POLICY RESEARCH WORKING PAPER 2162 Quantifying the F iscal A general equilibrium tax model estimated for -(0 Effects of Trade R1{eform countries provides a s mple but rigorous method for Sbantayanan Devarajan estimating the fiscal impact of trade reform Delfin S. Go Hongyz Li The World Bank Development Research Group Public Economics U August 1999 POLICY RESEARCH WORKING PAPER 2162 Summary findings Using a tax model of an open economy, Devarajan, Go, For imports to grow and tariff collection to and Li provide a simple but rigorous method for compensate for the tax cut, the import elasticity has to estimating the fiscal impact of trade reform. be high. Because of the balance of trade constraint, Both the direction and the magnitude of the fiscal however, imports cannot substitute for domestic goods consequences of trade reform depend on the elasticities unless supply is able to switch toward exports. Hence, of substitution and transformation between foreign and the export transformation elasticity has to be high as domestic goods, so they provide empirical estimates of well. those elasticities. As substitution possibilities between foreign and They also discuss the implications of their analysis for domestic goods increase, a tariff reform can theoretically public revenue. be self-financing. But if the elasticities are less than In general, they find that it matters what the values of "large," tax revenue will fall with tariff reduction and the two elasticities are relative to each other. If only one further fiscal adjustments will be necessary. of the elasticities is low (close to zero), revenue will drop Devarajan, Go, and Li provide empirical estimates of unequivocally as a result of tariff reform, reaching close the possible range of values for the elasticities of about to the maximum drop whether or not the other elasticity 60 countries, using various approaches. The elasticities is high. range from 0 to only 3 in most cases - nowhere near the point at which tariff reform can be self-financing. This paper - a product of Public Economics, Development Research Group - is part of a larger effort in the group to develop and apply tools to analyze fiscal reform. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Hedy Sladovich, room MC2-609, telephone 202-473-7698, fax 202-522- 1154, Internet address hsladovich@worldbank.org. Policy Research Working Papers are also posted on the Web at http:/ /www.worldbank.org/html'dec/Publications/Workpapers/home.html. The authors may be contacted at sdevarajan @worldbank.org, dgo@worldbank.org. August 1999. (53 pages) The Policy Research Working Paper Series dissemintates 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 quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do noy necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Produced by the Policy Research Dissemination Center Quantifying the fiscal effects of trade reform: A general equilibrium model estimated for 60 countries Shantayanan Devarajan, The World Bank Delfin S. Go, The World Bank Hongyi Li, Chinese University of Hong Kong* We thank Will Martin and DoDiinique van der Mensbrugghe for several helpful comments. 1. Introduction Despite compelling evidence of its many benefits, trade liberalization remains an unfinished business in many parts of the world, particularly Sub-Saharan Africa. One reason is that many developing countries today are still dependent on import tariffs for revenue. Governments fear trade reform will lead to significant revenue losses in the short run. In Sub-Saharan Africa trade taxes account for 27 percent of total revenue of governments.1 For some countries-C6te d'Ivoire, The Gam- bia, Madagascar, Mali, Mauritius, Niger, Sao Tome and Principe, and Swaziland, for example-the dependence on trade taxes is higher than 40 percent. A number of authors (Branson et al. (1992), Mitra (1992), and Greenaway and Milner (1991)) have therefore emphasized the importance of concomitant fiscal adjustment to make trade reform sustainable. But what should the size of the fiscal adjustment be? What determines the magnitude of the fiscal impact? How can it be minimized? Will the liberalization ever be self-financing? Answers to these questions typically require an elaborate tax model with its healthy, not to say rude, appetite for data. As a result, policymakers rarely have the answers when they need them. Trade reform is then either not undertaken, or done so with little knowledge of its fiscal consequences. 'Excluding South Africa and Nigeria; based on recent data (1992) from the World Bank's African Development Indicators 1996. 1 The purpose of this paper is to present the simplest structure of an open econ- omy that provides a quick but clear method of quantifying the fiscal consequences of trade reform. The model can be solved analytically and we provide intuitive formulae for understanding the fiscal effects of tariff changes. In addition, we provide empirical estimates of the model's key parameters for about 60 countries. The fiscal impact of a tariff reduction will depend directly on the size of the tariff cut, the response of imports to the tax change, and the relative importance of import tariffs as a source of government revenue. It will also depend indirectly on what happens to the other tax bases and how they in turn will affect revenue. The key to revenue performance, therefore, is how all the tax bases will change with the reform. To estimate the direct and indirect fiscal consequences, economists have often employed general-equilibrium tax models. These tax models can be quite complicated and difficult to build, particularly in view of the data constraints in many developing countries. Their complexity also makes it hard to sort out the relative importance of various factors.2 To seek an easier but still rigorous alternative, to better understand how tax models work, and to identify what affects public revenue, this paper takes a simple analytic representation of a large class of empirical general-equilibrium models, a prototypical framework that has 2A recent, extensive discussion of the structure of applied general equilibrium models may be found in Ginsburg and Keyzer (1997). Past surveys are also found in Robinson (1989) or Shoven and Whalley (1984). 2 been shown in Devarajan et al. (1993, 1997) to anticipate many of the significant results of trade-focused general-equilibrium models. In this paper, we extend the framework to analyze taxes and their fiscal and welfare impact. The goal is not to capture the detailed economic and tax structure of a particular country, nor to arrive at precise estimates of the direct and indirect fiscal and welfare consequences of trade reform, but rather to isolate key parameters and ascertain their empirical magnitudes, and to provide a quick but clear way to quantify the fiscal consequences.3 The rest of the paper is divided into three sections. Section 2 discusses the specifications of the simple open2-economy framework and its applications towards macro-fiscal policy. Key analytical results are derived. Section 3 provides esti- mates of these parameters for different countries and describe estimation issues and problems. A summary of conclusions follows in Section 4. 3This framework can be numerically implemented using widely available data from national and fiscal accounts and in a user-friendly spreadsheet format (Devarajan, Go, Lewis, Robinson, and Sinko (1997)). It has also been adapted to look at the fiscal impact of a regional custom union, see Devarajan, Go, Suthiwart-Narueput, and Voss (1997), as well as the dynamic impact of trade and macro policy, see Devarajan and Go (1998). 3 2. The model 2.1. The 1-2-3 model with taxation Total public revenue, R, is defined as the sum of revenue intakes R, from various taxes. Consider the two most important taxes in developing countries-an import tariff tm and a domestic indirect tax. The latter may be a tax on domestic spending, tq, which applies equally to the domestic goods and imports; or, it may be a tax on production of domestic goods, td, but not on the production of exports, in which case exports serve as an untaxed sector. That is, R =ERi = tm e 7rmM + tqPqQ + tdPdD (2.1) where irm is the world price, and M the quantity of imports; e is the foreign exchange rate; Pq is the price of domestic demand Q; and Pd is the producer price of domestic goods D. The model in which the tax instruments are considered is a simple general equilibrium model that has one country, two activities, and three goods, or "1-2- 3 model" (Devarajan et al. (1993, 1997)). The basic nature of the 1-2-3 model is a modified Salter-Swan methodology that separates the economy into three distinct goods: exports E, imports M, and a 'domestic' or nontraded goods D. The production of exports and domestic goods is defined by a transformation 4 process marked by a constant elasticity of transformation Q, X a (AL" + (1 - A)D-?) t (2.2) where Q = -i- is the elasticity of transformation and a is a shift parameter. Output X is exogenous and constant given fixed factor endowment. From profit- maximizing behavior, the relative amount of exports and domestic goods pro- duced is determined by their relative prices and Q: E (PE) (2.3) D Pf, where k is a constant equal to (19A)n ; PD is the tax-adjusted producer price PD(1 + td); given the world price of exports 7rE, pE = 7E e is their domestic price. Consumers' demand is satisfied by imports and domestic goods and is defined by a constant elasticity of substitution (CES) utility over the two goods Q = e (6M-P + (1 - 6)D-P) P (2.4) where a = is the elasticity of substitution and e is a CES shift parameter. The proportion of imports and domestic goods consumed is affected by their relative 5 prices and a from the following first-order condition: D (PM) (2.5) where k' is a constant. To buy Q, consumers receive income from pro- duction plus a lump-sum transfer from government, i.e., their budget constraint is PQ(1 + tq)Q = PxX + R (2.6) The expenditure tax tq that is imposed on Q applies equally to M and D. We assume the economy is in trade balance so that: tlM - 7teE = O (2.7) In addition, there are two price identities: PX peE+PdD (2.8) M 1D PQ Pm. Q +P (2.9) A few remarks are in order. The model implies imperfect substitution be- 6 tween domestic and foreign goods. Even if the 'law of one price' prevails and perfect substitution is found amDng specific commodities, the elasticities of the aggregated goods will still be less than infinity whenever aggregation requires dif- ferent weights because of preferences, endowment, and trade specialization.4 In more practical terms, the CET and CES formulations allow one to avoid corner solutions or complete specialization. Because they are well-defined and intuitive to use, the two functions are popular and serve as key relationships in more dis- aggregated models as well.5 In both the aggregated and disaggregated cases, it is obvious that the elasticities of the CES and CET functions will play key roles in the numerical outcomes of the models. The model requires a price numeraire and the foreign exchange rate e is used. There are eight variables (R, E, M, D, Q, Fd, FiX, FQ) in nine equations. But by Walras' law, one of the equation is redundant so that one equation may be dropped. 4Devarajan (1997) discusses the problem of defining tradable and non-tradable goods in the context of determining the real exchange rate misalignment in the CFA zone. 5With more disaggregation, domestic and foreign goods may be indexed over groups of com- modities, i.e., Ei, Mi, Di,where i =food, various manufactures etc. For each class of commodities, Qi, Xi are then defined by their respective CES and CET functions. Cobb-Douglas, a special case, is sometimes used. In the supply side, note that wi1h fixed factors, the single output X is also fixed. With disaggregation, resource reallocation from relative price shocks and interindustry purchases come into play; a production function fcr each category of output and the specification of factor markets are necessary. 7 2.2. Consequences of a tariff reform To derive the fiscal and welfare effects of a tariff reform, we first log-differentiate the system of equations (1)-(7), noting that (except for tm) the growth rates of other exogenous variables (X, e, rm, nre, td, tq) are zeroes:6 R=6m(tmn + M) + OR (p, + Q) + +OdR P+D)(.0 -? = R d+f)(2.10) where OR = R iS the relative weight of the tax revenue from source i and, hence, E = 1. Even in this simple set-up, it is clear that the fiscal impact of a tariff reduction depends not only on im, M, and kRn, but also on D, Q, Pd, P, and the revenue weights of domestic taxes Of"'s. The growth rate of X (= 0) and Q are expressed as 'weighted' averages of the growth rates of foreign and domestic goods: Ox= E + oxD (2.11) Q= oM + oD (2.12) where the various weights are defined by the relative importance of foreign and domestic goods and the CET and CES parameters: 6That , a a = 8 SX = AS(E )7 OQ= be(M)p 02= (1 - b)C-P (D)P Equation 2.12 is also the change in utility or welfare. The first-order conditions in turn are: E -D b - = Jd (2.13) M-D = a (Pd-WtJ.m) (2.14) In equation 2.14, wt, - is the price wedge created by t, The budget constraint yields Pq + Q (1 - OR)Pz + 6R (2.15) where O{ is the relative weiglat of the lump-sum tax transfer in household income. 9 The trade balance equation implies M=E (2.16) Finally, P x =pXt + (1 - PX)(d + ) (2.17) where Px = -PxI'E exports' share in GDP. 2.2.1. Impact on Pd, D, M, and E An important factor in the revenue impact is the change in the price of domestic good Pd, the key endogenous price of the model. Given that world prices are ex- ogenous, Pd determines the change in the real exchange rate, which in turn affect the allocation of the three goods in the model M, E,and D through the optimal conditions, 2.3 and 2.5. Through its effects on Pg and P:, Pd also influences the behavior of the domestic tax bases Q and X. A specific expression for Pd can be derived (from equations 2.13, 2.14, and 2.16):7 A 7For an expression of Pd in terms of possible terms-of-trade shocks and their implications, see Devarajan et. al. (1993). 10 Pd = + itm (2.18) All parameters in the coefficieni- ° are positive so that Pd varies directly with as expected; a rise (reduction) in tariff protection will raise (lower) the price of the import substitute. A precise expression for exports can also be obtained (from 2.11, 2.13, and 2.18): E =- i '1: t^m (2.19) 1+ /02 where P I and /d2 = E 8 Both 01 and P2 are positive so that im, working like an export tax, affects D negatively as expected. FRom the current account balance (2.16), the change in imports is the same as that of exports, i.e., M = E:. Solving for D, we note that the added protection afforded by a positive tm will encourage the production of domestic or 'non-traded' goods: 8From equation 2.11, define 02 = O . Using the usual budget equations for a CET function or the first-condition 2.3 to remove A from the O's, it can be shown that 62 = . Finally, initializing prices to 1 in the base year, 12 = D 11 31I02- b = +2 t+ (2.20) Because of its modified Salter-Swan formulation, the behavior of the 1-2-3 model is shown to be consistent with traditional trade theory. In addition, the impact on E, M, D, and Pd, depend on values of the trade elasticities a and Q, the size of the tax wedge wt., and the relative importance of trade 02. How the parameters affect revenue and welfare are examined next. 2.3. Revenue impact 2.3.1. Case 1: tm is the only tax Take the simplest case in which there are no domestic taxes, so that tq = 0, td =0, O, =0, O, o =0, and OR = 1. From 2.10, it is clear that revenue will depend on whether imports expand sufficiently to offset the cut in tariffs: f? = tm + M Substituting the previous solution for M, we find that 1= ki,m 12 where r1 (u +2 )(1 + 32) Can k, be negative? In other words, can there be a Laffer curve for tariffs? k1 is negative if O*Wtm~i > (a + Q!)(1 + 02) an (1> +2) (2.21) a +Q wt, For non zero exports (f2 -D > 0) and positive tariffs (0 < wtm < 1), note that D (1+12) > 1. Hence, the left-hand side, +Q, has to be sufficiently greater than 1, a condition that may be satisfied if both a and Q are high. How high should the elasticities be? Take some feasible numerical values, =2 2 and wLm = 0.2 (i.e., t,, = 0.25), Figure 2.1 shows the revenue impact for a 50 percent reduction in tarilfs for different values of a and Q. A few observations can be made. Even if just one of the elasticities is low (close to zero), revenue will decline unequivocally, reaching close to the maximum drop regardless whether the other elasticity is high. This is because for imports to grow and tariff collection to compensate for the tax cut, or has to be high. Because of the balance of trade constraint however, imports cannot substitute for domestic goods unless supply is able to switch towards exports. Hence, Q 13 A R 0 0~~~~~~~~~~~~ 40~~~~ R ' t w t<~~~~~40 Figure 2.1: tm is the only tax has to be high as well. In the example given, both a and Q have to be greater than 20 before revenue growth becomes really positive. Hence as substitution possibilities between foreign and domestic goods increase, a tariff reform may be self-financing. Conversely, if the elasticities are less than 'large', tax revenue will fall with tariff reduction and further fiscal adjustments are necessary. The precise value of a or Q is of course an empirical issue, which we will examine in section 3. Note further that the higher is the initial tariff (higher wtm) the more likely will the tariff reduction be self-financing. 14 2.3,2. Case 2: There is also a domestic expenditure tax tq What would happen to revenue if there were a domestic tax? Take the case in which the only domestic tax is an expenditure tax on Q, so that td =, OR = O and R R (tm" + M3) + (1 _ OR) ( )(.2 R=O 7n m + ~~~~~~~~~(2.22) There are two possible opposing effects on the domestic tax base - Pq will fall while Q may rise because of the cheaper imports. To investigate their net effects, note that (Pq + Q) is an average of P. and R in 2.15, and an expression for P, is available in 2.17. Using the previous solutions for Pd, D, M and E, which are unaffected by the introduction of tq, we are also able to derive R in terms of the policy change tm: f = k2tm and kz = Xi (2.23) X2 where OR.1 = P, (t-awt,sQa+ rSt'. ( ntg), Xi a j+~Ytn&+7t T I)/2) + 1 (awt (1 + 2 ( + Q) W 2 (1 0E) B26E) 15 and X2 = (1qOR9Y) (a + Q +Wtm a + Q ) X2 q Rk \o+W) /32) The denominator of the coefficient k2 is always positive, i.e., X2 > 0. The numer- ator X1 is generally positive unless a and Q are sufficiently high so that (-uWtmtQ) in the first part of the expression dominates. Substituting the same values of the parameters in Case I and further assuming that tariff revenue is now one third of total revenue, OR = 0.33 and that the tax rebate is 20 percent of income, OR = 0.33. Figure 2.2 shows that the revenue profile is the sanie as Case 1. Note that for low elasticities, the revenue decline is less than Case 1 since tariff revenue is now less important. In this case however, both elasticities now have to be a lot greater than 40 before the revenue impact is positive. The reason is that, although with a higher elasticity, the greater is the increase in M, the response of D goes in the opposite direction, rendering the effect on Q relatively neutral. Thus, policy reform generally has a negative net effect on the domestic tax base and consequently, imports have to expand a lot more to raise revenue. Nevertheless, like Case 1, a tariff reform can still finance itself with sufficiently high elasticities. 16 ARRO. 180 good u otoneprs. q- O,/- n Sub3 Cstitutin prevodusoltions for is, impoand on weufindothat -0.1 = OR 601_O) p 0k.i (2.25) 17 -0.2 kO.4 R 8 -0. -0.8 0 40 4 020/20 Figure 2.3: A tax on D but E untaxed where k3 = ( 1 + [(1(-IO)OWtm + 2 (a + Q Wt-mWe + (0 +rn2)] Here the result is more direct. If O6; < 0.5, that is, tariff revenue does not dominate total revenue, k3 is always positive, regardless of the values of a and Q. FRom equation 2.24 Pd and D, and thus revenue from the domestic tax base, will tend to fall as both ao and Ql rise. Retaining the parameter values in Case 2, we confirm this in Figure 2.3. Hence, unlike Case 1 and 2, public revenue will always fall and it will decrease more the higher the elasticities. 18 2.4. Welfare impact Using equations 2.19, 2.20, and 2.16 in the equation for Q (2.12), the growth rate in utility is derived as follows:9: Q k4m (2.26) where k4 (a + Q)(1 +2) ( 02a- ) a,,tn3fl is always positive. Using the familiar budget shares equations for a CES equation and initializing lprices to one, it can be shown that O Q ( ) (P..)M e (Q) e Likewise, OQ = (i) so that welfare increases if 2 <2) (D)( (1)" (E) D' 9Q,& Re may be interpretedi as a negative equivalent variation -EV as a ratio to GNP, since EV = Po(Qo - Q), Po = 1 and Qo is base-year GNP. 19 The intuition behind 2.27 is simple. If the curvature1o of the isoquant is suf- ficiently 'flat', a condition that is easily attainable if the armington function is elastic (a > 1), and if M > E (for example, when the current account is bal- anced), k4 will be negative and a tariff cut will lead to a welfare gain.11 Using the parameter values from the previous cases, we show that the welfare gain increases with the elasticities in Figure 2.4. 3. Empirical estimates of elasticities 3.1. Econometric issues As the previous section indicates, to be able to determine the direction and quan- tify the size of the fiscal impact of trade reform, we need estimates of the foreign trade elasticities . Although there is large literature on trade elasticities, the es- timation of trade elasticities generally applies to import demand. Exports from developing countries are modeled as import demand from importing industrial °0More precisely, the curvature of the CES function is aM _( -_6)_ AD 6 (V "1For any given set of elasticities, it is not clear whether Q >,e > C-2 > Qcase. In case 2, Q is initially less than optimal because of the tax tg so that increases in Q from a tariff cut might be higher; but this may be offset by a smaller income effect because of tax diversification. In case 3, a taxed D implies that it is underproduced. Shifting away from it towards the untaxed exports may not enhance welfare as much in a second-best sense. But equation 2.26 indicates that k4 may or may not be higher. Whatever the case, it is safe to say that Q will at least be higher for plausible values of the parameters. 20 0.06< A ) l / ~~~~~~~~~~~10 Figure 2.4: Welfare impact of tariff reform countries. Furthermore, in estimating import demand, while the imperfect sub- stitution model of Armington J1969) is the most prevalent specification, estimates of elasticities are usually in thLe form of demand price elasticities, not directly in terms of the substitution and transformation elasticities that appear in our model. A key issue in the empirical investigations is whether changes in the real exchange rate have a significant impact on the balance of trade. The litera- ture may be divided along how they answer the question. Most of the earlier literature, such as Branson (1.972) and Khan (1974),12 as well as others like Rit- tenberg (1986), and Bond(1987), and Marquez and McNeilly (1988) found that 12A survey can be found in Goldstein and Khan (1985). 21 trade flows respond significantly to changes in relative prices. However, they are today criticized for inference problems associated with time-series variables that have unit roots. Some recent empirical work that took into account the time-series properties of trade flows and prices, such as Rose (1990 and 1991) and Ostry and Rose (1992), found little evidence that relative prices matter in trade flows. The lack of theory in time-series techniques can of course create interpretation problems. Marquez (1994), for example, stressed the importance of optimizing behavior and simultaneity in determining the expenditures on do- mestic and foreign goods. For developing countries, Faini, Pritchett, and Clavijo (1988) discussed the importance of trade policy and restrictions, which are likely to understate the structural demand elasticities. One study by Reinhart (1995) uses time-series techniques and dynamic optimizing behavior; it finds significant trade relationships but that the aggregate price elasticities tend to be lower than unity. There are differences in findings at the disaggregated levels as well. Brenton and Winters (1992) avoid assuming separability between home and foreign goods and find low import price elasticities. Panagariya, Shah, and Mishra (1996), on the other hand, employing better data such as explicit competitors' prices (not proxies) find high elasticities. In any case, these elasticities apply to specific groups of commodities, i.e., not at the level of aggregation desired in our model. 22 Another issue is the assumpl,ion of homotheticity in the Armington function, which is violated by the time trends observed in trade shares. Import and export shares in GDP for many countries appear to be increasing and independent of the relative price movements. Alstcn et al. (1990), for example, note that while the implicit assumption of homotheticity in the CES and CET formulations is theo- retically appealing, it is also highly restrictive in CGE modeling. The standard correction is usually to employ a scale variable, such as an income term, to denote aggregate income activity. Alternative formulations like the almost ideal demand system (AIDS) or one of the flexible functional forms are often suggested.13 Using AIDS, Hanson, Robinson, and 'rokarick (1993) find the sectoral expenditure elas- ticities in United States are generally greater than one. Other AIDS estimations are found in Alston and Green (1990) and Shiells, Roland-Holst, and Reinert (1993). While it is certainly plausible that the capacity to import among countries rises with income, Petri (1984) and Ho and Jorgenson (1997) believe that the high income elasticities estimated are probably spurious. In fact, trade shares seem to be increasing over time for rich and poor countries alike, as would be the case with increasing globalization. A natural breakpoint is the 1970s when 13Another possibility is to use the CES version suggested by Brown and de Cani (1963), which permits a variable degree of homogeneity while retaining a constant elasticity of substitution. 23 wide-ranging changes in the international monetary and trading system were im- plemented. Even for large industrial countries like the United States, there is a sharp acceleration in the import share in the 1970s. For developing and transi- tional economies, periods of rapid economic and trade liberalization (particularly in the late 1980s) are crucial factors. Compared to the earlier periods of inward- orientation, changes in trade policy in the latter periods often lead to structural breaks in the trade shares. To account for the shifts in trade shares, a time trend or a function like the logistic curve is recommended. There are very few studies that investigate export supply explicitly. Diewert and Morrison (1988) employ a production-based approach originally developed in Kohli (1978) to obtain export supply and import demand. A recent cross-country estimation of export demand and price elasticities that account for nonstationar- ity is by Senhadji and Montegro (1998). Faini (1994) is one study that estimates transformation elasticities from a CET function directly and that takes into ac- count adjustment lags, factor prices, and the importance of capacity utilization. He finds that the CET elasticity to be less than one for Morroco but much greater than one for Turkey. In general, it would be difficult to replicate these studies for many countries without extensive micro data. Part of the problem is the measure- ment problems of factor accumulations and their returns. Another issue is the adjustment lags in supply that may require measurement of capacity utilization. 24 3.2. Estimation methods anad results In this paper, we employ a variety of methods to estimate our critical parameters, or and P. Tables 1 and 2 tabulate the Armington and CET elasticities estimated using these various methods. No single method provides uniformly good estimates for all countries in the sample. In the tables, we provide only those estimates that have positive (or correct) sign s and significant coefficients. Between the two elas- ticities, CET elasticities tended to be estimated more frequently with the wrong sign or insignificant coefficients. One problem is that the aggregate price indices of imports, exports, and domestic good tend to move in the same fashion for many countries, dominated by underlying inflation trends and nominal exchange rate depreciation. As a result, the real exchange rates or relative prices of exports and imports over the domestic good may be indistinguishable from one another statistically. The quantities of exports and imports also tended to move together in many countries, especially during periods with balance-of-payment problems; hence, the two functions may not be fully unidentified in many instances. In ad- dition, the CET elasticities miay be affected by possible adjustment lags in supply behavior that are difficult to handle without additional micro data. Nevertheless, there are sufficiently good estimates for both elasticities in the tables. In the next subsections, we briefly describe the econometric methods employed. 25 3.2.1. Simple OLS results Before considering other approaches, we estimate using OLS the two aggregate trade elasticities from the optimal conditions for import demand and export sup- ply (equation 2.3 and 2.5). By taking logarithms, the non-linear functions are transformed into log-linear form, from which the trade elasticities are easily re- covered: In(D pan(m (3.1) In (E) =a + Q ln (pde) (3.2) where a = ln( In )t) and or= ln ( i4 -)) are two constant terms. In general, the simple OLS estimation of the CET elasticities is the most problematic with the largest number of incorrect signs and insignificant coefficients. 3.2.2. Time trend, structural breaks, and scale factor To account for the changing ratios of foreign over domestic goods, we add a time trend to the OLS. We also test the hypothesis that there is a structural break or a non-linear acceleration in the time trend due to policy reform, policy reversal, or increasing globalization. In general, it will be difficult to decide on a specific 26 break point unless the timing of trade reform is known for specific countries. The following specification is used to test for possible structural changes InK(D = +t 7rit+ 72I(k)t + In (p +J v In (E) = a -t 6t + 62I(k)t + Qln (p) + u where I(k) is an indicator function which equals I for t > k and otherwise 0. The break point k is set to d'fferent values, i.e., every five years. This specifi- cation is also extended to test for structural breaks in the elasticities as well as combinations of a structural break in the mean and a structural break in the slope coefficients (trend or elasticities). The logistic trend employed by Ho and Jorgenson (1998) is another alternative but we believe it is probably better suited for industrial countries, where there is less chance of reversals of trend directions due to policy failures and reversals. For comparison, we also eEtimate the equations employing aggregate income as a scale variable. We employ the log of domestic GDP in the Armington function and the log of OECD GDP index in the CET function. The two income variables, which increase gradually, generally function and serve as time trends. Time trends 27 seem to work best in improving the CET estimates. 3.2.3. Primary products Given the plausible hypothesis that primary exports are not very responsive to relative prices in developing countries, the export data are adjusted to eliminate primary product export. Define PM = real primary product export. Then E* = E - PM. We replace E with E* where applicable. In general, this seems to improve the elasticities for a few countries. 3.2.4. Seemingly unrelated regressions Another method we used is that of seemingly unrelated regressions (SUR). SUR is designed to improve the efficiency in the variance estimation in cases when there is a correlation between the two sets of residuals. The parameter estimation may change, however, but not to a great extent. In our case, the signs most of the coefficients remain the same. Using subsamples after 1980 on the hypothesis that foreign trade is more responsive to prices during the widespread liberalization in eighties seems to lead to more positive CET estimates than using the full samples. However, the number of observations becomes very small and the estimates should be taken with great caution. 28 3.2.5. Other factors We also examine the premise that a country's resource balance in the balance of payment account or the relat'ive importance of the agriculture sector may af- fect export and import behavior by adding them as separate variables, with and without trend. A few of the estimation improve. 3.2.6. Simultaneity We also tried a number of approaches that did not give good results. Among them was simultaneous estimation of the export function and the import function using instrumental variables methods. We also tried using a reduced form equation for Pd from which the trade elasticities can be derived: ln(Pd) = c + oe ln(Pm) + lIn(PX) + ±ln(A) + u where a = , = , M/3 and 7 = 1. However, the coefficients are over- identified because we have three equations for two unknowns. Note that if we consider a + p 1 we can fuither write the above regression as In ( =In ( + ln() + u. 29 Based on the and estimated from the restricted regression, the elasticities Q = p -1 and C = + 1 can be uniquely determined. The results were not promising however. 3.2.7. Cointegration We also tried unit root test and cointegration regressions. We tried different model specifications, including different lags, different orders of time polynomial, etc. The general conclusion is that the unit root null cannot be rejected for most of the variables (each country has four time series, i.e., In (E), In ( I), ln (D) and ln (PEM,)). It should be noted however that proper cointegration analysis will require a deeper data analysis that is only possible with thorough examination of individual countries. Moreover, the absence of long time-series may increase the probability of spurious cointegration. On the other hand, because we are using annual data, longer time-series runs into the problem of structural change or breaks that are harder to handle in this methodology.14 Keeping in mind the caveats, the cointegration results are also presented. For the cointegration results of regressions 3.1 and 3.2, we tried the Johansen vector error-correction model (VECM) model (Johansen, 1988, 1991), the Phillips and Hansen fully-modified (FM) OLS model (Phillips and Hansen, 1990), and the 4See, for example, Gonzalo and Lee (1998), for a discussion of some of the pitfalls. 30 Park canonical cointegrating regressions(CCR) model (Park, 1992). However, not all results are consistent or improve with cointegration. Only the Johansen VECM estimation results are reported. 4. Conclusions In this paper, we present the simplest structure of an open economy that provides a quick but clear method of quantifying the fiscal consequences of trade reform. The model can be solved analytically and we provide intuitive formulae for un- derstanding the fiscal effects of tariff changes. We find in general that it matters what the values of the elasticites, a and Q, are relative to one another. Even if just one of the elasticities is low (close to zero), revenue will decline unequivocally from tariff reform, reaching close to the maximum drop regardless whether the other elasticity is high. This is because for imports to grow and tariff collection to compensate for the tax cut, the import elasticity a has to be high. Because of the balance of trade constraint however, imports cannot substitute for domes- tic goods unless supply is able to switch towards exports. Hence, the export transformation elasticity Q has to be high as well. In the example given, both a and Q have to be greater than 20 before revenue growth becomes really positive. Hence as substitution possibilities between foreign and domestic goods increase, a tariff reform can theoretically be self-financing. Conversely, if the elasticities 31 are less than 'large', tax revenue will fall with tariff reduction and further fiscal adjustments are necessary. The precise value of a or Q is an empirical issue and we provide empirical estimates of the possible range for about 60 countries using various approaches, including the Johansen method. These estimates can of course be refined with better data (e.g. export price indices) and better knowledge of the breakpoints and episodes of policy reforms and crisis. However, they do indicate that the elasticites only range from 0 to 3 in most cases, nowhere near the point at which tariff reform can be self-financing. References [1] Alston, J.M., C.A. Carter, R. Green, and D. Pick. 1990. Whither Armington trade models? American Journal of Agricultural Economics 72: 455-467. [2] Alston, J.M., and R. Green. 1990. Elasticities in AIDS models? American Journal of Agricultural Economics 72: 442-5. [3] Armington, P. 1969. A theory of demand for products distinguished by place of production. IMF Staff Papers 16: 159-78. [4] Bond, M. 1987. An econometric study of primary commodity exports from developing country regions to the world. IMF Staff Papers 34: 191-227. 32 [51 Branson, W. H. 1972. The trade effects of the 1971 currency realigmnents. Brookings Papers on Economic Activity I: 15-58. [6) Brenton, P.A and L.A. Winters. 1992. Trade Flows and Trade Policy After '1992'. Cambridge: Cambridge University Press. [7] Brown, M and J.S. de Cani. 1963. Technilogical change and the distribution of income. International Economic Review 4: 289-309. [8) Devarajan, S. 1997. Real exchange misalignment in the CFA Zone. Journal of African Economies 6: 35-53. [91 Devarajan, S. and D.S. Go. 1998. The simplest dynamic general equilibrium model of an open economiy. Journal of Policy Modeling 20(6): 677-714. [10] Devarajan, S., D.S. Go, J. Voss, and S. Suthiwart-Narueput. 1997. DiTect and indirect fiscal effects of the Euro-Mediterranean Free Trade Agreements. Paper presented at the Mediterranean Development Forum, Knowledge and Skills for Development in the Information Age, Marrakech, Morocco, May 12-17. [11] Devarajan, S., D.S. Go, J.D. Lewis, S. Robinson, and P. Sinko. 1993. Sim- ple general equilibrium modeling. In J.F. FRancois and K.A. Reinert, eds., 33 Applied Methods for Trade Policy Analysis - A Handbook. Cambrudge: Cam- bridge University Press. [12] Devarajan, S., J.D. Lewis, and S. Robinson. 1993. External shocks, pur- chasing power parity, and the equilibrium real exchange rate. World Bank Economic Review 7: 45-63. [13] Diewert, W.E. and C.J. Morrison. 1988. Export supply and import demand functions: a production theory approach. In R.C. Feenstra, ed., Empirical Methods for International Trade. Massachusetts: The MIT Press. [14] Faini, R. 1994. Export supply, capacity and relative prices. Journal of De- velopment Economics 45: 81-100. [15] Faini, R., L. Pritchett, and F. Clavijo. 1988. Import demand in developing countries. World Bank Policy Research Working Paper 122, Washington D.C. [16] Ginsburg, V. and M. Keyzer. 1997. The Structure of Applied General Equi- librium Models. Massachusetts: The MIT Press. [17J Goldstein, M. and M.S. Khan. 1985. Income and price effects in foreign trade. In R.W. Jones and P.B. Kenen, ed., Handbook of International Economics, vol. 2. Amsterdam: North-Holland. 34 [18] Gonzalo, J. and T.H. Lee (1998). Pitfalls in testing for long-run relationships. Journal of Econometrics 86(1): 129-545. [19] Greenaway, D. and C. Mihner. 1991. Fiscal dependence on trade taxes and trade policy reform. Journal of Development Studies 27: 95-132. [20] Hanson, K., S. Robinson, and S. Tokarick. 1993. U.S. adjustment in the 1990's: A CGE analysis of alternative trade strategies. International Eco- nomic Journal 7: 27-49. [21] Khan, M.S. 1974. Import and export demand in developing countries. MF Staff Papers 21: 678-93. [22] Kohli, U.J.R. 1978. A gross national product function and derived demand for imports and supply of exports. Canadian Journal of Economics 11(2): 167-82. [23] Marquez, J. 1994. The econometrics of elasticities or the elasticity of econo- metrics: an empirical analysis of the behavior of U.S. imports. Review oJ Economics and Statistics 76(3): 471-81. [241 Marquez, J. and C. McNeilly. 1988. Income and price elasticities for exports of developing countries. Review of Economics and Statistics 70: 306-14. 35 [25] Mitra, P. 1992. The coordinated reform of tariffs and indrect taxes. World Bank Research Observer 7: 195-218. [26] Johansen, S. 1988. Statistical analysis of cointegration vectors. Journal oj Economic Dynamics and Control 12: 231-255. [27] Ho, M.S. and D.W. Jorgenson. 1998. Modeling trade policies and U.S. growth: some methodological issues. In N. Benjamin, ed., The Economic Implications of Liberalizing APEC Tariff and Nontariff Barriers to Trade. Washington, D.C: U.S. International Trade Commission (Publication 3101). [28] Johansen, S. 1991. Estimation and hypothesis testing of cointegration vectors in Gaussian vector autoregression models. Econometrica 59: 1551-1580. [29] Ostry, J.D. and A.K. Rose. 1992. An empirical evaluation of the macroe- conomic effects of tariffs. Journal of International Money and Finance 11: 63-79. [301 Panagariya, A., S. Shah, and D. Mishra. 1996. Demand elasticities in inter- national trade - are they really low? World Bank Policy Research Working Paper 1712, Washington D.C. [31] Park, J.Y. 1992. Canonical cointegrating regressions. Econometrica 60: 119- 143. 36 [32] Petri, P.A. 1984. Modeling Japanese-American Trade. Cambridge: Cam- bridge University Press. [331 Phillips, P.C.B. and B.E. Hsmsen. 1990, Statistical inference in instrumental variables regression with I(1) process. Review of Economic Studies 57: 99- 125. [34J Reinhart, C. 1995. Devaluation, relative prices, and international trade. IMF Staff Papers 42: 290-312. [35] Rittenberg, L. 1986. Export growth performance of less-developed countries. Journal of Development Economics 24: 167-77. [361 Robinson, S. 1989, Multisectoral models. In H. Chenery and T.N. Srinivasan, eds., Handbook of Development Economics, vol. 2. Amsterdam: North- Holland. [37] Rose, A.K. 1991. The role of exchange rates in popular models of interna- tional trade: does the Marshall-Lerner conditions hold? Journal of Interna- tional Economics 30: 30:L-16. [38] Rose, A.K. 1990. Exchange rates and the trade balance: some evidence from developing countries. Economic Letters 34: 271-75. 37 [39] Senhadji, A. and C. Montenegro (1998). Time series analysis of export de- mand equations: a cross-country analysis. IMF working paper WP/98/149. [40] Shiells, C.R., D.W. Roland-Holst, and K.A. Reinert. 1993. Modeling a North-American free trade area: estimation of flexible functional forms. Weltwirtschaftliches Archiv 129: 55-77. [41] Shoven, J. and J. Whalley. 1984. Applied general equilibrium models of tax- ation and international trade: an introduction and survey. Journal of Eco- nomic Literature 22(3): 1007-51. 38 5. Appendices We compile time-series data of trade for selected countries from various sources: the World Bank's Economic and Social Database, the IMF's International Finan- cial Statistics (IFS), as well as trade taxes from the IMF's Governmuent Finance Statistics (GFS). Each variable in the model may be calibrated using different data sources and definitions. AiFter considering various alternatives, we settle on the most widely available nationaal and fiscal accounts, which give us the longest time-series:15 * X = total output of goocds and services or GDP at factor cost in constant local currency (base year 1987) - E = exports of goods and services in constant local currency from the national accounts (base year 1987) * M = imports of goods and services in constant local currency from the national accounts (base year 1987) * tm = import duty defined as the ratio of tariff revenue to imports * te = export duty (-export subsidy) defined as the ratio of export duty to exports 15Table 1 in the appendix summaries the list of variables we have tried and their alternative sources. 39 * P = price index of aggregate output or GDP at factor cost * PE = price index of exports (fob) from national accounts in local currency plus export duty rate * PM = price index of imports (cif) from national accounts in local currency plus import tariff rate * D=X-E. * Pd = PXX-P,PE * Q = D + M, total supply of goods and services, which should be equal to C + I + G, i.e., aggregate domestic demand in constant local currency in the national accounts p _ P,D+PE Note the exchange rate is not incorporated into the formula if all variables are based on the local currency.'6 Due to missing observations, we consider only 16For some countries such as Hong Kong and Singapore the domestic good price is not easy to define. Because of significant re-exports, the value of trade is greater than GDP (or aggregate output). Instead we can define Q = composite goods and Pq = price of composite goods such that pq = P-X - (PE - PM) Q Q = X-(E-M). 40 countries which long time series in In (D) In (-d), In (M) and In (pm) required for our estimation. Over long period of time, we also noted that the trends of the price indices dominate the effects of the trade taxes on the relative prices. These trends are driven by significant long-term inflation in the domestic and world markets as well as devaluations of foreign exchange rate. In order to maintain a longer time-series in PM and 2E, we decided to exclude the trade taxes, which are available only from the mid-seventies at the earliest (from the IMF GFS). This does not seem to affect the estimates of the elasticities when we tested most of the countries. 41 Table I Import Elasticities Elasticity Break MODEL Country ° t(e) Dummy SUM Year NOB Meftod Order ANGOLA 0.47 1.42 * 6 clsw/GOP ARGENTINA 0.79 3.71 10 ols w/o trend AUSTRIA 1.92 10.53 30 ols wlo trend 1.76 3.06 30 Johansen VECM, NDT 3 BAHAMAS; THE 1.50 1.91 * 8 olsawl GDP 1.44 3.62 8 SURw/trend,from 1980 BANGLADESH 0.57 2.31 22 ols w/o trend 0.55 2.83 22 ols wl trend 0.54 2.98 22 ols wl GDP 0.37 2.18 22 SUR wi GDP, all available years 0.35 1.99 22 SUR w/ trend, all available years BELGIUM 0.61 1.79 34 ols w/o trend 0.26 6.66 34 ols wl trend 0.26 6.67 70 34 ols wl trend break 0.26 3.09 34 ols wl GDP 0.27 3.27 34 SUR w/ GDP, all available years 0.24 6.41 34 SUR w/ trend, all available years BELIZE 0.66 2.04 * 15 olsawI GOP 0.81 3.20 ^ 15 SURw/trend,from 1980 BENIN 1.45 10.19 25 ols w/o trend 1.36 4.94 25 ols wl trend 1.41 5.56 25 ols w/ GOP 1.17 5.46 25 SUR w/ GDP, all available years 1.18 4.71 25 SUR w/ trend, all available years 2.50 13.84 25 Johansen VECM, LDT 4 BURUNDI 0.89 2.73 23 ols w/ trend 0.68 1.80 23 ols w/ GDP 0.65 1.74 23 SUR wl GDP, all available years 0.87 2.68 23 SUR w/ trend, all available years CAMEROON 0.22 2.83 22 ols w/o trend 0.22 2.45 22 ols w/ trend 0.23 2.67 22 ols w/ GDP 0.24 2.83 22 SUR w/ GDP, all available years 0.33 3.21 * 11 SUR w/ GDP, from 1980 0.24 2.59 22 SUR wl trend, all available years 0.52 3.43 * 11 SUR w/ trend, from 1980 CANADA 2.01 8.69 29 ols w/o trend 2.29 4.05 29 Johansen VECM, NDT 5 CHAD 0.75 3.90 * 14 ols wl GDP 0.91 4.67 ^ 14 SUR wl trend, all available years CHILE 0.43 6.45 35 ols wl trend 0.43 6.31 80 35 ols w/ trend break 0.23 2.58 35 ols w/ GOP 0.25 2.74 35 SUR w/ GDP, all available years 0.51 8.12 35 SUR wl trend, all available years 0.15 1.34 35 Johansen VECM, LOT 3 CHINA 0.06 0.35 * 15 ols wl GDP 0.06 0.47 * x 15 ols WI (X-M)IGDP wI trend Page 1 of 8 Table I Import Elasticities Elasticity Break MODEL Country o t(o) Dummy SUM Year NOB Method Order 0.06 0.60 * x 15 olswl AgrlGDPwlo trend 0.07 0.40 * x 15 SUR w/ GDP, from 1980 COLOMBIA 1.20 5.81 27 ols w/o trend 1.13 5.35 27 olsw/trend 1.14 5.39 27 ols w/ GDP 1.30 6.97 27 SUR w/ GDP, all available years 1.31 6.98 27 SUR w/ trend, all available years 5.18 3.29 27 Johansen VECM, QDT 6 CONGO 0.36 2.11 30 olsw/otrend 0.37 2.12 30 ols wl trend 0.36 3.14 85 30 ols wl trend break 0.38 2.11 30 olsw/GDP 0.36 2.15 30 SUR wl GDP, all available years 0.34 2.08 30 SUR w/ trend, all available years 0.25 1.27 x 30 Johansen VECM, NDT 4 COSTA RICA 0.87 4.65 35 ols w/o trend 0.56 5.42 35 ols w/ trend 0.54 6.04 75 35 ols w/ trend break 0.56 6.19 35 ols wl GDP 0.59 7.35i 35 SUR w/ GDP, all available years 0.62 6.97 35 SUR w/ trend, all available years CYPRUS 0.82 5.34 20 ols w/o trend 1.08 4.24 20 ols wl (X-M)IGDP w/o trend DENMARK 0.78 8.03 34 ols w/o trend 0.86 4.15 34 Johansen VECM, NDT 3 DOMINICAN REPUBLIC 0.23 2.58 32 ols w/o trend 0.27 2.80 32 ols wl trend 0.36 4.C5 85 32 ols w/ trend break 0.26 2.62 32 ols wl GDP 0.41 4.80 32 SUR wl GDP, all available years 0.41 4.87 32 SUR wI trend, all available years ECUADOR 0.27 1.!6 30 Johansen VECM, LDT 7 EGYPT 0.10 0.139 * x 13 ols wlo trend FINLAND 0.47 2.29 34 ols w/o trend 0.44 2.21 34 ols w/ (X-M)/GDP w/o trend 0.51 1.84 * 14 SUR w/ GDP, from 1980 1.33 2.46 34 Johansen VECM, NDT 3 GHANA 0.22 1.62 34 ols w/ (X-M)/GDP w/o trend 0.34 4,75 34 ols w/ AgrlGOP wlo trend GREECE 1.46 5.61 34 ols w/o trend, 0.67 10.75 80 34 ols w/ trend break 0.96 9.09 34 ols w/ GDP 1.01 9.78 34 SUR w/ GDP, all available years 0.68 11.20 34 SUR w/ trend, all available years 0.77 12.57 34 Johansen VECM, LOT 5 GRENADA 0.22 01.46 x 9 SUR w/ trend, from 1980 GUATEMALA 0.49 3.66 35 ols wlo trend 0.43 1.24 x 35 ols w/ GOP Page 2 of 8 Table I Import Elasticities Elasticity Break MODEL Country (v t(a) Dummy SUM Year NOB Method Order 5.14 0.93 x 35 Johansen VECM, NDT 7 GUINEA-BISSAU 0.05 0.32 * x 8 ols w/ GDP 0.08 0.46 * x 8 SUR w/ GDP, from 1980 0.06 0.37 * x 8 SUR wI trend, from 1980 HAITI 0.69 3.80 29 ols w/o trend 0.62 9.63 29 ols wI GDP 0.64 10.01 29 SUR w/ GDP, all available years 1.62 2.99 29 Johansen VECM, NDT 4 HONDURAS 0.29 2.24 35 SUR w/ GDP, all available years 0.28 2.05 35 SUR wl trend, all available years 0.43 1.80 35 Johansen VECM, NDT 2 HUNGARY 0.14 1.46 25 ols w/ (X-M)/GDP w/o trend 0.16 1.33 * x 15 SUR w/ GDP, from 1980 0.37 1.70 * 15 SUR w/ trend, from 1980 0.04 0.66 x 25 Johansen VECM, NDT 2 INDIA 0.29 1.45 75 35 ols w/ trend break 0.16 0.97 x 35 SUR w/ GDP, all available years 0.45 2.76 ' 15 SURw/ GDP, from 1980 0.14 0.89 x 35 SUR w/ trend, all available years 0.48 2.84 * 15 SUR w/ trend, from 1980 INDONESIA 1.39 5.94 28 ols w/o trend 1.32 8.91 28 ols w/ trend 1.26 8.90 28 ols w/ GDP 1.02 8.88 28 SUR w/ GDP, all available years 1.11 8.53 28 SUR w/ trend, all available years 1.14 4.76 28 Johansen VECM, NDT 4 IRAN; ISLAMIC REPUBLIC OF 0.06 0.71 x 19 SUR wl trend, all available years IRELAND 1.68 7.63 34 ols w/o trend 0.36 4.67 34 ols w/ trend 0.37 4.76 80 34 ols w/ trend break 0.43 5.16 -0.17 0.26 80 34 olsw/ elasticity dummy 0.43 5.31 34 ols w/ GDP 0.52 6.84 34 SUR wl GOP, all available years 0.38 5.02 34 SUR w/ trend, all available years 2.95 6.31 34 Johansen VECM, LDT 7 0.44 6.08 34 Johansen VECM, QDT 1 ISRAEL 0.36 5.02 27 ols w/o trend 0.35 4.81 27 ols w/ trend 0.33 4.32 27 ols w/ GDP 0.34 4.59 27 SUR w/ GDP, all available years 0.35 4.84 27 SUR w/ trend, all available years 0.32 6.21 27 Johansen VECM, LDT 2 JAMAICA 0.24 1.69 70 34 ols w/ trend break 0.37 1.69 * 14 SUR w/ GDP, from 1980 0.24 1.96 34 SUR wl trend, all available years 0.29 2.08 * 14 SUR wi trend, from 1980 1.80 2.44 34 Johansen VECM, QDT 1 JAPAN 0.14 1.42 0.47 0.61 75 33 ols w/ elasticity dummy Page 3of8 Table I Import Elasticities Elasticity Break MODEL Country c t(a) Dummy SUM Year NOB Method Order 0.91 4.81 33 Johansen VECM, LDT I KENYA 0.88 6.66 31 ols w/ trend 0.94 7.40 80 31 ols w/ trend break 0.93 5.21 -0.04 0.89 70 31 ols wl elasticity dummy 0.75 4.04 31 ols w/ GDP 0.90 5.63 31 SUR wl GDP, all available years 0.79 6.72 31 SUR wl trend, all available years 0.66 2.94 31 Johansen VECM, NDT 2 KOREA; REPUBLIC OF 0.06 0.29 x 35 SUR wI trend, all available years 0.88 4.30 * 15 SURw/ trend, from 1980 KUWAIT 1.44 5.48 * 19 olsw/trend 1.31 6.91 * 19 olsw/GDP 1.20 6.48 * 19 SUR w/ GDP, all available years 1.42 5.81 * 19 SUR w/ trend, all available years LAO PEOPLE'S DEMOCRATIC REPUB 0.84 5.76 * 10 SUR w/ trend, from 1980 LESOTHO 0.55 2.48 * 15 ols w/ GDP 0.48 1.75 * 15 SUR wI trend, from 1980 MADAGASCAR 1.08 11.60 34 ols w/o trend 1.18 6.91 34 ols w/ GOP 0.97 7.51 34 SUR w/ GDP, all available years 1.30 13.1, 34 Johansen VECM, NDT 3 MALAWI 0.84 8.52 31 ols w/o trend 0.55 3.37 31 ols w/ trend 0.75 5.27 80 31 ols w/ trend break 0.66 3.55 31 ols w/ GDP 0.66 3.78 31 SUR wl GDP, all available years 0.54 3.46 31 SUR wI trend, all available years 0.83 4.64 31 Johansen VECM, NDT 6 MALAYSIA 2.14 1.88 25 ols wlo trend 1.66 3.30 25 ols w/ trend 1.78 3.'4 25 ols w/ GDP 0.90 2.74 25 SUR w/ GDP, all available years 1.11 3.20 25 SUR w/ trend, all available years 2.37 4.36 25 Johansen VECM, LDT 2 MALDIVES 1.38 0.55 * x 6 SUR w/ trend, from 1980 MALI 1.06 6.61 25 ols w/o trend 0.79 3.87 25 ols w/ GDP 0.70 3 49 25 SUR w/ GOP, all available years 1.25 11.40 25 Johansen VECM, NDT 3 MALTA 2.86 1.43 33 Johansen VECM, LDT 5 MAURITIUS 0.69 3.92 34 ols w/ trend 0.72 4.03 80 34 ols w/ trend break 0.67 4.09 34 ols w/ GDP 0.53 3.68 34 SUR w/ GDP, all available years 0.61 3.93 34 SUR w/ trend, all available years 2.29 3.60 34 Johansen VECM, LDT 6 MEXICO 1.04 4.64 35 ols w/o trend 0.97 13.51 35 ols w/ trend Page 4 of 8 Table I Import Elasticities Elasticity Break MODEL Country a t(a) Dummy SUM Year NOB Method Order 1.07 9.61 85 35 ols wI trend break 1.30 9.59. -0.70 0.60 85 35 ols wI elasticity dummy 0.97 5.40 35 ols w/ GDP 1.17 8.01 35 SUR w/ GDP, all available years 1.16 8.95 35 SUR wl trend, all available years MOROCCO 0.26 1.72 30 ols wlo trend 0.46 2.54 30 ols w/ trend 0.63 3.89 75 30 ols wI trend break 0.46 2.47 30 ols wl GDP 0.45 2.44 30 SUR w/ GDP, all available years 0.44 2.42 30 SUR w/ trend, all available years MYANMAR 0.26 1.69 28 olsaw (X-M)/GDP wlo trend 0.48 5.51 28 ols w/ (X-M)IGDP w/ trend 0.83 3.64 15 SUR wl trend, from 1980 0.17 2.09 28 Johansen VECM, NDT 5 NAMIBIA 0.81 1.91 * 10 olsw/GDP 0.64 1.58 t 10 SURRwltrend,from 1980 NETHERLANDS 1.06 8.11 32 ols wlo trend 2.93 1.65 32 Johansen VECM, NDT 1 NIGERIA 0.63 6.59 22 ols w/ trend 0.77 8.04 22 ols w/ GDP 0.76 8.00 22 SUR w/ GDP, all available years 0.64 6.84 22 SUR wl trend, all available years NORWAY 0.49 5.76 34 ols w/o trend 0.36 2.36 34 ols w/ trend 0.67 4.82 85 34 ols w/ trend break 0.40 4.13 0.80 1.19 70 34 ols wl elasticrty dummy 0.33 2.47 34 ols w/ GDP 0.31 2.35 34 SUR wI GDP, all available years 0.32 2.12 34 SUR wI trend, all available years 0.63 5.67 34 Johansen VECM, NDT 3 PAKISTAN OA7 13.69 35 ols w/o trend 0.49 6.15 35 ols w/ trend 0.61 7.95 70 35 ols w/ trend break 0.43 6.16 35 olsw/GDP 0.43 6.22 35 SUR w/ GDP, all available years 0.92 4.72 15 SUR wl GDP, from 1980 0.49 6.14 35 SUR w/ trend, all available years 0.93 4.85 * 15 SUR w/ trend, from 1980 0.46 15.03 35 Johansen VECM, NDT 1 PANAMA 2.63 1.70 * 7 ols wl GDP PAPUA NEW GUINEA 0.28 1.92 21 ols w/o trend 0.34 2.47 21 ols w/ trend 0.31 2.10 21 als w/ GOP 0.31 2.10 21 SUR wl GDP, all available years 0.34 2.47 21 SUR w/ trend, all available years PARAGUAY 1.37 3.50 35 ols w/o trend 0.62 2.03 85 35 ols w/ trend break Page 5 of 8 Table I Import Elasticities Elasticity Break MODEL Country a t(a) Dummy SUM Year NOB Method Order 0.93 4.50 35 SUR w/ GDP, all available years 0.76 3.81 35 SUR wl trend, all available years 3.19 4.19 35 Johansen VECM, NDT 3 PHILIPPINES 0.73 6.31 35 ols w/ trend 0.73 6.37 75 35 ols wI trend break 0.99 6.37 35 ols w/ GDP 1.08 7.97 35 SUR w/ GDP, all available years 0.97 3.03 ^ 15 SUR w/ GDP, from 1980 0.75 6.85 35 SUR w/ trend, all available years 1.51 2.40 * 15 SUR wI trend, from 1980 PORTUGAL 0.63 4.73 27 ols w/o trend 0.61 7.63 27 ols w/ trend 0.65 7.63 27 ols w/ GDP 0.67 4.47' 27 ols wl (X-M)/GDP wto trend 0.59 6.40 27 ols wl (X-M)/GDP wl trend 0.67 8.75 27 SUR w/ GDP, all available years 0.63 8.38 27 SUR wI trend, all available years 0.41 3.03 * 13 SUR wltrend, from 1980 0.49 3.93 27 Johansen VECM, LDT 3 RWANDA 0.17 1.413 33 ols w/ GDP 0.16 1.44 33 SUR w/ GDP, all available years 0.40 3.95 * 13 SUR wlGDP, from 1980 0.40 4.09 * 13 SURw/trend, from 1980 0.46 4.91 33 Johansen VECM, LDT 4 SALVADOR, EL 0.40 1.06 x 35 Johansen VECM, NDT 6 SENEGAL 0.27 1.34 * 14 ols w/ GDP 0.27 1.24 * x 14 SUR wI trend, all available years SEYCHELLES 0.68 2.C9 * 7 olsw/GDP 0.85 2.27 * 7 SUR w/ trend, from 1980 SIERRA LEONE 1.10 7.53 24 ols w/o trend 0.50 2.49 24 ols w/ trend 0.99 5.94 24 ols wI GDP 1.11 6.93 24 SUR w/ GDP, all available years 0.64 3.!54 24 SUR wI trend, all available years SOMALIA 0.82 4.47 * 8 SUR wl trend, all available years 0.82 4.47 * 8 SUR w/ trend, from 1980 SOUTH KOREA 1.64 9.20 35 ols w/o trend 1.29 5.130 35 ols wi (X-M)/GDP wlo trend SRI LANKA 0.90 4.92 32 ots w/o trend 0.76 4.15 32 ols w/ trend 1.02 6.92 75 32 ols wI trend break 0.75 3.96 0.93 1.68 70 32 ols w/ elasticity dummy 0.66 3.55 32 ols w/ GDP 0.58 3.79 32 SUR wI GDP, all available years 0.62 4 32 32 SUR w/ trend, all available years 3.27 6 58 32 Johansen VECM, LDT 3 ST. KITTS AND NEVIS 2.80 1.59 * 6 ols wI GDP 2.67 1.41 * 6 SUR w/ trend, from 1980 Page 6 of 8 Table I Import Elasticities Elasticity Break MODEL Country a t(o) Dummy SUM Year NOB Method Order ST. VINCENT AND THE GRENADINES 1.09 2.30 t 11 ols w/ GDP 0.47 1.38 * 11 SUR w/ trend, all available years SWAZILAND 0.38 3.23 * 18 olsw/trend 0.38 3.36 * 18 olsw/GDP 0.40 3.57 18 SUR w/ GDP, all available years 0.35 3.02 t 15 SUR w/ GDP, from 1980 0.41 3.58 18 SUR w/ trend, all available years SWEDEN 0.41 1.62 34 ols wlo trend 0.17 2.68 70 34 ols w/ trend break 0.14 2.62 34 ols w/ GDP 0.15 2.96 34 SUR w/ GDP, all available years 0.14 2.13 34 SURw/trend, all available years 1.17 0.67 x 34 Johansen VECM, NDT 2 SWITZERLAND 1.38 17.92 34 ols w/o trend 1.58 19.85 34 ols w/ (X-M)/GDP w/o trend 1.61 10.33 34 Johansen VECM, NDT 1 SYRIAN ARAB REPUBLIC 0.12 1.35 17 ols w/ trend 0.09 1.21 * x 17 ols w/GDP 0.11 1.37 * 17 SUR w/ GDP, all available years 0.13 1.54 17 SUR w/ trend, all available years TAIWAN 1.41 3.50 32 ols w/o trend 0.18 1.86 1.24 1.41 70 32 ols w/ elasticity dummy THAILAND 0.97 4.82 35 ols w/ trend 0.93 5.05 85 35 ols wl trend break 0.89 4.88 35 ols w/ GDP 1.13 9.09 35 SUR w/ GDP, all available years 1.07 7.41 35 SUR w/ trend, all available years TOGO 0.97 3.45 * 18 ols w/o trend 0.47 1.60 * 18 olsw/trend 0.46 1.93 t 18 SUR w/ GDP, all available years 0.63 2.50 * 8 SUR w/ GDP, from 1980 0.58 2.22 ' 18 SUR w/ trend, all available years 0.77 2.86 * 8 SUR wI trend, from 1980 TRINIDAD AND TOBAGO 0.81 28.34 28 ols w/o trend 0.62 7.82 28 ols w/ trend 0.78 24.41 28 ols w/ GDP 0.76 24.34 28 SUR w/ GDP, all available years 0.61 8.35 28 SUR w/ trend, all available years 1.01 10.39 28 Johansen VECM, NDT 2 TUNISIA 3.14 3.61 27 Johansen VECM, NDT 6 TURKEY 0.33 1.46 * 8 SURw/trend,from 1980 UNITED ARAB EMIRATES 0.76 4.32 * 15 SUR w/ trend, all available years 0.71 2.32 10 SURw/trend, from 1980 UNITED STATES 0.71 2.02 34 ols w/ GDP 0.54 1.57 34 SUR wl GDP, all available years 2.61 7.72 14 SUR w/ GDP, from 1980 2.40 0.82 x 34 Johansen VECM, NDT 4 1.47 0.77 x 34 Johansen VECM, NDT 5 Page 7 of 8 Table 1 Import Elasticities Elasticity Break MODEL Country a t(a) Dummy SUM Year NOB Method Order VENEZUELA 0.32 1.12 rx 15 SURwl GDP, from 1980 YUGOSLAVIA 0.64 9.31 21 ols w/o trend 0.45 4.95 21 ols w/ trend 0.53 7.14 21 olsw/GDP 0.55 7.41 21 SUR wJ GDP, all available years 0.44 4.89 21 SUR wl trend, all available years ZIMBABWE 0.49 2.87 * 15 ols wlo trend 0.51 2.71 15 olsw/trend 0.51 2.75 15 olsw/GDP 0.51 2.72 * 15 SUR w/ trend, all available years Notes: o = Armington elasticity t(a) = t statistic Model: Model Specification NOB = Number of observations NDT - No deterministic trend in the data, VECM LDT - Linear deterministic trend in the data, VECM QDT - Quadratic deterministic trend in the data, VECM Order: Number of lags of the VECM model.VECM SUM - The sum of elasticity and dummy * = Less than 20 observations x = lnsignifcant at 90% confidence level Page 8 of 8 Table 2 Export Elasticities, Selected Estimates Break Model Country Q2 t(Q) Dummy SUM Year NOB Method Order ARGENTINA 0.31 1.43 ' 10 ols wl trend 0.37 2.20 * 10 ols w/ GDP 0.39 1.81 * 10 SUR w/ trend, from 1980 AUSTRIA 0.78 4.87 30 ols w/ trend 0.94 6.30 75 30 olswl trend break 0.98 5.27 0.15 1.13 75 30 ols w/ elasticity dummy 0.71 4.55 30 SUR w/ trend, all available years 0.47 4.42 30 Johansen VECM, LDT 4 BANGLADESH 0.60 1.94 22 ols w/o trend 0.72 2.55 22 ols w/ (X-M)/GDP w/o trend 0.32 1.58 20 ols excl. Agr exports w/o trend BELGIUM 0.39 0.13 x 34 JohansenVECM, NDT w/Agr/GDP I BOLIVIA 0.35 3.04 13 ols w/o trend CAMEROON 0.26 1.13 * x 11 SURw/GDP, from 1980 CANADA 1.57 1.19 x 29 JohansenVECM, NDT 5 CHAD 0.12 0.23' x 6 SURw/trend, from 1980 CHILE 0.05 0.54 x 15 SUR w/ trend, from 1980 CHINA 0.23 4.36 ' 15 ols w/o trend 0.21 4.77 * 15 ols wi (X-M)/GDP wlo trend 0.15 1.81 * 15 olsw/ Agr/GDPw/o trend COLOMBIA 0.32 2.01 * 15 SURw/GDP,from 1980 0.52 3.47 15 SUR w/ trend, from 1980 CONGO 0.95 1.87 30 Johansen VECM, NDT w/ Agr/GDP 5 COSTA RICA 0.70 1.55 35 Johansen VECM, NDT w/ Agr/GDP I CYPRUS 0.93 1.26 x 20 ols w/ Agr/GDP w/ trend DENMARK 0.03 0.31 x 34 ols w/ trend 0.02 0.32 x 34 ols wI (X-M)lGDP w/ trend 0.03 0.29 x 34 SUR wi trend, all available years DOMINICAN REPUBLIC 2.73 3.95 32 Johansen VECM, NDT 3 ECUADOR 12.69 3.09 30 Johansen VECM, NDT I EGYPT 0.11 1.72' 13 ols w/o trend 0.10 1.69 ' 13 ols w/ GDP 0.09 1.43 ^ 13 SURw/trend, all available years FINLAND 0.47 2.72 34 ols w/ trend 0.61 3.36 75 34 ols w/ trend break 0.32 1.90 34 ols w/ GDP 0.25 1.53 34 SUR w/ GDP, all available years 2.05 3.82 * 14 SUR w/ GDP, from 1980 0.43 2.51 34 SUR w/ trend, all available years 1.96 6.06 * 14 SURw/trend, from 1980 0.11 0.78 x 34 Johansen VECM, LDT 4 GHANA 0.10 1.96 34 ols w/ trend 0.11 2.02 80 34 ols w/ trend break 0.09 1.72 34 SUR w/ trend, all available years 0.42 4.78 34 Johansen VECM, QODT 4 GREECE 0.78 12.63 34 Johansen VECM, QODT 5 GUATEMALA 0.33 1.90 85 35 ols w/ trend break 1.70 5.25 35 Johansen VECM, LDT 2 Page 1 of 4 Table 2 Export Elasticities, Selected Estimaltes Break Model Country a t(n) Dunmmy SUM Year NOB Method Order HAITI 0.43 3.48 29 ols w/ trend 0.40 3.88 29 ols w/ GDP 0.37 3.62 29 SUR w/ GDP, all available years 0.32 2.74 29 SUR w/ trend, all available years 0.30 3.57 29 Johansen VECM, LOT 5 0.30 3.55 29 Johansen VECM, QDT 5 HONDURAS 0.19 1.14 x 35 JohansenVECM, LDT 4 HUNGARY 0.56 2.59 25 ols wl (X-M)IGDP wI trend INDIA 0.44 5.14 35 ols w/o trend 0.41 2.32 70 35 ols w/ trend break 0.45 2.66 35 ols wI GDP 0.43 2.58 35 SUR w/ GDP, all available years 1.02 1.94 * 15 SURwiGDP,from1980 0.33 1.80 35 SUR w/ trend, all available years 0.23 4.81 35 Johansen VECM, NDT 6 INDONESIA 0.75 1.62 28 Johansen VECM, NDT w/ Agr/GDP 2 IRAN 0.14 2.03 * 19 olsw/o trend 0.33 1.70 * 19 ols excl. Agr expors w/o trend IRELAND 1.01 2.17 28 ols excl. Agr exports w/o trend ISRAEL 0.66 2.55* 15 SUR w/ GDP, from 1980 JAMAICA 0.45 3.61 34 ols w/o trend 0.22 1A6 34 olswl trend 0.30 1.94 75 34 ols w/ trend break 0.29 1.96 34 ols w/ GDP 0.38 2.38 28 ols exd. Agr exports wlo trend 1.55 3.38 34 Johansen VECM, NDT 4 1.47 3.46 34 Johansen VECM, LDT 4 JAPAN 0.56 1.92 33 ols w/ trend 1.06 4.20 80 33 ols w/ trend break 0.90 5.87 33 ols w/ GDP 0.93 4.22 28 ols exd. Agr exports w/ trend 1.04 7.11 33 SURw/GOP, all available years 0.52 1.84 33 SUR w/ trend, all available years 3.46 2.05 33 Johansen VECM, NDT 3 2.01 1.33 33 Johansen VECM, NDT 2 JORDAN 6.10 2.67 * 10 ols w/ GDP 7.25 2.38 ' 10 SUR wI trend, all available years 7.25 2.38 * 10 SUR w/ trend, from 1980 KOREA; REPUBLIC OF 0.94 3.64 35 ols w/ GDP 1.17 2.81 35 SUR w/ trend, all available years KUWAIT 0.45 1.02 * x 9 SUR wl GDP, from 1980 MALAYSIA 0.53 1.47 22 ols excl. Agr exports w/o trend 2.43 1.68 25 Johansen VECM, NDT 3 3.24 3.27 25 Johansen VECM, NOT 5 MALI 3.78 0.77 25 Johansen VECM, NDT 3 MALTA 1.71 1.17 33 Johansen VECM, NDT 2 MAURITIUS 0.14 0.35 * 14 SUR w/trend, from 1980 MEXICO 0.42 2.95 85 35 ols wl trend break Page 2 of 4 Table 2 Export Elasticities, Selected Estimates Break Model Country Q t(Q) Dummy SUM Year NOB Method Order 0.37 2.09 35 ols w/ Agr/GDP wl trend 0.47 2.31 28 ols excl. Agr exports w/o trend 0.22 1.18 35 SUR w/ GDP, all available years 0.72 4.80 * 15 SUR w/ GDP, from 1980 0.76 6.13 t 15 SUR w/trend, from 1980 2.54 1.49 35 Johansen VECM, LDT 4 MOROCCO 0.89 3.76 30 Johansen VECM, NDT w/ Agr/GDP 4 MYANMAR 0.24 0.44 x 28 Johansen VECM, NDT w/ Agr/GDP 4 NETHERLANDS 0.09 1.81 ' 12 SURw/GDP, from 1980 NORWAY 0.27 0.49 x 34 Johansen VECM, NDT wI Agr/GDP 2 PAKISTAN 1.50 1.43 35 Johansen VECM, NDT 6 PAPUA NEW GUINEA 0.02 0.14 * x 21 ols wl AgrlGDP wl trend PARAGUAY 1.67 0.97 x 35 Johansen VECM, NDT w/ Agr/GDP 6 PHILIPPINES 0.23 2.57 35 Johansen VECM, NDT w/ Agr/GDP 6 PORTUGAL 0.23 2.24 t 13 SUR w/ GDP, from 1980 0.20 1.69 t 13 SURwltrend,from 1980 SALVADOR, EL 0.19 1.88 35 ols w/o trend 0.23 10.27 35 Johansen VECM, LDT 6 SENEGAL 0.35 1.63 * 14 ols w/ (X-M)lGDP w/o trend SOUTH KOREA 1.30 2.36 35 ols w/ trend 1.42 3.36 80 35 ols wl trend break 0.54 2.32 35 SUR w/ GDP, all available years SWEDEN 0.39 5.83 75 34 ols w/ trend break 0.44 2.04 t 14 SUR w/ GDP, from 1980 0.62 5.06 * 14 SUR w/ trend, from 1980 0.43 0.21 x 34 Johansen VECM, NDT 1 SWITZERLAND 0.42 1.22 x 80 34 ols w/ trend break 0.61 2.92 * 14 SUR w/trend, from 1980 SYRIAN ARAB REPUBLIC 0.09 1.43 t 17 ols w/o trend 0.09 1.35 ' 17 ols wl (X-M)/GDP w/o trend TAIWAN 0.60 3.25 32 ols w/ trend 0.69 4.36 75 32 ols w/ trend break 0.67 3.35 0.25 0.92 75 32 ols w/ elasticity dummy 0.17 1.65 32 ols w/ GDP 0.14 1.36 32 SUR w/ GDP, all available years 0.67 4.18 32 SUR w/ trend, all available years THAILAND 2.89 3.00 35 Johansen VECM, LDT w/ Agr/GDP 3 TUNISIA 0.31 1.91 27 ols w/o trend 0.39 2.41 27 olsw/ (X-M)/GDPw/o trend TURKEY 0.73 2.91 * 8 ols w/ GOP 0.70 2.61 * 8 SUR w/ trend, from 1980 UGANDA 0.08 1.09 t x 12 ols w/o trend 0.06 1.02 t x 12 ols w/GDP 0.06 0.94 * x 12 SURw/GDP,from 1980 0.05 0.74 * x 12 SURw/trend, from 1980 UNITED STATES 0.39 1.77 34 ols w/ trend 0.49 2.28 80 34 ols w/ trend break 0.39 1.78 34 SUR w/ trend, all available years Page 3 of 4 Table 2 Export Elasticities, Selected Estimates Break Model Country L2 t(£) Dummy SUM Year NOB Method Order 2.77 1.94 * 14 SUR w/ trend, from 1980 5.86 2.02 34 Johansen VECM, NDT 6 VENEZUELA 0.15 1.50* 15 SUR w/ GDP, from 1980 0.17 1.68 * 15 SUR wl trend, from 1980 Notes: a = Armington elasticity t(o) = t statistic Model: Model Specification NOB = Number of observations NDT - No deterministic trend in the data, VECM LDT - Linear deterministic trend in the data, VECM QDT - Quadratic deterministic trend in the data, VECM Order: Number of lags of the VECM model.VECM SUM - The sum of elasticity and dummy * = Less than 20 observations x = Insignifcant at 90% confidence level Page 4 of 4 Policy Research Working Paper Series Contact Title Author Date for paper WPS2149 Income Gains to the Poor from Jyotsna Jalan July 1999 P. Sader Workfare: Estimates for Argentina's Martin Ravallion 33902 Trabajar Program WPS2150 Who Wants to Redistribute? Russia's Martin Ravallion July 1999 P. Sader Tunnel Effect in the 1990s Michael Lokshin 33902 WPS2151 A Few Things Transport Regulators Ian Alexander July 1999 G. Chenet-Smith Should Know about Risk and the Cost Antonio Estache 36370 Of Capital Adele Oliveri WPS2152 Comparing the Performance of Public Antonio Estache July 1999 G. Chenet-Smith and Private Water Companies in the Martin A. Rossi 36370 Asia and Pacific Region: What a Stochastic Costs Frontier Shows WPS2153 The Mystery of the Vanishing Martin Ravallion July 1999 P. Sader Benefits: Ms. Speedy Analyst's 33902 Introduction to Evaluation WPS2154 Inter-industry Labor Mobility in Howard Pack August 1999 H. Sladovich Taiwan, China Christina Paxson 37698 WPS2155 Lending Booms, Reserves, and the Barry Eichengreen August 1999 S. Kpundeh Sustainability of Short-Term Debt: Ashoka Mody 39591 Inferences from the Pricing of Syndicated Bank Loans WPS2156 How Has Regionalism in the 1990s Isidro Soloaga August 1999 L. Tabada Affected Trade? L. Alan Winters 36896 WPS2157 How Regional Blocs Affect Excluded Won Chang August 1999 L. Tabada Countries: The Price Effects of L. Alan Winters 36896 MERCOSUR WPS2158 The Effect of Foreign Entry on George R. G. Clarke August 1999 P. Sintim-Aboagye Argentina's Domestic Banking Robert Cull 38526 Sector Laura D'Amato Andrea Molinari WPS2159 Provincial Bank Privatization in George R. G. Clarke August 1999 P. Sintim-Aboagye Argentina: The Why, How, and Robert Cull 38526 "So What?" WPS2160 Protecting the Poor from Francisco Ferreira August 1999 PREM Macroeconomic Shocks Giovanna Prennushi 87736 Martin Ravallion Policy Research Working Paper Series Contact Title Author Date for paper WPS2161 Will the Real "Natural Trading Maurice Schiff August 1999 M. Kasilag Partner" Please Stand Up? 39081