W4S i6 o POLICY RESEARCH WORKING PAPER 1410 Macroeconomic Effects The experience of 18oi exporting countries (from of Terms-of-Trade Shocks -- 1973 to 1989) suggests that permanent terms-of4rade shodcs have a significant The Case of Oil-Exporting Countries positive effe'z on :consumptionr investment N.kola S- t'fora and output, particularly of Nikola Spatafora .-- ;- Ade Wae nontradables. There is almost Andrew Warner -{ no effect on savings, and -at. advEnerese ect on the trade. balance. There- is no evidence -6of'Dutch disease effects. The World Bank IntrntioalEconomics Depatrtment Intermaional EcoDnomic Analysi and Prospects Divison January 1995 Poi.icy RESIEARC(I-I WORKIN; PAI'hiR 1410 Summary findings Spatafora and Warner investigate the impact on The shocks al;o have a significant positive effect on economic growthi and development of long-run consumption. Government consumption responds almost movements in the external tcrms of tradc, with special twicc as strongly as private consumption. reference to the experience of 18 oil-exporting countries The shocks have no effect on savings and adversely between 1973 and 1989. affect the trade and current account balances. They argue that this sample approximates a controlled There is a significant positive effect on the output of cxperiment for examining the impact of unanticipated- all main categories of nontradables. But Dutch disease but permanent - shocks to the terms of trade. They effects arc strikingly absent. Agriculture and analyze the sample econometrically using panel data manufacturing do not contract in reaction to an oil price techniques. increase. Dutch disease effects may be absent in part They find that permanent ternis-of-trade shocks have a because of policy-induced output restraints in the oil strongly significant positive effect on investment, which sector, or because of the "enclave" nature of the oil they justify theoretically on the grounds that countries in sector, which does not participate in domestic factor the sample import much of their capital equipment. markets. This paper-a product of the International Economic Analysis and Prospects Division, International Economics Department- is one in a series of background papers prepared in support of the analyses and scenarios in Global Economic Prospects 1994. Copies of this paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contactJackie Queen, room S8-216, extension 33740 (37 pages). January 1995. The Policy Research Working Paper Series disseminates the fidings of work mi progress to encourage the exchange of ideas ahout development isue An objecae of the series is to get the findings out quickly, even if the presentations are kess than fly polished The papers carry the names of t authors and should be used and cited accor&ngly. The findings. inteapretations, and condusions are the authors' oum and should not be attributed to the World Bank. its Exwtive Board of Directors. or an of its moneber countries. Produced by the Policy Research Dissemination Center Macroeconomic Effects of Terms-of-Trade Shocks The Case of Oil-Exporting Countries Nikola Spatafora and Andrew Warner MACROECONOMIC EFFECTS OF TERMS-OF-TRADE SHOCKS: THE CASE OF OIL-EXPORTING COUNTRIES. Nikola SPATAFORA & Andrew WARNER I. Introduction Most development economists agree that variations in world prices are an imnportant source of risk and instability for developing economies. Developing counaies derive about half their export earnings from primary commodities, whose world prices are extremely volatile. Nor is such volatility purely short-term: mauch data suggest that commodity prices undergo long periods of rise and decline. To provide one example of the magnitudes involved, the World Bank's index of non-oil commodity prices has exhibited a trend decline of about 1.5% per annum since 1948, cumulating to a 50% decline over 45 years. Over the next decade, the World Bank is forecasting that this trend will reverse and there will be a rise of about 0.7% per annum. What is the impact of these long-run movements in the external terms of trade on economic .growth and development? We think that the world has provided a natural experiment over the past 25 years that can be used to analyze this issue. Oil-exporting developing countries experienced a major rise in the world price of their main export between 1973 and 1981, followed by a smaller, but still substantial, decline between 1981 and 1989. Nominal oil prices rose from $2.70 in 1973 to $34.31 in 1981, and then fell back to $16.31 in 1989; deflating this by US producer prices, real oil prices rose from an index of 1.0 in 1973 to 5.83 in 1981 and 2.42 in 1989. From a research standpoint, we think that this episode represents a fortunate opportunity to better understand the impact of terms-of-trade shocks, for the following reasons. First, it seems important to distinguish between permanent and tansitory, or anticipated and unanticipated changes in the terms of trade, and we believe that oil prices come closer than any other data to measuring an unanticipated permanent change. The 1973 rise was not anticipated several years before, and it was quickly and widely perceived as a permanent feature of the economic landscape. The long decline page 2 beginning in 1981 was also not widely anticipated; it probably took longer for it to be accepted as a long- run phenomenon, but at least by 1986, when oil prices fell by 25%, the 1981 levels were long considered a thing of the past. Hence, it is credible that the major long run changes in oil prices were previously unanticipated events that were widely viewed to be permanent once they occurred. Second, the time span of the rise and decline in oil prices is fairly long-8 or 9 years-so that it is credible to believe that we can observe long-run effects. Third, the movements in oil prices were large. Hence, there is much statistical variation, and if effects are present, we should be able to estimate them more efficiently. Fourth, terms-of-trade changes were so dominant for these countries over this period that deciding what variables to control for, so as to avoid omitted variable bias, should constitute less of a problem Fifth, we have both a rise and a decline in oil prices, so that we can try to detect asymmetries in the responses. Sixth, measurement problems associated with index numbers are less of an issue for oil exporters, since oil is a relatively homogeneous product Seventh, simultaneity between the terms of trade and domestic economic variables, while not always absent, is at least a tractable problem because there is a fair consensus about the causes of the major oil price movements. Further, for many of the countries and most of the variables we examine, it seems reasonable to view the terms of trade as predetermined. Our broad conclusion is that an examination of the data for oil exporters leads to a different picture of terms-of-trade effects than can be obtained from an unbiased sample of terms-of-trade articles in the literature. We find that the impact on investment is crucial to understanding the response of the current account to, and the long-run growth effects of, terms-of-trade shocks; yet the literature is filled with current account models that hold investment constant, and long-run trade models that treat the capital stock as an endowment which is unaffected by terms-of-trade shocks. Conversely, Dutch disease effects are examined extensively in the literature, but we fail to find any evidence that the Dutch disease is a major phenomenon. We also find that the simple insights fmm the tradables-nontradables model are well supported by the data. Finally, we find that the response of expenditure to terms-of-trade shocks is not very sensitive to whether the expenditure comes from the public or private sector. page 3 The rest of the paper is divided as follows. Section II summarizes the relevant theoretical literature and empirical studies. Section m contains a description of the data. Section IV sets out the econometric framework and analyses some importnt econometric issues. Section V presents the main results. Section VI discusses some criticisms. Section VII concludes. II. Theoretical Jiterature and empirical studies For expository purposes, we find it useful to divide the theoretical literature on the impact of terms-of-trade shocks into two broad groups. One class of models is microeconomnic in nature and stresses the differing effects of terms-of-trade shocks on different sectors of the economy. Examples include the tradables- nontradables models originated by Meade, Salter (1959), and Swan; the Dutch Disease models found in, say, Wijubergen (1984, 1984b) and Neary & Wijnbergen (1984), and summarized in Corden & Neary (1982) and Corden (1984); and the computable general equilibrium model of Bruno & Sachs (1982). A second class is concermed with the behavior of broad macroeconomic aggregates, particularly saving and the current account, and recently has tended to stress intertemporal issues. A partial reference list for this literature includes Laursen & Metzler (1950), Harberger (1950), Obstfeld (1982), Persson & Svensson (1985), and Bean (1986). A. Sectoral Effects Many of the insights from the earlier literature stressing sectoral disaggregation can be obtained, following Corden, by thinldng in terms of a thee-sector, perfectly competitive neoclassical model, with fixed total factor endowments. Let the first sector (0) produce oil for export; let the second, 'Dutch Disease' sector (D) produce all other tradables, both exportables and importables; and let the third sector (N) produce non- tradables. The first two sectors produce tradables at given world prices, and the difference between the quantity supplied and demanded is made up by exports or imports; the price of nontradables is endogenously detwrmined by the condition that their supply equal their demand. Assume first that output in each sector is produced by a factor specific to tluht sector, and by labor, which is mobile between all three sectors and moves between sectors to equalize its wage in all three employments. page 4 Now let there be an oil price shock. The direct effect is to increase the aggregate incomes of the factors initially employed in 0, which in turn has two effects. First, if some part of the extra income is spent, whether directly by factor owneis or indirectly through being collected in taxes and then spent by the govenunent, and provided the income elasticity of demand for N is positive, demand for N rises. Given less than perfectly elastic supply. the price of N relative to tradables must rise. That is, there is a real appreciation which draws resources out of 0 and D into N. This is the spending effect. In addition, the price increase raises the marginal value product of factors in O so that, at a constant wage in tenns of tradables, the demand for labor in 0 rises, inducing a movement of labor out of D and N. This is the resource movenen effect. The output of D must finally be lower than before the shock, while the output of N could be higher or lower, the spending effect tends to increase it, the resource movement effect to decrease it. Assume now, as is common, that the oil sector does not in fact employ any factors that can be used by the rest of the economy. Here, the oil sector is an 'enclave' which does not participate in domestic factor markets. There is then only a spending effect, and the key mechanism of resource reallocation is the real appreciation. Provided spending on nontradables initially goes up, output of N must finally be higher than in the pre-shock situation. Assume instead that more than one factor is mobile between at least two of our three main sectors. For instance, say that capital is also mobile between D and N, so that these two sectors, employing labor and capital in different proportions, make up a mnini-Heckscher-Ohlin economy. Now the resource movement effect can have paradoxical result. Because of the price shock, labor moves out of this mini-economy into 0. If D is the capital-intensive industry, its output could on balance expand. If N is capital-intensive, the shock could cause a real depreciation. At this stage it is useful to compare the predictions of this neoclassical model with a Keynesian model. In the neoclassical model, with factors fixed in the short run, and full employment determining output through the production functions, there is no scope for 'aggregate output' to change much: the shock can at most change the composition of output. However, in a Keynesian economy with sticky prices, the demand stimulus can raise aggregate output even in the short run. page 5 Regarding longer-term supply-side effects, the neoclassical framework predicts that these hinge crucially on the relative price signals. Reproducible factors, such as physical and human capital, will accumulate in sectors whose relative prices rise following the terms-of-trade improvement (more precisely, the relevant price is a product price adjusted for movements in input prices, that is, a value added price). As we have seen, the relative prices in 0 and N typically rise in the short run, and thus factors accumulate in these sectors, expanding output in the longer run. By contrast, capital will shift out of, or be allowed to depreciate without replacement, in D. B. Macroeconomic effects Tuning now to the macroeconomic literature, early analyses, conducted within the framework of the static Keynesian saving and investment functions, include Laursen & Metzler (1950) and Harberger (1950). The driving mechanism is that terms-of-trade shocks affect real income, and hence savings. Given investment, this determines the evolution of the current account. Later research, typified by Obstfeld (1982), Persson & Svensson (1985), and Sen & Turnovsky (1989), is based on neoclassical models, typically involving dynamic optimization. Terms-of-trade shocks alter permanent income, intratemporal and intertemporal relative prices, and hence affect consumption, saving, and investment. One conclusion to emerge from this literature is that the effects depend critically upon whether the shock is permanent or temporary, and anticipated or urnanticipated. Let us first consider the possible impact of a permanent, unanticipated increase in the terms of trade on consumption. The shock increases both current and permanent income. All models therefore predict an increase in aggregate consumption. In sectoral terms, consumption of oil is probably insignificant for our sample, and consumption of other tradables must rise through income effects. Consumption of nontradables must equal production, which was discussed above. Investment is treated as exogenous by much of the literature. Older Keynesian models often postulate a simple accelerator mechanism whereby the shock raises expected demand and hence investment. Optimizing models usually conclude that aggregate private investment is determined by the ratio of Tobin's q (the present discounted value of future marginal products of capital) to the aggregate investment price deflator. Assuming that capital goods have a strong import content, the shock boosts domestic page 6 investment-Schmidt-Hebbel & Serven (1993). Yet most models treat investment goods as dcmestically produced, and hence obtain far more ambiguous result-Persson & Svensson (1985). Sen & Turnovsky (1989). For developing economies, the former assumption appears to be correct. Three factors may however reduce the incentive to invest. First, profits may be captured by unions (rent- sharing). Yet many of these countries lack an organized labor movement. Second, in OPEC countries production quotas may be seen as a permanent ceiling on oil production, leaving little reason to invest in the oil sector's productive capacity. On the other hand, an increase in such capacity might strengthen one's hand in the periodic quota negotiations; also, there is stil an incentive for cost-reducing investment. Finally, increased wealth may encourage consumption of leisure, and a contraction in labor supply will reduce the marginal product of capital; see Bean (1986), Sen & Turnovsky (1989). Yet many of these countries face an extremely elastic supply of potential immigrant workers. Regarding the current account, Laursen & Metzler postulated that the shock by raising real income, increases savings. For a given investment, this increases the current account. In an optimizing intertemporal framework, with domestically produced investment goods, Obstfeld (1982) and Svensson & Razin (1983) showed that with a constant rate of time preference the economy should jump immediately to its new steady state, with no effect on the current account. If instead the rate of time preference decreases with utility, there is a transitory rise in savings and current account surplus. On the other hand, if investment requires imports of capital machinery, it may be encouraged by the shock, leading to a temporary current account deficit. C. Empirical Studies Warner (1992) examined whether the international debt crisis which began in 1982 could explain the investment decline in 14 heavily indebted countries. He found that equations which omitted all debt- related information, but incorporated the effects of falling export prices and high world real interest rates, could forecast the fall in investment. This suggests the importance of the terms of trade in determining investment. Warner (1994) proposed and estimated a microeconomic investment model to determine the relative importance for Mexico's investment decline in the early 1980's of three explanations: the oil price decline, page 7 the terriination of capital inflows, and debt-overhang/uncertainty effects. Using quarterly investment data for 68 private-sector industries between 1981 and 1985, he found that the main mirroeconomic mechanism driving the investment decline was the rise in the relative price of imported investment goods, and further that the terms of trade decline (driven by falling world oil prices) explains much of the increase in this relative price. Morley (1992) examined stabilization programs in a broad sample of LDC's. Using panel data, he found that the terms of trade had a significant positive impact on investnent and output. The same was true of real appreciations, which as we have argued are a likely consequence of terms-of-trade shocks. Etherington & Yainshet (1988), using time-series methodology, found that the price of coffee, the key Ethiopian export, had a significant positive impact on Ethiopian capital good imports and domestic capital formation. De Gregorio (1992) analyzed growth determinants in twelve Latin American countries over the period 1950-1985. Controlling for investment and macroeconomic stability, he did not find a significant effect of the terms of trade on growth. However, he did not consider that investment itself might be affected by the terms of trade. III. The Data We consider the period 1965-1989. Our sample consists of the following 18 oil-exporting countries: Algeria Gabon Mexico Trinidad and Tobago Bahrain Indonesia Nigeria United Arab Emirates Congo Iran Oman Venezuela Ecuador Iraq Saudi Arabia Egypt Kuwait Syria For each country, on the expenditure side we examined data on consumption and investment (by the private sector, by the government, and in the aggregate), savings, the trade balance, and GDP. At a sectoral level, we analyzed value added in the oil sector and the non-oil sector. The non-oil sector was further broken down into the following categories: agriculture, manufacturing, construction, public utilities, and page 8 services. Services were in tum disaggregated into transportation & communications, wholesale & retail trade, and other services. Tradables are probably most appropriately identified with agriculture and manufacturing, nontradables with the other non-oil categories. All the above variables were in per capita, constant local currency terms. As explanatory variables, we used data on the terms of trade, a debt crisis dummy, and the world real interest rate. All variables were at an annual frequency. A full description of the data used and its sources is in Appendix 1. IV. Econometric framework and issues. A. Estimation Throughout, we use a panel ier than a time-series methodology since, if the underlying assumptions are satisfied, this allows for more precise estimates and renders omitted variable bias a less serious problem (see next sub-section). We have no strong priors, and dteory provides little guidance on the speed at which our variables adjust to terms-of-trade shocks. Therefore we take what we think is a flexible approach to the issue of dynamics. As a first pass, we try to pick out long-ran relationships by selecting years when sufficient time has passed under a given regime for the countries to be in long-run equilibrium. Specifically, we note that 1973 was the end of a long period of stable oil prices; in 1981, after eight years of rising prices, agents geneally accepted the higher prices as prmanent, and had had some time to adjust to them; in 1989, after an eight-year period of falling prices, agents likewise accepted and had had some time to adjust to the lower prices; and, after 1989, oil prices rosr. sharply around the time of the Gulf War. Thus in 1973, 1981, and 1989 the countries in our sample could arguably have achieved long-run equilibrium. We therefore use these three benchmark years to carty out fixed-effects estimation of the following panel regression equation: y -t a t,+ TOT,+ en , (1) where y denotes the natural log of each of the variables we study (except for the trade balance, which is left in levels), TOT denotes the log of the terms of trade, the i subscript refers to countries, and t refers to page 9 time. The elasticity of each variable with ruspect to t toerms of trde is given by P. The regression results are given in Appendix 11, table 1. In addition, we also use the full data set to carry out fixed-effects estimation of the following regression equation, which allows for lagged responses: y - g + ITOT1 + P2TOTIr, + yCO) OL1, + e(, !2) where CONTROL denotes the control variables, which potentially include the debt crisis dummy and the world real interest rate. Because there is little theoretical guidance available on when and whether we should include the debt crisis dummy, we excluded it where it was insignificant. The results for the world real interest rate were hard to interpret, and we always excluded it. The only exception were the investment equations where, given the strong theoretical priors, we always controlled for the debt crisis and for the world real interest rate. The short-run elasticity of each variable with respect to the tens of trade is given by J,, the steady-state elasticity by (P, + N3). The regression results are given in Appendix II, table 2, and are discussed in the next section. Note that our specification of equations (1) and (2) makes e important assumptions. First, the random efrors es, have zero mean and are i.i.d. over time and across countries. Second, the intercept a varies across countries (but not over time), so that there are country-specific cffects. I subject this hypothesis to the following Laange Multiplier test, due to Breuisch & Pagan (1980). Let a be a random term with cross- country variance a.. Let the null hypothesis be that there are no country-specific effects, that is, Ho: a. =0, versus HI: a. > 0. Let e,t denote the residuals from the standard OLS regression with homogeneous intercepts. Under Ho, the test statistic LAM - N T I3 d X21). 2(T-1) iS page 10 The asymptotic P-values are reported in Appendix . In all regressions, LM is significant at the .0001% level; we therefore reject the null. Note that, since the test statistic has a block-diagonal information matrix, this pre-test does not affect the standard errors of the other estimators we compute. Third, we assume that the elasticities with respect to the terms of trade (the p coefficients) are constant over time and across countries. We test both of these hypotheses. To investigate temporal stability, we split up the sample into the subperiods 1965-1980 and 1981-1989; since they roughly correspond to the periods of rising and of failing terms of trade, this also allows us to check whether the responses to positive and to negative terms-of-trade shocks differ. As detailed in Appendix IL table 3, the null hypothesis of identical b coefficients across the two subperiods is generally supported by the F-test described in Hsiao (1986), chapter 2.2, equation 2.2.20; where the differences across subperiods are statistically significant, they are typically quantitatively unimportant. This suggests there are no significant asymmetries in the response to rising and to falling terms of uade. In contrast, as detailed in Appendix m, the null hypothesis of stability across countries is always rejected at the 1% significance level by the F-test described in Hsiao (1986), chapter 2.2, equation 2.2.15. The rejection is hardly surprising, given that we are dealing with such a broad sample of countries, but it does imply that we must look at country-specific regressions to determine how representative the panel estimates are. These country-by-country regressions are accordingly presented in Appendix m. Finally, note that our fixed-effects estimator is only BLUE if we interpret the country-specific effects a1 as fixed regressors. To forecast an out-of-sample response to a terms-of-trade shock, we must however treat the a%'s as random errors; our estimator is then in general not BLUE, but is still preferable to the random-effects estimator since it is consistent even when the az's are correlated with the explanatory variables. B. Omitted variable bias The issue of bias from omitted variables needs to be treated differently in a panel context. To see this, it is convenient to define the mean across time for any given country and variable, x, as: page 11 Xi - (4) and the deviations from country-means as: Xs X. -X. (5) Re-written in terms of deviation data, equation (1) then becomes YU,- pTOT,+ 1 . (6) Say that the true model is instead given by .= . PTOT,, yOAM + X , - (7) If equation (5) were estimated by OLS for each country, the bias b; in our estimator of tating the regressors as fixed, would be T E T6Tt , y O°MEDI( ,bf T (8) E ToT2 Hence, a necessary and sufficient condition for the countrty-by-county OLS estimates of i to be consistent is that asymptotically the terms of trade be orthogonal to any omitted variables: Al. lim(T--) b,r = 0. page 12 Now say that equation (5) is estimated in a panel context. The bias in ouLr estmator of P is then Ir r E 7T&T * y 0MJZTAD,, b; I1 .1 Ir (9) T 6T - t*ib- I I (10) Hence, a necessary and sufficient condition for the panel esfimates of p to be consistent is that asymptotically a weighted average of the country-specific biases be zero: A:2. lim(T--) F,,aTUb r = 0. Clearly, Al implies A2 but not the reverse: there are many ways for assumption A2 to be satisfied without any of the countries individually satisfying assumption Al. The important point is that for the panel estimates to be consistent, it is unnecessary for the terms of trade to be asymptotically orthogonal to the omitted variables in any country: what matters is that the country-specfic asymptotic biases be negatively correlated across countries so that they average to zero. This latter assumption is more credible to the extent that the relationship between the omitted variables and the terms of trade is idiosyncratic to each country and not positively correlated across countries. Of course, to the extent that the omitted variables are world variables which affect all countries, then the b; terms will tend to have the same signs and assumption A2 may fail. We therefore control for the debt crisis and for world interst rates. Et might also be desirable to control for world output, but since periods of rising oil prices coincided with changes in OECD activity that would have depressed growth in our sample, and vice-versa, this would probably strengthen the links we find between terms-of-trade changes and activity variables. page 13 The additional issue of the weights, a,, is unlikely to be important for our sample. Countries have larger than average weights to the extent that the sample variability of their terms-of-trade data is larger than average. Since our sample consists of oil exporters, the variability of their terms of trade is similar. C. Non-stationarity There are two ways to inltpret our econometric model. Frst, we may view the terms of trade (which are driven mainly by the oil price) as a fixed regressor. Equations (1) and (2) are then the mechanism generating the rest of the data. The presence or absence of stationarity is not an issue in this framework. Altemvatively, we may treat the terms of trade as being generated by some time-series statistical model. Here, the order of integration of our data is an issue. Unfortunately, little is known about how to test for and deal with nonsaionarity in a panel data setting. When estimating equation (2) for each country in isolation, we carry out an EngleGranger co-inwgration test, using the augmented Dickey-Fuller statistics. For most countries and many variables, we cannot reject the null hypothesis that the regression residuals are integrated of order 1. However, given that such tests are well known for their very low power, and given the paucity of observations at the single-country level, we do not feel there is much strong evidence that our data is non-stationary in the sense that it trends over time. V. Results A. The importance of Investment Overall, we find fairly strong evidence that the relationship between terms-of-trade shocks and investment is positive. The panel regression suggests the 95% confidence interval (0.500,0.675) for the long-run elasticity. In country by country regressions, 14 of the 18 countries have positive elasticities and of the 14, 13 are stastically significant This section examines what the available evidence says about the sectoral breakdown and the causes of this investment effect, and how this investment effect is crucial to understanding other effects such as those on the trade balance and on growth. page 14 It is likely that much of this investmnent occurred in the nontradable sector of the economy. Although data on investment by sector is simply not available for most of these countries,' and therefore this claim cannot be supported directly, much of the otber data we do have is supportive. First, there is strong evidence that the relative price of nontradables were positively associated with the terms-of-trade shocks. So there seems to have been a clear price incentive for investment in these sectors. Second, we do have evidence that in the long run output expanded in nontradable sectors such as construction and transportation. It seems reasonable to think that this rise was made possible by a higher capital stock in these sectors. This investment effi- in nontradables can be explained formally in a variety of intertemporal optimizing models of investment (recent examples in the literue include Schmidt-Hebbel & Serven (1993) and Warner (forthcoming, 1994)). These models usually conclude that aggregate private investment is determined by the ratio of Tobin's q (the present discounted value of future marginal value products of capital in a given sector) to the relevant price deflator for investment goods. Since the countries in our sample import many capital goods, and the world prices of capital goods did not change dramatically over the period, the price of nontradables relative to physical capital probably rose in the 1970's and declined in the 1980's, causing similar changes in investment in the nontradable sector. Tuming to the oil sector, one might think that relative price changes would also stimulate investment there. Although this would probably hold for countries experiencing exogenous changes in their export prces, it is important to remember that OPEC raised oil prices in the 1970's partly through a deliberate policy of output restriction, so that for much of our period the oil sector was engineering price changes rather than passively responding to them. In such a setting, where countries exercise their power in the world oil market, it is a priori unclear which way the incentives work for investment in the oil sector. Oil production quotas may be seen as a permanent ceiling on oil production, leaving little reason to invest. On the other band, an increase in productive capacity may strengthen one's hand in the periodic quota negotiations. X An exception is Wamer (1994), where it is shown that investment in Mexican private non-tradables did decline over 1981-85. That study also suggests that an important mechanism for the investment decline was the fall in the ratio of nontradable pnces to imported equipment prices, and that the terms-of-trade decline can explain most of the reduction in this relative price. page 15 Therefore, we cannot deduce from theoly whether investment should have risen or declined in the oil sector. We do know that real oil output per capita did not rise over the long run between 1973 and 1981. It may be that countries invested in modernizing the oil sector rather than in expanding its capacity. It may also be that they did invest in expanding capacity but just did not use this capacity fully over the period. We simply do not have the data to pin this down very precisely. However, it is clear that overall investment responded positively to the terms of trade, and this helps explain why the trade balance was strongly negatively related to terms-of-trade shocks, even though the export price effect in isolation should cause a higher trade surplus. B. Consumption, Savings, and the Current Account As discussed above, tory suggests that after a permanent positive terms-of-trade shock, consumption should rise, and savings should as a first approximation be unaffected. The data broadly confirms these implications. The only exceptions were Ecuador, Egypt, Syria, Kuwait, and the Emirates, which displayed an insignificant or indeed negative consumption response to the terms-of-trade shock. However, the former three countries were not really oil exporters before the first oil price shock, which consequently had a smaDler positive impact on their wealth. For Kuwait and the Emirates, the shock seems to have induced a large increase in the consumption of leisure, made up for by extensive immigration which pushed down average levels of consumption per capita. The data also suggests government consumption may respond almost twice as strongly as private sector consumption, perhaps because govemments systematically view shocks as more permanent than is the case. Numerically, panel estimates suggest the 95% confidence interval (0.310, 0.426) for the elasticity of consumption with respect to the terms of trade. Given the lack of an impact on savings, and the strong response of investment, the trade balance and current account should be expected to decrease, and without exception these were the estimated responses. C. Nontradable Prices and Quantities Terms-of-trade shocks are very strongly associated with real exchange rate appreciation, i.e., increases in the relative price of nontradables. Panel estimates of the elasticity suggest the 95% confidence interval (- page 16 0.606, -0.695), and only Congo displayed an insignificant response. The shocks are also associated with significant increase in value added within each category of nontradables. This confirms the importance of spending effects as a key mechanism in the transmission of terms-of-trade shocks to the rest of the economy: higher wealth leads to greater demand for nontradables, and hence to an increase in their relative prices and quantities. At a country level, the main exception was Ecuador, which displayed insignificant responses in the value added of each nontradable sector. D. Dutch Disease Effects The literature strongly focuses on Dutch Disease effects; here, the two non-oil tradables (agriculare and manufacturing) are the closest approximation we have to a Dutch Disease sector. Yet we failed to detect clear contractions in either of them in response to a rise in the price of oil; the only exception was Nigerian agriculture. Oil value added did respond negatively over the subperiod 1965-198 1, but this simply reflects the imposition of OPEC quotas: the oil sector is best not viewed as facing exogenously given terms of trade, since part of OPEC's strategy at the time clearly was to engineer price rises by restraining output. The absence of Dutch Disease effects may be partly explained by either the compression in the output of the oil sector, or by its typically being an 'enclave' sector which does not participate in domestic factor markets. Both factors would act to reduce the pull of resources away from other sectors, as mentioned in the theoretical section. Nevertheless, as discussed above, the spending effect was operational, increasing relative prices and output in the nontradable sector. To the extent that the nontradable sector may be competing with agriculture and manufacturing for scarce factors of production, or if nontradables are themselves intermediate inputs into other sectors, one might have expected a small bout of the Dutch Disease. E. Growth Given that the relative prices between oil and other products changed so dramatically over the period, and that the oil sector is so important in our sample, extremely serious index number problems arise in even defining and measung aggregate output. We therefore feel that it is not analytically useful to talk about the aggregate economy, and we instead separate the oil from the non-oil sector. The former has already been mentioned. In the non-oil aggregate, we do see an effect on GDP, driven mainly by the expansion page 17 in nontradables; Ecuador and Nigeria were the only countries with insignificant responses. Given our evidence on investment, this response is best seen as being driven by capital formation in the nontradable sector. Numerically, panel esimates of the elasticity of non-oil value added with respect to the terms of trade suggest the 95% confidence interval (0.206, 0.297). This is however influenced by outliers; the median elasticity from the country regressions is 0.381. VI. Some Warnings regarding Extrapolation When interpreting and above aU extrapolating our results, several issues must be considered. First, the causes of the oil price shock were different from the causes of the forecast future increase in primary commodity prices. IJ our sample export prices rose because of oligopolistic coordination aimed at reducing output. Some responses might well be different for a truly exogenous terms-of-trade shock. In particular, both investment and output in the favored sector would probably rise; since output was no longer constrained, we should also expect a bigger increase in wealth and hence consumption. Second, our results need only be valid for a permanent change in the terms of trade. A temporary terms-of- trade shock may lead to some intertemporal substitution in consumption and investment, but is unlikely to cause a permanent change in output. Third, an economy's response to an increase in the price of primary commodities will clearly hinge on how dependent it is on primary exports. Its investment response will also depend on what share of its capital goods is imported. Fourth, we might expect the availability of external finance to be a crucial factor interacting with changes in the terms of trade, and oil exporters in the 1970's probably enjoyed better access to the world capital markets an is true of the developing world today. On the other hand, this may well change if the trends in commodity prices change. Fifth, in the 1970's and 1980's oil-exporters wasted a lot of their windfall revenues on prestige projects with little impact on output. Presumably, developing countries have by now learnt the lesson, and next time around will make better use of their luck. page 18 VII. Conclusions We considered a panel of oil-exporters that import a significant fraction of their capital goods. We found that permanent terms-of-trade shocks have no impact on savings, a strong positive impact on investment, and a negative impact on the current account. There is also evidence of a long-run effect on output, particularly of non-tradables. Real exchange-rate appreciations are a key mechanism in triggering the resource reallocation. We failed to find any evidence that the Dutch disease is a major phenomenon. Finally, the response of expenditure to terms-of-trade shocks is not very sensitive to whether the expenditure comes from the public or private sector. As discussed in section VL estimating an in-sample response to a terms-of-trade shock is much easier than forecasting an out-of-sample response. While our qualitative results may hold in quite general contexts, our quantitative estimates almost certainly will not. Let us nevertheless try to obtain some idea of the potential impact of the forecast increase in commodity prices on the growth rate of commodity exporters. As a lower-bound scenario, assume that the commodity sector does not grow at all, that the non-commodity sector accounts for 75% of GDP, and that it responds like the non-oil sector in our sample. An increase in the trend growth rate of commodity prices from -1.5% p.a to 0.7% p.a. then implies an increase in the order of 0.6% in the trend growth rate of per capita GDP at constant prices. To the extent that commodity output increases, and that this does not pull resources away from the rest of the economy, the growth effect is magnified. page 19 Appendix I: Full Data Description and Sources. We consider the period 1965-1989. Our sample consists of the 18 oil-exporters listed below. Algeria Gabon Mexico Trinidad and Tobago Bahrain Indonesia Nigeria United Arab Emirates Congo hran Oman Venezuela Ecuador Lrq Saudi Arabia Egypt Kuwait Syria For all countries, fuels accounted for over 50% of total exports over at least half the sample period; for most countries, fuels accounted for over 70% of total exports over at least three quarters of the sample period. We used data on the following variables: (1) Terms of Trade (TOT), computed as Merchandise exports deflator US Dollar (USD) I merchandise imnports deflator USD, base 1987 = 1.0. (2) A Debt Crisis dummy (DEBT), set to unity for Ecuador, Mexico, Nigeria and Venezuela over 1982- 1989, and to zero elsewhere. (3) World Real Interest Rates (INTRATE), as computed by Barro and Sala-i-Martin. (4) The Real Exchange Rate (REALEXRA), or relative price of tradables to nontradables. This is computed as USA GDP derlator USD I local GDP deflator USD. index 1987. (5) Gross Domestic Product (GDP), per capita, constant 1987 market prices, Local Currency Unit (LCU). (6) Gross National Product (GNP), per capita, constant 1987 market prices, LCU. (7) Consumption (CONS), per capita, constant 1987 prices, LCU. (8) Private Consumption (CONSPRIV), per capita, constant 1987 prices, LCU. (9) General Government Consumption (CONSGOVT), per capita, constant 1987 prices, LCU. (10) Gross Domestic Savings (SAVINGS), per capita, constant 1987 prices, LCU. (11) Fixed Investment (INVT), per capita, constant 1987 prices, LCU. (12) Private fixed Investment (INVTPRIV), per capita, constant 1987 prices, LCU. (13) General Government fixed Investment (INVTGOVT), per capita, constant 1987 prices, LCU. (14) Balance of merchandise Trade (TRADBALA), per capita, constant 1987 prices, LCU. (15) Value added in the Oil sector, constant prices (OILVA), LCU. (16) Value added in the Non-oil sector (NONOILVA), constant prices, LCU. (17) Value added in Agriculture (AGRICULT), constant prices, LCU. page 20 (18) Value added in Manufacturing (MANUFACT), constant prices, LCU. (19) Value added in Construction (CONSTRUC), constant prices, LCU. (20) Value added in Public Utilities (UTILITY), constant prices, LCU. (21) Value added in Services (SERVICES), constant prices, LCU. (22) Value added in Transportation and Communications (TRANSPORT), constant prices, LCU. (23) Value added in Wholesale and Retail Trade (TRADE), constant prices, LCU. (24) Value added in Other Services (OTHRSERV), constant prices, LCU. (25) Total Labor force (LABOR). Variables 1, 4.. 14 and 25 came from the World Ban's DAD database (except for Mexico's terms of trade, obtained from Mexico's central bank), and variables 2 and 3 are as described above, They are available for all countries for the whole period. Variables 17, 18 and 21 were from the World Bank's STARS database, and all other variables came from World Bank Country Economic Memoranda. They were not available for Iraq, nor before 1969. In all cases, the ultimate sources are national central banks, national statistical services, and estimates by World Bank missions. page 21 Appendix I: Full Panel Regression Results. Table 1. Panel regression: In yi, = a, + 1B In TOTi, + e, . Sample period: 1973, 1981, 1989. Dependent variable Terms of trade Terms of trade T- N2 N Coefficient ratio Ln Real exchange rate -0.5617 -6.8338 0.5716 0.3973 54 Ln Consumption 0.3594 4.4355 0.3598 0.3917 54 Ln Private Consumption 0.3413 4.2327 0.3386 0.3898 54 Ln Govt Consumption 0.4085 4.1049 0.3250 0.4810 54 Ln Savings 0.0312 0.2177 0.0014 0.6939 54 Ln Investment 0.4780 4.3860 0.3547 0.5267 54 Ln Private Investment 0.4895 3.5540 0.2652 0.6656 54 Ln Govt Investment 0.5731 4.5936 0.3761 0.6030 54 Trade Balance -15.1420 -2.8571 0.1936 24.9415 53 Ln GDP -0.0123 -0.1613 0.0007 0.3689 54 Ln Oil value added -0.3862 -1.7479 0.2763 0.9133 23 Ln Non-oil value added 0.2211 4.4454 0.7118 0.2055 23 Ln Agriculture 0.0706 0.7940 0.0220 0.3787 46 Ln Manufacturing 0.1710 1.2890 0.0767 0.5432 36 Ln Construction 0.3314 2.7997 0.4655 0.4852 25 Ln Public Utilities 0.2553 1.5724 0.2610 0.6039 21 Ln Services 0.0951 1.0117 0.0353 0.4003 46 Ln Transportation 0.1958 1.7262 0.2986 0.4323 22 Ln Trade 0.1419 0.8500 0.0936 0.5207 21 Ln Other services 0.2704 3.7993 0.6734 0.2713 22 page 22 Table 2. Panel regression: In Yi0 = cc + 3,B In TOTi, + 12 In TOTi,.I + .. + E,. Dependent variable Terms of trade sum Tenns of trade T- R2 I N of coefficients ratio Ln Real exchange rate -0.6508 -28.7336 0.6747 0.2488 431 Ln Consumption 0.3681 12.5055 0.2766 0.3248 431 Ln Private Consumption 0.2902 10.2829 0.2105 0.3114 431 Ln Govt Consumption 0.5206 12.9296 0.2892 0.4442 431 Ln Savings -0.0679 -1.1228 0.0636 0.6660 429 Ln Investment 0.5875 13.1991 0.3217 0.4706 413 Ln Private Investment 0.6831 11.7508 0.2902 0.6146 413 Ln Govt Investment 0.5970 12.3281 0.2874 0.5120 413 Trade Balance -15.4053 -9.2591 0.2145 15.9832 395 Ln GDP -0.0201 -0.7378 0.0158 0.3000 431 Ln Oil value added -0.3640 -6.3850 0.2470 0.3548 178 Ln Non-oil value added 0.2519 10.8723 0.4316 0.1441 178 Ln Agriculture 0.0701 2.2620 0.0691 0.2155 293 Ln Manufacturing 0.2306 3.7061 0.0803 0.4022 240 Ln Construction 0.3967 9.1806 0.3581 0.2709 193 Ln Public Utilities 0.3163 6.3042 0.3226 0.2787 158 Ln Services 0.0965 2.7620 0.0276 0.2438 293 Ln Transportation 0.2332 5.4244 0.1758 0.2467 175 Ln Trade 0.2331 4.6300 0.1578 0.2382 158 Ln Other services 0.2606 9.7659 0.3833 0.1531 175 For the real exchange rate, the trade balance, non-oil value added, agriculture, construction, utilities, and trade, the regressor matrix also includes DEBTj,; in all the investment equations, it also includes DEBT21 and In LNTRATE,. Sample period: 1965-1989. page 23 Table 3. Panel Regression: In y,1 = oc + 01 In TOT,, + 132 In TOT,,., + .. + ell. coeficients unrestricted across the subperiods 1965-1980 and 1981-1989. Dependent variable Terms of trade Terms of trade P-value for R2 N sum of coef. sum of coef. coef. 1965-1980 1981-1989 stability Ln Real exchange rate -0.6746 -0.5484 0.0235 0.6807 0.2471 431 Ln Consumption 0.3791 0.3110 0.25 0.2815 0.3245 431 La Private Consumption 0.3042 0.2276 0.403 0.2140 0.3115 431 Ln Govt Consumption 0.5395 0.4211 0.0931 0.2974 0.4428 431 Ln Savings -0.1439 0.1884 0.0894 0.0747 0.6637 429 Ln Investment 0.3124 0.3510 0 0.4505 0.4253 413 Ln Private Investment 0.4647 0.1246 0 0.3689 0.5819 413 Ln Govt Investment 0.2590 0.6564 0 0.4226 0.4627 413 Trade Balance -16.9939 -10.9972 0.419 0.2181 15.9906 395 Ln GDP -0.0488 0.0699 0.0893 0.0274 0.2990 431 Ln Oil value added -0.4854 -0.0789 0.0258 0.2807 0.3491 178 Ln Non-oil value added 0.2449 0.2710 0.703 0.4341 0.1448 178 Ln Agriculture 0.1125 0.0087 0.259 0.0784 0.2153 293 Ln Manufacturng 0.2021 0.2877 0.281 0.0908 0.4018 240 Ln Construction 0.3898 0.4083 0.76 0.3601 0.2721 193 Ln Public Utilities 0.3213 0.3142 0.713 0.3258 0.2802 158 Ln Services 0.0842 0.1241 0.399 0.0341 0.2439 293 Ln Transportation 0.1601 0.4157 0.0798 0.2020 0.2445 175 Ln Trade 0.2396 0.2250 0.936 0.1585 0.2400 158 Ln Other services 0.2597 0.2674 0.53 0.3883 0.1536 175 For the real exchange rate, the trade balance, non-oil value added, agriculture, construction, utilities, and trade, the regressor matrix also includes DEBTk; in all the investment equations, it also includes DEBT,, and In INTRATE. Sample period: 1965-1980, and 1981-1989. The P-value denotes the minimum significance level at which we can reject the null hypothesis of identical terms-of-trade coefficients across subperiods, using the F-test described in Hsiao (1986), ch. 2.2, equation 2.2.20. page 24 Table 4. Share of capital goods in total inports in 1973, in 1981, and in the latest avaiable year up to 1989 inclusive. Share of capital goods in the year Latest available year- Country 1973 1981 Latest 1989 if not stated Algeria 0.3720 0.3834 0.2570 Congo 0.3868 0.3874 0.3620 1986 Ecuador 0.4117 0.4582 0.3397 Egypt 0.2477 0.2816 0.2303 Gabon 0.4066 0.4094 0.4009 1983 Indonesia 0.4104 0.3536 0.3774 Iran 0.3779 0.2832 0.3884 1983 Iraq 0.3297 0.5328 0.4478 1983 Kuwait 0.3442 0.4103 0.2961 Mexico 0.4438 0.4707 0.3260 Nigeria 0.4012 0.4402 0.3812 1986 Oman 0.3101 0.3908 0.3685 Saudi Arabia 0.3524 0.4047 0.3870 Syria 0.2369 0.2171 0.2592 1986 Trinidad and Tobago 0.1313 0.2242 0.2729 United Arab Emirates 0.3801 0.3568 0.3025 1986 Venezuela OA688 0.4344 0.4682 1988 page 25 Appendix III: Country-by-country Regression Results. Sample period: 1965-1989. For each dependent variable, two regressions are carried out. Regression 1: In yj, = uxi + P(L) In xk, + e,1, j identical across countries. Results are in the form: Y variable TOT Coef TOT T-stat Debt Coef Debt T-stat IntRate Coef IntRate T-stat R2 a N P-value Regression 2: In yt = a; + 3 P(L) In xt + ej,, ,B unrestrcted across countries. Results are in the form: Country TOT Coef TOT T-stat Debt Coef Debt T-stat IntRate Coef IntRate T-stat R2 a N P-value The P-value in regression I refers to the Lagange Multiplier test of the null hypothesis of no country- specific effects; the P-value in regression 2 to the F-test of the null hypothesis of coefficient stability across countries. Where a coefficient and its T-statistic are not reported, the corresponding variable was not included in the regression. REALEXRA -0.651 -28.7 0.325 6.07 0.675 0.249 431 4.21e-17 Algeria -0.447 -6.82 Bahrain -2.53 -13.6 Congo -0.0493 -0.257 Ecuador -0.263 -1.89 0.166 1.87 Egypt -0.818 -3.69 Gabon -0.629 -6.29 Indonesia -0.467 -5.25 hran -0.691 -11.5 Iraq -0.866 -12.3 Kuwait -0.728 -13.4 Mexico -0.637 -2.29 0.178 1.53 Nigeria -0.731 -8 0.489 4.99 Oman -0.853 -15.5 Saudi Arabia -0.783 -14.8 Syria -0.214 -1.55 Tobago -1.14 -5 Emirates -0.577 -9.97 Venezuela -0.359 -5.48 0.354 3.28 0.804 0.203 431 2.7e-14 CONS 0.368 12.5 0.277 0.325 431 5.81e-07 Algeria 0.488 6.03 page 26 Bahrain 0.917 4 Congo 0.145 0.613 Ecuador -0.261 -1.57 Egypt -0.64 -2.34 Gabon 0.682 5.54 Indonesia 0.596 5.45 Iran 0.292 3.95 Iraq 0.186 2.15 Kuwait -0.00163 -0.0244 Mexico 0.276 0.993 Nigeria 0.327 3.25 Oman 0.671 9.9 Saudi Arabia 1.09 16.7 Syria -0.513 -3 Tobago 0.794 2.84 Emirates -0.0429 -0.603 Venezuela 0.225 3.32 0.609 0.25 431 0 CONSPRIV 0.29 10.3 0.21 0.311 431 1.94e-10 Algeria 0.48 6.25 Bahrain 0.538 2.47 Congo 0.202 0.898 Ecuador -0.23 -1.45 Egypt -0.666 -2.56 Gabon 0.539 4.6 Indonesia 0.544 5.23 Iran 0.256 3.64 Iraq 0.0561 0.682 Kuwait 0.00356 0.056 Mexico 0.297 1.12 Nigeria 0.312 3.27 Oman 0.425 6.59 Saudi Arabia 0.99 16 Syria -0.502 -3.09 Tobago 0.856 3.21 Emirates -0.232 -3.42 Venezuela 0.23 3.57 0.581 0.237 431 0 CONSGOVT 0.521 12.9 0.289 0.444 431 3.82e-08 page 27 Algeria 0.51 4.34 Bahrain 2.21 6.62 Conglo -0.0562 -0.163 Ecuador -0.454 -1.88 Egypt -0.501 -1.26 Gabon 1.03 5.74 Indonesia 0.995 6.25 Iran 0.418 3.89 Iraq 0.371 2.95 Kuwait -0.0119 -0.123 Mexico 0.106 0.263 Nigeria 0.414 2.83 Oman 1.03 10.5 Saudi Arabia 1.23 13 Syria -0.557 -2.25 Tobago 0.617 1.52 Emnirates 0.329 3.18 Venezuela 0.195 1.97 0.567 0.363 431 0 SAVINGS -0.0679 -1.12 0.0636 0.666 429 9.34e-16 Algeria 0.462 2.37 Babrain 2.76 4.5 Congo 0.015 o.0264 Ecuador -1.14 -2.83 Egypt -0.212 -0.322 Gabon 0.787 2.65 Indonesia 1.4 5.31 Iran 0.0746 0.419 Iraq -0.258 -1.24 Kuwait -0.76 -4.72 Mexico 0.256 0.383 Nigeria -0.186 -0.768 Oman 0.232 1.42 Saudi Arabia -0.288 -1.82 Syria -0.931 -2.27 Tobago 1.09 1.62 Enmirates -0.345 -2.01 Venezuela -0.271 -1.66 9.18e-11 0.303 0.601 429 --1 INVT 0.587 13.2 -0.368 -3.34 -o.00073 -11 page 28 0.322 0.471 413 4.16e-10 Algeria 0.743 6.04 -0.00298 -1.42 Bahrain 1.76 5.09 -0.00417 -1.99 Congo 0.282 0.618 0.00402 1.61 Ecuador -o.246 -0.798 0.0597 0.248 -0.00117 -0.39 Egypt -1.3 -2.81 0.00772 3.83 Gabon 1.41 7.3 -0.00653 -3 Indonesia 1.41 8.5 0.00233 1.13 Iran 0.401 3.57 -0.00636 -3.04 Iaq 1.03 7.63 -0.00198 -0.955 Kuwait 0.316 3.19 -0.00202 -0.975 Mexico 1.01 1.89 -0.0707 -0.264 o.oD0483 0.19 Nigeria 0.624 3.67 -0.389 -1.78 -0.00697 -2.69 Oman 0.68 6.84 -0.00163 -0.785 Saudi Arabia 1.33 13.7 -0.000509 -0.245 Syria -0.554 -2.09 0.00413 1.98 Tobago 1.77 3.87 -0.000137 -0.0675 Emirates -0.0034 -0.0322 -0.00318 -1.53 Venezuela 0.25 2.11 -0.487 -1.93 -0.00265 -1.05 0.661 0.359 413 3.61e-27 i'VTPlRV 0.683 11.8 -0.675 -4.7 8.03e-05 0.0938 0.29 0.615 413 3.Ole-10 Algeria 0.756 5.12 -0.00581 -2.31 Bahrain 1.74 4.2 -0.00449 -1.78 Congo 1.08 1.97 -0.000525 -0.175 Ecuador -0.191 -0.516 0.0579 0.2 -0.00218 -0.603 Egypt -0.749 -1.35 0.00223 0.92 G3abon 1.29 5.57 -0.00215 -0.825 Indonesia 1.36 6.82 0.00169 0.683 Inn 0.524 3.89 -0.00478 -1.9 Irq 1.62 9.97 0.00611 2.45 Kuwait 0.0359 0.301 -0.000757 -0.305 Mexico 0.448 0.702 -0.141 -0.439 0.00053 0.173 Nigeria 0.495 2.43 -0.635 -2.42 -0.0115 -3.71 Oman 1.15 9.62 0.0094 3.77 Saudi Arabia 1.74 14.9 -0.00121 -0.486 Syria -0.54 -1.69 0.00603 2.41 Tobago 1.54 2.8 0.00132 0.545 Emirates -0.0215 -0.169 -0.00389 -1.56 Venezuela 0.00987 0.0694 -0.488 -1.61 -0.0044 -1.46 0.701 0.432 413 7.07e-38 page 29 lNVTGOVT 0.597 12.3 -0.173 -1.44 -0.00084 -1.18 0.287 0.512 413 6.48e-15 Algeria 0.737 5.3 0.00205 0.865 Babrain 1.8 4.61 -0.0035 -1.48 Congo -0.747 -1.45 0.0096 3A1 Ecuador -0.332 -0.953 0.0624 0.229 0.000259 0.0761 Egypt -1.88 -3.59 0.0115 5.03 Gabon 1.51 6.9 -0.0139 -5.68 Indonesia 1.5 7.98 0.00315 1.35 Iran 0.289 2.27 -0.00791 -3.34 Iraq 0.973 6.36 -0.00273 -1.16 Kuwait 0.444 3.96 -0.00245 -1.05 Mexico 2.05 3.41 0.072 0.238 0.000309 0.107 Nigeria 0.699 3.64 -0.266 -1.08 -0.00326 -1.11 Oman 0.583 5.18 -0.004 -1.7 Saudi Arabia 1.1 9.97 -0.000543 -0.231 Syria -0.537 -1.79 0.00311 132 Tobago 2.06 3.98 -0.00255 -1.12 Emirates 0.00568 0.0475 -0.00271 -1.15 Venezuela 0.606 4.52 -0.471 -1.65 -0.00213 -0.747 0.616 0.406 413 4.86e-14 TRADBALA -15.4 -9.26 -11 -3.17 0.214 16 395 4.07e-13 Algeria -2.71 -1.21 Bahrain -2.55 -0.362 Congo -1.92 -0.267 Ecuador -3.2 -0.666 -0.246 -0.081 Egypt -0.615 -0.0809 Gabon -5.58 -1.64 Indonesia -85.3 -28.1 Iran -2.31 -1.03 Iraq -1.26 -0.506 Kuwait -1.37 -0.582 Mexico -1.96 -0.207 0.179 0.0449 Nigeria -5.06 -1.62 1.09 0.323 Oman -4.06 -0.781 Saudi Arabia -0.522 -0.277 Syria -0.784 -0.138 Tobago -9.82 -1.27 Emirates -1.73 -0.676 Venezuela -77.1 -32.2 -1.03 -0.281 0.868 6.93 395 3e-108 page 30 GDP -0.0201 -0.738 0.0158 0.3 431 Algeria 0.249 3.18 Bahrain 0.051 0.231 Congo -0.065 -0.284 Ecuador -0.478 -2.97 Egypt -0.776 -2.94 Gabon 0.607 5.1 Indonesia 0.516 4.88 Iran 0.0491 0.687 Iraq -0.198 -2.38 Kuwait -0.589 -9.13 Mexico 0.261 0.972 Nigeria -0.0206 -0.212 Oman 0.352 5.37 Saudi Arabia 0.17 2.69 Syria -0.35 -2.12 Tobago 0.69 2.55 Emirates -o.324 -4.71 VJenezuela -0.142 -2.17 0.419 0.241 431 0 GNP 0.0163 0.621 0.0112 0.289 431 2.04e-14 Algeria 0.255 3.34 Bahrain -0.189 -0.871 Congo 0.00967 0.0432 Ecuador -.419 -2.66 Egypt -0.753 -2.92 Gabon 0.589 5.06 Indonesia 0.495 4.79 hran 0.0917 1.31 Iraq -0.146 -1.78 Kuwait -0.472 -7.48 Mexico 0.274 1.04 Nigeria 0.000775 0.00816 Oman 0325 5.07 Saudi Arabia 0.288 4.68 Syria -0.336 -2.09 Tobago 0.797 3.01 Emirates -0.313 4.66 Venezuela -0.126 -1.96 0.399 0.236 431 0 page 31 OILVA -0.364 -6.39 0.247 0.355 178 4.32e-43 Algeria -0.122 -0.916 Bahrain 1.1 1.51 Congo 1.09 1.01 Ecuador -0.138 -0.398 Egypt 1.94 2.38 Indonesia 0.00744 0.0191 Iran -0.776 -6.38 Kuwait -0.537 -2.16 Mexico -0.173 -0.46 Nigena 0.187 0.524 Oman -0.202 -0.326 Saudi Arabia -0.198 -2.01 Tobago -1.16 -1.04 Venezuela -0.559 -4.93 0.462 0.33 178 0.00347 NONOILVA 0.252 10.9 -0.0682 -1.72 0.432 0.144 178 8.75e-44 Algeria 0.336 7.39 Bahrain 0.235 0.952 Congo 0.92 2.51 Ecuador -0.0364 -0.355 0.137 2.32 Egypt 0.754 2.71 Indonesia 0.591 3.74 Iran 0.28 6.75 Kuwait 0.091 1.07 Mexico 0.417 2.48 0.102 1.48 Nigeria -0.00423 -0.0332 -0.117 -0.938 Oman 0.847 4.01 Saudi Arabia 0.381 11.3 Tobago 0.836 2.21 Venezuela 0.0778 1.72 -0.246 -3.29 0.721 0.113 178 3.54e-10 AGRICULT 0.0701 2.26 -0.167 -3.39 0.0691 0.216 293 2.33e-18 Algeria 0.183 2.08 Babrain 0.607 1.36 Congo 0.00972 0.0466 Ecuador -0.108 -0.631 -0.00651 -0.0722 Egypt 0.001 0.00387 page 32 Gabon -0.0547 -0.208 Indonesia 0.151 1.29 Ian 0.0964 1.18 Kuwait 0.17 2.21 Mexico 0.228 0.744 -0.079 -0.601 Nigeria -0.193 -2.11 -0.216 -2.06 Oman -0.203 -1.25 Saudi Arabia 0.0405 0.533 Syria 0.63 2.41 Tobago 0.955 4.13 Emirates 0.14 0.709 Venezuela -0.0647 -0.756 -0.081 -0.768 0.304 0.2 293 0.000138 MANUFACT 0.231 3.71 0.0803 0.402 240 3.25e-28 Algeria 0.298 1.84 Bahrain -0.227 -0.239 Congo -0.0879 -0.188 Ecuador 0.232 0.737 Gabon 0.278 0.578 Indonesia 0.918 4.26 Iran 0.199 1.33 Kuwait 0.233 0.73 Mexico 0.193 0.468 Nigeria 0.188 1.2 Oman -0.571 -1.92 Saudi Arabia 0.102 0.733 Tobago 0.922 2.18 Emirates 1.53 4.24 Venezuela 0.0121 0.0833 0.332 0.368 240 5.92e-05 CONSTRUC 0.397 9.18 -0.306 -4.19 0.358 0.271 193 1.44e-2B Algeria 0.714 9.07 Babrain 0.382 0.893 Congo 2.4 3.78 Ecuador 0.00434 0.026 -0.0965 -1.03 Egypt 0.945 1.96 Indonesia 0.707 3.08 Iran 0.053 0.739 Kuwait 0.133 0.906 page 33 Mexico 0.475 1.63 0.0674 0.567 Nigeria 0.64 2.91 -0.488 -2.26 Oman 1 2.75 Saudi Arabia 0.704 12.1 Syria 0.79 1.96 Tobago 1.9 3.93 Venezuela 0.232 2.95 -0.825 -6.37 0.732 0.195 193 2.5e-15 UTILIIY 0.316 6.3 0.259 3.21 0.323 0.279 158 1.85e-50 Bahrain 0.074 0.131 Congo 0.235 0.28 Ecuador -0.271 -1.22 0.603 4.86 Egypt 1.02 1.6 Indonesia 0.492 1.62 Iran 0.354 3.73 Kuwait 0.138 0.709 Mexico 1.56 2.2 0.429 2.46 Nigeria 0.491 1.68 0.0379 0.133 Oman 1.25 2.59 Saudi Arabia 0.406 4.94 Tobago 1.17 1.83 Venezuela 0.202 1.94 0.11 0.639 0.536 0.259 1SS 0.00748 SERVICES 0.0965 2.76 0.0276 0.244 293 1.22e-20 Algeria 0.114 1.16 Bahrain 0.357 0.713 Congo 0.588 2.52 Ecuador 0.0865 0.45 Egypt -0.449 -1.55 Gabon 0.277 0.942 Indonesia 0.497 3.76 Iran 0.0653 0.713 Kuwait -0.296 -3.43 Mexico 0.17 0.676 Nigeria 0.316 3.31 Oman 0.0162 0.0887 Saudi Arabia 0.188 2.2 Syria 0.985 3.36 Tobago 0.489 1.88 page 34 Emirates 0.19 0.859 Venezuela -0.0328 -0.369 0.277 0.225 293 2.01e-05 TRANSPORT 0.233 5.42 0.176 0.247 175 5.47e-52 Bahrain 0.962 1.99 Congo 1.18 1.64 Ecuador -0.00442 -0.0236 Egypt 1.63 2.99 Indonesia 0.636 2A6 hban 0.428 5.28 Kuwait 0.106 0.641 Mexico 1.58 2.73 Nigeria 0.37 1.55 Oman 1.46 3.53 Saudi Arabia 0.144 2.18 Syria 0.507 1.12 Tobago 1.25 2.3 Venezuela 0.0341 0.451 0.46 0.22 175 0.000119 TRADE 0.233 4.63 -0.132 -2.05 0.158 0.238 158 4.43e-47 Bahrain 1.1 2.75 Congo 1.59 2.69 Ecuador -0.0376 -0.242 0.111 1.27 Egypt 1.25 2.8 Indonesia 0.304 1.42 Iran 0.569 8.51 Kuwait 0.16 1.17 Mexico 0.344 1.27 0.0616 0.556 Nigeria 0.00838 0.0409 -0.0632 -0.315 Oman 1.04 3.06 Syria 0.443 1.18 Tobago 0.914 2.03 Venezuela -0.0189 -0.26 -0.373 -3.09 0.61 0.182 158 6.55e-10 OTHRSERV 0.261 9.77 0.383 0.153 175 1.14e-50 Bahrain -0.554 -2.02 Congo 0.634 1.56 page 35 Ecuador 0.0369 0.347 Egypt 0.816 2.65 Indonesia 0.437 2.98 hran 0.305 6.64 Kuwait -0.0263 -0.28 Mexico 0.389 1.19 Nigeria 0.00246 0.0182 Oman 0.576 2.46 Saudi Arabia 0.42 11.3 Syria 0.7 2.72 Tobago 0.861 2.78 Venezuela 0.0801 1.87 0.663 0.125 175 1.53e-08 page 36 References. Backus, D. (1993), 'Interpreting comovements in the trade balance and the terms of trade,' Journal of International Economics 34, p. 375-387. Bean, C.R. (1986), 'The Terms of Trade, Labour Supply and the Current Account,' Economic Journal Supplement 96, p. 3846. Breusch, T.S. & Pagan, A.R. (1980), 'The Lagrange Multiplier Test and its Applications to Model Specification in Econometrics,' Review of Economic Studies 47, p. 239-253. Bruno, M. & Sachs J. (1982), 'Energy and Resource Allocation: A Dynamic Model of the Dutch Disease,' Review of Economic Studies 49, p. 845-859. Corden, W. M. (1984), 'Booming Sector and Dutch Disease Economics: Survey and Consolidation,' Oxford Economic Papers 36, p. 359-380. Corden, W. M. & Neary J.P. (1982), 'Booming Sector and Deindustrialisation in a Smal Open Economy,' Economic Journal 92, p. 825-848. De Gregorio, J. (1992), 'Economic Growth in Latin America,' Journal of Development Economics 39, p. 59-84. Etherington, D. & Yainshet A. (1988), 'The Impact of Income Terms of Trade for Coffee on Capital Goods hnports and Investment in Ethiopia,' Eastern Africa Economic Review 4, p. 48-52. Gavin, M. (1990), 'Structural Adjustment to a Terms of Trade Disturbance,' Journal of International Economics 28, p. 217-243. Harberger, A.C. (1950), 'Currency Depreciation, Income, and the Balance of Trade,' Journal of Political Economy 58, p. 47-60. Hsiao, C. (1986), Analysis of Panel Data. Laursen, S. & Metzler LA. (1950), 'Flexible Exchange Rates and the Theory of Employment,' Review of Economics and Statistics 32, p. 281-299. Matsuyama, K (1988), 'Terms-of-Trade, Factor Intensities and the Current Account in a Life-Cycle Model,' Review of Economic Studies 55, p. 247-262. Morley, S. (1992), 'On the Effect of Devaluation during Stabilisation Programs in LDCs,' Review of Economics & Statistics 74, p. 21-27. page 37 Neary, J. P. & S. Wijnbergen (1984), 'Can an Oil Discovery Lead to a Rcccession? A Comment on Eastwood and Venables,' Economic Journal 94, p. 390-395. Obstfeld, M. (1982), 'Aggregate Spending and the Terms of Trade: Is There a Laursen-Metzler Effect,' Quarterly Journal of Economics 97, p. 251-270. Persson, T. & Svensson L.E.O. (1985), 'Current Account Dynamics and the Terms of Trade: Harberger-Laursen-Metzler Two Gencrations Later,' Journal of Political Economy 93, p. 43-65. Salter, W.E.G. (1959), 'Internal and External Balance: the Role of Price and Expenditure Effects,' Economic Record 35, p. 226-238. Schmidt-Hebbel, K & Serven L. (1993), 'Dynamic Responses to External Shocks in Classical and Keynesian Economies,' CEPR Conference Paper. Sen, P. & Turnovsky S.J. (1989), 'Deterioration of the Terms of Trade and Capital Accumulation: a Re-Examination of the Laursen-Metzler Effect,' Journal of Intemational Economics 26, p. 227- 250. Svensson, L.E.O. & Razin A. (1983), 'The Terms of Trade and the Current Account: the Laursen- Harberger-Metzler Effect,' Journal of Political Economy 91, p. 97-125. Warner, A. (1992), 'Did the Debt Crisis Cause the Investment Crisis?', Quarterly Journal of Economics 107, p. 1161-1186. (forthcoming, 1994), 'Did the Debt Crisis or the Oil Price Decline Cause Mexico's Investment Collapse?'. Wijnbergen, S. (1984), 'The "Dutch Disease": A Disease After All?', Economic Journal 94, p. 41-55. (1984b), 'Inflation, Employment, and the Dutch Disease in Oil-Exporting Countries: A Short- run Disequilibrium Analysis,' Quarterly Journal of Economics, p. 233-250. Policy Research Working Paper Series Contact Title Author Date for paper WPS1384 Fiscal Decentralization and David Sewell November 1994 G. Langton Intergovemmental Finances in the Chdistine I. Wallich 38392 Republic of Albania WPS1385 Fiscal Federalism Dimensions of Tax Robin Boadway November 1994 C. Jones Reform in Developing Countries Sandra Robens 37754 Anwar Shah WPS1386 EU Bananarama Ill Brent Borrell December 1994 G. llogan 33732 WPS1387 Fiscal Decentralization and the Size Jaber Ehdaie December 1994 C. Jones of Govemment: An Extension with 37699 Evidence from Cross-Country Data WPS1388 Does Voice Matter? For Public Samuel Paul December 1994 B. Moore Accountability, Yes 35261 WPS1389 Ruble Overhang and Ruble Shortage: Patrick Conway December 1994 L. Suki Were They the Same Thing? 33974 WPS1390 Regional Integration and the Baltics: Pirita Sorsa December 1994 J. Ngaine Which Way? 37947 WPS1391 Where in the World Is Population Jeff Kling December 1994 S. Falon Growth Bad? Lant Pritchett 38009 WPS1392 Some Economic Consequences of Jean-Paul Azam December 1994 C. Jones the Transition from Civil War to Peace David Bevan 37699 Paul Colrier Stefan Dercon Jan Gunning Sanjay Pradhan WPS1393 The Interface of Trade, Investment, J. Luis Guasch December 1994 J. Troncoso and Competition Policies: Issues and Sarath Rajapatirana 37826 Challenges for Latin America WPS1394 Macroeconomic Reform and Growth Lawrence Bouton December1994 R. Martin in Africa: Adjustment in Africa Christine Jones 39026 Revisited Mi;uel Kiguel WPS1 395 A Typology of Foreign Exchange Janine Aron December 1994 R. Luz Auction Markets in Sub-Saharan Ibrahim Eibadawi 39059 Africa WPS1396 Foreign Exchange Auction Markets Janine Aron December 1994 R. Luz In Sub-Saharan Africa: Dynamic Ibrahim Elbadawi 39059 Models for Aucton Exchange Rates Policy Research Working Paper Series Contact Title Author Date for paper WPS1397 Are Private Capital Flows to Uri Dadush December 1994 J. Queen Developing Countries Sustainable? Ashok Dhareshwar 33740 Ron Johannes WPS1398 The Cost of Air Pollution Abatement Raymond S. Haytman December 1994 E. Schaper David Wheeler 33457 Manjula Singh WPS1399 How Important to India's Poor is the Martin Ravallion December 1994 P. Cook Urban-Rural Composition of Growth? Gaurav Datt 33902 WPS1400 Technical and Marketing Support Brian Levy with December 1994 D. Evans Systems for Successful Small and Albert Berry, Motoshige Itoh, 38526 Medium-Size Enterprises in Four Unsu Kim, Jeffrey Nugent, Countries and Shujiro Urata WPS1401 Colombia's Small and Medium-Size Abert Berry December 1994 D. Evans Exporters and Their Support Systems Jose Escandon 38526 WPS1402 Indonesia's Small and Medium-Size Albert Berry December1994 D. Evans Exporters and Their Support Systems Brian Levy 38526 WPS1403 Small and Medium-Size Enterprise Motoshige Itoh December 1994 D. Evans Support Policies in Japan Shujiro Urata 38526 WPS1 404 The Republic of Korea's Small and Linsu Kim December 1994 D. Evans Medium-Size Enterprises and Their Jeffrey B. Nugent 38526 Support Systems WPS1405 Growth and Poverty in Rural India Martin Ravallion January 1995 WDR Gaurav Datt 31393 WPS1406 Structural Breaks and Long-Run Javier Le6n January 1995 R. Luz Trends in Commodity Prices Raimundo Soto 31320 WPS1407 Pakistan's Agriculture Sector Rashid Faruqee January 1995 F. Willie Is 3 to 4 Percent Annual Growth 82262 Suslainable? WPS1408 Macroeconomic Management and Jun Ma January 1995 C. Jones Intergovemmental Relations in 37754 China WPS1409 Restructuring Uganda's Debt Kapil Kapoor January 1995 E.Spano The Commercial Debt Buy-Back 35538 Operation WPS1410 Macroeconomic Effects of Terms- Nikola Spatafora January 1995 J. Queen of-Trade Shocks: The Case of Oil- Andrew Waner 33740 Exporting Countries