lead us to question the relationship between chemical toxicity and chemical release levels (and the effect on the observed EKCs). If, as we find, chemical toxicity affects the shape o f the EKC, then understanding the nature o f this relationship will enable us to also better understand the relationship between freer international trade and environmental quality. O f special interest are chemicals o f even higher order toxicity than sulfur dioxide, as these are pollutants that often have the most pronounced and immediate impacts on welfare. Should the EKC income turning point be at lower income levels or the EKC experience a less rapid increase (more rapid decrease) before (after) the income turning point for more toxic chemical releases, this would further support the argument for the benefits of 10 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. free trade on environmental quality. Toxic release levels would start to decline at lower levels of income as releases increase in toxicity. Great care must be taken when extending our results to other environments. Nonetheless, consider a thought experiment. Suppose our result (the EKC is more concave for more toxic chemicals) holds internationally and for all chemicals. If this is true, then Antweiler, et al. ’s results should hold even more strongly for more toxic chemicals than for S 0 2. As the toxicity of the chemical increases, the EKC will be more concave with the “technique” effect dominating at lower income levels. As a result, the income turning point should be even lower for more toxic chemicals. Such a result would imply a generally positive effect o f freer trade on environmental quality as the dominance o f the welfare improving technique effect would begin at lower income levels for more toxic chemicals. Information is a powerful regulatory tool. In this dissertation we illustrate two examples o f the use of information by the government to improve market outcomes by providing or facilitating the provision o f information. References Antweiler, Werner, Brian Copeland and Scott Taylor, 2001, “Is Free Trade Good for the Environment?,” American Economic Review, 91, 877-908. Arora, Seema and Timothy Cason, 1995, “An Experiment in Voluntary Environmental Regulation: Participation in EPA’s 33/50 Program,” Journal o f Environmental Economics and Management, 271-286. Bagwell, Kyle and Robert Staiger, 1988, “The Role of Export Subsidies when Product Quality is Unknown,” Journal o f International Economics, 27, 69-89. Brooks, Nancy and Rajiv Sethi, 1997, “The Distribution of Pollution: Community Characteristics and Exposure to Air Toxics,” Journal o f Environmental Economics and Management, 233- 250. Cason, Timothy and Lata Gangadharan, 2002, “Environmental Labeling and Incomplete Consumer Information in Laboratory Markets,” Journal o f Environmental Economics and Management, 43(1), 113-134. 11 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Cavlovic, T.A., K.H. Baker, R.P. Berrens, and K. Gawande, 2000, “A Meta-Analysis of Environmental Kuznets Curve Studies,” Agricultural Resource and Economic Review, 29(1), 32-42. Dixit, Avinash. “Trade and Insurance with Moral Hazard,” Journal o f international Economics, 1987, 23,201-220. Dixit, Avinash. “Trade and Insurance with Imperfectly Observed Outcomes,” Quarterly Journal o f Economics, 1989a, 104, 195-203. Dixit, Avinash. “Trade and Insurance with Adverse Selection,” Review o f Economics Studies, 1989b, 56, 235-247. Donnenfield, Shabtai, Shlomo Weber, and Uri Ben-Zion, 1985, “Import controls under imperfect information,” Journal o f International Economics, 19, 341-54. Eaton, Jonathan and Gene Grossman. “Tariffs as Insurance: Optimal Commercial Policy when Domestic Markets are Incomplete,” Canadian Journal o f Economics, 1985, 18(2), 258- 272. Fama, Eugene. “Efficient Capital Markets: Review of Theory and Empirical Work,” Journal o f Finance, 1970,25, 386-416. Grossman, Gene and Henrik Horn, 1988, “Infant-industry protection reconsidered: The case o f informational barriers to entry,” Quarterly Journal o f Economics, 103, 767-87. Grossman, Gene and Alan Krueger., 1993, “Environmental Impacts o f a North American Free Trade Agreement,” in P. Garber, ed., The U.S.-Mexico Free Trade Agreement, Cambridge, MA: MIT Press, 13-56. ., 1995, “Economic Growth and the Environment,” Quarterly Journal o f Economics, 353-377. Hamilton, James., 1995, “Pollution as News: Media and Stock Market Reactions to the Toxic Release Inventory Data,” Journal o f Environmental Economics and Management, 98-113. Konar, S. and M.A. Cohen, 1997, “Information as Regulation: The Effect o f Community Right to Know Laws on Toxic Emissions,” Journal o f Environmental Economics and Management, 32, 109-124. Mayer, Wolfgang, 1984, “The infant-export industry argument,” Canadian Journal o f Economics, 17, 249-69. Myers, S.C. and N.S. Majluf, 1984, “Corporate Financing and Investment Decisions when Firms have Information that Investors do not Have, ” Journal o f Financial Economics, 13, 187- 221 . Newbery, David and Joseph Stiglitz. “Pareto Interior Trade,” Review o f Economics Studies, 1984, 51,1-12. 1 2 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Raff, Horst and Young-Han Kim, 1999, “Optimal export policy in the presence o f informational barriers to entry and imperfect competition,” Journal o f International Economics, 49, 99- 123. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Chapter 2 Capital Market Inefficiencies: Signaling Through Trade Policy, A Second Best Solution R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Abstract Consider an industry facing intensified competition from imports that needs to restructure its production process or build new plant and equipment in order to restore its competitiveness. If this investment project has a positive net present value (NPV), efficient capital markets will provide financing and retooling occurs. However, when managers have more information about market conditions than shareholders prudent investment may not occur. In this paper, we show that a financially sound investment may not occur at equilibrium in a two country, perfectly competitive model if shareholders do not have the frill set o f information available to managers. If there are capital market imperfections such as these, the act o f petitioning for protection by the industry (as well as a protective tariff itself) can facilitate investments in technological upgrading that would not otherwise occur. Petitioning for protection signals the quality o f the investment project while the tariff shifts rents to the domestic industry making the investment payoffs more favorable. Both are second best in the presence o f a domestic distortion - a capital market imperfection. Finally, we show that the equilibrium strategy of the domestic government is to allow for the lobbying (signal), but to respond with a free trade policy. 15 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. I. Introduction In this paper we demonstrate that trade policy can act as a second best solution to a capital market inefficiency. We define capital market inefficiencies as the rejection o f a positive net present value project. The government policy we consider includes two possible instruments: permitting lobbying for protection, and protection in the form o f a tariff. First, the government may choose to allow domestic industries to petition for trade protection. Second, faced with protection petitions, the government chooses the level of protection to grant. We illustrate that for our simple perfectly competitive, two country model that both the lobbying behavior o f domestic industries and the trade protection provided by the government can reduce the incidence o f the capital market inefficiency. We further show that a domestic government that maximizes domestic social welfare chooses a policy that allows lobbying for protection, but responds with a policy of free trade. Firms petition for protection in this free trade environment to signal their value. Returns to the lobbying signal are sufficient to induce lobbying even when the rent-shifting motive is absent The government policies we consider in this paper are included in those outlined in section 201 of the 1974 United States Trade Act. Under section 201, domestic firms or industries1 may petition for protection from increased imports that are the “substantial cause” o f “serious injury” (or threat thereof). The trade code specifies conditions constituting “serious injury” and includes the case in which domestic industries are unable to generate sufficient capital to facilitate the modernization of their plant and equipment in order to effectively compete with intensified import competition. If an International Trade Commission (ITC) investigation finds support for injury claims, then the commission recommends remedies to the President who then makes the final relief decision. Section 201 is fully consistent with the GATT, WTO, and NAFTA.2 Thirty-eight other countries have 1 Given the nature of the trade code, we allow for the domestic entity to be either a firm or the entire industry. We use the terms interchangeably in this paper. 2 The corresponding sections o f the GATT, WTO, and NAFTA are Article 19, the WTO Agreement on Safeguards, and Section 302 respectively. Only five safeguard remedies have come before the 16 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. specific safeguard legislation and eleven others use the WTO Agreement as their guiding authority. Seventeen o f these countries have undertaken safeguard investigations since 1995 for a total o f 50 cases, ten o f which were in the US. The history o f Harley-Davidson Inc. provides anecdotal evidence o f the potential effect of safeguard remedies. In the early 1980s Harley-Davidson faced significant competition from Japanese producers and experienced a significant decline in its market share. In 1983 President Reagan, on the recommendation o f the ITC, increased tariffs on imports of large Japanese motorcycles (from 4.4 to 49.4 percent). Under this protection, Harley-Davidson instituted manufacturing changes and engaged in other operational retooling. By 1987, Harley-Davidson had significantly reduced costs, increased the effectiveness o f its marketing, and increased stockholder equity. As a result, they requested that tariffs on Japanese bikes be removed a year earlier than scheduled. Suppose Harley-Davidson had approached the capital market for funding in 1982 to finance the revamping of their production processes. When capital markets function as described by the efficient market hypothesis (for a review o f the theoretical and empirical literature on the efficient markets hypothesis see Fama, 1970), government intervention disrupts the smooth working o f the market. In a laissez-faire environment, the expectation o f future profits is sufficient to induce the market to supply all capital necessary to finance investments. Similarly, any investment in capital and equipment with a negative expected net present value (NPV) is rejected. Facing intensified Japanese competition, Harley-Davidson was forced to modernize its plant and equipment to remain competitive. If we assume that the modernization project has a positive net present value with certainty, then the market should supply the needed capital. Government intervention is inefficient when capital markets function perfectly. Dispute Settlement Body (DSB) o f the WTO, three o f which were brought against the US. All five cases were eventually decided in favor o f the complainant. 17 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Now suppose that capital markets operate in an environment o f uncertainty and asymmetric information.3 All participants know the proposed project has a positive NPV, yet there exists uncertainty with regard to the exact return level. Further, we assume that firm managers have superior information to the rest o f the market. This informational asymmetry as modeled by Myers and Majluf (1984) induces the rejection o f some profitable investments, and the domestic industry may not undertake the necessary modernization to remain internationally competitive, i.e. Harley- Davidson would not modernize its plant and equipment and thus would continue to suffer a production efficiency disadvantage. We demonstrate the conditions for this scenario in our perfectly competitive, two country, one good model where the domestic country is a net importer o f the good. Then we investigate the effects o f government intervention under the inefficiency conditions. Governments have a variety of instruments with which to address market failures. For our purposes, we concentrate on a small subset o f trade policy instruments. Our results indicate that both protection and lobbying for protection can reduce the occurrence of the market failure. Lobbying without the granting of protection, however, improves overall social welfare. Whether protectionist policies are or are not optimal from a societal view, they are certainly desirable from a domestic industry’s viewpoint. The lobbying behavior o f industries provides sufficient evidence. Rent seeking behavior, however, is generally considered as socially undesirable and is dubbed as ‘DUP’ - Directly Unproductive Profit-seeking - activity by Bhagwati (1982). Bhagwati’s general reasoning is that industries expend scarce resources in lobbying to obtain protection and increased rents.4 An expenditure o f resources in this manner is wasteful and reduces social welfare. 3 While capital markets in developed countries are generally very high functioning, they (like capital markets in developing areas) are nonetheless subject to uncertainty and informational asymmetries. Thus, such an environment is plausible in all countries. 4 Rents need not be monopoly rents. Feinberg and Hirsch (1989) demonstrate that protection of capital and labor rents are more important than the protection o f monopoly rents in the decision to file antidumping petitions. 18 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. In this paper, however, we show that the lobbying behavior itself can reduce the occurrence of the inefficiency arising from the asymmetry of information between agents. Further, we show that when the improvement to social welfare from the reduction of the inefficiency is sufficiently large that it can offset the loss from the resources used in lobbying. Petitioning the government for import protection can result in a Pareto improvement instead o f its usual ‘DUP’ role. This result is especially clear since we show that industry managers will petition for trade protection even if they know that the domestic government follows a policy o f free trade with certainty. That is, managers petition as a signal to the capital market, not as a vehicle for rent shifting. This result is the main contribution o f this paper. In addition to demonstrating the potential Pareto improvement from lobbying for import protection, we also show that the granting o f protection is inferior to free trade even though it reduces the incidence of the capital market failure. Until now, the literature has not addressed the effect o f a tariff in the presence of a capital market failure. As such, our results on this point constitute a second contribution of our work. The literature does, however, address the potential role for trade policy to improve social welfare in the presence of asymmetric information. Much of this literature focuses on ‘experience goods’ for which incumbent producers have a reputational advantage over potential entrants. Bagwell and Staiger (1989) demonstrate that export subsidies can improve social welfare when they assist high quality producers in entering a new foreign market. In a model similar to Bagwell and Staiger, but with the additional problems o f moral hazard and adverse selection, Grossman and Horn (1988) show that trade protection is welfare reducing. Depending on the specifics o f the model, protectionist trade policy may or may not improve welfare (other papers include Donnenfeld, Weber, and Ben-Zion, 1985; Mayer, 1984; and Raff and Kim 1999). The trade literature also addresses the effect o f a tariff when a failure arises in the insurance market. Several different approaches drive the varying results in the literature and consideration o f 19 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. these results informs our current work. First, Newbery and Stiglitz (1984) develop a perfectly competitive, two-country model both producing a risky and a safe good. The output o f the risky good varies with known probability and the variance across countries is not perfectly correlated. The assumption driving their model is that there are no insurance markets by which producers can shield themselves from uncertainty. Newbery and Stiglitz demonstrate that under free trade, the producers o f the risky good are subject to greater income variance than under autarky. Producers are necessarily worse off under free trade. As a result, producers shift production away from the risky good, raising its price. Producers’ actions have dual and opposing effects on consumers. Free trade reduces the risk faced by consumers but also reduces their surplus; autarky has the opposite affect. When the value o f the reduced consumer risk is dominated by reduced consumer surplus, consumers also find free trade to be inferior to autarky. Like Newbery and Stiglitz, we find that protection reduces the incidence of the inefficiency by adjusting agents’ payoffs. In our model, however, the benefits from reduction o f the market failure do not dominate the costs. Second, Eaton and Grossman (1985) show that a small country subject to uncertain terms o f trade may improve domestic welfare by following protectionist rather than free trade policy. Several assumptions drive their results: all agents in the economy are equally uncertain as to the outcome o f the risky venture, and insurance markets are not simply ‘incomplete’ but entirely absent. Our model is not based on either of these assumptions. We assume asymmetric information between agents and a functioning risk-sharing market (here the capital market). Eaton and Grossman note that asymmetric information would reduce their demonstrated inefficiency and Dixit’s later work illustrates the effect of relaxing the absent insurance market assumption. Dixit (1987, 1989a, 1989b) demonstrates that when the assumption of absent risk-sharing markets is weakened to allow partial insurance markets that protection of the risky sector is no longer welfare improving. Dixit further shows that the optimal trade policy is one that disfavors the risky sector. 20 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. The market failure in our model arises from the asymmetry of information between groups of agents. This differs from the current literature that investigates market failures stemming directly from uncertainty. Our approach informs the reader to the effects o f lobbying and protection in the presence o f both asymmetric information and uncertainty. We also contribute to the literature by studying the capital market (in addition to insurance markets) as a risk-sharing market in this international trade environment. The key results o f our paper - the efficiency improving effects o f both lobbying for protection and protection itself - as well as the associated social welfare effects, are developed in the remainder o f the paper as follows. In section two we present the structure of the economy and develop the conditions under which there exists the inefficiency in the capital market. Sections three and four consider the government policies o f protectionism and lobbying costs (for protection), respectively. Finally we summarize our findings and make our concluding remarks in section five. n. Illustration of the Capital Market Inefficiency Structure of the Economy Consider an internationally produced good. In the home (or domestic) country, the good is produced and sold at the prevailing world price. Domestic firms are price takers. The domestic country demands a larger quantity o f the good at the world price than the domestic industry is willing to produce at this price, making the home country a net importer o f the good. Imagine a scenario where the foreign industry experiences a positive technological shock. As a result, the domestic industry finds itself at a technological disadvantage; its marginal cost is greater than its competitors’ rendering it less competitive. The domestic industry can remove the technological handicap through an investment at a sizable fixed cost ( / ) . The domestic industry possesses no financial slack where financial slack is defined as the amount o f cash, marketable securities, and risk 21 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. free borrowing available to the firm. Thus, in order to finance the all or nothing investment the firm must issue new shares of stock. We restrict our study to include only those investments with a positive NPV for this is the only case where the potential for a capital market failure (as we have defined it) arises. We further assume that all players in the market have free access to information (including the probability distribution over possible states of nature) and they can accurately compute the expected NPV of the investment. In the event o f a new equity issue, new stockholders purchase all new shares and the holdings o f the original stockholders are unchanged. Finally, we initially assume a laissez-faire government; the tariff rate is zero and there are no lobbying opportunities for the domestic industry. There are two sources o f uncertainty in the model. First, worldwide demand is subject to stochastic shocks causing shifts in the world price. Second, the effectiveness of the investment is uncertain. The proposed modernization of industry plant and equipment decreases the marginal cost o f the domestic industry’s production, but the magnitude o f the decrease is not known with certainty. Consequently, the profits o f the firm and returns to the investment are uncertain. In this paper we focus on the case where the shocks are perfectly correlated. We further reduce the possible outcomes by assuming that the world price and investment effectiveness are each realized at one of two levels - high (H ) or low (L). These assumptions focus our study on the best and worst case scenarios (i.e. high world price and highly effective investment versus low world price and minimally effective investment), but allow our conclusions to encompass the full range of possibilities associated with any level of correlation between stocks. Additionally, we impose an asymmetry between the information sets o f domestic industry managers and all other players (see Myers and Majluf, 1984).5 Managers often posses more 5 The problem o f asymmetric information is equally applicable to developing and highly developed capital markets. Jacobson and Aaker (1993) provide evidence suggesting that the U.S. stock market suffers from asymmetric information to a greater degree than the Japanese stock market. The issue 22 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. information regarding a project’s viability, expected return, and risk exposure than do outsiders. We assume the managers o f the domestic industry are informed as to the true state o f demand and the effectiveness o f the investment one period prior to other players. Upon the receipt o f this information, managers make their investment decision. If the managers decide to invest, then they issue new stock and the capital market provides the necessary funding, where it prices the stock based on its beliefs regarding the state o f nature and the actions o f the industry managers. Depending on the true state o f nature, the capital market’s value may represent either a loss or a gain to the current stockholders o f the domestic industry. We assume that the managers have a fiduciary duty to the shareholders - managers attempt to maximize profits, but do not act in a manner that harms current stockholders (i.e. lowers the stock value). Such an assumption is plausible since managers are often stockholders themselves and are thus acting to maximize their own returns. Alternatively, managers may be seen as behaving in the best interest o f the original stockholders in order to maintain their current positions. Either o f these assumptions is consistent with models o f management behavior that have shown that in the presence of asymmetric information managers have incentives to artificially inflate current term results even if this comes at the expense of longer term profitability so as to enhance their current stock price (Stein, 1989). The tension between managerial obligations and market valuation created by the informational asymmetry may result in the rejection o f some positive NPV investments. Discounting o f share value by the market, in some circumstances, leaves original shareholders worse off than if the firm bypassed the investment. Thus, for certain parameterizations of the model, management foregoes positive NPV projects due to the negative effect the required issue o f new equity has on the existing shareholders. Rejection o f the investment protects the current stockholders from dilution of their stock price at the cost o f the domestic industry continuing under its technological handicap. is more the distance between investors and managers than the malfunctioning o f the market in a more basic sense. 23 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. The informational asymmetry produces inefficient investment decisions. The model is illustrated in the game tree in Figure 2-1. First, the state o f nature is determined. Second, the industry makes its investment decision. If it invests, then the capital market provides the requested capital and reevaluates the value o f the industry’s equity. Finally, the industry selects its level o f output and payoffs are determined and distributed. Figure 2-1. Game Tree Nature State Domestic Industry Investment Decision Capital Market Equity Valuation Domestic Industry Output Payoffs Profits New Equity Value Not Invest Invest Invest Not Invest 7 1 L I I n l o 0 n h i I n h o 0 Preliminaries of the Model: Firm Quantity Decision and Profit We begin by establishing the levels o f output and the profit realized under each state of nature. World demand experiences stochastic shocks that cause fluctuations in the world price. For simplicity, we only consider world demand indirectly through world price. The world price is either at a high level ( p * H) with probability 6 or a low level ( p ™ ) with probability (1 — 8 ) . Domestic 24 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. demand for the good is constant, unperturbed by outside fluctuations. Preserving as much generality domestic consumers are unable to differentiate between imported and domestically produced goods. Their choice o f product is based solely on price. On the supply side, the domestic industry is a price taker. Their output decision does not affect world price and they can sell as many units as they wish at the prevailing price level. Absent government intervention, the domestic price for the good is the same as the world price. In subsequent sections, we consider the affect o f protectionist trade policy (in the form of a tariff) on the market. Protectionist trade policies raise the domestic price to the sum o f the world price and tariff, i.e. p f = p™ + t , where t is the tariff and i is the state o f nature ( / = L , H ). Following standard assumptions, the domestic industry maximizes profits through its quantity choice. The domestic industry incurs costs and has a production function represented by the variable cost function Cy { q { p ) ) , where j indicates the investment decision ( j = 0 , / , no investment and investment respectively). Since we assume shocks to the world price and investment effectiveness are perfectly correlated, when the investment is undertaken (_/ = / ) , it has effectiveness level i . Standard assumptions apply to the curvature of the variable cost function, as possible, we represent domestic demand by the function q d i p \ , which conforms to generally accepted curvature constraints, i.e. q d (/>) < 0 and q d (/?) < 0 . Further, we assume that Further, we assume that < 0 . The domestic industry’s supply function is derived by implicitly solving the first order condition of the industry’s objective function, m a x ? p q — Cy ( q ) . The domestic supply is 25 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. qt ( p ) where qs m (p) > qs L I (p) > qs H 0 (p ) > qs L Q (p) for all p , q*j (p)> 0 , and Given the demand and supply conditions, the industry is more profitable in both states of nature when it makes the investment. The profit earned by the industry rises from 7 Z i0(p) to nu(p) as a result of the investment. We define the difference between the two values as the net present value of the investment in state i , NPVi (pj = ^,/(p) — 7T;o(p) ■ Substituting the primitives of the model and given the demand and supply conditions, the net present value o f the investment is positive for all possible initial values. Lemma 1: N P V ^ p ) = ^ ( p ) — 7 T ,0(jt>) > 0 for i = L , H . Proof. Writing the NPV from lemma 1 as a function of its primitives, we get NPVt [p] = p t [qu (p) - qi0 (p)] + [c,0 (qi0 (p)) - ca (qu (/?))]. By definition, each component P i • Qh { p ) ~ Qio ip)> and c w (?,o ( p ) ) ~ c n (flu ( i7)) 7 S greater than zero and thus the NPV o f the investment in each state is greater than zero. In this model, investment always yields a positive return and thus should always be undertaken. Investing is part of any efficient strategy set for the domestic industry. Investment Game The true value o f the domestic industry depends on demand and supply conditions and the industry’s investment decision. For simplicity we assume that the industry’s current plant and equipment has no liquidation value. If no new investment takes place, the value o f the firm and thus 26 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. the value o f the asset-in-place is the profit the plant and equipment generates, i.e. 7 1 i0. If the stock issue and investment are made, the industry spends / on retooling its plant and equipment. The physical investment is assumed to have a liquidation value o f I at the end of the period.6 With the investment, the true value o f the firm is 7lu + I , i — L, H . The manager’s investment decision (j), however, is based only in part on the true value of the domestic industry with and without investment. In deciding whether to invest, the manager must also consider the effect that the issuance o f new equity (to fund the investment) will have on the value to the original stockholders. By issuing new equity, the ownership o f the original stockholders is diluted. Instead o f owning 100 percent o f the industry, the original stockholders own the reduced P fraction , where P is the market’s valuation o f the original stockholders’ equity. Thus, the P + I value o f the original stockholders’ holdings is --------- (tt,, + l ) when the investment is undertaken. p + r ' It is this value to which the industry managers compare the no investment option. As such the investment decision depends in part on the market’s valuation o f the industry’s equity. The capital market behaves in accordance with the efficient markets hypothesis when valuing the domestic industry, P . Upon observing the manager’s investment decision, the capital market updates its prior beliefs regarding the state o f nature from the original distribution W ( i = H with probability 6 and i — L with probability 1 — 6 ) to the posterior distribution £»(• |y ) using Bayes’ rule. Given the managers’ objective, we know that managers always invest in state L as it is in the original stockholders’ best interest to do so. That is, if they invest, the worst-case scenario is that the market accurately deduces that state L has occurred and the industry is valued at its true value. 6 No further loss of generality is implied - any liquidation value here would preserve the results. 27 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Alternatively, the capital market may either mistake the state for state H or value the industry at its expected value. Both possibilities clearly yield a higher return to the current stockholders than an accurate valuation by the market. All three possible scenarios leave the original stockholders with a higher value than had the managers rejected the investment, which would earn the original stockholders 71L0 . The question then is what behavior the managers will follow in state H . In games o f this type where players only take on pure strategies, two types o f equilibrium exist - separating and pooling. First consider the potential separating equilibrium. In order for the separating equilibrium to be a perfect Bayesian equilibrium (PBE), the separating equilibrium must be a set of strategies and beliefs for which, at any stage o f the game, players’ strategies are optimal given beliefs and the beliefs are derived from equilibrium strategies and observed actions using Bayes’ rule. Suppose that the capital market prior to industry’s investment action believes that industry always invests in state L and that it never invests in state H . Upon observing the industry’s investment decision, the capital market updates its beliefs about state i and bases its value o f the industry on its posterior distribution tw(- |j). The separating equilibrium exists when industry only invests in state H . Thus, a necessary condition for the separating equilibrium is that the original stockholders are better off not investing in state H (earning original stockholders 71H0) than investing and being valued at the industry’s expected value (earning original stockholders Suppose that the condition for the separating equilibrium is satisfied and consider the following {n H 1 + /)). That is, the separating equilibrium exists when strategies and beliefs. In state H industry does not invest and the capital market correctly infers 28 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. that state H has occurred and values the industry stock at 71H Q . Alternately, state L occurs and industry invests and the capital market correctly infers that the industry is in state L and values the stock as 71L1. Clearly, these strategies and beliefs form a PBE, so 9nm +(l-6)nu ( \ ------------1-------\------------y^rn + J — ^ ho * s a necessary and sufficient condition for the separating e7tHI+{l-0)nu +I equilibrium. In a pooling equilibrium both types of industry invest, so the capital market’s posterior beliefs regarding the state o f nature are the same as their prior when investing is observed. Since failing to differentiate itself from an industry in state L is costly to industries in state H , they will invest only if the investment is sufficiently profitable. As such, a condition for the existence o f a pooling equilibrium is that investing by industries in state H (where they are undifferentiated from state L ) is more advantageous to the original stockholders than not investing and being recognized as an industry in state H . That is, the pooling equilibrium exists when { +l)>rr Conversely, assume that the condition for the pooling equilibrium is satisfied and consider the following strategies and beliefs. In both states the industry invests. The capital market has posterior beliefs 6)(i = H \ j = 1^ = 0 and 6)(i = H \ j = o ) = 1. In state H , industry would earn d7im + \ \ - 6 p i n / r\ ------------T — \ — 1 71 hi + ■ * J fr°m investing and n H0 from forgoing the investment. Thus, if 0KHi+{l-0ptL i+ I the condition for the separating equilibrium is violated, the proposed strategies and beliefs form a pooling equilibrium. 29 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 071 rrt + ( l - 6 VZ„ / \ Proposition 1: I f ------------- r----------- 17tm + 1 )> 71HQ, then the perfect Bayesian equilibrium 071 H! + ( 1 - 0 ) 7 t LI + 1 is a pooling equilibrium with both types undertaking the proposed investment. From Lemma 1, investing is always an efficient strategy. The condition for managers to invest in retooling is —— ~ ( f tu + / ) > 7tj0 . When the new equity shares are issued, the capital market revalues the industry’s equity and provides the funding for the investment. The managers then select a quantity strategy. Managers produce the quantity o f the good that maximizes profits. Such a strategy is also efficient. Thus, the efficient strategy pair for the domestic industry is , q ; ; { p f ) , i = L ,H and results in a payoff of —— — (n a + / ) to the incumbent shareholders. Suppose the domestic industry always invests in retooling, i.e. j = I . In order for the efficient strategy pair to be part o f a perfect Bayesian equilibrium, industry managers must be behaving optimally under this strategy, i.e. { j , q l { p ) ) = { h q S u { p ) ) j = L ,H maximizes the value to the original stockholders. As we have seen, this strategy set is part o f a pooling equilibrium, and thus the condition for the efficient equilibrium is the same as for the pooling equilibrium. Proposition 2: In state H i f --------—-7— — ^ u— (ft hj + / ) > 7tH0, then the efficient pair o f 07tH i + { \ - 0 } 7 t u + I strategies is part o f a perfect Bayesian equilibrium. The case in which the equilibrium fails does so when the above condition is not satisfied; the positive NPV project is rejected. Managers know the true value o f the industry and this serves as a reservation price. The reservation price may exceed the market assigned (or expected) value when 30 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. the state of nature is the good (or high) state. The capital market inefficiency (rejection o f positive NPV projects) results from the discounting o f shares by the capital market and the asymmetry of information between the capital market and industry managers. Industry interests are always served by signaling that state H has occurred. Without some intervention by the government, the industry has difficulty credibly signaling the true state o f nature. Direct correction of the inefficiency is not a viable option. Other than refraining from any action and allowing the inefficiency, the only alternative is to implement a second best solution. In the remainder o f the paper we consider the affects o f trade policies on the inefficiency and social welfare. III. Affect of an Import Tariff on the Capital Market Inefficiency Consider a domestic government that provides domestic industries with protection from foreign competition. The government imposes a per unit tariff on imports competing with the good produced by the domestic industry. The game tree in Figure 2-2 updates the original game tree from Figure 2-1 to include the government. Initially, nature determines the state. Then unaware o f the actual state the government selects the tariff level. Upon observing the protection level, the domestic industry makes its investment (if it invests, the capital market provides the necessary capital and revalues industry stock) and production quantity decisions in turn. Although the government sets the tariff level at its choice node, we reserve discussion o f the government’s objective function and decision process until later. In section four, we model the government’s strategy set where its protectionist strategy depends on the strategies o f the industry. For now, we restrict our attention to exploring the affect that protectionist policies have on the domestic industry’s investment decision. 31 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Figure 2-2. Game Tree with Import Tariff Nature State Government Tariff Domestic Industry Investment Decision Capital Market Equity Valuation Domestic Industry Output Payoffs Profits New Equity Value Social Welfare High Low Invest Not Invest Not Invest Invest, XX XX XX XX n HI ( t ) n u (t) I Gu (t) 0 Gl o (0 I Gh i (t) 0 Gh o (t) The industry’s investment decision is a function of the price, and thus any protectionist measure that affects the price. With this in mind, we rewrite the condition for the pooling Pit) equilibrium as a function o f the tariff, —-rr------if f u (?) + / ) < 7ti0{ t) , where t represents the per P (t) + I unit import tariff, / is the fixed cost of the investment, Tty (?) the industry profits in state i with investment decision j , and P { t) the market value o f the industry conditional on the industry’s 32 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. investment decision. After some manipulation the condition reduces to (N P V { t) + / ) < —-pr (/)• By considering the specifications that these general functions r \ t ) represent we derive the following relationships between the tariff and components o f this condition. d p * n The domestic price for the good increases with the tariff ( --------> 0 ). Further, assuming that the d t domestic industry selects output in a competitive manner (i.e. marginal cost o f production equals the domestic price), the profits o f the domestic industry also increase with the tariff. Finally, given the competitive output assumption and the curvature o f the domestic supply curve, the net present value o f the investment is also an increasing function o f the tariff. A graphical representation of the investment decision condition characterizes industry conditions by the value to the original stockholders without investment and net present value o f the investment (see also Myers and Majluf, 1984). The depiction creates two regions - one where investment is undertaken and one where it is rejected. Figure 2-3 depicts the investment decision facing the industry manager. Point e is an example o f a coordinate set (initial value without investment, net present value) that results in the separating equilibrium strategy described in section 2 o f this paper. 33 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Figure 2-3. Investment Decision facing Industry Manager Net present value o f Investment Invest region Reject region Value to Original Stockholders without Investment Suppose the domestic industry is initially at point e, when the government adopts a protectionist trade policy. Two changes occur in Figure 2-3 and are illustrated in Figure 2-3a. First, the initial point e is no longer the point that describes the investment opportunity. Both the value to the original stockholders (without investment) and the net present value o f the investment increase. This moves the domestic industry’s graphical location in a northeasterly direction. Depending on the relative magnitudes o f the value adjustments, such a move could shift the domestic industry to a dP(t) . point in the investment region. Second, the investment decision line rotates downward ( ----------> (J, dt the slope o f the investment decision line decreases and its point o f intersection with the horizontal axis increases). This effect increases the size o f the ‘invest region’. The joint effects o f the tariff on the investment decision reduce the incidence o f the capital market inefficiency. Although other 34 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. distortions are created with the tariff (and these must be considered prior to recommending a protectionist trade policy) it positively affects the capital market inefficiency illustrated in this paper. Figure 2-3a. Effect of Import Tariff on Investment Decision Net present value of investment Invest region Increased investment region, i.e. reduced inefficiency Reject region Value to Original Stockholders without Investment Proposition 3: Protectionist trade policy in the form o f a tariff reduces the incidence o f the capital market inefficiency. IV. Affect of Lobbying on the Capital Market Inefficiency The inefficiency in the capital market is the result of the asymmetry o f information between the market and the managers o f the domestic industry. The first-best solution to the inefficiency is to bring the information sets o f the parties into parity. Implementation o f such a solution would be difficult for several reasons including, fiduciary conflicts, proprietary information, and the cost o f 35 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. disseminating information. Additionally, we must consider the credibility of information provided by the manager. Regardless o f the true state o f nature, managers have an incentive to indicate the H state to the market in order to boost the value of the original stockholders’ shares. In this section, we consider whether the government can provide a framework that allows managers to signal the state of nature in a credible manner. Signal of Exogenously Determined Cost Suppose the domestic government provides a format where by the domestic industry can take a perfectly observable action, the cost o f which would only be borne by managers in state H ? This format would allow managers to credibly signal the state. One possible framework allows managers to petition for trade protection (e.g. a tariff). Lobbying the government imposes costs on the domestic industry in terms o f time and financial resources (including petition filing, attorneys, and other consulting fees). We assume the lobbying cost ( K ) is determined exogenously and its magnitude is common knowledge.8 We add the lobbying strategy to the firm’s choice set as follows. As before, the true state is determined by nature and revealed to the managers. At this point, the managers decide whether or not to petition for import protection. If the managers lobby, then the government selects a tariff level. For now we assume that filing results in protection ( t > 0 ) with probability /i and in no protection ( t = 0 ) with probability 1 - ^ . If the firm does not lobby, then the government does not consider it for a protection 7 The domestic industry could potentially signal its type through a variety of costly activities. We do not assert that the lobbying activity that we propose here is the only such activity that could serve as a credible signal, but is simply one o f several possibilities. 8 For now we reserve discussion o f the affect such a framework would have on social welfare. This issue is addresses in the next section where we study the government’s objective function. 36 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 2-4. Game Tree with Exogenous Lobbying Cost Nature State Domestic Industry Lobbying Decision Government Tariff Domestic Industry Investment Decision Capital Market Equity Valuation Domestic ] Output Payoffs Profits New Equity Value Social Welfare Not Lobby Not Lobby Not Invest Invest Not Invest Invest Not Invest Invest Invest A q |k 4 | | 4 | | 4 | | 4 | | 4 | | q | | q Industry ^ ^ ^ ^ ^ ^ ^ ^ n li (0) I Gu (0) nu> (0) 0 Gl o (0) n li (t) / Gu (t) 7 1 L 0 (t) 0 Gl . (t) 7 1 HI (t) I G hi (t) 7 1 H O (t) 0 G ho (t) * H I (0 ) / G „ i (0 ) 7 1 H O (0 ) 0 G„0 (0) U ) award (i.e. t = 0 with certainty). The remainder o f the tree follows the game tree from Figure 2-1 where managers make their investment decision and the market provides capital revaluing the industry’s equity conditional on the industry’s actions (in particular now the lobbying decision o f the industry which serves as a signal of the state o f nature). The manager’s investment decision is followed by their quantity decision and the payoffs are determined and distributed. The revised game tree is presented in Figure 2-4. The manager’s strategy set now includes their lobbying strategy, ( / where / indicates the lobbying strategy ( f = F , N F representing lobbying and no lobbying respectively), j the investment decision, and (/?) the output level. The lobbying strategy o f the domestic industry depends on the cost of lobbying (K), the expected benefit (tariff rents), and the affect o f the signal sent to the market by lobbying. When a firm lobbies for protection, the cost of filing is reflected as a lump sum reduction in their profits. Lobbying also increases the probability of protection (from zero to ju ) increasing expected profits. Given these additional considerations we update the payoffs to the industry’s investment and lobbying decisions to determine the manager’s equilibrium strategy set. Absent lobbying and investment, the value realized by the original stockholders remains the profit in state i with a zero tariff, i.e. nj0( 0) . The industry may choose not to lobby but to invest, in which case the payoffs reduce to those from section 2, P —— — (/r;/ (o )+ / ) . Alternatively, the industry may file but not invest, and the original stockholders value reflects the potential gains from protection and the cost o f lobbying, 38 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. {pin i0 (/)+ (l — ju )n i0 (o))— K . Finally, the domestic industry lobbies and invests, and the value P — K to the original stockholders i s ---------------- ((,u n a P — K + 1 From section 2, we know that any efficient strategy set includes investment by the domestic industry. For the lobbying strategy to be efficient, it must induce the pooling equilibrium indicated in section 2, i.e. Two sets meet this efficiency requirement. The first is the trivial case where the capital market inefficiency does not exist, \N F ,I,q’ w (p)) . \ . Petitioning for protection in this case falls into the category (N F J .q p p j) ofD U Ps - directly unproductive profit-seeking activities - because lobbying serves no purpose other than to shift rents. We assume away this trivial case by requiring that if the industry lobbies for protection that it also invests (which is a condition set forth in the safeguard clause). The second 6nm + {l-6\iu / \ case is that in which the inefficiency exists ( ------------,------------------- 17tm + * ) > 71 ho ) ™ e n m ^ - o y u + 1 lobbying allows the market to distinguish the true state. Consider the case where only H state managers lobby for protection. If the market believes that managers only lobby in state H , then lobbying for protection signals that the realized state is H . An issue o f new stock (after lobbying) by the domestic industry would result in the stock being valued at the H level. Alternatively, failure to lobby signals to the market that state L has occurred and the market values the domestic The value for P in this section reflects the possibility o f protection. hi (o ) , if the market believes the investment is o f H return L m ,,\t) + (\ — ju)7rT A v ) ,i f the market believes the investment is o f L return + (l — 0 \ j l K u (f)+ (l - fXprijj (o)] , if the market values the firm at its expected value 39 P = R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. industry at the L level. Investing is a dominant strategy in state L , so again stock is issued and the firm invests. Thus, when the initial inefficiency existence condition is satisfied, the efficient strategy '{f . i . i ’ M ) set for the domestic industry is ~ (NF,I,q-u (p)) We now consider whether the efficient strategy set is part o f a perfect Bayesian equilibrium. Again we consider the possible equilibria, this time in lobbying. Suppose that the condition for a separating equilibrium in investing from section 2 is satisfied. Further suppose that the domestic industry lobbies for protection only if state H occurs. For the efficient strategy to be a part o f a perfect Bayesian equilibrium strategy managers must be behaving optimally when following the strategy (the efficient strategy maximizes the value to the original stockholders) given the capital market’s beliefs, which are derived from equilibrium strategies and observed actions updating using Bayes’ rule. Now with lobbying as the signal, the efficient strategy set is a perfect Bayesian equilibrium strategy if and only if the act of lobbying accurately signals the true state (thus differentiated, investment will be undertaken in both states). That is, for the signal to be effective, it must induce separating in lobbying and pooling in investing. Suppose that the capital market, prior to industry’s lobbying action, believes that industry will only lobby if state H occurs given the cost o f lobbying and the managers’ objective function. Upon observing the industry’s lobbying action, the capital market updates its prior using Bayes’ rule. The separating equilibrium in lobbying exists when industry lobbies only in state H . A necessary condition for separating in lobbying is that industry is better off lobbying in state H (and earning P — K „ r((pxm( 0 + 0 - p V hi(0)) — K + 1) ) than not lobbying and being taken for an P — K + 1 industry in state L (and thus not subsequently investing and earning nH O (0) ). Further, it is also 40 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. necessary that not lobbying and thus revealing state L (earning ^ ( 0 ) ) is more attractive to industries in state L than lobbying and trying to deceive the capital market into believing that the p — K “good” state has occurred (earning----------------(f)+ (l - f l) n Ll (o))— K + /) ) . Therefore, P — K + 1 the separating equilibrium exists when J Z ' k I j (0 + (l - (o))-K + l)>Jcm (o) ^ (f)+ (l - h ) k u (0))-K + / ) < n u (0). Suppose that both of the conditions for the separating equilibrium in lobbying are satisfied and consider the following strategies and beliefs. In state H industry lobbies and the capital market correctly infers that state H has occurred. The industry invests and the original stockholders earn since P = fin H I ( t) + (l — J U )?z H I (0 ) given the capital market’s beliefs. Alternatively, state L occurs and the industry does not lobby but invests and earns 7lLI{ 0 ). These strategies and beliefs are consistent with a PBE, so the above conditions are necessary and sufficient for the existence of a separating equilibrium in lobbying. Alternatively two possible pooling equilibria exist - lobbying occurs in both states or in neither state. If the capital market observes lobbying in both states or in neither state, then its posterior beliefs are the same as their prior beliefs. It is simple to show that the conditions for the pooling equilibria are satisfied when the conditions for the separating equilibrium are violated. Further, note that the conditions for the separating equilibrium form an upper and lower bound for P — K the lobbying costs. If — (f)+ (l - h )k h1 ( 0 ) ) - Z + / ) > ^ ho (o ) is violated, P — K + 1 then lobbying costs (AT) are too large and industries in neither state will lobby. On the other hand, 41 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. P — K if —---- ((/UXU ( t) + ( l- J U )tCl i (o))- K + l ) < 7 t Ll(p ) is violated, then lobbying costs are P — K + I too low to discourage industries in state L from attempting to pass themselves off as H state industries. By combining the conditions for the separating equilibrium, we derive a range for K in which the efficient strategy set is the equilibrium strategy. First, lobbying costs must be less than + n ) n Hl{0 )) — 7 Z f r 0(0) to maintain the efficient equilibrium. Let AT1 " 3 * represent the upper bound, jK ’m a x — jj ) n m (O))—^ ^ ( o ) . Second, lobbying costs must not fall below \jJ7l H I (V)+ (l — /i)n m ( o )+1 + ( .u(n u (f)~ 7 lLI ( o ) ) —? ] . 1 0 Below this level, the cost o f lobbying is not sufficient to deter state L managers from trying to deceive the market as to the true state. Let K™* represent the lower bound, K™* = \jd7lm (t) + ( l - ju )x m (o) + / + fj,{n u (t) - 7Z li (o)) - y \ . If there is some lobbying cost, K , such that ™ n < K < K 1 ™ * , then for lobbying cost K lobbying serves as an accurate signal of the true state. The efficient strategy is a perfect Bayesian equilibrium strategy. Proposition 4: I f the government provides a framework through which the domestic industry can lobby fo r trade protection, lobbying serves as an accurate signal o f the true state when lobbying costs (K ) falls within the range mm, K “ ). The efficient strategy is a perfect Bayesian equilibrium strategy. 1 0 See the appendix for derivation o f the value for the lower bound. 42 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. In the US, filing petitions for protection can be costly. The majority o f the costs consist of legal and economic consulting fees and construction of a structural adjustment plan, which are relatively fixed (Feinberg and Hirsch, 1989). The probability o f remedy (represented here by JU) varies depending on the code to which the petition appeals and over time. Forty-four percent of AD petitions between 1980 and 1985 resulted in remedies.1 1 During the same period, 38.5 percent o f safeguard petitions led to remedies or adjustment assistance. As the number o f safeguard petitions has fallen, the proportion winning protection has increased. Five o f the nine US petitions between 1995 and 2000 gained protection (55.5%). Global safeguard petitions during this period yielded remedies in 38 percent of cases. Special Cases Consider two special cases of the efficient equilibrium strategy set: the government follows a policy of free trade ( t = 0 ) , or the government always protects the domestic industry ( t > 0 ). Suppose the government’s trade policy is one o f free trade. Although the manager knows that lobbying will not result in protection, the signaling value o f lobbying can be sufficient to induce the efficient equilibrium. Updating the separating equilibrium conditions to reflect a zero probability of protection, we find that the signal will accurately distinguish between states when the cost o f the signal is in the range ^ - j ^ O ) + 1 - (n H M + 1 )2 - 4(/(% ,(<>) - ^ u (0 ))) 0) - x H O (0) condition for this set to be non-empty is that the return in state H meets a minimum level, A sufficient 1 1 This figure does not include those petitions that were filed and subsequently withdrawn. Many petitions are eventually withdrawn (38 percent between 1980 and 1985) after the disputing parties reach a settlement. If we consider these settlements as a type o f remedy, then the incidence of remedy for this period rises to 65 percent (Prusa, 1992). 43 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. «w(0) — ^ 0(0 ) > / . The free trade case illustrates that a framework using lobbying activity as a signal can reduce the incidence of the capital market inefficiency even if lobbying never results in protection. The signal works absent any rent-shifting motive. Suppose instead that the government follows a protectionist trade policy with certainty. In this case, the tariff and the signal both work to reduce the incidence of the capital market inefficiency. We update the conditions for the separating equilibrium to reflect the certainty o f protection, and find that lobbying accurately signals the state o f nature when lobbying costs fall within the range, / nH{t) +/+(%(*) - %(0)) - (% w+/+(% w -%(o)))2 ^ - 4 ((%W -%(o))(%(;)) +i(%W - % (o)))_ -%>( 0) Some simple manipulation shows that this set is non-empty when the return in state H exceeds the return in state L by a large enough margin. The sufficient condition for a non-empty set is - % 0(0)) - {rcu {t) - 7 ^ (0 )) > I . Although both the signal (lobbying) and tariff work toward reducing the inefficiency, the lure o f rents puts greater constraints on the signal than under free trade. Managers in state L are more likely to petition for protection now that the effort is rewarded with certainty. Consequently, the range o f lobbying costs where lobbying accurately indicates the true state is reduced from that in the free trade case. When the efficient strategy set fails to be an equilibrium, an alternative equilibrium exists in which investment occurs in only one of the two states of nature. From lemma 1, the investment has a positive NPV in both states. Thus, an efficient equilibrium involves investment in both states. The introduction o f the signal improves the markets ability to differentiate between the states. The signal fails, however, when the lobbying cost is inadequate to provide the proper incentives to the managers to correctly indicate the state. The alternative (pooling) equilibrium (in lobbying) strategy 44 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. (F,0,qUpj) { f .i .i M ) (NF,l,q\,(p)) . An analysis of these strategy sets indicates that the conditions for maintenance o f these equilibria are the same as the conditions for failure o f the separating (and efficient) equilibrium in lobbying. Signal of Endogenously Determined Cost In this section we endogenize determination of the lobbying cost and study the government’s equilibrium strategy. We allow the government discretion over the size o f lobbying costs giving it indirect control over the incidence o f the capital market inefficiency. Endogenizing lobbying cost adds another choice node to the game tree; see Figure 2-5. Nature moves first, determining the true state, which is then only revealed to the managers. The market and the government know the possible states of nature and the probability distribution over the states. At this point, the government sets the cost the domestic industry must incur should they choose to lobby for protection. The cost is common knowledge. Upon learning the lobbying cost, the managers decide whether or not to petition for protection. The remainder o f the game follows the previous structure set out in Figure 2-4. The strategies for firm managers (as they depend on government strategies) remain unchanged from preceding sections. Government’s efficient strategy set includes selecting the lobbying cost and tariff level that support investment by the domestic firm, (K,t). The government’s strategy results from the maximization of its objective function - domestic social welfare. Domestic social welfare is defined as the sum o f domestic consumer surplus, domestic producer surplus, and tariff revenue. We assume the government equally weighs the components of 45 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 2-5. Game Tree with Endogenous Lobbying Cost Nature State Government Lobbying Cost Domestic Industry Lobbying Decision Government Tariff Domestic Industry Investment Decision Capital Market Equity Valuation Domestic Industry Output Payoffs Profits New Equity Value Social Welfare Not Lobby Not Lobby Invest/ \ Not Invest Invest/ \ Not Invest Invest/ \N o t Invest Invest / \ Not Invest P \ / P XX XX XX XX XX XX XX XX # • # n l i(0,K) 7 1 lo (o, K) n u (t, K) 7 0 / Gl. (0,K) Glo(0,K) Gu (t, K) • • • « L0(t,K) /I H I (t, K) 7 1 H O (t,K) 0 / 0 Gl o (t, K) Gffl (t,K) Gh o (t, K) 7tffl(0,K) 7 1 „0 (0,K) I 0 G ffl (Q ,K ) G ho ( 0 , K ) 4^ 0\ social welfare in its objective function (Brander and Spencer, 1984).1 2 Each component o f domestic social welfare is a function o f the protection level (tariff). We represent domestic demand by Pm a x q d { p i t)) and define p m a 3 i as the price such that q d {p 'mx) = 0. The integral ^ q d {j> {tf)dp pf p f represents domestic consumer surplus. Profits o f the domestic industry are J qf {jp(tfdp, where „rain pj p™ m is the price for which j = 0 . Tariff revenue accruing to the government is t^}d (p(t))-qy (/?(?))]. Summing the components o f domestic social welfare, the government’s objective function is pf max, G{t) = \ q d(p{t))dp+ \q-j(p{t))dp + t\qd(p{t)) - ql(p{t))\ . Solving the first p f order condition for the optimal tariff, we find t = 0 . Free trade is the socially optimal trade policy for the domestic government to follow. We anticipate this result given the domestic country is a small, open economy. The government’s free trade strategy does not deter managers from lobbying when the cost of the signal falls in the range Aj r .i/t '\ 2" + 1 - (%7(0) + i f - 4(/(%7(0) - T C jj(0))) ^ j,^tf/(0) - 7T HQ(0) (see discussion o f exogenously determined lobbying cost). Now consider the government’s initial choice node, where it sets lobbying costs. Lobbying imposes a lump sum drain on social welfare. Further since lobbying is a drain on social welfare, if 1 2 We place an equal weighting on the government’s desire to maximize consumer surplus (based on reelection concerns) and its value o f industry profits (based on its desire for contributions and political support that also weigh heavily in reelection issues). Others in the literature have considered other weightings that reflect different emphasis o f government based on various assumptions. See Baldwin (1987) and Feenstra and Lewis (1991) for examples. 47 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. the government decides to set lobbying costs so that lobbying is an effective signal, the government implements the minimum cost within the effective range, K mui. Alternatively, lobbying costs the industry nothing and the inefficiency continues. These options form the branches o f the government’s objective function. max d t = ( \ K = K m ij = 6 d [t= (\K = 0 )= 6 jqd(j(0))dp¥ d „m in » » Pm + ! Pa ( i - e i j / W 0) ) ^ - Id dT \<fW ¥p+ P l dr Comparing the branches o f the government’s objective function, the government implements an effective lobbying cost if and only if the expected value from doing so (increased producer surplus) exceeds the cost o f correcting the inefficiency (lobbying cost). The condition for lobbying to serve as an effective signal o f the true state is 9 P h P h I <im (p (t))d p - J q s m (p (t))d p > ATm m . If the net present value o f the investment is sufficiently large, then the government will set lobbying cost so as to produce an effective signal. The government also sets K within the effective range for sufficiently large probability o f the “good” state ( H ) . Proposition 5: I f the government maximizes domestic social welfare, then it will follow a strategy o f setting lobbying cost such that the lobbying signal accurately indicates the true state and responds to lobbying activity with free trade policy. There is some anecdotal evidence that suggests that governments do attempt to adjust the cost to firms o f lobbying for protection through the use o f safeguard, antidumping, and 48 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. countervailing duties clauses. In the Uruguay round o f the WTO, changes were made in the antidumping and countervailing duties clauses to make them more costly to employ. These changes were ostensibly to reduce their use. V. Concluding Remarks In this paper we define an economy with an informational structure such that there exists an asymmetry of information between managers of industry and all other agents in the economy. Superior knowledge on the part of managers results in the rejection o f profitable investments. (Notably, the inefficiency only arises when the true state is the high profitability state.) The information available to managers allows them to accurately value the industry. The capital market, however, only knows the probability distribution over possible states and thus the industry’s expected value. When the true value is high, the expected value necessarily falls below the true value. Therefore, managers contemplating obtaining funding for an investment through the issuance and sale o f new equity are unwilling to accept the dilution in the value o f the original stock that a new issue would entail. We demonstrate that lobbying for import protection can reduce the incidence o f the capital market inefficiency. This result is in contrast to the usual DUP role attributed to lobbying. We further demonstrate that tariff protection can also reduce the inefficiency, but that imposition o f a tariff is not a necessary condition to induce lobbying or for lobbying activity to be effective in reducing the capital market inefficiency. Each instrument works independently to reduce the inefficiency. Lobbying behavior of the domestic industry can serve as a signal to the capital market as to the industry’s true state. The ‘signal’ provides information to the market and in so doing prevents the reduction in the value o f the original stock that a new issue would create. The opportunity cost o f the resources expended in the signaling process reduces the value o f the investment (as opposed 49 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. to the increase in value a tariff creates), but for certain levels this cost is dominated by the value o f the investment. A tariff shifts rents to the domestic industry further raising its value. At certain levels, this increases the attractiveness of the investment overriding the dilution that the sale o f new stock creates. As a result, protection reduces the capital market inefficiency. The social value of the reduced inefficiency, however, when weighed against the distortions created by the tariff is insufficient to induce the domestic government to abandon its free trade policy. Since lobbying behavior does not create the same price/consumption distortion as a tariff, it is less costly to social welfare. We show that a government maximizing domestic social welfare provides the appropriate lobbying opportunities providing a signaling mechanism, but follows a policy o f free trade. Between 1975 and 1987, 60 relief petitions were filed with the ITC o f which 20 resulted in relief. The number o f petitions has declined in recent years, to only 9 between January 1995 and November 2000; representing a 78 percent reduction from the six-year span two decades earlier. Petitions receiving relief, however, are up as a percentage of petitions filed to 55.5 percent. While section 201 (and the corresponding international safeguard clauses) is not the code of choice, firms and industries appealing to it have become more successful in obtaining protection over time. This may indicate that those that petition for protection under section 201 are those that truly need adjustment assistance (as determined by the ITC) as opposed to pure rent-seekers. Motives o f firms appealing to the AD and CVD codes are less clear. Prusa and Skeath (2001) find both economic and strategic motivation behind AD filings. We leave the discussion o f strategic (retaliatory) motivation for future research.1 3 1 3 The structure o f the WTO Agreement on Safeguards makes retaliation difficult and unlikely. Remedies under the safeguard provision are applied generally as opposed to on specific countries as are AD and CVD remedies. Further, upon the imposition of safeguard protection, affected parties must engage in consultation. If mutually acceptable agreement is not reached, then the party suffering safeguard regulation must wait 3 years before imposing retaliatory trade barriers. The waiting period will reduce the probability o f retaliation by giving the parties time to “cool-off.” 50 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Finally, we wish to make a clear distinction between the work in this paper and the infant industry literature. The infant industry literature argues that temporary protection o f new industries allows them the opportunity to acquire the skills necessary to effectively compete in developed global markets. Much of this literature focuses on the development of human capital and information (Hagen, 1958). The problems the literature addresses include labor mobility and the non-excludability o f information that prevent the investor from recouping expenses. Consequently, industries are either unwilling to invest or unable to acquire the necessary funding. This paper makes no comment on these arguments, but mentions them only to illustrate differences with our work. First, the industry in this paper need not and is not an “infant” industry. The industry can be thought o f has being long lived and recently finds itself losing competitive ground due to advances abroad. Second, retooling is not a necessary condition for the survival of the industry. The industry remains viable at its reduced profit level. Third, the type o f investment we consider is one of physical capital, which is not subject to the mobility and non-excludability issues, with which the infant industry literature is concerned. The condition preventing investment in this paper is not one o f inability to recoup investment costs, but instead arises from the protection o f one group at the expense of general efficiency. An asymmetry in the information available to the different groups of investors causes an inefficiency in the capital markets. Finally, the literature makes note that classification o f models such as ours as belonging to the infant industry literature is incorrect and misleading (Baldwin, 1969). We hope that this short comment helps to clearly distinguish between this paper and the existing infant industry literature. References Akerlof, G.A.. “The Market for ‘Lemons’: Quality and the market mechanism,” Quarterly Journal o f Economics, 1970, 84,488-500. Bagwell, Kyle and Robert Staiger, 1988, “The Role o f Export Subsidies when Product Quality is Unknown,” Journal o f International Economics, 27, 69-89. 51 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Baldwin, Richard. “Politically Realistic Objective Functions and Trade Policy,” Economics Letters, 1987, 24, 287-90. Baldwin, Robert E. “The Case against Infant-Industry Tariff Protection,” Journal o f Political Economy, 1969, LXXVII, 295-305. Barclay, Michael and Robert Litzenberger, 1988, “Announcement Effects of New Equity Issues and the Use of Intraday Price Data,” Journal o f Financial Economics, 21, 71-99. Bhagwati, J. N. “Directly-unproductive, profit seeking (DUP) activities,” Journal o f Political Economy, October 1982, 90, 988-1002. Brander, James and Barbara Spencer. “Tariff Protection and Imperfect Competition,” Monopolistic Competition and International Trade, ed. H. Kierzkowski, Oxford University Press, 1984. Dixit, Avinash. “Trade and Insurance with Moral Hazard,” Journal o f international Economics, 1987, 23,201-220. Dixit, Avinash. “Trade and Insurance with Imperfectly Observed Outcomes,” Quarterly Journal o f Economics, 1989a, 104, 195-203. Dixit, Avinash. “Trade and Insurance with Adverse Selection,” Review o f Economics Studies, 1989b, 56, 235-247. Donnenfield, Shabtai, Shlomo Weber, and Uri Ben-Zion, 1985, “Import controls under imperfect information,” Journal o f International Economics, 19, 341-54. Eaton, Jonathan and Gene Grossman. “Tariffs as Insurance: Optimal Commercial Policy when Domestic Markets are Incomplete,” Canadian Journal o f Economics, 1985, 18(2), 258- 272. Fama, Eugene. “Efficient Capital Markets: Review o f Theory and Empirical Work,” Journal o f Finance, 25, 383-416. Feenstra, Robert and Tracy Lewis. “Negotiated Trade Restrictions with Private Political Pressure,” Quarterly Journal o f Economics, November 1991, 106(4), 1287-308. Feinberg, R.M. and B.T. Hirsch. “Industry Rent Seeking and the Filing of ‘Unfair Trade’ Complaints,” International Journal o f Industrial Organization, 1989, 7, 325-340. Grossman, Gene and Henrik Horn. “Infant-industry Protection Reconsidered: The Case of Informational Barriers to Entry,” Quarterly Journal o f Economics, November 1988, 767- 787. Hagen, Everett E. “An Economic Justification o f Protectionism,” Quarterly Journal o f Economics, 1958, LXXII, 496-514. Hagen, Everett E. “Reply” to the “Comment” on “An Economic Justification o f Protectionism” by A. Koo in the same journal, Quarterly Journal o f Economics, 1958, LXXV, 145-151. 52 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Harris, Milton and Artur Raviv. “The Theory o f Capital Structure,” Journal o f Finance, 1991, 46, 297-355. Jacobson, Robert and David Aaker, 1993, “Myopic management behavior with efficient, but imperfect, financial markets,” Journal o f Accounting and Economics, 16, 383-405. Jensen, M.C. and W.H. Meckling. “Theory o f the Firm: Managerial Behavior, Agency Costs and Ownership Structure,” Journal o f Financial Economics, 1976, 3, 305-360. Johnson, Harry G. “Optimal Trade Intervention in the Presence o f Domestic Distortions,” Trade Growth, and the Balance o f Payments: Essays in Honor o f Gottfried Haberler, 1965, 3-34. Koo, Anthony Y.C. “An Economic Justification o f Protectionism: Comment,” Quarterly Journal o f Economics, 1961, LXXV, 133-44. Leland, H.E. and D.H. Pyle. “Informational Asymmetries, Financial Structure, and Financial Intermediation,” Journal o f Finance, 1977, 32, 371-387. Lipsey, R.G. and K. Lancaster. “The General Theory o f the Second Best,” Review o f Economic Studies, 1956, 24, 11-32. Mayer, Wolfgang, 1984, “The infant-export industry argument,” Canadian Journal o f Economics, 17, 249-69. Modigliani, Franco and M.H. Miller. “The Cost o f Capital, Corporate Finance and the Theory of Investment,” American Economic Review, 1958,48, 261-297. Myers, S.C. and N.S. Majluf. “Corporate Financing and Investment Decisions when Firms have Information that Investors do not Have, ” Journal o f Financial Economics, 1984, 13, 187- 221 . Newbery, David and Joseph Stiglitz. “Pareto Interior Trade,” Review o f Economics Studies, 1984, 51, 1-12. Peter, J.P. “Harley-Davidson, Inc. - Motorcycle Division, “ in S.C. Certo and J.P. Peter (Eds). Selected Cases in Strategic Management, New York: McGraw-Hill, 461-475. Prusa, T.J. “Why are so many antidumping petitions withdrawn?,” Journal o f International Economics, 1992, 33, 1-20. Raff, Horst and Young-Han Kim, 1999, “Optimal export policy in the presence o f informational barriers to entry and imperfect competition,” Journal o f International Economics, 49, 99- 123. Rosendorff Peter B. “Voluntary Export Restraints, Antidumping Procedure, and Domestic Politics,” American Economic Review, 1996, 86(3), 544-561. Spence, Michael. “Job Market Signaling,” Quarterly Journal o f Economics, 1973, 87, 355-374. 53 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Stein, Jeremy, 1989, “Efficient capital markets, inefficient firms: A model o f myopic corporate behavior,” Quarterly Journal o f Economics, 103, 655-69. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Chapter 3 Income, Disamenity, and Toxic Chemical Releases in the United States R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Abstract Building on the literature concerning the Environmental Kuznets curve, we develop and test a simple economic model o f toxic releases for counties within the United States. We expand upon previous literature by using a twofold measure of environmental disamenity associated with toxic releases: the level and toxicity of releases. We find that in the United States, a concave relationship between income and release levels exists for the 21 chemicals in our study as reported in the U.S. EPA’s Toxic Release Inventory. Further, we show that the sensitivity o f this relationship (i.e. the concavity) depends importantly on the toxicity of the release in question. Finally, our model expands on the literature by more fully specifying the political and economic factors that determine toxic releases. We find that the capital-labor ratio for a county is an important determinant o f toxic releases - ceteris paribus, more capital-intensive counties have higher releases. We also find that county characteristics influencing voting behavior affect release levels. 56 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. I. Introduction and Literature Review Increased public concern over environmental quality led the United States Congress to pass the Emergency Planning and Community Right-to-Know Act (EPCRA) in 1986. The EPCRA mandated the reporting o f the release and transfer of 337 different toxic chemicals,1 and outlined the process through which manufacturing facilities would file reports with the United States Environmental Protection Agency (EPA). The EPA was charged with then assembling and disseminating the information to the general public. The final report is the TRI or Toxic Release Inventory. Policy makers’ vision for the TRI was as an information source by which citizens could easily obtain information on the toxic releases in their neighborhood, city, county, or state. Much o f the analyses using TRI data have focused on aggregate measures o f releases. Large quantities of releases are thought to signal low environmental quality. Low rated or “dirty” states and companies (with relatively high aggregate releases) have experienced significant public scrutiny as a result. Public pressure has caused some companies to modify their production to reduce their release levels and/or release composition. States have also responded by instituting more stringent regulations on chemical releases. The result is a general downward trend in aggregate releases since 1987 (Arora and Cason, 1999). While this trend seems to signal improving environmental quality, aggregate release figures provide only a partial measure. Chemicals in the EPCRA mandate vary significantly in the degree o f their toxicity. For example, the TRI includes both the relatively benign isopropyl alcohol (rubbing alcohol) and the highly toxic mercury. Suppose we sum releases o f the two for a total o f X pounds. The releases degrade the environment significantly more when the proportions o f the two are 95 percent mercury and 5 percent isopropyl alcohol than if the proportions are reversed. Thus given the toxicity variation, analyses must control for variation in toxicity o f chemicals in order to capture the true variation in environmental quality. 57 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Environmental quality is generally assumed to be a normal good. As such, individuals demand more environmental quality as their incomes increase. We also assume that the utility consumers derive from environmental quality can be degraded both by the aggregate level and the damage (toxicity) caused by toxic releases. Given this dual concern, we include both the amount o f toxic pollution and the toxicity o f this pollution in our investigation o f the determinants of environmental quality in counties o f the United States. From these assumptions, we predict increases in consumer incomes increase demand for environmental quality. The demand for environmental quality is manifest as increased consumer activism to reduce or eliminate toxins from the local environment. This consumer activism may take a variety of forms: political pressure for regulations of toxic releases, heightened enforcement o f those regulations, boycotts o f companies with high levels o f releases and/or highly toxic releases, and reductions in the stock prices of companies reporting unanticipated high (and highly toxic) releases (Hamilton, 1995). Assuming regulators (and thus the regulation and observed level o f releases) are responsive to consumer activism,2 then regulation increases as consumer’s incomes increase. A considerable literature exists suggesting that variation in economic and demographic characteristics may, in part, explain geographic variation in pollution. Esty (2001) and Cavlovic et al. (2000) provide summaries o f this literature. In an early study on international levels o f pollution and income, Grossman and Krueger (1993) estimate releases by type (sulfur dioxide, particulate matter, and several measures of water quality). Their results demonstrate a quadratic relationship between per capita income and pollution levels, which has come to be known as the environmental Kuznets curve (EKC). Grossman and Krueger show that at low levels of income, increases in income are associated with increases in pollution. Then at some income level (know and the income turning point or ITP), further increases in income are associated with decreases in pollution. With 1 The EPCRA has been amended to currently include 643 chemicals. 2 This is often assumed, see Antweiler, Copeland, and Taylor (2001) and Arora and Cason (1999). 58 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. their panel o f macroeconomic measures and pollution data, the authors estimate curves that exhibit varying shapes over the measures o f environmental quality. Without a formal theoretical model, however, Grossman and Krueger were unable to explain the differences in the estimated relationship across pollutant measures. Others in the literature also find support for the EKC, some within a single country. A number o f these studies have used the TRI and socioeconomic data from the United States Census. Arora and Cason (1999), Helland and Whitford (2000) use unadjusted aggregated release values when performing analysis on TRI data. These studies attempt to explain the relationship between aggregate releases and income (and other social, economic, and political factors) based on an implicit assumption of constant toxicity. We suggest that this approach both potentially distorts their results and fails to address the importance o f toxicity in determining release levels. Arora and Cason (1999) analyze the economic factors behind toxic release levels as reported by the TRI. They estimate the relationship between pounds of total releases and socioeconomic characteristics o f US zip code areas. Arora and Cason find evidence of the EKC phenomenon for the US with income turning points that start from $30,000 (exact magnitudes vary depending on the specification o f their model and the sample). Their analysis does not include weights for chemical toxicity. They comment that in previous work (Arora and Cason, 1995) they compare aggregation with and without weighting for toxicity and find their results are not sensitive to the weighting scheme. They further note that many o f the TRI listed chemicals that are in wide use are o f similar toxicity or have not been assigned a toxicity rank. Since then, the EPA has considerably extended their toxicological reports now allowing for greater comparison and analysis. While Arora and Cason forego direct weights o f releases, they make two other important adjustments in their estimation procedure. First they note that previous studies failed to recognize that within a geographic area, releases and socioeconomic characteristics are simultaneously 59 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. determined. This creates a potential simultaneity problem.3 To address this issue, they use 1993 releases and 1990 socioeconomic variable values. Arora and Cason’s second methodological contribution is that they recognize and correct for possible Heckman selectivity bias. Seventy-two percent o f the U.S. zip codes in their study experienced no releases in 1993. To correct for possible selection bias, they begin their analysis with a probit estimation in order to construct and then use the inverse Mills ratio in their estimation o f the level o f releases. Their results, however, fail to support the presence o f selectivity bias. Like other studies, Arora and Cason estimate an inverted U-shaped relationship between income and releases. While the relationship between income and releases is as expected in the Arora and Cason study, the coefficients on other social and economic control variables are counterintuitive. These include a negative sign on the coefficient for the percent o f the population that is black, suggesting that releases fall as the black proportion o f the population increases. Further, Arora and Cason estimate coefficients not significantly different from zero for the following: the percent o f the population that is urban, the percent that work in the manufacturing sector, the percent in poverty, the percent with a bachelors degree, and the percent o f renter occupied housing. These results are not consistent with much of the literature. Brooks and Sethi (1997) also analyze toxic releases using TRI data. Unlike Arora and Cason, they control for the varying toxicity of the chemicals by creating a measure o f toxic chemical exposure for each zip code’s population. They construct this measure by taking the ratio o f releases o f each chemical in the TRI to its threshold limit value4 then summing these ratios over all values within 30 kilometers of zip code i . Their measures span the years 1988-1992. For each o f these 3 Causation moves in both directions. Socioeconomic characteristics influence release levels and release level may well influence an area’s socioeconomic characteristics. 4 A threshold limit value is the amount o f airborne concentration o f a substance to which a worker can repeatedly be exposed to for a standard 8-hour workday without suffering adverse health effects. These limits have been used by OSHA (Occupational Safety and Health Administration) in setting US workplace standards. 60 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. years, Brooks and Sethi find evidence o f the EKC, however, with surprisingly large income turning points of $67,000 and $143,000. Due to data constraints, many o f the explanatory variables used in the Brooks and Sethi analysis are measured only in 1990 (Census figures). Brooks and Sethi report their estimation results by year to show that coefficient magnitudes and significance do not vary significantly across years. This result shows a lack o f independence over time for either releases or socioeconomic variables and suggests the simultaneity problem posited by Arora and Cason is a not an issue. Brooks and Sethi’s analysis uses a similar set of explanatory variables to the Arora and Cason study to estimate environmental quality as measured by exposure levels. By weighting releases for toxicity and creating exposure values, Brooks and Sethi obtain more intuitive and literature consistent results. They estimate a positive, significant coefficient on the percent o f the population that is non-white (Arora and Cason estimate a negative coefficient for the percent of the population that is black). Brooks and Sethi also estimate coefficients significantly different from zero on the aforementioned variables for which Arora and Cason failed to find significance. Overall, Brooks and Sethi’s results are more consistent with intuition than the ones in the Arora and Cason study. One possible reason for these more intuitive results is that Brooks and Sethi’s measure o f environmental quality includes both aggregate releases and a weight for toxicity. Both studies use an underlying model that assumes consumers react to environmental quality through activism, and that regulators respond to consumers. This model predicts regulators increase controls V on more toxic chemicals faster than on less toxic ones; the rate o f change of release levels varies by a chemical’s toxicity. Brooks and Sethi’s use o f toxicity weights better captures these effects. However, in their creation o f exposure measures, Brooks and Sethi combine the level and toxicity effects. Separating the effects in their results is not possible. Thus, their analysis fails to provide the reader with a measure o f the toxicity effect. 61 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Recently researchers have begun again to incorporate differentiation across chemicals by toxicity or other important characteristics. This literature, while small, has tended to support the use o f toxicity weights. Dasgupta, Laplante, and Meisner (1998) demonstrate that in ranking Brazilian states by their release levels, the rankings were sensitive to the use o f a toxicity-weighting scheme. Dasgupta et al. (1998), Grossman and Krueger (1993), and Brooks and Sethi’s (1997) results should raise questions about results of other studies that use unadjusted aggregate values o f chemical releases. In this paper, we attempt to fill this gap in the literature. We begin by developing a simple theoretic model that links consumer utility to toxic releases. Our model is built upon the Antweiler, Copeland, and Taylor (2001) model that explains the relationship between trade liberalization and pollution across countries. Then we use our model to develop and test hypotheses that link consumer preferences with measures of environmental disamenity (toxicity o f releases) and pollution release levels. We find that the toxicity of a chemical release is an important determinant of the level o f releases. Our estimated income turning points range from $26,027 for the least toxic pollutants to $22,835 for the most toxic pollutants in the TRI. We further contribute to the literature by using explanatory variables not previously used with the TRI data. We create a capital-labor ratio for U.S. counties. Capital-intensive production techniques produce more pollution (on average) than labor-intensive techniques. The capital-labor ratio allows us to control for the capital-intensity o f a county. The international-environmental literature demonstrates capital-labor ratios have a significant, positive effect on release levels (Antweiler, Copeland, and Taylor (2001), Dean (1998), Tobey (1990)). We also use an expanded set o f data on chemical properties (carcinogen indicator and chemical’s physical state) in our analysis to better characterize the disutility o f toxic releases reported in the TRI. The remainder o f the paper is organized as follows. First we discuss the theoretical framework o f our study. We present our model and outline the hypotheses that we intend to test. Then, we 62 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. describe our data and results. The results are organized into four areas: biochemical characteristics as determinants o f releases, technological determinants o f releases, control variables, and determinants of releases over time. Finally, we summarize our results and present our concluding remarks. II. Theoretical Framework and Hypothesis Construction Producers, Consumers, and Government Antweiler, Copeland, and Taylor (2001) develop a theoretical model demonstrating the relationship between pollution levels and socioeconomic characteristics across countries. Antweiler et al. illustrate a release level determination mechanism through which pollution levels are determined. Three types o f agents comprise the mechanism -producers, consumers, and the government. In this paper, we adopt and extend their model to motivate our hypotheses and empirical analyses. The production side o f the economy consists o f a large number of firms producing a variety o f consumption goods and services. We assume that each firm’s production is characterized by constant returns to scale and they participate in a competitive market environment. Chemical releases are expelled as a by-product o f the production process, rel = B(q, abate) where rel = [rel; ... relm ] is a vector o f release levels for m chemicals, q is a measure of industrial output, and abate captures investment in pollution abatement technology. All firms have the opportunity to invest in abatement technology that will reduce their releases. For simplicity, we assume that the efficacy o f abatement technology is known with certainty. The firms choose their output and abatement levels to maximize profits taking prices and pollution taxes as given. The profit function of an individual firm is given by, 63 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. n = 7i(p,q,abate,regx regm ) where p represents a vector of input and output prices, q is the level o f output o f the consumption good, a b a te is a vector o f abatement spending levels undertaken by the firm to reduce releases o f chemical i , and regt is a vector o f costly pollution regulation (e.g. a “command and control” regulation or a tax) on chemical i . Investment in abatement technology for chemical i increases with regi, ceteris paribus. In the short-run, abatement is costly (i.e. Jtabate^). Consumers in our model maximize their utility conditional upon the level of environmental disamenity present. Every consumer has well-behaved preferences and their utility is a function of consumption and environmental disamenity. Generally, the aggregate consumer’s utility function is given by U = u ( c ,z r , . . . , z m, E ) subject to 7 = pc, where c is a vector o f consumption levels o f goods and services, z ; . is the environmental disamenity from chemical i, m is the number o f chemicals released into the environment, E is all other aspects o f environmental quality, Y is income, p is a vector o f prices o f consumption goods, and all income is spent on consumption, c. The utility function follows standard assumptions: u is concave in consumption (uc >0, m c c <0) and increasing in environmental quality (uE >0, uz<0). Consumption is a function of income and prices. On average, the consumer’s income depends on the income o f the local firm (e.g. through direct employment, stockholding, or tax transfers). Without further loss o f generality we assume that the consumer’s income depends directly on the profitability o f the local firm, Y = H I N where N is the populatioa Since utility is concave in c, we know that the marginal utility of consumption declines as income levels increase. We assume that the marginal utility o f environmental quality (a public good) remains constant or increases with increasing income as in Weitzman (1994). Thus the consumer’s preferred mix of consumption and environmental quality changes with increasing levels o f income. As income grows, consumers 64 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. prefer increases in environmental quality relative to increased quantities of consumption goods. (Note that at higher income levels, consumers may also begin to demand goods, c, produced with cleaner technologies. For simplicity, we do not model the demand for green products here, but note that the results o f such demand reinforce our model.) We assume that the level of disamenity corresponds with the health and environmental impacts o f the pollutant. The level o f environmental disamenity is a function o f release levels, the harm caused by these releases, and the obviousness o f the pollution. We assume specifically that the level o f disamenity associated with chemical i is given by the function where re l{ denotes the level o f releases o f chemical i , toxt is the toxicity o f chemical i , and ex, is the exposure suffered by consumers from r e l,. Release levels, toxicity, and exposure all influence the aggregate harm caused by a pollutant. The level o f environmental disamenity, therefore, is an increasing function o f re l, , to x ,, and exi . In our model, the government responds, to varying degrees, to the preferences o f both consumers and firms as in Peltzman (1976). The degree to which the government responds to the demand by consumers for better environmental quality depends upon the level o f political activism and representation o f consumers. As the consumers’ preferred mix of consumption and environmental quality changes, so too does the government’s willingness to regulate polluters. As income increases, the government responds to consumer demand for a cleaner environment by imposing more stringent regulations on the polluting firms (either through command and control or 65 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. market-based mechanisms). Specifically, the stringency o f regulation reflects the income and thus where reg =[regj...regm ] is a vector o f m regulations (or tax rates), rep is a measure o f consumer representation or activism, and G z > 0 for all i . Thus, reg increase with releases, toxicity, the level o f exposure, and political representation. The Release Level Determination Mechanism and Hypotheses Given the consumers’ demand for consumption goods and environmental quality, the government mandates pollution regulation to maximize consumers’ welfare. Firms make their output and abatement decisions based upon the demand for final goods and the pollution regulations (or taxes) set by the government. Chemical characteristics, consumer preferences, government legislation, and firm choices jointly determine release levels o f each chemical. The reduced form function for releases is where all variables are as previously defined and x is a vector of control variables including the physical forms o f the toxics released and proxies for consumer tastes and firm characteristics. The relationship between release levels and income can be derived from the model and have been demonstrated in detail in Antweiler et al. (2001). At low income levels, the increasing scale of production (sometimes known as the scale effect) dominates the effect o f increasing demand for environmental quality. When incomes are low, consumers are mainly concerned with acquiring larger quantities o f consumption goods. At low income levels, the marginal utility of additional 5 Without changing the model, we may assume that consumers are directly aware o f the disamenities posed by pollutants or that the government possesses more information about pollutant disamenity consumption o f consumers and the disamenity caused by the pollutants5 emitted by the local industry rel, = / (tox,., ex,, Y, rep, x) 66 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. consumption is likely to be higher than the marginal utility o f improved environmental quality and thus, the relative demand for environmental quality will be low. As incomes and consumption rise, the marginal utility o f consumption falls while the marginal utility o f environmental quality remains constant or increases. As a result, environmental quality becomes increasingly important to consumers. The demand for environmental quality grows with income and at some point dominates the scale effect, lowering release levels with further increases in income. Lower levels o f pollution are achieved by reducing the output of emissions per unit o f industrial activity (the technique effect) or by changing the composition o f local industry from more polluting firms (e.g. manufacturing) to less polluting firms (e.g. the service sector). This latter transition is known as the composition effect. In addition to the now standard hypothesis o f an inverted-U shaped relationship between income and pollution levels, we make four other specific hypotheses regarding pollution releases. Our hypotheses result directly from the release level determination mechanism. We test the following hypotheses6: Hypothesis 1: Greater toxicity increases the level o f disamenity and thus the Environmental Kuznets Curve is more concave for more toxic chemicals {relYtox>0, rely ttox<0). Hypothesis 2: Cancer causing chemicals have a higher disamenity and thus result in lower emission levels (relcancer<0). Hypothesis 3: Releases that result in greater consumer exposure to harm result in greater disamenity and thus should result in lower toxic releases (relex<0). Hypothesis 4: Consumers with greater political representation will have their preferences better represented by regulators and thus release levels will be lower in these localities (relrep <0). than consumers, but acts in the best interest o f the consumer. 67 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Technological Constraints The production functions of producers also influence release levels. One measure that we use to control for production differences is the capital to labor ratio. Capital-intensive industries are generally more polluting than labor-intensive industries as discussed in Antweiler et al. (2001) and Dean (1992). Thus, areas with higher capital to labor ratios are likely to have a greater concentration o f pollution-intensive industries (this can be considered the starting point for the composition effect). One would expect that counties that had more polluting industries in the early years of the TRI reporting would have higher levels o f pollution, all things being equal. Because these types o f industries are inherently more polluting than other industries, we hypothesize that areas with capital-intensive industries experience higher levels o f releases. In the short-run (e.g. in the ten years between the first release o f the TRI and 1996), it would be difficult to alter the capital- intensity of industry within a county. Of course, over time, regulators should be able to influence the capital to labor ratios of local industry through regulation and incentives. Control Variables Following the literature, we include unemployment and education as control variables. At the margin, unemployed consumers are less concerned with environmental quality than with gaining employment (i.e. the marginal utility of income is greater than the marginal utility o f environmental quality). Areas experiencing high rates of unemployment may be more likely to allow large releases in order to encourage firms to locate or to continue to operate there, ceteris paribus. Education should also be important in determining toxic releases. For environmental disamenities to affect consumer utility, consumers must be aware o f the effects o f emissions. Individuals with more education are more likely to be cognizant o f environmental disamenities. 6 Technically, we attempt to reject the null hypotheses. We present the alternative hypotheses here 68 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. We also control for several county-level characteristics, including population density, population location (urban versus suburban or rural), and level o f manufacturing influence. Releases in more densely populated areas are likely to lead to greater levels o f exposure, ceteris paribus. Urban areas tend to be more industrialized and industrialization is positively correlated with pollution. Manufacturing industries are some o f the most polluting industries. Finally, we also control for the physical form o f the pollutant. The physical form o f a pollutant may contribute to its obviousness to the consumer. Physical form also could be important in determining the ease o f clean up and control by emitting firms. Solid and liquid forms o f pollutants can be more easily captured and contained (e.g. in barrels), while gaseous emissions are impossible to capture once released. III. Data Toxic Release Data The EPCRA requires all facilities in standard industrial classification (SIC) codes 20-39, with 10 or more full-time employees, manufacturing or processing more than 25,000 pounds or using more than 10,000 pounds o f any of the designated chemicals, to file annual reports with the US EPA. Reports contain detailed information on plant level releases of toxins (in pounds) into the air, land, and water. The reports also contain data on transfers o f chemicals off-site, and information on the plants from which the chemicals are released. Currently available data consists o f release estimates for 1987 to 1996. Following Brooks and Sethi (1997), we use a simple transformation o f the pounds o f air releases for each o f 21 chemicals in each US county as our dependent variable, taking the natural log for clarity. 69 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. of (1+total releases of chemical j in county k ). This log-linear functional form is less sensitive to large outliers than the untransformed data. The log-linear transformation also allows us to include the counties that experience zero releases o f a particular chemical (while a natural log transformation of releases would not). Further, the difference between taking the log o f releases and the log of one plus releases is negligible for large levels of releases. We use all 10 years of the TRI air releases. Recognizing that the level of total releases is only a partial measure of the risk suffered by citizens, the EPA developed a toxicity ranking system. The TRI Relative Risk-Based Chronic Human Health Indicator (Indicator) was developed using a methodology similar to that used in the EPA’s Hazard Ranking System (HRS). The HRS scoring system rates the inherent toxicity of chemicals based upon measures of their cancer slope factors7, reference doses8 , and acute toxicity levels9. These individual measures are compiled and compared to produce a proportional weighting system. This system is considered superior to a simple ordinal ranking because it allows for comparison o f chemicals by their toxicity weights.10 The chemicals in our study cover the entire range o f chemical toxicities defined by the EPA. Table 3-1 presents the chemicals we examine, their toxicity weights, and mean release levels (and standard deviations). Due to the large range of toxicities, in our estimation of releases, we use ln(l+toxicity) as our explanatory variable. 7 A chemical’s cancer slope factor represents an individual’s incremental lifetime risk o f cancer per dose o f the chemical. 8 A reference dose is an estimate o f a daily exposure to a chemical that, when subjected to, the human population is unlikely to experience an appreciable risk o f deleterious effects over an average lifetime. 9 Acute toxicity level refers to the potential for short-term exposure by any media (inhalation, oral, etc.) to cause acute health effects or death. 1 0 The EPA publishes inhalation toxicity weights for over 130 chemicals. These weights range from 0 to 1,000,000 and represent the chemical’s relative inhalation toxicity. The toxicity weights are proportional, thus the increase in the numerical score reflects a magnitude o f difference between the impacts associated with chemical releases. For example, our study includes the chemicals isopropyl alcohol with a weight o f 0, chlorine with a weight of 10, mercury weighted 10,000 and lead weighted 100,000. Chlorine is 2 orders of magnitude, mercury 5 orders o f magnitude, and lead 6 orders o f magnitude more toxic than isopropyl alcohol. 70 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3-1. Chemicals Chemical name CAS Registry No. Toxicity weight (tw) Carcinogen1 1 Physical State1 2 Observations (10 years) Mean value Standard deviation Isopropyl Alcohol 67-63-0 0 L 886 8.3907 2.2842 Phosphoric Acid 7664-38-2 0 F 4591 5.6967 2.2651 Ammonium Nitrate 6484-52-2 1 F 472 6.8327 2.9630 Ethyl Chloride 75-00-3 1 L 360 8.8513 3.0854 Ethylene 74-85-1 1 G 1278 10.1299 2.4136 Chlorine 7782-50-5 10 G 6035 6.9266 2.8522 Ethyl Benzene 100-41-4 10 L 4069 8.4793 2.1780 Methyl Chloride 74-87-3 10 G 731 9.3719 2.7379 Ammonia 7664-41-7 100 G 10178 9.3871 2.6569 Benzene 71-43-2 100 K L 2340 8.8198 2.7143 Chlorobenzene 108-90-7 100 L 528 7.3762 3.1023 Asbestos 1332-21-4 1,000 K F 453 4.7916 1.9260 Hydrogen Chloride 7647-01-0 1,000 L 7992 8.3224 3.0207 Chlorine Dioxide 10049-04-4 10,000 G 996 8.7644 2.7182 1 1 K=known carcinogen, R=reasonably accepted to be a carcinogen 1 2 G=gas, L=liquid, F=crystal/fiber, and M=metal H- * Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3-1. Chemicals continued Chemical name CAS Registry No. Toxicity weight (tw) Carcinogen Physical State Observations (10 years) Mean value Standard deviation Mercury 7439-97-6 10,000 M 226 5.5751 2.1442 Nickel 7440-02-0 10,000 R M 4624 4.9208 2.1741 Aluminum 7429-90-5 100,000 M 1972 6.7248 2.6708 Arsenic 7440-38-2 100,000 K F 439 3.5447 1.8363 Lead 7439-92-1 100,000 M 3415 5.5580 2.2073 Sulfuric Acid 7664-93-9 100,000 F 8353 7.3217 2.7964 Dimethyl Sulfate 77-78-1 1,000,000 R L 238 4.3261 2.0921 < » to We also use chemical characteristics that expand upon those used in previous TRI studies. This includes whether the chemical is a known carcinogen (K) or “reasonably accepted to be a carcinogen” (R), and physical properties o f the chemicals. The chemicals in our study take on one o f four physical forms - crystals/fibers, gas, liquid, or solid/metal. We create dummy variables for the fiber, liquid, and metal forms (a dummy for chemicals taking a gaseous form has been excluded).1 3 We proxy for exposure by interacting population density with ln(l+toxicity). While the TRI provides a broad range o f data, the data are not without limitations. First, reported releases are estimates not actual measures. Release estimates are calculated using standard methodology, but are nonetheless estimates. Second, releases are aggregated over the entire calendar year and it is unknown if the releases occurred in a single discharge or were expelled evenly throughout the year. Timing of releases is important for it affects the level o f exposure or disamenity. Reference doses measure the daily exposure level that is safe to humans. If occurring in small doses some chemical releases, will cause no harm to nearby populations, but will produce significant harm if expelled in a single large emission. The TRI data does not allow for such a refinement on exposure. Third, only manufacturing facilities meeting the profile set forth by the EPCRA are required to file reports. Since only a selection o f manufacturing facilities file TRI reports, this poses potential data censoring problems. Within the group o f manufacturing industries, conditions on employee levels and chemical usage levels for reporting may be binding for some facilities. However, considering the average scale of production in manufacturing, we assume that these conditions are not binding for practical purposes. Further, since the TRI contains releases only from manufacturing 1 3 We also considered controlling for odor or atmospheric discoloration resulting from releases. Available individual chemical information, however, was insufficient and any categorization premised on this information would be arbitrary. 73 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. facilities, many other industries using toxic chemicals are excluded. Non-reporting industries that involve the use of toxic chemicals include mining, electric/gas/sanitary services, and agriculture. Others using the TRI data (Brooks and Sethi (1997), Arora and Cason (1999)) do not report these issues as important in their analysis. We follow the literature on this point. Finally, Levinson (1999) claims that censoring in the first two years of the TRI is high. Censoring in later years, however, is not reported in the literature. To control for the potential problem o f censored data in the early years o f the TRI, our analysis is performed in two parts - one containing all years o f data and the other excluding the first two years. Arora and Cason (1999) discuss an additional data issue - data selection bias. They hypothesize that industries may locate in communities based on a community’s characteristics. Consequently, community demographics influence releases levels and affect the probability that counties experience releases at all (e.g. manufacturing facilities may be less likely to locate in higher income areas). Arora and Cason use the Heckman correction in their analysis, but find they cannot reject the hypothesis that their data does not contain selection bias (their inverse Mills ratio is not significant). Based on Arora and Cason’s result, we do not include a Heckman correction in our analysis. Socioeconomic Data Our socioeconomic data are drawn from two main sources: the Bureau o f Census, and the Regional Economic Information System (REIS). Table 3-2 contains a list o f the socioeconomic variables we use in our analysis (and their summary statistics). As with the release data, we note the potential shortcomings o f our demographic data. A number of the variables in Table 3-2 are measured only once during the study period. The reason for this is that these measures are collected only during Census years. In spite o f this data constraint, we believe (as did Brooks and Sethi) that these measures did not change significantly over the period 74 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. o f our study and create no bias in our results. In addition, some o f our other variables also are limited in their measurement frequency. The variables vote and democrat are measured only periodically; both reflect measures of voting trends in national presidential elections. We assign the vote and democrat variables for 1988 to analyze the years 1987-1990, the 1992 measures for 1991 - 1994, and the 1996 measures for 1995-1996. The measures for the percent o f the population that is white and the percent female are only available for 1990-1996. We fill the 1987-1989 missing values with the 1990 measures for each county.1 4 Finally, capital-labor ratios are not available by county. Given the lack o f data, we create a proxy (capital/labor). We use new manufacturing capital expenditures as a measure of capital stock and the manufacturing labor force for labor. The one problem with our proxy is that expenditures measure a flow, not a stock. In using our expenditure measure, we assume that the flow measure is a good indicator o f the relative stock differences. In addition, we are only able to generate measures for capital/labor in 1987 and these measures are highly censored. We have capital/labor ratios for 70 percent o f US counties. Filling missing values with the mean from other counties is not appropriate because the non-reporting counties are not a random sample.1 5 Excluding non-reporting counties is equally problematic. Thus, we report both our results for the full data without capital/labor and our results on the partial data including capital/labor as an explanatory variable. 1 4 We create two variables each for white and female: one containing the true values and zeros for missing values, the other containing 1990 values if the true value is missing and zero otherwise. Estimation o f our model with these four variables yields significant negative coefficients on each o f these variables. Although the estimated variance is slightly larger for the filled values (possibly indicating greater measurement error), the general result supports the use o f the 1990 values where the true value is missing. 1 5 Reporting counties have significantly higher mean income, percent o f earning from manufacturing, proportion college educated population, population density, and are considerably more urban. 75 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Table 3-2. Socioeconomic Variables Variable Description Mean Std dev. college graduates Percent of the population with a college education (measured in 1990) 0.135 0.066 democrat Percent o f those voting in the Presidential election that voted for the Democratic candidate for president (1984,1988, 1992) 0.406 0.110 earnings Real earnings (thousands US$) 1302.373 5900.891 exposed Population density * toxicity weight 166028 4537217 female Percent of the population that is female 0.509 0.021 income Real per capita income (thousands US$) 17.619 4.207 capital/labor Ratio o f capital expenditures to labor force size in manufacturing (measured in thousands US$, data for 1992) 0.002 0.003 land Land area (hundreds o f square miles) 11.257 37.674 manufacturing Percent o f total county non-farm earnings from manufacturing 0.215 0.149 population density County population density (population/land area; thousands per square mile) 2.435 20.254 poverty Percent o f the population living in poverty (measured in 1990) 0.164 0.071 underage 18 Percent o f the population under the age o f 18 (measured in 1990) 0.269 0.036 unemployment Unemployment rate 0.068 0.034 urban Percent o f the population living in urban areas (measured in 1990) 0.365 0.299 vote Percent o f those eligible to vote that voted in the Presidential election (1984, 1988, 1992) 0.564 0.107 vote*democrat vote * democrat 0.227 0.069 white Percent o f the population that is white 0.887 0.161 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. IV. Results We test the null hypothesis that toxicity levels do not influence the income-pollution relationship. The pollution emission literature has not yet folly considered this relationship. Where this relationship receives attention, the literature often dismisses it as unimportant or addresses it only indirectly. Our theory, however, suggests a significant relationship between these measures does exist. Our full data sample includes 10 years o f toxic releases and demographic data. Given our panel time-series data structure, we use a traditional fixed effects estimation procedure estimating the model using feasible generalized least squares correcting for panel heteroskedasticity. We also estimated our model using a random effects procedure. The results, however, indicated zero variance in the random error component and thus the random effects model is inappropriate for our data. Therefore we present only the estimates from the fixed effects model. Our estimation results are presented in tables 3-3 and 3-5. Determinants o f Releases: Biochemical Characteristics Table 3 displays results for the estimation of the fixed effects model.1 6 Consistent with the literature, we find the inverted-U shaped relationship between releases and income. The coefficient on the interaction between income and toxicity tests Hypothesis 1. The estimated coefficient on income interacted with ln(l+toxicity) is positive and significant and the coefficient on income squared interacted with ln(l+toxicity) is negative and significant. As chemical toxicity increases, the quadratic relationship between income and release levels becomes more concave. 1 6 We also performed the same analysis on a reduced dataset (data for 1989-1996 only). A Chow test indicates that the estimates from the two subsets (1987-1988 and 1989-1996) are significantly different at the 5 percent level. The differences are mainly in coefficient magnitudes. With the exception o f the effect o f poverty, all coefficients have the same sign in both sets. Poverty has a 77 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Table 3-3. Fixed Effects, Feasible Generalized Least Squares Model o f Toxic Releases (correcting for panel heteroscedasticity), 1987-1996 Hypothesis Tested (and Expected Sign) Variable Coefficient Estimates'1 constant -1.821a -1.697 s literature (+) income (thousands) 0.146a © © technique eff. (-) income2 -2.80E-033 -2.90E-033 HI: Toxicity (+) income*ln( l+(tw)) 3.20E-033 3.11E-03a HI: Toxicity (-) income2 *ln( 1+(tw)) -9.83E-05a -9.77E-053 H2: Cancer (-) cancer -0.498a -0.483s H3: Exposure (-) exposed -7.48E-06 -1.85E-05 H4: Representation (-) white -0.0682a -0.173a H4: Representation (+) poverty 0.463a 0.185a H4: Representation (+) underage 18 1.133a 1.184a H4: Representation (-) vote -0.103a -0.029 HH4: Representation (?) vote*dem 0.336a 0.329s tech constraint (-) fiber -0.009 -0.006 tech constraint (-) liquid -0.243a -0.241s tech constraint (-) metal -0.276a -0.263s compos, effect (+) capital/labor 66.6a control variables manufacturing 1.47a 1.553s control variables urban 0.7503 0.844s control variables population density -0.005s -0.005s control variables unemployment 1.535a 1.769s control variables college graduates 0.282a 0.463s control variables female -0.907a -0.981s control variables land -8.90E-043 -8.67E-043 scale effect (+) earnings 5.08E-05a 5.19E-053 Observations 627002 639413 a = significant at the 1% level b = significant at the 5% level c = significant at the 10% level positive effect in the 1987-1988 period, but no significant effect in the later period. The results from the estimation on the reduced dataset are reported in the appendix. 78 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. The effect o f toxicity on the relationship between income and release levels is predicted, but not tested, by Antweiler, Copeland, and Taylor (2001). This outcome is illustrated in Table 3-4 by the income turning points (ITP, the per capita income level at which pollution starts to fall with increases in income) for various toxicity levels. As toxicity rises, the ITPs occur at lower income levels (Hypothesis 1). For further illustration, we estimate the ITPs for four o f the individual chemicals - chlorobenzene, hydrogen chloride, aluminum, and dimethyl sulfate with toxicities of 100, 1,000, 100,000, and 1,000,000 respectively. The ITPs for these four chemicals are $30,941 for chlorobenzene, $26,931 for hydrogen chloride, $24,050 for aluminum, and $21,197 for dimethyl sulfate. As the toxicity weight for the chemicals increases, the ITP falls. Note further that the ITPs we calculate for high toxicity pollutants is similar to that found by Arora and Cason (1999). The mean per capita income of counties in our sample is $17,619, indicating that the average county is in a situation in which increasing income leads to increasing levels o f pollution. We also find support for our other hypotheses. Five o f our chemicals are known or “reasonably accepted” as carcinogens. The coefficient on this dummy variable for carcinogens is negative and significant. That is, releases are lower when a chemical is a carcinogen. This result supports hypothesis 2 that the greater the disamenity the lower will be release levels. Hypothesis 3 predicts that releases that are likely to cause greater exposure result in significantly lower levels of releases. Our estimated coefficient on our exposure measure is not statistically significant, although the coefficient is negative as predicted in Hypothesis 3. Further, while the coefficient on our exposure measure in not significant, the coefficient on population density is significant and negative. We interpret these combined results to mean that as the population density increases (and thus the threat of exposure increases), releases decrease, but we find no differential effect o f toxicity on exposure. 79 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Table 3-4. Income turning points Toxicity weight (tw) Income Turning Point 0 $26,026.84 1 $25,794.60 10 $25,267.84 100 $24,663.99 1000 $24,119.04 10000 $23,638.90 100000 $23,213.95 1000000 $22,835.28 Finally, hypothesis 4 predicts that greater political activism reduces the level o f toxic releases within a county. Our estimates support this hypothesis, indicating that higher levels of voter participation are associated with lower toxic releases. We further find that higher rates of voting for the Democratic candidate in counties are associated with higher levels of toxic releases. A possible explanation for this finding is the support industrial workers often have for the Democratic Party as larger proportions of industrial workers are also associated with higher levels o f releases. Determinants o f Releases: Technological Constraints As we outline in section 2 of the paper, firms’ production functions also influence their release levels. We use a measure o f the capital to labor ratio in the manufacturing sector, an explanatory variable not previously used in association with the TRI data. We estimate a positive, significant coefficient for the capital to labor ratio. In international studies (Antweiler et al. (2001), Grossman and Krueger (1993)) demonstrate that capital-labor ratios exhibit highly significant positive effects. 80 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. “Dirtier” or more polluting industries tend to be more capital-intensive, while cleaner industries are labor-intensive. Thus, capital-labor ratios can be good indicators of pollution levels. As we explained in the data section, our capital-labor ratio data are highly censored. To account for the exclusion of 30 percent o f US counties from our analysis, we run a separate regression to estimate the model on the foil set o f counties without the capital to labor ratio variable. The results are reported in the last column of table 3-3. The significance for only one o f the estimated coefficients is affected by the exclusion o f capital/labor - the coefficient on vote is no longer significant. Control Variables We control for the effects of a number o f socioeconomic variables in our analysis. These explanatory variables coincide with many used in the literature, and allow us to control for differences across 3,141 US counties. Generally we find the same sign on our coefficients as Brooks and Sethi (1997). While Arora and Cason (1999) failed to find a significant effect of increased unemployment, we find that as unemployment rates increase, so do releases. We further find that releases are higher in counties with populations that have larger proportions o f young people (under 18 years o f age), non-whites, and people living in poverty. A possible explanation for these findings is that these groups may well be underrepresented in the political process. Finally, we find that chemicals in gaseous form have higher release levels than those in other physical states. 81 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Table 3-5. Release estimates over time 1987-1991 Coefficient 1987 1988 1989 1990 1991 income 0.227a 0.235a 0.257“ 0.264“ 0.282“ income2 -4.96E-03a -4.92E-03“ -4.92E-03 “ -4.89E-03“ -5.38E-03 “ income*ln( l+(tw)) 5.25E-03a 5.14E-03a 5.43E-03 “ 5.99E-03 “ 6.17E-03 “ Incomei *ln(l+(tw)) -1.07E-04a -9.95E-053 -1.25E-043 -1.40E-043 -1.52E-04 “ exposed 1.86E-04 7.76E-05 8.79E-05 6.89E-05 1.13E-04 cancer -0.639a -0.665“ -0.721“ -0.738“ -0.724 “ fiber -0.290a -0.275“ -0.261 “ -0.288 “ -0.300“ liquid -0.362a -0.384“ -0.309“ -0.330“ -0.299“ metal -0.827“ -0.840“ -0.786“ -0.859“ -0.888“ 1992-1996 Coefficient 1992 1993 1994 1995 1996 income 0.261“ 0.233“ 0.233“ 0.196“ 0.171“ income2 -4.83E-03a -4.31E-03“ -4.44E-03 “ -3.45E-03“ -2.96E-03 “ income*ln( 1 +(tw)) 6.22E-03 “ 5.58E-033 4.60E-03 “ 3.31E-03 “ 2.63E-03 “ Income2*ln(l+(tw)) -1.52E-04“ -1.23E-04“ -1.00E-043 -9.78E-05 “ -8.18E-05“ exposed 1.65E-04 1.63E-04 1.12E-054 1.91E-04 1.24E-04 cancer -0.743“ -0.725 “ -0.630“ -0.408 “ -0.325 “ fiber -0.275“ -0.258“ -0.358“ -0.517“ -0.557“ liquid -0.267“ -0.239“ -0.261“ -0.332“ -0.377“ metal -0.873“ -0.846“ -0.801“ -0.674“ -0.647“ a = significant at the 1% level b = significant at the 5% level c = significant at the 10% level 82 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Determinants o f Releases: Time Concerns exist in the literature over estimating TRI releases across time. First, measurement error in the TRI data is likely to vary over time. Second, it is possible that preferences for environmental quality (and thus the coefficients on income and toxicity in our model) change over time. We test for changes in our coefficients across time by running individual estimates for each year. Table 3-5 displays results for each o f the ten years o f data. In the interest o f space, we only include chemical characteristics in the table. We consistently find the EKC and greater concavity for more toxic chemicals and lower releases for carcinogens. While the sign and significance o f the coefficients on our income and income-toxicity interaction terms do not change over time, they do vary in magnitude. Finally, we note that the estimated effects of our remaining explanatory variables change significantly from year to year. Again, the differences (as with the reported estimates) are mainly in the magnitude of the coefficients rather than changes in sign or significance. V. Concluding Remarks The results that we present in this study support the idea that toxicity levels play an important role in explaining the level o f chemical releases. Not only do chemical characteristics have a significant effect on the level o f observed releases, but the toxicity of the release also affects the way in which income growth may impact pollution levels. Most significantly, we find that the environmental Kuznets curve is more concave for more toxic releases. The results imply that the transmission mechanism from consumer’s (who experience disutility from releases) to governments (who set regulation) to firms (who generate releases subject to regulation) is sensitive to the toxicity o f releases. The response o f the government in the release level determination mechanism is especially significant, because it is the link in the transmission mechanism in which we have the least confidence. Our model suggests the government responds to consumer demands with greater 83 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. regulation. Sensitivity of release levels to social, economic, and political pressures appears to be higher for more toxic chemicals. Additionally, we also demonstrate a relationship between release levels and capital-intensity of manufacturing production. Admittedly, our capital-intensity measure is a rough proxy for the capital-labor ratio in a county, but the highly significant, positive coefficient on this variable suggests that the inclusion o f such a measure is important in both national and international analyses. For now, we leave the construction of a better proxy for future research. Finally, we must be careful in extrapolating our results to other geographic areas and other chemical types. Our results can only be said to hold for U.S. data on toxic releases as reported in the TRI. With this restriction in mind, though, let us engage in a thought experiment. Suppose that these results were universally true - that the environmental Kuznets curve is more concave for more toxic chemicals. On the international level, this would imply that the findings o f Antweiler, Copeland, and Taylor (2001) - free trade leads to increases in environmental quality, when we measure quality in terms of sulfur dioxide emissions - might be more generally applicable. Antweiler et al. estimate the relative sizes o f the scale, technique, and composition effects o f freer trade on environmental quality. The scale effect measures the influence o f increased income on the scale of economic activity from trade on environmental quality. The technique effect also measures an effect o f income, but from increased income’s effect on consumer demand for environmental quality (and ability to afford cleaner production techniques). The composition effect reflects the effect o f a change in the composition o f production that occurs as a result o f freer trade. Antweiler et al. estimate that the technique effect dominates the scale effect and that the composition effect is negligible. Thus, if the inverted U-shaped relationship is more concave for more toxic releases, then the relative sizes of the scale and technique effects estimated by Antweiler, et al. would hold for releases that are more toxic than sulfur dioxide. For chemicals that are less toxic than sulfur dioxide, the technique effect may still dominate the scale effect, but this would occur at a higher level o f per 84 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. capital income than for sulfur dioxide. Again, while this thought experiment is purely speculative, if it were to hold this would have positive implications for our global environmental quality in the presence o f ever increasing trade liberalization. Proof o f this hypothesis is also left for future research. References Andreoni, James and Arik Levinson., 1998, “The Simple Analytics of the Environmental Kuznets Curve,” NBER Working Paper No. 6739. Antweiler, Werner, Brian Copeland and Scott Taylor., 2001, “Is Free Trade Good for the Environment?,” American Economic Review, 91, 877-908. Arora, Seema and Timothy Cason., 1995, “An Experiment in Voluntary Environmental Regulation: Participation in EPA’s 33/50 Program,” Journal o f Environmental Economics and Management, 271-286. 1999, “Do community characteristics influence environmental outcomes? Evidence from the toxic release inventory,” Southern Economics Journal, 65, 691-716. Bingham, Taylor, Donald Anderson and Philip Cooley., 1987, “Distribution o f the Generation of Air Pollution, " Journal o f Environmental Economics and Management, 30-40. Bowen, Howard R., 1943, “The Interpretation o f Voting in the Allocation o f Economic Resources,” Quarterly Journal o f Economics, 27-48. Brajer, Victor and Jane Hall., 1992, ”Recent Evidence on the Distribution of Air Pollution Effects,” Contemporary Policy Issues, 63-71. Brooks, Nancy and Rajiv Sethi, 1997, “The Distribution of Pollution: Community Characteristics and Exposure to Air Toxics,” Journal o f Environmental Economics and Management, 233- 250. Cavlovic, T., K. Baker, R. Berrens, and K. Gawande, 2000, “A Meta-Analysis of Environmental Kuznets Curve Studies.” Agricultural and Resource Economics Review, 29(1), 32-42. Chichilnisky, Graciela., 1994, “North-South Trade and the Global Environment,” American Economic Review, 851-874. Comanor, William S., 1976, “The Median Voter Rule and the Theory o f Political Choice,” Journal o f Public Economics, 169-177. Copeland, Brian., 1996, “Pollution content tariffs, environmental rent shifting, and the control of cross-border pollution,” Journal o f International Economics, 459-476. 85 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 1994, “International Trade and the Environment: Policy Reform in a Polluted Small Open Economy,” Journal o f Environmental Economics and Management, 44-65. Copeland, Brian and M. Scott Taylor, 1994, “North-South Trade and the Environment,” Quarterly Journal o f Economics, 755-787. 1995, “Trade and the Environment: A Partial Synthesis,” American Journal o f Agricultural Economics, 765-771. ., 1995, “Trade and Transboundary Pollution,” American Economic Review, 716-737. Dasgupta, Susmita, Benoit Laplante, and Craig Meisner, 1998, “Accounting for Toxicity Risks in Pollution Control: Does It Matter?,” World Bank Working Paper No. 2002. Dean, Judith., 1998, “Testing the Impact o f Trade Liberalization of the Environment: Theory and Evidence,” School o f Advanced International Studies, Johns Hopkins University Working Paper. ., 1992, “Trade and the Environment: A Survey o f the Literature,” in Patrick Low, ed., International trade and the environment. World Bank discussion Papers, Washington, DC: World Bank, 15-28. Dion, Catherine, Paul Lanoie and Benoit Laplante., 1998, “Monitoring of Pollution Regulation: Do Local Conditions Matter?,” Journal o f Regulatory Economics, 5-18. Esty, D., 2001, “Bridging the Trade-Environment Divide.” Journal o f Economic Perspectives, 15(3), 113-30. Gianessi, Leonard, Henry Peskin and Edward Wolff., 1979, “The Distributional Effects o f Uniform Air Pollution Policy in the United States,” Quarterly Journal o f Economics, 281-301. Gray, Wayne and Mary Deily., 1996, “Compliance and Enforcement: Air Pollution Regulation in the U.S. Steel Industry,” Journal o f Environmental Economics and Management, 96-111. Gray, Wayne and Ronald Shadbegian., 1997, “Environmental Regulation, Investment Timing, and Technology Choice,” NBER Working Paper No. 6036. Greene, William H., 2000, Econometric Analysis Fourth Edition, Prentice-Hall, New Jersey. Grossman, Gene and Alan Krueger., 1993, “Environmental Impacts of a North American Free Trade Agreement,” in P. Garber, ed., The U.S.-Mexico Free Trade Agreement, Cambridge, MA: MIT Press, 13-56. 1995, “Economic Growth and the Environment,” Quarterly Journal o f Economics, 353-377. Hamilton, James., 1995, “Pollution as News: Media and Stock Market Reactions to the Toxic Release Inventory Data,” Journal o f Environmental Economics and Management, 98-113. 86 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Hamilton, Sally, Hoje Jo and Meir Statman., 1993, “Doing well while doing good? The investment performance o f socially responsible mutual funds,” Financial Analysts Journal, 62-66. Harrison, David Jr. and Daniel Rubinfeld., 1978, “The Distribution o f Benefits from Improvements in Urban Air Quality,” Journal o f Environmental Economics and Management, 313-332. Helland, Eric and Andrew Whitford., 2000, “Pollution Incidence and Political Jurisdiction: Evidence from the TRI,” Working Paper. Jaffe, Adam, et al., 1995, “Environmental Regulation and the Competitiveness ofU .S. Manufacturing: What does the evidence tell us?,” Journal o f Economic Literature, 132- 163. Levinson, Arik., 1999, “State taxes and interstate hazardous waste shipments,” American Economic Review, 89(3), 666-677. Seldon, Barry, et al., 1994, “The effect of EPA enforcement funding on private-sector pollution- control investment,” Applied Economics, 949-955. Seldon, Thomas and Daqing Song., 1994, “Environmental Quality and Development: Is there a Kuznets Curve for Air Pollution Emissions?,” Journal o f Environmental Economics and Management, 147-162. Terry, Jeffrey and Bruce Yandle., 1997, “EPA’s Toxic Release Inventory: Stimulus and Response,” Managerial and Decision Economics, 433-441. Tobey, James A., 1990, “The Effects o f Domestic Environmental Policies on Patterns o f World Trade: An Empirical Test,” Kyklos, 191-209. US Department o f Commerce (Bureau o f Census), USA Counties 1998 CD-ROM. _., USA Counties 1994 CD-ROM. Current Population Reports, P20-523RV: Voting and Registration in the Election of November 1998, http://www.census.gov/prod/www/abs/vote.html. US Department o f Commerce (Bureau o f Economic Analysis), Regional Economic Information System (REIS) 1969-1997 CD-ROM. US Department o f Health and Human Services, 8th Report on Carcinogens: 1998 Summary, National Toxicity Program, http://nto- server.niehs.nih.gov/Main pages/NTP 8RoC pg.html. downloaded March 5, 2001. US EPA, Office o f Enforcement and Compliance Assurance, Sector Facility Indexing Project, Science Advisory Board Review Documents, http://es.epa.gov/oeca/metd/sabrev.html. downloaded October 5, 2000. 87 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Chapter 4 Conclusion R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Summary Through the chapters o f this dissertation we investigate the role o f information as regulation in capital markets and in the environment. We show that trade policy can be used to reduce the incidence o f an inefficiency in the capital market. In so doing, we demonstrate an alternative, productive use for lobbying - in contrast to its more common ‘DUP’ role. We also use data from the Toxic Release Inventory, which was developed to provide US citizens with greater information about the quality o f their surrounding environment. TRI data can and has been used by citizens and firms to make consumption and production decisions. Our utility theoretic model o f toxic releases illustrates the mechanism through which citizens’ demand for environmental quality is balanced against their desire for consumption goods. This demand (in the face o f environmental quality information) can be translated into consumer activism aimed at changing government regulation of toxic releases. Finally, government regulation shapes firm production decisions. We show that, for the US, the relationship between per capita income and pollution exhibits the standard quadratic relationship common in the literature and that this relationship is more concave for more toxic pollutants. Future Research These papers raise as many questions (if not more) as they answer. Each paper can be extended in a number o f ways and also are related to other interesting avenues for future research. Below are just a few o f the many opportunities open for future work. Trade There are some direct extensions o f chapter 2 that are of interest. First, chapter 2 investigates the role o f tariffs in reducing the incidence o f capital market inefficiencies. The section of the trade code to which chapter 2 refers also allows for relief in the form o f quotas. As such, a natural 89 R eproduced with perm ission o f the copyright owner. Further reproduction prohibited without perm ission. extension would be to illustrate the use o f quotas in the framework o f chapter 2. Second, the model does not currently incorporate the institutional process through which industries lobby. Extending the model to explicitly include the lobbying process would also be o f interest. Third, chapter 2 holds the foreign government as a passive agent. A considerable literature argues (convincingly) that foreign government may engage in retaliation in light o f protectionist actions by the domestic government. Allowing for strategic interaction between the governments would be an interesting and highly relevant extension to chapter 2. Related to the third extension above is the current literature on the optimal design o f trade institutions. This literature observes the significant variation in the design o f trade institutions and seeks to find explanations for it (see for example Rosendorff and Milner, 2001). That is, why do the safeguard, antidumping, and countervailing duties clauses exist from an international point of view. This literature seeks answers at an international level as opposed to the domestic level investigated in chapter 2. This line o f research would provide additional reasons for the escape clauses studied here. Finally, empirical support for the results found in chapter 2 would also be a promising avenue for future research. Investigation of stock market return affects for companies appealing for safeguard (and other escape clause) protection as well as capital market responses to new equity issues for investment purposes are of significant interest. Further, empirical investigations regarding the magnitude of lobbying costs ( K ) over time might suggest the deliberate shifting o f K for strategic purposes. Environment One possible extension of this dissertation considers the EKC from the consumption-side. In part, the production-side focus to the EKC literature assumes that production is mobile, but ignores the mobility of consumers. This assumption has less validity in domestic analysis than in international analysis given the considerable domestic labor mobility in most countries. Since our 90 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. study uses domestic pollution data, a natural extension would examine the possibility o f a consumption-driven EKC for the United States. Using data from the Toxic Release Inventory, estimates o f the relationship between the level of toxic releases and interregional migration in the U.S. would indicate the potential presence of consumption-side effects. Estimating migration as a function of releases would add the influence of pollution on the migration decisions of individuals to our current model. We hypothesize that the relatively mobile labor force in the U.S. would migrate away from areas with high levels of toxic releases. This extension is related to Gawande et al. (2000) who estimate the effect o f hazard site build-up on migration. Gawande et al. use hazard waste sites as their pollution measure and estimates inter-county migration. An extension using TRI data would reflect a more prevalent pollution problem (manufacturing releases). We also propose examination o f these relationships at the zip code level. By using a lower level o f aggregation even more o f the influence o f pollution on migration would be captured. Another avenue o f research could examine the long-term effect o f public disclosure programs (publicizing toxic releases for individual firms) on firm environmental performance. Hamilton (1995) and Konar and Cohen (1997) demonstrate that the initial report o f the Toxic Release Inventory in 1989 (and firms’ subsequent identification by the media in relation to their reported environmental performance in the TRI) resulted in significant negative abnormal returns for firms with poor environmental performance. Konar and Cohen (1997) also show that the forty firms with the largest negative abnormal returns improved their environmental performance significantly more than the “average” firm in their respective industries. These results are not surprising and they affirm the role that information provision can play in environmental regulation. While the initial effect o f public disclosure programs has been demonstrated, whether subsequent annual reports continue to provide the incentive to firms to improve their environmental performance is unproven. Assuming that capital markets function according to the efficient markets hypothesis, in order for subsequent TRI reports to have an effect on firm environmental performance (through capital market incentives) they must provide “new” information to the market If the initial TRI report in 91 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 1989 established environmental reputations for firms, then unless subsequent TRI reports provide information that changes these reputations the reports will have little to no effect on the stock returns. Further if public disclosure programs affect firm environmental performance through capital market effects, then without “new” information the reports will have no effect on firm environmental performance. As many countries adopt and rely on public disclosure programs as part o f their environmental policy,1 the long-term efficacy o f these programs becomes increasingly important. We propose investigating the effects of TRI reports in 1990-1996 to determine whether the information effects found by Konar and Cohen were maintained for later reports. These extensions are only a few o f the many research opportunities related to the work presented here. Pursuit o f these extensions would enrich our understanding o f economics especially with regard to the use o f information as a regulatory tool. References Gawande, K., A.K. Bohara, R.P. Berrens, and P. Wang, 2000, “Internal migration and the environmental Kuznets curve for US hazardous waste sites,” Ecological Economics, 33, 151-166. Hamilton, James., 1995, “Pollution as News: Media and Stock Market Reactions to the Toxic Release Inventory Data,” Journal o f Environmental Economics and Management, 98-113. Konar, S. and M.A. Cohen, 1997, “Information as Regulation: The Effect o f Community Right to Know Laws on Toxic Emissions,” Journal o f Environmental Economics and Management, 32, 109-124. Rosendorff, Peter and Helen Milner, 2001, “The Optimal Design o f International Trade Institutions: Uncertainty and Escape,” USC Center for Law, Economics, and Organization Research Paper CO 1-9. 1 Public disclosure programs have been adopted by a number o f countries including Canada, Mexico, the UK, Australia, the Philippines, Indonesia, Argentina, and Chile. 92 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Bibliography R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Akerlof, G.A.. “The Market for ‘Lemons’: Quality and the market mechanism,” Quarterly Journal o f Economics, 1970, 84,488-500. Andreoni, James and Arik Levinson., 1998, “The Simple Analytics o f the Environmental Kuznets Curve,” NBER Working Paper No. 6739. Antweiler, Werner, Brian Copeland and Scott Taylor, 2001, “Is Free Trade Good for the Environment?,” American Economic Review, 91, 877-908. Arora, Seema and Timothy Cason, 1995, “An Experiment in Voluntary Environmental Regulation: Participation in EPA’s 33/50 Program,” Journal o f Environmental Economics and Management, 271-286. ., 1999, “Do community characteristics influence environmental outcomes? Evidence from the toxic release inventory,” Southern Economics Journal, 65, 691-716. Bagwell, Kyle and Robert Staiger, 1988, “The Role o f Export Subsidies when Product Quality is Unknown,” Journal o f International Economics, 27, 69-89. Baldwin, Richard. “Politically Realistic Objective Functions and Trade Policy,” Economics Letters, 1987, 24,287-90. Baldwin, Robert E. “The Case against Infant-Industry Tariff Protection,” Journal o f Political Economy, 1969, LXXVII, 295-305. Barclay, Michael and Robert Litzenberger, 1988, “Announcement Effects o f New Equity Issues and the Use o f Intraday Price Data,” Journal o f Financial Economics, 21, 71-99. Bhagwati, J. N. “Directly-unproductive, profit seeking (DUP) activities,” Journal o f Political Economy, October 1982, 90, 988-1002. Bingham, Taylor, Donald Anderson and Philip Cooley., 1987, “Distribution o f the Generation of Air Pollution, "Journal ofEnvironmental Economics and Management, 30-40. Bowen, Howard R , 1943, “The Interpretation o f Voting in the Allocation o f Economic Resources,” Quarterly Journal o f Economics, 27-48. Brajer, Victor and Jane Hall., 1992, "Recent Evidence on the Distribution of Air Pollution Effects,” Contemporary Policy Issues, 63-71. Brander, James and Barbara Spencer. “Tariff Protection and Imperfect Competition,” Monopolistic Competition and International Trade, ed. H. Kierzkowski, Oxford University Press, 1984. Brooks, Nancy and Rajiv Sethi, 1997, “The Distribution o f Pollution: Community Characteristics and Exposure to Air Toxics,” Journal o f Environmental Economics and Management, 233- 250. Cason, Timothy and Lata Gangadharan, 2002, “Environmental Labeling and Incomplete Consumer Information in Laboratory Markets,” Journal o f Environmental Economics and Management, 43(1), 113-134. 94 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Cavlovic, T.A., K.H. Baker, R.P. Berrens, and K. Gawande, 2000, “A Meta-Analysis of Environmental Kuznets Curve Studies,” Agricultural Resource and Economic Review, 29(1), 32-42. Chichilnisky, Graciela., 1994, “North-South Trade and the Global Environment,” American Economic Review, 851-874. Comanor, William S., 1976, “The Median Voter Rule and the Theory o f Political Choice,” Journal o f Public Economics, 169-177. Copeland, Brian., 1996, “Pollution content tariffs, environmental rent shifting, and the control of cross-border pollution,” Journal o f International Economics, 459-476. ., 1994, “International Trade and the Environment: Policy Reform in a Polluted Small Open Economy,” Journal o f Environmental Economics and Management, 44-65. Copeland, Brian and M. Scott Taylor, 1994, “North-South Trade and the Environment,” Quarterly Journal o f Economics, 755-787. ., 1995, “Trade and the Environment: A Partial Synthesis,” American Journal o f Agricultural Economics, 765-771. ., 1995, “Trade and Transboundary Pollution,” American Economic Review, 716-737. Dasgupta, Susmita, Benoit Laplante, and Craig Meisner, 1998, “Accounting for Toxicity Risks in Pollution Control: Does It Matter?,” World Bank Working Paper No. 2002. Dean, Judith., 1998, “Testing the Impact o f Trade Liberalization o f the Environment: Theory and Evidence,” School of Advanced International Studies, Johns Hopkins University Working Paper. ., 1992, “Trade and the Environment: A Survey o f the Literature,” in Patrick Low, ed., International trade and the environment. World Bank discussion Papers, Washington, DC: World Bank, 15-28. Dion, Catherine, Paul Lanoie and Benoit Laplante., 1998, “Monitoring o f Pollution Regulation: Do Local Conditions Matter?,” Journal o f Regulatory Economics, 5-18. Dixit, Avinash. “Trade and Insurance with Moral Hazard,” Journal o f international Economics, 1987,23, 201-220. Dixit, Avinash. “Trade and Insurance with Imperfectly Observed Outcomes,” Quarterly Journal o f Economics, 1989a, 104, 195-203. Dixit, Avinash. “Trade and Insurance with Adverse Selection,” Review o f Economics Studies, 1989b, 56,235-247. Donnenfield, Shabtai, Shlomo Weber, and Uri Ben-Zion, 1985, “Import controls under imperfect information,” Journal o f International Economics, 19, 341-54. 95 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Eaton, Jonathan and Gene Grossman. “Tariffs as Insurance: Optimal Commercial Policy when Domestic Markets are Incomplete,” Canadian Journal o f Economics, 1985, 18(2), 258- 272. Esty, D., 2001, “Bridging the Trade-Environment Divide.” Journal o f Economic Perspectives, 15(3), 113-30. Fama, Eugene. “Efficient Capital Markets: Review of Theory and Empirical Work,” Journal o f Finance, 1970, 25, 386-416. Feenstra, Robert and Tracy Lewis. “Negotiated Trade Restrictions with Private Political Pressure,” Quarterly Journal o f Economics, November 1991, 106(4), 1287-308. Feinberg, R.M. and B.T. Hirsch. “Industry Rent Seeking and the Filing o f ‘Unfair Trade’ Complaints,” International Journal o f Industrial Organization, 1989, 7, 325-340. Gawande, K., A.K. Bohara, R.P. Berrens, and P. Wang, 2000, “Internal migration and the environmental Kuznets curve for US hazardous waste sites,” Ecological Economics, 33, 151-166. Gianessi, Leonard, Henry Peskin and Edward Wolff., 1979, “The Distributional Effects o f Uniform Air Pollution Policy in the United States,” Quarterly Journal o f Economics, 281-301. Gray, Wayne and Mary Deily., 1996, “Compliance and Enforcement: Air Pollution Regulation in the U.S. Steel Industry,” Journal o f Environmental Economics and Management, 96- 111. Gray, Wayne and Ronald Shadbegian., 1997, “Environmental Regulation, Investment Timing, and Technology Choice,” NBER Working Paper No. 6036. Greene, William H., 2000, Econometric Analysis Fourth Edition, Prentice-Hall, New Jersey. Grossman, Gene and Henrik Horn, 1988, “Infant-industry protection reconsidered: The case o f informational barriers to entry,” Quarterly Journal o f Economics, 103, 767-87. Grossman, Gene and Alan Krueger., 1993, “Environmental Impacts o f a North American Free Trade Agreement,” in P. Garber, ed., The U.S.-Mexico Free Trade Agreement, Cambridge, MA: MIT Press, 13-56. ., 1995, “Economic Growth and the Environment,” Quarterly Journal o f Economics, 353-377. Hagen, Everett E. “An Economic Justification o f Protectionism,” Quarterly Journal o f Economics, 1958, LXXII, 496-514. “Reply” to the “Comment” on “An Economic Justification o f Protectionism” by A. Koo in the same journal, Quarterly Journal o f Economics, 1958, LXXV, 145-151. H am ilton, James., 1995, “Pollution as News: Media and Stock Market Reactions to the Toxic Release Inventory Data,” Journal o f Environmental Economics and Management, 98-113. 96 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Hamilton, Sally, Hoje Jo and Meir Statman, 1993, “Doing well while doing good? The investment performance o f socially responsible mutual funds,” Financial Analysts Journal, 62-66. Harris, Milton and Artur Raviv. “The Theory of Capital Structure,” Journal o f Finance, 1991, 46, 297-355. Harrison, David Jr. and Daniel Rubinfeld., 1978, “The Distribution of Benefits from Improvements in Urban Air Quality,” Journal o f Environmental Economics and Management, 313-332. Helland, Eric and Andrew Whitford., 2000, “Pollution Incidence and Political Jurisdiction: Evidence from the TRI,” Working Paper. Jacobson, Robert and David Aaker, 1993, “Myopic management behavior with efficient, but imperfect, financial markets,” Journal o f Accounting and Economics, 16,383-405. Jaffe, Adam, et al., 1995, “Environmental Regulation and the Competitiveness o f U.S. Manufacturing: What does the evidence tell us?,” Journal o f Economic Literature, 132- 163. Jensen, M.C. and W.H. Meckling. “Theory o f the Firm: Managerial Behavior, Agency Costs and Ownership Structure,” Journal o f Financial Economics, 1976, 3, 305-360. Johnson, Harry G. “Optimal Trade Intervention in the Presence o f Domestic Distortions,” Trade Growth, and the Balance o f Payments: Essays in Honor o f Gottfried Haberler, 1965, 3-34. Konar, S. and M.A. Cohen, 1997, “Information as Regulation: The Effect o f Community Right to Know Laws on Toxic Emissions,” Journal o f Environmental Economics and Management, 32, 109-124. Koo, Anthony Y.C. “An Economic Justification o f Protectionism: Comment,” Quarterly Journal o f Economics, 1961, LXXV, 133-44. Leland, H.E. and D.H. Pyle. “Informational Asymmetries, Financial Structure, and Financial Intermediation,” Journal o f Finance, 1977, 32, 371-387. Levinson, Arik., 1999, “State taxes and interstate hazardous waste shipments,” American Economic Review, 89(3), 666-677. Lipsey, R.G. and K. Lancaster. “The General Theory o f the Second Best,” Review o f Economic Studies, 1956, 24, 11-32. Mayer, Wolfgang, 1984, “The infant-export industry argument,” Canadian Journal o f Economics, 17, 249-69. Modigliani, Franco and M.H. Miller. “The Cost o f Capital, Corporate Finance and the Theory of Investment,” American Economic Review, 1958,48,261-297. Myers, S.C. and N.S. Majluf, 1984, “Corporate Financing and Investment Decisions when Firms have Information that Investors do not Have, " Journal o f Financial Economics, 13, 187- 221. 97 R eproduced with perm ission o f the copyright owner. Further reproduction prohibited without perm ission. Newbery, David and Joseph Stiglitz. “Pareto Interior Trade,” Review o f Economics Studies, 1984, 51, 1-12. Peter, J.P. “Harley-Davidson, Inc. - Motorcycle Division, “ in S.C. Certo and J.P. Peter (Eds). Selected Cases in Strategic Management, New York: McGraw-Hill, 461-475. Prusa, T.J. “Why are so many antidumping petitions withdrawn?,” Journal o f International Economics, 1992,33,1-20. Raff, Horst and Young-Han Kim, 1999, “Optimal export policy in the presence o f informational barriers to entry and imperfect competition,” Journal o f International Economics, 49, 99- 123. Rosendorff, Peter and Helen Milner, 2001, “The Optimal Design of International Trade Institutions: Uncertainty and Escape,” USC Center for Law, Economics, and Organization Research Paper C01-9. Seldon, Barry, et al., 1994, “The effect of EPA enforcement funding on private-sector pollution- control investment,” Applied Economics, 949-955. Seldon, Thomas and Daqing Song., 1994, “Environmental Quality and Development: Is there a Kuznets Curve for Air Pollution Emissions?,” Journal o f Environmental Economics and Management, 147-162. Spence, Michael. “Job Market Signaling,” Quarterly Journal o f Economics, 1973, 87, 355-374. Stein, Jeremy, 1989, “Efficient capital markets, inefficient firms: A model of myopic corporate behavior,” Quarterly Journal o f Economics, 103, 655-69. Terry, Jeffrey and Bruce Yandle., 1997, “EPA’s Toxic Release Inventory: Stimulus and Response,” Managerial and Decision Economics, 433-441. Tobey, James A., 1990, “The Effects of Domestic Environmental Policies on Patterns o f World Trade: An Empirical Test,” Kyklos, 191-209. US Department o f Commerce (Bureau o f Census), USA Counties 1998 CD-ROM. ., USA Counties 1994 CD-ROM. ., Current Population Reports, P20-523RV: Voting and Registration in the Election of November 1998, http://www.census.gov/prodywww/abs/vote.html. US Department o f Commerce (Bureau of Economic Analysis), Regional Economic Information System (REIS) 1969-1997 CD-ROM. US Department o f Health and Human Services, 8th Report on Carcinogens: 1998 Summary, National Toxicity Program, httn://ntp- server.niehs.nih.gov/Main pages/NTP 8RoC pg.html. downloaded March 5, 2001. 98 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. US EPA, Office o f Enforcement and Compliance Assurance, Sector Facility Indexing Project, Science Advisory Board Review Documents, http://es.epa.gov/oeca/metd/sabrev.html. downloaded October 5, 2000. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Appendices R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Appendix to Chapter 2 The second condition for the separating equilibrium in lobbying, after some rearranging, yields the following condition for maintenance o f the efficient equilibrium. This condition is a quadratic function in K (call it f { K )), and reveals that the condition is satisfied in equality when 2 = ~ \m ^ hi (t)+ (l - m ) * h i (°)+ 1 + it) - 7 1 l i (°))±r] where r= (fma {t) +(l-ju)7rM(0)+I+^ 7vLr{ t ) - x u (oj)J Y i Let this lower bound be represented by F mn, = \ I ')**m (O+O - Ffrm (0)+/ + n(nu if)~^u (< >))± y\ Figure 2-6. Sketch o f f ( K } 101 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. When f ( K ) > 0 , the efficient equilibrium fails. Thus, for any filing cost outside the range ( £ 1 ™ , K f n ), where = \f in m (*)+ (l - j u } t m (o)+/ + ju (tv l i { t ) - 7 l u (0))— y ] , and K™n = \jlTCm ( / ) + ( l - //);X m (0) + / + }l{7tL I ('t) — 7tu (0)) + / ] , the managers o f the domestic industry defect from the efficient strategy set. Thus, the appropriate value to consider for the lower bound on the filing cost is . Let X 'm m = K ™ n . R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Appendix to Chapter 3 We report the estimation results for 1989-1996 in Table 3-6. The sets o f coefficients are significantly different from the 1987-1988 estimates at the 5 percent level. With the exception of the coefficients on fiber, white, and vote the differences in the estimates with and without the capital-labor ratio are in magnitude not sign or significance. The coefficients on fiber, white and vote differ in their significance with fiber’s coefficient becoming significant and white and vote’s coefficients exhibiting a lower level o f significance (significant at the 5 and 10 percent levels instead of at the 1 percent level). 103 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Table 3-6. Fixed Effects, Feasible Generalized Least Squares Model o f Toxic Releases (correcting for panel heteroscedasticity), 1989-1996 Hypothesis Tested (and Expected Sign) Variable Coefficient Estimates'1 constant -1.821a -1.577 s literature (+) income (thousands) 0.15 l a 0.153s technique eff. (-) income2 -2.84E-03a -2.9 IE-033 HI: Toxicity (+) income *ln( 1+(tw)) 2.98E-033 2.90E-033 HI: Toxicity (-) income2*ln( l+(tw)) -9.3 IE-053 -9.27E-05a H2: Cancer (-) cancer -0.4803 -0.464s H3: Exposure (-) exposed -4.75E-06 -1.45E-05 H4: Representation (-) white -0.05 l b -0.166s H4: Representation (+) poverty 0.5203 0.235s H4: Representation (+) under age 18 1.2603 1.273s H4: Representation (-) vote -0.2003 -0.095b HH4: Representation (?) vote*dem 0.4253 0.387s tech constraint (-) fiber -0.017b -0.013° tech constraint (-) liquid -0.2323 -0.229s tech constraint (-) metal -0.2593 -0.245s compos, effect (+) capital/labor 65.9a control variables manufacturing 1.4623 1.273s control variables urban 0.7403 0.837s control variables population density -0.0043 -0.005s control variables unemployment 1.388s 1.634s control variables college graduates 0.200s 0.368s control variables female -1.080s -1.327s control variables land -8.46E-043 -8.78E-043 scale effect (+) earnings 4.93E-053 5.04E-053 Observations 500826 511998 a = significant at the 1% level b = significant at the 5% level c = significant at the 10% level 104 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Managerial incentives and financial markets

PDF

Three essays on asymmetric information and public policy: Information, the environment, and the provision of defense goods

PDF

Leading indicators, institutions, and growth: An application to emerging markets in the Middle East

PDF

The role of information in economic and public policies: Empirical applications for measuring human capital and water regulatory contracts

PDF

Organizational representativeness and performance

PDF

Glutathione metabolism and transport in cystic fibrosis airway epithelia

PDF

The Japanese main bank system: A transaction cost approach

PDF

Two essays on time series econometrics: Multiple time series analysis of the Asian crisis and estimation and inference of the nonstationary structural vector autoregressive model

PDF

Neuro-fuzzy systems with nonuniform spectral structure for multichannel audio synthesis and delivery

PDF

An adaptive temperament -based information filtering method for user -customized selection and presentation of online communication

PDF

Spatial convergence patterns in the presence of information technology

PDF

Regulation of RNA polymerase I transcription by CK2 and TAF1

PDF

Regulatory agenda -setting and environmental risk

PDF

Prescription drug profiles as health risk adjusters in capitated payment systems: An applied econometric analysis

PDF

The role of the Korean middle class in democratic transition

PDF

Political economy of financial liberalization in emerging markets: A comparative study of South Korea and Taiwan in the 1990s

PDF

In and out of marriage: A study of the multiply divorced

PDF

Spectroscopy and photodissociation dynamics of the hydroxymethyl radical (CH(2)OH)

PDF

Empirical essays on exchange rate economics

PDF

The relationship between collective efficacy and task interdependence

Asset Metadata

Creator Emerson, Tisha Lin Nakao (author)

Core Title Regulation through information: Capital markets and the environment

Degree Doctor of Philosophy

Degree Program Economics

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

Tag Economics, Finance,economics, general,environmental sciences,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-247936

Unique identifier UC11339385

Identifier 3093756.pdf (filename),usctheses-c16-247936 (legacy record id)

Legacy Identifier 3093756.pdf

Dmrecord 247936

Document Type Dissertation

Rights Emerson, Tisha Lin Nakao

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