V. Terms of Trade and Economic Growth in the Former Soviet Union102
1. The economic literature on countries of the former Soviet Union (FSU) suggests that the transition process has been complex and experiences have varied. The actual onset of the transition differs among countries (Havrylyshyn, 1999), as does the pace of reforms. However, what is common to almost all the countries is that they experienced initial output declines, such that the first half of the 1990s can be broadly characterized as a period of output contraction. This period was followed by a recovery during 1996–97, followed, in turn, by a slowdown during 1998–99 as a consequence of the 1998 financial crisis in Russia, which coincided with a fall in the world prices of oil and other primary products. The subsequent pickup has been accompanied by an improvement in the terms of trade of almost all the countries in the region, especially net exporters of oil.
2. This sequence of events has raised an important question. This is whether the positive terms of trade shock, driven to some extent by rising oil prices, is the major factor behind the region’s improved growth performance, in particular in the Commonwealth of Independent States (CIS).103
3. Controlling for growth spillovers within the region might enhance the assessment of the impact of the terms of trade shocks on growth. The IMF’s World Economic Outlook (International Monetary Fund 2004, p. 47) suggests that “on the back of the strong regional economic growth momentum, economic activity in the low-income CIS-7 countries has generally expanded rapidly, although lagging reformers, including Uzbekistan, have gained considerably less.” Against this background, it would be useful to ascertain the extent of growth spillovers within the region and the link of such spillovers to reforms in individual countries.
4. This chapter looks into the impact of the terms of trade shocks on economic growth in the twelve countries of the CIS (the CIS-12) and the three Baltic states.104 The analysis excludes the tumultuous earlier years of the transition. It covers the period 1995–2004. The analysis differs from most of the growth studies related to transition economies on two fronts. First, it extends the analysis of economic performance in the FSU to the linkage to external terms of trade, an avenue that has not been explored very much, as far as we know.105 Second, it examines the growth impact of the terms of trade in an integrated framework that controls for trade links between countries in the region and the associated spillovers, as well as for progress in economic liberalization and other institutional reforms, and other important determinants of growth identified in the growth literature.
5. We try to answer five questions. First, what is the effect of the terms of trade shocks on economic growth? Second, have the growth impacts of the terms of trade shocks differed between net oil exporters and net oil importers? Third, are there spillovers from regional growth? Fourth, does intraregional trade matter for a country’s growth, including as a channel for regional growth spillovers? Finally, are the growth effects of the terms of trade shocks and the spillovers from regional growth different between lagging and advanced reformers?
6. The findings of the analysis can be summarized as follows. First, there is a significant positive impact of the terms of trade growth on economic growth, although it is not the only (and perhaps the major) factor. Second, the magnitude of the impact is almost 25 percent higher for net oil exporters than for importers, although statistically not different. Third, regional economic growth exerts positive externalities on individual countries’ growth. Fourth, after controlling for the terms of trade shocks, the share of regional trade in total trade is positively associated with growth. However, there is no evidence that trade is a channel through which regional growth spills over. Finally, progress in institutional and structural reforms does not affect the spillovers from regional growth. More specifically, there is no evidence that lagging reformers benefit less from regional growth than advanced ones. But progress in structural reforms affects the growth impact of the terms of trade shocks. In particular, the interaction of the terms of trade growth with an indicator of stabilization and reforms suggests that the more advanced the reforms, the greater the contribution of a terms of trade improvement to growth.
7. Our results are subject to a number of caveats, and, therefore, should be interpreted with caution. These caveats arise from biases associated with measurement problems, including the subjectivity of our reform measure, the potential nonstationarity of some variables, the short sample period, and the possible sensitivity to model specification.
8. The remainder of the chapter is organized as follows. Section B presents some stylized facts relevant to the analysis of the CIS-12 countries and the Baltic states. Section C reviews selected findings of the literature on terms of trade shocks and economic growth and on growth spillovers. Section D presents the empirical analysis. Section E concludes.
B. The CIS-12 Countries and the Baltic States: Stylized Facts
9. The 15 countries covered in this study have had growth experiences that show both similarities and differences. The similarities are in the output path—a substantial decline in the earlier years of the transition, followed by a gradual recovery. Great differences, however, have been observed in the depth and length of the decline, as well as the timing and strength of the subsequent recovery (Berg and others, 1999; and Havrylyshyn, 2001). By 1995, output had started to recover in the Baltics while still declining in the CIS-12 countries, except Armenia and Belarus. The first year of positive average growth for the 15 countries of the FSU is 1996. Since then, average real GDP growth has been positive and quite strong, except for the 1998–99 slowdown associated with the financial crisis in Russia and its aftermath.
10. Although many observers link the strong growth in many countries included in our analysis to the improvement in the external terms of trade, the evidence is not clear-cut. While the terms of trade have, on average, evolved favorably for the 15 countries during the ten-year period ended in 2004, there are noteworthy differences. The terms of trade have been less favorable to the Baltic countries than to the CIS-12 as a group (Table 1). Moreover, within the CIS, the seven net fuel importers have faced less favorable terms of trade than the five net fuel exporters—Azerbaijan, Kazakhstan, Russia, Turkmenistan, and Uzbekistan. Interestingly, although, they have faced less favorable terms of trade than the CIS-12 countries, the Baltic states have experienced a higher annual average rate of economic growth during the ten years ended 2004—5.6 percent compared with 4.5 percent for the CIS-12. Also, in the second half of the ten-year period to 2004, their growth rate tracked that of the CIS-12, despite significantly less favorable terms of trade (Figures 1 and 2; Table 3). Among the CIS-12 countries, Armenia has faced the most adverse terms of trade shocks, but its average growth has been above the average of both the Baltics and the CIS-12. Russia, meanwhile, is among the five countries with the largest average increase in the terms of trade and one of the three with the lowest average growth among the 15 during 1995–2004 (Table 1). This suggests that factors other than the terms of trade have shaped growth.
Figure 1.Real GDP Growth in the CIS and the Baltics, 1995–2004
Source: Authorities; and IMF staff calculations.
Figure 2.Terms of Trade Shocks and GDP Growth in CIS and Baltics, 1995–2004
(Average, in percent)
Source: Authorities; and IMF staff calculations.
|Average real GDP growth||Average terms of trade growth||Average inflation||Average BMGDP 1/|
|Average, excl. Russia||1.48||8.12||4.80||2.53||3.29||2.91||100.90||13.21||57.06||14.94||20.05||17.49|
|CIS, excl Russia||0.72||8.46||4.59||2.45||4.04||3.24||124.56||16.15||70.36||12.83||16.48||14.65|
|Oil exporters, excl. Russia||0.67||10.22||5.45||5.46||8.26||6.86||164.34||13.11||88.73||12.96||14.90||13.93|
|Oil importer CIS countries||0.74||7.46||4.10||1.16||2.23||1.69||101.83||17.89||59.86||12.75||17.38||15.07|
|Average real GDP growth||Average terms of trade growth||Average inflation|
|Average, excl. Russia||99.04||103.88||101.46||45.08||38.29||41.68||47.38||36.19||41.79|
|CIS, excl Russia||94.13||98.88||96.50||48.89||41.73||45.31||49.09||38.90||44.00|
|Oil exporters, excl. Russia||78.52||83.05||80.79||43.27||36.30||39.79||41.66||34.39||38.03|
|Oil importer CIS countries||103.04||107.93||105.48||52.10||44.83||48.46||53.34||41.48||47.41|
BMGDP is the ratio of broad money to GDP in percent. Average real GDP growth Average terms of trade growth Average inflation Average BMGDP
|All 15 countries||Baltics||CIS Excl. Russia||Russia|
11. Earlier studies have identified the strength and consistency of macroeconomic stabilization and structural reforms as important determinants of growth differences among transition countries (Havrylyshyn and others, 1999). Differences in the reform effort have been identified between the Baltic states and the CIS-12 at the onset of the transition and thereafter. First, the pace of reforms has been very different. The Baltic states, which welcomed the transition as a liberation from the Soviet empire and an opportunity to access the Western club of nations, were more open to reforms from the start of the transition (Roland, 2001). Accordingly, they have made remarkable progress compared with the CIS-12. The Baltic states are currently rated on a par with industrial countries in many structural and institutional areas.106 Their objective of joining the European Union has been an important factor behind the faster and more consistent reform process. Second, and related to the first point, following the disruption of traditional trade and distribution channels of the Soviet era, the Baltic states have been more open to, and successful in, reorienting trade from the FSU to other economies, Western European countries in particular. Their trade with countries of the FSU as a share of total trade is smaller than that of the CIS-12. However, most CIS-12 countries have also made progress in opening themselves to markets outside the FSU in recent years.
12. The differences pointed out above between the CIS-12 and the Baltic countries regarding macroeconomic stabilization and structural reforms and the direction of trade are highlighted in Tables 1 and 2. We choose the index of economic freedom, published by the Heritage Foundation, to illustrate differences in macroeconomic policies and structural reforms.107 Havrylyshn and Rooden (2003) indicate that the overall index of economic freedom rating is based on the evaluation of institutional changes spanning a broad range of areas. They find the index to be thorough and reasonable, albeit still fundamentally subjective. On a scale of 1 to 5, where 5 represents the most liberalized economic environment, the rating of all 15 countries included in our analysis averaged 2.5. The average was 2.2 and 3.5 for the CIS-12 countries and the Baltic states, respectively (Table 1 and Figure 3). Regarding trade partnerships, during 1995–2004, regional trade as a share of total external trade averaged almost 43.7 percent for all the 15 countries, while it was 46.8 percent and 31.1 percent for the CIS-12 countries and the Baltic states, respectively (Table 1). Exports to the region as a share of total exports were also smaller for the Baltic states than for the CIS-12. Distinguishing between the growth effects of the terms of trade and those of policies and reforms, as well as other factors, requires an econometric analysis, which Section D provides.
Figure 3.Index of Economic Freedom for the CIS and the Baltics, 1995–2004
Source: The Heritage Foundation.
|Average, excl. Russia||2.21||2.57||2.43|
|CIS, excl Russia||1.94||2.24||2.14|
|Oil exporters, excl. Russia||1.56||1.98||1.83|
|Oil importer CIS countries||2.16||2.39||2.31|
The index ranges from 1 to 5, with 5 being the most free. See footnote 6 for full explanation of index.
C. Brief Literature Review
13. Our analysis relates to at least three areas of the economic growth literature, the first of which is growth in transition economies.Havrylyshyn (2001) is a very informative review of empirical studies on growth in transition economies. It indicates that most studies identify macroeconomic stabilization, economic liberalization, and market-friendly institutions as important determinants of growth. It also highlights the lack of significance, thus far, of traditional factors, investment in particular, in explaining growth.
14. The second area is the growth impact of the terms of trade, whose predictions have varied on both the theoretical and empirical fronts. Some theoretical studies predict that rapid growth in the terms of trade will bring about windfall gains that can undermine economic growth through Dutch disease-type effects (Corden, 1984; and Sachs and Warner, 1999), rent-seeking behavior that breeds corruption, or disincentives to pursue reforms and a coherent economic policy framework needed for long-term growth (Auty, 2001; and Sachs and Warner, 2001). A contrasting view comes from Mendoza (1997). He develops a small-economy, endogenous growth model that predicts that while the average rate of change of the terms of trade has a positive impact on average growth, its variability can result in slower growth if the coefficient of relative risk aversion is small.108 Another view suggests that the effects of terms of trade shocks are ambiguous, depending not only on whether the shock is temporary or permanent, but also on the rate of time preference and access to capital markets (Svensson and Razin, 1983).
15. Some empirical studies of the impact of the terms of trade on growth have found positive relationships. Several studies have included the growth or volatility of the terms of trade as explanatory variables in growth regressions. One strand of literature finds conclusive one-way relationships between the terms of trade and growth. For instance, Easterly and Rebelo (1993) and Fischer (1993) find that the rate of change of the terms of trade has a positive significant impact on economic growth, after controlling for policy variables and other determinants of growth. Likewise, Mendoza (1997) tests his theoretical model on a sample of 40 industrialized and developing countries covering 1971–91 and finds support for a robust positive relationship between the rate of change of the terms of trade and growth. His findings also support the robustness of a negative relationship between terms of trade volatility and growth. As regards Russia, a World Bank report (World Bank, 2003) focusing on the growth impact of oil price increases rather than improvement in the terms of trade in general suggests that there is a positive relationship between oil price increases and growth.
16. Another strand of empirical literature, however, finds varied or less conclusive results. Using a sample of industrial and developing countries covering 1970–88, Lutz (1994) finds no significant impact of growth in the terms of trade on economic growth, and an insignificant—or, in some instances, a positive and significant impact—of the volatility of the terms of trade on growth. When he divides his sample into subgroups of countries according to their income levels or the structure of their exports, he finds a negative significant impact of terms of trade growth on output growth for some groups.109
17. Like Lutz (1994), Turnovsky and Chattopadhyay (2003) and Blattman, Hwang, and Williamson (2004) find mixed results. Using a sample of developing countries covering the period 1975–95, Turnovsky and Chattopadhyay find that neither the growth rate of the terms of trade nor its volatility has a significant impact on output growth. When they split the sample into two groups according to the degree of volatility of the terms of trade, their findings are in line with Mendoza (1997) for the high-volatility group and the full-sample results in Lutz (1994) for the low-volatility group. Assessing the impact of the terms of trade on economic development during 1870–1939, Blattman, Hwang, and Williamson (2004) find that the effects or the core and the periphery are asymmetric.110 First, an improvement in the terms of trade is associated with output growth in the core, but not in the periphery. Second, increased volatility in the terms of trade has a significant negative impact on growth in the periphery, but not in the core.
18. The third area of the economic growth literature that our analysis relates to is that of spillovers. Many studies suggest that externalities across countries are relevant in explaining growth. Coe and Helpman (1995) suggest that foreign research and development (R&D) has beneficial effects on a country’s total factor productivity, and that these are stronger the more open a country is to foreign trade. Arora and Vamvakidis (2004) suggest that a country’s economic growth is positively influenced by both the relative income and the growth of its trading partners. In a more recent paper—Arora and Vamvakidis (2005)—they find that South African growth had a substantial positive impact on growth in the rest of Africa. The literature also consider capital movements as a channel through which growth can spill across national borders.
D. Empirical Analysis
19. We use a standard growth model augmented by several variables of interest:
where GRW is real GDP growth in percent, and controls includes variables that were significant in many other growth regressions in the literature. Z is the vector of our variables of interest. The βs are country-specific intercepts, and ε is the error term. The indices i and t represent countries and time periods, respectively.
20. The regressions contains several control variables usually included in empirical growth models:
Initial income (IGDP) is proxied by 3-lags of the natural logarithm of real per capita GDP at 1995 U.S. dollars. The use of lagged real GDP in regressions with fixed effects circumvents the unsuitability for fixed-effects regressions of income at the beginning of the sample time period, which is time invariant.
The investment ratio, commonly used to capture the impact of factor inputs;
The broad money-to-GDP ratio is used as a measure of financial development.
Inflation, measured as the natural logarithm of one plus the rate of consumer price inflation, expressed as a decimal, is meant to capture the impact of macroeconomic stabilization.
The trade-to-GDP ratio (trade) aims to reflect the impact of trade openness. The exports-to-GDP ratio (exports) is also used as an alternative measure of trade openness, following Levine and Renelt (1992).
The ratio of net capital flows to GDP (capital flows) tries to capture the impact of external finance. Net remittances is an alternative measure of external financial flows.
21. In addition to the “controls,” we include our variables of interest, which can be grouped in four categories. The terms of trade is our primary variable of interest.111 Previous research suggests that the terms of trade affect growth through their rate of change and the variability of the rate of change, a measure of risk. Therefore, we include in the regressions the following:
the annual percentage change in the terms of trade; and
the volatility of the terms of trade growth, which, following Lutz (1994), is computed as a short-run variance within each cross section, based on subsamples updated period by period.112 (Mendoza uses the standard deviation over the entire time period covered by his sample, but such time-invariant measure does not suit our fixed-effect regressions).
22. We include some indicators of regional growth and countries’ trade with regional partners to control for their effects on growth, thereby capturing potential spillovers effects. Specifically:
growth of all, determined for a country j, as , i≠j, where i refers to the countries in the regional sample, wi is the purchasing-power parity (PPP) weight of country i’s GDP in the region’s GDP, and n is the number of countries in the sample excluding j.
trade with all and exports to all represent, respectively, a country’s trade with, and exports to, the other countries contained in the sample in percent of its total trade.
23. To account for the role of reforms and institutions in economic development—a factor increasingly emphasized in the literature—we include a measure of structural progress. As indicated above, we choose the index of economic freedom published by the Heritage Foundation because of its comprehensiveness. As it encompasses macroeconomic stabilization, among other indicators of institutional progress listed in Havrylyshyn and others (1999) as key measures of reforms, its use lessens the need to include indicators of macroeconomic management or stability—exchange rate, government consumption, and fiscal deficit for instance—whose effects on growth have been found in the literature to be mixed.113
24. We rescale and standardize the raw indices published by the Heritage Foundation. First, as indicated in footnote 6, we rescale the ratings so that the higher the score, the higher the level of economic freedom a country enjoys. Furthermore, following Wei (2001), we standardize the rating by subtracting the mean and dividing by the standard deviation. With such a transformation, the coefficient of economic freedom in the growth equation will be interpreted as the response of real GDP growth to a one-standard-deviation increase in the rescaled economic freedom, that is, to an improvement in economic freedom.
25. Finally, we introduce interaction terms to capture indirect effects and dummy variables in order to distinguish subgroups of countries. Specifically, the dummy variable DUMOIL equals 1 if the country is a net oil exporter and 0 otherwise. DUMNOIL is the reverse of DUMOIL. Furthermore, we construct the following interaction variables:
growth of all*trade with all or growth of all*exports to all are used to explore the idea that trade is a channel through which regional growth spills across borders.114
growth of all*economic freedom is used to test the hypothesis that the gains from regional growth may be linked to reforms in individual countries.
terms of trade growth*economic freedom is used to test relevance of the “curse of natural resources.”
trade*tradewith all and exports*exports to all are used to explore the indirect effects of regional trade partnerships on growth through openness.
26. The data come from various sources. The share of trade with regional partners in total trade is computed from data published by the IMF’s Direction of Trade Statistics (DOTS). Real per capita GDP at 1995 U.S. dollars is from the World Bank’s World Development Indicators (WDI). The Index of Economic Freedom is from the Heritage Foundation as indicated above. Data for the remaining variables are from the IMF’s World Economic Outlook (WEO). We use annual data, as in many of the previous studies of growth in transition countries. This contrasts with some other growth regressions, which use data averaged over several years. In the data base, we have 15 cross sections, covering the period 1995–2004. However, because of some missing data, Turkmenistan has been dropped from the regressions. We use fixed effects to control for differences in countries’ characteristics not properly captured by variables included in the regressions.
27. We begin the analysis with a basic regression including the control variables,economic freedom, and our primary variable of interest, the growth of the terms of trade, as well as the volatility of the terms of trade growth. Columns (1) and (2) of Table 4 present the regressions with growth of the terms of trade, as well as control variables in the set of explanatory variables. Among the variables that are statistically significant, capital flows, inflation, and initial income are negatively associated with growth, while growth of theterms of trade, trade, broad money-to-GDP, and economic freedom are positively associated with growth. Investment ratio and volatility of the terms of trade growth are not significant. They are excluded from the regression in column (3), which is our baseline.
|terms of trade growth||0.053||***||0.052||***||0.514||***||0.087||**||0.070||**||0.101||***||0.070||0.070||***||0.047||0.075||**||0.071||**|
|capital flows (net, in percent of GDP, lagged)||-0.181||***||-0.163||***||-0.152||***||-0.142||***||-0.092||***||-0.144||***||-0.092||***|
|broad money-to-GDP (lagged)||0.480||***||0.497||***||0.481||***||0.510||***||0.476||***||0.500||***||0.440||***||0.544||***||0.452||***||0.52||***||0.48||***|
|inflation [1n(1+0.01*CPI inflation)]||-5.137||***||-4.787||***||-4.677||***||-4.64||***||-3.8||***||-4.1||**||-3.89||***||-3.1||*||-3.381||**||-3.93||***||-3.8||***|
|net remittances (in percent of GDP, lagged)||0.151||0.142||**||0.040||-0.140|
|growth of all||0.416||**||0.610||***||0.461||***||0.529||***||0.458||***||0.553||***||0.433||**||0.609||***|
|trade with all (lagged)||0.174||**||-0.063||-0.073||0.181||***|
|exports to all (lagged)||0.290||***||0.163||***||0.177||***||0.286||***|
|growth of all* trade with all (lagged)||-0.004||-0.002||-0.004||*||-0.007||**|
|growth of all* exports to all (lagged)||-0.01||***||-0.006||**||-0.01||***||-0.012||***|
|growth of all* economic freedom||-0.100||-0.237||**||0.070||-0.044||-0.002||-0.163||*||-0.162||*||-0.237||**|
|terms of trade growth* economic freedom||0.060||***||0.050||**||0.055||**||0.030||0.039||0.017||0.053||**||0.05||**|
|initial income (3-lag, ln per capita GDP)||-0.009||***||-0.008||***||-0.009||***||-0.008||***||-0.006||**||-0.001||-0.009||***||-0.01||***||-0.008||***||-0.008||***||-0.006||**|
|trade (trade in percent of GDP, lagged)||0.071||***||0.074||***||0.074||***||0.110||***||***||0.003||-0.016||0.10||***|
|trade*trade with all (lagged)||-0.001||*||0.001||0.001||-0.0007|
|exports (in percent of GDP, lagged)||0.320||***||0.274||***||0.290||***||0.319||***|
|exports*exports to all (lagged)||-0.004||***||-0.003||-0.004||***||-0.004||***|
|volatility of the terms of trade growth||-46.78|
|investment ratio (in percent of GDP, lagged)||0.045|
|Number of cross-sections||14||14||14||14||14||13||13||12||12||13||13|
|Number of observations||117||118||118||118||118||111||111||102||102||109||109|
Dependent variable: growth of real GDP.
Estimation using panels with fixed effects *, **, and*** denote significance at 10 percent, 5 percent, and 1 percent, respectively; based on White heteroscedasticity-consistent p-values.
28. In subsequent regressions, we include our additional variables of interest to try to answer the five questions that were raised in the introduction. To explore whether the growth effects of the terms of trade shocks differ between net oil exporters and importers, in the regressions summarized in Table 5 we also include two separate terms of trade shock variables (See Appendix 1). Overall, the best performing equations are those in column (5) of Table 4 and column (4) of Table 5. They suggest that a 10 percentage points increase in the growth of the terms of trade is associated with a 0.7–1.0 percentage point increase in the growth rate of the economy. Also, a one-standard deviation increase in the index of economic freedom strengthens the growth impact of a terms of trade shock by 0.05 to 0.06 percentage point.
|terms of trade growth*DUMOIL||0.053||0.055||0.1||**||0.104||**||0.117||*||0.089||0.092||*||0.063||0.090||**||0.086|
|terms of trade growth*DUMNOIL||0.053||***||0.05||**||0.082||**||0.077||**||0.098||**||0.067||0.070||0.036||0.077||**||0.064|
|capital flows (net, in percent of GDP, lagged)||-0.183||***||-0.166||***||-0.161||***||-0.107||***||-0.162||***||-0.105|
|broad money-to-GDP (lagged)||0.481||***||0.5||***||0.519||***||0.486||***||0.499||***||0.439||***||0.543||***||0.448||***||0.531||***||0.492|
|inflation [1n(1+0.01*CPI inflation)]||-5.147||**||-4.707||***||-4.4||***||-4.275||***||-4.044||**||-3.541||**||-3.050||-2.87||*||-3.635||**||-3.466|
|net remittances (in percent of GDP, lagged)||0.132||0.127||*||0.008||-0.192|
|growth of all||0.477||*||0.538||**||0.489||**||0.535||***||0.479||***||0.536||***||0.494||**||0.621|
|trade with all (lagged)||0.159||**||-0.052||-0.065||0.157||**|
|exports to all (lagged)||0.249||***||0.135||***||0.131||**||0.247|
|growth of all *trade with all (lagged)||-0.005||-0.001||-0.004||-0.007||**|
|growth of all *exports to all (lagged)||-0.007||**||-0.006||*||-0.011||***||-0.012|
|growth of all *economic freedom||-0.077||-0.137||0.103||-0.011||0.033||-0.144||-0.136||-0.211|
|terms of trade growth *economic freedom||0.062||***||0.059||***||0.057||*||0.029||0.045||0.016||0.057||***||0.050|
|initial income (3-lag, ln per capita GDP)||-0.009||***||-0.008||***||-0.008||***||-0.007||***||-0.01||***||-0.009||***||-0.01||***||-0.008||***||-0.007||***||-0.006|
|trade (trade in percent of GDP, lagged)||0.071||***||0.075||***||0.105||***||0.014||***||-0.004||0.092||***|
|trade *trade with all (lagged)||-0.001||*||0.001||0.001||0.000|
|exports (in percent of GDP, lagged)||0.308||***||***||0.285||***||0.297||***||0.294|
|exports*exports to all (lagged)||-0.004||***||***||-0.003||**||-0.003||***||-0.003|
|volatility of the terms of trade growth||-50.03||-51.897||-39.14||-30.58||-35.89||-28.47||-41.79||-46.93||-39.26|
|investment ratio (in percent of GDP, lagged)||0.045|
|Wald test for equality of the two terms of trade growth coefficients|
|Number of cross-sections||14||14||14||14||13||13||12||12||13||13|
|Number of observations||117||118||118||118||111||111||102||102||109||109|
Estimation using panels with fixed effects *, **, and *** denote significance at 10 percent, 5 percent, and 1 percent, respectively; based on White heteroscedasticity-consistent p-values.
29. Our results, while subject to a number of caveats, provide useful insights into the growth experience of countries of the FSU during 1995–2004. Also, they are broadly in line with findings of previous studies. These results can be summarized as follows.
30. Higher terms of trade growth boosts economic growth, and even more so if policies and reforms are advanced. The indicator of progress in reforms and macroeconomic stabilization, economic freedom, is the single most important determinant of growth in the baseline model. The significance of the coefficient associated with the interaction of economic freedom and the change in the terms of trade suggests that reforms are a channel through which the growth effects of terms of trade improvement can be enhanced. Specifically, the growth effects of an improvement in the terms of trade are larger the more advanced are the reforms.
31. The magnitude of the direct effect of terms of trade growth on economic growth is higher for net oil exporters than for net oil importers. Net oil exporters gain 25 percent more from an increase in the terms of trade growth than do net oil importers. Nonetheless, a Wald test for the equality of the coefficients associated with the terms of trade growth of the two groups fails to reject the hypothesis that the two coefficients are equal (Table 5).
32. Regional growth is unambiguously beneficial to individual countries’ growth. We also test the importance of reforms and trade partnerships for growth spillovers. We find that the advancement of reforms in a country does not affect the magnitude of the spillovers from regional growth. This undermines the idea that lagging reformers benefit less from regional growth than advanced ones. In fact, some regressions suggest the opposite: advanced reformers benefit less from regional growth spillovers than lagging ones.
33. The evidence on regional trade partnerships as a channel through which regional growth spills over is mixed. Also, this suggests that there are other channels than regional trade linkages through which regional growth spills over.
Trade with all regional partners as a share of total external trade is positively associated with growth, but its interaction with regional growth does not have a significant impact on growth.
The share of exports to regional partners in total exports is a significant determinant of growth. However, in its possible role as a transmission channel for regional growth, it weakens the magnitude of the growth spillovers. In particular, the spillovers from regional growth to an individual country are smaller the larger is the share of the country’s exports to the region in its total exports. This could be interpreted as indicating that, for a transition country of the FSU, concentration on regional markets for exports reflects the country’s insufficient progress in channeling its exports toward new, more profitable markets.
34. Trade openness is good for growth, and the direction of trade matters. Using exports-to-GDP and total trade-to-GDP ratios as indicators of openness, our analysis confirms the findings of previous studies that trade openness is good for growth. However, the significance of the negative coefficient of the interaction of exports-to-GDP and exports to all suggests that the greater the share of total exports directed to regional markets, the smaller the contribution of openness to trade. Therefore, openness with a greater concentration on regional markets seems to be “limited openness.”115
35. Capital flows have a negative and significant impact on growth. This result is somewhat surprising, considering that the sample consists of low- and middle-income countries, which could be expected to benefit from external capital to finance reforms and productive investment. Nonetheless, the negative association of economic growth with capital flows could suggest that capital flows have not been used properly, or that they may have been used to postpone reforms rather than finance them (Havrylyshyn and others, 1999).
36. Net remittances do not have a robust impact on growth. A positive and significant impact is detected in a couple of regressions, but such significance disappears when Moldova is dropped from the sample. Overall, the fit of the regressions was better when total capital flows rather than net remittances was used as an indicator of external financial flows.116
37. Financial development, macroeconomic stabilization, and initial income are also relevant in explaining growth. However, the coefficient of initial income is small, suggesting that convergence plays a secondary role in explaining growth differences among FSU countries. Although inflation is taken into account in the determination of the economic freedom, it is highly significant, providing robust support to the idea that macroeconomic stabilization is important for economic growth.117
38. The improvement in the terms of trade does not appear to be the major factor behind growth recovery in the FSU. As the empirical analysis suggests, many factors have played a nonnegligible role. This explains why some countries have had a higher (lower) growth rates notwithstanding more (less) favorable terms of trade. For instance, as indicated above, Armenia’s growth has been above the average for both the Baltics and the CIS-12 countries, although it has faced the most adverse terms of trade shocks among all 15 countries. The answer seems to lie in Armenia’s macroeconomic stabilization and the depth of its structural and institutional reforms.118
39. Our results are subject to a number of caveats, and, therefore, should be interpreted with caution. First, measurement problems associated with key variables such as the reform measure and the economic growth measure could bias the results. In particular, our reform measure—the Heritage Foundation Index of Economic Freedom—is inherently subjective. Also, the calculated GDP growth rates may be biased owing to the failure to sufficiently account for the underground economy, which could represent a large share of total output in some countries. Second, the results could be biased owing to the possible nonstationarity of some variables.119 Third, the short sample period precludes general conclusions about long-term growth. Finally, in light of the modeling difficulties that characterize the growth literature in general, results could be sensitive to model specification.
E. Concluding Remarks
40. Subject to the caveats mentioned above, our findings lend support to the idea that improvements in the terms of trade have contributed significantly to economic growth in the countries of the FSU. They also indicate that the impact on growth of changes in the terms of trade is statistically the same for both net oil exporters and importers. Beyond terms of trade shocks, there have been significant spillovers from economic growth in the region as a whole. To the extent that regional trade linkages have not been found to be significant in enhancing growth spillovers, the analysis suggests that externalities may take place through channels other than regional trade.
41. The analysis also highlights the importance of other determinants of growth identified in the previous literature. In particular, financial development, macroeconomic stabilization, and structural reforms have a positive impact on growth. The analysis also establishes a link between reforms and the impact on growth of a change in the terms of trade, a very important finding from a policy standpoint.120 In particular, the advancement of structural and institutional reforms enhances an economy’s capacity to manage resources from terms of trade improvements in a manner that does not undermine economic growth. This suggests that, where reforms and institutional development are advanced, they provide a channel through which the positive impact of favorable terms of trade shocks is enhanced and the adverse impact of negative terms of trade shocks is mitigated.
1. The first three columns of Table 4 show our basic regressions, while in subsequent regressions we experiment with different combinations of variables. Specifically, in column (4) of Table 4, we introduce growth of all, trade with all, and some interaction terms. In column (5), trade and trade with all are replaced with exports and exports to all, respectively. Growth of all and trade with all are positive and significant, but their interaction is not significant. Exports to all is positively associated with growth, while its interaction with either growth of all or exports is negatively associated with growth. economic freedom loses significance, but its interaction with terms of trade growth is positive and significant. The remaining interaction terms are not significant. Except for economic liberalization, all variables included in the baseline regression maintain their significance. Columns (6) and (7) have the same variables as columns (4) and (5), respectively, except that net remittances replaces capital flows as a measure of external financial flows. Net remittances is not significant in column (6) and the other variables, except trade with all and its interaction with trade, lose significance. In column (7), the coefficient of net remittances is positive and significant, and the coefficients of the other variables change drastically. To ascertain to what extent the regressions including net remittances among the explanatory variables are affected by potential outliers, in columns (6a) and (7a) we exclude Moldova from the cross sections, as it is the only country in the sample with net remittances far in excess of the sample average. Excluding Moldova, the coefficient of net remittances, which was positive and significant in regression (7), becomes negative and non significant. Also the coefficient of growth of the terms of trade declines, while remaining non-significant in column (7a), as it was in column (7).
2. Net remittancesseems to perform more poorly thancapital flowsas a proxy for external financial flows. For the full sample, the fit of the regressions including capital flows—columns (4) and (5)—is better than that of the corresponding regressions—columns (6) and (7)—which include net remittances. The same is true when Moldova is excluded from the sample. Not only is the fit of the regressions in columns (4a) and (5a) better than that of the regressions in columns (6a) and (7a), respectively, but also the coefficients are almost the same as in columns (4) and (5), suggesting that Moldova is probably not an outlier when capital flows rather than net remittances is the indicator of external financial flows.
3. To explore whether the growth effects of the terms of trade shocks differ between net oil exporters and importers, we include two separate terms of trade shock variables in the regressions. Terms of trade growth is interacted with DUMOIL and DUMNOIL in the regressions presented in Table 5. In the first two columns of Table 5, terms of trade growth* DUMOIL is marginally not significant and volatility of the growth of the terms of trade is marginally significant in column (2), which is the baseline for the new series of regressions. When additional variables of interest are included, the results compare with those in Table 4. Columns (3) and (4) compare with columns (4) and (5) in Table 4. The direct growth effect of terms of trade growth is positive for both net oil exporters and importers. The regressions presented in the remaining columns follow the same steps of including net remittances and checking for its explanatory power after dropping Moldova. Again, regressions in which external financial flows are measured by capital flows perform better than corresponding ones in which net remittances replaces capital flows. In particular, the regressions in columns (3), (3a), (4), and (4a) have a better fit than that of the regressions in columns (5), (5a) (6), and (6a), respectively.
2004, How Much Do Trading Partners Matter for Economic Growth?” IMF Working Paper 04/26 (Washington: International Monetary Fund).
2005, “The Implications of South African Economic Growth for the Rest of Africa,” IMF Working Paper 05/58 (Washington: International Monetary Fund).
2001, “The Political Economy of Resource-Driven Growth,” European Economic Review, Vol. 45 (May), pp. 839–46
2004, “The Impact of the Terms of Trade on Economic Development in the Periphery, 1870–1939: Volatility and Secular Changes,” NBER Working Paper No. 10600 (Cambridge, Massachusetts: National Bureau of Economic Research).
1999, “The Evolution of Output in Transition Economies: Explaining the Differences,” IMF Working Paper 99/73 (Washington: International Monetary Fund).others,
1995, “International R&D Spillovers,” European Economic Review, Vol. 39 (May), pp. 859–87
1984, “Booming Sector and Dutch Disease Economics: Survey and Consolidation,” Oxford Economic Papers, Vol. 36, pp. 359–80
1993, “Fiscal Policy and Economic Growth: An Empirical Investigation,” Journal of Monetary Economics, Vol. 32 (December), pp. 417–58
1993, “The Role of Macroeconomic Factors in Growth,” Journal of Monetary Economics, Vol. 32 (December), pp. 485–512
2001, “Recovery and Growth in Transition: A Decade of Evidence,” Staff Papers, International monetary Fund, Vol. 48 (Special Issue), pp. 53–87
Institutions Matter in Transition, But So Do Policies,” Comparative Economic Studies, Vol. 45 (March), pp. 2–24“
1999, “Growth Experience in Transition Countries, 1990–98,” IMF Occasional Paper No. 184 (Washington: International Monetary Fund).others,
2005. Available via internet at: http://www.heritage.org/research/features/index/downloads.cfm
World Economic Outlook, September2004, World Economic and Financial Surveys (Washington).
1992, “A Sensitivity Analysis of Cross-Country Growth Regressions,” American Economic Review, Vol. 82 (September), pp. 942–63
1994, “The Effects of Volatility in the Terms of Trade on Output Growth: New Evidence,” World Development, Vol. 22 (December), pp. 1959–75
1997, “Terms of Trade Uncertainty and Economic Growth,” Journal of Development Economics, Vol. 54 (December), pp. 323–56
2001, “Ten Years After… Transition and Economics,” Staff Papers, International Monetary Fund, Vol. 48 (Special Issue), pp. 29–52
1995, “Natural Resource Abundance and Economic Growth,” NBER Working Paper No. 5398 (Cambridge, Massachusetts: National Bureau of Economic Research).
2001, “The Curse of Natural Resources,” European Economic Review, Vol. 45 (May), pp. 827–38
1983, “The terms of Trade and the Current Account: The Harberger-Laursen-Metzler Effect,” Journal of Political Economy, Vol. 91 (February), pp. 97–125
World Bank, 2003, Russian Economic Report No. 6. Available via internet: http://siteresources.worldbank.org/INTRUSSIANFEDERATION/Resources/305499–1094736798511/518266–1097661221042/RER6-English.pdf.
2001, “Domestic Crony Capitalism and International Fickle Capital: Is There a Connection?” International Finance, Vol. 4 (Spring), pp. 15–45
Prepared by Mwanza Nkusu.
The CIS-12 comprises the Russian Federation—referred to hereafter as Russia, Ukraine, Kazakhstan, Belarus, Armenia, Azerbaijan, Georgia, Kyrgyz Republic, Moldova, Tajikistan, Turkmenistan, and Uzbekistan. The last seven of these are referred to as the “CIS-7.”
The Baltic states are Estonia, Latvia, and Lithuania.
A 2003 World Bank report on Russia focused mostly on the growth impact of oil price increases.
Most notably property rights and government intervention in economic activity.
The Heritage Foundation rates countries according to progress achieved on factors considered as most influential for the institutional setting of economic growth. These factors cover the following categories: trade policy, fiscal burden of government, government intervention in the economy, monetary policy, capital flows and foreign investment, banking and finance, wages and prices, property rights, regulation, and informal market activity (Heritage Foundation, 2005). Countries are rated on a 1 to 5 scale, where 1 represents the most liberalized economic environment. We have rescaled the rating to make 1 and 5 represent the least liberalized and most liberalized environments, respectively. Therefore, an increase in the rating represents an improvement in economic freedom.
Smaller than two more specifically.
These groups include high- and middle-income countries, separating these from less-developed countries (LDCs); oil exporters, separating these from exporters of other primary products and manufactured goods; and LDCs exporters of oil and other commodities, separating these from LDCs exporters of manufactured goods.
In the sample, the core is made up of the industrial leaders and latecomers—the United States and Western Europe—and the periphery comprises primary products exporters, including Canada, New Zealand, Australia, Latin America, Asia and the Middle East, and some European countries (Greece, Spain, Portugal, the Soviet Union, and Yugoslavia).
In our regressions, terms-of-trade growth is part of the baseline regressions.
More specifically, for each cross section, the cyclical component of the terms of trade, et is obtained from the following regression: ln TOTG = c0 + c1trend + et, where trend is the linear time trend, whose upper limit is the number of observations in the cross section. The variance of et is a measure of volatility of the terms of trade (see Lutz (1994) for details).
Havrylyshn and Rooden (2003) indicate that the overall index of economic freedom rating is based on the evaluation of institutional changes spanning a broad range of areas. They find the index to be thorough and reasonable, albeit still fundamentally subjective. This criticism applies to alternative indices as well.
Capital flows constitute another channel through which regional growth can influence growth in individual countries. We do not examine the relevance of this channel in the analysis because of the lack of data on capital flows among the 15 countries.
This finding points to the weakness of trade flows as indicators of openness.
Moldova is an outlier because of the significantly large remittances it has received over the past several years,
The coefficients associated with inflation in our preferred regressions suggest that a 10 percent increase in inflation would reduce growth by almost 0.4 percentage point.
Armenia has the best index of economic freedom among the CIS-12 countries.
Owing to the small number of observations we could not formally test for nonstationarity. Nevertheless, it seems reasonable to assume that GDP growth and terms of trade growth are stationary.
As evidenced by the positive and significant coefficient on the interaction between terms of trade growth and economic freedom.