THE WORLD BANK ECONOMIC REVIEW Voluime 11 September- 1 997 Number . i 5- ISSN 0>58 6i70 THE WORLD BANK ECONOMIC REVIEW EDITOR Moshe Syrquin CONSULTING EDITOR Sandra Gain EDITORIAL BOARD Kaushik Basu, Cornell University and University of Delhi David Dollar Francois Bourguignon, DELTA, Paris Gregory K. Ingram Willem H. Buiter, University of Cambridge Lant H. Pritchett Gerald K. Helleiner, University of Toronto Jacques van der Gaag Mark R. Rosenzweig, University of Pennsylvania John Williamson L. Alan Winters The World Bank Economic Review is a professional journal for the dissemination of World Bank- sponsored research that informs policy analyses and choices. It is directed to an international readership among economists and social scientists in government, business, and international agencies, as well as in universities and development research institutions. The Review emphasizes policy relevance and opera- tional aspects of economics, rather than primarily theoretical and methodological issues. It is intended for readers familiar with economic theory and analysis but not necessarily proficient in advanced mathematical or econometric techniques. Articles will illustrate how professional research can shed light on policy choices. Inconsistency with Bank policy will not be grounds for rejection of an article. Articles will be drawn primarily from work conducted by World Bank staff and consultants. Before being accepted for publication by the Editorial Board, all articles are reviewed by two referees who are not mem- bers of the Bank's staff and one World Bank staff member; articles must also be recommended by at least one external member of the Editorial Board. The Review may on occasion publish articles on specified topics by non-Bank contributors. Anv reader interested in preparing such an article is invited to submit a proposal of not more than two pages in length to the Editor. The views and interpretations expressed in this journal are those of the authors and do not necessarilv represent the views and policies of the World Bank or of its Executive Directors or the countries they rep- resent. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility whatsoever for any consequences of their use. When maps are used, the bound- aries, denominations, and other information do not imply on the part of the World Bank Group any judg- ment on the legal status of any territory or the endorsement or acceptance of such boundaries. Comments or brief notes responding to Review articles are welcome and will be considered for publica- tion to the extent that space permits. Please direct all editorial correspondence to the Editor, The World Bank Economic Review, The World Bank, Washington, D.C. 20433, U.S.A. The World Bank Economic Review is published three times a year (January, Mav, and September) bv the World Bank. Single copies may be purchased at $12.95. Subscription rates are as follows: Individuals Institutions 1-year subscription US$30 US$50 2-year subscription US$55 US$95 3-year subscription US$70 US$130 Orders should be sent to: World Bank Publications, Box 7247-7956, Philadelphia, PA 19170-7956, U.S.A. Subscriptions are available without charge to readers with mailing addresses in developing countries and in socialist economies in transition. Written request is required every two years to renew such subscriptions. D 1997 The International Bank for Reconstruction and Development / THE WORLD BANK 1818 H Street, N.W., Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America ISBN 0-8213-3783-1; 15SN 0258-6770 Material in this journal is copyrighted. Requests to reproduce portions of it should be sent to the Office of the Publisher at the address in the copyright notice above. The World Bank encourages dis- semination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to make photocopies is granted through the Copyright Clearance Center, Suite 910, 222 Rosewood Drive, Danvers, MA 01923, U.S.A. This journal is indexed regularly in Current Contents/Social & Behavioral Sciences, Index to Interna- tional Statistics, Journal of Economic Literature, Public Affairs Information Service, and Social Sciences Citation Index. It is available in microform through University Microfilms, Inc., 300 North Zeeb Road, Ann Arbor, MI 48106, U.S.A. THE WORLD BANK ECONOMIC REVIEW Volume 11 September 1997 Number 3 Rationing Can Backfire: The "Day without a Car" in Mexico City 383 Gunnar S. Eskeland and Tarhan Feyzioglu Prices and Protocols in Public Health Care 409 Jeffrey S. Hammer Formal and Informal Regulation of Industrial Pollution: 433 Comparative Evidence from Indonesia and the United States Sheoli Pargal, Hemamala Hettige, Manjula Singh, and David Wheeler Capital Flows to Developing Countries: Long- and Short-Term 451 Determinants Mark P. Taylor and Lucio Sarno Determinants of the Export Structure of Countries 471 in Central and Eastern Europe Bernard Hoekman and Simeon Djankov A NEW DEVELOPMENT DATA BASE A New Data Base on State-Owned Enterprises 491 Luke Haggarty and Mary M. Shirley Index of Authors for Volume 11 515 Index of Titles for Volume 11 517 List of Referees 519 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3: 3 83-408 Rationing Can Backfire: The "Day without a Car" in Mexico City Gunnar S. Eskeland and Tarhan Feyzioglu A ban restricting each car from driving on a specified weekday is found to have in- creased total driving in Mexico City. Because of the ban, cars effectively represent -driv- ing permits," and some households have bought an additional car and increased their driving. Greater use of old cars, congestion effects, and increased weekend driving may also have contributed to the disappointing results. The ban has high welfare costs and does not deliver the intended benefits of reduced driving-quite the contrary. The experience provides an interesting lesson in applied welfare economics. Theory indicates that this is a costly way of reducing traffic and pollution. But the finding that the strategy is counterproductive could be made only with applied quantitative analysis. In November 1989 the Mexico City administration imposed a regulation ban- ning each car from driving a specific day of the week. Called Hoy no circula (this one does not circulate today), the "Day without a Car" regulation specifies that cars with license plate numbers ending with digits 0 or 1 do not drive on Mon- day, 2 or 3 do not drive on Tuesday, and so on. Restrictions do not apply on weekends. The regulation applies to all cars (except those of the fire depart- ment) and thus to firms as well as households. We use the term household for simplicity. Compliance is generally believed to be high: the police are visible and fines are heavy. The regulation has been both popular and controversial. Some people argue that it places a reasonable burden on car owners in order to alleviate congestion and pollution problems. Others argue that it is inefficient and unfair: inefficient in the way most rationing devices are inefficient; unfair because it is more easily avoided or accommodated by some than by others. Finally, some people are arguing that the regulation is counterproductive, actually increasing the levels of congestion and pollution because many households have purchased additional cars to circumvent the ban. This article analyzes the policy pragmatically. Section I shows how the results of rationing can be compared with those that would be obtained using market- Gunnar S. Eskeland is with the Development Research Group at the World Bank, and Tarhan Feyzioglu is with the European I Department at the International Monetary Fund. The authors thank Jorge Daude Balmer, Sergio Sanchez, Rodolfo Lasi, Gabriel Vera, and Lynda Baynham for help with making data available and Shanta Devarajan, Kenneth A. Small, and the anonymous referees for discussion and helpful comments. C 1997 The International Bank for Reconstruction and Development / THE WORLD BANK 383 384 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 based instruments. We illustrate why rationing entails welfare costs at least as high as those of a market-based mechanism producing the same reduction in trips. Section II presents an empirical framework for estimating the reductions in demand induced by the regulation. A model of gasoline demand is estimated using aggregate time-series data from before the regulation. This model is used to simulate a counterfactual scenario for demand in subsequent periods as if the regulation had not been introduced. The results of these simulations motivate further examination of three aspects of car ownership and trip generation. First, due to the integer nature of cars and the fact that cars effectively come bundled with "workday driving permits," households may want an additional car once theirs is made less useful by the regulation. Second, multiple drivers in a household could mean that total car use increases even though an additional car is purchased primarily to substitute for the household's existing car on its banned day. Third, increased congestion, substitution between trips, and differences in fuel efficiency all affect gasoline consumption per car. These issues are pursued in the following sections. Section III presents a model of car ownership based on household survey data. The regulation is modeled as a reduction in the service flow from each car. Some car-owning households would want an additional car to compensate for the lost service flow from the car they already own, while others-with lower incomes- would now find their car insufficiently useful and want to sell it. Our model indicates that the two groups of households should be of about the same size, with a few more car sellers than buyers. Thus, our model of car ownership is unable to capture the apparent reality of increased ownership. We proceed to discuss some known weaknesses of the car ownership model. If there are some transaction costs in the market for used cars, then our model will overestimate the number of "sellers" in Mexico City. Section IV considers additional information indicating major changes in the market for used cars. Mexico City traditionally exported used cars to the rest of the country but started importing them in the years following passage of the regulation. The role of used cars in the response of Mexico City households is significant in itself because increased use of old cars implies that gasoline con- sumption has increased more than total driving and that pollution has increased more than has gasoline consumption. We also examine the possibility that changes in driving patterns-increased driving on weekends-may have contributed to increased driving per car. This is relevant because some advocates of the regula- tion maintain that beneficial changes in driving patterns could justify the regula- tion even if reductions in total driving never materialize. I. MARKET-BASED AND REGULATORY DEMAND MANAGEMENT It is well known that pollution charges are first-best instruments because they achieve reductions with the lowest possible losses in welfare. However, often- and sometimes with good reason, as when the costs of monitoring individual Eskeland and Feyzioglu 385 emissions are high-such instruments are not used. Eskeland (1994) and Eskeland and Devarajan (1996) discuss how many real-world automobile pollution con- trol strategies could be improved by including instruments that directly discour- age car use, such as fuel taxes. Improvement is possible because existing pro- grams provide incentives to make cars and fuels cleaner (standards) but fail to discourage the use of cars. Fuel taxes could be effective in Mexico because de- mand elasticities for gasoline are estimated to be in the range of -0.8 to -1.25 (see Berndt and Botero 1985 and Eskeland and Feyzioglu 1997). Instruments to reduce pollution by reducing polluting trips include gasoline taxes, driving bans, parking fees, highway tolls, and subsidies for public transport. But what are the welfare costs for consumers who sacrifice trips in response to demand manage- ment instruments? We make the simplifying assumption that income can be trans- ferred costlessly between households and between the private and the public sectors, allowing us to abstract from income distribution effects and any pre- mium (or penalty) on public revenue generation. When a trip is sacrificed due to an increase in the tax on gasoline, the value of the sacrificed unit to the consumer is the retail price of gasoline. Thus although some inframarginal units of gasoline (and trips) are worth more to consumers, a higher gasoline tax systematically screens out the trips that are worth the least. This property allows the gasoline tax to reduce trips at the lowest possible wel- fare cost. Demand reductions resulting from a regulation rarely are this selective. The "Day without a Car" program may curtail trips in households with a very high willingness to pay, and it may block a household's driving on Tuesday, say, even if the household could more easily have sacrificed other trips. These effects re- sult from the fact that the regulation does not allow trading of the rationed commodity, thus curtailing both inframarginal and marginal trips. (Goddard forthcoming advocates solving part of this problem by making driving permits tradable.) Figure 1 compares the welfare losses from a regulation and a tax increase, with the two calibrated to give the same reductions in demand. With the rationing mechanism used in Mexico City, issues are slightly more complex because the effects on demand are not known. First, the ration applies to the utilization of a vehicle that was not fully utilized (24 hours a day, 7 days a week) at the outset. For this reason, if users can move trips from one day to another or exchange car services with others-on Tuesdays I drive twice my distance to pick you up, and on Thursdays you return my favor-vehicle kilo- meters and the number of vehicles may both remain constant. Second, house- holds can purchase an additional car, thereby purchasing four "workday driv- ing permits" and two "weekend permits." All of these escape routes place upper bounds on the costs of compliance for a household. Finally, redistribution of trips between weekdays, when restrictions apply, and weekends, when no re- strictions apply, affects use. Increased weekend driving would limit the curtail- ment of total driving but might be valued for redistributing traffic to less con- gested days. 386 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Figure 1. 7Te Welfare Losses from a Tax Increase and a Regulation on Drivinig Tax increase Driving ban Pesos Pesos P + t + dt. ... P+t + ...P+t..... p _ _ _ tPt q . q Gasoline. liters -I + LI+ HL= II Gasoline. dq dq Liters Note: The crossed area in the figure for the driving ban illustrates the extra welfare losscs when the demand reduction is found under different parts of the demand curve, rather tharn only unde- the right- most part of the demand curve as in the case of a tax increase. II. AGGREGATE GASOLINE CONSUMPTION In this section, we investigate the behavior of aggregate gasoline consumption in the Mexico City Metropolitan Area (MVCIMA) before and under the ban. We trace the consumption pattern with quarterly observations from January 1984 through December 1992. Actual consumption levels are given by the solid line in figure 2. The driving ban became effective at the end of 1989, where the dotted lines appear. We assume that aggregate gasoline consumption in Mexico City depends on the gasoline price and household income. Other variables like congestion, quality of cars, and number of cars are not available., The equation for gasoline consumption, c,, is: (1) ct = (co + ctjPt + cO2yt + et t = 1, . ,T where Pr is the weighted average of gasoline prices in constant pesos (types of gasoline, by share in total use), and Yr is a proxy for personal income in Mexico City. Gasoline consumption is total sales of all types from terminals in the met- ropolitan area, in millions of liters (diesel is used only for heavy-duty vehicles in Mexico). Outgoing international telephone calls from Mexico City are used as a proxy for income. All variables are in logarithms. Thus c, is a price elasticity, and oc2 is a transformed income elasticity (transformed because the proxy for income may itself have an income elasticity different from 1). The hypothesis of the policymakers and the analysts is, of course, that impo- sition of the restriction changes consumption patterns, that is, shifts the demand function (equation 1). Such a change can be a change in the constant term, U., or a change in the elasticities, cc and cxl, without any change in xo, or both. To 1. Eskeland and Feyzioglu (1997) discuss demand estimation based on annual data and a national focus. Eskeland and Feyzioglu 387 Figure 2. Actual and Simulated Gasoline Consumption in Mexico City, 1984-92 Log gasoline consumption Actual consumption Upper bound of 95 percent __ confidence interval Simulated consumption had there .- been no ban on driving Lower bound of 95 percent confidence interval 1984 1985 1986 1987 1988 1989 1990 1991 1992 Note: We calculate the simulated consumption from data before the regulation and substitute actual values for price and income for 1990 to the end of 1992. Source: Authors' calculations. capture these possible changes, it is standard to introduce a dummy variable that is 0 before the restriction is imposed and 1 after. The estimated coefficients for this dummy variable and its interaction with price and income indicate whether any statistically significant changes in the demand function are related to the restriction. First we analyze the time-series properties in terms of nonstationarity for the data on gasoline consumption, gasoline price, household income, and the re- sidual (see appendix A). We conclude that, for our purposes, it is sufficient to model the variables as stationary, and that ordinary least squares estimation is appropriate. Second, to test the hypothesis that the demand function has not changed, we estimate equation 1 with a dummy variable for the periods under regulation. The estimated elasticities are given in table 1. The key result is that a significant change in the gasoline demand function is associated with the regulation. The effect of the regulation on the demand func- tion for the relevant income and price ranges can be seen in figure 3. The con- stant term shifts downward, but in the relevant area the demand function under regulation is above the demand function without regulation. This is due to an increase in the absolute values of the price and income elasticities. For sensitivity analysis, when a model is estimated allowing only the constant term to change, this term increases significantly, confirming that an increase in gasoline con- sumption is the result of the regulation. 388 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Table 1. Elasticities from the Estimated Gasoline Demand Function, Mexico City, 1983-92 Without Under Variable regulation regulation Constant 7.27#* 5.30t Gasoline price -0.17* -0.05 Income 0.06* 0.24t Significantly different from zero at 5 percent. * Significantly different from zero at 1 percent. t Significantly different from the "without regulation" model at 1 percent. Source: Authors' calculations. To estimate how much gasoline consumption has increased because of the ban we simulate a counterfactual scenario for the periods of regulation. We use the demand function based only on data from periods prior to regulation and insert actual values for price and income for 1990 to the end of 1992 to simulate what gasoline consumption would have been without a ban. In other words, we assume that the structural change in demand would not have occurred but that gasoline prices and income would have developed as they did. The simulated demand is shown in figure 2, together with actual demand. Figure 2 also shows Figure 3. The Effect of the Driving Ban on the Demand Function for Gasoline in Mexico City, 1983-92 ~~~~~~~~~~wt Drvnhan Aote: The surfaces show the demand function estimated before regulation and how it is changed hy the regulation. Source. Authors' calculations. Eskeland and Feyzioglu 389 a 95 percent confidence interval for simulated demand without the regulation. The simulation indicates that, if the ban on driving had not subjected the demand system to a structural shift at the end of 1989, demand would have been lower in all but the first quarter following passage of the regulation. Fur- thermore, actual demand is outside the confidence interval for all but one observation.2 III. HOUSEHOLD BEHAVIOR: A VEHICLE OWNERSHIP MODEL We now investigate how regulation could have increased gasoline consump- tion. Although it is possible that the regulation has increased use per car, a more likely explanation is that regulation has provoked additional car purchases in the metropolitan area because each car implicitly comes with four workday driving permits. One way of explaining our results, therefore, is that many vehicle- owning households both wanted and were able to purchase an additional vehicle when part of the service flow from each vehicle was effectively expropriated. To examine this hypothesis, we analyze data from a general-purpose household expenditure survey from 1989 (INEGI 1989). The survey was conducted before the regulation, and we use it to study the socioeconomic determinants of vehicle ownership. We adopt a discrete choice model of car ownership with household characteristics and socioeconomic variables as determinants and use household data from Mexico City to estimate the model parameters. Model Households must allocate their scarce resources across durable goods, non- durable goods, and savings. Let us assume a durable good (such as a car) is owned because of the value of the service flow it offers and that households behave optimally given their preferences, constraints, and resources. Then, for all households owning a car, the net value of the service flow, after subtracting short-term variable costs, exceeds the fixed costs of owning the car. We concentrate on the household's decision to allocate income between car services and other goods and services. Each household's ownership decision de- pends on characteristics determining its desire for car services and its income. For example, we expect a car to be more useful to households with more than with fewer people due to economies of scale in using the car's capacity. At the same time, however, for a given household income, more individuals may make a car less affordable. We also expect the demand for cars to rise with wages because higher wages increase the value of the time-saving services of a car. Ben- Akiva and Lerman (1985) discuss the assumptions underlying discrete choice modeling of car ownership and travel mode. 2. In another simulation, we use univariate forecasts of price and income in the demand model. When taken individually, price looks like white noise, and income is stationary around a positive trend. We forecast price using only its mean and forecast income using the estimated trend. The results yield even larger estimates of demand increases due to the policy. 390 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 As distant analysts we observe the household's choices and a partial list of the household characteristics that could be associated with these choices. We pro- ceed in two steps. First, we identify which characteristics determine how many cars a household decides to own. Second, we use this understanding to predict how their choice would change when the service flow from each car is restricted. We assume that the household maximizes a household utility function subject to a budget constraint: (2) U(TCSiOCi) i = 1, 2, . . . , m (3) I, = p,Di + p.OCi where U is the utility function for household i, TCSj is the transportation ser- vices that household i obtains from its cars, Di, OCi is the consumption of all other goods and services, Ii is the total expenditure of household i, p, is the annualized cost of owning and using a car, and p0 is the price of all other goods and services. Prices in this cross-section of households from Mexico City are assumed to be uniform. We restrict the choices to three: no car, one car, and two or more cars. This simplification is consistent with our data because only 2 percent of 1,037 house- holds possess more than two cars. We assume that the value of the service flow that household i obtains from owning j cars, TCSji, is a function of the charac- teristics of the household: (4) TCSji = fi(zi) j = O, 1, 2 where zi is a vector of household characteristics, and fi(zi) allows differences among households in the utility gained from the services of j cars. A household would presumably choose to own one car only if possessing more cars or no car both would make it worse off. We allow for the possibility that a household's decision to own a car also depends on unobserved variables. The observed variables are determinants only of the probability of a household owning j cars, because some of the household's unobserved characteristics may lead it to choose a different number of cars. We assume that on average the effects of these unobserved characteristics add up to 0. Hence, the assumption that the observed choice, yi = j, is optimal implies that the probability of household i choosing j vehicles is the probability that its total utility is maximized by owning j vehicles.3 (5) Prob(yi = j) = Prob(Ui(j) > U1(k)), k,j = 0, 1, 2, k 1 i = 1, 2, . . . , m where k, j = 2 denotes two or more cars. Results The probability of car ownership in equation 5 can be expressed in terms of observables once we assume that household utilities are linear in their argu- 3. This is a joint probability distribution: U(j) > U(i), and U(j) > U(k), i r k, j X i. See Manski and McFadden (1981). Eskeland and Feyzioglu 391 ments and that errors have a log Weibull distribution (see appendix B for the derivation). (6) Prob(y, = j) = exp(XiIS) exp(Xi3k)j= 0, 1,2 i = 0,1,... ,m k=o where Xi includes household characteristics as well as total expenditure, which we may interpret as a proxy for disposable income. Household characteristics and total expenditure feed into the utility functions through the coefficients Po, 1l, and 12. For example, the vector 131 tells us the importance of each of the household characteristics and income in determining the value of one car, relative to none, and 32, for two cars relative to none. If Xi ,1 is greater than Xi 12 and Xi J0, house- hold i would choose to own one car. The coefficients indicate how the variable increases the probability of having j cars, as opposed to having no car. We estimate the parameters of the model by maximizing the multinomial logit likelihood function that is defined in appendix B. We use a two-step proce- dure: first maximizing the likelihood function with respect to all variables and subsequently reestimating the model using only the variables that were signifi- cant in the first step. The results are given in table B-1. The signs of the coeffi- cients are plausible a priori. The more children a household has, or the more members who have higher education, the greater their preference to have a car, given similar incomes. The importance of education and average wages is higher for the second car than for the first. For a cross-section, discrete choice model, the fit is reasonably good. We compare actual with predicted outcomes in table B-2. Out of the 694 households that do not own cars, the model predicts "no car" correctly for 94 percent, but it predicts car ownership correctly only for 43 percent of the car-owning households. Simulation of a Reduction in the Service Flow from Each Car Next we use our model of car ownership to examine the likely response to a ban on driving. We model the ban as reducing the service flow a household gets from each car it owns. Once the value of the car's service flow is restricted, a household's optimization problem has changed.4 Poor households for whom the value of the service flow initially only marginally exceeds the costs of car owner- ship might want to sell their car after the ban reduces its service flow. Wealthier households might find a second car justified because it can substitute for the expropriated service flow and perhaps provide additional services. We model the utility of having a car under the regulation by applying a re- striction factor, a, to the utility of the service flow: U'(TCS1,OC,) = U(aTCS1,OC1). But first we need to make assumptions about the utility of having one car, U(TCS1,OC,). We assume that having one car is additive in its 4. We assume, in these calculations, that prices do not change, including those of used cars. In Mexico data on used car values are used for insurance purposes and reflect standard depreciation factors rather than market conditions. 392 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 two arguments and that there is no constant term in the part corresponding to travel services. Restricting this constant to 0 means that a household-indepen- dently of income-does not draw any utility from a car if there are no adults and no children (thus no education or wages). A restrictive assumption is required to simulate the effect of the ban, and we believe this one is reasonable. The ap- proach is vulnerable to a "Lucas critique," because we assume that we know how the determinants of car ownership are changed by the regulation, that is, how parameters are changed. We perform some sensitivity analysis and discuss potential weaknesses in the subsequent sections. The following equations display the model of the utility of household car ownership. (7) U(TCS1,OC1) = V(TCS,) + W(OC1) where V(TCS1) = 31iChild + P32Adult + 313MedEd + P3,4HighEd + ,WagePW, W(OC1) = 010+ 316TotExp. where our household characteristics are number of children in the household (Child), number of adults (Adult), number of people with high-level and inter- mediary-level education (HighEd and MedEd, respectively), and average wages earned by the wage earners in the household (WagePW). The vector of exog- enous variables includes these household variables plus a constant (C), and total expenditure (TotExp), our proxy for disposable income. The assumptions imply that income does not influence the utility of car services directly but that it does so indirectly through the shadow price of income available for other consump- tion. With a negative constant and a positive income coefficient in W(OCI), car ownership becomes more likely as the total budget grows, because higher in- come allows for more other expenditures if allowance is made for the costs of owning and operating a vehicle. We simulate how the utility of owning one car is changed by the restriction as follows. We take the estimated utility function as given and simulate the car usage restriction by applying a restriction parameter ax to the value of the service flow from the car, that is, to parameters 31 through P,5: (8) U'(TCS,,OCI) = IO + L6TotExp + x!, P1, x where U' is the simulated utility under the restriction. The value of a is subject to sensitivity analysis. For households with one car, if travel days have no substitutability and the car is used onlv on workdays, aX is 4/5; if the car is used approximately evenly across the week and the weekend, then cx would be 6/7. If a household can comply with the restriction without losing any service value (say, it uses the car for some trips every week, but the trips can be moved from one day to another without any costs) then U. would be 1. At the other extreme, if the household needs the car only on the day the restriction is binding, a would be 0. This latter case is unlikely, given that the car Eskeland and Feyzioglu 393 registry process as well as the car market allows owners to influence which week- day is banned. For households that own more than one car, we assume that one car can substitute for the other on restricted days, so the restriction has little or no ef- fect. If the household has only one driver, this would be accurate. For a two- driver household, the regulation cuts the service flow from two cars on five workdays to two cars on three workdays and one car on two workdays. Thus, if a two-car household would otherwise use its second car only three days a week, this assumption is accurate. Otherwise, it is an approximation, and it is wrong if households with two cars have the same difficulty managing without every car on each workday as has a one-car household. Our modeling assumptions should be interpreted as the effective reduction in the value of the service flow for a one- car household in comparison with a multicar household. Finally, we assume that households without cars would not change their behavior, because optimi- zation theory predicts that an added constraint can change the optimal choice only if it restricts the originally optimal choice. Results for different restriction coefficients are given in table 2. The simula- tions produce an optimal number of cars for each household before and under the regulation. In table 2 we report as sellers the households that the model predicts will switch from being a one-car household to being a no-car household and as buyers the households that will have one car in the preregulation regime, but two under regulation. The model indicates that, for restriction factors in the range of 0.8 to 0.9, the number of "sellers" exceeds the number of "buyers" by 2 to 3 percentage points. Thus, the model predicts a slight increase in exports of used cars to the rest of the country as a result of the restriction. However, most observers believe that the opposite occurred. Many Mexico City households have bought an addi- tional car in response to the regulation. Increased purchases of used cars is con- sistent with our estimation that total gasoline consumption has increased, but this result is not indicated by our car ownership model. Figure 4 illustrates the main features and results of the car ownership model. The first graph in figure 4 is constructed by projecting the model to a two- Table 2. Simulating Car Ownership under the Regulation (percentage of predicted stock of cars) Restriction Net coefficient, al Sellers Buyers purchases 0.95 5 3 -2 0.90 8 6 -2 0.8s 11 8 -3 0.80 14 12 -2 0.75 18 16 -2 Note: Sellers are households that the model predicts will switch from one car in the preregulation regime to no car under the regulation. Buyers are households that the model predicts will switch from one car to two cars. Source: Authors' calculations based on INEGI 1989 household survey, Mexico City subsample. 394 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Figure 4. Income Distribution and the Utility of Car Ownership in Mexico City Utility Utility of car ownership 8 6 * Preregulaton household does not own a car, U0 - - Preregulaton household owns one car, U, , - Preregulaton household owns two cars. U 12 - - 4 ~ - - ~ S Under the regulaton, household owns one car, L' , - 2 -, < ,, . 00: bLRegion of .,,. ho~~~~bLyers -2 () ~ ,$s' q Region o sellers -4, 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 Income distribution of the households surveyed Number of households 240 200 160 120 Region of Region of sellers hoycers 80 40 0 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3,2 3.6 Total household expenditure (millions of pesos) Note: The curves are drawn as follows: U, is only a reference-the horizontal line. Fo- U., we use the estimated one-car coefficients, plugging in the variables for each household in the data set to calkulate 1,037 utility levels given that the household has one car. Next, wve use a univariate ordinary least squares model to regress these utility levels on constant and total expenditures. For 6,. we foll ow the same procedure, but with the two-car coefficients. fU'* utility as a function of income given one car under the restriction, is calculated using a shift parameter o of 0.8 (see text). Source: Authors' calculations. Eskeland and Feyzioglu 395 dimensional one: utility as a function of income given that the household owns no, one, or two cars. For each household a separate utility level is calculated under three different preregulation scenarios: the household owns no car (U0), one car (U1), or two or more cars (U2). We plot these utilities against the household's income level. The figure shows that each of the conditional utility functions U0, U1, and U2 has an income range for which it gives the highest utility. This is the income range for which that specific number of cars is optimal when there is no regulation. For the lowest income range, U0 is highest, so no car is the optimal choice for households in that range. As income increases, the utility of having one car increases (the slope of U0 is normalized to 0). In the income range to the right of the intersection of U0 and U1, households typically own one car. At an even higher income range, households typically own two or more cars. U1' shows the utility of having one car after the restriction is imposed, using a restriction factor of 80 percent (that is, 80 percent of the service flow remains). U1' is lower than U1; we thus have a reduction in the size of the income range for which one car is the optimal choice. For incomes around Mex$800,000, the optimal number of cars has shifted from one to zero. These households, according to our simple model, would sell their car. But house- holds earning about Mex$2.8 million would respond by expanding vehicle ownership from one to two (or more). Thus our model predicts that the in- come threshold a household must pass to buy its first car moves upward, while the income threshold for buying a second car moves downward. The second graph in figure 4 depicts the income distribution of the households surveyed. There is a greater density of households in the range of sellers than in the range of buyers, but the latter range is larger and is supported by greater per household incomes. We may conclude that our simple ownership model is successful in demon- strating a range of buyers and sellers in similar magnitude, but it does not ex- plain increased vehicle usage, because it does not point to increased ownership. Are there simplifying assumptions in our model that could plausibly cause an underestimation of net purchases? Two are worth considering: used car prices and sunk costs. In our model, used car prices are assumed to be unaffected by the ban. When part of the service flow from cars is expropriated, ceteris paribus the value of used cars should fall. However, under the ban a car also becomes a bundle of implicit driving permits, which should increase its price. Therefore the net effect of the regulation on (used) car prices cannot be known a priori. This does not mean that car prices have not risen or fallen in reality, as our model assumes. We do not know how to adjust used car prices to correct this weakness in the model because price variation is absent in our data. A related problem is that when we abstract from the sunk cost aspect of in- vesting in a car, we introduce an asymmetry if a regulatory change reduces the service flow from a car. Some owners who would want to sell in the absence of sunk costs will hesitate and not sell when there are sunk costs. No correspond- 396 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 ing hesitation applies to the households induced to purchase an additional car. This is thus a weakness that could also explain why our model, which ignores sunk costs, can underestimate the number of net purchases.5 Our analysis is overly simplistic in other ways, however, as in lumping all multicar households into a single category and in making assumptions about the specific way utility function parameters are changed by the regulation. Supporting Data: What Happened to Vehicle Ownership? We have recently obtained aggregate annual data that shed light on the effects of the ban on vehicle ownership. Table 3 shows data on sales of new vehicles and increases in the number of vehicles registered in Mexico City (the Federal District) and the rest of the country, averaged over seven preregulation years and four regulation years. An area's net import of used vehicles is inferred by subtracting the sales of new vehicles from the net increase in vehicles registered (assuming vehicle scrap- page is 0). Before regulation capital city households were consistent net export- ers of used vehicles to the rest of the country-about 74,000 vehicles per year. Such a flow is to be expected. Given that city households on average have higher incomes, poorer regions buy used vehicles from the capital. In Mexico City, the tradition has been that visitors from far away gather at the Aztec Stadium to bargain for used cars. Table 3 also shows that Mexico City's traditional exports of used vehicles turned to net imports under the regulation. During the years of regulation, Mexico City has been importing an average of 85,000 vehicles per year from the rest of the country. A statistical test does not confirm that a break exists at the intro- duction of the regulation. This is not surprising given anomalies in the data. For example, 300,000 vehicles-15-20 percent of the stock-appear to vanish in 1986, resurfacing in 1988. Expansion of car ownership through used cars has obvious implications. Older cars are typically less fuel efficient due to both initial design and deterioration. Part of the poorer fuel economy of older cars is due to incomplete combustion, leading older cars to be more polluting per kilometer and per liter of fuel con- sumption. Finally, the fact that the regulation artificially ties up and idles capital in the wrong places implies that it is costly to the nation. We include other data in table 3 that can be disaggregated by area. If phone calls and electricity consumption are good proxies for personal in- comes, then they indicate that there was no change in personal incomes in Mexico City and the rest of the country that would explain the observed turn in the flow of used cars. Finally, table 3 indicates that ridership of the sub- way system in Mexico City has declined in the years of regulation. This could 5. These effects are explored in option pricing theory (not buying a car leaves the option of buying one later), and this particular effect is called hysteresis (Dixit and Pindvck 1993: 136). In a market for used cars, transactions costs may be high due to asymmetric information about quality, giving a theoretical underpinning for the existence of sunk costs (Akerlof 1970). Eskeland and Feyzioglu 397 Table 3. Supporting Data for the Household Car Ownership, Mexico City, 1983-93 (annual average) Before regulation, Under regulation, Indicator 1983-89 1990-93 Sales of new vehicles (thousands) Mexico City (federal district) 80 154 Rest of Mexico 127 237 Increase in vehicles registered (thousands) Mexico City (federal district) 7 239 Rest of Mexico 174 250 Net import of vehicles (thousands) Mexico City (federal district) -74 85 Rest of Mexico 47 13 International phone calls (percentage growth) Mexico City 27.2 24.1 Whole country 19.5 29.7 Local phone calls (percentage growth) Mexico City 12.9 12.9 Whole country 19.5 29.6 Electricity consumption (percentage growth) Mexico City 7.0 7.0 Whole country 5.0 3.0 Metro ridership (percentage growth) Mexico City 5.7 -2.4 Note: We analyze the vehicle data using two different definitions of Mexico City, the federal district and the federal district plus the state of Mexico. The regulation applies to the Mexico City Metropolitan Area (MCMA), comprising the federal district plus part of the state of Mexico. The conclusions are very similar. We show only the data from the federal district here. For other data, the area covered by Mexico City varies. The area for electricity consumption is larger still than the MCMA. We use a long data series (1983-93) due to the noisiness of registration data, but this does not affect the conclusions. Source: Mexico's Automobile Manufacturers Association, Banco de Mexico, and INEGI. be interpreted as another indication that the regulation does not work ac- cording to intentions. It could also provide a partial explanation: if the sub- way and other public transport systems have little capacity to serve addi- tional passengers, this would contribute to the choice of additional vehicles as a compliance strategy. In this section we have described and estimated a theoretical model of car ownership showing how a regulation creates buyers as well as sellers; we have also considered other data supporting the hypothesis that households in Mexico City have responded to the driving ban by buying additional used cars from the rest of the country. 398 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 IV. CONGESTION AND WEEKEND TRAVEL While the above explanation explores possible increases in vehicle owner- ship, this section explores features that could break or reduce the presumed effect on driving per car. The ownership model assumes that the regulation re- duces the service flow from each car, other things being equal. In fact, as the service flow from other cars is reduced as well, congestion levels could decline and travel speeds could increase, making each car more useful. To explore this effect, let demand, vd (we now represent the vehicle and its services with one variable, abstracting from the distinction between ownership and use) consist of an exogenous component, k, and a component sensitive to travel time, t (say, the average time that it takes an individual to drive to work, given the conges- tion levels), zd = k + v(t). Further, to describe the road network's capacity, let travel time be a function of demand, t = t(v). If the regulation reduces the exog- enous component of demand, k, the equilibrium effect will be dampened by a rebound in demand due to increased speed (appendix C), as described by equation 9. (9) dv 1 The first elasticity (travel demand with respect to time) is negative, and the second (the road network's supply of travel time, with respect to additional entering vehicles) is positive. Thus the denominator is at least 1, dampening any direct effect that a regulation has on travel demand. We can see that if either (or both) of the elasticities is 0, the equilibrium effect on demand is the direct effect: a car removed from the streets on Tuesday simply reduces overall traffic on Tuesday by one car. This will in fact be the case on completely uncongested roads, but it is not realistic for workday conditions. By contrast, if the product of the two elasticities multiplied by each other is -1 or -3, then the equilibrium reduction after reducing traffic by, say, four cars on Tuesday is only two cars or one car, reflecting that other vehicles enter the roads to take advantage of the reduced congestion. Using plausible parameters we find that the multiplied elasticities could be larger than 1 in absolute value, so that the equilibrium reduction in workday traffic could be less than half of the initial reduction, due to a resurgence in speed-sensitive travel. For supply conditions, only a few estimates exist in the literature of how travel times (or speeds) respond to additional vehicles entering the road, and none exists for Mexico City. In severely congested conditions the elasticity is greater than 1, meaning that an additional vehicle reduces the total throughput of a road link. According to Small (1992), an elasticity of 2.5 re- flects conditions in the middle of a range. The demand elasticity with respect to travel time savings under plausible assumptions is three-quarters of the demand elasticity with respect to gasoline prices (appendix C), or 0.75>;(-0.8) = -0.6, Eskeland and Feyzioglu 399 using a conservative elasticity estimate of -0.8 from Mexico.6 Using these val- ues, a plausible estimate of the equilibrium reduction in traffic on a weekday is 0.40, or two cars for every five initially removed.7 But how could a reduction in traffic on weekdays-even if only 40 percent of the initial reduction-result in an increase in total traffic? Another unmodeled effect is the distinction between weekdays and weekends. Trips on different days will generally be imperfect substitutes, and less congestion and absence of regu- lation are only part of what makes weekend trips special. But if the equilibrium reduction in traffic is only 0.4 cars on a weekday when one car is initially re- moved from the road by the regulation and if as much as 40 percent of the cars originally removed from the street make an additional trip on the weekend to compensate, Mexico City could see total travel increase even if no additional cars are purchased. In this case much of the intended effect of the regulation might be undermined by increased weekend driving, and the additional car pur- chases then lead to a significant increase in car use. As attractive as the substitution of weekend for weekday driving hypothesis appears, we are unable to confirm it in empirical tests. We used daily observa- tions of carbon monoxide pollution data to provide a daily proxy for driving, because the link to emissions from cars is direct and quick. Using weekly obser- vations of the ratio of the pollution concentration in the weekend to that during weekdays, we find two results (see Baynham 1997). First, regressing the ratio on a regulatory dummy alone, the regulation has a significant effect, increasing the average ratio by 7 percentage points, from 86 to 93 percent. Second, once addi- tional variables, such as a time trend, are included, the effect becomes insignifi- cant. Thus the effect of the regulation on the relative importance of weekend driving is small if existent at all. It is rejected by a simple statistical test. So although it is conceivable that net weekday reductions are considerably less than direct weekday reductions and that weekend travel increased, our analy- sis of daily pollution data does not give firm indication of any major change in the relative importance of weekend travel. This leaves induced purchases of ad- ditional vehicles as the apparent cause of the counterproductive results of the regulation. V. SUMMARY AND CONCLUSIONS We estimated a gasoline demand function based on aggregate time-series data to analyze the effect of the driving ban in Mexico City. Surprisingly, our results indicated that total car use in Mexico City has been increased by the regulation. We focused on additional car purchases in the metropolitan area as the most likely explanation of increased driving, because households with unchanged car 6. Eskeland and Feyzioglu (1997) and Berndt and Botero (1985), both with pooled estimate results for Mexico, find long-term estimates in the range of -0.7 to -1.25. 7 dv 1 1 = 1/2.5 = 0.4. dk (1- t (1- (-0.6) * 2.5) 400 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 ownership would be unlikely to increase driving. Supporting data on vehicle ownership supported this interpretation. In preregulation years, Mexico City on average exported 74,000 used vehicles annually to the rest of the country. In the first four years of the regulation, Mexico City imported 85,000 vehicles annually. Assuming that car ownership is motivated by the service flow that cars offer, we estimated a model of household ownership decisions and simulated indi- vidual responses to the ban. Our model indicated that although some house- holds would want to buy more cars as a result of the regulation, a somewhat greater number of households would want to reduce their car ownership. An assumption-that cars are bought and sold without transaction costs-could be a reason why the car ownership model would overestimate the number of house- holds that would sell their car. We also noted additional features excluded in the ownership model. First, congestion-sensitive demand for travel would imply that net reductions in travel on weekdays might be significantly less than direct reductions. Second, some suppressed weekday trips could show up as additional weekend trips. Third, travel may have shifted toward less fuel-efficient old cars so that aggregate gaso- line consumption-and pollution-could increase even if travel were constant or slightly reduced. There is thus ample evidence that the ban imposed high compliance costs for households, much higher than those of alternative market-based policies such as gasoline taxes. Moreover, many individuals chose a compliance strat- egy that led to no reductions-or even increases-in car use and acquired a used car with lower technical standards. Thus results for accidents and pollu- tion could be worse than what is indicated for total gasoline consumption and congestion. Finally, we should mention that we have not analyzed how regulation and taxation would differ in terms of how costs are distributed. Such differences may exist-it is unclear whether they would favor one strategy-but are hardly relevant when a costly strategy is found to be counterproductive. APPENDIX A. TIME-SERIES PROPERTIES AND ESTIMATION RESULTS OF THE AGGREGATE DATA Time Series Properties The autocorrelation function and the augmented Dickey-Fuller unit root tests reject nonstationarity for all but the income data. The autocorrelation function rejects nonstationarity for income, but the augmented Dickey-Fuller does not. Therefore, for the income series we perform further tests. Income in itself is not the focal point, but rather a factor that affects consumption. We therefore run a regression with consumption, price, and income. Consumption and price are stationary; therefore, the residual has the same stationarity properties as the price. Our tests show that the residual is stationary. We therefore conclude that, Eskeland and Feyzioglu 401 Table A-1. Regression of Aggregate Gasoline Consumption Variable Coefficient Standard error 1-tail significance Constant -7.27 0.37 0.0000 ln(price) -0.17 0.06 0.0167 ln(income) 0.06 0.02 0.0050 Dummya -1.93 0.50 0.0060 Dummya*ln(price) 0.12 0.11 0.2820 Dummya4'n(income) 0.18 0.07 0.0092 Adjusted R2 0.942 Durbin-Watson statistic 2.28 F-statistic 115 Probability F-statistic 0 Note: The dependent variable is ln(total gasoline consumption). a. The value of the dummy variable is 0 for 1984-89 and 1 for 1990-92. Source: Authors' calculations. for our purposes, it is sufficient to model income as a stationary variable, and ordinary least squares estimation is appropriate. Regression Results We estimated equation 1 using ordinary least squares, and table A-1 reports the estimation results. The proxy for income in Mexico City is the number of international (outgoing) telephone calls. Others variables, such as quarterly fig- ures for gross national product, local industrial output, and local electricity con- sumption, performed less well statistically, judged in terms of a preregulation demand model. The changes in the constant and the income elasticity are statis- tically significant, while the changes in the price elasticity are not. APPENDIX B. DERIVATION AND ESTIMATION OF THE CAR OWNERSHIP MODEL Derivation First, we combine the budget constraint and the definition of total transpor- tation services from cars with the utility function by substituting equations 3 and 4 into equation 2: (B-1) Uii = U[f1(zi),(I, - p,Di)/p.] where, Uii is the utility of household i if it has j cars, f1(zJ) is a vector of household characteristics, Ii is the total expenditure of household i, PC is the annualized cost of owning and using a car, Di is the number of cars owned by household i, and p0 is the price of all other goods and services. Second, we assume that the utility function is linear in its arguments: (B-2) Uii=tji+eii j=0, 1, 2 i=1,2,...,m and ,uji = Xif3, where Xi includes household characteristics as well as total expenditure. 402 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Third, we establish the probability distribution of owning j cars. For the prob- ability of i's not owning any vehicle, we obtain: (B-3) Prob(yi = 0) = Prob(Uoi > Uki), k 0, = Prob[(e0i - eli > ti - t0:), (e0i - e2, > 2i - [0)l where k is the number of cars owned by household i. If ej has a log Weibull distribution, the probability of choosing j vehicles is a logistic function: (B-4) Prob(yi = i) = exp(Xir3;{) exp(Xif3k)J j = 0, 1, 2 i = 0, 1,. . . This is equation 6 in section III. For estimation, because utility is ordinal, we normalize utility to be 0 for the case of no cars, so that the model estimates the additional utility of owning a positive number of cars. The decision process is restated in terms of deviations from the utility of owning no cars: (B-5) Prob(yi =1) exp(Xi )r,exp(XiP3') , j = 0, 1, 2 i = 0,1,... , k =o where Xi31' = Xij - Xi4o. The decision process about the number of cars to own by all households can be put together into a standard multinomial logit likeli- hood function: mO mO +ml mC M +m2 (B-6) L =J P(Yi 0) H P(Yi =1) P(y, =2) 1=1 i=mO +1 i=mC + 7n, +1 where, mo, ml, and M indicate number of households in each category in the data set that is sorted with respect to number of vehicles owned. Results Results from household data are given in table B-1. The coefficient of wage income is 0.12 for one car, which means that if the average wage income of the household increases by 1, then the extra utility the household gets from owning a car as opposed to not owning one is 0.12. The corresponding coefficient for two cars is 0.19, which means that if wage income increases by 1, then the extra utility of owning two cars as opposed to one or no car is 0.07 and 0.19, respec- tively. Similar interpretations follow for the other coefficients. The predictive power of the model is illustrated in table B-2. Finally, using the model in simulation with a restriction factor of 0.8, as is used in both parts of figure 4, 33 households come out as sellers, while 38 of the one-car households want an additional car. As predicted percentages of the stock of cars among the 1,037 households indicated, purchases and sales come out as 12 and 14 percent, respectively (see table 2). Eskeland and Feyzioglu 403 Table B-1. Results from the Household Car Ownership Model for Mexico City One car, Two or more cars, Variable 1 o2 Constant -2.37 -4.92 (0.21) (0.27) Child 0.12 a (0.05) Adult -0.19 a (0.07) Secondary-level education 0.20 a (0.08) Tertiary-level education 0.79 1.32 (0.11) (0.12) Wage per worker 0.12 0.19 (0.03) (0.03) Total expenditure 1.62 1.87 (0.27) (0.28) Note: The estimated equation is U(j cars) - U(no car) = Xb, + e were X is the vector of explanatory variables. All the reported coefficients are significantly different from 0 with 95 percent confidence. Standard errors are in parentheses. a. The coefficient is not significantly different from 0 in the first-stage estimation. Source: Authors' calculations based on INEGI 1989 household survey, Mexico City subsample. Table B-2. The Predictive Power of the Household Car Ownership Model for Mexico City (number of households) Predicted Actual No car One car Two cars Total No car 654 36 4 694 One car 172 60 17 249 Two cars 25 38 31 94 Total 851 134 52 1,037 Source: Authors' calculations based on INEGI 1989 household survey, Mexico City subsample. APPENDIX C. EQUILIBRIUM CHANGE IN TRAFFIC WHEN THERE Is AN EXOGENOUS CHANGE IN DEMAND FOR TRAVEL How does travel respond in equilibrium if the road's capacity to supply rapid travel is declining as the level of traffic increases and if the demand for travel is sensitive to congestion levels? Assume that demand for vehicle use, v, is the sum of two components: f(t), which depends on congestion levels, represented by the time spent making a certain trip, t, and k, which is exogenously given: 404 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 (C-1) v = f(t) + k. Also assume that the road link's capacity to get the vehicle through within a certain time, t, depends on the number of vehicles entering the road, v: (C-2) t = g(v). For equilibrium, we have: (C-3) v= f [g(v)] + k dv== +gd+dk=> at av (C-4) dv 1 1 dk af ag 1dE Es at av where Ed and gs are the demand and supply elasticities of travel time. The demand elasticity of travel with respect to travel time is negative (if travel time goes up, people are less interested in traveling), while the supply elasticity of travel time with respect to entering vehicles is positive (as more vehicles are on the road, traffic slows, and travel time for a given trip rises). Thus unless one, or both, of the elasticities is 0, the equilibrium effect on travel is less than 1, that is, less than the initial, exogenous reduction in demand. Any combination of elas- ticities that yields a high product in absolute value would reflect conditions close to Downs law: an exogenous reduction of congestion levels would immediately lead to an increase in the demand for travel that swamps the initial effect, so that congestion levels are back at normal. Put differently, highly congested roads can easily be normal. More cautiously argued, what are plausible values? For supply, we can imme- diately establish that e' = 0 represents an extreme case of road conditions in which vehicle density is so low that an additional vehicle does not slow down other vehicles. Positive values represent natural conditions in which a positive shadow price for road capacity is natural for urban roads and intercity high- ways (see Hau 1992). Values greater than 1 represent heavily congested condi- tions in which an additional vehicle reduces the total throughput of the road link per time unit (a 1 percent increase in entering vehicles increases travel times for all vehicles by more than 1 percent). For the elasticity of travel demand with respect to travel times, what are plau- sible values? The following modeling framework, focusing on vehicle travel as a timesaving alternative, may shed light on that question. Let utility, u, be defined over cars, c, other goods and services, o, and leisure, 1: (C-5) u = u(c,o,l). Eskeland and Feyzioglu 405 And let the individual budget constraint be: (C-6) P6C+o=w(L+lcc -1) +1. where pc is the price of owning and using a car, the price of other goods and services is normalized to 1, w is the wage rate, 1c are the time savings offered per vehicle (c is interpreted as a continuous variable; we may view this as a model of a representative consumer), L is the endowment of human capital, and I is lump-sum income. The Lagrangian of the consumer's maximization problem can be written: (C-7) L = u(c,o,l) - A[(pc - wlc)c + o - w(L- 1) - I]. The first-order conditions are: IU = -(PC - wUC) w II -1 w III (PC - w1C)c + o - w(L - I) = I where u,1 and u01 are marginal rates of substitution. The interpretation of the first equation is that the value of time savings justifies part of the cost of cars, and this part is subtracted from the cost, pc/w, that would otherwise be equated with the marginal rate of substitution between cars and leisure. To study relationships between demand elasticities, differentiate the first- order conditions with respect to the time savings offered per car: auc, ac + auc, ao + auc al ac aJc a O aic al alc (C-8) au", ac + au", ao + aUCJ al 0 (ac ac aO alc a, alc (C- WIC) ac + ao + w al C c aic aic aic The coefficient matrix, A, is: auC1 auc6 auCl ac ao al A au., au01 au., (C-9) ac O Jal PC wlc 1 w 406 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Assuming that A is nonsingular, we may use Cramer's rule to solve for: 1 au', au' ao al ac JAI= 0 au"0 au"O (C-10) DIC aO ai WC 1 W Similarly, when we differentiate the first-order conditions with respect to car prices, we obtain: ao w (C-11) A = 0 apc Again using Cramer's rule, we have: O au'[ auC, ao a3 3AIA -1 au0I du01 (C-12) apc w aO ai -c 1 w For matrix A, let CkI be the cofactor of row k, column j. We may write ac AI= C21 +WC C31 a3c JI 1C2 '3 (C-13) apc =-WC21-cC31 ac ac D_c lC aDcpwlc -= -w = > - = c => aTc apC aic c aPc Pc Thus: WI' (C-14) G=c,l' (=-CPc pC Eskeland and Feyzioglu 407 Thus, the elasticity of car demand with respect to the time savings offered per car is wlC/p, times minus the elasticity of car demand with respect to car prices. Referring back to first-order condition I, we may see that wIN/p, is the share of time savings in justifying car purchases at the margin (the other share being the direct utility drawn from cars): _uclw + wIc 1 Pc (C-15) ac lC ac ac lc t act alc at =at aic c IC =at c = It remains to relate the concept of time savings to the concept of travel times. So let us note that: t (C-16) Gc,t= = G - The conversion between the two elasticities thus is simply the ratio of travel times to time savings offered by the car. It takes positive values and is often in the range of 0.5 to 2.0. For instance, if a regular trip takes half an hour, and the alternative is a combination bus ride and walk that takes one hour, then both time savings and travel times are half an hour, and e,t = - However, if a car trip is 20 minutes and the alternative mode takes 30, the elasticity of travel demand with respect to travel time is double the elasticity of travel demand with respect to time savings. REFERENCES The word "processed" describes informally reproduced works that may not be com- monly available through library systems. Akerlof, George A. 1970. "The Market for Lemons: Quality Uncertainty and the Mar- ket Mechanism." The Quarterly Journal of Economics 84(August):488-500. Baynham, Lynda. 1997. "Air Pollution and Transportation in Mexico City: Focus on the 'Hoy no Circula."' Ph.D. diss., Institute for Environmental Studies, University of Wisconsin, Madison. Processed. Ben-Akiva, Moshe, and S. R. Lerman. 1985. Discrete Choice Analysis: Theory and Ap- plication to Travel Demand. Cambridge, Mass.: MIT Press. Berndt, E. R., and G. Botero. 1985. "Energy Demand in the Transportation Sector of Mexico." Journal of Development Economics 17:219-38. Dixit, Avinash K., and Robert S. Pindyck. 1993. Investment under Uncertainty. Princeton, N.J.: Princeton University Press. Eskeland, Gunnar S. 1994. "A Presumptive Pigovian Tax: Complementing Regulation to Mimic an Emissions Fee." The World Bank Economic Review 8(3, September):373-94. Eskeland, Gunnar S., and Shanta Devarajan. 1996. Taxing Bads by Taxing Goods: Pol- lution Control with Presumptive Charges. Directions in Development Series. Wash- ington, D.C.: World Bank. 408 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Eskeland, Gunnar S., and Tarhan Feyzioglu. 1997. "Is Demand for Polluting Goods Manageable? An Econometric Model of Car Ownership and Use in Mexico. " Journal of Development Economics 53(2, August). Goddard, Haynes. 1997. "Urban Decentralization with an Application to Mexico Citv." Environmental and Resource Economics 10(1):63-99. Hau, Timothy. 1992. "Economic Fundamentals of Road Pricing: A Diagrammatic Analysis." Policy Research Working Paper 1070. World Bank, Policy Research Department, Wash- ington, D.C. Processed. INEGI (Instituto Nacional de Estadistica Geografica e Informatica). 1989. Encuesta de hogares. Mexico City. Manski, C. F., and D. McFadden. 1981. "Alternative Estimators and Sample Design for Discrete Choice Analysis." In C. F. Manski and D. McFadden, eds., Structural Analy- sis of Discrete Data with Econometric Applications, pp. 2-50. Cambridge, Mass.: MIT Press. Small, Kenneth A. 1992. Urban Transportation Economics. Gland, Switzerland: Harwood Academic Publishers. THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3: 409-32 Prices and Protocols in Public Health Care Jeffrey S. Hammer This article addresses the problem of how to determine the optimal allocation of pub- lic expenditure in the health sector. The first part poses the question: How should the set of services provided in the public health care system and the fees charged for them be chosen to maximize the health status of the population with a fixed budget? First, the findings show that policy reform should take into account the response of the private sector. Substituting for a reasonably well-functioning private sector is not as valuable as providing services the private sector cannot. Second, the assumptions needed to justify the cost-effectiveness of medical interventions as a criterion for setting priori- ties are so restrictive as to make this method usable in few, if any, circumstances. Third, prices for any one service should be set to balance the conflicting goals of en- couraging its use and of conserving the budget for more effective services. The second part broadens the objective of policy to cover the standard welfare economics concerns of utility and market failure, the latter being extensive in the health sector. It reexamines welfare maximization rules to show that only the market failure components of shadow prices are needed to calculate the welfare gains from public investments. Providers of health care in public clinics in the developing world work with tight budgets, requiring difficult choices. This article examines a few ways to make these choices using standard tools of economic analysis. It does not pretend to be a comprehensive treatment but does raise some core issues concerning the best use of public funds in health care. The standard method characterizes optimal allocations by posing the prob- lem: how can the government maximize a given objective subject to constraints? The answer depends on the choice of objective and the characterization of the constraints. Economists generally like to use social welfare or the sum of peoples' utilities as defined by their own preferences as the objective in such a problem. Often, they weight this sum by peoples' income so as to give more influence to the well-being of the poor. In the context of the health sector, this objective is somewhat problematic because people may not be able to define their prefer- ences due to lack of information about the nature of health under different dis- eases or the likelihood that a treatment will succeed. Using the standard ap- proach requires making some monetary valuation of health, which may be difficult, especially if people's demand functions for care are not believed to give true valuations. Further, ministries of health are more likely to think in terms of Jeffrey S. Hammer is with the Development Research Group at the World Bank. C) 1997 The International Bank for Reconstruction and Development / THE WORLD BANK 409 410 THE WORLD BANK ECONOMIC REVIEW, VOL. 11. NO. 3 improving the public's health status. Therefore it is of interest to identify the best policies for achieving the goal of improved health. Governments and their ministers of health face three sorts of constraints. The first is the state of the world that the government does not directly control. This includes available medical technology, of course, but also people's behavior in pursuing or providing health care outside the government sector. The govern- ment, while typically a large supplier of health care, forms only one part of the overall medical market. The private sector, especially in developing countries, is also quite large, often larger than the public sector (World Bank 1993). If not controllable, the response to public policy of the entire private market for health care represents one of the constraints on the government's ability to influence people's use of goods and services, and, hence, their health. I do not presume that the private market does a good job, merely that it reacts to and therefore partly determines the ultimate effect of policies. The second sort of constraint involves the budget. Analyses can focus on re- sources available to the public health care service, the ministry of health as a whole (which includes nonclinical interventions such as health education and vector-pest-control campaigns), the government as a whole (which includes sanitation, water supply, and female education), or the entire economy, includ- ing the private sector. For the economy, the range of policy instruments and regulatory powers must be quite extensive. Thus, the third kind of constraint is the range and flexibility of the policy instruments at the government's disposal. Optimal policies must, above all, be feasible and take into account restrictions irr. osed by administrative capacity or political acceptability. The first part of this article takes the particular view of a minister of health running a public health care system. I assume that the minister aims to maximize the health of the public. Decisions beyond the minister's control fix the budget. The policy instruments are of two sorts. First, a set of rules, or protocols, ration the range of services available in the public sector. The types of services or treat- ments offered and excluded comprise the policy choice. Second, a schedule of fees, or prices, recovers some of the cost of services and thereby stretches the budget. Charging fees might deter people, particularly poor people, from using effective services. At the same time, public health care providers might face re- strictions on setting fees. It is often not feasible, for example, to charge different rates for individual treatments or, sometimes, to charge at all. The basic model in section I focuses primarily on clinical care. Broader issues (and in the context of developing countries, very possibly the more important issues)-such as the provision of public health measures (safe water, sanitation, or vector control), the regulation of private providers, and public information campaigns-are ex- tensions to the basic model. The second part of the article is more consistent with standard public eco- nomics. The objective in this case is the overall welfare of the people. I argue that although health status as an objective can highlight certain principles of optimal resource allocation, this one dimension of health status fails to capture Hammer 411 many crucial policy decisions. Section I treats private markets as constraints on the effects of policy whether the markets work well or not. Section II follows standard public economics and goes one step further by examining the inconsis- tency between the behavior of markets and the maximization of social welfare. It searches for ways to correct these market failures or for policies that improve welfare given these imperfections in the market. Such policies include the loca- tion of facilities and the ability of the health care system to correct pervasive insurance market failures. The constraints include private markets with signifi- cant market failures. I treat relevant budget and policy constraints implicitly. I do not derive much of the standard results in the literature; however, I highlight their less obvious implications for the health sector. I contrast the results of the formal model with the method of cost-effectiveness for dealing with the problem of allocation in health. "Cost-effectiveness" pro- vides a decision rule suggesting that activities should be undertaken if they have the highest ratio of the amount of some unit of output (lives saved or, as in a recent World Bank study, disability-adjusted life years) per dollar spent on the intervention (Jamison and others 1993). This ratio can determine the elements of a policy of providing or financing a basic package of services (refer to World Bank 1993 or Jamison and others 1993). I call this allocation rule "medical inter- vention cost-effectiveness" and contrast it with the concept of "public sector cost-effectiveness." (I am grateful to Lant Pritchett for coining these terms.) The latter is the net effect of public intervention compared with the case in which the government forgoes providing or financing the treatment (leaving it to other pro- viders). In this article I show the conditions under which medical intervention cost-effectiveness provides an adequate guide to decisionmaking, demonstrating that the conditions are so restrictive as to make these calculations usable only in very special circumstances, which are unlikely to be fulfilled in any developing country. I also identify the types of information that the government needs to collect in order to make correct public health decisions. A consistent theme running through the results is the insight that the govern- ment can best make use of its own resources by focusing on activities that most improve health or welfare beyond what would happen in the private market alone. The use of medical intervention cost-effectiveness is inadequate because it ignores the counterfactual of private sector activity and the scope of comple- mentary instruments such as prices. Similarly, this method does not deal with the extent to which market failure varies across health activities. Such variation provides the basis of the welfare maximization approach. I. HEALTH AS THE OBJECTIVE This section presents the basic model for maximizing improvement in health for a given budget and discusses the solution with condition-specific charges for health care, simplifications needed to justify medical intervention cost- effectiveness, and further extensions of the model. 412 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 The Basic Model The government wishes to run its public clinics in such a way as to im- prove health status as much as possible for a given amount of money. There are N possible disease conditions and interventions indexed by i. The health centers do not actively seek cases, although they may perform outreach in the form of follow-up for certain disease conditions. Improvement in the health outcome depends on the number of people who show up for a given disease condition and the effectiveness of treatment currently available. Each of the treatments entails a known cost, and a given budget limits the level of services. Health centers can charge for services, thereby relaxing this con- straint on the budget. The next section deals with the complication of whether the fees can be levied at different levels for different disease conditions. Health centers can decide not to treat a particular condition (if, for example, it is too expensive and will deplete the available budget too rapidly). The govern- ment is not necessarily the only provider of health care. The private sector, nongovernmental (usually nonprofit) organizations, traditional healers, and self-care by the afflicted person can all substitute for the provision of ser- vices in public facilities. A formulation of the problem that reflects these considerations can be written: N (1) max = ,{L.t[D B(1iB,fv)] Ii +14 [Di (PiBPiV ,Ii)1} p~~~Y[C BB I ipB V FB]J subject to ,[(Cy -Pi) D (Pi, Pi) + FB] Ii < R where Li is health improvement in people visiting sector j-with j = B (public) or V (private)-with condition i, Di is the number of people visiting sector j with condition i, PJI is the price charged for treating condition i in sector j, Ii is a variable that indicates whether condition i is treated by the public clinic (1 if yes, O if no), CQ is the unit cost of treatment for service i by the public clinic, R is the total public budget for health care, and F? are the fixed costs of including treat- ment i in the menu of public activities. The first line in the expression sets out the main objective of the government, which in this case is to maximize the health status, A, of the population through policies related to the public health care system. Whether care is given by public or private practitioners is not a concern in and of itself. The second line is the budget constraint for the public health system. Total subsidies (the difference between costs of provision and the price charged summed across all services provided) cannot exceed the total resources, R, available for public health care. This is not the budget constraint for the whole government nor for society as a whole; its use can lead to unacceptable results. However, the minister of health may find it the most relevant budget constraint. Hammer 413 The demand for public health services DIP(-) depends on prices in both the public and private sector. Demand for a particular disease condition, i, needs to be interpreted with some care. I do not assume that the individual knows the cause of illness; demand is really a function of symptoms rather than of disease. However, an unexpressed relation linking disease conditions to demand for ser- vices underlies the demand relation. This is captured in full generality by having demand functions that are specific to disease i and that can vary according to the usual severity of symptoms and the likelihood that individuals will seek help when the symptoms appear. The demand for services will bear a relationship to the incidence of the disease to the extent that specific symptoms are more or less likely to induce a search for care. For example, schistosomiasis may go unde- tected so that the gap between incidence and demand for treatment is large. Incidence and demand will be connected more closely in illnesses with more acute symptoms. The model captures the degree of substitutability between the public and pri- vate sectors by the responsiveness of demand to their respective prices. High own- and cross-price elasticities reflect a great degree of substitution. Other de- terminants of demand, particularly income, are held constant for this analysis. The concluding comments explicitly recognize demand by different groups ei- ther for undertaking ethical valuation of outcomes or for describing the pattern of use of public and private sectors. The health improvement function, L,P(-), in the public sector can represent lives saved or healthy life years saved conditional on someone with disease i showing up for treatment. A substantial literature discusses the appropriate measure for health status. For example, World Bank (1993) and Murray and Lopez (1994) use the concept of disability-adjusted life years (see Anand and Hanson 1995 for a critique of this measure). For the purposes of this article the choice of measure does not matter. The same sort of function L,v(-) applies in the private sector, either the same function or one altered to reflect differences in clinical effectiveness between the two sectors. In a country with a well- functioning public sector with very little private modern care, treatment outside public clinics could include either mostly very poor care by traditional healers or self-care. In other settings, a sophisticated private sector may exist and offer more effective treatment than the public sector, L7(-) > L,P(-), for certain ill- nesses. For psychosomatic or culturally determined illnesses, even the traditional sector may exhibit this property. Sometimes health improvement due to treating a condition-the functions, LP(-) and L1V(-)-is simply proportional to the num- ber of people being treated for the condition. That is, the marginal benefits to treating any given condition once someone has already presented with symp- toms may not vary with the number of individuals seeking treatment. This as- sumption could fail to be true for a variety of reasons, for example if severely afflicted individuals are more prone to seek treatment at any level of prices and if the improvement in health that results from treatment varies with severity of the disease. 414 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 In the following I assume constant costs of treatment in the public sector, C I, for each treatment. Treatment costs should include the costs of all diagnostic tests needed to identify an eligible disease and its treatment. To some extent, then, the analysis could attribute costs to diseases that are not covered by the health service if they take up diagnostic or other resources before the ineligibility is discovered. I assume that this effect is small. Similarly, I assume away other issues raised by the form of the cost function. In particular, I do not consider joint costs. This is a potentially serious limitation in practical application be- cause some policy decisions involve packages of services whose complementarity may well determine the appropriate choices. Having a certain piece of equip- ment with multiple uses in place may make some treatments worthwhile that otherwise would not be justified and that alone would not justify the purchase of the equipment. I cannot assume away the complication of the explicit treatment of fixed costs F?B(.). These could be interpreted as the cost of introducing an expensive piece of equipment that is likely to have excess capacity, but for a large enough country or catchment area, the amortized value of the machinery captures most of the costs. Instead, I justify fixed costs in terms of the costs of attention of policymakers or the perpetual increase in training of health care personnel who must learn a larger and more complex set of procedures. The analysis must include signifi- cant fixed costs to obtain results in which there is rationing by type of service, a principal feature of the design of basic packages. Without this assumption, the health care system needs no rationing other than by price. The price in the public sector is determined within the problem as a matter of policy. I assume different degrees of specificity of this price. In the most general case, health care providers may charge a separate price for each condi- tion. However, this degree of flexibility in the pricing structure may not be possible; therefore, I examine various constraints on the allowable price rules. The government does not exercise direct control of the price in the private sector. Two versions of private price determination are possible. In one the price in the private sector is given and is not affected by either the price charged or the inclusion of the disease in the protocols of the public sector. Strictly speaking, this version requires strong assumptions concerning the nature of production throughout the economy (not just in health). It can be justified, however, if the health sector is a small part of the economy and takes all its prices as given from outside. Because this includes the price of doctors' ser- vices, this is still not a great assumption. In the other version of private price determination, private sector prices and inclusion of diseases can affect the price charged in the private sector and thus have a secondary effect on total demand. A careful analysis of the connection between the private sector price and performance as influenced by policies within public clinics is beyond the scope of this article; however, the analysis captures the basic point that private sector behavior may be influenced by competition in price and rationing rules in the public sector. Hammer 415 The policy parameter I: takes on the value 1 if the public clinic provides the service and 0 if it does not. For the simplest analysis in the article, people's demand for services is independent of the range of services offered by the public sector. However, if the public clinic does not provide certain services, this lack of comprehensiveness may discourage people from using the public sector even when they have conditions that the clinic would treat. People's demand is deter- mined primarily by symptoms and not the actual disease condition. Thus, when individuals originally seek treatment, they do not know whether they will be turned away. Solution with Condition-Specific Charges The first-order condition for prices is: ay LB' aDr + aD pV I.+LY r aD aDy apV) (2) aps- i tapB apIv aB ipV apBi (2)X~p Ci -Pi )V ( iB + apB + ] -x[C -PB)!aD DD ~ - B B] = aB aV DpB2D1 0 where L' is the marginal health improvement resulting from a visit for condition i to sector j (B = public; V = private), X is the Lagrange multiplier for the budget constraint of the ministry, and pB is the change in the private sector price for condition i with respect to a change in the public sector price. Rearranging terms to solve for optimal prices, PB= 1 Bc e! 2,! i (3) Pi [eCB - D Dv where EP is the elasticity of demand for public service i with private prices changing as a result of the change in the public price, and e-Y is the cross-price elasticity of demand for private care, again with price changing in the new equilibrium. The expression for the optimal price (equation 3) indicates, first of all, that the price charged for a service will be such that demand for the service is elastic (EIs < -1) as is the case with all monopoly pricing models. Therefore, prices will be higher under the following conditions: * The demand for the service is less elastic. The higher the elasticity of demand for a service (ignoring for the moment its health effects), the greater the sacrifice in earnings by the public's health care system (given high elasticities) from further raising the price. Therefore, relatively more elastic services have lower prices (higher subsidies) at the optimum. 416 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 * The health consequence of raising the price in the public sector is smaller. The second term in equation 3 reflects the health impact (evaluated in money terms by the multiplier X) of a price rise. The term involving the B superscripts reflects the direct discouragement of treatment in the public sector, the term involving the V superscripts reflects the offset to this effect due to increases in demand in the private sector. The cross-price elasticity captures the extent of this offset. The greater the net reduction in the health status of the population due to a price increase for a particular service, the lower the price for that service. * The budget constraint is more binding. That is, higher prices result with a higher X, the term that converts money into health gains. The k term reflects the amount of health gain that an additional dollar of budget will buy if optimally deployed. As X increases (which happens if the budget R is cut), the second term in equation 3 decreases in absolute value. Because this term represents the amount by which prices are discounted in pursuit of better health, a decrease raises the price. For the rationing rule, solving a problem having fixed costs within an optimiza- tion model would, as a practical matter, present a difficult combinatorial problem. However, characterizing the solution for the purposes of this article is straightfor- ward. An additional procedure is included in the list of covered treatments only if offering the treatment brings a gain in health status larger than the loss of health from reducing the subsidy on all previously included treatments. The loss is the total cost of subsidizing the treatment, both its fixed component and the variable cost when optimal prices are charged. The gain in health status can be expressed as: (4) gain = L [Dp(JiB,1iv I11)]+I44Di1(YI 1, 1)] - ] [JD;[ (PjP1 11 =L 0,0)] where piv II, =10 is the price charged by the private sector when a service is (Ii = 1) or is not (Ii = 0) offered by the public sector. The loss in terms of health forgone from restricting other services is the cost of treatment: (Ci - PjB) Di(iB,pV i i multiplied by the Lagrange multiplier, X, which con- verts money into health gains. These considerations lead to a rationing rule in which a treatment is included if the following is true: (5) jL D (Ii = 1)] + Li [Di (I = 1)] - LY [Dv (I = 0) where to simplify the notation, I write demand as a function of the indicator variable, that is, with prices (not presented) allowed to reflect whether the service is provided in the public sector. The objective function (health status plus a term reflecting the budget constraint) increases with a new treatment if the improvement in health resulting from extra demand for services in the public sector net of any reduced demand in the private sector is greater than a term that reflects the extra cost to the system of providing the service. Hammer 417 The left-hand side of equation 5 is the net improvement in health due to offering a service per unit of subsidy to the public sector. The numerator is the health impact due to the direct provision of services, LiP, net of the offset of (presumably) lower use of private services (the difference in use, Li, with and without competition from the public sector). The denominator is the net cost of providing the service at prices Pi, determined simultaneously in equa- tion 3, or the per unit costs (Ci - Pi) times the demand in the public sector. This ratio represents health improvement per unit subsidy and should be compared with X, the implicit value of health, or the amount of health status improvement that would result in a unit increase in the budget allocation, R. The public health system should treat only conditions having ratios of ben- efit per unit subsidy higher than this. Equation 5 shows the need for fixed costs to justify the policy of ration- ing. Without fixed costs, any procedure should be offered, regardless of its effectiveness, provided it has any net benefit at all. Increasing the price to nearly the cost would make the denominator of equation 5 arbitrarily close to 0 and the full benefit greater than any value of B. With fixed costs, total benefits would have a further hurdle to overcome even if the service covered its variable costs. For both the pricing rule (equation 3) and the rationing rule (equation 5), the model handles private sector prices with a little sleight of hand. The elasticities in the price equation as well as the levels of demand in both sectors with and without public provision all depend on the prices determined by the private market equilibrium associated with the public price and rationing decisions. There is little information about these markets and their response to public policy. This indicates an important set of questions for research. In the current context, lowering a price in the public sector could induce a fall in the price in the private sector (especially if competitive cost considerations did not determine the origi- nal private price that, thus, included excess profits). The accompanying fall in price may limit both the number of people who leave the private sector and the number who come to the public sector. The lower cross-price elasticity of de- mand for private sector services would be reflected in both a lower optimal price and a higher likelihood of inclusion in the public sector. The main results pertaining to the rationing rule are the following: * The public sector should choose interventions that lead to the greatest improvement in health relative to what is being done for the same disease condition outside the public sector. * This relative improvement must be assessed in comparison with the budget impact of the intervention when the price for the service is set correctly rather than relative to the resource cost of the service. * The net health impact depends on the manner in which private markets respond to policy changes in the public sector: pricing and rationing rules depend on the degree of competition in the private sector. 418 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 * The public sector should set prices to balance two competing ends: limiting the adverse, net (of private sector response) health effect of a reduction in use of the particular service from higher prices and increasing revenue to allow provision of more services. The point is to stretch the public budget as far as it can go in achieving health gains. If the private sector provides a service of comparable quality, the public sector should not provide it or, at least, the service should not absorb much public subsidy. Its price in the public sector should be closer to cost. For condi- tions for which high prices dissuade many people from getting effective treat- ment in either sector, the public sector should, sensibly, charge lower prices. On the revenue side, the less elastic is demand, the higher is the price that can be charged without affecting health status, providing revenue the public sector can use to extend services or to increase the subsidy to other ailments. Some of these results are quite standard in optimal pricing theory. Besley (1988) and Barnum and Kutzin (1993) discuss them in greater detail. The inno- vation here is to link the problem with the rationing of services and to specialize the problem to the decision of the ministry of health rather than address the broader question of improving overall societal well-being. This perspective runs into a few very serious problems, as discussed below, but mirrors the particular interest of health decisionmakers. The main contribution is to underscore the central role of demand and other aspects of private behavior in the setting of priorities and prices within the public sector. This formulation captures the private response in twvo ways. First, on the demand side, consumers can choose to switch between providers at given prices. This leads to measuring the appropriate health improvement by the difference between the medical outcomes in the public and private sector weighted by the change in use of each sector due to price changes or rationing rules. Second, on the supply side (or in market equilibrium, taking both supply and demand into account) possible endogenous change in the private sector price may result from the pricing and rationing rules within the public sector. If the public sector de- cides to provide a service at subsidized prices, it becomes harder for private providers to charge much higher rates. If the sector is competitive, this results in private doctors going out of business and is ambiguous in its social contribution. If the sector is not competitive and exhibits elements of monopoly or more com- plex equilibrium conditions, the reduced price (or even the full marginal cost) can lead to reductions in the private sector price (inclusive of market distortions) as well and may have a second-round effect of increasing service use in both the public and private spheres. Pricing needs to balance competing needs. On the one hand, higher prices discourage users of health facilities. They could lead to substitution bv an inad- equate private sector (especially if dominated by self-care or ineffective tradi- tional healers), and they may have the indirect effect of increasing prices and thus discouraging use in a modern private sector. On the other hand, higher Hammer 419 prices raise revenue that can help to relieve the ministry's budget constraint. Although this obviously does not expand the treatment of the disease whose price is being raised, it can help to expand services (or, more important in this most general case, to reduce prices) for other disease conditions. Jimenez (1987) illustrates this tradeoff for social sector spending in general. Equation 3 allows for the possibility that certain services may actually make money for the ministry, that is, the appropriate fee could exceed the cost of provision. This is most likely to happen under two conditions. First, it is likely if demand for the service in the public sector is inelastic in the neighborhood of marginal cost. In this case, higher fees have little impact on health status but generate revenues until the point at which demand becomes elastic. Second, it is likely if cross-elasticities with the private sector are high and the private sector is at least as effective as the public sector. The contrast of these results with often-suggested cost-effectiveness analysis is striking. Standard formulations compare the effectiveness of the technique with the resource cost and suggest that techniques with high ratios, Li I Ci, be the only ones offered. Although analysts sometimes suggest using this method to allocate resources in the public sector, it appears to bear little relation to the solution presented in equation 5. Unrelated to the choice of public or private provision or to price policy issues, the ratio Li / Ci appears nowhere in the solu- tion as such. The gain relevant in these calculations is the one net of private response (L - LY). The relevant cost is the net subsidy and depends on the per unit subsidy (Ci- Pi), the change in use of public facilities (the number of units that must be subsidized), and fixed costs. Therefore, even within the very nar- row objective of health status, the cost-effectiveness of medical interventions is not relevant to public decisions without further assumptions. Simplifications Needed to Justify the Relevance of Medical Intervention Cost-Effectiveness In this section I discuss the simplifications that are simultaneously required to make cost-effectiveness analysis a legitimate criterion for setting priorities for public spending: free public provision of health care, costs of curative care that are strictly proportional to cases seen, and absence of a private sector. The cor- rect criterion for public provision is given in equation 5. When is this consistent with a high ratio of Li / Ci alone? FREE PUBLIC PROVISION. Rather than assuming that health care providers can charge any price or that they can distinguish prices by type of service, assume the opposite: that health care providers do not charge any fees at all. Instead, the ministry of health must fund its entire expenditure out of a fixed budget. This version of the model puts greater stress on the rationing rules because any activity in the public sector is a drain on government funds with nothing recouped in user charges. The assumption of no fees limits the total health package that the government can offer to the public. This scenario differs from the case in which 420 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 the public health sector can provide even frivolous (yielding little change in health status) activities by charging for them at full cost. In this case, equation 1 is changed to remove all terms p,' from the problem, and the budget constraint becomes: (6) _[CIB D (0, Pi) + FiB] Ii < R Because there are no public prices in equation 6, the solution is only in terms of the decision whether to include a particular treatment of a disease condition in the set offered. Equation 5 becomes: (7) L, [Di (Ii =L1)] + Li [Di (I, =1)] - Li [D (Ii = 0)] (7) )p~~~~C D + FB< Even with the constraint of free services, the logic of the optimal policy is the same. A treatment should be offered for free in the public sector if the ratio of the net gain in health status (net of substitution from the private sector) to costs exceeds a certain cutoff level. The substitution from the private sector should take into account any price changes that the existence of free public care induces in the private sector. Two differences mark the interpretation of the model in the case with no prices compared with the case with individual prices. First, the rationing rule in the case with no prices compares net health benefits with actual resource costs, Ci, rather than with subsidy costs, (Ci - Pi). Second, the rationing rule has real bite, even without explicit fixed costs. Some services certainly will not be provided because their prices cannot be arbi- trarily raised to match small health benefits with small subsidy costs; subsidy costs are technologically determined (given zero price) rather than determined as part of the solution. FORM OF TECHNOLOGY. If, in addition to free care, the technology of curative care is such that costs are strictly proportional to cases seen, or if Li(Di) = Li Di and FiB = 0 for all i, the rationing rule becomes: LB +v Di (Ii = 1) - Div (Ii = ) (8) D: (Ii = 1) > X ci< ABSEN-CE OF A PRIVATE SECTOR. In addition to free care and no scale effects of costs, let us assume one of three conditions is true. First, there is no private sector (Div = 0 for all i), or, second, the private sector is useless; that is, it generates no improvement in health status (Lv = 0 for all i), or, third, there is no cross- price elasticity of demand at all between public and private sectors, in the sense that private demand for services is completely unaffected by whether or not services are offered in the public sector: D7'(I = 1) = Dv(Ii = 0), for all i. If any of Hammer 421 these cases is combined with the assumptions of free care and proportional costs, the rationing rule for the public sector becomes: LB (9) i < ci Equation 9 denotes the rationing rule associated with standard cost- effectiveness analysis of curative care options. The health improvement associ- ated with a technology is divided by its resource cost. This ratio is higher for any included procedure than for any excluded one. Note that within this fairly gen- eral model, this decision rule is appropriate only when all of these assumptions hold-free care, public monopoly of all health services, and proportional costs. Even with all these assumptions a serious problem occurs in applying the cost-effectiveness ratios. This problem of the effectiveness per dollar of the least attractive procedure included involves society's valuation of life (or health). People's own evaluation of their own lives would likely differ from this number. Thus, if people's valuation exceeds that implicit in the ministry's budget and the ministry is doing the best it can with the budget R, society could do much better in terms of health and welfare than the monopoly position of the government allows. People with higher valuations of their own health status would want to pay more for services not allowed by this allocation mechanism. Serious ineffi- cierncy can result because this allocation has no place for personal preferences. Some Further Extensions of the Model This section discusses two extensions of the model: uniform user fees and public health interventions. UNIFORM USER FEES. I again abandon the assumption that prices charged in the public sector can differ by disease. Rather than having free care, health care providers charge a common fee for all disease conditions (or certain classes of conditions). The common fee is then a matter of policy as well. Musgrove (1986) performs a similar analysis with more aggregate disease groups. The solution for the single optimal price becomes: -X [cov(L, Th) + cov(I, l) +e L:B + L, P] + cov(cB,3 ,B) + cB _B (10) PB = cvL B yt iy I where ~ ~ ~ ~ ~ Eg 0s- Oi = Dl is the share of public visits accounted for by condition i, rqj = FIP Oi is the elasticity of demand for public visits for condition i weighted by the fraction of visits accounted for by this condition (the sum of these terms is v v D7 the elasticity of demand for all public visits), Ili = FpB Oi DB is the cross-price 422 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 elasticity of demand for private care for condition i with respect to the price of care in the public sector weighted by the ratio of private sector visits for this condition to total public sector visits, and X is the mean of X. Start by solving the general problem for a single service (the average of all services). That solution will, as in the general case, have a higher price for the service the lower is the elasticity of demand, taking private supply into account. Because demand is for a group of services, the uniform price should depend on the type of illnesses for which higher prices discourage treatment. If effective treatments are available and significantly improve health for the kinds of ill- nesses that fail to be treated (in either the public or the private sectors) because of an increase in the common price, this argues for limiting those increases. If price increases happen to discourage treatment for ailments with ineffective treat- ments, raising those prices more extends treatments for ailments with more ef- fective treatments. Similarly, if price increases happen to dissuade people from using particularly expensive treatments (hotding their effectiveness constant), this too argues for higher prices (because it results in greater savings that can be used to extend services). The relevant piece of information is the covariance between the elasticity of demand for public service and the marginal effective- ness (or cost) across treatments. All of these considerations depend on the degree to which the private sector picks up treatments. The common (public sector) price can be higher, the higher is the covariance between the cross-price elasticity of demand for private services and the effectiveness of those services. The public sector should charge a higher price if doing so disproportionately pushes people into the private sector for treatments that the private sector is capable of providing. The public sector should not charge a higher price if doing so pushes people with eminently treatable con- ditions (in the public sector) into the hands of unqualified private practitioners (traditional healers or self-care). If patients happen to sort themselves out in ways helpful to the public service (that is, they know the illnesses for vhich the public sector is most useful, or, by luck, they stop using the most expensive services), the public sector can charge a higher price. For the most part, this entirely empirical question depends on the knowledge of the general public or on cultural patterns. To some extent, though, the term cov(cB, r,B) would usually argue against higher prices. A larger gap between private and public prices would likely occur for expensive treatments (private practitioners charge closer to costs) than for cheap treatments. Therefore, the change in demand in the public sector would likely be greater for cheaper services than for expensive ones. The emphasis on covariances between elasticities and costs or effects parallels results in the optimal tax literature. When government assesses taxes on com- modities taking into account the effects on different income groups, it modifies the tax rates on each commodity according to the covariance of income shares of that commodity by different income groups and the marginal social valuation of income going to that group (presumably higher for the poor). This results in lower taxes on items disproportionately consumed by the poor (see Feldstein Hammer 423 1972). The objective function examined here is not sensitive to distributional concerns because it is hard, ethically, to distinguish between health outcomes per se by income group. If it were, however, those treatments that dispropor- tionately go to the poor would have lower prices. In the case with uniform fees, the common price would be lower if diseases afflicting poor people were dispro- portionately included in the package of provided services. The rationing rule that corresponds to the uniform price case is not different in form from those given in the previous sections. With a single price, the ration- ing rule does have real consequences even without fixed costs as in the case of free care because prices cannot be raised on individual services to cover less effective or more expensive treatments. The higher the price charged, the more services can be provided. Improved sorting of people (that is, lower covariances between service reduction and relative effectiveness in public facilities) can stretch the public health subsidy further. The appropriate price, and simultaneously the appropriate choice of treat- ments to offer, depends on the behavior of people and varies by area. Some stud- ies have examined the pattern of service use before and after price increases. There appears to be a wide variety in this pattern with demand for quite essential services falling in some areas (see Bennett 1989 on Lesotho) while more appro- priate selectivity (that is, less use of less effective services) occurs in others (see Gertler and Melnick 1993 on Indonesia). In deciding the type of price increase and the types of services offered, the government should consider the following features of the demand (and supply) structure of the medical services market: Own- and cross-price elasticities of demand for public and private services * The private sector response to public sector price and rationing policies * Covariances between elasticities of demand and the costs and effectiveness (again relative to the private sector) of treatments. PUBLIC HEALTH INTERVENTIONS. This framework can be modified to handle public health initiatives that do not rely on a clinic-based delivery system. Such initiatives include vector control (killing disease-bearing pests); information, education, and communications activities; and provision of public goods such as safe water or sanitation services. These can be incorporated into the analysis by modifying equation 1 in the following way: max Y = 1L. [D (PIB, F,K)] I, + Lv [I(1iBPV, K), K] + L(K)} subject to (Ci - piB) [Di( iB ,Iv) Ii] + C(K) < R. The direct investment, K, enters the equation in at least four possible places. First, it enters through the direct improvement in health from the intervention. The number of cases of malaria avoided by killing mosquitoes, for example, translates into a direct health benefit through the function L*(K). Second, the number of cases of a disease avoided translates into the demand for health ser- 424 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 vices in both the public and private sectors. This translation need not be one for one, that is, some people may suffer (and possibly die) from a disease without ever seeking help, and so the number of cases avoided need not be the same as the reduction in the number of people seeking help. Certain kinds of interven- tions may have no direct effect on disease but may increase the demand for services. For example, public information campaigns that make mothers more aware of symptoms of disease and more prone to seek help for them have no direct impact on health but increase the demand for services (public or private). Stimulating demand for preventive care such as immunization is another ex- ample. Third, information campaigns directed toward private providers can improve treatment currently given in the private sector (for an example of this in the context of improving the accuracy of drug prescriptions, see Hammer 1992). Fourth, the intervention entails a cost C(K). For many of the clinical interven- tions, the costs attached to the provision of service can be assumed to be con- stant without much problem. However, when direct, population-based inter- ventions are considered, the cost structure can be very important. Large fixed start-up costs for pest-control operations or information campaigns can gener- ate decreasing costs. On the other hand, costs of successfully providing services such as health information or water networks to more and more remote and sparsely settled groups within a population can lead to increasing costs. Equation 11 implies an additional first-order condition to determine the ap- propriate level of public health interventions: iFfL 1 ~ ~~ v D1e' L aD _ (12) C'(K) l aL aDB + aL= aD+ a + D. - (C, - P.) aDBaK aDjv K I aK aJKa The marginal cost of provision should be set equal to the benefit due to im- proved health (represented by the term multiplied by the conversion factor 1 / X) plus that due to budget savings. Two types of effects compose the health ben- efits. The expression aK captures direct health benefits from the investment itself. The expression Di aK reflects any improvement in the effectiveness of the private sector as a result of the investment, such as information, education, and communications activities directed at private providers. Against these direct improvements in health status must be counted the health improvement of cases that could be handled by the curative care system if people become ill and seek aL B aDP treatment. The terms involving DL dD and their equivalent for the private sector are negative because the effect of prevention, say, on demand for services should be negative. Prevention of diseases in which effective curative care is Hammer 425 being used is less important than prevention of diseases with no cure. However, the existence of curative care is irrelevant if people are not using it, and there- fore, the offset to the health effects of the prevention activity only comes from that part of the reduced incidence that is reflected in the reduced demand for care. Those who never seek care to begin with still reap the benefits. The budget- ary benefits for the ministry are captured in the last term in equation 12. Preven- tion is beneficial to the extent that it saves the ministry its subsidy to services. The existence of public health interventions can change the optimal pattern of subsidies and services provided, although there is no general rule concerning what these changes will be. Direct investments may lead to a disease condition either appearing or disappearing from the list of treated illnesses because the budget constraint binds more tightly and the cutoff level of effectiveness, X, rises. The disease would disappear from the list, of course, if the investment leads to eradication of the disease. Alternatively, reducing the demand for a particular type of treatment to a low level may make it worthwhile to offer that treatment because the budgetary impact will be much less (the denominator in equations 6 or 11 will be lower). In the case of disease-specific charges, a reduction in de- mand will show up as a possibly higher subsidy rate on treating the residual cases of the disease. Public health interventions can also affect the pattern of own- and cross-price elasticities if the people (say, remote rural dwellers) who are most affected by the public health interventions vary systematically from other people in the sector in which they seek treatment or in their response to prices. For example, if the demand for public services in remote areas is less elastic (because fees are a smaller fraction of the total cost of seeking treatment due to transport and time costs), market demand elasticity will rise in response to reduced incidence of disease. The control activities can change the market- level elasticities of demand for treatment and therefore the appropriate price and rationing rule for those treatments. Four propositions summarize the basic results of including direct interven- tions in the analysis of a health care system. Preventing health problems for which there is effective care is less valuable than preventing problems with no solution. Conversely, the existence of effective primary prevention methods can influence the degree of subsidy and the decision to treat an illness. The direction of this effect is not obvious, however. The use of funds for primary prevention makes the budget constraint bind more tightly and knocks some diseases off the treatable list. At the same time, effective prevention may make treating any remaining cases less burdensome when prices are not fully controllable. * Preventive activities are more valuable when care for the conditions they prevent is heavily subsidized. * Public health investments can change the appropriate pricing and rationing rules for health care delivery. Improving the quality of private care may obviate the need to provide free public care for the same condition. 426 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Improvement in the sorting of patients by disease conditions (in the case of uniform user fees) may allow a higher fee and a greater range of services to be provided. Information campaigns that increase the demand for effective services in the private sector (relative to subsidized services in the public sector) are better than those that increase demand for public sector services. The last proposition is particularly sensitive to the initial statement of the problem in which only overall health outcomes and the public budget concern the relevant policymaker (the minister of health). In the following section, in which I consider welfare, other, less appealing aspects of the private market may modify this result. II. WELFARE AS THE OBJECTIVE Up to this point I have based my argument on the assumption that only health outcomes and only the budget of the ministry of health matter in public decisionmaking. This position is not tenable. Consider the following proposed investment in the context of a network of public health facilities (and no private sector). Everyone goes to a clinic when they are ill. However, some people who live far from the closest clinic have to expend much money and forgone earnings to get there. The ministry is considering adding one clinic in a particularly re- mote area that would decrease the travel time significantly for many people (travel time is worth much more than the cost of the facility) but would not improve health at all (everyone already gets needed treatment). Should the min- istry build the facility? The decision rule implicit in the analysis above answers unequivocally no because the investment obtains no extra health from the scarce public health budget. From society's point of view, however, the answer is cer- tainly yes because, as assumed, the savings from reduced travel time and money outweigh the cost of the facility. Maximizing health status alone does not an- swer a large set of questions a policymaker must address. Rather than press the health maximization model too far by adding more complications, I reinterpret standard formulations of welfare economics. An es- sential feature of the health sector is the pervasiveness of market failures. Some of these market failures compromise health, but some manifest themselves in financial or utility losses not captured by health status. The different versions of the preceding model establish the theme that governments should focus on ac- tivities that make the greatest improvement relative to the status quo of private markets. I extend this argument to the more general case of utility maximization by correcting market failures characteristic of the health sector or making in- vestments that most improve welfare given uncorrected market failures. The literature on the welfare economics of policy reform and project evalua- tion identifies conditions in which welfare improves either as a result of policy reform (changes in prices essentially, although other reforms can be interpreted Hammer 427 in this way) or as a result of a direct investment (see Boadway 1975, Dreze and Stern 1987, Squire 1989, and Kanbur 1991). In the evaluation of investments, the method through which welfare is improved determines the appropriate prices (or the shadow prices) used to value the outputs and inputs of the investment. If the investment makes a profit at these prices, it increases social welfare and should be made. Here I use the notation in Kanbur (1991) and write a change in welfare as: Kax v~( ax 1 (13) dw= t E-1 ay dt + +t-1 dz _ aq ap) aq where W is a measure of welfare, q is a vector of consumer prices (marginal benefit) for a commodity, p is a vector of producer prices (marginal private cost) for a commodity, t is q - p or the distortion of prices in the economy, x is a vector of levels of consumption for each commodity, y is a vector of production levels for each commodity, E is the matrix of elasticities of net demands, and z is the vector of net inputs and outputs associated with a project. Based on the existence of the distortions, t, the government justifies both policy reform, dt, and direct project investment, dz. The first term in equation 13 indicates that there will be no improvement in welfare from any change in t if all distortions were originally 0. On the project evaluation side, it is less obvious because the term describing the change in welfare due to projects (dz) is com- posed of two parts. The second term in the shadow price calculation is the distortion-correcting component of the valuation. Investments receive a premium in the calculation if they lead to an expansion of consumption of goods with higher social than private valuation. The first term, p, is the actual private resource cost. If there are no distortions in the system, the value of a project is p - dz. This argument seems to indicate that an increase in welfare results from a public project if it turns a profit at private prices. However, the private profit criterion raises some questions as to the source of projects for evaluation. The usual, zero-profit condition for competitive equilibrium is (in the current nota- tion) p * dz = 0. In the absence of any distortions (induced by policy or private marrket failure), the condition that government must turn a profit is sufficient to justify an investment; however, such an investment is not likely to be found in a zero-profit equilibrium. We may agree with Hammond (1988) who says, "[Those], with more open minds, will at least wish to consider the possibility of there being some desirable projects which private sector corporations and entre- preneurs have overlooked." However, even if it can identify such a project, the government may not want to implement projects that are viable at private prices. Instead, it may simply inform potential investors about profit-making opportu- nities. Lack of private investors taking advantage of the information may point to distortions in the capital markets or to lack of confidence in the calculations. In the presence of distortions, the public sector has a clear role, and the lack of private investors in the project is less mysterious. 428 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 In the welfare literature, distortions usually result from policy-induced taxes (hence the mnemonic t) that drive a wedge between producer and consumer prices. In the case of the health sector, however, distortions are more likely to be caused by market imperfections in the private sector. For some health-related items, such as treatment of communicable diseases, the social value of consuming the good is greater than the private value. When evaluating the need for treatment, people do not consider the risk of infecting others. Therefore increases in the consumption of such goods, whether by policy changes (say, by subsidizing treatment) or by direct investment (by providing subsidized treatment or, in some cases, vaccination) should be evaluated at a premium over the private price of the service. In the case of pure public goods, or cases in which no private market can exist due to the inability to exclude a nonpaying user (vector control, some water supply problems), the whole value of the investment is attributable to the public intervention. Other market failures commonly associated with the health sector revolve around the problem of imperfect information. Here some difficult conceptual issues could arise, but the main problem is that people may not perceive the true value of the commodity because they lack knowledge of the effectiveness of treatment or the consequences of going without it. The health care literature frequently assumes that people undervalue the benefits of preventive activities (immunizations and lifestyle changes including cessation of smoking). These activities yield a social value greater than the perceived private value. How this value is determined is a major question, both in principle and in actual measure- ment. As far as the principle goes, the value needs to capture the effect of provid- ing the consumer with more complete information. The value is the marginal benefit under this better set of information. Medical professionals know certain kinds of information, such as the change in the probable incidence of disease with and without vaccination or the kinds and degrees of health improvements from alternative treatments. The patient knows other kinds of information, such as tolerance for pain, tolerance of uncertainty of outcome, or burdens put on family members due to disability, death, or financial cost. The appropriate in- formation set for the model with welfare as the objective combines personal situations and preferences with professional knowledge. Practically speaking, there could be two ways to identify and measure rel- evant information. One is to approximate the social value of a service under the augmented information set by (a) guessing (or researching) how many more people would use a given service if they were fully informed of the consequences of using it, (b) determining the elasticity of demand for the service (either using demand studies for the service under the status quo or using studies from better- educated populations or subpopulations within the same area), and (c) inferring how much higher the price under the new demand curve would be at the old, status quo, level of demand. Alternatively, the analyst could make an explicit valuation of the service using more than one dimension of valuation, thus giving the decisionmaker the freedom to use different weighting systems. Monetary Hammer 429 changes (or changes that may be easily converted into monetary values) could be added up in one dimension with various sorts of health outcomes left in a sepa- rate account. Social and private prices may diverge indirectly due to the lack of information on the part of consumers as a result of having to rely on medical professionals to suggest treatment. This principal-agent problem occurs because the incentives of the medical professionals differ from the incentives of a perfectly informed consumer. The principal-agent problem, identified 30 years ago by Arrow (1963), is at the core of attempts to model behavior in markets for medical care. Several researchers have advanced models of this phenomenon (Ellis and McGuire 1986, 1990; Selden 1990; and Pauly 1988). However, there is little consensus as to the most salient features to include or how these models might best be adapted to conditions in developing countries. In terms of the supplier-induced demand problem, an increase in the supply of doctors may not reduce the price of medi- cal services as the providers induce more, and more expensive, procedures in order to maintain income (Evans 1974). The providers can get away with this because consumers cannot second-guess the professional. Here, again, the true value of a service to the consumer differs from the supply price due to the decisionmaking of an agent, probably with different values and motives than satisfying the consumer. Imperfect information in the health sector has its greatest effect on insurance markets. One characteristic of health problems is that very expensive problems are relatively rare. This is why there should be great demand for insurance against catastrophic losses. However, most of the developing world has no market for health insurance except in small niche areas. The combined problems of adverse selection and moral hazard, which make health reform so difficult in industrial countries, prevent the very emergence of such markets in developing countries. A complete treatment of the insurance market and its effect on public health priorities is beyond the scope of this article (see Hammer and Berman 1995 for further discussion). However, the absence of insurance markets can impose a large gap between private and social benefits for different types of health care, particularly for expensive, catastrophic illnesses. Health services are almost always nontraded goods, that is, they are provided and used in the same place as opposed to ordinary commodities that can be sent. This distinction is important because the proper ratio of output to value is the net addition to consumption of an item, that is, net of adjustments in the market for the service that already exists. As in the analysis of health outcomes dis- cussed above, policymakers should value only net additions above what the market will supply. In evaluating public investment in the health sector, essen- tial information comes from changes in demand, supply, and the market equilib- rium in the private sector that results from the investment. This would be true even if all medical markets were competitive simply due to the nature of nontraded goods. When combined with the argument of the preceding paragraph, that the markets have noncompetitive characteristics (in ways that go beyond the stan- 430 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 dard noncompetitive models of markets), the need for convincing models of behavior takes on central importance. III. CONCLUSIONS This article attempted to derive price and rationing rules for public health facilities. It discussed the effect on these rules of different assumptions concern- ing the objectives of government (health versus welfare), the limitations on avail- able policy instruments, and the market environment in which the public system operates. The article highlights the need for analysts to assess policy reform in relation to the changes it induces relative to the status quo before reform. An obvious point, this finding identifies a distinct gap in the literature on resource allocation in health. In order to assess changes, analysts need to know the behavior of the private sector both in terms of the type of care that it provides and in terms of how this care will change as a response to policy reform. Substituting for a reasonably well-functioning private sector is not as valuable as providing ser- vices that the private sector cannot sustain. Research is needed into the charac- terization of market equilibrium for medical care and its response to policy mea- sures. Among the issues not examined here, the most important relate to uncertainty and insurance. In further research, these issues will have to figure prominently as major determinants of the demand for care. Originally identified by Arrow (1963), this line of research requires much more work. I have not focused my analysis here in terms of preventive or curative care. Instead I argue for the assessment of interventions on the basis of changes in the stated objectives of a public system. However, my analysis could connect with the preventive/curative dichotomy if there were reason to believe that preventive care would systematically lose out to curative care in a market setting. On the basis of people's generally acknowledged undervaluation of preventive services, this may well be the case. Various prevention activities also have many public good features with few private alternatives. Such activities will look good when analysts examine all interventions for improvements over the status quo. How- ever, analysts should evaluate all activities in terms of their improvement over market provision and not prejudge each case for certain types of intervention. REFERENCES The word "processed" describes informally reproduced works that may not be com- monly available through library systems. Anand, Sudhir, and Kara Hanson. 1995. "Disability Adjusted Life Years: A Critical Review." Working Paper 95.06. Center for Population and Development Studies, Harvard University, Cambridge, Mass. Processed. Arrow, Kenneth. 1963. "Uncertainty and the Welfare Economics of Medical Care." American Economic Review 53:941-73. Hammer 431 Barnum, Howard, and Joseph Kutzin. 1993. Public Hospitals in Developing Countries: Resource Use, Cost, Financing. Baltimore, Md.: The Johns Hopkins University Press for the World Bank. Bennett, Sara. 1989. "The Impact of the Increase in User Fees: A Preliminary Investiga- tion." Lesotho Epidemiological Bulletin 4:29-37. Besley, Timothy J. 1988. "Optimal Reimbursement Health Insurance and the Theory of Ramsey Taxation." Journal of Health Economics 7:321-36. Boadway, Robin. 1975. "Cost-Benefit Rules in General Equilibrium." Review of Eco- nomic Studies 42(September):361-73. Dreze, Jean, and Nicholas Stern. 1987. "The Theory of Cost-Benefit Analysis." In Alan Auerbach and Martin Feldstein, eds., Handbook of Public Economics, pp. 909-89. Amsterdam: North-Holland. Ellis, Randy, and Thomas McGuire. 1986. "Provider Behavior under Prospective Reim- bursement." Journal of Health Economics 5:129-52. . 1990. "Optimal Payment Systems for Health Services." Journal of Health Eco- nomics 9(4):375-96. Evans, Robert G. 1974. "Supplier-induced Demand: Some Empirical Evidence and Im- plications." In Mark Perlman, ed., The Economics of Health and Medical Care. Lon- don: Macmillan. Feldstein, Martin. 1972. "Distributional Equity and the Optimal Structure of Public Prices." American Economic Review 62:32-36. Gertler, Paul, and Glenn Melnick. 1993. "Does the Price Elasticity of Demand for Medical Care Depend on Severity of Illness?" Rand Corporation, Santa Monica, Calif. Processed. Hammer, Jeffrey S. 1992. "To Prescribe or Not to Prescribe: On the Regulation of Phar- maceuticals in Developing Countries." Social Science and Medicine 34(9):959-64. Hammer, Jeffrey S., and Peter Berman. 1995. "Ends and Means in Public Health Policy in Developing Countries." In Peter Berman, ed., Health Sector Reform in Developing Countries. Cambridge Mass.: Harvard University Press. Hammond, Peter. 1988. "Principles for Evaluating Public Sector Projects." In P. G. Hare, ed., Surveys in Public Sector Economics. Oxford: Blackwell. Jamison, Dean T., W. Henry Mosley, Anthony R. Measham, and Jos&Luis Bobadilla, eds. 1993. Disease Control Priorities in Developing Countries. New York: Oxford University Press for the World Bank. Jimenez, Emmanuel. 1987. Pricing Policy in the Social Sectors: Cost Recovery for Edu- cation and Health in Developing Countries. Baltimore, Md.: The Johns Hopkins University Press. Kanbur, Ravi. 1991. "Projects versus Policy Reform." Proceedings of the World Bank Annual Conference on Development Economics 1990, pp. 397-413. Washington, D.C.: World Bank. Murray, Christopher J. L., and Alan D. Lopez. 1994. Global Comparative Assessment in the Health Sector: Disease Burden, Expenditures, and Intervention Packages. Geneva: World Health Organization. Musgrove, Philip. 1986. "What Should Consumers in Poor Countries Pay for Publicly Provided Health Services?" Social Science and Medicine 22:329-33. Pauly, Mark. 1988. "Market Power, Monopsony, and Health Insurance Markets." Jour- nal of Health Economics 7:111-28. 432 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Selden, Thomas. 1990. "A Model of Capitation." Journal of Health Economics 9:397- 409. Squire, Lyn. 1989. "Project Evaluation in Theory and Practice." In H. B. Chenery and T. N. Srinivasan, eds., Handbook of Development Economics, pp. 1093-137. Amsterdam: Elsevier Science Publishers B.V. World Bank. 1993. World Development Report 1993: Investing in Health. New York: Oxford University Press. THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3; 433-50 Formal and Informal Regulation of Industrial Pollution: Comparative Evidence from Indonesia and the United States Sheoli Pargal, Hemamala Hettige, Manjula Singh, and David Wheeler Economic theory and recent empirical work suggest that when formal regulation of pollution is absent or less than 100 percent effective, affected communities are often able to negotiate abatement from plants in their vicinity through -informal regula- tion." Using a model of equilibrium pollution, this article confirms the existence of significant informal regulation for unregulated pollutants in both Indonesia and the United States as well as for regulated pollutants in the United States. Combining plant-level data with community data in both countries, regressions reveal that even after controlling for traditional economic variables such as output levels and input prices as well as for plant characteristics such as industrial sector and age, the per capita income of affected communities significantly affects pollution intensities. Higher-income communities win significantly lower emissions in both countries and for both unregulated and regulated pollutants in the United States, presumably be- cause income affects both preferences for environmental quality and the ability to bring pressure on polluting factories. This article starts from the premise that governments, using the various in- struments at their disposal, act as agents of the public in regulating pollution. However, when formal regulatory mechanisms are absent or ineffective, com- munities seek other means of translating their preferences into reality. Recent empirical work indicates the widespread existence of such "informal" regula- tion through which communities are often able to negotiate with or informally pressure polluting plants in their vicinity to clean up. Formal regulatory mechanisms include both command and control instru- ments (effluent concentration standards, technology standards) and market-based instruments (emissions charges, abatement credits, tradable permits). Informal regulation also takes many forms, including demands for compensation by Sheoli Pargal, Hemamala Hettige, and David Wheeler are with the Development Research Group at the World Bank. Manjula Singh is with AT&T and was formerly a consultant at the World Bank. Support for this work was provided by the World Bank's Research Support Budget (RPo 680-20). The authors thank Robert McGuckin and staff at the Center for Economic Studies, U.S. Census Bureau, who collaborated with them on the U.S. work; their colleagues in BPS (Central Statistics Bureau) and BAPEDAL (National Pollution Control Agency, Environment Ministry), government of Indonesia, for sharing their time, experience, and information; and, in particular, Mr. Nabiel Makarim (BAPEDAL) and Mr. Rifa Rufiadi (BPS). C) 1997 The International Bank for Reconstruction and Development/ THE WORLD BANK 433 434 THE WORLD BANK ECONOMIC REVIEW, VOL. 11. NO. 3 community groups, social ostracism of the firm's employees, the threat of physi- cal violence, boycotts of the firm's product, and efforts to monitor and publicize the firm's emissions (for examples, see Pargal and Wheeler 1996). Implicitly, such actions force firms to recognize the community's property rights in the local environment. They frequently work because firms do not operate in a so- cial vacuum. When informal regulation is effective, local factories face a positive expected penalty for polluting. Informal regulation need not be limited to cases where formal regulation is absent. If formal regulatory standards and institutions exist, the most effective informal regulatory tactic may be to report violations of legal standards. This is easier if information on standards is widespread, monitoring is relatively costless, and violators are easy to identify. A second "formal" channel for informal regu- lation is to pressure regulators to tighten their monitoring and enforcement. Profit-maximizing firms reduce pollution to the point where the marginal cost of abatement equals the expected marginal penalty for noncompliance. Because regulators have limited resources, polluters know that many regulatory viola- tions will go undetected or be lightly penalized. For recent evidence from North America, see Deily and Gray (1991); Dion, Lanoie, and Laplante (1996); Magat and Viscusi (1990); and Russell (1990). Similar evidence for Asia can be found in Dasgupta, Huq, and Wheeler (1997); Hartman, Huq, and Wheeler (1997); O'Connor (1994); and Wang and Wheeler (1996). In this context, public pres- sure on regulators can be an important countervailing force. Our thesis is that such informal regulation will be likely wherever formal regulation leaves a gap between actual and locally preferred environmental qual- ity. If our hypothesis is correct, we would expect widespread informal regula- tion in developing countries where formal regulation of pollution is absent or ineffective. However, informal regulation may also be common in industrial countries that have nationally uniform regulatory standards. In this case, formal and informal pressure on plants to abate may vary significantly with local pref- erences and organizational capabilities. We would expect informal regulation to be strongest in richer communities because they have stronger preferences for environmental quality and more knowledge about the risks of pollution. In ad- dition, such communities are more capable of exerting political, social, and eco- nomic pressure. In this article, we use plant and community data from the United States and Indonesia to test for the effect of informal regulation in a general model of "equilibrium pollution." This model, first developed in Pargal and Wheeler (1996), posits the existence of community-level emissions equilibria at points determined by local environmental demand and supply schedules. Industry's environmental demand (or pollution) schedule reflects the marginal cost of abate- ment. The position and slope of the schedule are pollutant-specific and depend on three sets of variables: the expected cost of pollution, determined by formal and informal regulation, standard economic factors such as abatement scale economies and relative input prices, and a variety of plant and firm characteris- Pargal and others 435 tics such as vintage of the equipment, efficiency, and ownership. The positions and slopes of environmental supply schedules for specific pollutants also reflect three sets of factors: community perception of damage, valuation of that dam- age, and the ability to impose costs on polluting facilities via formal or informal regulatory channels. The intersection of environmental demand and supply schedules is, of course, a familiar conceptual device in environmental economics textbooks (see, for example, Tietenberg 1992). However, the conventional treatment identifies "op- timal" pollution at the intersection of marginal cost and marginal damage sched- ules, implicitly assuming transactions costs to be zero for the regulator. By con- trast, we assume that formal regulation and informal regulation are affected by the costs of information, organization, and enforcement. Thus local equilibrium emissions may differ significantly from "optimal" emissions, which are derived solely from marginal cost and damage functions. Of course, they may also differ significantly from the emissions that are mandated by formal regulatory standards. The United States and Indonesia provide good test cases, because they occupy nearly opposite ends of the regulatory and socioeconomic spectra. In the United States, many air and water pollutants have been regulated at the national level for more than two decades. The U.S. Environmental Protection Agency (EPA) has a large staff, good technical facilities, and a record of tough enforcement. Indonesia, by contrast, had no regulatory standards before 1992, and at present, enforcement of these standards is mostly limited to a few water pollutants. The monitoring and enforcement capabilities of Indonesia's National Pollution Con- trol Agency (BAPEDAL) are extremely limited, technical staff are almost nonexist- ent, and very few actions have been taken against noncompliant facilities. The same extremes apply to socioeconomic data. The United States is one of the world's richest, best-educated societies, while Indonesia has only begun to reach lower-middle-income status after three decades of rapid growth and in- dustrialization. Both societies are, however, extremely diverse both geographi- cally and socially. In addition, both have large differences between incomes in different geographical communities. Thus it is plausible to assume that informal regulation may play an important role in both societies. We would certainly expect equilibrium emissions per unit of output to be higher in Indonesia. How- ever, we would also expect substantial intercommunity differentials in both coun- tries and, possibly, an overlap in the tails of the distributions. In this article, we compare econometric analyses of plant-level emissions across communities in the two countries. Our geographic units are counties in the United States and kabupaten (subprovincial units) in Indonesia. For the United States, we include regressions for emissions of two regulated air pollutants (total sus- pended particulates, TSP, and sulfur dioxide, SO2); two regulated water pollut- ants (biological oxygen demand, BOD, and total suspended solids, TSS); and total emissions of toxic compounds, many of which are not formally regulated at present. For Indonesia, we reproduce results for BOD that were previously re- 436 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 ported in Pargal and Wheeler (1996). BOD is the only pollutant for which large- sample data are currently available in Indonesia. Section I briefly describes the institutional setting in the United States and Indonesia. Section II describes our model of equilibrium emissions under formal and informal regulation. Section III specifies the regression equation and de- scribes our data. Section IV reports results and implications, and section V pre- sents a summary and conclusions. I. INSTITUTIONAL BACKGROUND Socioeconomic conditions and regulatory history are quite different in Indo- nesia and the United States. Yet both countries have conditions that give rise to large pollution differentials among communities. Regulation in the United States Criteria air and water pollutants have long been regulated by command-and- control methods in the United States. Air pollution control has historically been based on uniform emissions standards, starting with creation of the EPA and pas- sage of the Clean Air Act Amendments of 1970 (Tietenberg 1992). EPA regula- tory acts have included ambient air quality standards, the designation of areas as nonattainment regions, the application of guidelines for "prevention of signifi- cant deterioration," and mandated installation of specific control technologies at the plant level. Substantial noncompliance penalties exist, and there is little flex- ibility in the imposition of environmental regulations under the Clean Air Act. The control of conventional water pollutants has also been based on national effluent standards at the industry level, which are derived from technological specifications. There is less emphasis on uniform ambient standards than there is for air quality regulation, possibly reflecting the different impacts of pollution in the two media. At present, responsibility for regulating the manufacture and use of toxic substances is shared by several agencies including the EPA, the Food and Drug Administration (FDA), and the Occupational Safety and Health Ad- ministration (OSHA). The EPA regulates the handling, shipping, and disposal of hazardous or toxic pollutants in the United States. Legal recourse for damage under common law is a valuable complement to these regulations. Some toxics are clearly very dangerous, but many others have uncertain effects on human health and ecosystem functioning. At present, the EPA does not have safety or "emission/effluent" standards for the majority of toxic pollutants. As an aid to community awareness of exposure risk, however, it has since 1987 published an annual Toxic Release Inventory (TRI). The TRI, which was legislated in 1986 by the Emergency Planning and Community Right-to-Know Act, is based on man- dated disclosure of toxic releases and transfers by U.S. industrial facilities. Sub- stantial penalties can be imposed for failure to comply with reporting require- ments, but the TRI is basically a public information tool. Local governments or community groups must assess the performance of listed firms in their vicinity Pargal and others 437 and act on this information as they deem fit-by negotiation, public appeals, citizen suits, and so forth. To summarize, the United States has administered a system of formal, command-and-control regulation of criteria air and water pollutants for more than two decades. Limited use of tradable pollution permits has begun, notably in the case of the national trading program for S02 emissions permits under the revised Clean Air Act. For the most part, however, the U.S. system continues to rely on command-and-control regulation rather than on pollution taxes or trad- able pollution permits. Analysis of plant-level emissions of these pollutants should therefore provide a good test of the degree to which local environmental demand- supply considerations complement the enforcement of national uniform stan- dards in pollution control. By contrast, the lack of formal regulation for toxic emissions provides a good case for analyzing the influence of informal regula- tion unaided by national standards. Regulation in Indonesia Indonesia began formal regulation in 1992, with establishment of maximum allowable volumes and concentrations (in kilograms per ton of output) for emis- sions of BOD and other water pollutants from 14 broadly defined industry sec- tors (such as textiles and wood pulping). Although self-reported BOD emissions are now mandated by law, reporting has been extremely sparse. Expansion of self-reporting has recently begun under the PROPER program for rating and dis- closing the environmental performance of industrial facilities (see Afsah and Wheeler 1996). Until 1995 the only program of monitoring and pressuring for compliance was a voluntary arrangement instituted in 1989. This PROKASIH, or Clean Rivers, program covers about 5 percent of Indonesian manufacturing fa- cilities in 11 river basins on the islands of Java, Sumatra, and Kalimantan. Al- though it has succeeded in eliciting significant pollution reductions from some of Indonesia's largest polluters, PROKASIH represents only the first stage of regu- lation. See Afsah, Laplante, and Makarim (1996) for a detailed analysis of the PROKASIH program. Formal regulation of air and toxic pollution has only re- cently been introduced. Thus Indonesia provides a good test of the significance of informal regulation in a developing country where formal regulation is still in its infancy. II. A MODEL OF EQUILIBRIuM EMISSIONS UNDER INFORMAL REGULATION This article presents a summary version of our model of informal regulation as developed in Pargal and Wheeler (1996). The model follows convention in defining emissions as the use of "environmental services"-an additional factor of production in an augmented KLEM (capital, labor, energy, materials) frame- work. The implicit "price" of pollution is the expected penalty or compensation exacted by the affected community. It is different from other input prices in that it may be plant specific. Optimizing communities may tolerate polluting facto- 438 THE WORLD BANK ECONOMIC REVIEW, VOL. II, NO. 3 ries when they provide significant employment, local contracts, or tax revenues. Conversely, they may pay particular attention to plants whose location makes them easy to monitor, such as large, isolated facilities, or whose emissions are particularly damaging to the local environment, such as pulp mills immediately upstream from local fisheries or irrigated fields. Informal Regulation and Coasian Economics There is clearly a relationship between the concept of informal regulation and the Coasian view of environmental economics. Both consider the process by which externalities are internalized in the absence of regulatory agents, and both acknowledge that an externality can be created by the action of either a polluter or pollutee. An externality can be generated when a household or firm moves into proximity with a polluter. In the law, this is called "coming to the nui- sance" (Cooter and Ulen 1988: 181). For further discussion, see Calabresi (1970), Coase (1960), Coelho (1975), Demsetz (1967 and 1988), Diamond (1974), Hartman (1982), and McKean (1970). However, the traditional Coasian solu- tion depends on a well-defined legal and institutional environment. Efficient outcomes require accurate information about pollution problems, clear delinea- tion of environmental property rights, and courts that are able and willing to enforce legal agreements. We do not question the relevance of these conditions in many cases, but our view of informal regulation extends the Coasian view in two ways. First, com- munity pressure clearly affects polluters' behavior even where there is little ac- curate information, no explicit agreement about environmental property rights, and no court system able to adjudicate settlements. Second, our research and other studies suggest that the conceptual distinction between Coasian arrange- ments and traditional regulation is often blurred in practice. Regulators must confront strategic behavior by polluters and frequently renegotiate compliance schedules and penalties. And community pressure may significantly affect these interactions, creating de facto informal regulation even in countries where in- formation is plentiful, formal regulations are clearly defined, and the legal sys- tem functions effectively. Environmental Supply Informal regulation reflects local factories' implicit acceptance of the community's property rights in the environment. Communities use their lever- age to impose penalties (costs) on firms whose emissions are judged to be unac- ceptable. As factories use up more local environmental quality, affected commu- nities impose higher costs. From the viewpoint of industry, the result of informal regulation is an environmental supply schedule that shifts inward as average community income increases. Field survey evidence from Southeast Asia sug- gests that this schedule depends on several factors: the level of community orga- nization, information, legal or political recourse, media coverage, the presence of nongovernmental organizations, the efficiency of existing formal regulation, Pargal and others 439 and the opportunity cost of time. Many of these factors are correlated with community income levels. For more detailed discussion, see Huq and Wheeler (1993); Hartman, Huq, and Wheeler (1997); and Hettige and others (1996). This supply schedule is expected to be flat with respect to income only if formal regulation is completely successful in equalizing emissions across locations. Environmental Demand Faced with an environmental supply schedule, each plant adjusts pollution to the optimal point along its pollution demand schedule, derived from its cost mini- mization exercise. As noted in Pargal and Wheeler (1996), potentially significant determinants of the environmental demand schedule include industrial sector, output levels, relative input prices, vintage and efficiency of technology, and, in the case of Indonesia, ownership. The latter variable is not a significant source of variation in the United States given the preponderance of domestic, private firms in the economy. However, in Indonesia we might expect the effect on pollution intensity to be positive for state-owned firms and negative for multinationals. Pollution seems likely to be complementary to material inputs, but its cross- price relationships with labor, capital, and energy are not clearly signed. For this study, we construct local labor and energy price indexes for both the United States and Indonesia and a proxy for materials price variation across regions in Indonesia. The absence of capital price information is not a major concern, be- cause both of our samples are cross-sectional and we would not anticipate much within-country variation in the price of this factor. Well-managed plants should generate fewer waste residuals per unit of out- put and respond more readily to incentives for pollution control. To the extent that profitability reflects efficiency, well-run plants should also have more dis- cretionary funds with which to satisfy the demands for cleanup. Profitability, however, is a double-edged sword in this context, because firms that have avoided pollution abatement should have lower operating costs, other things being equal. Thus although efficient management should have an unambiguously negative effect on a plant's pollution intensity, proxies based on measures of profitability might have a "perversely" positive effect in regressions if the cost-saving compo- nent is dominant. Equilibrium Pollution Following the supply/demand derivation in Pargal and Wheeler (1996), we solve for the following reduced-form equation, which characterizes a plant's equilibrium pollution, Pit: (1) Pj1 = f(Wi, wVej, Wmp, Qp, si, vi, f, mi, g,, n1, a, yi). Right-hand variables for the regression equation are defined as follows, with expected signs of estimated parameters indicated in parentheses. See Pargal and Wheeler (1996) for a detailed discussion. There are five standard demand vari- ables: W,, is the manufacturing wage in county j (uncertain), Wej is the energy 440 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 price index in county j (uncertain), Wmj is the material input price index in county j (negative), Qi is the total output of plant i (positive, with elasticity less than 1), and si is the sector of plant i (uncertain). There are four firm-specific variables: vi is the age of plant i (positive), fi is the factor productivity of plant i (uncertain), mi is the multinational status of plant i (1 if multinational; negative), and gi is the public/private status of plant i (1 if public; uncertain). There are three infor- mal regulation variables: ni is the share of plant i in county j's manufacturing employment (uncertain), a, is the population density in county j (uncertain), and yi is the per capita income in county j (negative). The Roles of Industrial Sector and Location As previously mentioned, broadly defined industry sectors differ greatly in average pollution intensity of production (see Hettige and others 1994). How- ever, even in sectors with high pollution potential, emissions can often be sub- stantially reduced by modifying the process or installing end-of-pipe abatement equipment. Investors simultaneously choose products, processes, and abatement levels, taking relative prices at different locations into account. In general, we would expect emissions per unit of output to be relatively elastic with respect to the local "price of pollution." Possible Endogeneity of Income Our model employs community mean income as an exogenous variable, but we recognize the risk of endogeneity. Within an urban region, where residential mobility is comparable to factory mobility, an increase in pollution-intensive manufacturing in some areas may induce a decline in average income as richer people move away and lower property prices attract poorer people. This may well have been a significant factor in some U.S. metropolitan areas. If so, the result would be an upward-biased estimate of the impact of informal regulation (proxied by community income) on plants' abatement decisions. However, our results (section IV) suggest that simultaneity bias is not a significant problem for our U.S. regressions. In the Indonesian case endogeneity is far less likely. The units of analysis are kabupaten drawn from a broad spectrum of urban and rural areas in Java, Sumatra, and Kalimantan whose relative social and economic status has changed little since 1975. However, most of Indonesia's manufacturing has developed during the past two decades. Therefore industrial location clearly dominates residential migration in the Indonesian case. If there is any bias in our estimates, we are confident that it is small. Econometric Specification We have no strong prior views on appropriate specification of the estimating equation for equilibrium pollution. Because our theory of informal regulation has not been extensively tested, we start with a relatively simple and tractable empirical exercise. The pollution price variable is endogenous, with many deter- Pargal and others 441 minants, and there are many plant-specific demand-shift variables in the model. We therefore limit ourselves to estimating log-log regressions, using dummies for categorical variables. Heteroscedasticity, often a problem with cross-sectional analyses, is not a significant problem in our U.S. data. For Indonesia, we have reported White heteroscedasticity-consistent results. Although the correlation between differ- ent groups of variables is fairly significant in our Indonesia data set, multicollinearity is apparently not a problem for estimation. III. THE DATA U.S. data used for this study were obtained by merging establishment-level manufacturing data from the U.S. Census Bureau's Longitudinal Research Data- base (LRD), county income and population data from the U.S. Census Bureau's "U.S.A. Counties on CD ROM" (COSTAT), and EPA data from various sources: the TRI for toxics, the Aerometric Information Retrieval System (AIRS) for air emis- sions, and the National Pollutant Discharge Elimination System (NPDES) for water pollutant discharges. Indonesian manufacturing and socioeconomic census data were combined with observations on plant-level water pollution measured as part of the En- vironment Ministry's PROKASIH program during the period 1989-90. Our plant-level emissions data were provided by BAPEDAL. Data on plant charac- teristics and socioeconomic characteristics of communities were provided by BPS (Indonesia's Central Statistics Bureau) and are described in Pargal and Wheeler (1996). U.S. Variables For the U.S. case the dependent variables in our analysis include two air pol- lutants (SO2 and TSP), two water pollutants (BOD and TSS), and the total quantity of toxic releases. We measure the volume of emissions in kilograms per day for both BOD and TSS; and toxics, SO2, and TSP in pounds per year. Our data on U.S. plant characteristics, drawn from the LRD and COSTAT, in- clude measures of output value, plant age, and county-level employment share. As in the Indonesian case, we use value added per employee as a proxy for productive efficiency. We also include dummy variables for all nine two-digit U.S. Standard Industrial Classification (sic) manufacturing sectors. Data for our U.S. energy price index were obtained for each state for 1987 from the State Energy Price and Expenditure Report of 1991, published by the Energy Information Administration (1991). This is a composite index for the industrial sector created from coal, natural gas, petroleum, and electricity prices, in dollars per million British thermal units. The local manufacturing wage in the United States is computed from the LRD as the mean plant wage for each county. We lack direct measures of relative materials prices by sector and exclude them from the U.S. regressions. 442 THE WORLD BANK ECONOIVIIC REVIEW, VOL. 11, NO. 3 Data Description The U.S. data set used for toxics is the largest, with 12,005 observations com- mon to the LRD and TRI. The characteristics of more than 2,000 matching coun- ties vary significantly, with first and third quartiles of population density and per capita income ranging from 109 to 1,403 people per square mile and $10,000 to $13,075, respectively. Plant characteristics for the toxics sample are repre- sented by a mean age of 25 years, mean wage of $22,000 a year, and a mean local employment share of 5.6 percent. The air pollution data for the U.S. include 1987 emissions information from 878 plants in the AIRS data base that could be matched with the LRD. The match- ing community characteristics are less widely distributed, ranging between 139 and 1,281 persons per square mile for first to third quartiles of population den- sity and between $10,000 and $12,500 for per capita income. Plants in this sample have a mean age of 27 years, mean wage of $26,000 a year, and mean local employment share of 8.5 percent. The U.S. water pollution data come from a sample of 1,368 plants in EPA'S NPDES data base that could be matched with the LRD. Here the matching coun- ties are less densely populated: first and third quartiles range between 74 and 742 people per square mile. Per capita income is similar to that in the other data sets, with an interquartile range of $9,607 to $12,168. The plants have a mean age of 27 years and employ a mean of 4.1 percent of county workers at a mean annual wage of $26,000. For Indonesia, out of a total sample of 253 plants, we have ownership infor- mation on 246. Three are wholly foreign owned, 13 are completely owned by the government, and 178 are owned by Indonesians. Factory age ranges from 0 years (2 firms) to 90 years (2 firms), with the median age of firms being 10 years. The geographic spread of the data is restricted to three islands-Java (189 plants), Sumatra (40 plants), and Kalimantan (24 plants)-and eight prov- inces. Forty-one International Standard Industrial Classification (isIC) codes or sectors are represented in the data set, and firms range in size from 22 to 41,821 employees, with share in kabupaten employment varying from 0.02 to 91 per- cent. The kabupaten represented in the data are quite varied as well: 1990 popu- lation density ranges from 3.4 to 53,876 persons per square kilometer, the pro- portion of the population with more than a primary education varies between 6.85 and 48.5 percent, and mean annual per capita expenditure varies from Rp256,447 to Rp837,277 (1990 rupiah). IV. RESULTS The econometric results for Indonesia from Pargal and Wheeler (1996) are reproduced in table 1, while comparative results for the United States are re- ported in table 2. Because we regress log (emissions) on log (output), the regres- sion coefficients should be interpreted as partial effects on pollution intensity. Pargal and others 443 Table 1. Determinants of Pollution Intensity in Indonesia Coefficient Variable estimate Intercept 17.479* * (2.38) Economic variables Output 0.712*** (3.78) Wage -0.316 (0.52) Fuel price -1.267 (0.52) JAVA* -1.530*** (2.98) Plant/firm variables Value added per worker -0.312* (1.79) Age 0.179 (1.07) Foreign ownershipb 0.0004 (0.06) State ownershipb 0.021*** (2.91) Textilesb 1.247** (3.41) Leather tanningb 1.961** (3.14) Foodb 2.480*** (4.52) Pulp and paperb 2.265*** (3.98) Wood productsb -0.930 (1.02) Community variables Local employment share -0.313* (1.84) Income per capita -2.811*** (3.02) Percentage of population with -0.668 greater than primary education (1.17) Population density 0.128 (0.62) Number of observations 250 Adjusted R2 0.3805 'Significant at 10 percent. * *Significant at 5 percent. * " ' Significant at 1 percent. Note: The dependent variable is the log of biological oxygen demand (BOD) emissions volume in kilograms per day. Variables are in logs. t-statistics are in parentheses. a. A dummy variable that is a proxy for the crude materials price. b. Dummy variable. Source: Pargal and Wheeler (1996). 444 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Table 2. Determinants of Pollution Intensity in the United States, 1987 Biological Total Total oxygen suspended Sulfur suspended Toxic demand solids dioxide particulates Variable emissions (BOD) (TSS) (S0O) (TSP) Intercept 6.18*X8 3.38 5.54 21.73 16.77 (4.58) (0.85) (1.21) (1.78) (1.56) Economic variables Output 0.69**x 0.37*** 0.50**' 0.91*'. 0.88Y- (25.76) (4.86) (5.68) (3.64) (4.04) Wage 0.84*** 0.21 2.28*- - -3.03 1.39 (4.18) (0.39) (3.60) (1.56) (0.81) Fuel price -0.012 -0.09 -0.96** 4.48*' -0.13 (0.10) (0.27) (2.52) (2.69) (0.09) Plant/firm variables Value added per worker -0.16*** -0.03 -0.14 0.94** 0.39 (4.51) (0.31) (1.20) (2.91) (1.38) Age 0.05 -0.11 -0.17 -0.31 0.42 (0.83) (0.50) (0.66) (0.50) (0.76) Textiles and leathera 1.77*** 0.58 1.41*x. 2.11* 3.26* (13.16) (1.53) (3.24) (1.72) (3.02) Wood productsa 0.99**x -0.10 -0.02 -1.69 1.14 (7.67) (0.16) (0.03) (1.57) (1.20) Pulp and papera 1.78*** 2.98*** 2.93x * 0.78 -0.99 (14.76) (11.54) (9.85) (0.92) (1.32) Chemicalsa 1.44*** -0.53** 0.16 1.62*' 1.03 (16.01) (2.57) (0.66) (2.18) (1.58) Nonmetallic mineralsa 0.51*x -1.64*** -0.48 1.01 3.53 ' (3.50) (4.15) (1.05) (1.09) (4.36) Metalsa 1.50*** -2.03Xx. 0.11 -0.05 2.04x (13.02) (8.13) (0.39) (0.06) (2.69) Machinerya 1.03**x -1.87"" x -1.55x x -3.39x < 3.04x (11.78) (8.54) (6.14) (4.49) (4.58) Miscellaneous 0.92 -1.33 3.38""" -1.10 -3.33 manufacturing' (4.29) (1.37) (3.01) (0.39) (1.35) Community variables Local employment share 0.08"gx 0.04 0.10 0.27 -0.34 (3.28) (0.58) (1.17) (1.10) (1.56) Income per capita -0.59x- -0.51 -1.27X" -2.85" -2.69" (3.58) (1.08) (2.35) (1.90) (2.03) Population density 0.02 -0.04 -0.03 0.13 -0.43"" (0.54) (0.54) (0.35) (0.56) (2.02) Number of observations 11,827 1,343 1,343 869 869 Adjusted R2 0.196 0.315 0.273 0.211 0.205 *Significant at 10 percent. x Significant at 5 percent. *xSignificant at 1 percent. Note: Dependent variables are in logs of emissions volume. Toxic emissions, SO,, and TSP are measured in pounds per year, and BOD and TSS are measured in kilograms per day. Variables are in logs. t-statistics are in parentheses. a. Sectoral dummy variable. Source: Authors' calculations. Pargal and others 445 The single exception is the result for log (output) itself, which should be reduced by I for interpretation in emissions-intensity form. Standard Demand Variables Among the standard demand variables, only scale economies emerge consis- tently in our results. In all cases except S02 in the United States, the output elasticity of emissions is significantly less than 1. Thus emissions intensity gener- ally declines with plant output, reflecting scale economies of abatement. The input price results show no consistent pattern of complementarity or substitutability with labor and energy across pollutants. For Indonesian BOD emissions, only the crude materials price proxy (JAVA in table 1) is significant. In the United States, there is evidence of emissions-labor substitutability for TSS and toxic emissions, emissions-energy substitutability for SO2, and emissions- energy complementarity for TSS. Otherwise, cross-price effects are not signifi- cant. Within countries, labor and energy price variation seems to have had little effect on emissions intensity. Firm- and Plant-Specific Variables Except for sectoral differences, plant and firm characteristics show little evi- dence of consistent and significant impacts in these regressions. Value added per worker does have a negative, significant association with emissions intensity for the two variables that are not formally regulated: U.S. toxics and Indonesian BOD. For the formally regulated U.S. pollutants, however, the results for this variable are inconsistent. We find no effect for water pollutants (BOD and TSS), but a strong and significant positive association for S02. The association for TSP is also positive, although weaker. These results suggest a different balance in air and water emissions between pure efficiency effects and profitability from lower abatement costs. The effects of plant vintage are not strong, with some possible effect only for water pollution in Indonesia. This may not be too surprising, because even older facilities in the United States have operated under strict regu- lation for a long time. Our results for sectoral dummy variables are generally consistent with expec- tations about relative pollution intensity. In the case of Indonesia, where five sectors are amply represented in the data set, our results show that four are more BOD intensive than other manufacturing sectors: textiles, leather tanning, food products, and pulp and paper. The BOD intensity for wood products is not significantly different from the average for other manufacturing. In the U.S. case, we include dummy variables for all two-digit SIC sectors, using food prod- ucts as the excluded sector. This sector is known to have high intensity in or- ganic water pollution (BOD) and relatively low intensity in toxics and standard air pollutants. Our results follow this pattern and indicate the following relative intensities for other sectors: textiles and leather, pulp and paper, chemicals, and metals have the highest intensity in toxics, while food products have the lowest; textiles, nonmetallic minerals (principally cement), and metals are the most in- 446 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 tensive in TSP, while machinery and miscellaneous manufacturing are the least intensive; textiles and leather and chemicals are the most intensive in SO2, while machinery is the least intensive. Not surprisingly, only pulp and paper are sig- nificantly more intensive in organic water pollution (BOD) than food products, although textiles and leather tanning are also relatively BOD intensive, as in the Indonesian case; metals and machinery are the least BOD intensive. Pulp and paper and miscellaneous manufacturing are the most intensive in TSS, while machinery is the least intensive. To summarize, our results suggest that plant and firm characteristics are more important determinants of emissions intensity in weakly regulated economies than in strongly regulated ones. In the United States we find strong scale and sectoral effects, but no consistent effect for vintage and efficiency. However, the Indonesian results for vintage, efficiency, and public ownership are consistent with results obtained for Bangladesh, India, and Thailand in other studies (Huq and Wheeler 1993 and Hartman, Huq, and Wheeler 1997). Informal Regulation or Community Variables Our results suggest that informal regulatory forces are pervasive, even when strong formal regulation is in place. In both the United States and Indonesia, the elasticity of all emissions intensities with respect to community income is nega- tive, large, and generally highly significant (the only exception is U.S. BOD). The estimated elasticity with respect to community income for the regulated U.S. air pollutants (SO2 and TSP) is approximately equal to the elasticity for Indonesian BOD; U.S. water pollutants and toxic emissions have lower, but still substantial, elasticities. As discussed earlier, these results may reflect income-related differ- ences in both preferences and power. The income result is critical for our central hypothesis, and as previously noted there is some risk of upward bias in the estimated impact when per capita income is used as a proxy for local informal regulation. As might be expected, community income is highly correlated with community education in both coun- tries. After controlling for income, education does not emerge as significant in either set of results. We reproduce the estimated coefficient for schooling in Indonesia but drop education from the final U.S. regressions. The argument for endogeneity is based on co-location of poor communities with pollution-intensive industry sectors. By introducing sectoral dummy vari- ables, our equations measure the impact of income on pollution intensity after sector choices have been made. As we noted in the previous discussion, our results confirm what is generally known about the relative pollution intensity of industry sectors. With the sector controls in place, our regressions allow us to focus on determinants of within-sector pollution abatement. In an independent check for significant bias, we have analyzed the relation- ship between county income per capita and the location of the most highly pol- luting U.S. industry sectors. Using very different criteria, several empirical stud- ies (Robison 1988, Tobey 1990, Mani 1996, Hettige and others 1994, and Mani Pargal and others 447 and Wheeler 1997) have identified the same five three-digit SIC sectors as large outliers in air, water, and toxic pollution intensity: iron and steel (SIC 371), non- ferrous metals (sic 372), industrial chemicals (sic 351), pulp and paper (sic 341), and nonmetallic mineral products (sic 369). For the year 1987 (the same year as our plant sample), we estimate a log-log function that relates per capita income in all U.S. counties to the ratio of value added in "dirty" production (our five outlier sectors) to value added in other manufacturing activities. In this context, there can be a serious problem of simultaneity bias only if the estimated effect of income on dirty-sector share is large and negative. However, our results (table 3) show no relationship at all. They reinforce our view that the econometric result in table 2 reflects the impact of income (and informal regulation) on in situ abatement, not location. Results for the other two community-related variables in tables 1 and 2- plant share of local employment and population density-are mixed. Under in- formal regulation, there may be a significant "visibility effect" when polluting facilities can hide among other plants in urban/industrial areas. This would im- ply that population density should be positively associated with emissions inten- sity. When formal regulation ensures visibility by requiring emissions reports, however, another effect should dominate: more densely populated areas should exert pressure for greater cleanup because more people are adversely affected. This effect should be particularly strong for conventional air pollutants. In the United States, our results are consistent with this interpretation for TSP, the most visible air pollutant. However, no other U.S. pollutant has a significant relation- ship with population density. In the case of Indonesia, the positive association betwveen BOD intensity and population density is consistent with a strong "vis- ibility effect" in an unregulated economy, but the measured effect is weak. Our results for plants' share of local employment are also mixed and have no ready interpretation. Under informal regulation, the estimated impact of rela- tive size will reflect two considerations: the plant's visibility as a polluter and the benefits it brings to the community as an employer. In the United States, we would expect emissions reporting under formal regulation and, for toxics, the Table 3. County Income per Capita and Location of Pollution-Intensive Production in the United States, 1987 Variable Coefficient estimate Intercept -1.199 (0.66) Log (income per capita) -0.17 (0.88) Number of observations 2,464 Adjusted R2 -0.0001 Note: The dependent variable is log (VAD / VA,), where VAD is value added in the five most pollution- intensive three-digit sectors, and VAo is value added in all other manufacturing sectors. t-statistics are in parentheses. Source: Authors' calculations. 448 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 TRI to reduce or eliminate the visibility effect. We might therefore expect the employment benefit effect to dominate, with larger employers having greater emissions intensity, other things being equal. This is indeed the case, particularly for toxics. The exception is the most visible pollutant, airborne TSP. This has a marginally significant negative association, as does informally regulated BOD in Indonesia. V. SUMMARY AND CONCLUSIONS In this article, we used plant-level data from the United States and Indonesia to test a model of equilibrium pollution under informal regulation. The model yields a reduced form in which emissions intensity is related to variables in three categories: standard demand variables (scale, input prices), plant and firm char- acteristics (sector, efficiency, vintage, ownership), and informal regulatory vari- ables (community income, population density, plant size relative to local economy). We estimated emissions regressions for four formally regulated U.S. pollutants (TSP, S02, BOD, and TSS) as well as for unregulated pollutants (toxic pollution from U.S. factories and BOD emissions from Indonesian factories). Thus the full set of regressions also includes implicit controls for formal regulation and level of economic development. Our results suggest three common elements across countries and pollutants: (1) abatement is generally subject to significant scale economies, (2) within- country variations in labor and energy prices have little impact on pollution intensity, and (3) community incomes have a powerful negative association with pollution intensity. Although the plant and firm characteristics we have mea- sured seem important in Indonesia (and other Asian developing economies we have studied), only scale and sector have consistent, significant effects in the United States. Our findings on community income are particularly important, because they suggest a powerful role for informal regulation whether or not formal regula- tion is in place. The impact of income disparity on intercounty differences in U.S. pollution intensities seems to match the impact in Indonesia. Undoubtedly, this reflects differences in both preference for environmental quality and abilitv to exert pressure on polluting factories. The fact that such disparities exist in the United States, even for traditionally regulated pollutants, shows that U.S. regu- lation has not been able to ensure uniform environmental quality for all citizens regardless of income. REFERENCES The word "processed" describes informally reproduced works that may not be com- monly available through library systems. Afsah, Shakeb, Benoit Laplante, and Nabiel Makarim. 1996. "Programme-Based Pollu- tion Control Management: The Indonesian PROKASIH Programme." Asian Journal of Environmental Management 4(2, November):75-93. Pargal and others 449 Afsah, Shakeb, and David Wheeler. 1996. "Indonesia's New Pollution Control Pro- gram: Using Public Pressure to Get Compliance." East Asian Executive Reports 18(6):9-12. Calabresi, Guido. 1970. The Costs of Accidents. New Haven, Conn.: Yale University Press. Coase, R. H. 1960. "The Problem of Social Cost." Journal of Law and Economics 3(1):1-44. Coel]ho, P. R. P. 1975. "Externalities, Liability, Separability, and Resource Allocation: Comment." American Economic Review 65:721-23. Cooter, Robert, and Thomas Ulen. 1988. Law and Economics. Glenview, Ill.: Scott, Foresman, and Company. Dasgupta, Susmita, Mainul Huq, and David Wheeler. 1997. "Bending the Rules: Deter- minants of Discretionary Pollution Control in China." Working Paper 1761. Policy Research Department. World Bank, Washington, D.C. Processed. Deily, M. E., and W. B. Gray. 1991. "Enforcement of Pollution Regulations in a Declining Industry." Journal of Environmental Economics and Management 21:260-74. Demsetz, Harold. 1967. "Toward a Theory of Property Rights." American Economic Review 57(2, May):347-59. - -.1988. "When Does the Rule of Liability Matter?" In Organization of Economic Activity Series. Oxford: Blackwell. Diamond, Peter A. 1974. "Single Activity Accidents." Journal of Legal Studies 3(1):107- 64. Dion, Catherine, Paul Lanoie, and Benoit Laplante. 1996. "Monitoring Environmental Standards: Do Local Conditions Matter?" Working Paper 1701. Policy Research Department, World Bank, Washington, D.C. Processed. Energy Information Administration. 1991. State Energy Price and Expenditure Report of 1991. Office of Energy Markets and End Use, Energy Information Administra- tion, U.S. Department of Energy, Washington, D.C. Hartman, R. S. 1982. "A Note on Externalities and the Placement of Property Rights: An Alternative Formulation to the Standard Pigouvian Results." International Re- view of Law and Economics 2:111-18. Hartman, Raymond, Mainul Huq, and David Wheeler. 1997. "Why Paper Mills Clean Up: Results of a Four-Country Survey in Asia." Working Paper 1710. Policy Re- search Department, World Bank, Washington, D.C. Processed. Hettige, Hemamala, Mainul Huq, Sheoli Pargal, and David Wheeler. 1996. "Determi- nants of Pollution Abatement in Developing Countries: Evidence from South and Southeast Asia." World Development 24(12, December):1891-904. Hettige, Hemamala, Paul Martin, Manjula Singh, and David Wheeler. 1994. "lPPS: The Industrial Pollution Projection System." Working Paper 1431. Policy Research De- partment, World Bank, Washington, D.C. Processed. Huq, Mainul, and David Wheeler. 1993. "Pollution Reduction without Formal Regula- tion: Evidence from Bangladesh." Working Paper 1993-39. Environment Depart- ment, World Bank, Washington, D.C. Processed. Magat, W. A., and W. K. Viscusi. 1990. "Effectiveness of the EPA'S Regulatory Enforce- ment: The Case of Industrial Effluent Standards." Journal of Law and Economics 33(October):331-60. 450 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Mani, Muthukumara. 1996. "Environmental Tariffs on Polluting Imports: An Empiri- cal Study." Environmental and Resource Economics 7:391-41 1. Mani, Muthukumara, and David Wheeler. 1997. "In Search of Pollution Havens? Pol- luting Industry in the World Economy, 1960-1995." Working Paper 16. Research Project on Social and Environmental Consequences of Growth-Oriented Policies, Policy Research Department, World Bank, Washington, D.C. Processed. McKean, Roland. 1970. "Products Liability: Implications of Some Changing Property Rights." Quarterly Journal of Economics 84:611-26. O'Connor, David. 1994. Managing the Environment with Rapid Industrialization: Les- sons from the East Asian Experience. Paris: Development Centre, Organisation for Economic Co-operation and Development. Pargal, Sheoli, and David Wheeler. 1996. "Informal Regulation of Industrial Pollution in Developing Countries: Evidence from Indonesia." Journal of Political Economy 104(6, December):1314-27. Robison, D. H. 1988. "Industrial Pollution Abatement: The Impact on the Balance of Trade." Canadian Journal of Economics 21(February):187-99. Russell, C. S. 1990. "Monitoring and Enforcement." In P. R. Portney, ed., Public Poli- cies for Environmental Protection, pp. 243-74. Washington, D.C.: Resources for the Future. Tietenberg, Thomas. 1992. Environmental and Natural Resource Economics. New York: Harper Collins Publishers, Inc. Tobey, James A. 1990. "The Effects of Domestic Environmental Policies on Patterns of World Trade: An Empirical Test." Kyklos 43(2):191-209. Wang, Hua, and David Wheeler. 1996. "Pricing Industrial Pollution in China: An Econo- metric Analysis of the Levy System." Working Paper 1644. Policy Research Depart- ment, World Bank, Washington, D.C. Processed. THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3: 451-70 Capital Flows to Developing Countries: Long- and Short-Term Determinants Mark P. Taylor and Lucio Sarno This article focuses on the determinants of the large portfolio flows from the United States to Latin American and Asian countries during 1988-92. Cointegration tech- niques reveal that both domestic and global factors explain bond and equity flows to developing countries and represent significant long-run determinants of portfolio flows. The article also investigates the dynamics of portfolio flows by estimating seemingly unrelated error-correction models. Global and country-specific factors are equally important in determining the long-run movements in equity flows for both Asian and Latin American countries, while global factors are much more important than domes- tic factors in explaining the dynamics of bond flows. U.S. interest rates are a particu- larly important determinant of the short-run dynamics of portfolio, especially bond, flows to developing countries. International capital flows have recently been marked by a sharp expansion in net and gross capital flows and a substantial increase in the participation of foreign investors and foreign financial institutions in the financial markets of developing countries (World Bank 1997).l This expansion has been much greater than that of international trade flows (Goldstein, Mathieson, and Lane 1991 and Montiel 1993). It has been reinforced by the ongoing abolition of impedi- ments and capital controls and the broader liberalization of financial markets in developing countries since the late 1980s. Feldstein and Horioka's (1980) find- ing of low international capital mobility based on the correlation between the shares of savings and investment in gross domestic product (GDP) still remains a puzzle (see Obstfeld 1995). However, studies based on interest rate differentials generally provide evidence that there is a high and increasing degree of interna- tional capital mobility among the major industrial countries and among interna- tional and developing countries (Montiel 1993). 1. For a comprehensive review of some recent prospects and developments concerning capital flows to developingcountries, seeWorld Bank (1995, 1997), IMF (1994a, 1994b, 1995), and, for Latin American countries, United Nations Economic Commission for Latin America and the Caribbean (1994). Mark P. Taylor is with the Department of Economics at Oxford University and the Centre for Economic Policy Research, and Lucio Sarno is with the Department of Economics and Finance at Brunel University. This article was begun while Mark Taylor was a visiting scholar at the World Bank. The authors are grateful to Stijn Claessens, Ning Zhu, Eduardo Fernandez-Arias, Kausik Chaudhuri, and two anonymous referees for helpful and constructive comments on a previous version of this article. 1997 The International Bank for Reconstruction and Development / THE WORLD BANK 451 452 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Another main feature of recent capital flows to developing countries is that private (bond and equity) flows, as opposed to official flows, have become a crucial source of financing large current account imbalances: "This surge of portfolio investment combined with the large amounts of foreign direct invest- ment has meant that in the early 1990s, close to half of all aggregate external financing of development economies comes from private sources and goes to private destinations" (Bruno 1993, foreword). These trends raise important issues concerning the factors that motivate capi- tal flows and their effect on the performance of developing countries. This ar- ticle identifies the main determinants of capital flows to developing countries. It sheds light on the relative importance of the improved economic performance of these countries-country-specific or "pull" factors-and of the stimulus pro- vided by the decline of U.S. interest rates and the slowdown of the U.S. economy in the early 1990s-global or "push" factors (Chuhan, Claessens, and Mamingi 1993; Fernandez-Arias 1996; and Agenor forthcoming). Identifying the relative importance of push and pull factors is important for designing effective policy and therefore worthy of investigation. This has re- cently been shown by Fernandez-Arias and Montiel (1996), who first summa- rize a number of arguments describing why large capital flows may, under vari- ous circumstances, adversely affect developing countries unless policies designed to neutralize such effects are adopted. They then point out that if the causes of capital flows are largely exogenous to the developing country, compensatory policies are appropriate. But if the causes are largely domestic, then direct policy design may be more appropriate and effective. 1. THE DETERMINANTS OF INTERNATIONAL CAPITAL FLOWS Net capital flows arise when saving and investing are unbalanced across coun- tries, resulting in a transfer of real resources through a trade or current account imbalance. Gross capital flows, in contrast, need not involve any transfer of real resources. In fact, they may be offsetting across countries. Nevertheless, they allow individuals and firms to adjust the composition of their financial portfo- lios and are therefore important for improving the liquidity and diversification of portfolios. Both net and gross capital flows respond to economic fundamentals, official policies, and financial market imperfections. International capital flows play an important role in increasing economic efficiency, assuming that international financial markets can correctly evaluate the portfolio preferences of savers, identify and fund investments that have the highest expected rate of return, appropri- ately price financial assets on the basis of their underlying risks and returns, and provide information to reduce uncertainty. Following the traditional literature in financial economics, assets are priced, in the absence of distortions, so that the riskiest assets offer the highest rates of return. Also, although unsystematic risk can be reduced to a minimum by appropriately diversifying portfolios, sys- Taylor and Sarno 453 tematic risk cannot be diversified and should be reflected in the price of assets- investors should be compensated for holding a portfolio of assets whose returns have a high covariance (see Taylor 1991 for a survey of this literature). These arguments imply that among the fundamental determinants of interna- tional capital flows are factors such as the investment opportunities available in the global economy, the covariances between the expected returns on various investment projects, and the preferences of individuals for present and future consumption, as well as their attitudes toward risk. There is a major problem, however, in measuring empirically the effect of those factors on capital flows: international capital markets may react to a shock in one country through a change in capital flows, through a change in the prices of the country's financial claims, or through a mix of the two. Moreover, as the international financial system becomes more integrated and portfolios more diversified, asset prices are more likely to change than are net capital flows to restore market equilibrium. Thus most econometric models try to express financial linkages across countries in terms of interest rate parity conditions. That is, they specify the asset price linkages that are the outcome of arbitrage between financial markets rather than the capital flows that are part of the arbitrage process (Goldstein, Mathieson, and Lane 1991). In addition to these economic fundamentals, government policies and capital mar]ket imperfections also determine international capital movements. It is, how- ever, extremely difficult to assess the impact of these policies and distortions because they generally overlap, creating both impediments and stimuli to capital flows. The processes of deregulation, globalization, and innovation have increased both the efficiency of and volatility in financial markets. Volatility adds another source of risk, not only making the pricing of financial assets more difficult but also generating portfolio flows that are potentially more unstable (Corrigan 1989; Claessens, Dooley, and Warner 1995; Grabel 1995; and Clarke 1996). Also some evidence suggests that volatility is not correlated with any measure of fi- nancial integration and that it does not rise because of financial liberalization (see, for example, Tesar and Werner 1995 and Bekaert 1995). Push and Pull Factors The recent literature usually distinguishes between two sets of factors affect- ing capital movements (see, for example, Claessens, Dooley, and Warner 1995; Chuhan, Claessens, and Mamingi 1993; Fernandez-Arias 1996; Ferndndez-Arias and Montiel 1996; and Agenor forthcoming). The first are country-specific- pull-factors reflecting domestic opportunity and risk. As developing countries' creditworthiness is restored, capital (bond and equity) flows are likely to be- come an increasingly prominent source of external finance. For example, equity- related capital flows could be very large and come in the form of either foreign direct investment (FDI) or portfolio investment in equities. FDI may be attracted by the opportunity to use local raw materials or employ a local labor force. 454 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Although portfolio equity flows to developing countries have increased sharply in recent years, they are expected to be extremely sensitive to a country's open- ness, particularly to rules concerning the repatriation of capital and income (Williamson 1993). The right to repatriate dividends and capital may be the most important factor in attracting significant foreign equity flows (Goldstein, Mathieson, and Lane 1991). The International Finance Corporation differenti- ates between countries that allow foreign investors to repatriate capital and in- come freely and without restriction from countries that are "relatively open," which apply some restrictions on the repatriation of capital and income, and countries that are "relatively closed," which apply very strict restrictions. Rates of return-obviously a crucial determinant of capital flows-are often very high in the financial markets of developing countries relative to many ma- jor markets in industrial countries, reflecting the high risk generated by their typically high volatility. In particular, rates of return rose significantly in devel- oping countries in the late 1980s relative to those available in the major indus- trial economies (Calvo, Leiderman, and Reinhart 1993 and Chuhan, Claessens, and Mamingi 1993). Credit ratings and secondary-market prices of sovereign debt, reflecting the opportunities and risks of investing in the country, are likely to be important in determining capital flows as well (Bekaert 1995). Those indicators also rose in the late 1980s (Mathieson and Rojas-Suarez 1992 and Chuhan, Claessens, and Mamingi 1993). The second set of determinants of capital flows to developing countries are global-push-factors. For example, the sharp increase in U.S. capital flows, which represent a significant share of the portfolio flows received by emerging markets, may have been induced to some extent by the fast and marked fall of U.S. interest rates (short, medium, and long term) in the late 1980s. Moreover, the slowdown of the U.S. economy in the late 1980s mav also have attracted U.S. capital flows, especially because during that period macroeconomic poli- cies, labor market conditions, and exchange rate policies in many developing countries were becoming noticeably more stable (Calvo, Leiderman, and Reinhart 1993 and 1996). One would expect that as the governments of developing coun- tries make macroeconomic and institutional reforms, international investors will gain confidence and be more willing to direct capital flows toward the new markets (Papaioannu and Duke 1993). Chuhan, Claessens, and Mamingi (1993), using panel data for 1988-92, find that portfolio flows to a sample of Latin American and Asian countries are about equally sensitive to push and pull factors. They also find that equity flows, relative to bond flows, are more responsive to global factors; bond flows, however, are more responsive to a country's credit rating and to the secondary-market price of debt. Using a model with partial irreversibility of investment, Daveri (1995) derives a negative relationship between foreign investment and costs of entry anid exit from financial markets. His theoretical framework is consistent with the results of Chuhan, Claessens, and Mamingi (1993). Taylor and Sarno 455 A Simple Theoretical Framework Fernandez-Arias and Montiel (1996) have developed a useful analytical frame- work that incorporates the effect of domestic and global factors on capital flows.2 They separate potential domestic causes into those operating at the project level and those operating at the country level. Assuming that capital flows may be transactions in n different types of assets, indexed by s (s = 1, . . . , n), the domes- tic return on an asset of type s is decomposed into two components: a project- level expected return (GC) and an adjustment factor dependent on the creditwor- thiness of the country (C,). The project-level return is assumed to be a function of a vector of net flows (F) going to projects of all types, while the creditworthi- ness factor is assumed to be a function of a vector of the end-of-period stocks of liabilities of all types, S: S = S + F, where S-l denotes initial stocks of liabilities. Given that external creditors will diversify their portfolios, the opportunity cost of assets of type s, V, ,is a function of S. Fernandez-Arias and Montiel (1996) establish an arbitrage condition-from which F may be solved for-of the form: (1) G,(g,F)C5(c,S_1 + F) = V,(v,S + F) where g, c, and v represent shift factors associated with the domestic economic environment, domestic creditworthiness (pull factors), and the financial condi- tions of the creditor country (push factors). G5, Cs, and V, are assumed to be increasing functions of g, c, and v. The equilibrium or "desired" value of the vector of net flows F, F* say, determined implicitly by equation 1, may be ex- pressed as: (2) F* = F*(g,c,v,S.1) where F* is increasing in g and c but decreasing in v and S-,. Holding Sl con- stant, totally differentiating equation 2, and approximating total derivatives by first differences yield: (3) AP = F*Ag + F-Ac + F3Av where subscripts denote partial derivatives. Equation 3 describes the pattern of changes in desired capital flows, determined by changes in the pull factors g and c and the push factors v and by the initial value of S. Increases in g and c and decreases in v may induce prolonged growth in capital flows to developing coun- tries. This simple model is clearly consistent with both the push and the pull view of the surge in capital inflows, although the relative importance of the two factors will depend on the relative magnitude of the partial derivatives of F* as well as on the relative magnitude of changes in the factors themselves. Equation 3 states that differences in short-run and long-run capital move- ments might arise in accordance with the types of changes in g, c, and v: perma- nent changes in g, c, and v may cause long-run, permanent changes in the pattern 2. An alternative way to model capital flows is to use the international capital asset pricing model of Bohn and Tesar (1996). 456 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 of net flows, whereas transitory changes in these factors may generate transi- tory, short-term changes in net flows, which may be reversed over time. For example, the gradual, permanent removal of capital controls and liberalization of restrictions on FDI may reduce the adjustment costs that foreign investors face in diversifying their portfolios and thus give rise to a gradual stock adjustment (flow) over time. This gradual adjustment also implies a complex dynamic pat- tern of net flows moving toward their long-run equilibrium value and is consis- tent with an estimation methodology that distinguishes the short- from the long- run determinants of capital flows. Dynamic adjustment can be formally introduced into the Fernandez-Arias- Montiel framework by assuming a simple cost-of-adjustment model. In this model factors such as market imperfections, informational asymmetries (Stiglitz and Weiss 1981), and entry and exit costs to emerging financial markets (Daveri 1995) are captured in the assumption that creditors face costs in adjusting their portfolios that are increasing in the size of the adjustment. The desired vector of capital flows is given by equation 2. Assume that agents want to minimize the difference between desired and ac- tual flows, subject to adjustment costs. A simple way of modeling this is to assume a simple quadratic loss function for investors: (4) = (F - F-')'Ml (F - F-) + (F - Fl)M2 (F - F 1) where M1 and M2 are positive definite weighting matrices. From the first-order conditions for minimizing , we can derive a simple equation for changes in F: (5) AF = (Ml + M2)-l Mp(F* -F1) which, rearranging and using equation 3, can be equivalently expressed in the error-correction form: (6) AE = Ao(F* - F)_1 + A1Ag + A2Ac + A3Av where Ao = (M1 + M2)-M1 and Ai = (M1 + M2)-1M1Fi, (i = 1, 2, 3). According to equation 6 changes in current capital flows are determined partly by the difference between desired and actual capital flows in the previ- ous period and partly by changes in the factors determining the desired level of capital flows. Again, changes in push and pull factors can be decomposed into permanent and transitory components, with only the permanent ones affecting the long-run level of F. Transitory movements, which are reversed over time, will generate transitory movements in F, which are also reversed over time. For example, a temporary reduction in U.S. interest rates, which might be interpreted as a downward movement in v, will, other things being equal, generate a rise in capital flows to the developing country equal to A3Av (which is positive because F" is decreasing in v and Av is negative). If this change persists over time, then the long-run level of F will be raised because of the permanent effect on F* operating through equation 2. If, how- ever, the change in v is ultimately reversed, then although AF will be affected Taylor and Sarno 457 for several periods, the net long-run effect on both desired and actual capital flows will be zero. Although the simple theoretical framework developed in this section should not be taken too literally, it does suggest, together with a reading of the litera- ture presented, that shifts in capital flows may be determined by both push and pull factors and by both permanent and transitory factors. But the issue as to which of these factors is relatively more important is difficult to determine theo- retically. It therefore remains largely an empirical matter. II. DATA The data set used here is identical to that employed by Chuhan, Claessens, and Mamingi (1993). We use monthly data on U.S. portfolio flows, defined as gross and net purchases of foreign long-term securities for a group of nine Latin American countries-Argentina, Brazil, Chile, Colombia, Ecuador, Jamaica, Mexico, Uruguay, and Venezuela-and nine Asian countries-China, India, In- donesia, the Republic of Korea, Malaysia, Pakistan, the Philippines, Taiwan (China), and Thailand. For a full statistical description of the data set employed in this study, see Chuhan, Claessens, and Mamingi (1993). The data on capital flows are taken from the International Capital Reports of the U.S. Treasury Department and, according to the computations of Chuhan, Claessens, and Mamingi (1993), cover a substantial portion of U.S. portfolio flows to develop- ing countries. Note, however, that the U.S. share in total portfolio flows is far larger for the Latin American countries than for the Asian countries. Also, fol- lowing Chuhan, Claessens, and Mamingi (1993), we use net equity flows (ef and gross bond flows (bf) to developing countries, which cover a substantial share of their portfolio inflows. Even if we are concerned with modeling net capital flows in principle, using gross measures for bond flows is preferable in order to abstract from the effect of sterilizations and other types of reserve op- erations by central banks.3 There are two sets of explanatory variables: country-specific factors and glo- bal factors. For country-specific factors we use the country credit rating (cr) and the black market exchange rate premium (bm), which are available for both groups of countries considered.4 The credit rating variable is constructed on the 3. These data are collected by the U.S. treasury from financial intermediaries in the United States through the International Capital Form S reports. Hence, the data do not include direct dealings of U.S. investors with foreign intermediaries, because these transactions bypass the system. Note also that the data on bonds cover transactions of foreign securities in the United States from and to developing countries; transactions in bonds not issued by the developing country concerned nor by U.S. parties are expected to be insignificant (see Chuhan, Claessens, and Mamingi 1993: 6-7). 4. Another possible country-specific variable is secondary-market debt prices, which are a continuous random variable likely to be an integrated process of order one. Chuhan, Claessens, and Mamingi (1993: 10-11) use secondary-market prices only for the countries for which Salomon Brothers provide data. Because secondary-market prices are not available for all the countries considered in this article, we do not use them. 458 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 basis of the Institutional Investor's semiannual country credit rating, while the black market exchange rate premiums are calculated from data taken from the International Monetary Fund's World Currency Yearbook and the official exchange rates of the International Monetary Fund's International Financial Statistics (IFS). Finally, we also take from the IFS data base a short-term and a long-term U.S. nominal interest rate-the treasury bill rate (i) and the government bond yield (r)- and the level of real U.S. industrial production (y), which we consider to be global factors that potentially determine U.S. capital flows to developing countries. The sample period examined runs from January 1988 to September 1992, corresponding exactly to that considered by Chuhan, Claessens, and Mamingi (1993).5 An immediate avenue for research is to extend this sample period. Some of the global factors considered here, as well as in Chuhan, Claessens, and Mamingi (1993), have changed over the past five years. In particular, U.S. inter- est rates have started to rise after a prolonged period of decline, and the U.S. economy is recovering after the slowdown in the early 1990s. Although these changes may have caused structural breaks, generating additional estimation difficulties, it would be interesting to investigate whether and how they have affected portfolio flows to developing countries. III. ESTIMATION TECHNIQUES In this section we describe the estimation methods used in the analysis. We provide, first, a formal treatment of these methods and, then, a briefer, more intuitive discussion. A summary is given at the end. A Formal Statement Consider a panel of N countries, indexed by i (i = 1, . . ., N), with portfolio flows at time t denoted f,t, assumed to be an integrated process of order one, 1(1). Also, define a vector of country-specific factors as xi, and a vector of global factors as wi, and assume that both vectors contain at least one 1(1) variable and no higher-order integrated process. We can analyze the long-run behavior of portfolio flows by investigating cointegrating relationships of the kind: (7) (it= f xit+ y'Wit + ei-, i = i, . . ., N where ft may be either equity flows, say ef,, or bond flows, say bf1,. If cointegration is established in equation 7, that is, the error term eit is ap- proximately stationary, then the I(1) variables in xi, and wi, may be thought of as 5. Chuhan, Claessens, and Mamingi (1993: 13-15) provide a detailed description of the data employed here and also include illustrative graphs of all the data series. In addition, they present a panel correlations table, finding that most correlations are of the expected sign: negative between U.S. interest rates and flows, negative between black market exchange rate premiums and flows, negative between U.S. industrial production and flows, and positive between credit ratings and flows. These preliminary results are encouraging in that they support the underlying rationale of empirical models employed here. Taylor and Sarno 459 capturing the long-run or permanent components in f, , while ei, captures the short-run or temporary movements. Given that the dependent variable is I(1), there must be at least one I(1) variable among the explanatory variables; if all of the explanatory variables are I(0), then equation 7 will be misspecified (Pagan and Wickens 1989: 1002; Banerjee and others 1993; and Baffes 1997). In fact, after we execute preliminary unit root tests on the series in question in order to identify their order of integration, we immediately test to see if the residuals of the cointegrating regressions described by equation 7 are nonstationary using the two-step procedure of Engle and Granger (1987). In order to ensure that equation 7 is a cointegrating relationship, we also employ the relatively more powerful technique described by Johansen (1988) in a vector autoregression (VAR) context (see Kremers, Ericsson, and Dolado 1992). The Johansen estimation method is based on an error-correction representation of the pth order vector autoregression model [VAR(p)] with Gaussian errors of the form: (8) Aqi, = i + ]FiAqit-l + Fi2Aqit-2 + ... + rip-lAq,t-p,l + nqit-p + Bizit + uit where qi, is an m x 1 vector of 1(1) variables, Xi is an m x 1 vector of constants, zi, is an s x 1 vector of I(0) variables, the 7ijs and H are m x m matrices of parameters, Bi is an m x s matrix equation, and ui, is normally distributed with mean zero. The Johansen maximum likelihood procedure is based on the esti- mation of equation 8 subject to the hypothesis that the matrix 11 has reduced rank r < m, which may be written formally as H(r): H = ax', where oc and 6 are m x r matrices. 8'qit - p represents the cointegrating vectors. In the Johansen cointegration framework we can also test the hypothesis that the estimated coef- ficients on the country-specific or global factors are zero. These tests may in fact shed light on the relative importance of the two sets of factors, indicating whether push or pull factors are the major determinants of longer-run capital flows. Because ei, may alternatively be interpreted as the deviation from long-run equilibrium (ei, = ft- P'xi,- y'Wit), it may be used in an analysis of the short-run dynamics of capital flows through estimation of the error-correction model: At,, = vi - Pi (f - f3X - y W)i,t-i + Oitf,t-i p p (9) + XMAxxi'j + X6 5Awzt-j + WiP i = 1 N... j=O j=O where i is a country index, NYi is a constant term, j = 0, . .. , p denotes the number of lags, and wt,t is approximately white noise. Equation 9 is a panel data gener- alization of the error-correction representation of cointegrated variables estab- lished by Engle and Granger (1987) and follows directly from the Granger rep- resentation theorem (Granger 1987 and Engle and Granger 1987). The equation may be interpreted as a statistical approximation to the theoretical error- correction form in equation 6, on the assumption that desired capital flows F' 460 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 are achieved in the long run and are well approximated by the cointegrating vector. The adoption of a seemingly unrelated error-correction system, employed in other dynamic modeling contexts in international finance, allows us to compare the dynamics and long-run determination of bond and equity flows, and differ- entiates the present study from earlier studies on how portfolio flows are deter- mined. Equation 9 provides the full dynamic interaction of the determinants of capital flows. In particular, the analysis of the parameters pi may shed some light on the degree of persistence in the flow series considered (Claessens, Dooley, and Warner 1995; Dooley, Fernandez-Arias, and Kletzer 1996; and Sarno and Taylor 1997), conditional on the dynamics of the underlying fundamental determinants. An Intuitive Interpretation If over the sample period we find that capital flows do not tend to settle at any particular level-that is, they are nonstationary-then at least some of their determinants must also be nonstationary. Thus if we believe that flows to coun- try i, fti, are affected by a vector of pull factors xi, and a vector of push factors wi,, then we would expect to find a relationship of the form written in equation 7. Thus rapid changes in flows are determined by rapid changes in some of the push or some of the pull factors, or both. It is possible, however, that some of the push and pull factors are relatively stable over the sample period but still enter into equation 7. Given that ei, is expected to be highly stable over time, equation 7 can be interpreted as a long-run relationship because, on average, we would expect to find f t = f'xit + 7'wi,. In the econometric literature the relationship in equation 7 is termed a cointegrating relationship, and there are well-developed econometric techniques for testing for such relationships, estimating the parameters, and de- termining if some of the parameters are zero. These methods allow us to deter- mine which of the long-run determinants of capital flows are important. If we assume that actual and desired capital flows coincide in the long run, then we can think of the cointegrating relationship as determining the desired level of capital flows: fJtt = P'x, + y'wi,. Under this interpretation equation 7 is the empirical analog of the theoretical equation 2. Hence the error term e,, can be thought of as the difference between desired and actual flows, ei, = ft - ft/I This suggests that, having estimated the long-run parameters D and y, we can then estimate an error-correction equation that is the empirical counterpart of the theoretical error-correction model (equation 6), in which changes in flows are a function of changes in the variables determining the desired level of flows- that is, changes in xi, and wit-as well as of the error-correction term itself, ei,. In fact, econometric theory shows that if a cointegrating relationship exists, then such an error-correction model must also exist. Estimating the dynamic error- correction form then allows us to determine which factors are important in de- termining short-run movements in capital flows. Taylor and Sarno 461 We estimate these dynamic error-correction models jointly for our sample Asian and Latin American countries because the exogenous disturbance terms in these equations are likely to covary across countries. Also, exploiting the panel nature of the data by accounting for this covariation in our estimation method allovvs us to obtain more efficient-broadly speaking, more precise-estimates of the parameters because we are using more information than is used in single- equation estimation. Systems of equations of this kind are termed seemingly unrelated regressions. Note also that, in principle, country-specific factors could be influenced by global factors. For example, global interest rates may affect the pattern of secondary-market debt prices and credit ratings (see, for example, Dooley, Fernandez-Arias, and Kletzer 1996 and Fernandez-Arias 1996). To the extent that country-specific factors affect portfolio flows only insofar as they are col- linear with global factors, country-specific factors would appear to be insignifi- cant when both sets of explanatory variables are included. IV. LONG-RuN. FUNDAMENTAL DETERMINANTS OF PORTFOLIO FLOWS As a preliminary step to testing for cointegration in equation 7, we execute augmented Dickey-Fuller (ADF) unit root test statistics on the series used. The results (available from the authors on request) show that all of the series appear to be realizations from integrated processes of order one. The null hypothesis of nonstationarity is never rejected for both portfolio flows (equities and bonds) and global factors, while it is rejected once-Chile-for the credit rating and 5 times out of 18-Argentina, Malaysia, the Philippines, Taiwan (China), and Uruguay-for the black market exchange rate premium. In general, the ADF test statistics are relatively close to the rejection region for credit ratings and black market premiums, perhaps indicating, as one would expect, less evidence of nonstationarity in these series. In fact, credit ratings are discrete random vari- ables that cannot assume a very large number of outcomes. Also, black market premiums tend to widen before crises but may become very thin over time as a result, for example, of international capital market integration. Overall, how- ever, the unit root tests suggest that a permanent component is statistically sig- nificant in credit ratings and generally statistically significant in black market premiums at the 5 percent nominal significance level. Thus these variables can potentially contribute to the long-run determination of portfolio flows. Given the finding that the series considered appear to be realizations of I(1) pro- cesses, we are justified in testing for cointegration in equation 7. We test for nonstationarity of the residuals in equation 7, considering first equity flows and then bond flows as dependent variables (table 1). In all but 6 of the 36 regressions examined we were able to reject the null hypothesis of no cointegration at the 5 percent nominal level of significance, suggesting that a combination of domestic and global factors share a common trend with equity and bond flows. In order to gauge which of the two sets of variables, xi, and wit, are more important in determining 462 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Table 1. Results of Tests for Nonstationarity of the Residuals of the Cointegration Equation for 18 Countries, 1988-92 Dependent variable Augmented Dickey-Fuller test statistics and country ADFi ADF2 ADF3 Asian equity flows China -8.4265 -8.2182 -8.4364 (-5.0141) (-3.8963) (-4.2988) India -6.2328 -6.4854 -4.4082 (-5.0257) (-3.9053) (-4.3064) Indonesia -7.1897 -6.8962 -5.6371 (-5.0141) (-3.8963) (-4.2988) Korea, Rep. of -6.7943 -4.4005 -6.4559 (-5.0141) (-3.8962) (-4.2988) Malaysia -7.0122 -6.5245 -6.0072 (-5.0141) (-3.8962) (-4.2988) Pakistan -1.7677' -2.3677a -0.911 ga (-5.0382) (-3.9085) (-4.3146) Philippines -6.1994 -5.2261 -6.1861 (-5.0141) (-3.8963) (-4.2988) Taiwan (China) -8.5079 -2.7677 -2.6200 (-5.0141) (-3.9085) (-4.3146) Thailand -5.9094 -4.1257 -5.8286 (-5.0141) (-3.8963) (-4.2988) Asian bond flows China -6.4286 -6.1033 -6.0844 (-5.0141) (-3.8963) (-4.2988) India -8.2003 -7.0188 -7.3904 (-5.0141) (-3.8963) (-4.2988) Indonesia -5.4419 -5.3780 -5.1129 (-5.0141) (-3.8963) (-4.2988) Korea, Rep. of -5.7214 -7.4051 -8.0958 (-5.0198) (-3.8963) (-4.2988) Malaysia -7.6344 -7.4688 -7.6126 (-5.0141) (-3.8963) (-4.2988) Pakistan -7.8331 -7.7325 -7.7011 (-5.0141) (-3.8963) (-4.2988) Philippines -4.2752a -4.1574 -3 .9047a (-5.0198) (-3.8992) (-4.3025) Taiwan (China) -5.4774 -2.8194a -4.1504a (-5.0141) (-3.8992) (-4.2988) Thailand -7.8268 -7.7779 -8.4181 (-5.0141) (-3.8992) (-4.2988) Latin American equity flows Argentina -7.3331 -3.5075 -5.1805 (-5.0318) (-3.8992) (-4.3146) Brazil -5.5815 -5.3729 -6.2050 (--5.0198) (-3.8963) (-4.2988) Chile -6.0774 -4.7966 -4.9482 (-5.0141) (-3.8963) (-4.2988) Colombia -7.5776 -6.1239 -6.9158 (-5.0141) (-3.8963) (-4.2988) Ecuador -10.6141 -10.7085 -10.1457 (-5.0141) (-3.8963) (--4.2988) Taylor and Sarno 463 Table 1. (continued) Dependent variable Augmented Dickey-Fuller test statistics and country ADF1 ADF2 ADF3 Jamaica -7.9104 -7.7771 -7.5235 (-5.0141) (-3.8963) (-4.2988) Mexico -3.8560 -5.0159 -3.6606a (-5.0318)a (-3.9053) (-4.3146) Uruguay -8.1749 -7.5328 -8.1380 (-5.0141) (-3.9053) (-4.2988) Venezuela -2.7387a -2.5070a -2.2916a (-5.0382) (-3.9085) (-4.3146) Latin American bond flows Argentina -7.6526 -5.5257 -7.3757 (-5.0141) (-3.8963) (-4.3146) Brazil -8.7683 -7.4884 -8.7535 (-5.0198) (-3.8963) (-4.2988) Chile -8.8537 -8.3542 -8.6428 (-5.0141) (-3.8963) (-4.2988) Colombia -7.8772 -6.9796 -8.0583 (-5.0141) (-3.8963) (-4.2988) Ecuador -8.2530 -7.4536 -8.1222 (-5.0141) (-3.8963) (-4.2988) Jamaica -6.2529 _3.7397a -4.3974 (-5.0318) (-3.9053) (-4.2988) Mexico -2.3368a -3.8186a -2.3687a (-5.0382) (-3.9022) (-4.3146) Uruguay -5.8362 -0.8025a -0.8631a (-5.0198) (-3.9022) (-4.2988) Venezuela -3.6914a -5.1953 -3.2371a (-5.0257) (-3.8963) (-4.3146) Note: ADF1, ADF2, and ADF3 are augmented Dickey-Fuller test statistics computed on the residuals from the regression (equation 7) of equity/bond flows on both country-specific and global factors, only country-specific factors, and only global factors, respectively. Critical values are reported in parentheses. The number of lags included is such that the error term is approximately white noise. a. The null hypothesis of no cointegration cannot be rejected at standard nominal levels of significance. Source: Authors' calculations. long-run movements of capital flows, we employ the ADF test statistic on a regres- sion including only country-specific factors, and then only global factors. In fact, a test of the null hypothesis Ho: ,B = 0 or Ho: y = 0 cannot be executed in such a framework because of the strong bias of the estimated standard errors. The results show that cointegration is often established in both regressions (table 1). Neverthe- less, in three of the six cases for which the ADF fails to reject the null hypothesis, the results suggest that if we cannot reject the null of no cointegration in a regression of capital flows on xi, (wi,), we cannot reject no cointegration in a regression of capital flows on wit (xit)-;perhaps supporting the view that it is appropriate to consider both xir and wit as explanatory variables. It must be noted, however, that results are subject to the problem of small- sample bias in cointegrating relationships originally highlighted by Banerjee and 464 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 others (1986) and by Stock (1987). Therefore, in order to strengthen the finding of cointegration in equation 7, we employ a relatively more powerful cointegration technique, the Johansen (1988) procedure, in a VAR system including capital flows and country-specific as well as global factors (Kremers, Ericsson, and Dolado 1992). The results of the Johansen procedure are available from the authors on request. In this framework we can also check, using likelihood ratio tests, for zero restrictions on the coefficients of the push or pull factors. The results enable us, based on both of the test statistics suggested by Johansen, to establish cointegration in all 36 systems analyzed at conventional nominal levels of significance. Thus the failure of the Engle and Granger procedure to detect cointegration in several cases may simply be a result of the low power of the test. In particular, the evidence of cointegration is impressive because it is based on the estimation of individual regressions with a relatively small number of obser- vations. This obviates the need to employ more powerful panel cointegration tests (Quah 1994 and Im, Pesaran, and Shin 1995). The Johansen procedure also suggests that there are multiple cointegrating vectors among the variables examined, as may be the case whenever cointegration is investigated among more than two variables. Further, the zero restrictions on the coefficients of country-specific or global factors are always strongly rejected, pointing to the long-run importance of both push and pull factors. This result also suggests that pull factors are important determinants of capital flows in their own right, not only because they may be correlated with push factors. V. A DYNAMIC ANALYSIS OF PORTFOLIO FLOWS The existence of at least one cointegrating relationship between a set of vari- ables implies that an error-correction model exists, because, as established by the Granger representation theorem, for any set of 1(1) variables error- correction and cointegration are equivalent representations. Therefore, the re- siduals from the equilibrium regressions (equation 7) can be used to estimate, by generalized least squares, the error-correction models (equation 9) as seemingly unrelated regressions. The speed-of-adjustment coefficients p, have important implications for the dynamics of the model. In fact, for any given value of the deviations from long-run equilibrium (f, - 3xi, - y'wi,), a large value of pi is associated with a large value of the change in capital flows, Aft. If pi is zero, the change in capital flows does not respond at all to the deviation from long-run equilibrium in period t - 1. If pi is zero and all 4i, = 0 (8ii = 0), then Axi, (Awit) does not Granger-cause Aft. We know, however, that one or both of these coef- ficients should be significantly different from zero because the variables are found to be cointegrated (Engle and Granger 1987). We examine the panel data generalization of the error-correction representa- tion of cointegrated variables (equation 9) using the cointegrating residuals re- trieved from the cointegrating regressions (equation 7). Note that the estimation of seemingly unrelated error-correction mechanisms exploits the nature of the Taylor and Sarno 465 panel data and is an efficient technique for analyzing the effect of a common set of global factors across a group of countries in a dynamic framework. In fact, the disturbances in the individual regressions are expected to be very highly correlated because they include some factors that are common to all of the coun- tries (Dwivedi and Srivastava 1978). We adopt the conventional general-to-specific procedure to estimate a parsi- monious error-correction model, as suggested by Davidson and others (1978) and Hendry (1983).6 The resulting models appeared to be quite adequate in terms of high coefficients of determination and residuals that are approximately white noise (table 2). Although space considerations preclude us from reporting each of the estimated equations in detail, the equation estimated for bond flows to Colombia is reasonably representative: Abf, = -0.976e,l +0.197Acr, +0.555Acr,l -1.811Ai,1 (10) (0.097) (0.083) (0.083) (0.687) R2 = 0.79, Q(24) = 15.138 [0.917] where R2 denotes the coefficient of determination and Q(24) denotes the Ljung- Box statistic for residual autocorrelation computed for 24 autocorrelations. The figures in parentheses are estimated standard errors, and the figure in square brackets is the marginal significance level (a constant term was also included). The variable bf denotes the bond flow, cr the country credit rating, i the U.S. short-term interest rate, and e the cointegrating residual or error-correction variable. The equation shows a strongly significant and relatively large error- correction coefficient-indicating rapid adjustment-and demonstrates the im- portance of both push and pull factors. The full set of results with the point estimates is available from the authors. The underlying dynamics of the error-correction models for Asian capital flows are, especially for equities, slightly more complicated than those for Latin Ameri- can capital flows, in the sense that more variables are found to be statistically significant on the right-hand side of the error-correction models. Global and country-specific factors seem to have roughly the same statistical significance in determining the change in equity flows for both Asian and Latin American coun- tries. The change in bond flows, however, appears to be relatively more strongly determined by global factors than by domestic factors, while equity flows are relatively more responsive to changes in country-specific factors. Even if significant lags in country-specific factors are included in the error- correction model, overall bond flows appear to react predominantly to changes in global factors. From table 2 we can see that a change in U.S. interest rates explains the dynamics of bond flows better than the other global factor consid- ered in this study, the growth of U.S. industrial production. It must be pointed out, however, that the two interest rate series considered are not expected to have the same cyclical properties. The finding that the dynamics of bond flows, 6. See Cuthbertson, Hall, and Taylor (1992) for a textbook exposition of general-to-specific modeling. 466 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Table 2. Number of Significant Push and Pull Factors in the Error-Correction Models for Asia and Latin America, 1988-92 Region and Credit Black market U.S. interest U.S. real industrial capital flow type rating premium rates production Total Asia Equities 11 14 15 9 49 Bonds 5 11 16 8 40 Total 16 25 31 17 89 Latin America Equities 10 7 17 4 38 Bonds 7 6 18 8 39 Total 17 13 35 12 77 Source: Authors' calculations. relative to those of equity flows, are determined more by global factors than by country-specific factors contrasts with the finding of Chuhan, Claessens, and Mamingi (1993), who conclude that equity flows are more sensitive than bond flows to global factors, while bond flows are more sensitive to country-specific factors. However, they are primarily interested in identifying the long-term de- terminants of the large capital flows to developing countries rather than in fully modeling the dynamics of capital flows. Hence their conclusions are drawn for illustrative purposes, using a simpler approach based on the computation of standardized coefficients and elasticities. The relative importance of the global factors suggested by our results is consis- tent with those of Calvo, Leiderman, and Reinhart (1993 and 1996), who first suggested the importance of U.S. interest rates and of the slowdown in U.S. industrial production over 1988-92 in explaining portfolio flows to emerging markets. They also argued that a reversal of the global conditions could induce a fast outflow of capital from developing countries. Our results go further in pointing out that interest rates are likely to be the most important determinant of the dynamics of portfolio flows (especially bonds) to Asian and Latin Ameri- can countries. Finally, note that interest rates are a more important short-term determinant of portfolio flows in Latin American countries than in Asian countries: interest rates are statistically significant 31 times (34 percent of the total number of significant terms) in the Asian error-correction models, while they are signifi- cant 35 times (45 percent of the total number of significant terms) in the Latin American error-correction models. Put another way, Latin American inflows are as sensitive as Asian inflows to interest rates, but they are less sensitive to all the other factors. VI. CONCLUSIONS In this article we examined the determinants of U.S. capital flows directed to nine Latin American and nine Asian countries over 1988-92, extending previ- Taylor and Sarno 467 ous work by Chuhan, Claessens, and Mamingi (1993). In particular, we investi- gated whether bond and equity inflows were induced by push or pull factors, differentiating between short- and long-run determinants. We considered in our set of country-specific factors the domestic credit rating and the black market exchange rate premium as well as a set of global factors including two U.S. interest rates and the level of U.S. real industrial production (Calvo, Leiderman, and Reinhart 1993 and 1996). We examined the long-run determinants of portfolio flows by employing two complementary cointegration techniques. The results provide unequivocal evidence that long-run equity and bond flows are about equally sensitive to global and country-specific factors and, therefore, that both sets of variables help to explain U.S. portfolio flows to the developing countries considered. Moreover, we also estimated seemingly unrelated error-correction models for equity and bond flows in order to shed light on the underlying short-run dynam- ics of U.S. portfolio flows. The models for portfolio flows to Asian developing countries, especially for equity flows, are more complicated in terms of their short-run dynamics than the error-correction models for portfolio flows to Latin American countries. A count of the number of significant push and pull factors appearing in the error-correction forms, classified by type of flow and geographic area, revealed that both seem to be equally important in determining short-run equity flows for Asian and Latin American countries. When bond flows are considered, how- ever, global factors seem to be much more important than domestic factors in explaining the short-run dynamics of flows. In particular, changes in U.S. inter- est rates are found to be the single most important determinant of short-run movements in bond flows to developing countries. REFERENCES The word "processed" describes informally reproduced works that may not be com- monly available through library systems. Agenor, P. R. Forthcoming. "The Surge in Capital Flows: Analysis of 'Pull' and 'Push' Factors." International Journal of Finance and Economics. Baffes, John. 1997. "Explaining Stationary Variables with Nonstationary Regressors." Applied Economics Letters 4(1):69-75. Banerjee, Anindya, J. J. Dolado, J. W. Galbraith, and D. F. Hendry. 1993. Co-Integra- tion, Error Correction, and the Econometric Analysis of Nonstationary Data. New York: Oxford University Press. Banerjee, Anindya, J. J. Dolado, D. F. Hendry, and G. W. Smith. 1986. "Exploring Equilibrium Relationships in Econometrics through State Models: Some Monte Carlo Evidence." Oxford Bulletin of Economics and Statistics 48(3):253-77. Bekaert, Geert. 1995. "Market Integration and Investment Barriers in Emerging Equity Markets." The World Bank Economic Review 9(1):75-107. Bohn, Henning, and L. L. Tesar. 1996. "U.S. Equity Investment in Foreign Markets: Portfolio Rebalancing or Return Chasing?" American Economic Review 86(1):77-81. 468 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Bruno, Michael. 1993. "Foreword." In Stijn Claessens and Sudashan Gooptu, eds., Port- folio Investment in Developing Countries. Discussion Paper 228. Washington, D.C.: World Bank. Calvo, Guillermo A., Leonardo Leiderman, and Carmen M. Reinhart. 1993. "Capital Inflows and the Real Exchange Rate Appreciation in Latin America: The Role of External Factors." IMF StaffPapers 40(1):108-51. . 1996. "Inflows of Capital to Developing Countries in the 1990s." Journal of Economic Perspectives 10(2):123-39. Chuhan, Peter, Stijn Claessens, and Nandu Mamingi. 1993. "Equity and Bond Flows to Asia and Latin America." Working Paper 1160. Policy Research Department, World Bank, Washington, D.C. Processed. Claessens, Stijn, Michael P. Dooley, and Andrew Warner. 1995. "Portfolio Capital Flows: Hot or Cold?" The World Bank Economic Review 9(1):153-74. Clarke, Roland. 1996. "Equilibrium Interest Rates and Financial Liberalisation in De- veloping Countries." Journal of Development Studies 32(3):391-413. Corrigan, E. G. 1989. "Reflections on the 1980s." In Federal Reserve Bank of New York Seventy-Fifth Annual Report {for the Year Ended December 31]. New York: Federal Reserve Bank of New York. Cuthbertson, Keith, S. G. Hall, and M. P. Taylor. 1992. Applied Econometric Tech- niques. Exeter, U.K.: Harvester Wheatsheaf. Daveri, Francesco. 1995. "Costs of Entry and Exit from Financial Markets and Capital Flows to Developing Countries." World Development 23(8):1375-85. Davidson, James, D. F. Hendry, F. Srba, and Stephen Yeo. 1978. "Econometric Model- ling of the Aggregate Time Series Relationships between Consumers' Expenditure and Income in the United Kingdom." EconomicJournal 88(December):661-92. Dooley, Michael, Eduardo Fernandez-Arias, and Kenneth Kletzer. 1996. "Is the Debt Crisis History? Recent Private Capital Inflows to Developing Countries." The World Bank Economic Review 10(1):27-50. Dwivedi, T. H., and K. A. Srivastava. 1978. "Optimality of Least Squares in the Seem- ingly Unrelated Regressions Model." Journal of Econometrics 7(2):391-95. Engle, R. F., and C. W. J. Granger. 1987. "Co-integration and Error Correction Repre- sentation, Estimation, and Testing." Econometrica 55(2):251-76. Feldstein, Martin, and Charles Horioka. 1980. "Domestic Saving and International Capital Flows." Economic Journal 90(June):314-29. Fernandez-Arias, Eduardo. 1996. "The New Wave of Capital Inflows: Push or Pull?" Journal of Development Economics 48:389-418. Fernandez-Arias, Eduardo, and P. J. Montiel. 1996. "The Surge in Capital Inflows to Developing Countries: An Analytical Overview." The World Bank Economic Review 10(1):51-77. Goldstein, Morris, D. J. Mathieson, and Timothy Lane. 1991. "Determinants and Svs- temic Consequences of International Capital Flows." In iMNF Research Department, Determinants and Systemic Consequences of International Capital Flows. Occasional Paper 77. Washington, D.C.: International Monetary Fund. Grabel, Ilene. 1995. "Assessing the Impact of Financial Liberalisation on Stock AMarket Volatility in Selected Developing Countries." Journal of Development Studies 31(6):903-17. Taylor and Sarno 469 Granger, C. W. J. 1987. "Cointegration and Error Correction: Representation, Estima- tion, and Testing." Econometrica 55(2):251-76. Hendry, D. F. 1983. "Econometric Modelling: The Consumption Function in Retro- spect." Scottish Journal of Political Economy 30(3):193-220. Im, K. S., M. H. Pesaran, and Y. H. Shin. 1995. "Testing for Unit Roots in Dynamic Heterogenous Panels." Working Paper 9526. Department of Applied Economics, University of Cambridge, Cambridge, U.K. Processed. IMF (International Monetary Fund). 1994a. International Capital Markets: Develop- ments, Prospects, and Policy Issues. Washington, D.C. -. 1994b. World Economic Outlook, World Economic and Financial Surveys. Washington, D.C. 1995. Private Market Financing for Developing Countries. Washington, D.C. -. Various years. International Financial Statistics. Washington, D.C. Various years. World Currency Yearbook. Washington, D.C. Johansen, Soren. 1988. "Statistical Analysis of Cointegration Vectors." Journal of Eco- nomic Dynamics and Control 12(June-September):231-54. Kremers, J. J. M., N. R. Ericsson, and J. J. Dolado. 1992. "The Power of Cointegration Tests." Oxford Bulletin of Economics and Statistics 54(August):325-48. Mathieson, D. J., and Liliana Rojas-Suarez. 1992. "Liberalization of the Capital Ac- count." IMF Working Paper WP192/46. International Monetary Fund, Washington, D.C. Processed. Montiel, Peter. 1993. "Capital Mobility in Developing Countries." PRE Paper 1103. Policy Research Department, World Bank, Washington, D.C. Processed. Obstfeld, Maurice. 1995. "International Capital Mobility in the 1990s." In P. B. Kenen, ed., Understanding Interdependence: The Macroeconomics of the Open Economy. Princeton, N.J.: Princeton University Press. Pagan, A. R., and M. R. Wickens. 1989. "A Survey of Some Recent Econometric Meth- ods." Economic Journal 99(December):962-1025. Papaioannu, M. G., and L. K. Duke. 1993. "The Internationalization of Emerging Eq- uity Markets." Finance and Development (September):36-39. Quah, Danny. 1994. "Exploiting Cross-Section Variation for Unit Root Inference in Dynamic Data." Economics Letters 44(1):9-19. Sarno, Lucio, and Mark P. Taylor. 1997. "Hot Money, Accounting Labels, and the Persistence of Capital Flows to Developing Countries: An Empirical Investigation." Department of Economics, Oxford University. Processed. Stiglitz, J. E., and Andrew Weiss. 1981. "Credit Rationing in Markets with Imperfect Information." American Economic Review 71(3):393-410. Stock, J. H. 1987. "Asymptotic Properties of Least Squares Estimators of Co-integrating Vectors." Econometrica 55(5):1035-56. Taylor, M. P. 1991. Money and Financial Markets. Cambridge, Mass.: Blackwell. Tesar, Linda, and Ingrid M. Werner. 1995. "U.S. Equity Investment in Emerging Stock Markets." The World Bank Economic Review 9(1):109-29. United Nations Economic Commission for Latin America and the Caribbean. 1994. Economic Panorama of Latin America, 1994: Argentina, Brazil, Colombia, Chile, Ecuador, Mexico, Peru, Uruguay, Venezuela. Santiago. 470 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 U.S. Treasury Department. Various years. International Capital Reports. Washington, D.C. Williamson, John. 1993. "Issues Posed by Portfolio Investment in Developing Coun- tries." In Stijn Claessens and Sudashan Gooptu, eds., Portfolio Investment in Devel- oping Countries. Discussion Paper 228. Washington, D.C.: World Bank. World Bank. 1995. Trends in Developing Economies 1995. Washington, D.C. . 1997. The Road to Financial Integration: Private Capital Flows to Developing Countries. Washington, D.C. THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3: 471-87 Determinants of the Export Structure of Countries in Central and Eastern Europe Bernard Hoekman and Simeon Djankov The growth in exports from Central and Eastern Europe to Western markets suggests that entrepreneurs have responded to changed incentives by restructuring their pro- duction to capture new markets. The absence of change in the structure of exports, however, suggests that these restructuring efforts have not been significant. This ar- ticle analyzes the magnitude of the change in export structure across the Central and Eastern European countries in 1990-95, focusing in particular on trade with the Eu- ropean Union. It finds that imports of intermediate inputs and machinery are an im- portant determinant of the changes in export structure. Sourcing of inputs from abroad is a major factor underlying the expansion of exports to the European Union. Out- ward processing (subcontracting) arrangements and foreign direct investment have a smaller impact. Except for Poland, inflows of foreign direct investment are statistically insignificant or negatively associated with measures of revealed comparative advan- tage. This suggests that foreign investors have chosen sectors in which the Central and Eastern European countries were not relatively specialized under central planning. Following the demise of central planning, Central and Eastern European coun- tries experienced severe economic shocks. The Council of Mutual Economic Assistance (CMEA), which had governed the international trade relations of mem- ber countries, collapsed in 1989. Since then, analysts have done a significant amount of work investigating developments in the trade of the countries in Cen- tral and Eastern Europe. This literature presents several stylized facts (see, for example, Drabek and Smith 1995; Kaminski, Wang, and Winters 1996; and World Bank 1996). First, exports from countries in Central and Eastern Europe to Western Europe have grown very rapidly. Second, the composition of these exports has changed relatively little (Halpern 1995). Third, an increasing share of the trade between many Central and Eastern European countries and the European Union is intra-industry, that is, it involves exchanges of similar prod- Bernard Hoekman is with the Development Research Group, and Simeon Djankov is with the Private Sector and Finance Department of the Europe and Central Asia Region, both at the World Bank. The authors are indebted to Ying Lin for his excellent research assistance in compiling and extracting the data used in this article. They are also grateful to Alan Winters and two anonymous referees for insightful and constructive comments and to Alan Deardorff, Zdenek Drabek, Simon Evenett, Bart Kaminski, Will Martin, Gerhard Pohl, Maurice Schiff, Matthew Tharakan, and Zhen Kun Wang for helpful suggestions and discussions. C) 1997 The International Bank for Reconstruction and Development / THE WORLD BANK 471 472 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 ucts (Neven 1995). Fourth, inflows of foreign direct investment (FDI) are heavily concentrated in specific sectors and countries. The Visegrad countries (Hun- gary, Poland, and the Czech and Slovak Republics) have attracted more than 80 percent of all FDI into the region (European Bank for Reconstruction and Devel- opment 1996). Much of this FDI has gone into services, food processing, and the motor vehicle industries. The growth in exports to Western markets suggests that entrepreneurs are responding to the changed incentives by engaging in restructuring to capture new markets. The increase in intra-industry trade and the industry concentra- tion of FDI also may indicate that firms are adjusting by specializing in narrower production lines. The absence of change in export structure, however, suggests a lack of significant restructuring. Much of the increase in exports may simply be due to the redirection of goods to Western countries. We cannot determine ex ante the extent to which Central and Eastern European countries must realign their historical production structures according to their comparative advantage. This depends on the divergence between initial conditions (the pattern of spe- cialization under the CMEA) and the allocation of resources in conformance with market-determined prices. Some of the countries may need to improve the effi- ciency of existing industries; others may need to improve the allocation of re- sources across industries. Baldwin (1994) surveys the literature in an attempt to determine the extent to which the volume and direction of trade will change once the transition to a market economy has been completed. Such analysis is often based on historical data on trade patterns before World War II or on gravity models of trade. The change in the composition of exports that occurred during 1990-95 pro- vides information on how much the pattern of specialization under central plan- ning diverged from what would have emerged under a market system. The pe- riod is long enough for the countries to have undertaken sufficient reform. Absence of change in the structure of exports in the last six years suggests that the initial structure of production may have been appropriate; a lot of change would sug- gest the opposite. However, even if we observe little change, this does not neces- sarily imply that little restructuring occurred. Even relatively efficient firms would be likely to have improved productivity following the opening and liberalization of the economy. The greater the difference between best-practice production techniques in the global economy and best practices in the context of a largely closed planned economy, the greater the scope for enhancing efficiency. In prac- tice such improvement might be reflected in a rise in imports of technology, components, and machinery. In this article we analyze the magnitude of change in the export structure in Central and Eastern European countries. We investigate the relative importance of processing (subcontracting) trade, imports of inputs, and FDI as determinants of the countries' export performance in European Union markets. We undertake a statistical analysis of the extent to which these variables are associated with the countries' export composition during 1990-95. Hoekman and Djankov 473 Section I briefly summarizes the stylized facts of the trade performance of countries in Central and Eastern Europe since 1990, including the extent of change in export composition. Section II discusses the possible relationships among subcontracting or processing trade, imports of inputs more generally, and FDI and examines the change in export structure across these countries. Sec- tion III reports the estimation results. Section IV concludes. I. REORIENTATION AND CHANGE IN EXPORTS It is very difficult to obtain accurate data on trade flows prior to 1990 be- cause of highly distorted cross-exchange rates and the prevalence of barter trade. For this reason the analysis in this article starts in 1990. Although this may miss part of the adjustment process, the data difficulties make it much more problem- atical to interpret any observed changes starting from an earlier date. In prin- ciple, data after 1990 should not be subject to the valuation and measurement problems that affect data under central planning. Until the end of the 1980s, most of the countries in Central and Eastern Europe traded extensively with one another and with the Soviet Union. As of 1990 these countries shipped 30-45 percent of total exports to former mem- bers of the CMEA (table 1). After 1990 the share of total exports going to Western Europe increased significantly for all the countries in Central and Eastern Europe. As of 1996 exports to Western Europe accounted for 50-80 percent of total exports (table 1). A similar phenomenon occurred on the import side. For most of the countries, some 70 to 80 percent of total im- ports originated in Western Europe (International Monetary Fund, Direc- tion of Trade Statistics). How much of the shift in the direction of trade is associated with a change in the composition of exports? Revealed comparative advantage (RCA) is an easily interpretable measure of the change in the structure of exports. The RCA is the share of a commodity in a country's total exports relative to the average share for the world.' We measure the change in the composition of exports by calculating the simple correlation between RCAS for each country in 1990 and 1995, the most recent year for which disaggregated data are available. A higher correlation indicates that less change has occurred. Be- cause the Czech and Slovak Republics became separate countries in 1993, we calculate the RCA correlations for exports between 1990 and 1992 and between 1993 and 1995. We calculate the RCAs at both the two- and four- 1. This measure is due to Balassa and is defined as: (Xi, /X, )/( x=-1 X; /E1 =I )where x,, are exports of commodity i by country j, x; are country j's total exports, and N is the number of countries. In what follows, RCAS in exports to the European Union are defined as the European Union's reported imports of a commodity from a Central and Eastern European country divided by total reported imports relative to total imports by the European Union of that commodity divided by total European Union imports. Table 1. Share of Exports to Former Centrally Planned Economies and Western Europe, 1990-96 Total exports, Export growth Sbare of exports to former 1996 (billions (average annual centrally planned economies Share of exports to of U.S. percent) (percent)a Western Europe (percent) Country dollars) 1990-96 1993-96 1990 1993 1996 1990 1993 1996 Albania 0.3 3.9 30.6 31 3 6 49 70 82 Bulgaria 4.8 13.4 22.3 30 16 19 40 46 51 Czech Republic b 18.8 - 21.2 - 31 38 - 61 60 Czechoslovakiab - - - 44 - - 40 Hungary 15.7 7.1 14.3 34 14 21 50 56 71 Poland 22.8 5.9 16.2 33 11 21 51 70 69 Romania 8.5 6.2 16.7 35 11 10 36 40 54 Slovak Republic 9.3 - 34.0 - 57 56 - 42 47 - Not available. a. Includes Bulgaria, Czech Republic, Slovak Republic, I lungary, Poland, Romaniia, and the former Yugoslavia and Soviet UJnion. b. Excludes intra-Czech-Slovak trade. Source: Data from the International Monetary Fund, Direction of 't'rade Statistics. Hoekman and Djankov 475 digit levels of disaggregation for trade with the world and for trade with the European Union.2 Between 1990 and 1992 little change occurred in the composition of exports at the two-digit level. The correlation coefficients are 0.80 or higher for Bul- garia, Hungary, Poland, and Romania (table 2). For exports to the world (total exports), Albania has the lowest correlation (0.62), and Poland has the highest (0.88). For most of the countries exports to the European Union changed even less; correlation coefficients are above 0.9 for all of the countries except Czecho- slovakia (0.73) and Albania (0.54). Between 1993 and 1995 greater changes occurred for most countries, and greater differences emerged across countries. The Czech Republic, Hungary, and Poland continued to experience very little change in the structure of exports at the two-digit level (correlation coefficients are higher than 0.9). Conversely, Albania and Bulgaria experienced a substan- tial change in their export structure (coefficients of 0.44 and 0.69, respectively). Comparator countries such as Indonesia, Mexico, Morocco, Spain, and Turkey have correlation coefficients in the 0.7-0.8 range over an analogous period (U.N. COMTRADE data base). The Slovak Republic also changed its export mix more than average. Most of the Central and Eastern European countries experienced more change in their exports toward the European Union than toward the rest of world during 1993-95. Based on the absence of change at the two-digit level of disaggregation, most of the Central and Eastern European countries exported the same products in the early 1990s as in the late 1980s (see, for example, Halpern 1995 and Drabek and Smith 1995). However, enterprises may change their export mix within two-digit categories. For example, a paint factory may continue to produce and export paint, but switch from selling oil-based paints to a wholesaler in large drums to selling water-soluble paints that are packaged for retail sale. Such changes will not show up at the two-digit ievel. When we analyze the correlation at the four-digit level (1,238 commodities) for exports to the European Union, we obtain similar conclusions as in the two-digit analysis, with one significant exception. Although Hungary and Poland continued to show little change in export composition, the Czech Republic experienced a substantial amount of change within the two-digit product categories (table 2). In principle, we must use more disaggregated data to track the extent to which enterprises managed to differentiate their output from that produced under central planning. Unfortu- nately, the available data are not very reliable. (For many commodities beyond the four-digit level no trade is reported for either 1990 or 1995. It is often not clear whether this reflects reality-there really was no trade-or simply a re- porting or measurement problem.) 2. 'There are 99 commodity groups at the two-digit level of the Combined Nomenclature, the classification of trade statistics used by the European Union. Excluding so-called special codes, there are 63 two-digit categories in the Standard International Trade Classification, which we use to report statistics on world trade. There are therefore 99 and 63 sectors, respectively, in the RCA correlations for trade with the European Union and with the world. There are 1,238 four-digit items in the Combined Nomenclature. Table 2. Change in the Composition of Exports, 1990-95 Measure Destination Czech Slovak and period of exports Albania Bulgaria Republic Czechoslovakia Hungary Poland Romania Republic Correlation coefficients of RCAs at the two-digit levela 1990-92 World 0.62 0.83 - 0.78 0.85 0.88 0.82 1990-92 European Union 0.54 0.92 - 0.73 0.95 0.90 0.94 - 1993-95 World 0.44 0.69 0.91 - 0.90 0.96 0.84 0.77 1993-95 European Union 0.36 0.61 0.95 - 0.96 0.91 0.81 0.71 Correlation coefficients of RCAs at the four-digit level" 1990-92 European Union 0.77 0.28 - 0.66 0.88 0.83 0.35 - 1993-95 European Union 0.69 0.41 0.58 - 0.89 0.80 0.32 0.68 Herfindahl index of concentration of exportsc 1990 European Union 0.073 0.044 - - 0.042 0.043 0.139 - 1993 European Union 0.114 0.043 0.042 - 0.051 0.051 0.122 0.054 1995 European Union 0.123 0.062 0.064 - 0.074 0.051 0.096 0.072 - Not available. Note: RCA is revealed comparative advantage. It is the sharc of a commodity in a country's total exports relative to the average share for the world or, in trade with the European tJnion, relative to the share for the European Union. a. 63 and 99 categories for the world and Europcan Union, respectively. b. 1,238 catcgories. c. The index is calculated for commodities at the two-digit level. The flcrfindahl index is defined as X, (Si)2 where s is the share of sector i (i = I ... 99) in total exports to the European tJnion. Source: Authors' calculations based on the European Union c;OMtXt and U.N. COMTRADF, data bases. Hoekman and Djankov 477 Although a high correlation between RCAs across time suggests that little change occurred in the broad structure of trade, there may have been significant changes in the relative importance of individual items. Even at the two-digit level, sub- stantial changes in the value of RCAs occurred for most of the countries. In part this reflects changes in the volume of exports, with some commodities registering large increases in exports and others registering decreases. One effect of these changes was an increase in the concentration of exports to the European Union. The Herfindahl measure of concentration suggests that the export composition of most of the countries became more specialized during 1990-95 (table 2). The exception, Romania, is largely explained by the collapse of exports of oil products. Here we are interested in the determinants of changes in export structure. Because the European Union is by far the largest trade partner of-and direct investor in-the Central and Eastern European countries and because detailed data on trade are available, the analysis that follows focuses on the export per- formance of these countries with the European Union. The high correlation be- tween changes in export structure to the European Union and changes in export composition to the world (which in large part reflects the large share of total exports going to the European Union) suggests that little will be lost by limiting the analysis to trade with the European Union. II. POSSIBLE FACTORS UNDERLYING CHANGES IN EXPORT COMPOSITION An increase in intra-industry trade accompanied the changes in both the direction and composition of exports.3 Although still below the levels regis- tered for advanced industrial countries in the region, the level of intra- industry trade grew rapidly for the Central and Eastern European countries. Slovenia, the Czech Republic, and Hungary currently have indexes of intra- industry trade that exceed those of Portugal and Greece. Two major dimen- sions may underlie such exchange. First, the textbook explanation maintains that intra-industry exchange results when firms specialize in similar but dif- ferentiated products, driven by the need to realize economies of scale or scope. Second, in the early stages of the transition to a market economy, firms are likely to have incentives to source inputs from the rest of the world, thereby obtaining access to know-how and technologies. Such exchanges may be arm's-length, or they may occur in the context of joint ventures or other contractual relationships. Such vertical intra-industry trade may well be more important than exchanges involving similar but differentiated products, especially in the early stages of 3. Analysts often use the Grubel-Lloyd index of intra-industry trade. It is defined as: 1 - ( I X - M I)/[(X + M)] where X~ and M, are a country's exports to and imports from, respectively, a trading partner of commodity i. See Helpman (1987) for a theoretical analysis of such trade and Faini and Portes (1995) and Drabek and Smith (1995) for a discussion of intra-industry trade developments between the Central and Eastern European countries and the European Union. 478 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 transition. As in the literature on industrial organization and FDI, we use the term vertical to refer to the geographic fragmenting of the production process by stage of production (see, for example, Markusen 1995). The trade literature uses the term vertical intra-industry trade to describe intra-industry bilateral exchanges of very similar goods where the unit values of exports and imports exceed a particular threshold. The term horizontal intra-industry trade describes bilateral trade flows in the same commodity classification where unit values are below this threshold (Greenaway, Hine, and Milner 1995). For Central and Eastern European enterprises seeking to export to Western Europe, European Union firms (be they potential partners or customers) are an obvious source of information on quality standards, packaging requirements, tastes (design of goods), and production techniques as well as suppliers of machinery and high- quality intermediate goods. Intra-industry trade is a mechanism for the transfer of technology. The share of intermediate inputs and capital goods in total imports from the European Union is in the 55 to 65 percent range for Bulgaria, Hungary, Poland, and Romania. For the Czech and Slovak Republics the figure is 80 percent (fig- ure 1). In Hungary and Poland the growth in imports of capital goods is particu- larly strong, rising from 12 and 16 percent of total imports from the European Union in 1990 to 20 and 30 percent, respectively, in 1995. Capital goods ac- count for some 30 percent of total Czech and Slovak imports from the European Union. Albania stands out for its very low share of capital goods in total imports. Enterprises may use inward FDI or nonequity-based relationships with Euro- pean Union suppliers or customers (including contracts and joint ventures) to obtain intermediate inputs and capital goods. Alternatively, this may be the re- Figure 1. Composition of Importsfrom the European Union, 1990 and 1995 Percent 100 90 0 80s R Rep 70 111 60- 50 40 30 20 10 1990 1995 1990 1995 1990 1995 1995 1990 1995 1990 1995 1990 1995 Albania Bulgaria Czecho- Czech Slovak Hungar-y Poland Romania slovakia Repub. Repub. * Consumer goods a Intermediate goods Ol Capital goods Source: Authors' calculations. Hoekman and Djankov 479 Table 3. Foreign Direct Investment in Central and Eastern Europe, 1995 Billions Industry share Ratio to Country of dollars (percent) GDP Albania 0.2 64 3.5 Bulgaria 0.3 51 0.8 Czech Republic 5.5 56 5.6 Czechoslovakia 1.1 49 - Hungary 11.5 44 10.2 Poland 2.4 38 0.7 Romania 0.9 46 1.0 Slovak Republic 0.6 41 1.1 -Not available. Note: Foreign direct investment includes inflows of goods and services. GDP is gross domestic product. Data are cumulative flows giving a 1995 stock figure. Source: European Union COMEXT; International Monetary Fund, Direction of Trade Statistics; and European Bank for Reconstruction and Development (1996). sult of the independent decisions of managers to upgrade production processes. FDI may in part be driven by relative cost considerations that make it attractive to produce in a host country but will also have other motivations related to ownership and knowledge. Without such advantages it is usually assumed that a foreign investor does not have a competitive advantage over local incumbent firms. Trade barriers that raise the cost of direct exports, the perception that consumers prefer locally produced goods, or incentive policies of the host gov- ernment might also drive FDI. Empirical work on the motivations of foreign in- vestors in the Central and Eastern European countries suggests that production costs are not a significant factor (Meyer 1995). FDI flows into Eastern Europe after 1989 were heavily concentrated in specific countries and sectors. Hungary alone accounted for more than 50 percent of the total stock of inward FDI in the region in 1995 (table 3). Much of the FDI went into services (distribution, tourism), but between two-fifths and three-fifths of the total went into industry. Joint ventures are an alternative to FDI. The key difference is that joint ven- tures imply no controlling equity stake by the foreign partner. From the perspec- tive of an enterprise in Central and Eastern Europe, joint ventures with Western firms may result in the provision of intermediate inputs, know-how, equipment, or a variety of services ranging from design, to production and management techniques, to distribution and marketing. FDI implies a longer-term commit- ment to the domestic firm and may give rise to greater transfer of (proprietary) technology as well as capital. Although numerous joint ventures have been es- tablished between firms in Central and Eastern Europe and the West, compre- hensive data on this are not available. Imports of intermediate inputs and capital goods may also occur through outward processing trade (OPT). Independent trade involves no cooperative rela- tionship and less communication between domestic and foreign firms. Instead, 480 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 domestic enterprises find sources for inputs and equipment and seek to obtain export contracts independently. OPT, or subcontracting arrangements, involve greater communication with foreign firms and may be an important avenue for the transfer of technology, especially soft technology. OPT is a customs regime under which enterprises based in the European Union may ship components abroad for processing and reimport the processed commodities free of duty or quantitative restrictions, if applicable. Corado (1995) and Naujoks and Schmidt (1994) describe and dis- CUSS OPT and European Union rules on OPT. Helleiner (1973) and Keesing and Lall (1992) discuss the potential benefits associated with subcontracting. For example, the foreign partner will require that production meet specifications (both with respect to design and maximum defect rates), and this may require the implementation of quality control systems. The partner will also require timely delivery of production and will therefore need to be convinced that management can deal with possible disruptions in the supply of inputs from local suppliers. Interviews with Central and Eastern European enterprises that have subcontracting agreements with European Union firms reveal that for- eign buyers frequently provide information on possible sources of equipment and inputs and make strong recommendations to source from a limited num- ber of possible suppliers. The European Union collects statistics on OPT, which consists mostly of sub- contracting (European Union, COMEXT data base). In 1994, goods entering the European Union under outward processing customs regimes accounted for about 17 percent of total Central and Eastern European exports to the European Union, up from 10 percent in 1989. Similarly, imports from the European Union for processing grew from 7 to 12 percent of the total. Processing activities generated almost one-third of Romania's exports to the European Union in 1995, up from 13 percent in 1989. OPT for the other Central and Eastern European countries accounted for 10-20 percent of total exports. Most of the processing occurs in leather and footwear (20-30 percent of total exports) and in textiles and cloth- ing (60-80 percent), both of which are sensitive to pressures for protection by European Union industries. Other industries with significant OPT include electri- cal machinery (10-16 percent), precision instruments (16-18 percent), and fur- niture (15-20 percent). Most of the furniture processing is concentrated in the Visegrad countries. OPT also occurs for agricultural goods. In 1993 almost 5 percent of Poland's agricultural exports to the European Union entered under the OPT regime. This was due in part to the processing of raw crustaceans and other fish in Poland (Naujoks and Schmidt 1994). The European Union tariff provides the incentive to use the OPT customs re- gime. For many industrial products, tariffs are zero or negligible for Central and Eastern European exporters as a result of the Europe Agreements. Thus OPT measures only part of the more general phenomenon of two-way trade in goods that make up an industry's production chain. Unfortunately, data are not avail- able with which to estimate the importance of such intra-industry trade. In prin- Hoekman and Djankov 481 ciple, we could use input-output tables if data are reported on the origin of imported intermediate inputs that are used in production. Unfortunately, the input-output tables of the Central and Eastern European countries do not do so, making it impossible to relate imports of intermediates from and exports of goods to the European Union. However, the available input-output tables do report information on total imports of intermediates used by industries. This allows us to calculate the ratio of imported to total intermediate consumption. We regard this ratio as a measure of integration into the world economy and as a reflection of the upgrading process. Producers use the imported inputs in pro- duction for both the home and foreign markets. In section III we analyze the relative importance of three variables-OPT, im- ports of inputs more generally, and FDI-as determinants of the observed export structure. FDI, joint ventures, and subcontracting may all be associated to a greater or lesser extent with an increase in imports of inputs. Although we control for FDI, this is not possible for joint ventures and subcontracting, although to some extent subcontracting is captured by OPT. Thus the import variable is not limited to arm's-length, independent exchanges. Whether we should expect the three variables to have a positive or negative association with changes in export struc- ture is unclear. For example, firms may use OPT to keep existing facilities in operation by engaging in subcontracting activities, or they might use OPT to diversify production and penetrate new export markets. In the first case OPT would not be associated with a change in export composition; in the latter it would. The same ambiguity pertains to integration through more general sourc- ing of inputs from foreign providers. Even established sectors that entrepreneurs consider viable most likely will require substantial efforts to upgrade quality and improve productivity. In principle, the same ambiguity arises with respect to FDI, because foreign investors should invest in those sectors where positive returns are expected, which may or may not be traditional activities. If foreign investors are risk averse, they may target sectors where export capacity and associated human capital already exist. Even if the existing capital stock has little value, the availability of a quali- fied and experienced labor force may provide an incentive for investors to prefer such sectors over others. Governments may have an interest in encouraging FDI in existing facilities so as to maintain employment. In a related vein, countries may attract FDI by offering policy-based incentives such as tax concessions or guaranteeing some margin of protection against import competition. For ex- ample, a number of Central and Eastern European governments have granted such incentives to investors in the automotive industry. These incentives may be in sectors in which the country does not have a comparative advantage. III. ESTIMATION RESULTS In this section we investigate the association between changes in export struc- ture and OPT, imports of inputs more generally, and FDI. The dependent variable 482 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 is the level of RCAs in trade with the European Union for 23 industries in 1990- 95. We use RCAs rather than the shares of commodities in total exports for each country because RCAs control for changes in both demand by the European Union for that commodity and supply by the rest of the world. Because none of the Central and Eastern European countries is a major supplier of a particular good in the European Union, their export performance will not affect the denomina- tor of the RCA. If we use export shares as the dependent variable, we obtain results that are very similar to those obtained with RCAs. We define industries at the two-digit level of the International Standard Industrial Classification (IsIc). The 23 sectors are those distinguished in chap- ter D of the isic, rev 3, that is, food; tobacco; textiles; clothing; leather; wood products; paper; publishing; coke and petroleum products; chemicals; rub- ber and plastics; other nonmetallic products; basic metals; metal products; other machinery; office and computing machinery; electrical machinery; ra- dio, television, and communications equipment; medical and precision in- struments; motor vehicles; other transport equipment; other manufacturing; and recycling. The European Union reports detailed statistics on the value of imports that enter under the OPT regime. We categorize these data at the industry level using a mapping developed by Eurostat. (We use the concor- dance included in the software accompanying the European Union's trade data base, COMEXT.) We also categorize detailed annual data on the value of FDI by sector collected by the national authorities of the Central and Eastern European countries for the 23 isic sectors. The FDI data reflect actual invest- ment, not planned, approved, or committed flows.4 We obtain data on the share of imports in total intermediate consumption (IMP) for each of the 23 sectors from national input-output tables, as reported in the statistical year- books of the countries (Slovenia Statistical Office 1996; Bulgarian Statistical Office 1996; Kozponti Statisztikai Hivatal 1996; Glowny Urzad Statystyczny 1996; Cesky Spisovatel 1996; Comisia Nationala Pentru Statistica 1996; Slovenska Spisovatel 1996). Unfortunately, only Bulgaria, the Czech Repub- lic, Hungary, Poland, and Romania report this information. The latest year for which Albania and the Slovak Republic report such data is 1993; they are therefore excluded in what follows. FDI and IMP are not specific to the European Union in the sense that FDI or imports of intermediates are restricted to being of European Union origin. In an earlier paper, Hoekman and Djankov (1996), we attempt to map trade data on imports of intermediates into exports by Central and Eastern European indus- tries to the European Union, which requires strong assumptions regarding the share of European Union inputs in total imports, as well as the European Union import content of exports to the European Union. The ratio of imported to total 4. Sources for FDI data are as follows: Albania, Bulgaria, Poland, and the Czech and Slovak Republics: national foreign investment agencies; Hungary: Ministry of Finance; anid Romania: Romanian Development Agency. Data are available upon request from the authors. Aggregate data on FDI are reported in European Bank for Reconstruction and Development (1996). Hoekman and Djankov 483 intermediate consumption has the advantage of being transparent and not re- quiring such assumptions. OPT is by definition a subset of the IMP measure (the simple correlation coeffi- cient between the two variables is 0.43). (Simple correlations between OPT and FDI and between IMP and FDI suggest that these variables are uncorrelated.) In order to reduce standard errors of the parameter coefficient estimates, we first regress the IMP variable on OPT. We then use the residual resulting from this procedure, IMP* (the part of IMP not explained by OPT) as the integration variable in the analysis of the change in export structure. (If we use IMP instead of IMP*, the fit of the estimating equation is very similar, but standard errors increase.) Our data set is a so-called panel, that is, it contains observations across indus- tries for a relatively short period of time. Using ordinary least squares on the pooled data is only appropriate if parameter values are common to all industries at all times. An F-test rejects the hypothesis of such a common intercept. We therefore follow the standard panel approach, using generalized least squares to estimate a random effects model. This approach assumes that industry-specific effects vary over time and across industries; it treats these effects as random variables in the sense that they are assumed to be drawn from a given distribution for each year. An alternative is to use a fixed effects model, where it is assumed that the industry-specific effects are fixed parameters over time. The choice between fixed and random effects models in the current situation is unclear. A Hausman specification test suggests that either a fixed or random effects model could be used. Fixed effects models are costly in terms of degrees of freedom. In the absence of more information, we consider it appropriate to treat industry-specific effects as random variables. Hsiao (1986) provides a detailed discussion of the econometric issues and tradeoffs. There are 23 sectoral observations for six years (1990-95) for Bulgaria, Hun- gary, Poland, and Romania and for three years (1993-95) for the Czech Repub- lic. In addition to OPT, FDI, and IMP*, we include annual dummies to control for macroeconomic and external shocks that are common to all the countries in the sample. For the sample as a whole, including annual dummies but not sector dummies, both IMP and OPT are statistically significant, while FDI is not. Includ- ing sector dummies improves the fit of the equation somewhat, but given that by definition the inclusion of an additional 23 variables will increase the R2, the small rise suggests that sector-specific forces are not that important (table 4). However, the significance level of the FDI variable drops substantially, suggest- ing that FDI is correlated with specific sectors. Of the 23 sector dummies, only five are statistically significant at the 0.99 level. Two are positive (basic metals and office and computing machinery); the other three are negative (electrical machinery; motor vehicles; and radio, television, and communications equip- ment). In the first two there is very little FDI, while the last three generally attract a substantial share of total manufacturing FDI. The magnitude and significance of the coefficients on OPT and IMP* are not affected by the inclusion of sector dummies, suggesting that they are not driven by sector-specific forces. 484 THE WORLD BANK ECONOOMIC REVIEW, VOL. 11, NO. 3 Table 4. The Determinants of Export Structure in Central and Eastern Europe, 1990-95 Model without Model with Variable sector dummies sector dummies Outward processing trade, OPT 1.21 1.40 (2.72) (2.91) Foreign direct investment, FDI -0.36 -0.18 (1.57) (0.84) Imports in intermediate consumption,a ImP 8.02 6.71 (11.1) (9.79) R2 0.21 0.31 Note: The dependent variable is revealed comparative advantage (RC(A) in trade with the European Union. RCA is the share of a commodity in a country's total exports relative to the average share of that commodity in total European Union imports. The sample includes annual data for 23 sectors for 1990- 95 for Bulgaria, Hungary, Poland, and Romania and for 1993-95 for the Czech Republic, giving 621 observations. Both models include annual dummies. t-statistics are in parentheses. a. The part of the share of imports in intermediate consumption that is not explained by outward processing trade. See text for details. Source: Authors' calculations. Regression results across individual countries reveal substantial differences in the relative importance of OPT, FDI, and IMP (table 5). The country regres- sions again include annual dummies to control for shocks that affect all sec- tors in the economy but do not include sector dummies because there are in- sufficient degrees of freedom. At the country level, OPT is significant only for the Czech Republic and Romania. FDI is statistically significant and negative in sign for Bulgaria and Hungary, insignificant in the Czech Republic and Roma- nia, and significant and positive in Poland. A significant negative coefficient Table 5. The Determinants of Export Structure in Five Countries in Central and Eastern Europe, 1990-95 Czech Variable Bulgaria Hungary Republic Poland Romania Outward processing trade, OPT -0.96 0.60 0.69 0.36 2.63 (0.85) (1.16) (1.92) (0.60) (3.l1l Foreign direct investment, FDI -0.10 -0.69 -0.24 0.77 0.69 (2.59) (6.50) (1.12) (2.26) :1.03) Imports in intermediate 7.36 4.94 6.98 4.78 36.95 consumption,,IMP' (7.20) (7.66) (4-56) (3.15) (11.10) Number of observations 138 138 69 138 138 R2 0.34 0.49 0.48 0.27 0.55 Note: The dependent variable is revealed comparative advantage RC:A) in trade with the European Union. RCA is the share of a commodity in a country's total exports relative to the average share of that commoditv in total European Unioni imports. The sample includes annual data for 23 sectors for 1 990- 95 for Bulgaria, Hungary, Poland, and Romania and for 1993-95 for the Czech Republic, giving 621 observations. The model includes annual dummies. t-sratistics are in parentheses. a. The part of the share of imports in intermediate consumption that is not explained by outward processing trade. See text for details. Source: Authors' calculations. Hoekman and Djankov 485 on the FDI variable indicates that investment is going into industries where host countries do not have a revealed comparative advantage. Only in Poland does FDI appear to be associated with traditional export activities. IMP*, or integration, is by far the most important explanatory variable for all four coun- tries. The coefficient estimate is particularly large for Romania. OPT, FDI, and IMP* explain some 50 percent of the variation in the dependent variable for Hungary and Romania. In a panel setting of the type used here, the explana- tory power is quite high. IV. CONCLUSIONS Although on average relatively little change occurred in the composition of Central and Eastern European exports between 1990 and 1995, there were sig- nificant differences across countries. Hungary apparently exported very much the same products, while others such as Bulgaria and Romania changed signifi- cantly the composition of their exports, especially to the European Union. In this article we analyze the impact of FDI, OPT, and imports of inputs on RCAS in five countries-Bulgaria, the Czech Republic, Hungary, Poland, and Roma- nia. We find that imports of inputs are highly correlated with the composition of exports. With only one exception (Poland), FDI is either negatively correlated with the host country's RCA in an industry or statistically insignificant. In Hun- gary in particular, foreign investors apparently took equity stakes in sectors that do not have a comparative advantage in European Union markets. And, with the exception of Romania, we find that outward processing activities-under which firms in Central and Eastern Europe process components received from European Union partner firms and reexport these back to the European Union- are not a significant factor. Our analysis suggests that in most countries imports of intermediate goods and machinery drove the changes in export structure. Local enterprises appar- ently exploited the opportunity to acquire foreign inputs and know-how in or- der to improve production quality, thereby expanding their export market share in the European Union. FDI did not play a large role in this upgrading process. In this respect our results are consistent with those of Tharakan and Kerstens (1995), who in a case study of intra-industry trade find a negative relationship between such trade and FDI. Indeed, FDI was concentrated in sectors where the Central and Eastern Euro- pean countries do not have a revealed comparative advantage (that is, they are not relatively specialized in terms of their export share in European Union mar- kets): this is the case for Bulgaria, the Czech Republic, and Hungary. Of the five countries for which data are available, Poland is the only one with a significant positive association between FDI and RCAs. The negative relationship between RCAs and FDI for many of the countries implies that FDI could be a force for change. Foreign investors must perceive the industries concerned to be viable in the medium term, and over time this FDI may lead to greater changes in the 486 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 countries' export composition. Thus, FDI complements efforts by domestic in- dustries to restructure and upgrade production facilities. REFERENCES The word "processed" describes informally reproduced works that may not be com- monly available through library systems. Baldwin, Richard. 1994. Towards an Integrated Europe. London: Centre for Economic Policy Research. Bulgarian Statistical Office. 1996. Statisticheski Godishnik 1996, Izdanie 86. Sofia. Cesky Spisovatel. 1996. Statisticka Rocenka Ceske Republiky 1995. Prague. Comisia Nationala Pentru Statistica. 1996. Anuarul Statistic al Romaniei 1995. Bucharest. Corado, Cristina. 1995. "The Textiles and Clothing Trade with Central and Eastern Europe." In Riccardo Faini and Richard Portes, eds., Eu Trade with Eastern Europe: Adjustment and Opportunities. London: Centre for Economic Policy Research. Drabek, Zdenek, and Alasdair Smith. 1995. "Trade Performance and Trade Policy in Central and Eastern Europe." CEPR Discussion Paper 1182. Centre for Economic Policy Research, London. Processed. European Bank for Reconstruction and Development. 1996. Transition Report. London. Faini, Riccardo, and Richard Portes, eds. 1995. EU Trade with Eastern Europe: Adjust- ment and Opportunities. London: Centre for Economic Policy Research. Glowny Urzad Statystyczny. 1996. Rocznik Statystyczny 1996, Rok LVI. Warsaw. Greenaway, David, Robert Hine, and Chris Milner. 1995. "Vertical and Horizontal Intra-Industry Trade: A Cross-Industry Analysis for the United Kingdom." Economic Journal 105(433):1505-18. Halpern, Laszlo. 1995. "Comparative Advantage and the Likely Trade Pattern of the CEECS." In Riccardo Faini and Richard Portes, eds., Eu Trade with Eastern Europe: Adjustment and Opportunities. London: Centre for Economic Policy Research. Helleiner, Gerald. 1973. "Manufactured Exports from Less Developed Countries and Multinational Firms." EconomicJournal 83(329):21-47. Helpman, Elhanan. 1987. "Imperfect Competition and International Trade: Evidence from Fourteen Industrial Countries." Journal of theJapanese and International Econo- mies 1(1):62-81. Hoekman, Bernard, and Simeon Djankov. 1996. "Intra-Industry Trade, Foreign Direct Investment, and the Reorientation of Eastern European Exports." Policy Research Working Paper 1652. Policy Research Department, World Bank, Washington, D.C. Processed. Hsiao, Cheng. 1986. Analysis of Panel Data. Cambridge, U.K.: Cambridge University Press. Kaminski, Bart, Zhen Kun Wang, and Alan Winters. 1996. "Export Performance in Transition Economies." Economic Policy 23(3):423-42. Keesing, Don, and Sanjay Lall. 1992. "Marketing Manufactured Exports from Devel- oping Countries: Learning Sequences and Public Support." In Gerald Helleiner, ed., Hoekman and Djankov 487 Trade Policy, Industrialization, and Development: New Perspectives. Oxford: Clarendon Press. Kozponti Statisztikai Hivatal. 1996. Magyar Statisztikai Evkonyv 1995. Budapest. Markusen, James. 1995. "The Boundaries of the Multinational Enterprise and the Theory of International Trade." Journal of Economic Perspectives 9(2):169-89. Meyer, Klaus. 1995. "Direct Foreign Investment in Eastern Europe: The Role of Labor Costs." Comparative Economic Studies 37(4):69-88. Naujoks, Petra, and Klaus-Dieter Schmidt. 1994. "Outward Processing in Central and East European Transition Countries." Kiel Working Paper 631. Institut fur Weltwirtschaft, Kiel. Processed. Neven, Damien. 1995. "Trade Liberalization with Eastern Nations: How Sensitive?" In Riccardo Faini and Richard Portes, eds., EU Trade with Eastern Europe: Adjustment and Opportunities. London: Centre for Economic Policy Research. Slovenian Statistical Office. 1996. Statisticni Letopis 1996. Letnik XXXV. Ljubljana. Slovenska Spisovatel. 1996. Statisticka Rocenka Slovenska Republiky 1995. Bratislava. Tharakan, P. K. M, and Birgit Kerstens. 1995. "Does North-South Intra-Industry Trade Really Exist? An Analysis of the Toy Industry." Weltwirtschaftliches Archiv 131(1):86- 105. World Bank. 1996. World Development Report 1996: From Plan to Market. New York: Oxford University Press. A NEW DEVELOPMENT DATA BASE The following article is the second in an occasional series introducing new data bases. The series intends to make new development data bases more widely avail- able and to contribute to discussion and further research on economic develop- ment issues. The data bases included in the series are selected for their potential usefulness for research and policy analysis on critical issues in developing and transition economies. Some are drawn from micro-level firm or household sur- veys; others contain country-level data. The authors describe the data contents, criteria for inclusion or exclusion of values, sources, strengths and weaknesses, and any plans for maintenance or updating. Each data base is available from the author, at the address provided in the article. THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3: 491-513 A New Data Base on State-Owned Enterprises Luke Haggarty and Mary M. Shirley Privatization of state-owned assets was a central economic reform of the 1980s and 1990s. Lack of data has made it impossible to judge the extent to which privatization has diminished the importance of state ownership in the economy or changed the per- formance of state-owned enterprises. This article introduces a new data base on state- owned enterprises, the first since the mid-1980s, which partly fills this information gap. The Bureaucrats in Business data base provides time-series data for up to 88 countries on the share of state enterprises in the overall economy, investment, employ- ment, and internal and external credit as well as their overall surplus or deficit and the size of transfers to and from national treasuries. The article presents the rationale for the data base, describes its seven measures of state enterprise size and performance, and explores possible uses of the information. Privatization of state assets is one of the defining characteristics of economic change in the 1980s and early 1990s. Suleiman and Waterbury (1990: 4) take as their leitmotiv "the belief that the breadth of the phenomenon [privatization] reveals that we are witnessing a fundamental shift in industrial and financial ownership and in the management of economies." Yet although there is evi- dence that the incidence of privatization has increased dramatically, the lack of data on such fundamentals as value added, investment, or employment in state- owned enterprises has made it impossible to judge how important privatization has been in changing the role of state ownership in the economy. To address some of these issues, the World Bank (1995) developed the Bu- reaucrats in Business (BIB) data base detailing the size, nature, and performance of state enterprises in industrial and developing market economies during 1978- 91.1 This data base is the most complete of its kind and is the only significant addition to the pool of state enterprise data since the early 1980s. Although it does not answer all of the questions posed above, it addresses many of them and produces some surprises. Section I reviews the rationale for the data base and the choice of indicators. Section II compares the existing data bases with the BIB data base and explores 1. The BIB data base is freely available on the Internet at http://www.worldbank.org/html/prdfp/bib/ bibdata.htm. Luke Haggarty and Mary M. Shirley are with the Development Research Group at the World Bank. Bharat Nauriyal is the analyst primarily responsible for development of the data base described in this article. The authors thank the Fiscal Affairs Department of the International Monetary Fund for help on the data base. They are also grateful to Ross Levine for helpful comments. C) 1997 The International Bank for Reconstruction and Development / THE WORLD BANK 491 492 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 its strengths and weaknesses. Section III discusses each of the seven indicators in more detail. Section IV illustrates potential uses of the data base. Section V concludes. I. THE RATIONALE FOR DEVELOPING THE DATA BASE An important motivation for developing the BIB data base was the need to assess the impact of privatization on state enterprises. The upsurge in privatization in the late 1980s and early 1990s was accompanied by an upsurge in rhetoric claiming that divestiture was minimizing the importance of state-owned enter- prises as an economic issue. Two existing data bases on privatization (Candoy- Sekse 1988 and World Bank 1997) show nearly seven times as many privatization transactions in the eight years from 1988 to 1995 as during the previous eight years (1980-87). Although comparable data on the value of transactions are not available for 1980-87, the shift of sales to larger firms in higher-value sectors (from manufacturing and services to infrastructure and banking) suggests that the value of privatized industries increased substantially as well. It thus seemed that divestiture was reducing the economic importance of state ownership. But without data on the stock of state enterprises, we could not evaluate that trend. To analyze the impact of privatization on government ownership, we col- lected data on three measures of economic size and importance: share in eco- nomic activity, investment, and employment. Despite some problems (described below), these indicators are useful for assessing the relative importance of state ownership by country or region. Surprisingly, the data suggest that the average share of state ownership in developing market economies did not change much over 1978-91 (figure 1). We also wanted the data base to provide information on the performance of state enterprises and their contribution to economic growth. The upsurge in privatization spurred a sizable literature debating the effects of government ownership and privatization on efficiency and growth (see, for example, theo- retical studies by Shapiro and Willig 1990; Sappington and Stiglitz 1987; Vickers and Yarrow 1988; and Yarrow 1986; and empirical studies by Galal and others 1994; Boardman and Vining 1989; Kikeri, Nellis, and Shirley 1992; Megginson, Nash, and van Randenborgh 1994; and Pollitt 1995). The theoretical debate is inconclusive. Hence, Yarrow holds that "it cannot be expected that one form of ownership will be superior to the other in all indus- tries and in all countries" (1986: 332). As for the empirical evidence, analysis has been hampered by lack of comprehensive information. Studies comparing public and private firms (reviewed in Yarrow 1986; Galal and others 1994; and World Bank 1995) generally favor private ownership in competitive markets. But these studies fail to adjust for differences not attributable to ownership (see Borcherding, Pommerehne, and Schneider 1982; Millward and Parker 1983; and Millward 1988). Studies of privatized firms (surveyed in Kikeri, Nellis, and Shirley 1992 and World Bank 1995) generally favor private ownership in com- Haggarty and Shirley 493 Figure 1. State-Owned Enterprises' Shares in Gross Domestic Product, 1978-91 Percent 16 . ~~~~~~~~~~~~~~15 low-income countries 14 12 .,40 developing countries 12 10 ,-- . 25 middle-income countries 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 Note: Values are unweighted averages. Source: World Bank (1995, statistical appendix). petitive and monopoly markets but are based on relatively small samples. Galal and others (1994) find that privatization improved welfare significantly over a counterfactual in 11 of 12 cases, 9 of which were monopoly firms. In a study of 41 firms in 15 countries Megginson, Nash, and van Randenborgh (1994) find that privatized firms in competitive sectors increased their returns on sales, assets, and equity; raised internal efficiency; improved their capital structure; and increased investment (although no counterfactual is constructed). Empirical analyses of the macroeconomic effects of ownership are limited to country case studies. II. A COMPARISON WITH EXISTING DATA BASES Until development of the BIB data base, Short (1984) and Nair and Filippides (1989) had developed the most extensive data sets (table 1). Short's data base has the largest number of indicators and includes up to 78 countries, but cover- age for many indicators is less.2 Also, Short's data fall between 1960 and 1980, making them unsuited for addressing questions about the effects of privatization, because the significant increase in divestiture dates from the late 1980s. Another problem is that many of the indicators are presented as three- and four-year averages rather than as yearly figures. These averages mask significant policy 2. The indicators are state enterprise shares in GDP, gross fixed capital formation, output and investment by industrial sector, overall balances and their component parts, state enterprise financing by source, state enterprise budgetary burden, and share of gross domestic credit. 494 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Table 1. A Comparison of Data Bases on State-Owned Enterprises in Developing Countries (total number of developing countries with more than five years of observations) Value Savings- Budgetary Data base and region added Investment investment gap' burdenh Short (1984) Total 33 51 37 33 Latin America and the Caribbean 9 23 22 17 Africa 14 16 8 8 Asia 10 12 7 8 Nair and Filippides (1989) Total 30 39 27 37 Latin America and the Caribbean 8 16 17 18 Africa 16 12 4 12 Asia 3 10 6 7 World Bank (1995) Total 65 70 63 65 Latin America and the Caribbean 22 28 26 24 Africa 31 27 27 32 Asia 12 15 10 9 Note: Although each data set has additional indicators, only the four most important are used for this comparison. Africa includes Sub-Saharan and North Africa; Asia includes South Asia, Southeast Asia and the Pacific, and Central Asia. Transition economies and industrial economies are excluded. a. Overall balances of state enterprises before transfers. b. Net financial flows from governments to state enterprises. changes and can be misleading when trying to identify the timing of policy changes. They also make it difficult to extrapolate the data forward because it is impossible to compare overlapping years. Nair and Filippides (1989) generally cover 1980-85 (although some countries and indicators also cover the 1970s) for up to 67 countries. The scope of their data base is narrower than that of Short, but similar to ours. Because Nair and Filippides present annual data, we were able to check whether the data were similar to our series and to incorporate consistent information into our data base. The BIB data base covers a much larger sample of countries than the earlier data bases, providing seven measures of state enterprise size and performance for up to 88 countries (tables 2 and A-1). In addition to yearly data, it gives individual country averages for 1978-85, 1985-91, and 1978-91 (using three- year moving-average estimates to extrapolate for the years without data) as well as averages by region, income group, and individual country. The regions are Africa (all of North Africa and Sub-Saharan Africa), Asia (South Asia, Southeast Asia and the Pacific, and Central Asian nontransitional economies), and Latin America and the Caribbean. Averages weighted by gross domestic product (GDP) in current U.S. dollars are also given. The income groups were calculated ac- cording to the income categories in the World Bank's 1992 World Development Indicators: low income, those with a 1992 per capita income of $675 or less; middle income, those with a 1992 per capita income of $676 to $8,355; and Haggarty and Shirley 495 industrial, those with a 1992 per capita income of $8,356 or more (World Bank 1992). When we calculated world averages or averages by region and income level, we excluded countries that did not have at least six years of data. For example, we have GDP shares for 76 countries but world averages for only 40 developing countries and 8 industrial countries. In addition, we excluded the transition countries of Central and Eastern Europe and Asia because of data discontinuities. Before the transition these countries did not use accounting prin- ciples comparable with those in market economies, making it impossible to con- struct a time series similar to the rest of our sample. Table 2. Definitions and Country Coverage for Indicators in the Bureaucrats in Business Data Base Number of Unweighted 1978-91 developing countries average for with more developing countries Total number than 5 years of with more than 5 Indicator of countries observations years of observations Share in economic activity (percentage of GDP)a 76 40 10.9 Share in investment (percentage of GDI)b 88 55 4.6 Share in employment (percentage of total employment)c 30 21 10.4 Savings-investment deficit before transfers (percentage of GDP)d 64 46 -1.6 Net financial flows from the government (percentage of GDP)e 68 37 -0.3 Share of domestic credit (percent)' 39 36 11.1 Share of total external debt (percent)g 82 74 13.6 a. Aggregate value added of country's state enterprises expressed as a percentage of GDP at current market prices. Value added is calculated as sales revenues minus cost of intermediate input (or as sum of operating surplus and wage payments). The data base also provides share in economic activity expressed as a percentage of nonagricultural GDP. b. State enterprise gross fixed capital formation (includes changes in stocks) as a percentage of gross domestic investment (GDI) at current market prices. The data base also provides share in investment expressed as a percentage of GDP. c. Number of full-time state enterprise employees as a percentage of total employment. The data base also provides share in employment expressed as a percentage of formal sector employment for 13 developing countries. d. State enterprise savings (sum of net operating and net nonoperating revenues) minus net capital expenditures (including stock changes) as a percentage of GDP. Operating savings are operating revenues minus intermediate inputs, wages, factor rentals, and depreciation, or gross operating profits. Net nonoperating revenues are revenues from such sources as earnings on financial assets minus nonoperating expenditures, such as losses on foreign exchange accounts. e. Flows to state enterprises from governments, including government loans, equity, and subsidies, minus flows from state enterprises to government, including taxes and dividends, as a percentage of GDP. f. Stock of domestic credit outstanding to state enterprises as percentage of gross credit outstanding to private sector, state enterprises, and governments. The data base also provides share of domestic credit expressed as a percentage of GDP for 39 countries, 3 of which are developing countries. g. State enterprise total external debt outstanding and disbursed as percentage of total external debt (both calculated in U.S. dollars). The data base also provides share of external debt expressed as a percentage of GDP (converted to U.S. dollars at current exchange rate). Source: World Bank (1995). 496 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Time Period Unlike the earlier data sets, the BIB data base covers 1978-91 and hence in- cludes a period of intense privatization (1985-91), although the 1991 cutoff date means that recent privatization and reform efforts cannot be assessed. Our goal was to assemble a time series with at least five years of data for every indicator in at least 35 countries. The data end in 1991, the last year for which information was available for a meaningful number of countries when the data base was built. The most important factor affecting state ownership since 1991 is the contin- ued growth in the number of privatizations. According to one privatization data base (originally presented in Sader 1994 and updated in World Bank 1997) the value of privatization transactions in developing countries grew sharply from 1988 to 1993 and then fell in 1994 and 1995 (figure 2). While there was almost $30 billion in sales during 1988-91 (the period covered by the BIB data base), there was almost $70 billion in the subsequent period (1992-95). One might think, therefore, that any conclusions concerning the impact of privatization on the share of state enterprises in GDP would be flawed because the data exclude the recent period of intense privatization. Although information on privatization flows is important for analyzing fiscal changes, private investment, and other issues, extrapolating changes in the stock of state enterprises from privatization data alone can be misleading, as the BIB data base has shown. If we compare yearly sales as a percentage of the privatiz- ing countries' GDP in 1988-91 (the period covered by the BIB data base) and in 1992-95 (the subsequent period), we find that the figure scarcely increased, from 0.475 percent (1988-91) to 0.519 percent (1992-95). The value of trans- Figure 2. Values of Privatization Proceeds in Developing Countries, 1988-95 Millions of U.S. dollars 25,000 t 20,000 15,000- - 10,000 5,000-- 1988 1989 1990 1991 1992 1993 1994 1995 Note. Based on data from 89 countries. Source: World Bank (1997). Haggarty and Shirley 497 actions rose after 1991 because the number of privatizing countries grew-from 38 in 1988-91 to 64 in 1992-95. In both periods most sales were concentrated in a few countries, while most countries sold relatively few firms. Thus the top five privatizing countries in 1988-91 (Argentina, Brazil, the Republic of Korea, Mexico, and Mozambique) generated 77 percent of the value of transactions, while during 1992-95 the top five (Argentina, Brazil, China, Malaysia, and Mexico) were responsible for 57 percent of the value of total sales. Also, in some cases sales revenues may not be related to trends in state ownership. For ex- ample, India sold $3.64 billion in nonfinancial state enterprise assets from 1991 to 1995, yet because both sellers and buyers were often part of the public sector, the impact on state ownership was negligible. There are plans to update a subset of the sample focusing on larger countries and major indicators for the 1998 issue of World Development Indicators (pub- lished by the World Bank). Some countries (for example, India, Korea, and Mexico) publish annual information on their state enterprises with about a year's delay. Thus interested scholars can also selectively update the BIB data base. Assembling current information for a large sample is difficult, however. Many governments do not publish fully consolidated data on their state firms, and it can be difficult to put the data in comparable form because countries vary in their definition of state enterprises. Collecting and assembling data for the BIB data base took more than a year. Definition and Sources Following Jones (1975) we defined state-owned enterprises as government- owned or -controlled economic entities that generate the bulk of their revenue from selling goods and services. This definition limits the set to commercial activities for which the government is able to control management decisions by virtue of its ownership stake. This definition has several advantages. First, it provides an objective way of distinguishing state enterprises from other gov- ernment activities that might be officially labeled state enterprises in some coun- tries but that receive the bulk of their income from general revenues (social security systems, road maintenance agencies, or agricultural research institutes) or government transfers (public health or university systems). Second, by de- fining a state enterprise as a government-controlled as well as a government- owned entity, we included enterprises directly operated by a government de- partment or indirectly owned through other state enterprises as well as enterprises in which the government owns a minority share but, given the dis- tribution of the other shares, has effective control. (Effective control can be ascertained by examining who designates the majority of the board and ap- points senior management.) Although we attempted to correct for deviations, we did not always have the information needed to make an individual country's data consistent with our definitions; these cases are detailed in the footnotes and specific country notes in the data base. On average, our data tend to understate rather than overstate the 498 THE WORLD BANK ECONOMIC REVIEW, VOL. 11. NO. 3 size of the state enterprise sector because governments rarely include enterprises that are not commercial, such as agricultural research institutes, on their lists of state enterprises. More frequently, they omit enterprises that clearly are state enterprises by excluding a particular legal form (for example, departmental en- terprises), state enterprises owned by a local rather than the national govern- ment (for example, most water and many power utilities), and smaller state enterprises (in terms of size or demand for fiscal resources). In addition, we had to limit the sample to central or federal government enterprises, because data on enterprises owned by local governments are almost nonexistent. We excluded financial enterprises because we lacked the time and resources needed to deal with their differing character. Another limitation of the data is the lack of break- down by sector. Despite considerable effort, we were unable to assemble sector information for a meaningful number of countries-an unfortunate outcome because many hypotheses about state ownership distinguish competitive firms from natural monopolies. The main sources for state enterprise data from developing countries were individual country publications (described in detail in World Bank 1995), country reports from the World Bank and International Monetary Fund, and Nair and Filippides (1989). Other sources included World Bank (1994); World Bank, African Development Indicators (various years); and World Bank data. Industrial-country data were from CEEP, Annales du CEEP (vari- ous years) and OECD, National Accounts Statistics (various years). GDP was taken from World Bank (1994) and state enterprise and total credit, from IMF (various years). Accounting Anomalies Users of this-or indeed any-data base on state enterprises should keep in mind that many state enterprises fail to use the Generally Accepted Accounting Principles and that the accounting rules applied can vary from country to country and from enterprise to enterprise. Often smaller state enterprises are not audited by internationally accredited accounting firms, meaning that there may be no in- dependent check on their record-keeping and reporting. Some of the indicators (such as state enterprise savings) are more vulnerable to the problems this presents than others (such as employment), and in general our measures are less vulnerable than the profitability measures widely used by other analyses. Because these prob- lems are widespread and highly varied, we assume that they do not bias the data and do not unduly affect comparability, which is not to minimize the serious prob- lems that accounting weaknesses create for objective analysis. III. THE BIB DATA BASE INDICATORS In this section we describe the seven measures of state enterprise size and per- formance in more detail and consider some of the issues raised by our findings. The indicators are defined in table 2. Appendix table A-I gives country coverage. Haggarty and Shirley 499 Share in Economic Activity This indicator was calculated as the state enterprise value added as a percent- age of GDP and, because the size of the agricultural sector (in which state enter- prises are normally minimal) varies substantially among countries, as a percent- age of nonagricultural GDP. The exclusion of agriculture has a large effect on some predominately agrarian economies, such as many African economies. For example, the share of state enterprises in Mali more than doubles when agricul- ture is excluded-from 15 percent of total GDP to 34 percent of nonagricultural GDP in 1978. The average state enterprise share of GDP in 40 developing countries with market economies (11 percent) was almost unchanged from 1978 to 1991 (fig- ure 1). During the same period, state enterprise share of GDP in 8 industrial countries fell from 9 to 7 percent. Employment shares were also steady. This surprising result in light of the extent of privatization led us to ask whether the trend in the share of nominal state enterprise value added in current price GDP is a meaningful measure of the change in the relative importance of state enterprises. Many countries keep state enterprise price increases below cost increases, particularly for infrastructure prices, meaning that the sector's nominal value added may be understated. However, a number of countries increased state en- terprise prices as part of their reform efforts in the second half of the 1980s, which could overstate value added. In other words, it could be that privatization did shrink the real share of state enterprises, but that price increases kept the nominal share from falling. For this to be true, state enterprise product prices would have to have risen faster than the average of all other prices in the economy. Evidence suggests that this did not occur. First, company information cited in the data base suggests that state enterprise price increases often failed to keep pace with inflation, as measured by the consumer price index. Second, countries that were decontrolling state enterprise prices were also liberalizing trade, which forced state enterprises in competitive markets to cut prices or at least restrain increases. This indicator can shed light on other issues. For example, looking across income levels, we find that state enterprises were more important in the econo- mies of lower-income countries, where they produced an average of almost 14 percent of GDP over the period studied, than in those of middle-income countries (9 percent; figure 1). The variance is large, despite the exclusion of the transition economies of Europe and Asia. At one extreme is Algeria, where state enter- prises produced an average of 65 percent of GDP between 1978 and 1991; at the other is the United States with 1.6 percent. Share in Investment We calculated aggregate state enterprise investment as a percentage of both gross domestic investment (GDI) and GDP. The first calculation measures the 500 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Figure 3. State-Owned Enterprises Shares in Gross Domestic Investment, 1978-91 Percent 35 - 30 - -income countrics 25 - 20 _ ~~~~~~~~~~~~~55 developing countries 20- 37 middle-income countries 15 _ iustrial countries 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 Note. Values are unweighted averages. Source. World Bank (1995, statistical appendix). importance of state enterprises compared with other investors, and the second measures the importance of state enterprise investment to the economy overall. The average share of GDI for 55 developing countries was 21 percent (figure 3), while the share of GDP was 4.6 percent. The variance is large: in 15 countries state enterprises accounted for more than one-third of all investment, including Egypt (59 percent), Zambia (55 percent), Venezuela (46 percent), Algeria (45 percent), India (41 percent), Turkey (36 percent), and Tunisia (35 percent). Over time this indicator behaves differently than value added. State enter- prise investment in a sample of 55 developing countries fell from 23 percent of GDI in 1978 to 19 percent by 1991, compared with 11 percent on average for a sample of 10 industrial countries (figure 3). Again, we can analyze patterns across income groups. State enterprise investment was 29 percent of GDI in low-income countries and 17 percent in middle-income countries. We can also identify re- gional differences: although the average state enterprise share of total invest- ment fell steadily in most countries, and all regional averages dropped between 1978 and 1991, the fall was much sharper in Latin America (28 percent) than in Africa (16 percent) or Asia (8 percent). Because in Latin America state enter- prises rely more on external debt to finance their investments than they do in other regions, the fall in investment may partly reflect the fall in the region's access to foreign credit during and after the debt crises of the mid-1980s. Share in Employment Share in employment was defined as full-time employees as a percentage of total employees. For 13 of the African countries in the sample we could report only formal or modern sector employment. The smaller denominator biases the Haggarty and Shirley 501 state enterprise employment shares upward: these 13 countries reported average employment shares of 22.8 percent compared with 15.3 percent for the 11 Afri- can countries that reported state enterprise employment as a percentage of total employment. The African average for this variable was based on a sample of nine countries that meet our time-series requirement. Five of those countries reported state enterprise shares of total employment, and four reported only formal sector shares. The weighted average for Africa, 16.4 percent, is closer to the average for countries reporting shares of total employment. The unweighted average, 20.6 percent, is closer to that of countries reporting only formal em- ployment and hence is less comparable with the rest of the world. The employment data support the premise that the share of government own- ership in economic activity did not change: shares were virtually steady through- out the period (figure 4). Again, there is wide variance among countries at dif- ferent income levels, although this is due partly to the upward bias in the African data. Because many state enterprises are capital intensive, we would expect their share of employment to be small relative to that of private firms. And, indeed, in Latin America and Asia their share in employment was only about one-third the size of their share in GDP. In Africa, however, the state enterprise share of em- ployment was similar to its share of GDP, even when countries reporting only shares of formal sector employment are excluded. It may be that state enter- prises were less capital intensive in Africa than in other regions, more over- staffed, or both. The indicator does illustrate that even in Africa, state enter- prises were not major employers. Indeed, only in 1 of our 43 sample countries- Guinea-did state enterprise employment exceed 50 percent of formal sector employment (68 percent). Figure 4. State-Ouned Enterprises' Sbares in Employment, 1978-91 Percent 18 10 low-income countries 16 - 14 - 12 - 21 developing countries 10 - 8- 11 middle-income countries 6 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 Note: Values are unweighted averages. Source: World Bank (1995, statistical appendix). 502 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Figure 5. Trends in State-Owned Enterprises' Balances and Government Fiscal Deficits, 1978-91 Percentage of GDP 2 0 , l -" State enterprise -2 savings minus investment -4 \ / G~~~~overnment fiscal cdeficit -6 -8 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 ANote: Sample is 38 developing market economies. Values are unweighted averages. Source: World Bank (1995, statistical appendix). Overall Balances before Transfers The overall balance, or savings-investment deficit, tells us the extent to which the savings of state enterprises cover their net capital expenditures, that is, how much state enterprises demand from the rest of the economy to finance invest- ment and service their debt. A temporary deficit caused by productive new in- vestments would not be a problem, because such investments would provide the enterprise with the resources to repay its debt in the future. Although private firms tend to finance the bulk of their investments from retained earnings, large state enterprises, which make major infrastructure investments, need not do the same. A persistent deficit could signal a problem, however, and we would want to know the cause. Because many state enterprise sectors include some of the country's major revenue earners (petroleum companies, tobacco monopolies, gold mines), a zero balance or a very small surplus might also be cause for con- cern. Is there a deficit or small surplus because state enterprises are investing heavily in projects with high future returns, or because productivity is low, prices are set below costs, past investments are yielding low returns, or state enter- prises are borrowing to cover current losses? We would also want to know how the deficit is being financed. Is it contributing to fiscal deficits, domestic debt, or foreign debt, each of which can cause problems? The average overall balance for a sample of 46 developing countries was -1.6 percent of GDP. Again, we see a variance across countries. Although the average deficit of the entire sample improved after the early 1980s (figure 5), it deterio- Haggarty and Shirley 503 rated in low-income countries after the mid-1980s (figure 6). We could not ascertain the cause of the deficit from the aggregate information in the BIB data base. However, case studies of countries and firms in BIB find persistently higher deficits in enterprises where productivity is low; tariffs are administered in an erratic, ad hoc fashion; and public banks are forced to lend to state enterprises. The savings-investment deficit moved closely in tandem with the government fiscal deficit, averaging about 35 percent of the fiscal deficit from 1978 to 1991 for 38 developing countries (figure 5). We interpreted this parallel movement as an important indicator of the impact of state enterprise performance on growth. Other studies find fiscal deficits unambiguously bad for growth (Easterly, Rodriguez, and Schmidt-Hebbel 1994; see also Fischer 1993). In calculating state enterprise savings we eliminated all transfers, which in- clude, for example, subsidies on the revenue side and dividends on the expendi- ture side. We did so because they would obscure the measure we were trying to capture: the resources that state enterprises require from the economy. We could not, however, exclude hidden subsidies-those that do not appear on any in- come statement or budget, including loans at below-market rates, nonpayment of interest charges, conversions of state enterprise loans into government equity, duty exemptions, procurement preferences, forgiven taxes or arrears between state enterprises, access to goods and services produced by other state enter- prises at below-market prices, and use of land and buildings rent free. Country data suggest that these hidden subsidies are sometimes more significant than overt subsidies. In Kenya, for example, indirect subsidies during 1991-92 were three to four times larger than direct subsidies and increased as direct subsidies fell (Investissement Developpement Conseil 1993: 17). Because these hidden Figure 6. Trends in State-Owned Enterprises' Overall Balances, 1978-91 Percentage of GDP 29 middle-income countries 0 17 low-income countries -2 . -3 -4 . 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 Note: Overall balances are savings-investment balances. Values are unweighted averages. Source: World Bank (1995, statistical appendix). s04 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 subsidies generally add to the income of an enterprise (or lower expenses), they make the sector appear financially better off than it actually is. Net Financial Flows from the Government to State Enterprises Net financial flows from the government indicate the net burden of state en- terprises on the budget or their contribution to it.3 In two-thirds of our sample countries state enterprises initially received more from the government than they paid. But net financial transfers to state enterprises declined over the period to the point where, by 1991, state enterprises were making transfers to govern- ments equal to about 1 percent of GDP. State enterprises have three ways of financing their deficits: transfers, domes- tic borrowing, and foreign borrowing. There were marked differences in the form of finance by country income level. State enterprises in low-income coun- tries relied much more heavily on government transfers than did middle-income countries. In fact, in middle-income countries state enterprises made net trans- fers to governments (averaging 0.8 percent of GDP from 1978 to 1991), even when they ran a net deficit. These numbers are strongly influenced by the data from Chile and Venezuela, where large extractive state enterprises (copper and petroleum) made sizable net transfers. Share of Domestic Credit We compared state enterprise credit with total credit and calculated state enterprise credit as a percentage of GDP.4 These measures tell us how successful state enterprises are at capturing credit compared with other borrowers and how important that credit is in the economy. The state enterprise share of do- mestic credit is understated because governments frequently convert state enter- prise domestic debt to equity. Again, different patterns emerge among different groups of countries. Do- mestic credit was an important source of deficit financing in low-income coun- tries, where state enterprises captured about 15 percent of gross domestic credit compared with about half that in middle-income countries. The average credit share of state enterprises for all developing countries (about 11 percent) fell only slightly over 1978-91, although it dropped more sharply in low-income coun- tries (from 16 to 13 percent in a sample of 16 countries). Share of Total External Debt The third form of state enterprise financing is foreign borrowing. These num- bers understate state enterprise foreign debt because they do not capture debt that is assumed or incurred by the government on behalf of state enterprises-a 3. This definition differs from that of Nair and Filippides (1989) in that we treat taxes paid by state enterprises as a transfer to the government. 4. Again, our definition differs from that of Nair and Filippides (1989), who use net rather than gross credit. By using gross credit we avoid the negative ratios that would arise if we used ner credit and government deposits with the central bank exceeded central bank credit to the government. Haggarty and Shirley 505 relatively frequent occurrence. State enterprises in middle-income countries re- lied more on external credit (accounting for an average of 16 percent of total foreign debt) than did those in low-income countries (10 percent on average). After rising in the early 1980s, state enterprise foreign borrowing fell sharply as a share of total borrowing. IV. POTENTIAL USES FOR THE DATA BASE BIB used the data base to answer the questions posed at the outset of this article by analyzing trends in the magnitude and performance of state enter- prises. But many other potential uses have yet to be exploited. One is to compare outliers with the rest of the sample. For example, in 11 countries the state enter- prise share of GDP ranges from two to six times the average for developing mar- ket economies. How do these countries fare against the rest of the sample on other economic variables, such as those in Summers, Heston, and Nuxoll (1994)? Another is to analyze regional or income differences. Scholars interested in specific countries could also use the data base to assess the impact of state enterprises on a specific economy. For example, in some countries state enterprises command high levels of credit relative to their value added. Outstanding credit to state enterprises in Bangladesh averaged 6.5 per- cent of GDP over 1978-91, even though their value added and investment aver- aged only 2.7 and 2.6 percent of GDP, respectively. Or scholars could consider trends in employment and value added to get a sense of movements in labor productivity. In Algeria, for instance, the share of state enterprises in GDP fell 31.1 percent from 1980 to 1989, while their share of employment fell only 8.8 percent. In Ghana state enterprise value added as a share of GDP fell more than 50 percent from 1986 to 1991, while employment shares rose almost two-thirds over the same period. The data can also shed light on the role of the private sector in economies where direct information is limited. National accounts and other statistics often lump state enterprises and private firms into a nongovernment category. BIB data make it possible to separate state enterprise from private value added, invest- ment, employment, or credit statistics. V. CONCLUSIONS The state enterprise statistics assembled in our data base represent a signifi- cant increase in coverage and time span over previous data bases. Nevertheless, serious gaps remain. Only two-thirds of the developing countries for which we have any observations have enough annual data to be part of the period aver- ages. The omission of transition economies and financial state enterprises leaves out an important region and sector. The problems we faced in filling these gaps were partly inadvertent but may also have been partly intentional. On the one hand, the fact that we 506 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 could find many more observations for middle-income than for lower- income countries suggests that the poor statistics in the latter are partly a result of underdeveloped and underfunded statistical offices. On the other hand, the fact that some middle-income countries also had very poor data and that many countries relied on hidden subsidies suggests that some gov- ernments may have intentionally underreported their state enterprises. One reason for this may be that state enterprises are an off-budget expense, less subject to scrutiny and control by the legislature. For example, an important and often overlooked consequence of Mexico's privatization program was the elimination of hundreds of state enterprises that were not functional, consisting principally of an act of creation and a bank account. Another reason may be that countries wish to avoid angering trading partners that might view state enterprise subsidies as an unfair trading practice. Regard- less of the motive, the net effect is to make it difficult for researchers to collect reliable information about this important aspect of government activ- ity. We hope that future researchers will begin to fill these gaps. (Table A-1 begins on the followinig page.) Table A-1. Countries and Indicators in the Data Base Overall Share of Share balancesa Net financial Share of Share of economic of gross before flows from gross total activity domestic Share of transfers government domestic external (percentage investment employment (percentage (percentage credit debt Country of GDP) (percent) (percent) of GDP) of GDP) (percent) (percent) Low income Bangladesh x x x x x Benin x x x x x Bhutan x Burkina Faso x x Burundi x x x x x x x Central African Republic x x x x x Comoros x x x Egypt, Arab Republic of x x x x x x x Ethiopia x Gambia, The x x x x x x Ghana x x x x x x x Guinea x x x x Guinea-Bissau x x Guyana x x x x x x Haiti x x x x x Honduras x x x x x India x x x x x x Indonesia x x x x x x Kenya x x x x x x x Liberia x x Madagascar x x x x Malawi x x x x x x Maldives x (Table continued on tbefollowingpage.) Table A-1. (continued) Overall Share of Share balances, Net financial Share of Share of economic of gross before flows from gross total activity domestic Share of transfers government domestic external (percentage investment employment (percentage (percentage credit debt Country of GDP) (percent) (percent) of GDP) of GDP) (percent) (percent) Mali x x x x x x Mauritania x x x x x Myanmar x x x Nepal x x x x x x Nicaragua x Niger x x x x x Nigeria x x x x x Pakistan x x x Rwanda x x x Sao Tome and Principe x Sierra Leone x x x x x x Somalia x Sri lanka x x x x x Sudan x x x Tanzania x x x x x x Togo x x x x x x Uganda x Zaire x x x x x x x Zambia x x x x x x Zimbabwe x x Middle income Algeria x x x x x Antigua and Barbuda x Argentina x x x x x x Barbados x x x x Belize x x x x Bolivia x x x x x x Botswana x x x x x x x Brazil x x x x x x x Cameroon x x x x x Cape Verde x x Chile x x x x x x x Colombia x x x x x x x Congo x x x x x x Costa Rica x x x x x C6te d'Ivoire x x x x x x Djibouti x Dominica x x x x x x Dominican Republic x x x x x x Ecuador x x x x x x El Salvador x x x x x x Fiji x x Gabon x x Greece x x Grenada x x x x x Guatemala x x x x x Jamaica x x x x x Korea, Republic of x x x x x x Malaysia x x x x x Mauritius x x x x x x Mexico x x x x x x x Morocco x x x x x Namibia x x x x Panama x x x x x (Table continued on the following page.) Table A-1. (continued) Overall Share of Share balances' Net financial Share of Share of economic of gross before flows from gross total activity domestic Share of transfers government domestic external (percentage investment employment (percentage (percentage credit debt Country of GDP) (percent) (percent) of GDP) of GDP) (percent) (percent) Papua New Guinea x x Paraguay x x x x x Peru x x x x x x Philippines x x x x x x x Portugal x x Senegal x x x x x x Seychelles x x x x Solomon Islands x South Africa x x St. Kitts and Nevis x x x St. Lucia x x x St. Vincent and the Grenadines x x x x x Swaziland x Thailand x x x x x x x Trinidad and Tobago x x x x x x x T unisia x x x x x Turkey x x x x x x Uruguay x x x x x x Van uatu x Vcnezuela x x x x x x Western Samoa x High income Australia x Austria x x Belgium x x Denmark x x France x x Germany x x Ireland x Italy x x Japan x Netherlands x Norway x Spain x x Sweden x Taiwan (China) r x x x United Kingdom x x United States x x Note: For detailed definitions of the indicators, see table 2. a. Savings-investment balance. Source: Authors' calculations. 512 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 REFERENCES The word "processed" describes informally reproduced works that may not be com- monly available through library systems. Boardman, Anthony E., and Aidan R. Vining. 1989. "Ownership and Performance in Competitive Environments: A Comparison of the Performance of Private, Mixed, and State-Owned Enterprises." Journal of Law and Economics 32(1):1-33. Borcherding, Thomas, W. W. Pommerehne, and Friedrich Schneider. 1982. "Compar- ing the Efficiency of Private and Public Production: The Evidence from Five Coun- tries." Zeitschrift fur Nationalokonomie supplement 2: 127-56. Candoy-Sekse, Rebecca. 1988. Techniques of Privatization of State-Owned Enterprises. Vol. 3, Inventory of Country Experience and References. Technical Paper 90. Wash- ington, D.C.: World Bank. CEEP (Centre Europ6en de l'Entreprise Publique). Various years. Annales du CEEP. Brussels. Easterly, William, Carlos Alfredo Rodriguez, and Klaus Schmidt-Hebbel, eds. 1994. Public Sector Deficits and Macroeconomic Performance. New York: Oxford Univer- sity Press. Fischer, Stanley. 1993. "Role of Macroeconomic Factors in Growth." Journal of Mon- etary Economics 32(December):485-512. Galal, Ahmed, Leroy Jones, Pankaj Tandon, and Ingo Vogelsang. 1994. Welfare Conse- quences of Selling Public Enterprises: An Empirical Analysis. New York: Oxford University Press. IMF (International Monetary Fund). Various years. International Financial Statistics. Washington, D.C. Investissement D6veloppement Conseil. 1993. "Kenya: Study of Subsidies and Other Financial Flows in Favor of Major PEs." Paris. Processed. Jones, Leroy. 1975. Public Enterprise and Economic Development: The Korean Case. Seoul: Korean Development Institute. Kikeri, Sunita, John Nellis, and Mary Shirley. 1992. Privatization: The Lessons of Expe- rience. Washington, D.C.: World Bank. Megginson, William, Robert Nash, and Matthias van Randenborgh. 1994. "The Finan- cial and Operating Performance of Newly Privatized Firms: An International Empiri- cal Analysis." Journal of Finance 49(2):403-52. Millward, Robert. 1988. "Measured Sources of Inefficiencies in the Performance of Public and Private Enterprises in LDCs." In Paul Cook and Colin Kirkpatrick, eds., Privatiza- tion in Less Developed Countries. Sussex, U.K.: Wheatsheaf Books. Millward, Robert, and D. M. Parker. 1983. "Public and Private Enterprise: Compara- tive Behavior and Relative Efficiency." In Robert Millward, D. M\. Parker, L. Rosenthal, M. T. Summer, and N. Topman, eds., Public Sector Economics. New York: Longman. Nair, Gouindan, and Anastasios Filippides. 1989. "How Much Do State-Owned Enter- prises Contribute to Public Sector Deficits in Developing Countries-and Why?" Working Paper 45. World Development Report Office, World Bank, Washington, D.C. Processed. OECD (Organisation for Economic Co-operation and Development). Various years. Na- tional Accounts Statistics. Paris. Haggarty and Shirley 513 Pollitt, Michael G. 1995. Ownership and Performance in Electric Utilities: The Interna- tional Evidence on Privatization and Efficiency. Oxford: Oxford University Press for the Oxford Institute of Energy Studies. Sader, Frank. 1994. "Privatization Techniques and Foreign Investment in Developing Countries, 1988-93." Foreign Investment Advisory Services, World Bank, Washing- ton, D.C. Processed. Sappington, David E. M. 1991. "Incentives in Principal-Agent Relationships." Journal of Economic Perspectives 5(2):45-66. Sappington, David E. M., and Joseph E. Stiglitz. 1987. "Privatization, Information, and Incentives." NBER Working Paper 2196. National Bureau of Economic Research, Cambridge, Mass. Processed. Shapiro, Carl, and Robert D. Willig. 1990. "Economic Rationales for the Scope of Privatization." John M. Olin Program for the Study of Economic Organization and Public Policy Discussion Papers 41. Woodrow Wilson School of Public and Interna- tional Affairs, Princeton University, Princeton, N.J. Processed. Short, Peter. 1984. "The Role of Public Enterprise: An International Statistical Com- parison." In R. H. Floyd, C. S. Gray, and R. P. Short, eds., Public Enterprise in Mixed Economies: Some Macroeconomic Aspects. Washington, D.C.: International Mon- etary Fund. Suleiman, Ezra N., and John Waterbury, eds. 1990. The Political Economy of Public Sector Reform and Privatization. Oxford: Westview Press. Summers, Robert, Alan Heston, and Daniel A. Nuxoll. 1994. "Differential-Productivity Hypothesis and Purchasing-Power Parities: Some New Evidence." Review of Interna- tional Economics 2(October):227-43. Vickers, John, and George Yarrow. 1988. Privatization: An Economic Analysis. Cam- bridge, Mass.: MIT Press. World Bank. Various years. African Development Indicators. Washington, D.C. 1992. World Development Indicators. Washington, D.C. 1994. World Tables. Baltimore, Md.: The Johns Hopkins University Press. 1995. Bureaucrats in Business: The Economics and Politics of Government Ownership. New York: Oxford University Press. 1. 1997. Global Development Finance 1997. Washington, D.C. Yarrow, George. 1986. "Privatization in Theory and Practice." Economic Policy: A European Forum 32(April):323-77. THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3: 515-16 Index of Authors for Volume 11 Aryeetey, Ernest, Hemamala Hettige, Machiko Nissanke, and William Steel, "Fi- nancial Market Fragmentation and Reforms in Ghana, Malawi, Nigeria, and Tanzania" (2, May):195-218 Assaad, Ragui,"The Effects of Public Sector Hiring and Compensation Policies on the Egyptian Labor Market" (1, January):85-118 Banerji, Arup, and Hafez Ghanem, "Does the Type of Political Regime Matter for Trade and Labor Market Policies?" (1, January):171-94 Batra, Geeta (see Tan, Hong) Chen, Shaohua (see Ravallion, Martin) Coleman, Jonathan R. (see Faruqee, Rashid) Devarajan, Shantayanan, Hafez Ghanem, and Karen Thierfelder, "Economic Re- form and Labor Unions: A General-Equilibrium Analysis Applied to Bangladesh and Indonesia" (1, January):145-70 Dinar, Ariel (see Yacov, Tsur) Diwan, Ishac, and Michael Walton, "How International Exchange, Technology, and Institutions Affect Workers: An Introduction" (1, January):1-15 Djankov, Simeon (see Hoekman, Bernard) Eskeland, Gunnar S., and Tarhan Feyzioglu, "Rationing Can Backfire: The 'Day without a Car' in Mexico City" (3, September):383-408 Faruqee, Rashid, Jonathan R. Coleman, and Tom Scott, "Managing Price Risk in the Pakistan Wheat Market" (2, May):263-92 Feyzioglu, Tarhan (see Eskeland, Gunnar S.) Ghanem, Hafez (see Banerji, Arup) Ghanem, Hafez (see Devarajan, Shantayanan) Haggarty, Luke, and Mary M. Shirley, "A New Data Base on State-Owned Enter- prises" (3, September):491-513 Hammer, Jeffrey S., "Prices and Protocols in Public Health Care" (3, September): 409-32 Hettige, Hemamala (see Aryeetey, Ernest) Hettige, Hemamala (see Pargal, Sheoli) Hoekman, Bernard, and Simeon Djankov, "Determinants of the Export Structure of Countries in Central and Eastern Europe" (3, September):471-87 Isham, Jonathan, Daniel Kaufmann, and Lant H. Pritchett, "Civil Liberties, Democ- racy, and the Performance of Government Projects" (2, May):219-42 S1S 516 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 Kaufmann, Daniel (see Isham, Jonathan) Mody, Ashoka, and Fang-Yi Wang, "Explaining Industrial Growth in Coastal China: Economic Reforms . . . and What Else?" (2, May):293-325 Nissanke, Machiko (see Aryeetey, Ernest) Pargal, Sheoli, Hemamala Hettige, Manjula Singh, and David Wheeler, "Formal and Informal Regulation of Industrial Pollution: Comparative Evidence from Indonesia and the United States" (3, September):433-50 Pissarides, Christopher A., "Learning by Trading and the Returns to Human Capi- tal in Developing Countries" (1, January) :17-32 Pritchett, Lant H. (see Isham, Jonathan) Rama, Martfn, "Organized Labor and the Political Economy of Product Market Distortions" (2, May):327-55 Ravallion, Martin, and Shaohua Chen, "What Can New Survey Data Tell Us about Recent Changes in Distribution and Poverty?" (2, May):357-82 Sarno, Lucio (see Taylor, Mark P.) Scott, Tom (see Faruqee, Rashid) Shirley, Mary M. (see Haggarty, Luke) Singh, Manjula (see Pargal, Sheoli) Squire, Lyn, and Sethaput Suthiwart-Narueput, "The Impact of Labor Market Regu- lations" (1, January):119-43 Steel, William (see Aryeetey, Ernest) Suthiwart-Narueput, Sethaput (see Squire, Lyn) Tan, Hong, and Geeta Batra, "Technology and Firm Size-Wage Differentials in Colombia, Mexico, and Taiwan (China)" (1, January):59-83 Taylor, Mark P., and Lucio Sarno, "Capital Flows to Developing Countries: Long- and Short-Term Determinants" (3, September):451-70 Thierfelder, Karen (see Devarajan, Shantayanan) Walton, Michael (see Diwan, Ishac) Wang, Fang-Yi (Mody, Ashok) Wheeler, David (see Pargal, Sheoli) Wood, Adrian, "Openness and Wage Inequality in Developing Countries: The Latin American Challenge to East Asian Conventional Wisdom" (1, January):33-57 Yacov,Tsur, and Ariel Dinar, "The Relative Efficiency and Implementation Costs of Alternative Methods for Pricing Irrigation Water" (2, May):243-62 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3: 517-18 Index of Titles for Volume 11 "Capital Flows to Developing Countries: Long- and Short-Term Determinants," by Mark P. Taylor and Lucio Sarno (3, September):451-70 "Civil Liberties, Democracy, and the Performance of Government Projects," by Jonathan Isham, Daniel Kaufmann, and Lant H. Pritchett (2, May):219-42 "Determinants of the Export Structure of Countries in Central and Eastern Eu- rope," by Bernard Hoekman and Simeon Djankov (3, September):471-87 "Does the Type of Political Regime Matter for Trade and Labor Market Policies?" by Arup Banerji and Hafez Ghanem (1, January):171-94 "Economic Reform and Labor Unions: A General-Equilibrium Analysis Applied to Bangladesh and Indonesia," by Shantayanan Devarajan, Hafez Ghanem, and Karen Thierfelder (1, January):145-70 "The Effects of Public Sector Hiring and Compensation Policies on the Egyptian Labor Market," by Ragui Assaad (1, January):85-118 "Explaining Industrial Growth in Coastal China: Economic Reforms...and What Else?" by Ashoka Mody and Fang-Yi Wang (2, May):293-325 "Financial Market Fragmentation and Reforms in Ghana, Malawi, Nigeria, and Tanzania," by Ernest Aryeetey, Hemamala Hettige, Machiko Nissanke, and Wil- liam Steel (2, May):195-218 "Formal and Informal Regulation of Industrial Pollution: Comparative Evidence from Indonesia and the United States," by Sheoli Pargal, Hemamala Hettige, Manjula Singh, and David Wheeler (3, September):433-50 "How International Exchange, Technology, and Institutions Affect Workers: An Introduction," by Ishac Diwan and Michael Walton (1, January):1-15 "The Impact of Labor Market Regulations," by Lyn Squire and Sethaput Suthiwart- Narueput (1, January):119-43 "Learning by Trading and the Returns to Human Capital in Developing Countries," by Christopher A. Pissarides (1, January):17-32 "Managing Price Risk in the Pakistan Wheat Market," by Rashid Faruqee, Jonathan R. Coleman, and Tom Scott (2, May):263-92 "A New Data Base on State-Owned Enterprises," by Luke Haggarty and Mary M. Shirley (3, September):491-513 "Openness and Wage Inequality in Developing Countries: The Latin American Challenge to East Asian Conventional Wisdom," by Adrian Wood (1, January):33- 57 "Organized Labor and the Political Economy of Product Market Distortions," by Martin Rama (2, May):327-55 517 518 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3 "Prices and Protocols in Public Health Care," by Jeffrey S. Hammer (3, Septem- ber):409-32 "Rationing Can Backfire: The 'Day without a Car' in Mexico City," by Gunnar S. Eskeland and Tarhan Feyzioglu (3, September):383-408 "The Relative Efficiency and Implementation Costs of Alternative Methods for Pricing Irrigation Water," by Yacov Tsur and Ariel Dinar (2, May):243-62 "Technology and Firm Size-Wage Differentials in Colombia, Mexico, and Taiwan (China)," by Hong Tan and Geeta Batra (1, January):59-83 "What Can New Survey Data Tell Us about Recent Changes in Distribution and Poverty?" by Martin Ravallion and Shaohua Chen (2, May):357-82 THE WORLD BANK ECONOMIC REVIEW, VOL. 11, NO. 3: 519 List of Referees The Editorial Board of The World Bank Economic Review thanks the following referees for their contribution to the editorial process during the past year. John Abowd Alan Gelb Paul Mosley Pierre-Richard Agenor Christopher L. Gilbert John D. Nash John Antle Indermit S. Gill Stephen O'Connell Arup Banerji Jack D. Glen Howard Pack Edward B. Barbier Paul Glewwe Enrico Perotti Robert J. Barro David Greenaway Guy P. Pfeffermann Michael Beenstock Reuben Gronau Christopher A. Pissarides Jere R. Behrman Dominique Hachette Thomas J. Prusa Clive Bell Jeffrey S. Hammer Graham Pyatt Linda Bell Gordon Hanson Martin Rama Dan Ben-David Eric Hanushek Vijayendra Rao Dwayne H. Benjamin Thomas F. Hellmann Martin Ravallion Timothy Besley Leonardo Hernandez Carmen M. Reinhart Eduardo Borensztein Arye L. Hillman E. Liliana Rojas-Suarez Ralph Bradburd Karla Hoff Xavier Sala-i-Martin Kenneth M. Chomitz R.G. Hubbard T. Paul Schultz Stijn Claessens Harry Huizinga Nemat Shafik Christopher Clague William F. Hyde Cora Shaw Daniel Cohen Kwang Jun Anne C. Sibert Fabrizio Coricelli Homi Kharas Khalid Siraj Alejandra Cox-Edwards Elizabeth M. King Kenneth A. Small Francesco Daveri Lori Kletzer Alasdair Smith Steven J. Davis Anjini Kochar T. N. Srinivasan Jaime de Melo Carsten Kowalczyk Sethaput Suthiwart- Ishac Diwan Aart Kraay Narueput Juan J. Dolado Donald F. Larson Duncan Thomas Ronald Duncan Victor Lavy Aaron Tornell William Easterly Danny M. Leipziger Wim P. M. Vijverberg Antonio Estache Ross Levine Dimitri Vittas Riccardo Faini Justin Yifu Lin Milan Vodopivec Asif Faiz David Lindauer Beatrice Weder Peter R. Fallon Marlaine E. Lockheed Larry E. Westphal Gary S. Fields Ali Mansoor John Whalley Deon Filmer Andrew D. Mason Jeffrey G. Williamson Ariel Fiszbein William L. Megginson John Williamson Bruce L. Gardner Branko L. Milanovic Stephen Younger 519 New from the World Bank I~ I I I !nzE The State in a Changing World overnment is in the spotlight in this twentieth annual edition of GWorld Development Report. This year the World Bank's flag- ship publication is devoted to the role and effectiveness of the state, a topic that ranks high on the agenda in developing and industrial coun- tries alike. The Report looks at what the state should do, how it should do it, and how it can do it better in a rapidly changing world. June 1997. 354pages. English editions: Paperback: Stock no. 61114 (ISBN 0-19-521114-6); $25.95. Hardcover: Stock no. 61115 (ISBN 0-19-521115-4). $49.95. Translations forthcoming in paperback. Publishedfor the World Bank by Oxford University Press. Private Capital Flows to Developing Countries: The Road to Financial Integration T he world's financial markets are rapidly integrating into a single global Tmarketplace. Developing countries are being drawn into this process starting from different points and moving at various speeds-some are pre- pared for the change but others are not. This World Bank report looks at the important challenges both sets of countries face in a new age of global capital. The book presents new and compelling evidence that private capital flows have entered a new phase, driven by increasedfinancial inte- gration. The report analyzes the causes and effects of integration, with a particular emphasis on how developing countries in the nascent stages of integration can learn from the experiences of the more rapidly integrating developing countries. M, ay 1997 300pages. Stock no. 61116 (ISBN 0-19-521116-2). $40.00. Publishedfor the World Bank byX Oxford University Press Visit our Website: http://www.worldbank.org f loFor US customers, contact The World Bank, PO. Box 7247-8619, = Philadelphia. PA 19170-8619. Phone: (703) 661-1580, Fix: (703) = World Bank 661-1501. Shipping and handling: uS$5.00. Airmail delivery outside - Publications *****.*the US is US$13.00 for one item plus US$6.00 for each additional 1 W Ji t Mublications item. Pavment bv US$ check drawn on a US bank payable to the World Bank or by VISA, MasterCard, or American Express. Customers out- = side the US, please contact your World Bank distributor. ER97 Coming Th Wrd `Baxk' this Fall... -Ni this pathbrealng report on ... Tthe global AIDS epidemic outlines the strategic role that governments must play in slowing N F ON the spread of HIV and mitigating the impact of AIDS on morbidity PUBUC and mortality. Drawing on a PRIORITIES wealth of knowledge that has IN A GLOBAL accumulated in the 15 years since EPIDEMIC the virus that causes AIDS was first identified, the report high- lights policies that are most likely m to be effective in managing the epidemic. These include early actions to minimize the spread of the virus, aiming preventive inter- ventions at high risk groups and evaluating measures that would assist households affected by AIDS according to the same stan- dards applied to other health and P:* j . poverty issues. This new policy research report from the World Bank will be a valuable resource for the public health community, policy- makers, researchers, and anyone with an interest in this devastating global health crisis. Publishedfor the World Bank by Oxford University Press October 1997. 300 pages (approx.). Stock no. 61117 (ISBN 0-19-521117-0). $30.00. Visit our Website: http:/www. worldbank.org - Bank For US customers, contact The World Bank, P.O. Box 7247-8619, Philadephia, World Bank PA 19170-8619. Phone: (703) 661-1580, Fax: (703) 661-1501. Shipping and LV~FJJJ Publications handling: US$5.00. Airmail delivery outside the US is US$13.00 for one item plus US$6.00 for each additional item. Payment by US$ check drawn on a US bank payable to the World Bank or by VISA, MasterCard, or American Express. ER97 Customers outside the US, please contact your World Bank distributor. Distributors of COLOMBIA GERMANY ISRAEL NEPAL PORTUGAL SWEDEN DIStributors of InfoenlaceLtda. UNO-Vedag Yozmot Literature Ltd. Everest Media Intemational Services (P) Livrana Poolugal Wennergren-Williams AB W orld Bank Carrera6No.51-21 PoppelsdorierAllee55 PO. Box 58055 Ltd. Apaoado2681,Rua DoCarm 70-74 P 0. Box 1305 Apartado Aereo 34270 53115 Bonn 3Yohanan HasandlarStreet GPO Box 5443 1200 Lisbon S-17125 Solna Publications Santafd de Bogota, D.C. Tel: (49 228) 949020 Tel Aviv 61560 Kathmandu Tel: (1) 347-4982 Tel: (46 8) 705-97-50 Prices and credit terrns vary Tel: (57 1) 285-2798 Tel: (972 3) 5285-397 Tel: (977 1) 472 152 ftoni cotintry to coiiiitry, GREECE ~~~~~~~~~~~~~~~~ROMANIA SWITZEAiLAND fom 5o1nt slno(a distr/i/ COTE DIVOIRE Papasatiriaa S.A. RBOA tatamatinal NETHERLANDS Caolpani eD Libnrii BucurestiS.A Libraine Payat Service Instituionnel yom. ul oesclclde!c usr Center dEdition etlde Dittusion 35, Stoumara Sir. PO Boa 13056 De LindeboomttnOr-Publikaties Str, Lipscani ao. 26, sector 3 C65tes-de-Montbenon 3D before placing an order. Aincaines (CEDA) 106 82 Athens Tel Aviv 61130 PO. Box 202, 7480 AE Haaksbergen Bucharest 1002 Lausanne ARGENTINA A n4 BP 541 Tel: (30 1) 364-1826 Tel: (972 3) 5461423 Tel: (31 53) 574-0064 Tel: (40 1) 613 9645 Tel: (41 21) 341-3229 OfiiAR dElNIbrAIteaioa Abidjan 04 Oicina delLibro ntemacional Tel: (225) 24 6510:24 6511 HAITI Palestinian Authority/Middle East NEWZEALAND RUSSIANFEDERATION ADECO Van Diermen Av. Cordoba 1877 Cu4ure Diftusion Index Infotmation Services EBSCO NZ Ltd. Isdatelstvo cVes Mir> EditionsTechniques 1120 Bueos Aires CYPRUS 5,RueCeapois PO.B.19502 Jerusalem PrivateMail Bag 99914 9a,Lolpacniy Pereulok Ch. deLacuez 41 Tel: (54 1)815-8354 CenterlorApplied Research C.P 257 Tel: (9722) 6271219 New Market, Auckland Moscow 101831 CH1807 Blonay AUSTRALIA,FIJI,PAPUAsEW CYgiollnegstree1Engomi Pon-au-Prince ITTel: (649) 524-8119 Tel:(V095)9178749 Tel: (41 21)943 2673 AUSTRALIA, FIJI, PAPUA NEW 6, Diogenes Street, Engomi Tel: (509) 23 9260 ITALY GUINEA, SOLOMON ISLANDS, PO Box 2006 Lcosa Commissionaria Sansoni SPA NIGERIA SINGAPORE, TAIWAN, THAILAND VANUATUI, AND WESTERN SAMOA Niwcsia HONG KONG, MACAO Via Duca Di Calabia, t/t Universiy Press Limited MYANMAR,BRUNEI Cental Books Distribution D.A. Infonmation Services Tel: (357 2) 44-1730 Asia2000 Ltd. Casella Postale 552 Three Crowns Building Jercho Asalgaite PutishitgAsiaParcdicPte. Wd. 306 Silom Road 648 Whiehorse Road Sales &Circulation Depaurment 50125 Firenze Private Mail Bag 5095 41 Kallang Pudding Road #04.03 Bangkok 10500 Mtcham 3132 CZECH REPUBLIC Seabird House, unit 1101 -02 Tel: (55)645-415 lbadan Golden Wheel Building Tel: (662) 235-5400 icl:oria National lnformation Center 22-28 Wyndham Street, Central Tel: (234 22) 41-1356 Singapore34931T Tel: (61) 3 9210 7777 prodejna, Konviktska 5 Hong Kong JAMAICA Tel: (65) 741-5166 TRINIDAD&TOBAGO, CS -11357 Prague 1 Tel: (852)2530-1409 lan Rande Publishers Ltd. NORWAY ANDTHE CARRIBBEAN AUSTRIA Tel: (422) 2422-9433 206 Old Hope Road, Kingston B SIC Into ArS SLOVENIA Syslematics Studies Unit Gerold and Co. HUNGARY Tel: 876-927-2065 Book Depanment, Postboks6512 Gospodarski Vestnik Publishing Group 9 Waus Street Weihburggasse 26 DENMARK Euro Into Service Eterslad Dunajska cesta 5 Curepe A-l0l1 Wien SamfundsLitteratur Margitszgeti Europa Haz JAPAN N-0660slo 1000 Ljubljana Tdnidad, West Ilndies el: (43 1)512-47-31-0 Rosenoerms A11611 H-tt38Budapest Eastem Book Service Tel: (4722)97-4500 Tel: (386 61) 1330347: 132 1230 Tel: (809)662-5654 BANGLADESH -1DO 970 Frederiksberg C Tel: (36 1)111 6061 3-13 Hongo 3-chome, Bunkyo-ku MicroAINGLADusES evlpet Tel: (45 31)351942 Tokyo 113 PAKISTAN SOUTH AFRICA, BOTSWANA UGANDA Micro Industries Development INDIA Tel: (81 3) 3819-0661 Mirza Book Agency For single titles: Gustro LId. Assistance Society (MIDAS) ECUADOR Allied Publishers Ltd. 65, Shahrah-e-Quaid-e-Azam Oxford Unversity Press PO Box 9997, Madhvani Building House 5, Road 16 Libn Mundi 751 Mount Road KENYA Lahore 54000 Southem Attica Plot 1674 Jinja Rd. Dhanmonda R1 Area Libreria Intemacional Madras -600 002 Afsca Book Service (EA.) Ltd. Tel: (92 42) 735 3601 Vasco Boulevard, Goodwood Kampala Dhaka 1209 RPO.Boxt7-01-3029 Tel: (91 44)852-3938 QuaranHouse,Mfangano Street PO. Box 12119 Tel: (256 41) 251 476 Tel: (800 2) 326427 Juan Leon Mets 851 P0. Box 45245 Oxiord University Press Ni City 7463 Quito INDONESIA Nairobi 5 Bangalore Town Cape Town UNITED KINGDOM ED LU y Tel: (5932) 521-606; (593 2) 544-185 Pi. Indira Limited Tel: (2542)223641 SharaeFaisal, PO Box 13033 Tel: (27 21)595 4400 Microinfo Ltd. Jean De Lannoy Jalan Borobudur 20 Karachi-75350 PO. Eox 3, Ahon, Hampshire GU34 2PG Av, du Rol 202 EGYPT,ARABREPUBLICOF P.O. BOX 19 KOREA, REPUBLIC OF Tel: (9221) 446307 Fot subscrotion orders: England 1060 Brussels Al AhramnDistnbulion Agency Jakarta 10320 Daejon Trading Co, Ltd. lntemational Subscription Service Tel: (441420)86848 Tel:(322)538-5169 Al Galaa Street Tel: (62 21) 390-4290 PO. Box 34, Youida, 706 Seoun Bldg Pak Book Corporation PO. Box 41095 Cairo 44-6 Youido-Dong, Yeongchengpo-Ku Aziz Chambers 21, Queen's Road Craighall VENEZUELA BRAZIL c T Tel: (20 2) 578-6083 IRAN Seoul Lahore Johannesburg 2024 Tecni-Ciencia Libros, SA. PubbicacoesTecnicas Intemaconais Ketab Sara Co. Publishers Tel: (82 2) 785-1631/4 Tel: (9242) 636 3222; 636 08a5 Tel: (27 11)88O-1448 CentroCuidad ComercialTamanco Ltda. The Middle East Observer Khaled Eslamboli Ave., 6th Street Nivel C2, Caracas RuaPeixoltoGomide, 209 41,SherilStreel Delafrooz Alley No. 8 MALAYSIA PERU SPAIN Tel: (582)9595547; 5035;0016 01409 Sao Paulo, S. Caro P.. Box 15745-733 University ol Malaya Cooperative EditorialIDesarnMloSA 4Mundi-Prensa Libros, S.A. Tel: (5511)259-6644 Tel: (20 2)393-9732 Tehran 15117 Bookshop, Limned Aparlado 3824, Limau1 Caslelb 37 ZAMBIA Tel: (98 21) 8717819;8716104 PO. Box 1127 Tel: (S114) 285380 28001 Madrid University Bookshop, UniSversiyol CANADA FINLAND Jalan Pantai Baru Tel:(341)431-3399 Zamba RenSuf Publishing Co. Ltd. Akateemeven Kiriakauppa Kowkab Publishers 59700 Kuaa Lumpur PHILIPPINES Great East Road Campus 5369 Canotek Road PO. Box 128 PO. Box 19575-511 Tel: (60 3) 756-5000 Intemalional Booksource Center Inc. Mundi-Prensa Barcelona Po. Box 32379 Otawa, Ontaon KtJ 93 FIN-00101 Helsinki Tehan 15117 1127-A Antipolo St, Barangay, Consell de Cent, 391 Lusaka Tel: (613) 745-2665 Tel: (3580)121 4418 Tel: (98 21) 258-3723 MEXICO Venezuela 08009 Barcelona Tel: (260 1)252 576 CHINA INFOTEC Makati City Tel: (34 3) 488-3492 CHioa Financial & Economic FRANCE IRELAND Av. San Femando No. 37 Tel: (63 2) 896 6501; 6505; 6507 ZIMBABWE CiaFnnil& oc Wodd Bank Publications Govemment Supplies Agency Col. Tonello Guerra SRI LANKA, THE MALDIVES Longman Zimbabwe (Pvt.)Ud. Publishing House 66, avenue dIlna Odig an ISoldthair 14050 Mexico, D.F POLAND Lake House Bookshop Tourle Road, Ardbennie 8, Da Fo Si Dong Jie 75116 Pads 4-5 Harcousl Road Tet: (52 5) 624-2800 Ileratonal Publishing Service 100, Sir Chaamplarm Gaudiner PO, DoX STt25 Beijing Tel: (33 1)40-69-30-56/57 Dublin 2 Ul. Piekna 31/37 Maweatha Southenron Tel: (86 10) 6333-8257 Tel: (353 1) 661-3111 00-677 Wareawa Colombo 2 Harare Tel:(482)628-089 Tel: (941)32105 Tel: (2634)6216617 sWims Coming in the next issue of THE WORLD BANK ECONOMIC REVIEW January 1998 Volume 12, Number I * Technological Capability and Firm Efficiency by Bee-Yan Aw and Geeta Batra * The Tragedy of the Commons in Cote d'Ivoire Agriculture: Empirical Evidence and Implications for Evaluating Trade Policies by Ramon Lopez * Foreign-Aid Fungibility: A Panel Data Analysis by Tarhani Feyzioglu, Viniaya Swaroop, and Min Zhu * Does Mercosur's Trade Performance Raise Concerns about the Effects of Regional Trade Arrangements? by Alexander Yeats