Internal Discussion Paper ASIA REGIONAL SERIES Report No. IDP15 The Role of Exchange Rate Policy in Four East Asian Countfries Sang-Woo Nam May 1988 The vies pres areth ose of the author, and they should not be interpretedas reecting those of the World Bank TU ROLE OF UXCHANGE ATE POLICY IN FOUR EAST ASIAN COUNTRIES SANG-WOO K4N' *Macro-economist, Country Department II. I am grateful to Danny Leipsiger for his valuable conments and Inbom Choi for his efficient assistance with the simulation exercises. THE ROLE OF EXCHANGE RATE POLICY IN FOUR EAST ASIAN COUNTRIES SANG-WOO NAM Abstract This paper attempts to shed some light on exchange rate management options in Korea, Malaysia, the Philippines and Thailand on the basis of quantitative findings based on macroeconomic models. The econometric model designed for this purpose is fully dynamic, incorporating the evolution of domestic inflation and absorption triggered by an exchange rate change. Simulation exercises with the model for 1986-90 show that exchange rate devaluation is both inflationary and expansionary, though the expansionary effect depends largely on the stance of monetary policy. Devaluation also improves the current balance (except for Malaysia), but the improvement is generally relatively small, and the favorable effect is very short-lived for Korea. Exchange rate management in these countries is shown to be consistent with both the medium-run constancy of the real effective exchange rate (REER) and the short-run optimality of macroeconomic policy adjustments. One possible exception is Malaysia whose currency seems to have undergone rather steady real depreciation. In the three countries where devaluation is found to improve the current balance, the REER was observed to be respsnsive to changes in foreign exchange reserves, while the Malaysian REER was responsive only to recent growth performance. The REER is also affected by current inflation in Korea and Thailand where devaluation is found to be most inflationary. These exercises are valuable for policymakers confronted with highly variable international exchange rates which can have profound effects on open economies. Particularly in the East Asian setting, exchange rate management is becoming an increasingly important policy tool to affect major macroeconomic variables. This paper aims at showing the possibilities and limits of those interventions. THE ROLE OF EXCHANGE RATE POLICY IN FOUR EAST ASIAN COUNTRIES Table of Contents Page No. II. THE EXCHANCE RATE AND ADJUSTMENT MECHANISMS.................. 4 III. AN ECONOMETRIC MODEL ...................................... 8 IV. ESTIMATION RESULTS........................................... 23 V. SIMULATION EXERCISES................ ......... 33 A. Nominal Devaluation...................................... 34 B. Real Devaation......... .......................... ...... 44 C. Counterfactual Simulation: Constant REER for 1980-86.... 48 VI. EXCHANGE RATE MAACEMENT. .................................... 51 A. Currency Basket.......................................... 51 B. Trend of the Real Effective Exchange Rate................ 52 C. Evaluation of Exchange Rate Management......*........... 60 VII. SUMMARY AND 00NCLUSIONS........ .............................. 67 REFERENCES I. INTRODUCTION 1.1 As most international commodities have suffered from prolonged price declines since the early 1980s, countries heavily dependent on natural resources for their exports have experienced a steady aggravation of the terms of trade. Slow export growth in these countries was one of the major hindrances to economic recovery not only because it directly repressed income growth but also because it go-arally entailed restrictive demand management policies in the face of growing difficulty in financing the external payment deficit in the international financial market. Due to the large and sustained government deficit in many countries, expansion of fiscal spending is not a feasible approach to pulling the economies out of recession. 1.2 In these circumstances, promotion of manufactured exports was considered the most promising avenue for improving both the balance of payments and growth performance. For countries like Korea which has no major natural resources to export, the development strategy since their initial take-off stage was firmly based on export promotion through industrializa- tion. Expansion of manufactured exports requires strong technological and infrastructural bases, dynamic entrepreneurs, a trained labor force as well as appropriate incentives in terms of the structure of inter-industry protection and the exchange rate. While it may be difficult to greatly enlarge manufac- turing export capacity in the short run, the most recent export boom in countries like Korea and Thailand seems to demonstrate that manufactured exports can indeed expand rapidly when export incentives or competitiveness are strengthened sufficiently. Of course, their new competitive edge came - 2 - mainly from the substantial depreciation of their currencies on a real effective basis as the US dollar was declining sharply against other major currencies. 1.3 The effect of a given exchange rate change on overall exports is very different depending on a country's export structure. When an economy relies heavily on several primary commodities for its exports, the country is likely to be one of the major suppliers of those commodities and face rather inelastic demand curves with respect to prices. In this case, an exchange rate depreciation will shift supply schedules to decrease prices in foreign currency, with little increase in the quantities traded. On the other hand, a country is typically "small" in manufactured exports and faces elastic demand curves with respect to prices, although the latter can hardly be considered flat (infinitely elastic) due to product differentiation. In this case, an exchange rate depreciation is most likely to increase export value in foreign currency. 1.4 It is no wonder why many economies endowed with rich national resources to export have been afflicted with the "Dutch disease" syndrome, in which the overvalued exchange rate during a commodity boom has made manu- factured exports unprofitable and uncempetitive. Of the four countries in this study, i.e., Korea, Malaysia, the Philippines and Thailand, all except Korea have been heavily dependent on natural resources for their exports: in 1985, the share of total merchandise exports accounted for by primary commodities was as high as 72% for Malaysia (1984), 61% for Thailand and 43% for the Philippines, but was only 8% for Korea. Interesting question is to what extent the role of exchange rate policy is determined, and the pattern of exchange rate management influenced, by the export structure and other structural characteristics of the economy. - 3 - 1.5 This paper attempts to evaluate the exchange rate management of these forr southeast Asian countries. As a groundwork, major quantitative work is devoted to assessing the effect of exchange rate policy in these economies. The effect on the balance of payments and economic growth is set against its inflationary effect, with intercountry differences explained by their trade and economic structure. The pattern of exchange rate management is then analyzed to determine whether it is consistent with the study's quantitative findings on the effect of devaluation. 1.6 In Section II, alternative approaches to explaining the role of exchange rate policy in the adjustment of the trade balance will briefly be reviewed. Against this background, a small economywide econometric model geared to assessing the effect of exchange rate policy will be presented in Section III. The estimation results will be discussed in Section IV, and si-ulation exercises will be undertaken in Section V to quantify the effect of nominal as well as real devaluation. A counterfactual simulation will also be performed to evaluate the role of exchange rate policy in the recent economic performance of these countries. Section VI focuses on evaluating exchange rate management. The real effective exchange rate is calculated in a consistent manner for all four countries, and the behavior of exchange rate management is assessed in light of the simulated effect of exchange rate policy. - 4 - II. THE EXCHANGE RATE AND ADJUSTMENT MECHANISMS 2.1 By now, there are fairly well established alternative approaches explaining the adjustment mechanism through which an exchange rate change affects the trade balance. They include the elasticity approach, the absorption or expenditure approach and the more recent monetary approach. These approaches are not necessarily mutually inconsistent, the differences lying rather in their emphasis. They, therefore, may be complementary to one another in understanding the effect of an exchange rate change. We hope to be able to synthesize these alternative approaches into a consistent analytical framework, although some fundamental differences in their focuses may limit this effort. Elasticity Approach 2.2 Based on a two-goods model, in which two countries specialize in one of the two goods and trade with each other, Bickerdike (1920), Robinson (1937) and Metzler (1948) derived the condition for an exchange rate devaluation to result in an improvement in the trade balance. The condition, called the Bickerdike Condition, is as follows: (2.1) (-ed) (-e) (s + es+ 1) + es.es ((_ed) + (-e) - 1> 0 where ea and b are supply elasticities for country A and B, respectively, a b and e d and e d are their demand elasticities. When the demand elasticities a b are low enough in absolute values, the condition may not be satisfied. However, it can easily be seen that, with infinite supply elasticities, the Bickerdike Condition reduces to the familiar Marshall/Lerner Condition: -5- (2.2) (_sd) + (_ed) 1 Furthermore, for a small open economy A facing given world market prices, that is, - d and b *, the Bickerdike Condition reduces to the following equation, which is always satisfied. (2.3) (-Cd) + ss > 0 Absorption or Expenditure Approach 2.3 While the elasticity approach sees the trade balance as the difference between exports and imports, the absorption approach pays its attention to the trade balance as the difference between aggregate income and expenditure. From the income-expenditure identity of y= ya + x - m, the trade balance is express.. as: (2.4) x - m - y - y, where y is income, ya is absorption, x is exports and m is imports in real terms. Unlike the elasticity approach, which is solely concerned with the price effects of devaluation on demand and supply, the focus of the absorption approach is on the effects on income and expenditure. With the absorption equation: (2.5) ya- ya(y, e) where e is the exchange rate, *6* B(x-u) O(Y-ya) O_a . -6- Se 8e e Sy s . This pure absorption approach, however, is insufficient to determine income or the trade balance. Bringing back the trade balance (t) as x - m, where m is a function of y, can do just this. (2.4) x - a * y - y (2.5) y a y a (y, e) (2.6) x - m = t (y, e) In the above model, the trade balance t is determined jointly with y, and the effect of an exchange rate change can uniquely be determined. Monetary Approach 2.4 The monetary approach understands the adjustment mechanism of devaluation as a monetary phenomenon. Devaluation raises the domestic price level to reduce the real value of money stock, and induces people to save more to replenish lost real balances. At the same time, deficits in the balance of payments are considered to be a symptom of monetary disequilibrium, which is only transitory and self-correcting without sterilization of the accompanying change in international reserves. If this self-correcting process is to be accelerated, deliberate monetary contraction can be pursued, *nd devaluation is viewed as a substitute for monetary contraction. The framework of the monetary approach may be written as follows: (2.7) Msu &*(R + D) (2.8) (!)d i4(y) i) (2.9) y a y&+ t (2.10) y = y + s d a p r where M SO(A) = Nominal money supply and real demand for money, respectively p p = Price level 4 a Money multiplier, and R,D = International reserves and net domestic credit of the central bank, respectively. When devaluation takes place, the accompanying price increases reduce the real balance of money, which, in turn, lowers absorption below real income. This excess supply of domestic goods is matched by net exports. This basic monetary argument is based on the assumption of factor price flexibility, which allows an economy to maintain full employment. In this case, however, devaluation is 'superficid,' since any deficits in the balance of payments may be eliminated by expenditure reduction polcies. 2.5 Summing Up. The monetary approach is different from other approaches in that it describes the dynamics of devaluation while others are basically static. The price elasticities in the elasticity approach should ultimately converge to zero in the real world, as the relevant dynamics are set in motion. The dynamic factors might be real money balances, domestic prices, or aggregate income and expenditures. The monetary approach concentrates on demand behavior for money, though it may be just one of many dynamic factors working in reality. The absorption or expenditure approach also ?rovides a static picture in the sense that domestic prices are fixed. In the absorption approach, where infinite supply elasticities are assumed, the focus is on the effect of devaluation on expenditure and the effect of variable income on absorption and imports. These are left implicit in the - 8 - elasticity approach for an explicit analysis of relative prices. The alternative approachei should not, however, be viewed mutually inconsistent, since they largely emphasize different aspects of the adjustment process. III. AN ECONOMETRIC MODEL 3.1 The model to be estimated in this report combines some features of the approaches sketched in the previous section. Export and import functions include both relative prices and income variables. The determination of absorption is an integral part of the model, along the lines of the absorption or expenditure approach. The model, however, departs from the elasticity or absorption approaches, in that it introduces essential dynamics. ;n the medium run, the effect of an exchange rate change should die away. For the monetary approach, reduction in real money stock due to higher domestic prices resulting from devaluation discourages absorption to improve the balance of payments, i.e., to increase the supply of money. As the real money otock approaches the original position, so does the balance of payments. 3.2 The model presented here does not depend on the critical role of money stock. As is a relative price change (between foreign and domestic prices) resulting ftom devaluation responsible for any short-run positive effect on the trade balance, so is it for the medium-run ineffectiveness of devaluation. The initial inflationary impact of devaluation magnifies as wage earners try to maintain their real purchasing power through wage adjustments in step with inflation. In this process, the relative price eventually converges to the pre-devaluation level to erode the initial favorable effect on the trade balance. Even though in this approach, unlike the monetary approach, the money supply is treated as exogenous, the monetary effect is reflected in the model. As devaluation leads to a reduction in real money stock, it will tend to improve the trade balance by discouraging absorption. 3.3 Though small, the model incorporates all the essential dynamics of determining the current balance. Few models geared to the evaluation of exchange rate policy give careful attention to these dynamics. Since the policy implications derived from such partial equilibrium models can be very misleading for the fairly open economies under study, it is essential to incorporate the feedback effects of changing income or economic activity (resulting from changes in real exports and imports) on import demand and domestic prices. Also important is how domestic prices, affected directly by a change in import cost and indirectly through the change in economic activity and inflationary expectations, influence export competitiveness and the relative price for imports over time. Finally, the export market is not properly treated in many models. At least for the countries under study, the small open economy assumption whereby the export price is given at the world price is too restrictive. Exports will, thus, be viewed as determined by the intersection of a downward sloping demand curve and a positivei- gloped supply curve. Exports 3.4 The quantity and price of merchandise exports are viewed to be determined by the intersection of their demand and supply curves. If the demand is infinitely price-elastic, the price is determined exogenously in the world market, and, for a given supply curve, it will determine the export - 10 - quantity. However, the a priori assumption of horisontal demand curve does not seem to be justified for the countries under study. Countries relying heavily on a few primary commodities for exports are usually a significant supplier of those commodities and certainly not price-takers. For most manufacturing exports as well, the number of competitive suppliers is typically limited due to product differentiation or noncompetitive industrial organization, making the price-taker assumption Anrealistic. 3.5 For the model being proposed, exports of major primary commodities are taken as given. The market structure and demand and supply behavior are so diverse from commodity to commodity that any analysis based on the aggregation of these primary goods is not assumed to be very meaningful. For instance, it is difficult to estimate the demand curve facing a country without explicitly considering the supply situation of other major suppliers. And the supply behavior is likely to be very different depending on the existence and effectiveness of cartels, the degree of current and potential competition, as well as the nature of the commodities (crops or exhaustible goods). 3.6 In general, exchange rate devaluation is not believed to be a useful tool for increasing exports of major primary commodities, though it may be effective for manufactured products. In the short run, demand is likely to change little, if other suppliers follow suit by adjusting their prices down- ward. In the long run, fairly price-elastic supply can also precipitate the price. The potential gain from an international suppliers' cartel together with the relative ease in administering subsidies and taxes and other produc- tion adjustments targeted for specific commodities might be other reasons for relying less on devaluation to increase exports. The analysis is, thus, - 11 - focused on the export market only for non-major primary commodities and manufactured products, for which the effect of devaluation is supposed to be more favorable and stable. 3.7 Table 3.1 shows that major primary commodities, as appear in the International Financial Statistics of the IMF, dominated merchandise exports until very recently in Malaysia and Thailand, and until the late 1970s in the Philippines. Weak commodity prices in recent years together with the countries' stepped-up industrialization efforts, however, reduced the export share of major primary commodities sharply to 46% for Malaysia, 32% for Thailand and as little as 151 for the Philippines in 1986. Even excluding the major primary commodities, exports of other primary commodities exceeded those of manufactured goods until the mid-1970s for Thailand and the Philippines and the early 1970s for Malaysia. In Korea, however, manufactured products dominated even from the initial stage of the export-oriented growth strategy in the mid-1960s. 3.8 Export demand is determined by foreign income or total world trade volume for relevant commodity groups and the relative export price of the country compared to the rest of the world. The exchange rate affects demand only by changing the export price. Export supply is mainly dependent on the profit margin (export price compared with production cost) as well as production capacity. (3.1) xd = xd (W f * n n n n x.n (3.2) Ks axs (p .e/p,p /p, xc xn x,n n where ad, x * Reil demand and supply of exports excluding major primary n commodities. - 12 - Table 3.1: TRENDS IN EXPORT COMPOSITION (Z) Major primary Others Items of commodities Subtotal major primary (xp) (xU) (Primary) (Manu!actured) commodities Korea Kor5- 100.0 43.7 56.3 None 1970 - 100.0 23.0 77.0 1975 - 100.0 17.9 82.1 1980 - 100.0 9.9 90.1 1986 - 100.0 7.7 92.3 Malaysia 1963 77.0 23.0 17.2 5.8 Petroleum 1970 80.8 19.2 11.6 7.6 Rubber 1975 71.9 28.1 9.3 18.8 Logs and 1980 72.8 27.2 7.3 19.9 timber 1984 57.5 42.5 14.6 27.9 Palm oil 1986 46.4 53.6 - - Tin Philippines 1965 71.9 28.1 22.4 5.7 Copra and 1970 73.9 26.1 18.9 7.2 coconut oil 1975 61.7 38.3 20.7 17.6 Sugar 1980 37.6 62.4 24.0 38.4 Copper 1985 17.4 82.6 25.4 57.2 Wood 1986 14.7 85.3 - - Thailand 1965 73.3 26.7 23.4 3.3 Rice 1970 67.7 32.3 24.1 8.2 Tapioca 1975 62.1 37.9 20.3 17.6 Rubber 1980 51.8 48.2 19.0 29.2 Tin 1985 36.9 63.1 23.6 39.5 Corn 1986 32.3 67.7 - - Sugar Sources: IMF, International Financial Statistics and World Bank database. - 13 - WTn = World trade volume of aon-oil products p *= World unit import value of non-oil products in dollars n pVn = Unit value of exports excluding major primary commodities xn in dollars e = Exchange rate per dollar p = Domestic wholesale prices p a Unit value of imports in dollars, and xn - Supply capacity of exports excluding major primary comodities. The supply equation (3.2) may be rewritten as follows with its price as the dependent variable. (3.3)p = p(ple, p , x /xc) (3.) x9n px9n a n n In the above, export profit margin is captured by the unit value of exports relative to the major costs represented by domestic wholesale prices and unit value of (intermediate) imports. For our purpose, the export capacity utilization ratio (x /xc) may be approximated by the growth rate of export a n volume (an Imports 3.9 Modelling merchandise imports is rather straightforward. Since the import structure of the countries under study is much more similar one another than that of exports, import volume may be estimated with a single aggregate equation. For oil-importing countries, however, a disaggregation into oil and non-oil imports may be desirable. Contrary to the case of exports, each country can be considered to be 'small' as far as imports are concerned, so - 14 - that unit value of imports is assumed to be exogenously given. The import demand function may be written as follows: (3.4) M = M (y, p .elp, ATOT, so/y) where ATOT is change in the external terms of trade. 3.10 Real imports are a function of real income or output and the cost of imports relative to domestic wholesale prices. Change in the terms of trade is also likely to affect real import demand, since worsening terms of trade results in a decline in real purchasing power for a given level of measured income. To the extent that the balance of payments situation is associated with the terms of trade movements, this variable may also reflect any import compression due to the deterioration of foreign exchange conditions. Finally, each component of aggregate demand has a varying degree of import intensity. Since manufactured exports are in general more dependent on imports of inter- mediate or capital goods, for instance, the GNP share of manufactured exports may also help to explain import demand. Absorption 3.11 Exchange rate devaluation affects exports and imports, and the resulting change in income will, in turn, affect absorption as well as imports. Thus, endogenizing absorption is essential for determining equili- brium income and providing important dynamics to the system. Absorption is composed of private consumption and investment and public expenditure. Private consumption (c ) may be viewed to depend on permanent income (yP) approximated by the weighted average of current and tagged income. If the weights are assumed to be geometrically declining, with the Koyck transforma- - 15 - tion, consumption function can be estimated by current income and the lagged dependent variable. If, however, current income is not fully known to consumers, the difference between actual and perceived income is likely to be positively correlated with change in income (Ay), and private consumption is expected to be negatively affected by hy, to result in the following equation. (3.5) c = c (y, Ay, c ) Private investment (i p) may be explained by a partial adjustment to the desired capital stock. If desired capital can be approximated by current income and annual capital depreciation is ignored, the investment equation can be written as a function of change in income and the lagged dependent variable.l (3.6) p = (y, p1 1/ From i 0 (K* - (1-d) K1) and - p1 K =. (1-d)i .i where K* and K are, respectively, desired and actual capital stock, and d is the constant annual depreciation rate, the following investment function is obtained. i * 0 (K* - (1-d) K1) + (1-8) (1-d) i p p-1 - 16 - Finally, public expenditure (g) is budgeted on the basis of expected aggregate income. To the extent that the discrepancy between actual and projected income is positively correlated with Ay, public expenditure will be negatively affected by Ay. (3.710g g (y, AD) 3.12 In the above, components of absorption are dependent only on income and own lagged values. However, consumer financing or general liquidity condition may also influence private consumption. Credit availability and cost variables will certainly be critical for private investment activities, and the investment speed toward desired capital stock will be variable depending on credit situations. On the other hand, real income in the above is measured in units of domestic goods. Thus, deterioration of the terms of trade (TOT) for a given level of measured income represents a decline in true real income, since more income would be spent to buy a given volume of foreign goods. Taking these considerations into account and combining the equations for cp, ip and g above, the absorption (ya) function may be written as follows.2/ 2/ With alternative specifications of ep = ep (yP) and ip = 8 (K* - (1-d) K-1 where K* = I*(yP) and K- -l 1 yY-1 y-2 ... ) ya may also be written as follows: y5 4 (y, 7-1 -2, *v , rr, ATOT) - 17 - (3.8) y C + i +g a p p * Ya (Yt hyya amr, rr, ATOT) where mr = Real money supply, and rr - Real interest rate. Inflation 3.13 As already emphasized, how domestic inflation is affected over time by exchange rate devaluation is ctitical to the dynamics of the whole system. Devaluation is generally believed to be effective, if at all, only in the short run. In the longer run, inflation accelerates such that the rela- tive price of domestic goods tends to return to the pre-devaluation level. In this ?rocess, the behavior of wages plays a critical role. The faster wages adjust to the initial inflationary impact of devaluation, the more short-lived will be the effect of devaluation on the trade balance. This s Aation will be more likely when an economy is near full employment, inflation has been a chronic problem, or labor workers are better organized to protect their interests. 3.14 Assuming that the cost of imports and unit labor costs are the major sources of inflation, inflation in the wholesale price index may be written as follows: (3.9) p =6(pm + e) + (146) w where () indicates the rate of change, w is the unit labor cost, and 6 and (1-6) are, respectively, the cost shares of wholesale products for imports and labor. The unit labor cost is nominal wages adjusted for labor - 18 - productivity, and may be estimated as the following expectation-augmented Phillips curve. (3.10) v = v ( p A where pe -'Expected inflation rate U = Unemployment rate, and Rate of labor productivity growth. The growth rate of labor productivity will have a role only to the extent that it is not fully reflected in current wages. If we choose not to directly estimate the equation of unit labor cost, given generally poor data on wages and employment, we may combine the above equations (3.9) and (3.10) to get the following inflation equation. (3.11) pp(p + e, p, y) a where GNP growth rate y is introduced as an instrumental variable for 1Iu and Since high economic growth is associated with both low unemployment and faster growth of labor productivity, the Phillips curve relationship will be weakened by the offsetting effect of productivity change. 3.15 The next task is to estimate the expected inflation rate p, . To make the expectation formation as general as possible, the following adaptive- and-extrapolative expectation is postulated, where the first and second parts, respectively, represent adaptive and extrapolative processes. (3.12) pe . (ap + (1-6) (ap1 + bp-2)) + 0 (F-1 - 2 - 19 - In the adaptive process, p may be considered as either a normal rate of inflation or a representative inflation rate three years before and earlier, and a + b = 1.1 Since the slowly changing p can be treated as almost constant, the inflationary expectation in Equation (3.11) may be represented by p_1 and p-2 The Complete Model 3.16 Though very small in sise, the model contains the current account balance, aggregate income and inflation, all endogenously determined and their interactions duly incorporated. The complete model including identities is as follows. 3/ The adaptive process can formally be derived from the following normal expectation, where recent (two years) inflation is partially corrected for the normal inflation rate ^* p * (a-1 + bP-2 - 1+ b -2 The process is also obtained from the more familiar adaptive expectation of Ss e A A A p , p-1* L P1- 1J which may be written as pe 1£ (1-m)£ p + Lp-1+ (1-f)tp-2 if the inflation rate three years back and earlier is represented by ^* . In this case, a > b is expected in equation (3.12). p - 20 - The Current Balance and External Debt: CB= E - M - i*D_l +En D D_1 - CB + Od E x .p + x .p p x,p n x,n x=x (WT , p */p n n n n x,n PNon Pxqn (plef pV X x) M m .p m - m (y, pm.e/p, ATOT, xn/y) where CS - Current account balance E Total nominal merchandise exports M = Total nominal merchandise imports i* = Average interest rate on external debt D * Gross external debt En * Net services balance except for interest payments on gross external debt Od = Residual item in the external debt identity, which includes changes in both foreign asset holdings and the value of non- dollar denominated debts outstanding due to their exchange rate changes against the dollar X = Real exports of major primary commodities, and px,p = Unit export value of major primary commodities in dollars. Aggregate Incomes y y + (x + xn -m) e* i* D.e*/p + In a p nx s ya. ya (Y, AyP ya, Am , rr, ATOT) - 21 - p* El (x + x) where e* * Average exchange rate per dollar in the base year of national income accounts px Composite unit value of exports in dollars, and Xn = Residual item in the real GNP identity, which includes net services exports (except for interest payments on gross external debt) and statistical discrepancies. Inflation: p = p (p + e, p-1 P2, y) 3.17 TD the above model, the effect of an exchange rate change can be traced as presented in Figure 3.1. Having presented the model, several points concerning limitations of the model are in order. In an effort to keep the model as small as possible, the services balance is treated as exogenous except for the interest payments on gross external debt. As was already made clear, exports of major primary commodities are also exogenous. These commodities are distinct from most manufactured goods to make the exchange rate policy relatively unimportant or unreliable as a tool for stimulating exports. However, even though devaluation may be ineffective in raising the export proceeds of major primary commodities, it is most likely to encourage their production and export in real terms. Thus, the exogeneity of these exports will understate the effect of devaluation on aggregate income and imports. This underestimation of imports will tend to offset any under- estimation of balance of payments effect resulting from exogeneity of the services balance. Figure 3.1 : FLOW OF THE EFFECT OF ANf EXCHANGE RATE CHANGE -~ c~wRen almcP iteret Ext0ma1 Oebt Real h t n Real - 23 - IV. ESTIMATION RESULTS 4.1 The model is estimated with annual data from the mid-1960s to 1986 using the ordinary least squares method. Equations with a sign of serial correlation are estimated by the Cochrane-Orcutt procedure. For some equations, with estimated parameters that do not seem to be constant during the sample period, a shorter sample of 1970-86 is chosen. That is the case for Korea's export, import and absorption equations, which is understandable in light of the rapid structural changes of the country's economy. 4.2 The Data. The IMF's International Financial Statistics (IFS) is the major data source for this model. Some data, including external debt, its average interest rate and national income statistics, are from the World Bank's data base, and national interest rate data are from the respective national sources. The world trade volume of non-oil products and its unit value are obtained from the information on total imports and oil exports. The real exports and export unit values for major primary commodities and other goods are obtained as follows for countries other than Malaysia. I = E (z1 (1980) *q) x n= E/p - x p EE.i/x x,p ( p where Ei - Export value of major primary commodity i in dollars, qi - Export quantity index (1980 - 1.00) of major primary commodity i, - 24 - and the major primary commodities are those listed in the international trans- actions section of the IFS's national tables. For Malaysia, where only the unit value of major primary commodities (p.,p) is available in the IFS rather than the unit value of total exports (px x,n is obtained from the export price indexes available in the Economic Report, Ministry of Finance, Malaysia. Two major corresponding categories in the export price indexes are food and manufactured goods classified by material, to which fixed weights of 16.9% and 83.1% are given, respectively, to get a Pxvn series. Real exports and the unit value of total exports are obtained as follows. x ZE./p p . x,p x = (E-ZE.) /p n 1 x,n p = E/(x + x ) 4.3 Real Exports. The demand elasticity of exports (excluding the major primary commodities) with respect to the world non-oil trade volume is estimated to be around 2.5 for Korea and Malaysia (see Table 4.1), 1.7 for the Philippiies, but as high as 3.3 for Thailand. This indicates that, for a given price competitiveness, Thai exports (excluding the major primary commodities) are most sensitive to world market conditions and grow faster with the growth of world trade. These income elasticities were fairly stable regardless of the sample periods. 4.4 The relative price elasticity is higher and has a longer lag for Korea. For the shorter sample of 1970-86, the estimated elasticity sums to 3.4, while it is 2.5 for the longer sample of 1966-86. The relative price elasticity is estimated to be 1.7 for the Philippines, 2.1 for Thailand and 0.85 for Malaysia. The high price elasticity of Korean exports is believed to result from the country's export structure which is dominated by manufactured - 25 - Table 4.1: REAL EXPORTS EXCLUDING MAJOR PRIMARY COMMODITIES (in xn) - -Korea--- Malaysia Philippines Thailand Constant -8.133 -8.858 -10.549 -4.401 -16.954 In WT. 2.451 2.557 2.534 1.688 3.346 (8.73)/b (39.7) (4.50) (7.46) (4.78) in (p */Pxn) 1.093 1.143 0.853 1.305 1.485 (3.31) (3.33) (4.30) (6.07) (5.93) in (p*n/Pxn),1 1.001 1.351 - 0.440 0.596 (3.29) (4.11) (2.12) (2.41) in (p*t/pxtn*2 0.390 0.897 - - - (1.36) (2.97) 0.711 - 0.833 0.647 0.873 (5.29) (10.8) (6.55) (23.1) R2 0.997 0.995 0.977 0.993 0.991 D.W. /a 1.74 1.48 1.50 2.41 1.91 Sample 1966-86 1970-86* 1965-86 1965-86 1965-86 /a Durbin-Watson statistic. /b Numbers in parentheses are t-values. - 26 - goods facing keen international competition in a wide range of products. The Malaysian price elasticity of less than 1.0 is somewhat disturbing, which suggests a decline in export proceeds in reaction to a decline in export price. Although poor data on the unit value of Malaysian exports may be partly responsible for this estimate, it may also be explained by the Malaysian export structure, characterized by the substantial controls of multinational corporations on manufactured exports. In electronics, the most active manufacturing export area in Malaysia, most exports are parts and components supplied by multinational firms to their affiliated companies. Price elasticity cannot be high when the market for exports lacks open competition, particularly when the parts and components constitute only a small portion of the total cost of the finished products. 4.5 Unit Value of Exports. Domestic wholesale prices in dollar terms (pie) is a significant variable explaining the unit value of exports (excluding major primary commodities) for all the countries (see Table 4.2). This variable, which serves as a measure of export production cost, may also be understood as the opportunity price in the competing (domestic) market. For a given increase in domestic wholesale prices, the export price adjustment is largest for Malaysia and smallest for the Philippines, although this relationship is statistically less tight than that of Korea and Thailand. The unit value of imports is significant only for Korea, which is understandable since Korean exports have relied heavily on imported intermediate and capital goods. The lagged export growth rate, introduced as an instrumental variable for the export capacity utilization rate, is marginally significant for Malaysia and Korea. Two statistically significant dummy variables are added to explain Korean exports. The 1973 dummy variable (D73) reflects the - 27 - Table 4.2: UNIT VALUE OF EXPORTS EXCLUDING MAJOR PRIMARY COMMODITIES (in px9n): 1965-86 Rorea Malaysia Philippines Thailand Constant 4.801 -0.493 2.249 1.817 In (p/e) 0.666 0.933 0.11 0.498 (7.73) (2.29) (2.88) (5.71) in pa 0.214 - * * (2.83) in (px,n1 - 0.267 0.321 0.437 (0.86) (1.45) (4.10) 4 In(x,1 0.149 0.299 0.076 - (1.65) (1.72) (0.61) Others 0.122 D73 * - * (3.31) 0.022 D(83-86) (3.02) R2 0.995 0.962 0.858 0.958 D.W. 1.34 1.49 1.35 1.83 D73 - I for 1973, 0 otherwise D (83-86) - 1 (1983), 2 (1984), 3 (1985), 4 (1986) and 0 otherwise. - 28 - prevalent price controls in 1973, which distorted the pie variable. The dummy variable D(83-86) is introduced to test the reaction of Korean exporters to a less favorable domestic policy regime to exports, with reduced interest subsidies and tax benefits. 4.6 Real Imports. The income elasticity of imports is estimated at 1.4 for Malaysia and around 1.0 for both the Philippines and Thailand (see Table 4.3). Korean income elasticity is almost 1.4 for 1966-86, but less than 1.0 for 1970-86. The price elasticity of imports is generally weak in statistical significance except for the Philippines. Still, the estimated price elasticities are remarkably similar across the countries: 0.42-0.62 for Korea, 0.59 for Malaysia, 0.52 for the Philippines and 0.46 for Thailand. A positive effect of terms of trade change is not found for Korean imports, though it is marginally significant for other countries. The relatively large coefficient, almosts 0.6, for Thailand seems to indicate that import management has been much influenced by the foreign exchange conditions resulting from terms of trade movements. The GNP share of exports excluding the major primary commodities (xnly) is significant only for the shorter sample of Korea, which is also explainable by the import-dependent Korean export structure. Finally, the significant Philippine dummy variable for 1984-86 (D84-86) indicates that import compression has been substantial since the country faced difficulty in external debt repayments. 4.7 Absorption. The long-run elasticity of absorption demand with respect to income is estimated at about 1.0 for all the countries (see Table 4.4). The reaction of absorption to income seems to be rather quick for the Philippines and Korea, but is slow for Malaysia. Change in income (Ay) was not significant for any of the countries and showed some signs of multi- - 29 - Table 4.3: REAL IMPORTS (ln m) - Korea Malaysia Philippines Thailand Constant -1.715 1.345 -5.503 5.399 -1.700 ln y 1.369 0.934 1.416 1.008 0.957 (8.67) (4.24) (7.80) (13.2) (2.88) ln (pm.e/p) -0.083 -0.185 -0.052 -0.520 -0.361 (0.37) (0.92) (0.16) (3.58) (0.53) ln (pm.e/p)_- -0.341 -0.303 -0.542 - -0.097 (1.71) (1.55) (1.39) (0.26) Aln TOT - - 0.055 - 0.276 (0.42) (0.82) A ln TOT. - - 0.137 0.307 0.319 (1.02) (2.02) (1.65) ln (7n/y) 0.122 0.252 - - (0.90) (2.12) Others - 0.208 ln m_j - -0.335 D(84-86) - (1.73) (6.86) 0.443 - 0.684 - 0.422 (3.48) (4.07) (1.53) R2 0.996 0.994 0.984 0.942 0.966 D.W. 2.18 2.08 1.92 1.99 2.07 Sample 1966-86 1970-86* 1966-86 1966-86 1966-86 D(84-86) - 1 for 1984-86, 0 otherwise. - 30 - Table 4.4: REAL ABSORPTION (in ya) =-os Malaysia Philippines Thailand Constant 0.222 0.155 0.377 -0.043 0.334 0.683 In y 0.750 0.623 0.430 in y 0.280 0.607 0.423 (9.87) (6.30) 0.346 In y-i (2.46) (3.80) (1.98) 0.202 in y-2 In (y).-1 0.236 0.371 - 0.723 0.318 0.524 (3.11) (3.65) (7.21) (1.99) (2.48) In ar 0.019 - - 0.196 - (0.75) (1.72) &in a ( 0.149 0.207 0.235 0.545 0.176 0.548 (2.89) (2.66) (2.81) (3.11) (1.81) (2.12) in (100+rr) -0.095 -0.129 -0.092 -0.275 -0.118 - (1.48) (1.94) (1.35) (2.23) (0.94) in (100+rr).I -0.101 -0.133 -0.162 - -0.206 -0.485 (1.01) (1.14) (1.46) (1.97) (2.47) &"in TOT - * - 0.104 - 0.044 (1.99) * (0.59) a TOT1 - - - 0.073 0.023 - (1.21) (0.45) Others * -0.049 08O -0.051 D8O 0.052 Dfis - - (1.82) (1.94) (4.17) R2 0.999 0.999 0.999 0.998 0.997 0.997 De. 2.43 2.14 1.99 2.57 1.43 1,57 Sample 1965-86 1970-86 1970-86* 1965-86 1964-86 1964-86 at = Real stock of M2 deflated by the wholesale price index (year average). rr a Interest rate minus wholesale price inflation rate. Interest rate is curb loan rate for Korea and representative bank lending rate for other countries (year average). D80 1 for 1980, 0 otherwise. Dfis - Dummy variable for fiscal policy stance (1 for 1981-83, -1 for 1985-86, 0 otherwise). - 31 - collinearity with other explanatory variables. This variable is excluded in favor of the stability of parameter estimates and comparability among the countries. When lagged income variables are used in place of the lagged dependent variable to estimate the permanent income version of the specifica- tion, the results are reasonable, actually marginally superior to the other specification, only for Korea. 4.8 The effect of monetary policy also proves to be generally significant. Liquidity condition, measured by a change in the real balance of broadly-defined money, has a relatively stronger impact on absorption for Malaysia and Thailand. Interestingly, the real interest rate also has a stronger long-run effect on absorption in these two countries, where financial markets are better developed and have probably been better functioning than in the other two countries. Change in the terms of trade, however, is signifi- cant only for Malaysia. Finally, the 1980 dummy variable (D80) for Korea's political turmoil is only marginally significant, but the Malaysian dummy variable for the change in fiscal policy stance during the 1980s (Dfie emerges as important in explaining absorption. 4.9 Inflation. The short-run impact of import cost on wholesale price inflation is remarkably similar across countries, with elasticity ranging from 0.41 for Malaysia to 0.48 for the Philippines (see Table 4.5). The lagged inflation rate, supposed to be associated with expected inflation in the determination of wages, is significant for Korea and the Philippines, where inflation has been much higher than in the other two countries. Though statistically insignificant, an extrapolative process of expectation formation seems to be in operation for Thai inflation. The Phillips curve relationship is estimated to be very weak except for Thailand, which is anticipated since a - 32 - Table 4.5: WHOLESALE PRICE INFLATION RATE (0) Korea Malaysia Philippines Thailand Constant 1.438 -0.342 -0.362 -5.037 (1.02) (0.22) (0.10) (1.23) Pm + 8 0.469 0.412 0.484 0.458 (17.8) (6.43) (6.62) (5.03) 0-1 0.320 0.169 0.144 - (8.86) (1.43) (1.50) 0-2 0.039 -0.004 0.106 - (1.10) (0.04) (1.20) 4 0-1 - - 0.154 (1.12) 9 0.172 - - 1.116 (1.44) (2.08) 9-1 - 0.173 0.486 (0.72) (1.07) Others 17.85 D(73,74,80) 35.78 D84 (11.2) (5.18) -5.22 D(81-86) (5.24) R2 0.988 0.859 0.900 0.708 S.E/Nean 1.54/12.02 2.33/3.34 5.61/14.53 5.09/6.10 D.W. 1.84 1.77 2.69 1.88 Sample 1966-86 1965-86 1964-86 1964-86 D(73,74,80) - -1 for 1973, 0.5 for 1974 and 1980, 0 otherwise. D(81-86) = 1.0 for 1981-and 1982, 0.9 for 1983, 0.8 for 1984, 0.7 for 1985, 0.6 for 1986. D84 = 1 for 1984, 0 otherwise. - 33 - high GNP growth is likely to be associated with fast labor productivity growth in manufacturing. The Thai inflation equation shows that domestic cost factors including wages and profits are mainly affected by economic growth, a proxy for capacity utilization of the economy, not by past inflation. 4.10 Two dummy variables are also tried to explain Korean inflation. D(73.74.80) is supposed to reflect the wide-ranging price controls in 1973 and the subsequent easing of controls during the two periods (1974 and 1980) of high imported inflation. D(81-86) is introduced to test whether the comprehensive stabilization policy pursued steadily throughout the 1980s had any effect. The latter effort, including an income policy stressing forward- looking expectations and consistent demand management, indeed seems to have been an important factor in Korea's successful price stabilization in the 1980s. The 1984 dummy (D84) for the Philippines suggests that the socio- political turmoil after the assassination of Mr. Aquino was responsible for more than half of the 67Z wholesale price inflation in that year. V. SIMULATION EXERCISES 5.1 Even though the model proposed has only five structural equations, the dynamic interactions among variables over time make it impossible to evaluate the effect of an exchange rate change without running simulations. In this section the following three simulation exercises are undertaken: - A 10% nominal devaluation in 1986 - A 10% real devaluation in 1986 - 34 - - A counterfactual simulation assuming that the real effective exchange rate was constant during the 1980-86 period. In case alternative equations are presented for a variable, the ones with an asterisk on the back of the sample period are used for simulations. A. Nominal Devaluation 5.2 For the basic scenario of the first simulation, which is run for the period of 1986-90, we have to input the values of exogenous variables in our model. These values for 1987-90 are assumed as follows: - World trade volume of non-oil products (WTn) grows 51 p.a. - World import price of non-oil products (pn*) and anit value of imports for each country (p.) increase at an annual rate of 41. - Average interest rate on external debt (i*) is fixed at 7% for all countries. - The year-average exchange rate (e) is the same as that of 1986, except for Korea, whose exchange rate is assumed to appreciate 6.5% annually during 1987-88 and 3.5% during 1989-90. - Real balance of broadly-defined money (mr) grows at an annual rate of 10% (5% for the Philippines). - Real lending rate (rr) does not change from the 1986 level. Although some of these assumptions may not be very realistic, our simulation results were little affected with some changes in the assumptions. 5.3 Unresponsive Nominal Money Supply. In the above, we assumed that the supply of money and interest rate are fixed in real terms. This means that nominal monetary growth varies with inflation due to devaluation, although it can hardly be called accommodating monetary policy for its non- responsiveness to income growth. If we choose to maintain money supply growth in nominal terms, the real variables have to be endogenized. - 35 - (5.1) m r Ms/p, and (5.2) rr a rb- where M' is nominal supply of the broadly-defined money, and rb is the representative tending rate. It is unreasonable, however, to treat both H' and rb as exogenous. For a given M' growth rate, devaluation will reduce the at growth rate when demand for mr growth is actually likely to increase due to the stimulatory effect of devaluation. Since keeping the nominal interest rate rb constant does not make sense in this case, we endogenize rb* 5.4 Although interest rates in the cotatries under study have been heavily regulated by the governments, they may still respond to money market conditions. Only for Korea, a free market rate is available for our use. The nominal interest rate is estimated in the framework of money market equilibrium. However, if capital transactions in and out of the country are free, the domestic interest rate is likely to be forced to move together with the international interest rate. From the following demand for real money balance (5.3) ad md (y, prb) r r together with a partial adjustment of the interest rate toward the money market equilibrium and the influence of the international interest rate, (5.4) rb rb rmy,e, rb-1i1 -36- where the expected inflation rate (pe) may be represented by p and p_,1 and a change in the LIBO rate (Ai*) is introduced as a proxy for change in the overseas borrowing cost, ignoring any exchange rate risk. The estimation results are presented in Table 5.1. As expected, the result is relatively good for Korea where the free market rate is used, and poor for the Philippines where interest rate policy was very rigid until interest rates were liberalized in 1981. Except for Thailand, inflation has a significant effect on the nominal interest rate, and inflationary expectation seems to be forued along the adaptive-and-extrapolative process or the extrapolative process as discussed before. The influence of the international interest rate is significant only for Thailand. Financial markets in these countries seem far from being closely integrated with the world market, although the result is somewhat surprising for Malaysia. 5.5 To actually get the effect of a 10% nominal devaluation, the following three simulations were run: - Base simulation, - A simulation with a 10% nominal devaluation from the exchange rate postulated in the base simulation, exogenous real money balance and real interest rate, - A simulation with a 101 nominal devaluation, exogenous nominal monetary growth (10% annually for all countries) and endogenous nominal interest rate. The results of the second and third simulations were then compared with the base simulation results. The differences are the net effect of devaluation, and are presented in Table 5.2 and 5.3. The following discussions of the results will mainly concentrate on those in Table 5.2 and Figure 5.1. 5.6 Nominal devaluation leads to an improvement in the current account balance for Korea, the Philippines and Thailand. In general, the effect on - 37 - Table 5.1: NOMINAL LENDING RATE (rb) Korea Malaysia Philippines ---- Tha 1and------- Constant -20.18 0.779 -10.70 -52.88 -50.13 (0.48) (0.65) (1.05) (2.20) (2.46) in ar -4.683 -8.274 an r - -3.129 in (ad_-1 -2.780 in (mr)-, (2,46) (3.56) (2.05) (2.10) in y 4.730 - 4.167 in y., 4.723 in y., 4.488 ln y. (1.10) (1.57) (2.26) (2.57) 0.252 - 0.210 0.015 (4.47) (3.96) (0.71) -0.148 - -0.152 -0.001 - (2.52) (2.10) (0.06) 0.044 -(0.010 (2.16) (0.61) - - - - 0.128 (2.35) (rb)-1 0.808 1.045 0.495 0.939 0.922 (5.81) (7.29) (2.52) (6.33) (7.95) R2 0.959 0.791 0.699 0.894 0.918 S.E./Mean 2.58/40.9 0.44/8.25 3.58/14.5 0.67/15.6 0.59/15.6 D.W. 1.63 1.37 2.15 1.80 1.96 Sample 1965-86 1965-86 1964-86 1964-86 1964-86* - 38 - Table 5.2: EFFECT OF A 10% NOMINAL DEVALUATION IN 1986 WITH EXOGENOUS REAL MONEY AND REAL INTEREST RATE (2 Difference from the base simulation results) Unit value Real Real Real Real Wholesale of exports exports imports Current balance GNP absorption prices (p) x) ) $ million Z of imports () (ya (P) Korea 1986 -3.18 3.76 1.04 -158 -0.50 1.47 0,63 4.79 1987 -1.40 6.16 2.10 1,173 3.13 3.53 2.02 6.80 1988 -0.97 6.10 3.70 979 2.18 4.45 3.42 7.84 1989 -1.09 3.91 3.74 -195 -0.37 4.26 4.08 8.22 1990 -1.29 3.91 3.90 -508 -0.80 4.29 4.21 8.38 -2.69 2.32 0.97 -155 -1.43 1.01 -0.00 3.84 1987 -2.46 2.11 -0.85 18 0.17 2.06 0.40 4.65 1988 -3.26 2.83 Q.58 -160 -1.40 2.53 0.92 4.96 1989 -3.25 2.82 0.96 -231 -1.89 2.75 1.38 5.09 1990 -3.55 3.09 1.54 -357 -2.63 3.05 1.81 5.15 Philippines 1986 -1.28 1.73 -1.71 112 2.08 1.05 0.63 4.28 1987 -1.30 2.42 -0.99 127 2.22 1.60 1.14 5.43 1988 -1.20 2.31 0.01 85 1.41 1.78 1.44 6.35 1989 -1.17 2.21 0.35 73 1.12 1.91 1.62 6.69 1990 -1.12 2.11 0.31 71 0.99 1.98 1.72 6.91 Thailand 1986 -1.16 1.94 -1.22 179 1.95 0.78 0.28 5.07 1987 -1.27 3.21 -0.55 264 2.59 1.46 0.76 6.72 1988 -1.46 3.66 0.32 259 2.37 1.91 1.19 6.68 1989 -1.53 3.87 0.80 284 2.35 2.33 1.60 6.91 1990 -1.49 3.81 1.26 287 2.12 2.59 1.93 7.27 Figure 5.1: EFFECT OF A IO NOKINAL DEVALUATION IN 1986 WITU EXOGENOUS REAL WUNEY AND REAL INTEREST RATE (Z) Real Exports Current Balance a (% of Imports) . 86 87 88 89 90 86 87 88 89 90 Real GNP wholesle Prices 7& 5as as 86 87 88 89 90 6 87 88 89 93 gS Korea + Malaysia * Philippines å Thailand - 40 - the current balance shows a U-curve pattern with the exception of Thailand. The current balance deteriorates or improves slightly in the first year of devaluation, followed by the biggest improvement in the next year and smaller improvements or deterioration thereafter. This pattern results mainly from the following dynamics in operation. - Real exports respond to a change in unit value of exports with some time lag, and, though not significant except for Thailando even the export unit value responds to a change in the exchange rate with a time lag. - As time passes, the initial impact of devaluation on domestic prices amplifies due to inflationary expectation as well as higher economic activity induced by devaluation. It weakens the initial price competitiveness of exports and the cost disadvantage of imports. - Absorption also reacts to devaluation-induced income growth with some time lag to make income/output grow steadily over time and to lead to rising demand for imports. For Korea, for instance, real exports are 6.2% higher in the second year of devaluation, but the gain reduces to 3.9% by the fourth year. For Thailand, the positive effect of devaluation on the current balance is sustained longer, since the downward adjustment of export unit value resulting from devaluation takes a longer time. 5.7 In the first three years of a 10% nominal devaluation, the average improvement in the current balance ranges from 1.6% of total imports (Korea) to 2.3% (Thailand), but, thereafter, the effect remains relatively strong only for Thailand. The favorable effect is particularly short-lived for Korea, as growth-induced import demand eventually surpasses declining export gains. Devaluation does not improve the current balance of Malaysia, because export value drops with a lower unit value (due to the inelastic price inelasticity of export demand) and that income effect is generally stronger than the price effect for import demand. Devaluation results in the largest decline in - 41 - export unit price for Malaysia, which does not help the trade balance, although it contributes to the expansion of real exports and income growth. 5.8 The growth effect of devaluation is strongest for Korea, followed by Malaysia. By the second year of devaluation, real GNP is 1.5-3.5% higher, which becomes 1.9-4.3% higher by the fourth year. The effect on absorption is basically similar to that on income, since absorption grows together with income with some time lag. Finally, devaluation is very inflationary. Through inflationary expectation and the Phillips curve relationship, a 10% nominal devaluation leads to initial inflation of 4.8% for Korea and 8.4% higher prices by the fifth year of devaluation. For Thailand and the Philippines, the first-year price impact starts with 4.3-5.1%, which increases to 6.9-7.3% by the fifth year of devaluation. By this time, the gap between domestic wholesale prices and export unit value in domestic currency is almost closed. The inflationary effect of devaluation is relatively small for Malaysia, where inflation has been minimal, since neither inflationary expectation nor the Phillips curve relationship is significant. 5.9 The picture is little changed in Table 5.3, which shows the effect of devaluation without an accompanying adjustment in the nominal stock of broadly-defined money. In this case, the real money balance becomes smaller and smaller as prices keep rising, thus squeezing absorption. This tight credit in the presence of growing economic activity raises the nominal interest rate and, in time, the real rate as well, which also exerts a contractionary effect on absorption. In the first year of devaluation, absorption is actually higher for Korea and Malaysia than the previous case of exogenous real money and real interest rate. This is so because the positive effect of a sharp drop in the real interest rate dominates the negative effect -42 - Table . EFFCT OF A 101 NOMINAL DEVALUATION IN 1986 WITR EXOGENOUS NOMINAL mogy AND EDOGENOUS INTEREST RATE (I Difference from the base simulation results) Unit value Real Real Real Real Wholesale of exports exports imports . Current balance GNP absorption prices (px) (x) (a) $ million I of iports (Y) (ya) (P) Korea 1986 -3.13 3.71 1.30 -239 -0.76 1.86 1.12 4.86 1967 -1.46 6.16 1.82 1,244 3.30 3.02 1.43 6.72 1988 -1.04 6.23 3.28 1,160 2.54 3.94 2.74 7.71 1989 -1.18 4.18 3.20 157 0.29 3.60 3.11 8.04 1990 -1.39 4.23 3.34 -28 -0.04 3.65 3.19 8.19 Malaysia 1-8 -2.69 2.32 1.96 -251 -2.32 1.70 1.12 3.84 1987 -2.41 2.07 -2.03 141 1.48 1.20 -0.94 4.78 1988 -3.32 2.89 -0.85 5 0.11 1.46 -0.94 4.83 1989 -3.34 2.90 -0.34 -61 -0.43 1.89 -0.35 4.88 1990 -3.67 3.20 0.46 -191 -1.30 2.36 0.24 4.97 Philippines 1986 -1.28 1.73 -2.37 147 2.72 0.38 -0.15 4.28 1987 -1.37 2.53 -1.62 167 2.92 1.13 0.55 5.10 1988 -1.28 2.46 -0.61 133 2.14 1.31 -0.83 6.07 1989 -1.26 2.38 -0.03 112 1.64 1.68 1.27 6.39 1990 -1.19 2.26 0.28 103 1.37 1.85 1.49 .6.73 Thailand 1986 -1.16 1.94 -1.22 179 1.95 0.78 0.28 5.07 1987 -1.38 3.39 -1.02 318 3.15 1.13 0.28 6.33 1988 -1.63 4.04 -0.20 345 3.19 1.60 0.66 6.25 1989 -1.71 4.33 0.35 388 3.25 2.04 1.05 6.57 1990 -1.71 4.33 0.72 430 3.23 2.22 1.22 6.85 - 43 - of a lower real money balance. Except in the Philippines where the nominal interest rate is not affected by the real money balance, the real interest rate becomes higher than the base case from the third year. In general, the contractionary effect of fixed nominal money supply is fairly strong. Lower GNP growth means fever imports, a little lower domestic inflation and a greater drop in export unit value, higher real exports and, thus, an improved balance of payments. 5.10 Before turning to the simulation of real devaluation, it may be of interest to see how different the effect of nominal devaluation in 1986 would be from that of, say, 1981. From Table 3.1, we have noted that the export structure of the countries other than Korea changed drastically during the first half of the 1980s, which, together with the changing GNP share of exports, will certainly make differences in the effect of devaluation. The effect of nominal devaluation for the two periods is compared in Table 5.4. This shows that the effect on exports and growth is weaker for the earlier period, when exports other than the major primary commodities accounted for a much smaller share of total exports and CUP. Naturally, the differences in the effect between the two periods are larger for Malaysia and Thailand, whose export structures underwent more rapid changes in recent years. In the case of Malaysia, the effect on the current balance is much more favorable for the earlier period. Because of the weaker impact of devaluation on growth for the earlier period, the import-reducing price effect more than compensates for the income effect as well as the slight reduction of export value due to the low price elasticity of demand for Malaysian exports. - 44 - Table 5.4: COMPARISON OF THE EFFECT OF A 10% NOMINAL DEVALUATION IN 1981 AND 1986 WITH EXOGENOUS REAL MONEY AND REAL INTEREST RATE (% Differences from the Basic Simulation Results) Current balance Real exports Real GNP (% of total imports) /b 1981-85 1986-90 1981-85 1986-90 1981-85 1986-90 Korea First Year 3.73 3.76 1.28 1.47 -0.45 -0.50 2nd & 3rd Year /a 5.89 6.13 3.22 3.99 1.71 2.66 4th & 5th Year 3.38 3.91 3.71 4.28 -0.74 -0.59 Malaysia First Year 1.28 2.32 0.36 1.01 -0.39 -1.43 2nd & 3rd Year 1.19 2.47 1.31 2.30 0.84 -0.62 4th & 5th Year 1.55 2.96 1.81 2.90 -0.22 -2.26 Philippines First Year 1.27 1.73 0.95 1.05 1.56 2.08 2nd & 3rd Year 1.61 2.37 1.40 1.69 0.92 1.82 4th & 5th Year 1.61 2.16 1.55 1.95 0.70 1.06 Thailand irst Year 1.38 1.94 0.57 0.78 1.30 1.95 2nd & 3rd Year 2.07 3.44 1.03 1.69 1.47 2.48 4th & 5th Year 2.73 3.84 1.69 2.46 1.44 2.24 /a The 2nd & 3rd year and 4th & 5th year values are averages of the individual year figures. /b For 1981-86, percent of actual imports. B. Real Devaluation 5.11 In the above case, the nominal exchange rate was the policy variable. However, if the government believes that its exchange rate is 10% overvalued, it may want to maintain the 10% devaluation over time in real terms. Since devaluation brings about higher domestic prices, a given percent of real devaluation means that the nominal exchange rate must be continually depreciated in step with rising prices. Simulating the effect of real devaluation, thus, requires endogenizing the nominal exchange rate in our - 45- model. From the real effective exchange rate (REER) formula (expressed in indexes), REER e/p H(ei/p)wi (5.6) e = p*REER H(e I/p.)i where ei and pi are, respectively, the exchange rate per dollar and wholesale prices of the major trading partner i, and wi is the weight (trade share) for country i. Now, the REER is exogenous and the nominal exchange rate moves with domestic wholesale prices. 5.12 The following two simulations were run for 1986-90 to evaluate the effect of real devaluation. - Base simulation: the REER for 1987-90 is assumed to be the same as that of 1986, except for Korea where 6.5% and 3.5% annual real appreciation is assumed for 1987-88 and 1989-90, respectively. The composite real exchange rate of the major trading partners, n(e./p.)wi, for 1987-90 is assumed to be constant and 8% lower than that ot 1986. - A simulation with a 10% real devaluation: the REER is 10% devalued throughout 1986-90 from the rate assumed in the base simulation. The effect of a 10% real devaluation is obtained as the differences between the above two simulation results, and is presented in Table 5.5 and Figure 5.2, 5.13 We note that maintaining a given rate of real devaluation is very inflationary. For instance, a 10% real devaluation for Korea requires the nominal exchange rate to be depreciated 551 by the fifth year and domestic wholesale prices raised by 41% compared with the base case. For Malaysia where the long-run inflationary effect of devaluation is the smallest, a 101 real devaluation is achieved with a much lower 22% nominal depreciation by the fifth year. - 46 - Table 5.5: EFFECT OF A 101 REAL DEVALUATION IN 1986 WITH EXOGENOUS REAL MONEY AND REAL INTEREST RATE (2 Difference from the base siaulation results) Unit value Real Real Real Real Wholesale of exports exports imports Current balance GNP absorption prices (pX) (x) (a) $ million of imports (y) (ya) (P) Korea 1986 -6.15 7.52 2,06 -319 -1.01 2.94 1.25 10.07 1987 -5.13 15.71 4.99 2,238 6.10 8.26 4.51 16.75 1988 -5.12 .20.69 9.88 2.609 6.14 12.97 8.95 27.12 1989 -5.56 20.15 12.79 999 2.10 15.96 12.96 34.39 1990 -6.21 21.85- 15.81 387 0.71 18.96 16.24 40.89 - *4.38 3.84 1.59 -254 -2.35 1.65 -0.01 6.23 1987 -4.51 3.92 -1.26 -4 -0.03 3.61 0.65 8.58 1988 -6.08 5.40 0.89 -287 -2.50 4.81 1.62 9.83 1989 -6.39 5.69 1.69 -435 -3.61 5.50 2.54 10.57 1990 -7.11 6.38 2.89 -687 -5.16 6.26 3.47 11.00 Philippines 1986 -2.28 3.12 -3.05 199 3.69 1.86 1.13 8.54 1987 -2.79 5.06 -2.39 273 4.76 3.28 2.29 12.89 1988 -3.12 5.83 -0.95 253 4.23 4.22 3.27 17.56 1989 -3.38 6.37 -0.11 262 4.08 5.04 4.09 20.88 1990 -3.54 6.70 0.54 279 3.97 5.69 4.74 23.72 Thailand 198- -2.48 4.17 -2.53 371 4.04 1.65 0.58 11.84 1987 -3.63 8.55 -1.92 703 6.97 3.78 1.84 20.88 1988 -4.40 11.14 0.19 815 7.41 5.53 3.25 23.62 1989 -5.03 12.87 1.91 953 7.81 7.25 4.71 25.95 1990 -5.56 14.13 3.47 1,116 8.09 8.87 6.16 29.67 Figure 5.2: EFFECT OF A 10% REAL DEVAILUATION IN 1986 VITN EXOGENOUS REAL NONEY AND REAL INTEREST RATE (2) .Real Exports * Current Bolonce (Z of Imports) see ts '4 glKra aayi Piipie Taln - 48 - 5.14 As anticipated, the effect of real devaluation on the economy is much larger. For Korea, with a 10% real devaluation, the current balance improves $2.2 billion and $2.6 billion in the second and third years, respectively, each representing 6.1% of total imports. As income and import growth continues while export expansion levels off, the current account starts to lose most of the earlier gain from the fourth year. The effect of real devaluation is the strongest and longest lasting for Thailand. By the third and fifth years, the annual current balance improvement is, respectively, over $800 million and $1.1 billion, representing 7.4% and 8.1% of total imports, respectively. The current balance improvement for the Philippines is somewhat smaller, and the Malaysian current balance deteriorates with real devaluation. The growth stimulation effect of real devaluation is the strongest for Korea, followed by Thailand, Malaysia and the Philippines. C. Counterfactual Simulation: Constant REER for 1980-86 5.15 The final stimulation exercise is designed to show what the economy might have been if the country's exchange rate policy during 1981-86 was to maintain the RER constant at the 1980 level. Again, we endogenize the nominal exchange rate as we did in the previous simulation. (5.6) e = p*REER H (e /p.)wi. The effect of deviation of the REER from the 1980 level can be evaluated by comparing the base simulation results using the actual REER with those of the policy simulation with the constant REER. The results are presented in Table 5.6, where the actual movement of the REER (1980=100.0) during this - 49 Table 5.6: COUNTERFACTUAL SIMULATION: WHAT IF THE REER WAS UNCHANGED DURING 1980-86 (% Difference from the base simulation results) Real /a Real Exchange exports Current balance GNP (Actual) Prices rate Actual REER (x) CD (Actual) (y) () (p) (a) (1980 - 100) Korea 9IN1 3.19 -101 (-4,646) 1.09 (7.41) 4.06 8.45 96.0 1982 7.28 436 (-2,650) 3.09 (5.68) 8.31 13.92 95.1 1983 5.19 609 (-1,606) 3.67 (10.94) 6.20 5.28 100.9 1984 -1.33 -373 (-1,372) 1.85 (8.65) 2.00 -1.57 103.6 1985 -9.62 -976 (-887) -2.20 (5.40) -7.16 -15.71 110.1 1986 -23.56 -1,520 (4,617) -10.60 (11.90) -24.16 -40.98 128.5 Malaysia 1981 0.82 -29 (-2,061) 0.23 (3.93) 2.66 6.27 96.6 1982 2.68 35 (-3,200) 1.35 (2.90) 9.25 22.23 89.4 1983 3.40 251 (-2,751) 3.19 (0.93) 13.18 29.37 87.5 1984 5.14 104 (-1,268) 4.76 (-5.52) 16.41 35.73 85.8 1985 4.11 81 (-18) 5.56 (-4.20) 13.96 26.25 90.3 1986 -2.50 153 (996) 2.70 (1.32) 1.20 -11.05 113.8 Philippines 1981 1.54 160 (-2.569) 1.15 (6.29) 6.10 12.37 94.4 1982 3.93 305 (-1,003) 3.09 (4.10) 13.40 12.80 89.4 1983 1.93 -27 (-2,874) 1.99 (5.82) 5.23 4.73 100.5 1984 4.22 173 (-2,109) 3.46 (6.12) 15.42 28.42 89.9 1985 8.43 400 (-1,537) 5.86 (3.65) 26.99 50.86 84.2 1986 3.03 -23 (247) 3.45 (3.79) 19.14 15.38 103.2 Thailand 1981 1.23 116 (-2,486) 0.51 (6.94) 4.91 8.81 96.4 1982 1.98 168 (-3,601) 0.99 (5.94) 7.76 11.94 96.3 1983 3.59 251 (-3,497) 1.65 (6.25) 10.76 18.19 93.7 1984 3.52 191 (-1,671) 2.02 (7.76) 7.85 10.70 97.4 1985 -1.86 -331 (-734) 0.14 (-1.02) -8.82 -17.84 110.0 1986 -12.12 -1,288 (-296) -4.53 (1.00) -28.21 -44.46 129.3 /a In million US dollars. - 50 - period is shown in the last column. (For details of calculating the REER, see the next section.) 5.16 We note that the exchange rates of these countries, which generally appreciated in the early years of the 1980s, have depreciated sharply since 1984 on a real effective basis. For Malaysia and Thailand, the depreciation of the REER was as large as 33% during the two-year period of 1985-86, while it was somewhat smaller for Korea (24%) and the Philippines (15%). The table shows that, with the unchanged REER, Korea's current account surplus in 1986 could have been $1.5 billion smaller than the actual $4.6 billion, domestic wholesale prices 242 lower, and real GNP 10.6% smaller in 1986. For Thailand, with the constant REER throughout 1980-86, the current account deficit could have been $1.3 billion larger than the actual deficit of $300 million in 1986, wholesale prices 28% lower and real GNP 4.5% smaller. In spite of a large real depreciation in 1986, the Philippine REER generally appreciated until 1985 from the 1980 level, costing a $300 million deterioration in the current balance in years like 1982 and 1985 and accumulated real GNP of 3.5% by 1986. The Malaysian REER was also kept fairly strong during the first half of the 1980s. Our counterfactual simulation result shows that the unchanged REER could have improved the Malaysian current balance slightly. - 51 - VI. EXCHANCE RATE MANAGEMENT 6.1 Having assessed the effects of exchange rate devaluation in the previous section, we now turn to the evaluation of exchange rate policy. In the countries under study, the exchange rate is heavily managed by the authorities. Under normal circumstances of a managed exchange rate system, a country is supposed to maintain rather constant real effective exchange rate (REER). Thus, the focus of investigation here will be on how the REER in each country has behaved, and whether or not the observed behavioral pattern is consistent with our findings on the effects of devaluation. A. Currency Basket 6.2 In order to calculate the REER, a basket of currencies should first be formed. 'As a compromise between economic reality and convenience, only the currencies of the United States, Japan and several major European trading partners are included in the basket. Although some countries have a fairly large trade share with nearby developing economies, these trading partners may be left out without much cost, since they generally do not represent competi- tive trade relationships, like the Malaysian exports of crude oil to Singapore for refining. Weights given to each currency are based on two-way trade values with the country. For simplicity and practicality, fixed weights are used on the basis of trade performance during the 1980s. In Europe, four countries are chosen as the major trading partners: the United Kingdom, the Federal Republic of Germany, France and the Netherlands. Currencies of these four countries are given "all Europe" weight proportionately distributed among them according to their trade values. Otherwise, Europe as a whole is under- represented vis-a-vis the United States or Japan. Implicitly assumed is that - 52 - the "all Europe" exchange rate and prices can be approximated by these four European trading partners. 6.3 As shown in Table 6.1, the United States, Japan and Europe accounted for 60% of the total trade for Korea and the Philippines during 1981-85, 55% for Thailand and 53% for Malaysia. Trade with Singapore was as much as 18% of the total for Malaysia and 7% for Thailand. Korea and the Philippines have very similar geographical trade structurest the United States accounts for 27-28%, followed by Japan with 18-20% and Europe with only about 13%. For Malaysia and Thailand, the trade share for the United States is about 14%, half the US share for Korea and the Philippines. Japan is by far the biggest trading partner for Malaysia, accounting for 23% of the total. For Thailand, Japan and Europe have about equal trade shares of 21%., Currency baskets are formed on the basis of these geographical trade structures. Weights for the US dollar is as high as 46-47% in the Korean and Phlippine currency baskets, while the Japanese yen is most important in the baskets of Malaysia and Thailand. B. Trend of the Real Effective Exchange Rate 6.4 Given the composition of the currency basket, the REER is calculated as follows. REER = Effective Exchange Rate/Relative Price (6.1) =e P _ lie.i Up. i .e/p B(ei pi)wi - 53 - Table 6.1: TRADE SHARES AND WEIGHTS IN CURRENCY BASKET BY MAJOR TRADING PARTNER /a (2) Korea Malaysia Philippines Thailand Trade Currency Trade Currency Trade Currency Trade Currency share weight share weight share weight share weight US (dollar) 27.4 45.6 14.4 27.0 28.4 47.3 13.9 25.2 Japan (yen) 19.8 33.0 23.2 43.5 18.4 30.7 20.7 37.5 Europe 12.8 21.4 15.7 29.5 13.2 22.0 20.7 37.3 UK (pound sterling) 2.7 7.3 3.3 8.3 2.9 6.0 2.3 6.4 West Germany (sark) 3.0 8.1 3.7 9.1 3.9 8.0 4.0 11.3 France (franc) 1.2 3.2 1.6 3.9 1.7 3.5 1.9 5.3 Netherlands (guilder) 1.0 2.8 3.3 8.2 2.2 4.5 5.1 14.3 Others 40.0 - 46.8 - 39.9 - 44.7 - (Singapore) (1.3) (18.1) (3.1) (7.3) /a Trade shares are five-year (1981-85) averages of the two-way trade weights. - 54 - where i represents each of the six countries whose currencies are included in the basket. The calculation of the composite exchange rate and wholesale prices of the major trading partners in the REER formula is based on geometric rather than arithmetic averages, because the former has superior properties. For one thing, in the case of the geometric average, the REER result is unchanged whether the exchange rate is measured in units of foreign currency for one unit of own currency or the other way around. Thus, the geometric average is more reliable, particularly in the medium- to long-run analysis where the value of a currency can undergo significant changes. For Malaysia where wholesale prices are not available, they are estimated by using consumer prices and utilizing the relationship between wholesale and consumer prices in the other three countries.4/ 4/ The WPI series for Malaysia was created on the basis of the following equation estimated with the pooled data of Korea, the Philippines and Thailand. - 0.008 + 0.968 C + 0.181 a, - 0.161 P-1 (0.05) (12.6) Pc (5.00)(n*e)-1 (1.38)yd_ -1l- R2 = 0.918 S.E. - 0.038 Sample * 1964-86 for Korea, the Philippines and Thailand where pc is consumer prices and Yd is real gross domestic product. The equation shows that import costs (p *e) affect wholesale prices more than consumer prices, and that, thouh statistically insignificant, the Phillips curve relationship is weaker for wholesale prices than it is for consumer prices. Table 6.2: TEND OF T1E REML EFFECTIVE ECI1CE RATE /a (1980 - 100.0) Korea alayuta Philippines Thailand Nant- Effer- Relative nmi- f fec- Relative Mt- Effec- Relative W5nl- E fec- Relative nal tive price REER nal tt¥* price aEEm Dal tive price IER mat tive price REER 1963 21.4 17.1 22.8 75.1 140.6 102.9 148.3 69.3 52.1 41.5 40.1 103.5 101.7 72.1 74.0 97.4 1964 35.2 28.1 30.4 92.4 140.6 102.9 144.9 71.0 52.1 41.5 41.5 100.1 101.6 72.0 68.6 104.9 1965 43.9 35.0 32.9 106.6 140.6 102.9 139.8 73.6 52.0 41.5 41.8 99.3 101.6 72.0 69.5 103.6 1966 44.7 35.7 34.8 102.6 140.6 102.9 135.0 76.2 51.9 41.4 43.4 95.3 101.6 72.0 77.0 91.5 1967 44.5 35.5 36.8 96.6 140.6 102.8 137.6 74.7 51.9 41.4 44.2 93.6 101.6 71.9 82.3 87.3 1968 45.5 36.0 39.1 92.0 140.6 101.6 134.1 75.4 51.9 41.0 44.2 92.9 101.6 71.3 77.5 91.9 1969 47.4 37.5 40.4 92.6 140.6 101.5 128.9 78.7 51.9 41.0 43.4 94.4 101.6 71.2 78.1 91.3 1970 51.1 40.5 42.4 95.5 140.6 101.9 124.3 82.0 78.6 62.3 50.9 122.3 101.6 71.5 74.6 95.9 1971 57.1 46.0 45.0 102.3 140.2 103.8 123.2 84.2 85.6 68.9 57.5 119.8 101.6 73.2 73.1 100.1 1972 64.7 55.4 49.6 111.6 129.5 104.2 122.2 85.3 88.9 75.9 61.3 123.8 101.6 79.5 76.5 101.8 1973 65.6 59.4 47.0 126.5 332.2 97.6 120.1 91.3 89.9 81.4 67.2 121.0 100.7 85.8 83.9 102.3 1974 66.6 58.8 54.4 308.1 110.6 93.1 119.0 78.3 90.4 79.8 81.2 98.3 99.5 82.7 88.4 93.5 1975 79.7 70.4 64.3 109.6 110.0 92.9 117.3 79.2 96.5 85.5 80.0 106.9 99.5 83.7 86.6 96.7 1976 79.7 68.9 68.2 101.1 116.8 96.0 111.2 86.3 99.1 86.0 86.2 99.8 99.6 81.5 84.9 96.0 1977 79.7 71.6 70.6 101.3 113.1 98.0 109.7 89.4 98.6 88.9 88.1 100.9 99.6 86.0 87.3 98.4 1979 79.7 79.5 76.1 104.5 106.4 106.2 111.4 95.3 98.1 97.8 89.1 109.8 99.3 98.2 92.0 106.8 1979 79.7 79.9 82.4 97.1 100.5 101.1 105.5 95.8 98.2 98.6 96.5 102.1 99.7 100.2 94.5 105.9 1980 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 1981 112.1 108.5 113.1 96.0 105.8 100.9 104.5 96.6 105.2 101.4 107.4 94.4 106.6 99.5 103.2 96.4 1982 120.4 109.2 114.8 95.1 107.3 94.0 105.2 89.4 113.7 103.1 115.3 89.4 1t2.3 96.5 100.3 96.3 1983 127.7 115.2 114.2 100.9 106.6 92.8 106.0 87.5 147.9 133.3 132.7 100.5 132.3 95.1 101.5 93.7 1984 132.7 116.7 112.6 103.6 107.7 90.4 105.4 85.8 222.3 195.1 217.1 89.9 115.4 93.5 96.0 97.4 1985 143.2 124.9 113.4 110.1 114.1 94.7 104.9 90.3 247.7 215.6 256.2 84.2 112.6 106.0 95.5 111.0 1986 145.1 149.3 116.1 128.5 118.6 123.2 108.3 113.8 271.4 277.6 268.9 103.2 128.4 129.2 99.9 129.3 la The exchange rates are per US dollar. - 56 - 6.5 The trend of the REER is shown in Table 6.2 together with that of the nominal exchange rate, the effective exchange rate based on the currency basket and the relative wholesale price. The table shows that the nominal exchange rates were fixed to the US dollar until 1969 for the Philippines, until 1971 for Malaysia and until 1972 for Thailand. For Korea, the only major period of a fixed exchange rate was during 1975-79. The fixed exchange rate regime was associated with a tendency toward real appreciation in Korea and the Philippines where the inflation rate was relatively high, and of real depreciation in Malaysia where inflation was lower than the average of the major trading partners. The magnitude of real appreciation was about 10% for both Korea (during 1975-79) and the Philippines (between 1963 and 1969). For Malaysia, the depreciation was more than 20% between 1963 and 1971 on the real effective basis. The fixed exchange rate regime of Thailand was associated with some real depreciation during the initial period, followed by steady real appreciation and correction of the appreciation. 6.6 Perhaps another notable pattern is the movements of the REER in relation to the strength of the US dollar against other major currencies. The sharp depreciation of the REER in 1986 for all four countries suggests that their exchange rate management is biased toward the US dollar rather than being strictly based on the basket of currencies. In 1986, the depreciation on the real effective basis was 16.5% for Korea and Thailand, 23% for the Philippines and 26% for Malaysia. This pattern, though less pronounced, is also found in earlier periods of the weak dollar. Real depreciation during 1972-73 and the subsequent reversal, most notable for Korea and the Philippines, are mainly the results of exchange rate realignments among the advanced countries. The steady trend of appreciation between 1978 and 1982 - 57 - except for Malaysia also coincides with the period of the stronger US dollar (1981-82) and relatively high inflation after the second oil price shock. 6.7 This observation makes us question whether exchange rate management in these countries was responsive at all to currency realignments among major trading partners. Thus, before we proceed to evaluate the exchange rate policy further in terms of the REER, it seems appropriate to look into the basic behavior of the nominal exchange rate. From the REER formula, we can rewrite the nominal exchange rate as: (6.2) e - REER * (p/p)*Re where Rp = Ap. and Ile = le. . If the countries maintain the REER constant over time, a change in either the relative price (p/p) or the composite exchange rate of the major trading partners (Re) will translate into the same percentage change in the nominal exchange rate. 6.8 If, however, the REER is not constant but is negatively correlated with the relative price and the composite exchange rate of the major trading partners, i.e., (6.3) REER - a (p/lp)-b. (fe)-d then, the nominal exchange rate is written as (6.4) e = a (p/lp)1-b (le)-d We will estimate this equation in logarithmic form. Coefficients of b or d significantly different from zero will suggest that the exchange rate policy - 58 - in these countries is not guided by the principle of maintaining the REER constant under normal circumstances. The estimation results are presented in Table 6.3. Although one may object to including the period of fixed exchange rate regime in the sample, securing a minimum required sample size dictated the choice. Estimates based on the pooled sample (not presented), however, showed no substantial differences between the whole sample and that of the floating rate regime. Our justification is that, even under the fixed rate system, a developing country could undertake devaluation or revaluation with relative ease if it chose to. 6.9 The estimation results show that exchange rate management of the countries under study responds rather quickly to the difference in inflation rate between home and the major trading countries, while the exchange rate realignment among the trading partners is reflected with a time lag of about a year. One exception may be Thailand where our version of the purchasing power parity seems to hold with about a six-month lag. Concerning the size of the elasticities, all but Malaysia's relative price elasticity are estimated very close to 1.0, indicating that b=d-0 is indeed the case. 6.10 The relative price elasticity ranges from 0.92 (Philippines) to 1.01 (Korea), ignoring Malaysia's 0.30. The elasticity with respect to the composite exchange rate of the currency basket is estimated between 0.75 (Thailand) and 0.90 (Korea). The weak response of the Malaysian exchange rate to the relative price may simply reflect a lack of such interest when their inflation is generally lower than that of the major trading partners. In conclusion, the REER is believed to be an appropriate indicato- of exchange rate policy, which could also be confirmed by the behavior of our sample countries. The seemingly apparent contemporaneous relationship between the - 59 - Table 6.3: NOMIAL EXCHANGE RATE (ln e ) _a Sample a 1964-86 Constant in (,TT) la ( m-) I in TT* in Tre., &2 -1.601 0.95i - 0.413 - 0.965 (0.89) (15.5) (1.24) Korea -4.128 0.741 0.269 - 0.899 0.978 (2.13) (2.73) (1.16) (2.52) -0.402 0.631 - 0.455 - 0.853 (0.83) (3.62) (2.85) -Malaysia -0.639 0.301 - 0.204 0.627 0.914 (1.66) (1.84) (1.43) (3.66) -1.673 0.890 - 0.486 - 0.963 (0.96) (17.6) (1.46) Philippines -3.705 0.918 - - 0.886 0.973 (2.54) (22.6) (3.13) 0.076 0.620 - 0.378 - 0.366 (0.05) (3.22) (1.96) Thailand -3.279 0.491 0.478 - 0.751 0.645 (2.12) (1.85) (2.05) (4.27) la a* - Index of nominal exchange rate per dollar (1980-100.0) - 60 - REER and value of the US dollar mainly comes from a slow exchange rate adjustment to the realignment of the dollar against other major currencies. C. Evaluation of Exchange Rate Management 6.11 Maintaining the REER roughly constant is certainly an important policy guideline in the long run. In the short and medium run, however, the exchange rate is one of the macroeconomic policy tools which can be used to stimulate exports and economic growth, affect the cost of imports and hopefully improve the balance of payments. Thus, exchange rate management may be viewed as the result of policy adjustments in the face of changing macroeconomic conditions. When economic growth performance deteriorates or foreign exchange reserves dwindle substantially, policy makers, giving a higher priority to improving these situations, may rely on exchange rate devaluation to do the job. This policy adjustment will be more likely when inflation is not a serious problem. 6.12 The REER may be estimated with three sets of explanatory variables as follows. A A A A A (6.5) REER - REER (y, p, Res_1; He, He_1, p/Hp ; REER1) where Res is the level of foreign exchange reserves measured in import- months. The first set of variables represents (current or recent past) major macroeconomic performance, upon which depends the probability and magnitude of exchange rate adjustment. We have already observed about a one-year time lag for exchange rate management in these countries to respond to currency realignments among the major trading partners, and that the relative price is - 61 - not fully reflected for some countries. The second set of variables is introduced to reflect this systematic pattern of exchange rate management. Finally, the lagged dependent variable is included since exchange rate management is typically sticky as it tries to avoid excessive fluctuation. Only a partial adjustment is made each year to whatever optimal level. 6.13 The estimation results presented in Table 6.4 show that exchange rate management in these countries has been responsive to some variables of macroeconomic performance. They also confirm the earlier finding that the exchange rates of these countries have not been adjusted quickly to the composite exchange rate of the major trading partners and, for some countries, to the relative price change. According to the equation based on the pooled sample, a 10% appreciation of the composite exchange rate of the currency basket (He) leads to a 8.6% depreciation of the REER in the same year, with a correction of 2.9% made in the next year. However, 83% of the first-year depreciation is carried over due to the stickiness of exchange rate manage- ment, so that 4.1% depreciation still remains. Depreciation of the REER, however, is most likely to be smaller than this, since it will produce self- correcting macroeconomic results such as higher inflation and more foreign exchange reserves. Another observation is that the REER seems to be more sticky for countries with lower inflation. 6.14 Turning to the reaction of exchange rate management to macroeconomic performance, the level of foreign exchange reserves (Res-1) emerges as a significant factor except in Malaysia. When this variable was replaced with the GDP ratio of the current balance or the terms of trade, the results were basically the same, although they were a little less significant. One may interpret the smaller but equally significant coefficient for Thai Res_ Table 6.4: THE REAL BPFECTIVE EXCHANGE RATE (Sample 1964-86) Constant 9 B-1 RTe (TTe)_ (P/TTp) REERl DJr. All countries 22.24 - 0.007 -0.175 -0.911 -0.858 0.294 - 0.826 0.790 (3.92) - (0.04) (2.22) (3.26) (7.05) (1.67) - (14.2) 1.64 (Floating rate 37.08 - 0.028 -0.222 -1.920 -0.825 - - 0.717 0.763 period) (5.41) - (0.15) (2.70) (4.78) (6.79) - - (10.3) 1.28 Korea 91.94 -0.290 -0.135 -0.378 -4.517/a -1.216 0.270 - 0.336 0.819 (5.31) (1.16) (0.56) (2.37) (2.75) (4.89) (0.67) - (2.56) 1.61 Malaysia 5.25 -0.719 -0.889 -0.111/b 0.367 -0.528 - -0.563 1.057 0.945 (0.72) (2.48) (2.84) (0.63f (0.31) (2.93) - (1.93) (12.9) 1.95 Philippines 14.61 - -0.234 -0.247/b -1.964/c -1.353 0.798 - 0.889 0.700 (0.74) - (0.38) (1.43) (1.95) (3.60) (1.22) - (4.29) 1.95 Thailand 16.07 -0.459 - -0.244 -0.607 -0.806 - -0.657 0.932 0.921 (0.99) (1.19) - (1.91) (2.31) (6.88) - (3.06) (5.95) 2.10 /a Korean Res in this case is foreign exchange reserves held by all domestic banks. /b The variable is lagged one period (P ). /c The variable is difference from the previous year ( &Res.,). - 63 - compared with the Philippines and Korea, as the confirmation of our finding that devaluation is most effective for Thailand in improving the current balance. Interesting is the absence of response of exchange rate management to foreign exchange reserves or other indicators of external balance for Malaysia--the only country showing little positive effect of devaluation on the current balance in our quantitative analysis. 6.15 Inflation (0) is a significant factor only for Korea, where not only was the inflation rate higher historically but also the inflationary effect of devaluation was the biggest. With marginal significance, inflation also influences the REER of Thailand where, in spite of relatively low inflation, our model indicates the substantial (next to Korea) inflationary effect of real devaluation. 6.16 Finally, apart from Malaysia, where income growth is found to be the only favorable effect of devaluation, the reaction of exchange rate management to economic growth performance (9) seems to be fairly weak.V This weak response is perhaps not a surprise, since there are other alternative options, like monetary and fiscal policies, for stimulating the economy. Overall, the mode of exchange rate management is seen to have been generally consistent with what our model suggests. The evidence seems to be that the exchange rate was used mainly as a policy tool for external balance where it is found to be effective, while exchange rate policy was utilized as a tool of general demand management in other countries (Malaysia). 5/ There is a possibility of underestimating the responsiveness of exchange rate management to 9 and ), because the causality could run positively from the REER to t and P. - 64 - 6.17 In connection with exchange rate setting behavior, some may be interested to know whether current exchange rates are overvalued or undervalued. Since this paper is not concerned with the complex issue of the optimal exchange rate, we will simply attempt to see whether current (or recent past) rates are over or below rates anticipated by the past behavioral pattern of the countries. If one believes that it is unlikely for the exchange rate of a country to deviate too much from the optimum in the long run, the residual from an estimated REER equation may serve as a proxy indicator for evaluating the optimality of an exchange rate. 6.18 For this purpose, the REER equation used above is slightly modified to exclude the own lagged variable and constrain the coefficient for the composite exchange rate of the major trading partners (Re) to be the same for all four countries. Inclusion of the variable He may be justified if the slow or partial adjustment of the REER to a change in He is associated with the expectation that other competing countries will do the same. The equation is estimated with the pooled data as follows. (6.6) REER (a Opi + a y. + a21j p + a 3. Res._ ).D. + a 4e + a 5e- where Dj = Country dummy variable (1 for country j, 0 otherwise), and al . 0, a 2. 5 0 , a . S 0. After trying both current and one-period lagged values for yj and pj, the one with the higher statistical significance was chosen by country. - 65 - Table 6.5: REGRESSION RESULT FOR THE REER so a1 a2 a3 a4 a5 R2 Korea 131.0 -0.325 -0.531 -5.70 Malaysia 103.1 -0.024 y_1 - -4.32 -1.000 -0.585 0.665 A Philippines 102.6 - -0.142p-1 -0.37 A Res-1 Thailand 116.0 -1.031 -0.430 -1.35 Res-2 The actuaal and fitted values of the REER based on the above equation are shown in Figure 6.1. The residuals are positive in 1985 and 1986 except for the Philippines, suggesting some possibility of currency undervaluation. The magnitude of "undervaluation" in 1986 is 5% for Korea and 8% for Malaysia. By contrast, the Philippine REER is shown to have been substantially "overvalued" (about 9%) during 1984-86. V Another notable feature is the tendency for the Malaysian residuals to grow with time. This might be due to the omission of some important factors in the Malaysian exchange rate equation or an under- estimation of the wholesale price inflation in Malaysia. Still another expla- nation is that the exchange rate has been purposely managed to be steadily depreciated with a view to diversifying exports into manufactured goods. 6/ In light of the unusual magnitude of the dollar weakening in 1986, one may prefer excluding 1986 in the sample of the REER regression, on which the judgment of undervaluation or overvaluation of a currency is based. When this was done, the regression coefficients changed very little, and the predicted values for 1986 REER declined by 2.3-2.7% points for Korea, Malaysia and Thailand and 0.5% point for the Philippines. It indicates that 1986 REER was "undervalued" by 7% for Korea, 10% for Malaysia and 3.5% for Thailand, and "overvalued" by 10% for the Philippines. - 66 - Figure 6.1: THE ACTUAL AND FITTED REAL EFFECTIVE EXCHANGE RATE 13 130 Korea lie lie 100 10 Malaysia 90 nap ,,. -Philippines 1104 so- 90a 130 I=o- Thailand 110* 100 * 90 640BG676 70 71 72 73 74 7 76 77 78 796a60 836 848 88 a Aad + pst a/ The fitted values are from the equation in Table 6.5 estimated with the pooled data. - 67 - VII. SUMMARY AND CONCLUSIONS 7.1 This paper is concerned with assessing the role of exchange rate policy and evaluating exchange rate management in four East Asian countries. Right exchange rate management requires reliable information on the effects of exchange rate policy, particularly in countries undergoing a rapid change in export structure. The econometric model proposed in this paper for quantify- ing the effects of an exchange rate change gives due attention to the changing export structure by disaggregating exports into major primary commodities and other products. In the Philippines, for instance, four major primary commodities took the lion's share (74%) of exports in 1970, but accounted for only 15% in 1986. Under the circumstances, aggregate export behavior in the Philippines around 1970 is not comparable to that in 1986. The approach adopted concentrates on exports other than major primary commodities, since the former are more responsive to an exchange rate change and are expected to dominate exports in the coming years, even in countries depending heavily on several key commodities. 7.2 Our analytical framework has two other characteristics making the quantitative estimates more reliable. First, the model is fully dynamic. Though small, the model incorporates all the essential dynamic interactions among variables, which have a significant bearing on the trajectory of the current balance over time. The results of this model unmistakably demonstrate the error of statically estimating the effects of exchange rate changes using only the price elasticities of exports and imports. Indispensable are the dynamics of the domestic inflation set in motion by an exchange rate change, the impact of domestic inflation on export prices, and the absorption respond- - 68 - ing to income growth. Second, the comparative nature of our analysis guards itself, to a degree, against serious mistakes in estimating parameters. Com- parison of the estimated parameters among the four countries in light of their structural differences makes it easier to detect any misspecification or other econometric problems in the equations. 7.3 The simulation results for 1986-90 show that devaluation would improve the current balance, except in Malaysia. The improvement, in terms of percent of total imports, is largest and most sustainable in Thailand but very short-lived in Korea. In general, however, the improvement is not impressive. A 102 nominal devaluation leads at most to an improvement of 2-3% of total imports even in the years of strongest effect. A 10% real devalua- tion, whicb raises domestic prices by as much as 18-27% by the third year of devaluation, improves the current balance by just 4-7% of total imports in the same year. Devaluation proves to be very inflationary, except in Malaysia. A 101 nominal devaluation results in an almost 7% increase in domestic wholesale prices by the second year for Korea and Thailand, and a sustained real deval- uation requires sharp acceleration of both nominal devaluation and inflation. Devaluation is also very expansionary, particularly for Korea, followed by Malaysia. With fixed nominal money balance, however, the expan- sionary effect is noticeably weaker and, thus, the effect on the current balance is more favorable. This indicates that the effect of devaluation depends significantly on accompanying monetary and fiscal policies. The counterfactual simulation results for 1981-86 suggest that Korea and, to a less degree, Thailand owed much of their export and GNP growth in 1986 to a depreciated currency, although the contribution of depreciation to the current balance was relatively small. - 69 - 7.4 Exchange rate management in these counties was seen to be consistent with maintaining a constant REER in the medium run, though the adjustment to the composite exchange rate of the currency basket typically lags by a year, and the Thai exchange rate adjusts slowly to the relative price change. One exception is probably Malaysia whose currency has not been fully appreciated to reflect its price stability compared with its major trading partners. There is no sign that any of the countries fell victim to the "Dutch disease" syndrome. Tn the short run, exchange rate management also seems to have been efficient in the sense that it was consistent with our findings on the effect of an exchange rate change. The REER was significantly responsive to changes in foreign exchange reserves in all four countries except in Malaysia where devaluation has been least effective in improving the current balance. Exchange rate management also appears to have been interfered with by inflationary trends in Korea and possibly Thailand, where devaluation is most inflationary. Finally, REER management was significantly responsive to recent economic growth only in Malaysia where growth stimulation was found to be the only substantive positive effect of devaluation. Concerning the possibility of currency undervaluation/overvaluation, Malaysian and Korean exchange rates seem to be a little undervalued, while the Philippine peso was overvalued in 1986. 7.5 The comparative framework of our study enables us to anticipate shifts in some key parameters of the model. In general, the price elasticity of export demand will become higher as manufactured goods account for a larger share of exports. At the same time, however, exports of manufactured goods are likely to require more imports of intermediate and capital goods as well as other investment on facilities and infrastructure. It seems difficult to - 70 - judge whether the shifts in parameters caused by the changing export structure will strengthen or weaken the effect of devaluation on the current balance. Apart from the parameter shifts, the changing export structure itself and the growing share of exports in the economies of Korea, the Philippines and Thailand are seen to have strengthened the effect of an exchange rate change. For countries like Korea where the nominal exchange rate has always been depreciating, some caution may be needed in using the estimated para- meters in the situation of the appreciating exchange rate. With domestic prices rigid downward, an appreciation will aggravate the relative profit- ability of exporters more severely, and the ultimate effect will depend on how much of the profit squeeze they can absorb. 7.6 Finally, some policy implications may be derived from our analyses. Korea, together with other newly industrializing countries (NICs) in Asia, is heavily pressured to appreciate its exchange rate as a way of correcting the bilateral trade imbalance with the United States. Our simula- tion results show that exchange rate appreciation, though not ineffective in the second and third year of the appreciation, is a very costly way of correcting the trade imbalance, both for the country and the world, and may not help in the longer run. A surplus reduction by an annual average of $1.5 billion for three years through a 10% real appreciation requires 131 lower real income and 10% lower real imports by the third year. An obvious global implication is that excessive appreciation pressure on most dynamic economies like the Asian NICs may have a serious bias toward recession in the world economy without much success in reducing the trade imbalance. 7.7 Analysis of REER behavior indicates that the Philippine peso is relatively overvalued. More flexible exchange rate management may be desired, - 71 - as inflation has subdued substantially and revitalising the economy is likely to accompany a large increase in imports. For the Malaysian economy, the challenge seems to be in making export demand more price elastic as well as making import demand less income-elastic. Heavy reliance on multinational corporations, which has been an easy way of obtaining the needed technology and markets for manufactured exports, may not facilitate the development of competitive and dynamic export industries in the long run. Despite an indica- tion of currency undervaluation, the poor Malaysian growth performance in recent years makes appreciation unwise. in the coming years, without the undervalued exchange rate, Malaysia will be in a better position to compete and cooperate with the Asian NICs faced with increasing trade protection and pressure for currency appreciation. The key to success in Malaysia is likely to be strengthening the technological base and providing efficient and adequate industrial incentives. References Alexander, S.S. (1952). "Effects of a Devaluation on a Trade Balance." International Monetary Fund Staff Papers, 2:263-78. Bickerdike, C.F. (1920). "The Instability of Foreign Exchange." Economic Journal 30(l):118-22. Johnson, 8.G. (1972). "The Monetary Approach to Balance-of-Payments Theory." Journal of Financial and Quantitative Analysis, 7(2):1555-72. Khan, M.S. (1986). "Exchange Rate Responses to Exogenous Shocks in Developing Countries." Development Research Department, World Bank, Discussion Paper DRD 146. Metzler, L.A. (1948). "The Theory of International Trade." in H.S. Ellis (ed.), A Survey of Contemporary Economics. Philadelphia: Blakiston. Miles, N.A. (1978). Devaluation, the Trade Balance, and the Balance of Payments. New York and Basel; Marcel Dekker. Niehans, J. (1984). International Monetary Economics. Baltimore and London: The Johns Hopkins University Press. Robinson, J. (1937). Essays in the Theory of Employment. Londons Macmillan. World Bank (1987). Trade and Industrial Policies in the Developing Countries of East Asia. Report No. 6952.