Information about Asia and the Pacific Asia y el Pacífico
Chapter

II. Long-Term Growth Performance

Author(s):
Erik Offerdal, Kalpana Kochhar, Louis Dicks-Mireaux, Jianping Zhou, Mauro Mecagni, and Balázs Horváth
Published Date:
December 1996
Share
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details

Thailand’s economic growth performance during the past three decades has been impressive. In 1960, per capita gross domestic product (GDP) was only slightly higher than that in India, Pakistan, Ghana, and Morocco and lower than that in Sri Lanka, the Philippines, Argentina, Chile, and Peru. By 1992, however, with per capita GDP estimated at nearly US$4,000 (valued at purchasing-power-parity—PPP—prices), Thailand had far surpassed the average for South Asia as well as that of many Latin American countries that began the period with higher per capita incomes, and it had reduced the gap with its high-performing East Asian neighbors.

Thailand has also been successful in diversifying its economic structure in a relatively short period of time. Between 1970 and the late 1980s, the agricultural sector’s contribution to GDP declined from 25 percent of GDP to 15 percent, while that of the manufacturing sector rose from 25 percent to about 35 percent. Significant diversification took place in the structure of exports, with the share of agricultural products falling more than half (to below 30 percent) in the late 1980s, when manufactured goods, especially textiles and garments, surpassed rice as the highest export earner. The manufacturing sector’s share of the labor force remained virtually unchanged, however, reflecting trade and industrial policies that, until the mid-1980s, had favored capital-intensive manufacturing.

Key Features of Economic Policies

The most striking feature of Thailand’s economic policy environment is its record of macroeconomic stability. For most of the past three decades, Thailand has followed conservative macroeconomic policies in support of its policy of maintaining (de jure or de facto) a fixed exchange rate (with the exception of two nominal devaluations in 1981 and 1984). Except for two brief periods immediately following the two oil price shocks, inflation has been in the single digits, and since 1982 it has averaged less than 4 percent. Although it could be argued that the adjustment to the external shocks of the 1970s and early 1980s was initially slow (see Section III), the imbalances were not permitted to become so acute as to undermine the credibility of policymakers.

The prevailing view among Thai policymakers has always been that the government should play only a limited role in the economy and that the private sector should be the engine of growth. Therefore, Thailand’s public enterprise sector is small, more focused on traditional public sector activities, and more profitable than in many developing countries (see Section VII for details). Furthermore, over time, public sector activity has shifted away from direct involvement in industrial production toward the provision of public infrastructure and services. This, together with the consistently pro-business orientation of the government reflected in its tax laws and industrial policy, has served to create a dynamic private sector.

Thailand’s trade and investment policies have been characterized by a liberal attitude toward foreign investment inflows, and by a trade system that, albeit oriented toward import substitution until the mid-1980s, had overall protection levels that were moderate in comparison with those of many other developing countries (see Section VII). The government pursued an activist policy of intervening to promote particular industries, with incentives granted by the Board of Investment (BOI) as the most powerful tool of industrial policy. Import substitution and promotion of capital-intensive manufacturing dominated the strategy of industrial development during the 1960s and 1970s. Tariff rates were progressively increased during the 1970s as a response to growing balance of payments difficulties, and there was a proliferation of exemptions and surcharges on specific products and some nontariff barriers. Although the strategy of industrialization through import substitution began to be deemphasized in the early 1980s, it was not until the mid-1980s that a decisive break was made away from import substitution and toward export promotion.

Implications of Empirical Studies on the Determinants of Growth

What can we learn about Thailand’s growth record from empirical studies of the determinants of long-term growth? In what follows, three questions are addressed. The first seeks to conduct an “accounting for long-run growth” exercise for Thailand, based on cross-country regressions and using data averaged over the period 1960-92. The second uses cross-country estimates of correlations between growth, capital accumulation, and productivity changes, on the one hand, and macroeconomic and structural policies and conditions, on the other, to examine whether Thailand’s policy environment has generally been conducive to growth and investment. The third examines the behavior of growth during Thailand’s “adjustment” phases after controlling for the long-run determinants of growth identified by the first approach.

More specifically, the first approach is to relate the average annual growth of real per capita GDP to the rates of physical and human capital accumulation, while controlling for population growth and initial real per capita income in each country relative to that of the United States (to capture the prediction that less technologically advanced countries tend to catch up over time with more advanced ones).2 The results of such an exercise are shown in Table 1. On the basis of this cross-section regression, Thailand’s rate of factor accumulation and its relatively low initial income level could have been expected to yield average per capita income growth of about 2.6 percent a year between 1960 and 1992; actual average annual growth over this period was 4.5 percent, suggesting that about 40 percent of long-term growth in Thailand is attributable to influences other than the accumulation of factors of production, including productivity gains not related to those arising from the “catching-up” phenomenon. This outcome is similar to that for most of the rest of East Asia.

Table 1.Growth and Factor Accumulation in Thailand and Selected East Asian Countries, 1960-92
ThailandMalaysiaSingaporeHong Kong
Parameter EstimateSample meanContributionSample meanContributionSample meanContributionSample meanContribution
CONSTANT– 0.069–0.070.07–0.07–0.07
INV0.12818.352.3523.783.0431.814.0719.762.53
POPG–0.1902.46–0.472.560.491.68–0.322.01–0.38
GAP60–3.7710.10–0.360.140.540.17–0.630.23–0.86
PRIM601.2280.831.020.961.181.111.360.871.07
SEC600.7420.120.090.190.140.320.240.240.18
Estimated total2.63.34.72.5
Actual4.54.46.46.2
KoreaTaiwan Province of ChinaPhilippinesIndonesia
Parameter EstimateSample meanContributionSample meanContributionSample meanContributionSample meanContribution
CONSTANT–0.069–0.07–0.07–0.07–0.07
INV0.12824.243.1022.222.8515.441.9817.432.23
POPG–0.1901.80–0.342.11–0.402.61–0.502.12–0.40
GAP60–3.7710.09–0.340.13–0.480.1 1–0.430.06–0.24
PRIM601.2280.941.150.961.180.951.170.670.82
SEC600.7420.270.200.280.200.260.190.060.04
Estimated total3.73.32.32.4
Actual6.76.21.33.7
Source: IMF staff estimates.Note: The following equation was estimated:GRTH=0.069+.128INV(***).190POPG+1.228PRIM60(**)+.742SEC603.771GAP60(***)(0.12)(4.40)(1.18)(0.72)(3.58)AdjustedR2=.45;numberofcountries:103.

where GRTH stands for per capita GDP growth, INV refers to the ratio of investment to GDP, POPG is the growth rate of the population (all averaged over the 1960–92 period), GAP60 is the income of country I relative to that of the United States in 1960, and PRIM60 and SEC60 refer to primary and secondary school enrollment rates. Data for GRTH, INV, GAP60, and POPG were taken from the Penn World Table, Mark 5 (Summers and Heston, 1991), and PRIM60 and SEC60 data are from the World Bank’s World Tables (various years).*** Implies significance at the 1 percent level, and ** implies significance at the 5 percent level.

Source: IMF staff estimates.Note: The following equation was estimated:GRTH=0.069+.128INV(***).190POPG+1.228PRIM60(**)+.742SEC603.771GAP60(***)(0.12)(4.40)(1.18)(0.72)(3.58)AdjustedR2=.45;numberofcountries:103.

where GRTH stands for per capita GDP growth, INV refers to the ratio of investment to GDP, POPG is the growth rate of the population (all averaged over the 1960–92 period), GAP60 is the income of country I relative to that of the United States in 1960, and PRIM60 and SEC60 refer to primary and secondary school enrollment rates. Data for GRTH, INV, GAP60, and POPG were taken from the Penn World Table, Mark 5 (Summers and Heston, 1991), and PRIM60 and SEC60 data are from the World Bank’s World Tables (various years).*** Implies significance at the 1 percent level, and ** implies significance at the 5 percent level.

Many caveats apply when interpreting the results of such empirical analyses: first, the regression explains less than one-half of the variation in per capita income growth across countries (see Table 1); second, the approach is based on the assumption that all countries have the same production function; third, the quality of the data for several countries is poor, and it is impossible to identify the direction of biases so introduced. Although the results need to be interpreted with care, they do imply that—compared with world “average” growth, adjusted for factor inputs and initial conditions—Thailand’s performance has been good, suggesting that, inter alia, the policy environment has by and large been growth enhancing.

The second approach, following Fischer (1993), seeks to identify the effects of macroeconomic and structural policies on output growth, and the channels—capital accumulation or productivity growth—through which these effects are transmitted. From a growth-accounting perspective, output growth can be attributed to increases in the supply of factors and to a residual total factor productivity (TFP).3

Several proxy measures were used for the impact of policies, and correlations were examined between these proxies and output growth, factor accumulation, and TFP growth, respectively, in a cross-country and time-series regression analysis.4 The panel regressions indicate that macroeconomic instability—measured by the rate of inflation—is negatively associated with growth, and that this link works through both capital accumulation and productivity (Table 2). Overall policy uncertainty or intervention in the exchange system—indicated by the parallel market exchange rate premium—is also negatively correlated with growth, with most of the effect acting through capital accumulation.

Table 2.Impact of Policies on Growth, Capital Accumulation, and Factor Productivity, 1970-92
Estimated Correlation with Growth1Estimated Correlation with Capital Accumulation1Estimated Correlation with TFP Growth1ThailandWhole SampleAsian Developing CountriesEast AsiaOECD
Macroeconomic policies
Inflation–--–--–--6.753.89.98.19.3
Budget surplus 2+–--++–2.6–4.3–3.3–1.6–4.1
Capital expenditure2++++++3.64.74.85.32.8
Government saving2++++++++1.10.51.43.7–1.2
Exchange rate premium–--–--75.448.73.8
External conditions Change in the terms of trade++++++–1.8–0.10.9–0.2
Structural conditions
Primary school enrollment rate+++++81.085.182.895.7106.0
Ratio of broad money to GDP++47.239.037.447.061.4
Average effective trade tax rate3–-7.613.414.712.111.6
GDP in 1970 (in logarithms)–-–--3.23.33.13.33.9
Memorandum:
Real GDP growth6.53.45.67.42.7
Growth in real capital stock8.34.26.79.93.6
Growth in total factor productivity (TFP)1.80.31.41.80.7
Source: IMF staff estimates.1 The correlations were estimated from three panel regressions with growth in real GDP, capital accumulation, and TFP as dependent variables. TFP residuals were calculated as follows: TFP = ZGDP - 0.4 ZKAP -0. 6 ZLAB, where ZGDP is the rate of growth of real GDP, ZKAP is the rate of growth in the real capital stock, and ZLAB is the growth rate of the labor force. The data for ZGDP and GDP in 1970 were measured at purchasing-power-parity (PPP) prices and were taken from the Penn World Tables, Mark 5 (Summers and Heston, 1991); data for ZKAP are from King and Levine (1994); data for the budget surplus, government capital expenditure and saving, and the average effective import tax rate are from IMF, Government Finance Statistics Yearbook (various years); data for inflation and broad money are from IMF, International Financial Statistics (various years); and all other variables are from Fischer (1993). The symbols ++ + and mean that the correlations are significant at the 1 percent level; ++(–) and +(-) imply significance at the 5 and 10 percent levels, respectively.2 In percent of GDP. The effects of the budget balances, on the one hand, and capital expenditures and government savings, on the other, were estimated in separate panel regressions. The estimated effects of the other macroeconomic variables did not differ significantly between the two regressions.3 Defined as the ratio of total trade taxes to the value of exports and imports.
Source: IMF staff estimates.1 The correlations were estimated from three panel regressions with growth in real GDP, capital accumulation, and TFP as dependent variables. TFP residuals were calculated as follows: TFP = ZGDP - 0.4 ZKAP -0. 6 ZLAB, where ZGDP is the rate of growth of real GDP, ZKAP is the rate of growth in the real capital stock, and ZLAB is the growth rate of the labor force. The data for ZGDP and GDP in 1970 were measured at purchasing-power-parity (PPP) prices and were taken from the Penn World Tables, Mark 5 (Summers and Heston, 1991); data for ZKAP are from King and Levine (1994); data for the budget surplus, government capital expenditure and saving, and the average effective import tax rate are from IMF, Government Finance Statistics Yearbook (various years); data for inflation and broad money are from IMF, International Financial Statistics (various years); and all other variables are from Fischer (1993). The symbols ++ + and mean that the correlations are significant at the 1 percent level; ++(–) and +(-) imply significance at the 5 and 10 percent levels, respectively.2 In percent of GDP. The effects of the budget balances, on the one hand, and capital expenditures and government savings, on the other, were estimated in separate panel regressions. The estimated effects of the other macroeconomic variables did not differ significantly between the two regressions.3 Defined as the ratio of total trade taxes to the value of exports and imports.

The results also suggest that budget surpluses are positively correlated with productivity growth but negatively correlated with capital accumulation. However, examination of the separate influences of government saving and government capital spending reveals that—as would be expected—both components of the overall fiscal balance are positively correlated with capital accumulation. The negative correlation of the budget surplus is accounted for by the coefficient on capital spending being larger than that on government saving. Moreover, favorable trends in the terms of trade are also strongly associated with output growth. Additional results suggest that high effective trade taxation and an underdeveloped financial system (as measured by the ratio of broad money to GDP) are also negatively correlated with capital accumulation, and that low human capital formation (provided by the primary school enrollment ratio) is negatively correlated with output and productivity growth.

Table 2 presents the estimated correlations of macroeconomic and structural policies with the growth of output capital and TFP growth, and it compares sample averages for the right-hand-side variables for Thailand with those for other countries. By most of the measures, Thailand’s macroeconomic policies were—on average during the period, and compared with other regions and countries—conducive to rapid output growth, capital accumulation, and productivity gains. It would be reasonable to conclude from this evidence that, relative to the behavior of the same variables for other countries, the stable macroeconomic environment—combined with an outward-oriented trade system; a reasonably well-developed and undistorted financial system; and relatively high levels of education, especially at the primary level5—have all had a positive influence on Thailand’s long-run growth performance. In contrast, the marked deterioration in Thailand’s terms of trade could be expected to have had a dampening effect on output growth relative to a situation in which the terms of trade remained stable or improved.6

Developments in Output and Productivity During Phases of Adjustment

What happened to output and productivity during the different stages of Thailand’s adjustment process, and how do these developments compare with what would be expected on the basis of the long-term determinants of growth set out in Table 1? Following an approach suggested by Bruno and Easterly (1995), this question was addressed by estimating the basic growth regression across a panel of countries during subperiods corresponding to Thailand’s adjustment phases (Table 3). The indicated “growth residuals” are a measure of how much Thailand’s per capita growth exceeded the world “average” in each subperiod, after controlling for the long-run determinants of growth specified in the equation.

Table 3.Per Capita Growth Residuals After Controlling for Long-Run Determinants of Growth1(Average growth differential; in percentage points)
1970–801981–861987–93
Growth residuals after controlling1.8***2.7***6.2***
for investment and other long-run(7.3)(9.1)(15.2)
determinants other than “policies”
Growth residuals after controlling2.3***3.4***8.1***
for long-run determinants other(8.7)(10.8)(21.7)
than investment and “policies”
Growth residuals after controlling2.4***3.0***7.6***
for all factors other than investment(9.2)(9.4)(15.7)
Source: IMF staff estimates.1 Following an approach suggested by Bruno and Easterly (1995), the growth residuals were derived by including dummy variables for each country in Barro-style growth regressions (Barro, 1989) estimated using pooled data on the average growth rates over the respective subperiods and across 92 countries. The estimated equations took the following form:GRTHit=Σt=1rλtDUMit+Σt=1rβtDUMIt+α1INV+α2POPG+α3GAP70+α4PRI70+α5SEC70.In addition, the following policy-related factors were added to (and INV dropped from) the regressors underlying the third set of results reported for each country:α6INFL+α7BUDSUP+α8PREMIUM+α9DTOT+α10HINFL.where T is the number of subperiods in each case, GRTH is the growth rate of real per capita GDP measured in terms of constant domestic prices, INV is the share of investment in GDP, POPG is the growth rate of the population, GAP70 is the relative income gap between country / and the United States in 1970, PRI70 and SEC70 are primary and secondary school enrollment rates in 1970, INFL is the average annual rate of inflation in the consumer price index, BUDSUP is the government budgetary surplus as a percent of GDP, PREMIUM is the average premium between the parallel market and official market exchange rates, and DTOT is the change in the ratio of export prices to import prices. Possible nonlinearities in the effects of inflation on growth are taken account of by including the variable HINFL, which takes the value of the inflation rate when it is greater than 40 percent. DL/A1, are the dummy variables for each period for all countries; and DUMI are dummy variables for the country being examined, one for each period. Absolute values of heteroscedastic-consistent t-statistics are in parentheses. The symbol ** * denotes significance at the 1 percent level, and * * and * at the 5 and 10 percent levels, respectively.
Source: IMF staff estimates.1 Following an approach suggested by Bruno and Easterly (1995), the growth residuals were derived by including dummy variables for each country in Barro-style growth regressions (Barro, 1989) estimated using pooled data on the average growth rates over the respective subperiods and across 92 countries. The estimated equations took the following form:GRTHit=Σt=1rλtDUMit+Σt=1rβtDUMIt+α1INV+α2POPG+α3GAP70+α4PRI70+α5SEC70.In addition, the following policy-related factors were added to (and INV dropped from) the regressors underlying the third set of results reported for each country:α6INFL+α7BUDSUP+α8PREMIUM+α9DTOT+α10HINFL.where T is the number of subperiods in each case, GRTH is the growth rate of real per capita GDP measured in terms of constant domestic prices, INV is the share of investment in GDP, POPG is the growth rate of the population, GAP70 is the relative income gap between country / and the United States in 1970, PRI70 and SEC70 are primary and secondary school enrollment rates in 1970, INFL is the average annual rate of inflation in the consumer price index, BUDSUP is the government budgetary surplus as a percent of GDP, PREMIUM is the average premium between the parallel market and official market exchange rates, and DTOT is the change in the ratio of export prices to import prices. Possible nonlinearities in the effects of inflation on growth are taken account of by including the variable HINFL, which takes the value of the inflation rate when it is greater than 40 percent. DL/A1, are the dummy variables for each period for all countries; and DUMI are dummy variables for the country being examined, one for each period. Absolute values of heteroscedastic-consistent t-statistics are in parentheses. The symbol ** * denotes significance at the 1 percent level, and * * and * at the 5 and 10 percent levels, respectively.

The results indicate that (1) growth was much stronger in Thailand relative to the world average throughout the period under consideration; (2) there was no decline in this growth differential during the period when Thailand was undergoing adjustment—on the contrary, growth remained stronger than the world average; and (3) the growth differential widened markedly during the “postadjustment” period. The equation was reestimated with investment excluded from the set of explanatory variables, to examine whether the change in the growth residual over time reflected influences acting primarily through investment or through productivity. All of these conclusions held true even after omitting investment from the regression. Thus, the impressive economic performance in Thailand, especially in the postadjustment period, could be attributed to productivity gains.

Finally, the equation was reestimated by incorporating a number of macroeconomic policy-related variables—inflation, the budget deficit, and the size of the parallel market premium—as well as changes in the terms of trade. The residuals from this regression give an indication of how much of the differentials in growth between Thailand and the “world” over the adjustment phases can be accounted for by these policy-related variables. The results suggest that, even after taking into account Thailand’s macroeconomic conditions, growth was significantly higher than the world average, indicating that other policies and conditions contributed to the consistently better-than-average growth performance in Thailand. The size of the growth residuals in each period indicates that the spurt in growth is unlikely to reflect merely a reversion to trend following a temporary slowdown.

Alternative measures of the contributions to growth of TFP, shown in Figure 2, indicate that the acceleration in output growth during the postadjustment phase owed much to a sharp improvement in TFP, in addition to the surge in investment.7

Figure 2.Contribution of Total Factor Productivity to Growth

Sources: International Monetary Fund, World Economic Outlook (various issues); World Bank, World Tables, “Social Indicators of Development” (various issues); IMF staff estimates; Barro and Lee (1993); Nehru and Dhareshwar (1993);and Sarel (1995).

1Residual GDP growth after subtracting the estimated contributions of capital, age-adjusted population growth as an alternative measure of labor force growth, and human capital inputs.

2Residual GDP growth after subtracting the estimated contributions of capital and labor, as measured by labor force growth.

    Other Resources Citing This Publication