Economic Growth and the Asian Financial Crisis

Economic Growth and the Asian Financial Crisis

Barro, Robert J

1. Background on Asian Economic Growth

The reason for the term “Asian miracle” is clear from Table 1, which presents the record of economic growth since 1960 for ten Asian countries. The four Asian tigers – Hong Kong, South Korea, Singapore, and Taiwan – had average growth rates of real per capita gross domestic product (GDP) from 1960 to 1995 between 5.6 and 6.6 per cent per year. This performance placed these countries at the top of the results for over 100 countries in the world.

China became the fastest growing country more recently, achieving 10.5 per cent per year per capita growth from 1990 to 1995. Also outstanding was the achievement of growth rates over 4 per cent per year from 1960 to 1995 for Indonesia, Malaysia, and Thailand. The Philippines performed much more poorly than the others, with a growth rate of only 1.3 per cent per year from 1960 to 1995. Japan’s performance was strong through the early 1970s but growth rates then diminished – the per capita growth rate from 1980 to 1995 was 2.7 per cent per year and that from 1990 to 1995 was 1.1 per cent per year. In contrast to the strong Asian results, the US growth rate from 1960 to 1995 was only 1.9 per cent per year, that for the average of 24 OECD countries was 2.7 per cent, and that for the average of 114 countries in the world was 1.8 per cent.

The Asian growth record was still strong for 1995-97, including China at 8.3 per cent per capita growth, Indonesia at 5.6 per cent, South Korea and Malaysia at 5.4 per cent, and Taiwan at 5.2 per cent. The Philippines performed somewhat better than before, with a growth rate of 3.2 per cent. Some diminution of growth rates did apply to Hong Kong (3.0 per cent), Singapore (2.4 per cent), and Thailand (3.2 per cent).

The Asian growth outcomes for 1995-97 were actually somewhat better than the forecasts of growth for 1996-2006 that I recently generated from a cross-country analysis of the determinants of economic growth. These projections were based on a model fit through data on around 100 countries from 1960 to 1995. Information was used through 1996 (or, for some variables, through an earlier date) to make the forecasts. These values are shown in column 6 of Table 1. Note that China is at 5.3 per cent projected per capita growth, South Korea at 4.2 per cent, the Philippines at 3.8 per cent, and Thailand at 3.7 per cent. Much lower are the projections for Singapore (2.1 per cent), Malaysia and Taiwan (1.9 per cent), and Hong Kong (1.2 per cent). Some of the factors that underlie these growth forecasts are discussed later.

As is well known, the Asian, and now global, financial crisis has been accompanied by recession in most of the Asian countries. Column 5 of Table 1 shows that per capita growth rates in early 1998 were negative in Hong Kong, Indonesia, Japan, South Korea, Malaysia, and the Philippines. (Thailand would be included here but the data were not available.) Only China, Taiwan, and Singapore sustained positive, though reduced, growth. It is too early to say whether the growth forecasts shown in column 6 of the table will turn out to be seriously overoptimistic. A recovery from the current recession could easily produce growth rates between now and 2006 that accord reasonably well with the ten-year projections.

2. Historical Determinants of Economic Growth

2.1. Conditional convergence

The growth-rate forecasts shown in Table 1 come from a framework with the now familiar property of conditional convergence (see Barro (1997) for an exposition). For given policies and national characteristics, diminishing returns eventually set in and cause growth rates to decline. This mechanism makes poor countries grow faster than rich ones and tends thereby to promote convergence of the poor toward the rich.

The conditional aspect of this process is critical because, in practice, countries are typically poor as they have been pursuing bad policies for a long time. Thus, it turns out that the cross-country data do not reveal a simple pattern of convergence. Figure 1 shows that, for over 100 countries observed from 1960 to 1995, there is virtually no relation between the growth rate of per capita GDP and the starting level of per capita GDP. My interpretation of these data is that the convergence tendency – the poor tending to grow faster than the rich – is roughly offset by the selection effect whereby the poor have weak policies and other characteristics.

The pattern of conditional convergence shows up in the cross-country data when the analysis holds constant an array of variables that measure polices and other factors. The resulting partial relation between the growth rate and the starting level of per capita GDP is shown in Figure 2. This nonlinear relation is strongly and increasingly negative once a country gets beyond the level of per capita GDP that prevails in the world’s poorest places (around $600-700 for purchasing-power-parity (PPP) adjusted values in 1985 US dollars). An interpretation of this pattern is that, if a poor country can get its policies and basic institutions in good order, then the fact of its being poor does tend to make it grow rapidly.

At the other end, as a country gets richer and richer, it becomes increasingly difficult to offset the “iron law of convergence” by generating enough policy improvements to offset the tendency for diminishing returns. Hence, the United States and the OECD group shown in Table 1 are projected to grow in the long run at per-capita rates of only around 1 per cent per year. By 1996, this force also became important in holding down the projected growth rates for some of the Asian countries – notably for Japan and Hong Kong but also for Singapore and Taiwan and to a lesser extent for South Korea and Malaysia.

2.2. Policy influences

Countries can, to some extent, offset diminishing returns by implementing better policies-such as strong enforcement of property rights, low levels of wasteful public expenditure, low tax rates, macroeconomic stability, and efficient public education. Given the level of per capita GDP, an improvement in policy causes a higher rate of growth for a long time but not forever. Figures 3 to 5 show some of the relationships that have been isolated between economic growth and policy variables. In Figure 3, growth is negatively related to the fraction of GDP absorbed by government consumption (intended as a measure of relatively unproductive public outlays). In Figure 4, growth is positively related to a subjective indicator of the extent of maintenance of the rule of law. In Figure 5, growth is negatively related to the average rate of inflation, which is an indicator of macroeconomic instability. Other variables that have been found to have significant explanatory power for growth include the ratio of investment to GDP (positive), the fertility rate (negative), the initial level of secondary and higher educational attainment (positive), and the growth rate of the terms of trade (positive). The relationship of growth to democracy (political rights and civil liberties) is weak when the other explanatory variables are held constant.

2.3. International openness

In recent years, researchers have used cross-country data to study the effects of numerous variables on economic growth. One area that has received a lot of attention is the extent of openness to world goods and capital markets. These results are of particular interest for assessing policy reactions to the ongoing global financial crisis, including Malaysia’s recent decision to implement capital controls.

Sachs and Warner (1995) have stressed openness to trade in goods and capital as a key policy determinant of growth. Although I am sympathetic with their underlying reasoning, I have found it difficult to make an empirical case to support their findings quantitatively. However, the cross-country evidence does suggest that openness to trade is good for growth and investment. Moreover, none of the evidence supports the notion that limitations on access to global markets for goods or capital would encourage growth and investment.

One obvious openness variable that researchers have examined is the volume of trade, measured as the ratio of exports plus imports to GDP. Policy variables that influence this trade volume have been difficult to measure in a consistent way across a large number of countries. However, variables that have been used with some success are the black-market premium on foreign exchange (a reflection of capital controls and other distortions in the foreign exchange market), measures of tariff and non-tariff barriers, variations in the prices of investment and other goods across countries, and subjective indicators of openness.

Figures 6 to 9 show the kinds of empirical relations that emerge from consideration of openness variables. Figure 6 shows the relation between the growth rate of per capita GDP and the trade variable (exports plus imports as a ratio to GDP). The result is a positive, but statistically not very reliable, effect1. The point estimate implies that a rise in the trade ratio by 10 percentage points would raise the growth rate on impact by 0.1 per cent per year. For extreme cases, such as Malaysia’s trade ratio of 1.83 in 1996 compared to Myanmar’s of 0.03, the growth effects would be large: Malaysia’s adoption of Myanmar’s trade openness would be projected to depress the Malaysian growth rate on impact by almost 2 per cent per year.

The interpretation of the results in Figure 6 requires an understanding of the other variables that are being held constant. As already mentioned, the analysis holds fixed a number of policy influences, including the size of government, subjective indicators of the maintenance of the rule of law and the extent of political rights, the inflation rate, and a measure of educational attainment. Also held constant are initial per capita GDP, the ratio of investment to GDP, the fertility rate, and the change in the terms of trade. Greater openness tends to go along with ‘favorable’ policies of other types, such as better maintenance of the rule of law. Therefore, if these other variables were not held fixed, then the openness variable would tend to get credit for the favorable growth effects of other policies. In addition, the trade variable has been measured after filtering out the normal strong negative relation between the trade ratio and country size, as measured by population and area. If this filtering is not carried out, then the estimated relation between growth and trade is weaker than that shown in Figure 62.

Figure 7 shows a similarly weak, but positive, relation between the trade variable and the ratio of investment to GDP. Thus, there is some evidence that trade stimulates growth indirectly by raising investment, in addition to the direct positive effect shown in Figure 6. (Recall that the result shown in Figure 6 applies when the ratio of investment to GDP has been held constant.)

We can also look more directly for an influence of exchange controls on growth by using the black-market premium on foreign exchange as a proxy for these controls. At the end of the 1980s (the last period for which I have assembled these data), the black-market premium was at or near zero for many of the ten Asian countries considered in Table 1-Hong Kong, Japan, Thailand, Singapore, Malaysia and Taiwan. The premium was between 5 and 12 per cent for the Philippines, South Korea and Indonesia. Only China had a high premium, around 50 per cent. In contrast, the value for Myanmar – one of the world’s most closed economies – was about 500 per cent.

Figure 8 shows that the relation between the growth rate and the black-market premium is close to zero. (A negative relation does tend to appear if some of the other explanatory variables are not held constant.) Thus, surprisingly, there is no evidence here that capital controls, as reflected in a high black-market premium on the foreign exchange, depress the growth rate. Of course, there is also no evidence from these data that capital controls are desirable. Figure 9 shows that the relation between the black-market premium and the investment ratio is also weak.

Policies such as tariff and non-tariff barriers and exchange controls may have adverse effects by distorting prices, such as those of investment goods. These effects could be important even if the impact of the policies on trade volume were minor. However, as with the black-market premium, the relation between the growth rate and the prices of investment goods turns out to be negligible3.

Thus, overall, the cross-country empirical analysis is less strong than I would have expected for documenting favorable growth effects from more openness. With respect to trade volume, the weight of the evidence is that more openness stimulates growth and investment. With regard to exchange controls, there is not much evidence for a systematic relation with growth and investment.

3. International Openness and Prosperity: Conceptual Issues

3.1. Benefits from international trade in goods and assets

On a theoretical level, openness to trade in goods and financial assets conveys several types of economic benefits. The first is that openness allows countries to specialize in production activities that match with their endowments of physical capital, skilled and unskilled labor, and natural resources. International trade in goods allows each country to produce efficiently, that is, to maximize the value of its output at world prices. The country can then consume a market basket of products that accords with preferences, even if this basket differs greatly from the one that the country produces. Without this trade possibility, the efficient matching of productive activities to factor endowments would not occur. Instead, each country would have to produce all the goods that it wished to consume.

Second, if there are economies of scale in production over some ranges, then specialised production activities in these countries can take advantage of these economies. This element applies even within an industry, for example, to the specialisation of automobile producers on different types of cars. This consideration leads to substantial international trade of an intra-industry type. This factor reinforces the first one in the sense that international trade promotes productive efficiency.

Third, international trade conveys benefits to developing countries by promoting the absorption of superior technologies from leading countries. Empirical evidence shows that increased trade volume with technologically more advanced countries promotes productivity growth. This argument is especially strong for intra-industry trade.

Fourth, international commerce can sustain healthy competition for domestic businesses. In the absence of foreign trade, inefficient domestic monopolies would be more likely to thrive.

Fifth, with respect to international capital markets, the potential to borrow from other countries can help to promote economic growth. In particular, developing countries with strong investment prospects can sustain high rates of capital accumulation without requiring an equal amount of national saving. The potential to borrow internationally can also help to smooth out temporary crises that affect the home country. On the other side, the opportunity to lend abroad (either financially or through direct foreign investment) can provide better rates of return than those available domestically.

Sixth, access to international capital markets can reduce a country’s overall risk position by allowing for global diversification of asset holdings. However, in some cases, a country’s linkage to world markets – and hence world shocks – will lead to losses that would not have occurred under autarchy.

3.2. Possible shortcomings of international borrowing

Because of the recent global financial crisis, some recent theories have stressed the possible downside aspects of openness to international capital flows. In the context of private borrowing and investment, a moral-hazard problem arises when private risks have social dimensions. One situation in which private borrowing creates social risks is when the government yields, ex post, to the temptation to bail out failing enterprises. For example, investors in risky Asian construction projects might be partly shielded from losses by public intervention. A weak bankruptcy law makes this situation particularly important. This problem often applies especially to banks and other financial institutions when the government explicitly or implicitly guarantees the deposits or other liabilities of these institutions. If governments are insuring liabilities of financial firms, then governments have the right and responsibility to monitor and regulate, ex ante, the risks taken on by these institutions.

The extent of the problems created by foreign indebtedness is probably greater for public borrowing. The potential to borrow from foreigners may motivate governments to overspend and to delay economic reforms. This adverse effect is particularly likely if foreign loans can be defaulted on in stressful situations or if needy governments tend to be bailed out by the IMF or the United States. This mechanism is clear in places such as Mexico, Russia and Brazil, where actual or prospective international bailouts worked politically against reforms, including spending cuts, privatisations, and enhanced tax collections. In terms of moral-hazard problems, the role of international institutions and rich countries in bailing out failing countries is analogous to that of governments in bailing out failing companies.

Restrictions on foreign borrowing – hence, on inflows of capital from abroad – may be justifiable as a second-best policy because of these considerations. The argument for these restrictions is stronger when the domestic legal and regulatory structure is primitive with regard to bankruptcy law, banking supervision, and transparency of corporate and financial structure. Hence, this second-best case for capital controls may be stronger for developing countries than for developed ones.

Even if capital controls successfully restrict the ability of domestic enterprises to borrow from foreigners, businesses would still be motivated to assume excessive risks domestically when public bailouts were likely. Hence, exchange controls would never be a good substitute for domestic legal and regulatory reforms. The restrictions on foreign borrowing would be compelling mainly if governments were more inclined, ex post, to bail out foreign creditors rather than domestic ones. This pattern seems to apply to the Mexican bailout of 1994, where the US led bailout mainly helped foreign financial institutions. However, the ongoing situation in Russia is different, because the government and the central bank seem to be favoring domestic banks and other enterprises relative to international creditors. More generally, it is hard to see from the standpoint of political economy why the bailout temptation would typically be greater for foreigners than for residents. Hence, it is hard to see why the restriction of private borrowing from foreigners is the paramount issue. Moreover, the curtailment of international borrowing by private entities does not address the likely more serious issue of excessive foreign borrowing by governments.

Other difficulties with capital controls involve enforcement issues and the tendency for corrupt practices to develop4. These problems tend to be especially serious for developing countries because the basic institutions of regulation and administration are typically primitive. Thus, this consideration works against the argument that capital controls are more likely to be a reasonable second-best policy for poor countries.

4. Exchange Rate and Other Policies

At times of crises, such as recently in Asia and Latin America, central banks are often preoccupied with defending the exchange rate. Such a defense is usually characterised by high domestic interest rates. Some of the recent arguments for exchange controls seem to be aimed at avoiding the problems of volatile exchange rates and high domestic interest rates.

High interest rates should be viewed mainly as a symptom of the likelihood of devaluation, rather than as an indicator that a country is mounting a strong defense against this devaluation. If participants in financial markets believe that a devaluation is coming, then it is not surprising that they have to be bribed with high interest rates to hold assets denominated in the domestic currency. Moreover, the maintenance of high interest rates entails large economic costs, which policymakers typically and reasonably are unwilling to bear for long. Because everyone knows this, high interest rates are usually viewed as only temporary – hence, market participants are rationally skeptical that this kind of defense will be maintained for long.

The real problem is that central banks have too many responsibilities, and these are often contradictory. In particular, the maintenance of a fixed exchange rate is sometimes incompatible with the desire to bail out domestic financial institutions, loan money to the government, provide subsidised credit to favoured industries, and so on. Central banks never have enough international reserves to maintain the world market value not only of the domestic currency but also of all bank deposits and government debt.

A central bank – or, more accurately, a currency board – can maintain a fixed exchange rate with a designated foreign currency if it takes this objective as its sole reason. All that is required, aside from a commitment to the rules, is an initial stock of international reserves that at least equals the central bank’s liabilities, typically in the forms of currency and reserves of domestic banks. Even Russia today has just about enough reserves to qualify. In this system, inflows or outflows of international reserves are not sterilized but are instead reflected as corresponding changes in the stock of high-powered money. Hong Kong and Argentina have been successfully operating an approximation to this kind of regime for some time.

If domestic residents find the system credible, then they tend to hold and use the domestic money and are disinclined to hoard foreign currency. Thus, for example, Argentine residents have dramatically reduced their holdings of US dollars. If the residents find the system unreliable, then the domestic money would be driven out by foreign currency (provided by the currency board), and the economy would be “dollarized.” Although this outcome implies the loss of seigniorage income, it is otherwise not terrible (aside, perhaps, from the political embarrassment). That is, the country ends up with a satisfactory payments mechanism and low inflation.

One frequently mentioned shortcoming of a currency board is that it entails an absence of an independent monetary policy. The loss of this independence is surely accurate-the country basically buys into the US dollar (if it pegs to the US dollar) and, hence, indirectly to the US monetary and inflation policy. Whether the loss of an independent monetary policy is a bad thing is debatable. One could instead say that a country that adopts this regime is no longer subject to the whims of the domestic central bank’s discretionary monetary policy. That is, the country has managed to buy cheaply into a system of commitment to low and stable inflation.

If some domestic nominal variables, such as wages, were sticky downward, then there would sometimes be benefits from an independent monetary policy. This situation would arise when economic disturbances raised the equilibrium real exchange rate for the home country in relation to the pegged country. In this circumstance, a monetary expansion and a corresponding nominal depreciation of the currency would substitute for reductions in domestic prices and wages. If these domestic prices and wages are not flexible, then the monetary expansion might avoid a recession.

In practice, however, the main thing that a country seems to lose from the loss of an independent monetary policy is the option to use currency devaluation to effect partial default of public and private obligations that are denominated in the domestic currency. A functioning currency board does not preclude these defaults, but they have to be explicit.

The system also does not rule out subsidies to banks or favoured industries. But these subsidies have to come directly from the government and be financed by taxes or public borrowing. The subsidies cannot be paid for and perhaps hidden by using the central bank and, hence, the printing press.

Although a currency-board system has many attractions, it has to be remembered that the system is not a cure-all. It would not prevent, for example, a large drop in stock-market values associated with reduced prospects for long-term economic growth. It also does not lessen the significance of sound fiscal and regulatory policies. In this respect, one problem in Indonesia some months ago was that the proposal to adopt a currency board seemed to be viewed as an excuse not to carry out other reforms. Given this unrealistic outlook, it was probably just as well that a currency board was not implemented at that time.

It is also not surprising that central bankers and international organizations such as the IMF tend to dislike currency boards. A key point of a currency board is to limit the discretionary powers of the monetary authority. For someone such as a central banker or an IMF official – for whom discretionary authority is the ultimate happiness – such restrictions would be distasteful. Even aside from issues of the joy of power, a rule-like system goes against the interventionist philosophy of most public officials5.

Malaysia ought to consider a currency board as an alternative approach to exchangerate and monetary policy. Although this system would not cure all problems, it looks much more attractive than the IMF’s usual prescription to defend a traditional fixed exchange rate with high interest rates and tax increases. It also looks a lot better than exchange controls. It is even preferable to have freely floating exchange rates, although this regime does have some appeal. With an operating currency board, the problems of exchange rates and monetary policy are basically settled, and government officials can worry about other matters.

1 The estimated coefficient of the trade variable is 0.011 with a t-statistic of 2.8.

2 The positive estimated relation between growth and trade still emerges if one uses as instruments some possibly exogenous variables that predict trade volume. These instrumental variables are measures of tariff and non-tariff barriers (observed in the early 1980s), prior values of the black-market premium on foreign exchange, and prior values of the prices of investment goods. If these instruments are used (and the trade-volume variable is dropped from the instrument list), then the estimated coefficient of the trade variable becomes 0.014 with a t-statistic of 2.1. This result suggests that the positive relation between growth and trade reflects the influence of trade on growth, rather than the reverse.

3 The investment price data are the PPP-adjusted values generated by Summers and Heston in their Penn-World Table, version 5.6. See Summers and Heston (1991) for a conceptual overview of their data set. Updated figures are available on the internet from the National Bureau of Economic Research.

4 However, in some cases, corrupt practices that allow an escape from bad rules are superior to strong enforcement of the rules.

5 For example, Larry Summers once told me that if he believed in my laissez faire philosophy he would find a different occupation.


Barro, R.J. 1997. Determinants of Economic Growth: A Cross-Country Empirical Study. Cambridge, Massachusetts: MIT Press.

Sachs, J. and A. Warner. 1995. Economic reform and the process of global integration. Brookings Papers on Economic Activity. 1: 1-95.

Summers, R. and A. Heston. 1991. The Penn World Table (Mark 5): An expanded set of international comparisons, 1950-1988. Quarterly Journal of Economics 106(2): 327-369.

Robert J. Barro*

Harvard University

* Department of Economics, Littauer Center, Harvard University, Cambridge, MA 02138-3001 USA.

Copyright Malaysian Economic Association Jun-Dec 1998

Provided by ProQuest Information and Learning Company. All rights Reserved