LS)Ps 3C)3 POLICY RESEARCH WORKING PAPER 3 031 Imports, Entry, and Competition Law as Market Disciplines Hiau Looi Kee Bernard Hoekman The World Bank Development Research Group Trade F COP April 2003 U POLICY RESEARCH WORKING PAPER 3031 Abstract Since the early 1990s numerous countries have adopted competition law is insignificant. However, once or strengthened competition legislation. Kee and alLowance is made for the endogeneity of both domestic Hoekman investigate the impact of competition law on competnton (number of firms) and the adoption of a industry markups over time and across a large number of competition law, the authors find that competition laws countries. They find both domestic and foreign have an indirect effect on equilibrium markups by competition to be major sources of market discipline in promoting a larger number of domestic firms. concentrated markets, but that the direct effect of This paper-a product of Trade, Development Research Group-is part of a larger effort in the group to study the links between trade and competition policies. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Paulina Flewitt, room MC3-333, telephone 202-473-2724, fax 202-522-1159, email address pflewitt@worldbank.org. Policy Research Working Papers are also posted on the Web at http:// econ.worldbank.org. The authors may be contacted at hlkee@worldbank.org or bhoekman@worldbank.org. April 2003. (30 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quijckly, even if the presentations are less than fully poleshed The papers carry the names of the authors and should be cited accordingly The findings, interprctations, and concltusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent Produced by the Research Advisory Staff Imports, Entry and Competition Law as Market Disciplines' Hiau Looi Kee2 and Bernard Hoekman3 JEL classzfication: F12, F13, Lll, L44 Keywords: Competition law, import penetration, domestic entry, industry markup. 1 The authors thank Mary Amiti, Simon Evenett, Caroline Freund, Peter Holmes, Marcelo Olarreaga, Caglor Ozden and David Tarr for comments and feedback. The findings, interpretations and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. 2 Development Economics Research Group, The World Bank, 1818 H Street N.W., Washington, DC 20433, USA. E-mail: hlkee0worldbank.org. 3 Development'Research Group, the World Bank, 1818 H Street N.W., Washington, DC 20433, USA, and CEPR, London, UK. E-mail: bhoekmna©nworldbank.org. 1. Introduction An important focus of discussions in international fora such as the WTO and the OECD is whether and how to expand the reach of competition legislation across members, strengthen cooperation between national agencies and explore the scope for the adoption of common norms. For example, multilateral dialogue in the WTO is currently focusing on an agreement that members abide by 'core principles' - nondiscrimination, national treatment and trans- parency (due process) - as well as, possibly, provisions banning 'hard core' cartels (Hoekman and Mavroidis, 2003). Relatively little cross-country empirical work has been done to identify the effect of com- petition law on the contestability of markets. For OECD countries this is an interesting research question, but not of major policy significance given that competition enforcement is long established in the main jurisdictions (US, EU, Canada, Japan). For these countries the policy questions revolve around issues such as the appropriate approach to dealing with vertical restraints and merger control. In many developing countries, however, the question is much more fundamental and centers on determining the maglitude of the net benefit of competition law. Adoption of such mechanisms is costly, requiring the allocation of skilled lawyers and economists that are in scarce supply. It may well be that a larger 'bang for the buck' can be obtained through less administratively costly policy measures to increase competition on markets. Of course, trade economists have long argued that trade liberaliza- tion is a powerful and administratively very simple way of enhancing competition (Bhagwati, 1968)." Other economists have emphasized the importance of removing government created 4 It is often argued that an open trade regime is a powerful instrument to discipline the behavior of firms which have market power. The empirical literature investigating the impact of import competition on the pricing behavior of domestic firms has concluded that trade liberalization forces firms to set prices closer to marginal costs. That is, there is a negative relationship between price-cost margins (markups) and the openness of the economy. Indeed, Levinsohn (1993), Harrison (1994), Grether (1996) and Djankov and Hoekman (2000) all find some support for the hypothesis that imports are a source of market discipline in 2 barriers to entry and exit for firms (Djankov et al, 2002). This paper attempts to determine the relative impact of competition law on competition outcomes. Specifically, we investigate empirically the contribution of competition law relative to alternative types of policies that enhance the contestability of markets, in particular import competition and measures to ease entry and exit of firms. FRom a policy perspective this analysis is relevant both in terms of informing decision makers on the relative importance of alternative national mechanisms to promote competition, and in terms of identifying where the priority areas for action may lie in terms of international cooperation. We develop a simple model where the markup by an industry of price over marginal cost is positively related to the size of domestic sales and negatively related to the number of domestic firms that are active in the industry, the magnitude of imports and the demand elasticity of the industry. Industry markups are estimated as a semi-translog function of the number of firms, value of imports and domestic production, using country, year and industry fixed effects to control for the demand elasticity of the industries. Having determined the impact of foreign and domestic competition (the number of domestic firms in an industry), a dummy variable approach is used to capture the impact of the introduction of a competition law. The results of cross-industry, cross-country, time series regressions using a sample of 28 industries, 42 countries and 18 years indicate that controlling for import competition and the number of firms in each industry/country, competition law has no direct impact on industry markups. These results suggest that from a competition viewpoint, policy priority should be given to measures that directly increase competition on markets - such as trade liberalization or the removal of entry barriers. However, once we control for the endogeneity of competition studies of domestic firns behavior in Turkey, the Ivory Coast, Mexico, and Bulgaria respectively. For reviews of the literature, see Levinsohn (1996), Roberts and Tybout (1996) and Evenett, Lehmann, and Steil (2000). 3 law adoption, we also find that industries that operate under a competition law tend to have a larger number of domestic firms, suggesting that in the long run, competition laws may have an indirect effect on domestic industry markups by promoting entry. This paper is organized as follows. We discuss our empirical model in Section 2, and present the data set in Section 3. Section 4 shows the estimation results, and Section 5 concludes the paper with some policy discussions. 2. Model One of the biggest difficulties in studying the effectiveness of competition law on industries is to define a measurable outcome variable. Given that the main objective of any competition law is to promote competition, variables that capture the level of competition, or the market power of firms in the industries are natural candidates. One such variable is the markup of price over marginal cost of production by firms in an industry.5 In perfect competition, price equals marginal cost, so that the equilibrium markup equals one. When firms have some market power, so that price is greater than marginal cost, we observe markups that are greater than one in equilibrium. Thus, in principle, the markup of price over marginal cost provide a simple way to measure the level of competition. However, in practice, given that marginal cost is itself not a well measured variable, the use of markups as a measure of competition has been limited. Hall (1988) developed a simple way to estimate industry markup from the production function of firms. Relaxing the neoclassical assumptions of perfect competition and constant returns to scale, Hall showed that by estimating the parameters of a production function, we can interpret the coefficient associated with the weighted growth rate of labor as the FX Other such variables include total revenue over total cost (Roberts and Tybout, 1996) and entry threshold (Bresnahan and Reiss, 1991). 4 implied equilibrium markup. Based on his model, Levinsohn (1993) and Harrison (1994) showed that trade liberalization is associated with lower industry markups in Turkey and Cote D'Ivoire, respectively. Subsequent papers by Norrbin (1993), Roeger (1995), and Basu and Fernald (1997) update Hall's approach to account for the usage of intermediate input and returns to scale. To address endogeneity issues of Hall regressions which are also common across most production function estimations, Olley and Pakes (1996) use a polynomial of capital and investment as a control for the unobserved productivity. They show that such a nonparametric correction is successful in reducing the upward bias on the labor coefficient, without using instrumental variables that may be questionable. This is the approach taken by this paper. In addition, we incorporate an industry markup function, derived from a short-run sym- metric Cournot equilibrium, into a Hall-type regression in this paper. In a short-run sym- metric Cournot equilibrium (that is, not allowing for entry), industry markups depend on the number of domestic firms, the share of imports in the domestic market, and the mag- nitude of total domestic sales. By introducing the industry markup function into the Hall regression directly, we are able to interpret the estimated coefficients associated with the re- sulting interaction terms between these variables and the weighted growth rate of labor per capital as the marginal effects of these variables on industry markups. In other words, we are able to directly estimate the effects of domestic and foreign competition on industry markups without explicitly estimating the industry markups themselves. This allows us to avoid some econometric complications arising from the use of an estimated dependent variable, and at the same time improve the efficiency and degrees of freedom of the estimations. In the empirical model developed below the role of competition law is twofold. In the short-run, given a fixed number of domestic firms and import penetration, the introduction 5 of a competition law can be regarded as a structural change in the economy that may lower industry markups directly by shifting down the industry markup functions. However, in the long-run, when firms are free to enter and exit, a competition law may affect the number of domestic firms by enhancing the contestability of markets through facilitating entry (via enforcement of provisions regarding restrictive business practices, the abuse of dominant positions, the potential for creating such dominance through mergers, etc.). It is important to recognize that countries will have different incentives to adopt competition laws, depending on the competition environment that prevails in their industries. In the long-run, both the number of firms and the adoption of competition law is endogenous. This is taken into account in our estimations. 2.1 Hall Regression For each country, let the output of industry i in period t be characterized by a production function of labor input, L,t, and capital input, Ktt, Qit = A.tFi (Lt, (1) Differentiating Equation (1) with respect to time and dividing both sides by qt yields the growth rate version of Equation (1):6 Q= A,t + aiLL%t + CeiKKit, where (2) Lt aFi nd Kit F3 F,t aL,t= F tK (3) are the elasticity of output with respect to labor and capital inputs, respectively.7 ; Here we adopt the convention to denote the growth rate of a variable with -aInX, 1 ax6 6 xU & 6 For each industry i, assume that the production function Fi is homogeneous of degree Si. F, shows increasing, constant, or decreasing returns to scale with respect to all inputs when St is greater than, equal to, or less than unity. Subtracting the growth rate of the capital input from both sides of Equation (2) and applying Euler's theorem for homogeneous functions, we can re-express Equation (2) as:8 qit = A,t + Ck,Litt + (Si - 1) K,t, (5) with the convention that x = X, (i.e., small caps express variables in per unit of capital terms). Let Pit = C X (6) be the price over marginal cost markup of industry i, and let OttL be the share of labor in total revenue. Given that a%L = AitO.ttL, Equation (5) becomes qit = Alt + J.t (0,JtLit) + (S, - 1) K,t (7) Equation (7) can form the basis for estimation of industry markups by regressing the growth rate of value added per unit of capital on the weighted growth rate of labor per unit of capital and the growth rate of capital.9 As suggested by Basu and Fernald (1995), one may be concerned regarding empirical analyses that use the growth rate of real value added instead of the growth rate of real output, given that due to the construction of value-added statistics, the growth rate of real value added will not be independent of the growth rate of intermediate inputs if the market is not perfectly competitive (even when production functions are weakly separable). However, the UNIDO industry level data set only provides real output for a few countries. Thus, due to data constraints, we have to rely on real value added data rather than real output data. 8 According to Euler's theorem, if a production F, (L,t, Kit) is homogeneous of degree S, with respect to its inputs, then Ct,L + Ct2K = Si (4) 9 Note that while maintaining the assumption that Cf,L and atiK are parameters of the production function of industry i that are constant over time, we allow industry markup, 1,L, and labor share, 0.tL to vary. This is consistent with the empirical data, as we observe some fluctuation in OitL from year to year. 7 Complications arise when using Equation (7) due to the fact that productivity growth, A,t, is unobservable. It is crucial to control for A,t since it enters the firm's first-order conditions for profit maximization that determine both input demand and output supply. Not controlling for A,t will bias upward the least squares estimates for the coefficients of the growth rate of labor per unit of capital and the growth rate of capital - a classical endogeneity problem. Olley and Pakes (1996) develop an empirical strategy to control for the endogeneity prob- lem. They introduce a polynomial of capital and investment as a control for the unobserved productivity. They assume that at the beginning of every period, firms know their pro- ductivity but this is not observable by the researcher. Based on the realized productivity, firms decide to stay in business or to exit. Providing that all surviving firms have positive investment, their investment can then be used as a control for productivity.'0 In other words, Olley and Pakes assume that firms with higher investment are those that realize higher productivity growth. They show that by introducing a polynomial of investment and capital stock as a control for productivity in the estimation of the production function, the upward bias on the coefficient of labor input is reduced." Given that our analysis centers on industry markups (that is, the coefficient on labor input), getting consistent estimates of the labor coefficients is critical. We therefore adopt the Olley and Pakes correction and use a polynomial of capital and investment to control for the unobserved industry productivity growth. l(Levinsohn and Petrin (1999, 2000) show that instead of investment, intermediate input could be a good instrument for productivity growth, especially for those firms that stay in business but do not have positive investment every year. "1Olley and Pakes (1996) also discuss bias on the capital coefficient due to entry and exit behavior of firms, and use a selection model to control for it. 8 2.2 Markup Fumction Leaving out time and industry subscripts for ease of exposition, assume that for each industry, domestic and foreign firms are Cournot players in the domestic market for a homogenous goods. Given homogeneity we assume further that there is a world market for the good and that imports can be characterized as being provided by one importing foreign firm, with a share in the domestic market of m. There are N identical domestic firms. Domestic firms face a positive fixed cost of entry, F, associated with government imposed entry and/or exit regulations of the type documented by Djankov et al (2002). Taking the quantity produced by other firms, Q-,, as given, each domestic firm n chooses its output by maximizing its profits: 7r* (N, F, m) -=max I 7rn (Qnt Q-n,) = P (Q) Qn,- C (Qn,)- F}, Vn = ,,N, where p(Q) is the inverse demand function, Q = QD + QM =ZZ Qn + QM, QD iS total domestic production (sales), and QM is the import quantity. (For simplicity, throughout what follows we assume that domestic firms do not export, so that domestic production equals domestic sales. In the empirical analysis below we take into account exports by domestic firms in the calculation of import market shares.) The first order condition for profit maximization implies: p(Q) [1_1 Q-] ( (Qn) ,(8) where e-- (OQ/ap) (p/Q) > 0 is the price elasticity of demand. In a symmetric Cournot equilibrium, total domestic production is given by Q*, = N Qn. Rearranging Equation (8) according to the definition of the markup in Equation (6) yields 1 ( Q- EN QD+QM~) 9 Thus, the equilibrium markup in the domestic industry is inversely related to the magnitude of imports and the number of firms, while it is positively related to size of domestic produc- tion." In addition, given the homogenous good assumption, the equilibrium price set by the domestic firms and the importer is identical, which means that the quantity ratio equals the volume ratio: Q P *QD 1 QD + Q P*QD* + P*Q 1T+ mI where m = PQf denotes the ratio of imports to domestic sales of the industry. Thus, given a fixed demand elasticity, industry markups are lower when there are more domestic firms and when the ratio of irnports to domestic sales is larger: 11 (6,N,m) = 11 1 (9) c N 1+m How effective are domestic and import competition in reducing industry markup, given the industry demand elasticity? Figure 1 plots industry markup against number of firms, given a hypothetical demand elasticity of 2 and an import ratio of 0.3. When there is only one domestic firm, the industry markup is about 1.6. The markup falls rapidly as the number of domestic firms increases, falling below 1.1 once there are more than 5 firms in the industry. With 30 firms or more, entry by an additional firm has only a negligible impact on the industry markup. Similarly, Figure 2 plots industry markup against the ratio of imports to domestic sales, given a hypothetical demand elasticity of 2 and 10 firms in the industry. We again observe that imports reduce industry markups at a declining rate. Moving from zero imports to an import volume that equals sales by domestic firms, markups drop from 1.05 to 1.025. The markup of the industry drops below 1.1 once imports are more than 4 times the value of total domestic sales. 12See Jacquemin (1982) for a similar derivation. 10 Figure 1: Industry Markup vs. Number of Firms (e = 2 & m = 0.3) 1.2: 1.15 1.1 1.05- 1 -4-0' '2b 40 n 60 eb ido Figure 2: Industry Markup vs. Imports (e = 2 & N= 10) 1.05 1.04\ 1.03- 1.02\ 1.01 .... . .... ...1 b' i ' 5 j 4 *- Figure 3 plots industry markup against the ratio of imports to domestic sales and the number of domestic firms, given a hypothetical demand elasticity of 2. Figure 3 clearly mustrates how import competition can act as a substitute for domestic competition when there are a limited number of domestic firms in the industry. Figure 3: Industry Markup vs. Number of Firms and Imports (E = 2) *1 Markup\ Summing up, both domestic and import competition have larger effects on industry markup when the existing industry markup is higher, as can be expected in highliy con- centrated markets. If an existing industry is already near perfect competition, additional domestic firms (entry) or imports would have only minimal impact on markups. These find- ings are consistent with those of Bresnahan and Reiss (1991), where markets are shown to approximate a competitive one when there are 5 or more firms. 12 Equation (9) implies that industry markup is a non-linear function of the number of domestic firms and the share of imports in domestic sales. We therefore approximate the non-linear relationship with a second order semi-translog function: i4C(N, m; Z) = 0 + 1c + 1t + ±i + INN (ln Ntc)2 +3Nln NitC +13MM (In mit.)2 + 3M ln m,tc +I3NM ln NitC In mrtC + g(Z), (10) where the demand elasticity of industry i in period t of country c is assumed to be the sum of country, time and industry fixed effects and matrix Z represents a set of controls which may help reduce the bias of the estimated coefficients. Variables in Z include the total value of domestic output and labor cost in each industry, GDP and GDP per capita of the economy. 2.3 Short-Run Empirical Model We incorporate the industry markup function into the Hall regression by substituting Equa- tion (10) into Equation (7): qite = Cc+ C, + ct + P5 (kttc, Iitc) + [13 + 8 + /3t +±+ i + fNN (ln NitC) +,AN ln NtC +OMM (lnmitc) +/Mlnm,tC + /NMlnNitclnm,tc +g(Z)j (OitLJt) (11) where P5 (Kitc>I tc) denotes the 5th order polynomial of capital and investment, and to- gether with country-industry fixed effects and year fixed effects, are the controls for industry productivity growth. Equation (11) shows that the coefficients of the interaction terms between the arguments of the semi-translog function and the weighted growth rate of labor, 0.ttLit, can be interpreted as the marginal effects of those factors on industry markups. For example, the effect of a one percent increase in import penetration on industry markup and the effect of a one percent 13 increase in the number of domestic firms on industry markup would be respectively t9gt (N, m; Z) _ ai 8 - a = dM ± 2MM in m19t ± /iNM in 92t a In mt,1 a In mit, A930tLida In mt., Dgt (N m; Z) a = ON + 23NN In Nt. + /NM In m,ti. a9 In Nt, a9 In Nt a90itLl,9 In Ntc~ Thus, assuming that our application of the Olley and Pakes (1996) methodology is successful in correcting for the endogeneity of the regression errors, and that the current value of imports and number of firms are exogenous, the estimated coefficients of the interaction terms will be unbiased estimates of the effects of import penetration and domestic entry on industry markups. The theoretical model is considered to be not rejected by the data if the estimated values of both OiM and OiN are negative, while /3MM, fiNN, and /3NM are positive. Finally, it is clear that even in the short-run, import penetration is likely to be en- dogenous. When there is a positive productivity shock or a favorable endowment shock, output of the industry may increase and reduce the import demand of that industry, given a fixed consumption pattern. While we already use the 5th degree polynomial of capital and investment to control for industry specific productivity shock, we have yet to control for endowment shocks. Following Tlefler (1993) and Goldberg and Maggi (1999), we model the import penetration ratio as a function of factor shares in each industry/country, in addition to all the exogenous variables of the model (W): In m,tC = 6 + 6 + ft + Si + 5fc factor _share,t, + h (W) (12) Together, Equations (11) and (12) form a system of two equations to be simultaneously estimated to determine the short-run effects of imports and domestic competition (number of firms) on the level of industry markups. Given this framework, we then consider the introduction of a competition law as a struc- tural change to the economy, one that is expected to enhance competition between domestic 14 firms and reduce industry markups without specifying the channel. To test for whether com-. petition law has this effect, we introduce a dummy variable, Dtc, in the markup equation, Lt, (N, m; Z), which equals one if there is a competition law in the country in a given year. In other words, we run the following system of panel regressions: qitc = Cc + Ci + Ct + P (kitci fitc) + [P/DDIC + 16 + 13c + 1t +i + +NN (ln NtC) +/N In Nttc + /MM (In mtc)2 + 8M In mttc + INM In Nstc In mitc +g(Z)] (Oitdit) (13) lnmt + =S+c+at + 6t + fCfactor _shareitc +h(W). (14) If the introduction of competition law has an effect on industry markup, we would expect the coefficients of the competition law dummy to be negative and statistically significant. 2.4 Long-Run Equilibrium and the Role of Competition Laws In the short-run, it is reasonable to assume that the number of firms is fixed, and thus is exogenous to, among other variables, the industry markups and the policy environment. However, in the longer run, the number of firms is endogenous, a function of the profitability of the industry as well as the ease of entry. We capture the latter by a fixed cost, F in the model. Specifically, given a fixed cost of entry and the prevailing import share, the equilibrium number of firms is determined by the condition that the profit obtained by an additional firm is smaller than the fixed cost of entry: N* (F, m) = arg max {O, 7r* (N, F, m) }. As the number of domestic firms is a discrete variable, in equilibrium it is possible for all of the N* existing firms to make a (small) positive profit. 15 A major role of competition law is to enhance the contestability of markets. Competition laws may affect the number of domestic firms in the long run by prohibiting anti-competitive behavior that raises entry costs. This suggests that the existence of a competition law should, cetems pa7ibus, lead to a higher number of domestic firms. To the extent that the number of domestic firms affects industry markup, competition law would then indirectly affect markups.13 On the other hand, in the long-run (that is, allowing for entry and exit), there may also be a country 'self-selection' effect in that the adoption of competition law is a function of the overall level of competition prevailing in an economy. Specifically, if the industry import penetration ratio is high, or there is a large enough number of domestic firms in each industry, the need to establish a competition law is less, controlling for the stage of development and size of the countries. Conversely, if there is a small number of domestic firms, it may be politically more difficult for countries to set up a competition law as the incentive for firms to lobby against such a law will be higher. Overall, the impact of domestic market structure on the probability for countries to adopt a competition law is therefore an empirical question. Both entry and the decision by governments to adopt competition laws are endogenous in the long-run. Estimating the long-run effect of competition law on domestic industry markup via its effects on the number of domestic firms thus requires estimation of a self-selection model, where domestic entry depends in part on the existence of competition law, conditioned on countries developing a competition law. Specifically, we test the following two-step model '.'We recognize that in practice the enforcement of competition law may be such as to raise the costs of entry for new efficient entrants. This type of capture of competition enforcement by incumbents, as well as enforcement mistakes that conclude that active price competition is predatory, would lead to an opposite conclusion. The empirical analysis that follows can be seen as providing a test of the hypothesized effect of competition law on the number of firms in an industry. 16 for the long-run equilibrium: In Nit+lc = YDDtc + YAtc + f (X) (15) Dt = { ° if D (Nttc, tc; ) 0 where Atc is the estimated hazard rate of the country adopting a competition law in year t, based on the first step selection model which specifies the decision rules of the government. Without controlling for A, the estimated treatment effect of adopting a competition law, _YD, is likely to be biased and inconsistent. Note that the dependent variable of the second step regression, Equation (15), is the one period lead value of the number of domestic firms. This captures the effect of a competition law on domestic entry, since in the short-run, entry- related fixed costs are likely to keep the current period number of firms constant. On the other hand, Equation (16) specifies the decision making process of the government, which depends on the average industry characteristics, such as the current number of firms and import penetration. Both X and w are the matrix of control variables in the two-step model. 3. Data Our data set comprises 28 industries in 42 developed and developing countries for 18 years (1981-1998). Industries are defined at the 3 digit level of the International Standard Indus- trial Classification (ISIC). The total number of observations in the data set is only 11,484, due to missing values for either industries, countries or years. Industry level production and trade data (exports and imports) are obtained from UNIDO and the Trade and Production Database compiled by Nicita and Olarreaga (2002).'4 We utilize World Bank (2002) for country level data on variables such as GDP and GDP per capita. Information on the ex- istence and year of adoption of competition legislation is drawn from national sources and 14As noted earlier, we use export data to calculate import shares in total sales. 17 the OECD. Table 1 reports sample averages of the key variables used in the regressions by country. Countries and industries in our data set vary quite significantly. On average, each indus- try in each country has some 1,500 firms, ranging from less than 50 firms on average in each industry in Panama to more than 15,000 firms on average in a Japanese industry. The size of industries also varies substantially across countries - with average sales of US$70 million, Cyprus has the smallest average industry size, while at nearly US$90 billion, industries in the U.S. are the largest. Import competition is weakest in Japan, where the ratio of imports to domestic produc- tion for a typical industry is only 6 percent, compared to 80 percent for a typical industry in Hong Kong. On average, the ratio of imports to domestic output is around 18 percent in the sample. In terms of the year in which competition law was first passed, Canada and U.S. have long-standing enforcement dating back to the turn of last century, while Egypt, Hong Kong and Singapore have yet to adopt any form of competition law. Nineteen countries in the data set had a competition law prior to 1981. A number of developing and transitional economies subsequently adopted competition laws during the sample period. 'rable 1 also presents data on GDP and GDP per capita for the countries in the data set. Here again there is substantial variation. The largest country in the sample, the U.S., is more than 1000 times larger than Jordan, the smallest country in the sample. On the other hand, the richest country in the sample is Japan, with a per capita GDP more than 100 times greater than the poorest countries (India and Kenya). Table 2 provides sample averages of the main variables used by industry. The food in- dustry has the largest average number of firms, with an average of more than 5200 firms in each country. On the other hand, with an average of only 33 firms, the petroleum refining 18 industry is the most concentrated. The largest industry in the sample is the transport equip- ment industry, with average sales of more than US$20 billion. The pottery and earthware industry lies at the other end of the spectrum with average sales of only US$550 million per year. Firms in the petroleum industry on average are the largest in size - the typical firm has average sales of US$275 million a year. On the other hand, the footloose apparel industry has the smallest average firm size - only about US$1 million on average. In terms of total volume of imports (trade), machinery and transport equipment indus- tries rank first. However, judging by the ratio of imports to domestic production, leather products, scientific instruments and miscellaneous manufactures face the most intense import competition. With an import ratio of only 4 percent, the printing and publishing industry faces the least import competition. Given the heterogeneity of the countries and industries in the data set, it is clearly important to control for country and industry specific effects in the estimation of the industry markup function. We also include year specific effects to control for any general movement of international prices and development trends. 4. Results Table 3 presents the regression results for the short-run case, where the number of domestic firms is assumed to be fixed. The dependent variable is the growth rate of real industry value added relative to the capital stock, 4it,. The top part of Table 3 presents the estimated semi-translog function, itc (estc, Nitc,)mitc), as defined in Equation (10) . Column (1) shows the baseline ordinary least squares regression using the full sample, without a competition law dummy variable. The estimated first order effect of imports on industry markup is negative and significant; the same holds for the estimated coefficient on 19 domestic sales. The second order effects of these variables do not seem to matter in the full sample. On the other hand, we do not find the effect of the number of domestic firms on industry markups to be significant. This is not surprising, as in the full sample the average number of firms per industry is around 1,500. Thus, an additional firm in an industry should not have any significant effect. As shown in Figure 1 and Bresnahan and Reiss (1991), entry by new firns should have the greatest effects on competition when the existing market is highly concentrated. We will come back to this point later. The lower part of Table 3 reports the variables that pertain to the industry production function, including a-5th order polynomial of capital and investment to control of industry productivity growth and a full set of industry, year and country fixed effects. The estimated coefficients are not reported due to space limitations, but are available upon request. The competition law dummy variable representing the structural change due to the adop- tion of a competition law is introduced in Column (2). Not only is the dummy variable not significant, it is clear that adding the competition dummy does not change the previous result. In other words, we do not observe a significant effect of the competition law dummy variable on industry markups in the full sample - the primary policy-determined variable is foreign competition. Given that the impact of both foreign and domestic competition on markups is most likely to be important in concentrated markets, the next two columns of Table 3 report the results of a three-stage least squares regression for a subset of industries with high concentration, specifically, those with no more than 30 domestic firms. Both the production and import penetration functions are simultaneously estimated. Column (3) reports the results of the production function estimation, and Column (3') reports those for the import penetration function. 20 Column (3) indicates that for this sub-sample both imports and the number of domestic firms have a statistically significant effect in reducing industry markups. In fact, the marginal effect of a 10 percent increase in the import share in domestic sales is quantitatively equivalent to the marginal effect of an additional domestic entrant when there are only 9 firms in the market. In other words, the regression results suggest that both foreign and domestic competition are important in reducing domestic market power. Domestic sales and labor costs are also significant determinants of industry markups - a larger domestic market or lower cost of production are associated with higher industry markups. Column (3') reveals that it is important to control for the endogeneity of import penetra- tion (Equation (14) above). Factors that are negatively correlated with import penetration include the number of domestic firms and their sales, and the size of the overall economy. The effects of domestic endowments, proxied by the industry factor shares are also included in the regression, but coefficient estimates are not reported in the Table due to space limitations. The effects of adding the competition law dummy to the system of equations are reported in column (4). Once again, we find that after controlling for the effects of domestic and foreign competition, the direct effect of competition law is not statistically significant, although it has the right sign, even if the analysis is restricted to more concentrated industries with less than 30 firms. Table 4 presents results for the long-run equilibrium case, where both competition law and the entry and exit of domestic firms are endogenous. The number of domestic firms in an industry is responsive to, among other factors, the (fixed) cost of entry, which in turn is affected by the competition law of the countries. On the other hand, the government's decision to adopt a competition law may depend on the average level of competition in domestic industries, as well as the stage of development of the economy (proxied by GDP and 21 GDP per capita). This suggests that the 'treatment effect' of competition law on domestic competition could be underestimated if we do not control for the self-selection bias.15 T'o correct for the endogeneity of competition law, we use a two-step procedure developed by Heckman (1979). Specifically, assume that for any period, a country's decision to adopt or abandon a competition law depends on the perceived level of industry markups, which are affected by the current level of imports, total domestic output, and total number of firms in the industry. We then first estimate a selection model by regressing the competition law dummy on average industry imports, domestic sales, and the number of firms, controlling for level of GDP and GDP per capita of the countries (Equation (16) above). We use the predict probability to construct the hazard rate, Atc, which wil then be included in our entry regression (Equation (15)). We expect the estimated hazard rate to be negatively correlated with domestic entry, since the larger the number of domestic firms, the less likely the government would need to adopt a competition law to promote entry. Once the self- selection bias is controlled for by the estimated hazard rate, we expect the competition law dummy variable to have a positive effect in promoting entry (that is, the future (t+1) number of firms in each industry). Table 4 presents the results of the two-step selection model, where Column (1) shows the estimated effects of competition law and the self-selection bias on domestic entry, and Column (2) shows the estimated decision rule of governments in adopting competition laws. The regression results suggest self-selection bias is indeed important. Correcting for this bias, industries operating under a competition law tend to have a larger number of domestic I ISpecifically, while some countries may choose to pass and keep a competition law as a response to current high industry markups, others may not need such a law given low industry markups. This would lead to a contemporaneous positive correlation between industry markup and the status of the competition law. Not controlling for such a self-selection bias will lead to under-estimation of the effectiveness of competition law in reducing industry markups. In other words, given that we expect the treatment effect of competition law on industry markup to be negative, the least squares estimate would have an upward bias; if the bias is large enough it could result in a positive estimate. 22 firms - on average 7.2 percent more. Moreover, countries' decisions to adopt competition laws do appear to be associated with variables that reflect the average level of competition in industries. For example, countries that have a higher level of import penetration are less likely to adopt competition laws, while countries that have a higher level of domestic sales in a typical industry are more likely to adopt competition laws. The results in Tables 3 and 4 together suggest that while competition law may not have a direct effect on industry markups, even in more concentrated markets, it may affect industry markups in the long run via its effects on domestic entry and thus the long run equilibrium number of domestic firms. 5. Conclusion For competition law to be a priority, it must yield a higher pay-off in terms of fostering competition than other policy options. The analysis in this paper suggests that dealing with trade barriers and govermnent regulation that restrict domestic competition by impeding entry and exit by firms will generate a higher rate of return than the adoption of a competition law. Indeed, the regression results obtained here suggest that the direct effect of competition law on industry markup is not significant, even if the analysis is limited to the sub-sample of more concentrated industries. However, once account is taken of the endogeneity of competition law adoption, we find that competition laws have an effect on entry by domestic firms, which may indirectly affect the long run level of industry competition (markups).'6 Any assessment of whether and how to adopt antitrust disciplines must of course con- sider factors that have been ignored in this paper. One such factor relates to the costs of '6Some authors have found that competition policy paradoxically reduces the number of firms-see e.g., Bittlingmayer (1985)-because prohibitions on price fixing and similar arrangements among firms encourage mergers. Our cross-country results suggest that while such incentives may exist the overall effect is the opposite. 23 enforcement. Import liberalization not only has a more powerful and direct effect on compe- tition, it also is a lower cost policy alternative, especially in the long run given no recurrent administrative enforcement and compliance costs. Another factor not considered here are possible international externalities associated with the enforcement (or non-enforcement) of antitrust by foreign countries. It must also be recognized that the analysis has been limited to industries producing tradable goods. Many products are non-tradable (e.g., many ser- vices). Even if tradable, competition may be limited to local markets for other reasons (e.g., transport costs). Certain products may be produced by (natural) monopolies or by firms where 'unnatural' (government-made) barriers to entry restrict contestability. In determin- ing whether to make the adoption and enforcement of competition law a domestic priority, a wider focus will be required. However, it should also be recognized that in many cases competition law may not be the appropriate instrument to deal with such issues either. For example, in the case of services it may well be the case that the impact of government poli- cies that restrict competition in services dominate (or are a major element allowing) private restrictive business practices. 24 1REFER1ENCES Basu, Susanto, and John G. Fernald (1997), "Returns to Scale in U.S. Production: Estimates and Implications," Journal of Political Economy 105, no. 2, 249-283. Bittlingmayer, George (1985), "Did Antitrust Policy Cause the Great Merger Wave?" Journal of Law and Economics 28, 77-118. Bresnahan, Timothy F., and Peter C. Reiss (1991), "Entry and Competition in Con- centrated Markets," Journal of Political Economy 99, no. 5, 977-1009. Bhagwati, Jagdish N. (1968), T7e Theory and Practice of Commercial Policy. Prince- ton, N.J.: Princeton University Press. Djankov, Simeon, and Bernard Hoekman (2000), "Market Disciplines and Corporate Efficiency: Evidence from Bulgaria" Canadian Journal of Economics, 33, 190-212. Djankov, Simeon, Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Shleifer (2002), "The regulation of entry", Quarterly Journal of Economics. Evenett, Simon, Alexander Lehmann, and Benn Steil (2000), Antitrust policy in an evolving global marketplace, Brookings Institution Press, Washington, DC. Grether, J. (1996) "Mexico, 1985-1990: Trade Liberalization, Market Structure, and Manufacturing Performance," in M. Roberts and J. Tybout (eds.), Industrial Evolution in Developing Countries, Oxford: Oxford University Press. Goldberg, Pinelopi K., and Giovanni Maggi (1999), "Protection for Sale: An Empirical Investigation," American Economic Revzew 89, no. 5, 1135-1155. Hall, Robert E. (1988), "The Relation between Price and Marginal cost in U.S. Indus- try." Journal of Political Economy, vol. 96, no. 5, p. 921-947. Harrison, Ann E. (1994), "Productivity, Imperfect Competition and Trade Reform: Theory and Evidence," Journal of International Economics, vol. 36, p. 53-73. Heckman, James J. (1979), "Sample Selection Bias as a Specification Error," Econo- metrica, vol. 47, no. 1, 153-161. Hoekman, Bernard, and Petros Mavroidis (2003), "Economic Development, Competi- tion Policy and the WTO", Journal of World Trade, February. Hoekman, Bernard, Hiau Looi Kee, and Marcelo Olarreaga (2001), "Markups, Entry Regulation and Trade: Does Country Size Matter?," Policy Research Working Paper. 2662, The World Bank. Levinsohn, James (1993), "Testing the imports-as-market-discipline hypothesis", Jour- nal of International Economics 35, 1-22. Levinsohn, James (1996), "Competition policy and international trade policy", in J. Bhagwati and R. Hudec, eds. Fair Trade and Harmonization: prerequisites for free trade?, MIT Press, Boston. 25 Levinsohn, James, and Amil Petrin (1999). "When Industries Become More Productive, Do Firms? Investigating Productivity Dynamics." NBER Working Paper, no. 6893. Levinsohn, James, and Amil Petrin (2000). "Estimating Production Functions Using Inputs to Control for Unobservables." NBER Working Paper, no. 7819. Nicita, Allessandro, and Marcelo Olarreaga (2002), "Trade and Production, 1976-99," in B. Hoekman, A. Mattoo and P. English (eds.), Development, T1rade and the WTO: A Handbook, World Bank. Norrbin, Stefan C (1993), "The Relationship between Price and Marginal Cost in US Industries: A Contradiction," Journal of Polztical Economy 101, no. 6, 1149-1164. Olley, 0. Steven, and Ariel Pakes (1996). "The Dynamics of Productivity in the Telecommunications Equipment Industry." Econometrica, vol. 64, no. 6, p. 1263-1297. Roberts, Mark, and James Tybout (1996), Industrial evolution in developing countries, Oxford University Press. Roeger, Werner (1995). "Can Imperfect Competition Explain the Difference between Primal and Dual Productivity Measures? Estimates for U.S. Manufacturing." Journal of Political Economy 103, no. 2, 316-330. World Bank (2002), World Development Indicators. 26 Table 1: Data at a Glance by Country, 1981-1998 Numbers of Value in Million of US$ US$ Comp. Law Growth Rates (%) of Domestic Country Sample' Firms Imports Production GDP GDPPC Passed in Y/K L/K K I Australia 138 1008 742 2830 248000 16146 1906 -0.34 -1.28 0.41 -1.09 Austria 392 313 1120 2380 195300 25362 1951 -1.88 -2.67 3.19 -0.02 Bulgaria 60 577 86.9 286 12570 1485 1991 -47.46 -19.31 44.69 57.70 Canada 276 1325 2660 8260 468100 17841 1889 -2.93 -2.50 4.56 2.70 Chile 376 55 265 839 42260 3180 1959 -1.02 -1.53 6.91 6.20 Colombia 402 271 213 798 69710 2050 1959 -0.59 -0.63 3.92 -3.31 Cyprus 335 278 74.2 70 6049 8978 1989 -4.01 -2.20 6.56 -2.46 Denrark 297 381 707 1300 150000 29267 1937 -0.31 -0.61 2.22 3.08 Egypt 366 268 268 786 45400 907 No Law 4.12 -0.34 3.85 5.13 Finland 454 337 577 2020 116500 23419 1958 -0.96 -1.93 1.69 1.80 Greece 336 294 407 797 102200 10210 1977 1.35 -0.69 -0.08 -3.69 Hong Kong 350 2042 6660 1610 106700 18265 No Law -8.99 -6.23 4.58 -10.30 Hungary 351 199 27.5 876 48150 4588 1990 -3.23 -2.23 0.19 -3.38 India 462 3846 553 4410 271600 320 1969 -5.36 -4.53 12.66 8.78 Indonesia 365 570 626 1430 129600 737 1999 -6.28 -3.47 23.83 15.18 Ireland 257 194 412 801 40790 11605 1991 -2.05 -2.62 2.43 -0.37 Italy 330 1266 3820 13300 959700 16919 1990 -4.86 -3.37 5.34 1.03 Japan 471 15044 4830 73500 4452000 36206 1947 -1.37 -1.27 2.85 2.90 Jordan 325 424 99.7 109 5339 1603 2000 1.19 -0.51 6.29 -7.59 Kenya 91 198 97.7 480 9283 333 1988 41.42 -19.26 43.90 16.96 Korea 494 2397 1930 8490 333900 7704 1980 -1.89 -2.81 9.96 8.27 Mexico 252 111 679 1870 246900 3180 1992 -4.06 -1.95 8.35 1.01 Morocco 209 282 217 586 34130 1317 1999 0.29 -0.56 4.50 3.10 Netherlands 208 327 3420 5150 340700 22943 1957 -1.59 -2.96 3.99 4.59 New Zealand 80 843 385 974 51500 15370 1986 -5.14 -3.03 0.36 -14.09 Norway 433 283 761 1530 122200 28823 1926 2.34 -1.11 -0.01 -1.14 Pakistan 82 163 183 421 39590 402 1970 -25.27 -8.38 36.04 -8.25 Panama 146 45 51.4 89.2 6784 2733 1996 -6.91 -4.28 7.36 9.28 Peru 295 522 82.4 498 48420 2394 1991 -2.01 0.26 -1.63 -19.62 Poland 190 216 318 1870 105900 2800 1990 -25.60 -5.65 12.15 -1.09 Portugal 241 488 322 725 77670 7799 1983 -5.07 -2.52 4.26 0.88 Romania 17 656 145 716 30250 1320 1991 -64.40 -18.42 43.11 18.83 Singapore 406 156 2120 1430 54550 19848 No Law -3.31 -3.00 6.67 5.01 Spain 386 5380 1800 6900 477000 12340 1963 -6.62 -3.14 6.12 3.96 Sri Lanka 253 294 93.9 99.9 10030 593 1987 -2.01 -2.34 13.07 -0.17 Sweden 168 383 953 2530 197400 23641 1953 0.32 -1.30 0.68 0.82 Thailand 77 537 1430 3570 128900 2258 1979 0.50 -5.42 38.78 -13.69 Turkey 196 274 608 2610 148500 2607 1994 -22.39 -6.75 30.18 6.09 United Kingdom 417 5026 5740 16900 969600 16881 1948 -1.79 -2.13 0.98 0.62 United States 84 13952 12400 88100 5716000 23456 1890 2.14 -0.68 1.16 -0.15 Venezuela 416 360 303 1030 66710 3508 1991 -1.59 -1.01 3.98 -1.34 Average 11484 1502 1419 6414 406973 10520 1970 -7.33 -3.76 10.00 2.25 Notes: ' Denotes the total number of observations for each country. Unless otherwise stated all numbers denote simple averages across industries and years for each county. 27 Table 2: Data at a Glance by Industry, 1981-1998 ISIC Numbers of Value in Million of US$ Growth Rates (%) of I ~~~~~~Domestic Industry Descrnption Samplel Firms Imports Production Y/K L/K K I 311 Food 455 5244 2660 18800 -1.39 -1.53 6.16 2.61 313 Beverages 417 508 307 3700 -3.24 -1.67 6.86 0.87 314 Tobacco 402 383 222 1850 -2.47 -1.67 7.47 1.29 321 Textiles 455 3414 1970 6620 -3.32 -2.30 2.98 -1.85 322 Apparel 442 2702 1150 3030 -3.32 -2.81 7.01 2.36 323 Leatherproducts 405 441 503 702 -3.53 -2.41 3.28 -2.18 324 Footwear 422 426 318 798 -5.73 -3.93 5.70 -2.61 331 Wood products 426 2826 696 3180 -3.22 -1.99 4.43 -1.93 332 Fumiture . 427 1655 254 1980 -3.17 -2.38 6.53 1.99 341 Paper and products 453 892 974 5480 4.17 -2.52 6.66 -0.28 342 Pnnting and publishing 450 3337 249 6650 4.49 -3.39 9.07 5.37 351 Industrialchemicals 394 414 2960 8580 -1.16 -1.47 5.20 0.05 352 Otherchemicals 405 829 1340 7560 -3.05 -2.36 8.63 4.62 353 Petroleumrefinenes 311 33 1400 9020 -6.31 -2.35 7.88 -1.31 354 Petroleumandcoalproducts *238 174 112 1090 -5.11 -2.45 8.26 2.26 355 Rubberproducts 435 412 308 1890 4.02 -2.54 5.13 2.29 356 Plasticproducts 437 1763 774 5050 4.02 -2.30 9.34 5.31 361 Potteryandearthenware 356 346 82.4 550 -5.28 -3.98 6.91 -1.20 362 Glass and products 376 233 223 1240 -5.95 -3.71 8.55 0.74 369 Non-metallicminmralproducts 411 1935 241 4650 -3.77 -2.21 7.20 2.63 371 Ironandsteel 374 764 1360 9870 -3.51 -2.59 3.55 -2.02 372 Non-ferrous metals 371 425 1280 3940 -3.14 -2.13 5.43 2.03 381- Fabricated metal products 451 4757 1380 9280 -3.68 -2.46 6.42 1.02 382 Machineryexceptelectncal 418 4903 5640 19100 -1.97 -2.39 7.51 1.08 383 Electricalmachmery 451 2415 5340 18500 -3.81 -2.75 7.20 3.59 384 Transportequipment 451 1374 4660 20500 -3.25 -3.53 6.82 3.27 385 Scientificequipment 411 707 1650 2820 -3.96 -3.00 8.90 3.02 390 Othermanufacturedproducts 440 1530 1220 2100 -5.33 -3.31 7.98 2.28 Notes: Denotes the total number of observations in each industry. Unless otherwise stated all numbers denote simple averages across countries and years of each industry 28 Table 3: Regression Results OLS OLS 3SLS 3SLS Explanatory Variables (in log) (1) (2) (3) (3') (4) (4') Competition law 0.420 -0.146 (0.244) (0.201) Imports -0.223°° -0.232** -0.139°° -0.136°° (0.107) (0.107) (0064) (0.064) Imports squared -0.002 -0.002 (0.009) (0.009) Firms 0.372 0 370 -0.138*** -0.544*** -0 1400 .(0 543000 (0.213) (0213) (0052) (0.149) (0.052) (0.149) Firms squared -0.025 -0.025 0 006*** 0.159°°° 0.0060** 0.159°°° (0.018) (0.018) (0002) (0.032) (0.002) (0.032) Firms * Imports 0.030 0.032 (0.018) (0.018) Sales -1.056°* -1.133** 2 304*** -2 692*** 2 320*** -2.692*0* (0.486) (0.482) (0.657) (0.341) (0.657) (0.341) Sales squared 0.024 00260* -0.0760** 0.051P * 0.077*** 0051 ** (0.013) (0.013) (0.018) (0010) (0.018) (0.010) Labor cost -0.258 -0 597 -3 3750°¢ 0 427 -3.458 0 428 (0.959) (0.983) (0.999) (0.418) (1.006) (0.418) Labor cost squared 0.015 0.034 0.199°°° -0.046° 0 2050** -0.046° (0 058) (0.060) (0.056) (0.025) (0.057) (0.025) GDP 3 054** 3.790** -2 939*0* -2.940°°° (I 051) (1.496) (0.610) (0.610) GDPPC -1.298 -I 878 4.142°°° 4.142°°° (I 504) (1.502) (0.552) (0.552) Constant -54.393*s -66.330°° -1.126 71 319*0* -0.950 71.342°0° (27.209) (26.882) (7.185) (12.344) (7.188) (12.344) Factor shares Yes Yes Industry Fixed Effects Yes Yes Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Yes Yes Country Fixed Effects Yes Yes Yes Yes Yes Yes Polynomial (K,l) 5th order 5th order 5th order 5th order Industry Fixed Effects Yes Yes Yes Yes Year Fixed Effects Yes Yes Yes Yes Country Fixed Effects Yes Yes Yes Yes Sample Size 10447 10447 1488 1488 1488 1488 R-squares 0.4219 0.4222 0.3499 0.9601 0.3501 0.9601 notes: Column (I) and (2) are performed using the full sample. Column (3) and (4) only use data on the highly concentrated markets (finns<="30). Both sets of 3SLS estimating a system of 2 equations on industry markups and import penetration. Column (31) and (4') report results of the import penetration regression where factor shares are used as controls for endowments and productivity differences. 29 Table 4: Selection Models Dependent Variable: Firns in next period Dependent Variable: Competition Law (1) (2) Competition law 0.072** Average imports -1.185*** (0 017) (0.055) Lambda (hazard rate) -0.066*** Average imports squared -0.128*** (0.009) (0.008) Imports -0.002 Average firms -0.201 (0.002) (0.488) Sales 0.023*** Average firms squared -0.067 (0.005) (0.047) Labor cost -0.025** Average lagged firms 4.392*** (0.010) (0.536) GDP -0.525*** Average lagged firms squared -0-334*** (0.074) (0.052) GDPPC 0.311*** Average sales 2.814** (0.078) (1.260) Firms 0.891*** Average sales squared -0.096*** (0.010) (0.031) Lag firms 0.066*** Competition agencyl 4.317*** (0.010) (0.132) Year 0.015*** GDP -4.349** (0.002) (1.743) Constant -18.363*** GDP squares 0.101*** (4.342) (0.035) GDPPC -8.553*** (0.379) GDPPC squared 0.576*** Industry Fixed Effects Yes (0.023) Year Fixed Effects Yes Country Fixed Effects Yes Year 0.127*** Sample Size 8277 (0.008) Notes: Heckman two step selection model is performed. Column (2) reports the competition law selection model and Column (1) reports the domestic entry model, controlling for selection bias (lambda) estimated from Column (2 lDummy variable which equals one if the country has a competition agency in that year. 30 Policy Research Working Paper Series Contact Title Author Date for paper WPS3011 Renegotiation of Concession J Luis Guasch April 2003 J Troncoso Contracts in Latin America Jean-Jacques Laffont 37826 Stephane Straub WPS3012 Just-in-Case Inventories: A Cross- J Luis Guasch April 2003 J Troncoso Country Analysis Joseph Kogan 37826 WPS3013 Land Sales and Rental Markets in Klaus Deininger April 2003 M Fernandez Transition Evidence from Rural Songqing Jin 33766 Vietnam WPS3014 Evaluation of Financial Liberalization Xavier Gin6 April 2003 K. Labrie A General Equilibrium Model with Robert M Townsend 31001 Constrained Occupation Choice WPS3015 Off and Running? Technology, Trade, Carolina Sanchez-Paramo April 2003 H Sladovich and the Rising Demand for Skilled Norbert Schady 37698 Workers in Latin America WPS3016 Partisan Politics and Stuti Khemani April 2003 H Sladovich Intergovernmental Transfers in India 37698 WPS3017 Why is Unemployment so High Jan Rutkowski April 2003 J Rutkowski in Bulgaria? 84569 WPS3018 Integrating the Least Developed Paul Brenton Apnil2003 L Tabada Countries into the World Trading 36896 System The Current impact of EU Preferences under Everything but Arms WPS3019 Foreign Aid, Conditionality, and Ghost Thilak Ranaweera April 2003 P Rathan Raj of the Financing Gap A Forgotten 33705 Aspect of the Aid Debate WPS3020 Consumption, Health, Gender, Anne Case April 2003 D Ballantyne and Poverty Angus Deaton 87198 WPS3021 Institutional Arrangements for Public Elizabeth Currie April 2003 M Rosenquist Debt Management Jean-Jacques Dethier 82602 Eriko Togo WPS3022 Sending Farmers Back to School. Gershon Feder Apnl 2003 P Kokila The Impact of Farmer Field Schools Rinku Murgai 33716 in Indonesia Jaime B Quizon WPS3023 Productivity Growth and Product Douglas M Addison April 2003 D Addison Variety Gains from Imitation and 31188 Education WPS3024 R&D and Development Daniel Lederman April 2003 P Soto William F Maloney 37892 WPS3025 Trade Structure and Growth Daniel Lederman April 2003 P Soto William F Maloney 37892 WPS3026 Strategic Approaches to Science Robert Watson Apnl 2003 R Watson and Technology in Development Michael Crawford 36965 Sara Farley Policy Research Working Paper Series Contact Title Author Date for paper WPS3027 Financial Intermediation and Growth: Genevieve Boyreau- April 2003 P. Sintim-Aboagye Chinese Style Debray 38526 WPS3028 Does a Country Need a Promotion Jacques Morisset April 2003 M. Feghali Agency to Attract Foreign Direct 36177 Investment? A Small Analytical Model Applied to 58 Countries WPS3029 Who Benefits and How Much? How Alessandro Nicita April 2003 P. Flewitt Gender Affects Welfare Impacts of a Susan Razzaz 32724 Booming Textile Industry WPS3030 The Impact of Bank Regulations, As11 Demirguc-Kunt April 2003 A Yaptenco Concentration, and Institutions on Luc Laeven 31823 Bank Margins Ross Levine