Trade adjustment policies and income distribution in three archetype developing economies

Journal of Development Economics 10 (1982) 67-92. North-Holland

Publishing Company

TRADE ADJUSTMENT POLICIES AND INCOME DISTRIBUTION IN THREE AKXETYPE DEVELOPING ECONOMIES* Jaime de MELQ and Sherman ROBINSON The World Bunk, Washington, DC 20433. USA Received November 1980, final version received April 1981

This paper explores quantitatively the macroeconomic and distributional impacts on non-oil semi-industrial developingcountri= of external shocks originating in the world economy -- in particular, rising costs of imports and shrinking export markets. The empirical analysis is done with a computable general equilibrium (CGE) model. The effects of the same external shock are modelled for threedifferentarchetypeeconomies: a primaryexporter, a manufacturing exporter, and a dosed economy. Threedifferentpolicy-adjustment regimes are considered:devaluation, premium rationing of imports (import ficenses), and premium rationing in an environment with a fixed real wage for unskilled labor. By making simple assumptions about the way socioeconomic groups operate to influencedecision-making,the paper also examines how the struggle between the gainers and losers is likely to affect the policy regime to be chosen.

1. Introduction

Shocks emanating from the international economy in the decade of the seventies have had a major impact on both developing and developed countries. The resulting structural adjustments have required, and will continue to require, changes in market conditions which necessitate changes in resource allocation and in product and factor prices to restore equilibrium. In turn, these changes in relative prices alter the distribution of income among socioeconomic groups. In the structuralist environment characterizing many developing countries, there is little doubt that the attempts of different socioeconomic groups to protect their relative position in society greatly affect the choice of government policies. Thus, in evaluating the feasibility of different adjustment policies, one must consider their impact on income distribution. A good deal of insight about the distributional impact of alternative adjustment policies can be obtained from the standard trade model with limited factor mobility when appropriately modified to incorporate both traded and *An extended version of this paper is available in the World Bank Staff Working Paper series (no. 842). De Melo wishes to acknowledge financial supyort from the Agency for International Development and from the World Bank. Our work has greatly benefited from other work undertaken jointly with Kemal Rervis, and we should also like to thank Hollis Chenery, Martha de Melo, Joan Nelson, Larry Westphal, and a referee for comments on an earlier draft. Neither they. AID, nor the World Bank are responsible for any views and interpretations expressed in this paper, which are wfely those of the authors.

03&J--3878/82/

$02.75 Q 1982 North-Holland

68

J. de Melo and S. Robinson,

Trade adjustment

policies

home goods. See, for example, Jones ( 197l), Mussa (1974), and Cordcn and Neary (198()). However, e.\len when this model is reduced to its bare essentials by ignoring the existence of intermediate inputs and assumirlg all demanders have identical tastes, the effect of a change in the relative prices of traded goods on the real income of the mobile factor is ambiguous. Attempts to incorporate realistic extensions rapidly lead to intractable algebra, and it becomes impossible to maintain any unambiguous qualitative results. One must consider the relative magnitudes of various offsetting effects and so one is forced to rely on empirical analysis. In this situation, empirical simulation models can be used to provide insights into the relative importance of different mechanisms at work. In this paper, we start from an empirical simulation model whose theoretical structure is quite close to what has become known as the Australian model.’ The model structure is given by a computable general equilibrium (CGE) model which extends the simple trade model to include intermediate inputs and greater disaggregation across sectors, factors of production and consumers. The model incorporates imperfect substitutability between imported and domestically produced goods, limited factor mobility and the possibility of fixed real wages for certain labor categories.2 The model includes eight sectors, four labor categories and seven socioeconomic groups. This disaggregation represents the simplest we could devise which nonetheless captures essential structural features of semiindustrial countries. Rather thzn basing the study on the historical, political and economic circumstances in a particular country, we have instead created three representative or archetype economies whose empirical structures reflect a range of semi-industrial countries. Our goal is to investigate how a common external shock and trade policy response might have different effects in different environments. We thus subject our three archetype economies to an identical external shock and to identical choices of policy response and then compare the different results. In constructing the three representative economies, we draw heavily on a variety of cross-section work describing the trade, production and demand structures in semi-industrial countries3 To keep the analysis and interpretation of results manageable, we emphasize distinguishing characteristics of our archetype economies and make them quite similar in other, non-essential aspects. The insights from the simulation *,llalysis thus arise from the impact ofdifferences in factor endowments, compr qiti,)n of output, and structure and volume of trade in an otherwise;:identical model.4 *See Salter ( 1959). There are some differences, especially ii? the incorporation of foreign capital inflows and savings. ‘See, for example, Brecher (1974) for a simple model with fixed real wages. %ee Chene r_v and SyrqKin (1975), Lluch et al. (1973, and Kubo and Robinson (1979). 4A complete description of how the data sets were constructed and of the equations of the model is given in de Melo and Robinson (1980b). A related model applied to Colombia is discussed in de Me10 and Robinson (1980a) A simplified sum)nary of the model equations is given in the appendix.

69

The next section describes briefly the external shocks and adjustment mechanisms exam ned in the paper. Se∨ 3 summarizes the characterist:cs or the archetype economies and describes the experiments. Section 4 analyzes the impact of alternative adjustment mechanisms on the distribution of income in each economy and contrasts the effects of each policy response across the three representative economies. Section S discusses how the political struggle among socioeconomic groups is likely to influence policy selection. Conclusions follow in section

2. External

6.

shocks and macroeconomic adjustment

The common external shock simulated below takes the form of an increase in the world price of imports and a decrease in 1he demand for exports (see section 4 below). We also assume that the country cannot borrow abroad to finance the gap between desired expenditures and incclmc and so must adjust fully to the shock.” The first and by no means least. painful adjustment requires that aggregate expenditure be cut to reflect the fall in the terms of trade. Second, a rise in the effective exchange rate for imports will be necessary to remove the excess

demand for foreign exchange. Through changes in relative 1 roduct prices and wages resulting in an increase in the relatilvle price of sectors in which there is scope for import substitution, there will be a movement of resources into these sectors away from the more exportable sectors and away from the pure nontraded sectors. In addition, the relative price shifts necessary to attain macroeconomic equilibrium will also affect the cost of investment goods and hence future productive capacity. How adjustment takes place is inff uenced by a variety of factors. In the first place. it depends on the structure of production and &a@::and on the volume of trade. Secondly, it depends on how the various socioeconomic groups percck that they are affected by the adjustment and the extent to which they manage 10 protect themselves by influencing the seIection of adjustment policy. For example, if labor is sufficiently well-organize:d so that real wages are sticky. then adjustment will rely more heavily on expenditure reduction rather than OH expenditure switching? The sim;llation analysis with the CGE model provides :I quantitative assessment of the role of expenditure-reducing and expendit urcswitching effects in achieving external and internal balance. As in thz Australian model referred to above, it does not capture inflationary mechanisms.’ The ‘For an analysis of borrowing to cushion the adjustment costs IO an external shock, see Martin and Selowsky ( 19X1I. ‘For a model with fixed real wages and markup pricing in the home goods sector where adjustment is entirely by expenditure switching SLL’ F’indlay(1973, ch. 10). ‘This is not to say that the distribution of income, broadly defined. is only affected by 1he slruclurill changes described above. For example. one can postulate disequilibrium adjustment with Sell inflationary spiral that will, of course, affect income distribution. See Rodriguez ( 1978). However. 1I)(, recent external shocks suffered by developing countries are structural rather than monefary and. 111 spite of their monetary consequences, they call for structural adjustments in the economy. It is lhc distributional constquences of these structural adjustrrents that we focus on in this analysis.

emphasis on income and relative price adjustments makes the structural differences built into the three archetype economies all the more important since they largely determine the comparative results. 3. The three archetype wonomies The three economies are all assumed to be ‘semi-industrial’ countries with an income per capita around $500 (1970 US$). Such countries are in a transitional phase with their economic structure closer to that of the industrialized countries than to the very poor low income countries in which the overwhelming bulk of

economic activity is in the primary sector. While they have quite ditI?rent structures of trade, production and employment, we have also assumed them to be similar in a number of important respects. First, they are identical in size with the same total physical outpu; and size of the labor force. Second, they share virtually the same technology. This assumption is common in cross-country comparisons where universal access to a common technology is often assumed. The inputoutput coefficients differ very little across the three countries, the capital structure (I?)matrices are identical and capital/output ratios are the same. We do assume different labor endowments by skill categories which implies that labor/output and hence capital/labor ratios differ? It is a difficult matter requiring compromise to come up with a satisfactory definition of socioeconomic groups which is useful both for economic and political analysis and is also closely related to the variables generated endogenously by a CGE model. The socioeconomic classification described below represents one such compromise. Table 1 summarizes the socioeconomic classification of individuals, the sectors from which their income is derived and whether that income is from wages or profit income. Finally, it provides the assumed within-group logvariance of the distribution of income within each socioeconomic group.g By ‘farmers’ we mean agricultural workers and small landowners or ‘minifundistas’ who own a plot of land that provides enough income to support a family. What distinguishes them from marginal labor, besides their slightly higher income, is that they are tied to the land and earn, in ‘In addition to the same sectora’ clas &cation, uniformity across the archetypeswas also imposed in the selection of parameters descr.bing consumer and producer response to relative price changes. Trade elasticities, elasticities of substitution in production, and income and price elasticities of demand for consumption by socioeconomic groups are the same. Each socioeconomic group makes expenditure decisions according to the Linear Expenditure System (LES). The parameters of the LES vary across socioeconomic groups, but are the same for a given type of group across countries. To simplify further, production functions are assumed to be Cobb-Douglas in capital and labor, with fixed input-output coefficients for intermediate inputs. Finally, the logvariances describing the distinction of income within socioeconomic groups are the same for a given socioeconomic group in all three archetype economies. ‘The distribution of income within each socioeconomic group is given by a tognormal distribution with an exagenously specified logvarance. The rIverall distribution is derived by numerically aggregating the within-group distributions. The technique is described in Adelman and Robinson ( 1978).

Table 1

Socioeconomic classification of individuals. -___TYF

seciots of activity

Farmers Marginal laborers Industrial laborers Service sector laborers Agricultural capitalists industrial capitalists Service sector capitalists

I l-8 2-7 8 1 2 ; 8

Income source __l_l_II Wages + profits” Wages Wages Wages Pr0E t -9 Profits PkOfiIS _I_ - _ I--~-~-_-l_-

“25 percent of agricultural profits go to farmers, the xst to agricultural

Within-group logvariance -x__-0.50 0.20 0.40 0.25 0.60

0.60 0. SO

__.-_________ _

capitalists.

addition to their wage income, a share of agricultural profits. Because there is considerable variation in the size and quality of land owned by smaller farmers, we have assumed this group has a fairly high within-group logvariance. By contrast, marginal labor - the only group which is mobile across all sectors - is fairly homogenous. With an income a shade above half of the economy-wide mean income, they are the largest and poorest group in society and their income is fairly equally distributed. Industrial laborers, or organized labor, are employed in the manufacturing sector. The smallest wage-earning group in society, they are fairly disparate in composition and include white collar workers, clerical workers, technicians and engineers. Their income is about 1.5 times that of marginal labor. Service sector laborers include government workers and, finally, we have the three capitalist groups MO get their income from profits. Their number is fixed exogenously with the income of the indust:-ial capitalists about ten times the economy-wide mean income and about 30 percent above that of agricultural capitalists. Service sector capitalists include a large group of self-employed (such as retai! traders, etc.) and have an income about five times the mean income. These groups whose incomes are determined by the model represent the minimrv* necessary to capture the diversity of socioeconomic groups found in _7.1a semi-industrial countries. Clearly there is coasiderable range of variation in rneaq income and, at this level of aggregation, a considerable degree of heterogeneity within socioeconomic groups. The countries are characterized by a fairly high degree of inequality, with Gini coefficients of approximately 0.50 (or ovcr~~ll logvarizlncc of 0.60). About 45 percent of the overall logvariance is due to between-group variation and corresponds to that part of total inequality whicfl is endogenous to the model. We now come to the distinguishing characteristics built into the production

side of the three archetype economies. The distinctions introduced into the otherwise similar structures fall under three categories. First, the economies are distinguished by their volume of trade. Two of the economies are open in the sense that the volume of exports as ;2 proportion of GDP is large (15 percent).

72

J. de Melo and S. Robinson, Trade adjustment policies

Second, they are distinguished by their structures of production and trade. Of the open economies, one is a primary exporter and the other is a manufacturing exporter. Accordingly, the structure of production is biased toward the primary sector in the first and towards manufacturing in the second. The third distinguishing characteristic, which provides the basis for the structural differences described above, is a variation in factor endowments, tariffs and subsidies that are assumed to account for the differences in production and trade structures between the three economies. There are thus three archetype economies: primary exporter (FE), manufacturing exporter (ME) and closed economy (CL)?’ They differ in: sectoral structure of production, employment and capital; volume of trade; sectoral composition ofexports and imports; skill composition of the labor force; and past trade policies reflected in different sets of tariffs and subsidies. These differences are summarized in tables 2 to 4. Table 2 compares the Ftructure of production and trade in the Base Run (prior to any shock) across the three archetypes. All of the variation in structure comes from the relative sizes of the industrial and primary sectors, and from variations within the manufacturing sector itself. The PE economy, with a larger primary sector (and hence a smaller manufacturing sector), has a distinctively different gross output structure. The differences between the CL and ME structures are less pronounced, although the ME economy has a much larger consumer goods sector which provides its main source of foreign exchange while the CL economy has larger primary and food sectors to assure self-sufficiency. The overall export and import ratios (see also table 3) indicate the sharp difference in volume of trade between the closed and open economies. Note that with world prices exogenous (and equal to unity), the foreign capital inflow is the same for all three economies and is maintained at its Base Run level throughout the experiments. The variation in exports and imports is most obvious at the 3sector level of aggregation. The PE and ME economies have opposite structures and the CL economy resembles the PE economv on the importing side, while on the export side it has a balanced structure with half of its exports originatin the primary sector. Within the manufacturing sector, note that the CL economy is almost as dependent on imports of intermediates and capital goods as the other economies.’ ’ Table 3 provides inform,. !iolb on the sectoral trade orientation of the three two

“Note that the import vector describing the CL economy corresponds more closely to that found in countries like Mexico and Turkey than to the itnport structure of Brazil or India. “Accompanying, and responsible for, this foreign trade structure are the following trade incentives: tariffs of 25 percent, 35 percent, 20 percent for consumer goods, inte,.mediates and capital goods for ME and PE; 15 percent across the boir,rdc;p?rt subsidies for PE and 75 pe;rwct export subsidies for manufacturing in ME; no export subsidies for CL and the gbilowing tari& cawumec goods (75 percent), intermediates (7s percent) and capital goods (20 percent). All three economies have the following trade substitution elast cities: primary (3.0), food (l.S), consumer goods (1.25j, intermc:diates(0.75),capital goods (0.25~

73

-

--

13.8

5.7 15.1 19.2 49.3 98.8 -

PE

15.8

f 3.2

14.8 9.5 49.2 69.2 --.

9.7 6.1 12.0 17.1 28.0 24.8 25.1 17.0

CL

20.7

13.7 9.9 19.2 25.1 38.5 30.9 31.6 20.5

PE

Imported inputs/totaf intermediate inputs

11.0

ME

production -exports.

6.7

2.3 2.0 3.0 29.8 39.8 -

CL

Imports/ domestic supply”

aDomestic supply =domestic

Average

!h-ViLX%

Consumer intermediates Capital goods Construction Social overhead

FOOd

Primary

Sector

-

19.5

14.7 12.2 15.7 23.5 32.2 27.3 27.4 17.8

ME

4.2

9.8 4.0 8.1 5.0 2.0

CL

10.2

30.3 4.0 10.2 5.1 2.0

PE

Exports/ domestic production

Table 3 Three archetype econormes: Sector&i trade ratios in the base run (percent).

9.7

4.9 5.0 35.7 20.2 15.4

ME

-

75

Table 4 Three archetype economies: Composition of the labor force (percent). -Archetype Rkrarcn

Total

CL

I%

ME

30 30 10 30

40 27 9 24

25 33 15 27

.I00

100

loo

--

archetype economies. It is crucial for understanding how the common external ts transmitted across sectors and across archetypes. The most tradable sectors in the sense that their domestic prices are closely tied to the exogenous world prices are the sectors which have high ratios of imports to domestic supply and/or high ratios of exports to domestic production. Note also that although there are three pure non-traded sectors in the economy, these are among the most import dependent sectors due to the high import content of total intermediate inputs. The final distinction between the three archetypes relates to labor endowments which are given in table 4. All three economies have a total labor Corce of 10 million. As one would expect on the basis of the factor endowmerlt theory of international trade, the pattern of trade is closely related to each archetype’s factor endowments. Thus the PE economy has the largest supply of farmers while the ME economy has the greatest supply of marginal labor and especially of organized labor which are intensively used in the manufacturing sector. As is the case with most other indicators, the CL economy, which has the most balanced trade and production strkture, has an employment pattern between those of the two open economies. The macroeconomic impact of sn external shock and its effect on the distribution of incame -. The external shock we impose on each of the archetype economies has two components. On the import side, the exogenous world prices of imports are raised by 25 percent across the board. On the export side, physical exports are lowered by 25 percent in all sectors and are essentially fixed to their new values.’ 2 World prikof exports, however, remain fixed. This experiment is intended to capture theessenceofthecventsthat occurredin thesecond halfoftheseventies. First, the increase in the purchase price of developing countries imports spurred by the rise in energy prices and compounded by the inflationary situation that developed in ‘“In the model, it is convenient to cut exports by lowering all the sectoral export supply ratios (see table 3). This, however, does permit a slight export supply response through changes in the sectoral composition and level of output. In the experiments, this effect is very small.

76

J. de Mefo ml S. Robinson. 7hufe udjustvvmt policies

the industrialized countries; second, the world recession that led to restrictionist policies in industrialized countries which in turn resulted in both a sharp decline in their demand for developing countries’exports along with the imposition of quantitative restrictions. We have deliberately chosen to ignore the induced terms-of-trade losses that occurred to The extent that developing countries fact downward sloping demancl curves for tl leir exports. Rather, we have emphasized the effects associated with export c.luantity restrictions coupled with an exogenous change in the foreign terms of trade.’ 3 TQ study the interactions between acljustment policies and the distribution of income, we compare the following rhree adjustment policies: devaluation, premium rationing and premium raticlning with a fixed real wage for marginal labor. There are important political as well as economic differences between the three adjustment mechanisms. A policy of devaluation, because of its fairly even spread across sectors, could be viewed as a political compromise reflecting some minimal degree of consensus among socioeconomic groups. A policy of granting import licenses giving rise to premia reflects a situation where political power is concentrated in the hands of the capitalists and would accentuate income concentration. Finally, a policy of premia on imports coupled with fixed real wages for marginal labor reflects a situation where no socioeconomic group is firmly in cor!rol of policymaking. Businessmen manage to appropriate the premia associated with a system of import control by licensing but, on the other hand, workers are sufficiently well organized to prevent a fall in real wages. The macroeconomic effects of alternative adjustments to the external shock arc reported in table 5 which contrasts the magnitudes of changes in the majot components of GDP at constant prices across the three countries for each of the alternative adjustment mechanisms. Consider first the case of devaluation. At constant prices, the fall in real GDP ranges between 8.7 percent for CL to 2.3 percent for PE. I4 The ME economy is harder hit than the PE economy because the latter has a production structure biased toward the primary sector which. next to food, i$ the least import dependent sector in the economy. As expected, under devaluation, private consumption falls the least in the PE economy since agricultural and food exports are now supplied to the domestic market, dampening the cut in aggregate consumption expenditure. With a 73 pl:rcent decline, the fall in pr.%; consumption expenditures is highest in the ME ei:oIlomy which has to bear the brunt of higher prices for food imports. ’ 31nselecting the external shock, we have tried to faciiitate the comparison acr&s archetypes in a number of ways. First, combining equal across-the-board rises in import prices with equal across-theboard reductions in export ratios yields an undifferentiated external shock for all three economics so that the impact effect depends only on the total volume oftrude in the economy. Second, by fixing the world price of exports, the decline in the foreign terms of trade is the same for all three economies so that the resulting income loss depends on the initial volume of trade and not on its composition. ‘“At first sight this result seems counterintuitive since the economy which trades the least is the most severely afficted. This is because the CL econk>myhas the highest ratio of intermedinte input requirements and, moreover. its imports are concentrated in the intermediate sector which has a low tmde substitution elasticity. This is also the reason why devaluation is highest in , ‘meCL economy.

J. de Melo and S. Robinson. Trade adjustment policitis

77

3 c

t-1

X

=t =

c

CJ

X

g

CT

X I

r-1

G

KJ I

-

1

Gcvernment consumption rises by approximately 10 percent across a11three eccnomies, largely because government domestic currency revenues from tariffs increase both because they are levied on higher foreign prices and because the currency has devalued. While, under devaluation, the differences in adjustments between countries are not great for exports and imports, they are quite significant for inves model, fixed savings rales are assumed for the government and for each socioeconomic group which implies that the model is ‘savings driven’. Foreign capital inflows are assumed to be part of savings. TO understand why real investment falls by mlzre than 16 percent in PE but only by 7 percent in ME we need more information about the composition of investment and the sources of savings. First, note tha,, one expects to find variations in relative prices and hence differences in the prices of capital goods. As can be seen from looking at the investment goods price deflator at the bottom of table 5, the same nominal investment would, after the shock, yield different real investment.’ 5 There is also a second efVectdue to the change in the con &positionof savings in current domestic prices. The trade deficit, measured in world prices, is identical and remains fixed across the three economies and also is assumed to be part of savings. But this saving takes place in domestic prices and hence varies with devaluation. The foreign component of domestic savings - 22 percent in the base run - rises dramatically across all three economies with devaluation. And of course it rises most rn CL where devaluation is greatest. Thus real investment falls less in CL than in PE, even though both economies experience essentially the same rise in capita! goods prices.’ 6 Whether and tc what extent foreign capital inflows are channelied into investment is an issue which has often been debated in the literature. After having been much discussed in the context of two-gap models, the issue has resurfaced in the debate on the role of alternative closure rules on the distribution of income. Likewise, how investment is determined may have a significant impact on the distribution of income. For example, had real investment been fixed, with increases in the prices of capital goods from 6 to 19 percent, there would have been significant forced savings which might have worsened the distribution of income, depending on the nature of the forced-saving mechanism assumed.” ’ Since exports are essentially Exed, the main difference between adjustment by devaluation and adjustmemt ?y. premium rationing is the income transfer to capitalists. Thus, many of the differences in the pattern of adjustment between the three economies remains when adjustment takes place by premium rationing. ’ “The investment goods price deflator varies widely across the three economnes. changing least for the ME economy because of the increased availabihty of capital goods on the domestic market. ’ ‘This effect, often referred to as the Hirschman (1’249)effect,has been noted and investigated in the context of r?eveloping countries by Cooper (19’71).The foreign savings component of total savings rises by 59 percent in CL, by 41 perce.lt in PE and 36 percent in ME. ’ -See Ta;Aor (1980).

Premium payments as a percent of GDP range from 6.6 percent for ME to 9.2 percent for CL. The potential rents accruing to the private sector froril a system ol‘ licenses can therefore be enormous and it is no surprise that it is important what method of adjustment is used. Although more sensitivity analysis would be needed before one could establish a correspondence between devaluation and associated premia, these results are consistent with the often-made observation that huge rents are commonplace in economies which choose to maintain external balance with a system of import controls? Finally, we come to adjustment by premia combined with fixed real wages. The fall in GDP due to the contractionary impact of the external shock is exacerbated by induced unemployment ranging from 1.6 percent for PE to 7.1 percent for ME. As a result, GDP falls substantially more, except in the primary exporter.‘g In addition to variations in real investment, the different policy adjustments have important variations in incidence among socioeconomic groups. Which groups bear the brunt of adjustment across the three economies and is it the Same groups that lose the most under both devaluation and premium rationing? Can any group benefit in absolute terms and experience a rise in real income? Before examining the magnitudes of distributional shifts, it is useful to review the various mechanisms whereby the external shock affects the distribution of earned income. First, on the supply side the change in relative net prices affects the distribution of value added petween sectors and, within sectors, the distribution of value added between wages and rents. Tiese shifts determine, after taxes, the distribution of earned nominal income among the socioeconomic groups. Second, the distribution of income is affected by differential changes in the cost of‘ living across groups. The relevant prices are those of composite goods (which include imports). The evolution of the net price agricultural terms of trade (see table 5) varies across experiments and adjustment mechanisms; so does the composite price agricultural terms of trade which follows a similar course as the net price terms of trade and is important in determining real income. Not surprisingly, the agricultural terms of trade increases the most in the ME economy which must replace agricultural products previously imported.20 Third, the distribution of income is affected by transfers which here arise from the distribution of import licenses and hence premia. ‘Wote that the premium rate is higher than the devaluation rate even though the cornposit on and volume of imports and exports are similar. This is entirely because of the price normaliz?rion rule which maintains a weighted sum of composite prices constant. Thus the domestic currency price of exports does not rise in the case of premia rationing whereas it does in the case of devaluation. It follows that to maintain the same price level, the domestic prices of imports do not rise as much. ‘“If we had instead made the short-run assumption that nominal rather than real wages remain fixed, the impact ofadjustment on the distribution of income would have been quite different. Indeed, uith real wages falling, the demand for labor would have increased. Such an assumption is not justified in a medium-turn model and; in any case, would require an analysis of inflationary mechanisms which is beyond the scope of the present analysis. “‘There is also a large spread in the shifts in cost of living indices between socioeconomic groups. On average, the spread is over 10 percentage points.

80

.I. de Melo and S. Robinson, Trade adjustmmt policies

Table 6 provides the group mean real incomes in the base run along with the percent changes from the values in the base run. It thus indicates which groups gain or lose in absolute terms. In turn, table 7 provides information on the relative distribution of incomes by giving changes in group shares in total income. It is noteworthy that some groups gain in absolute terms after the external shock in spite of the sizable fall in the economy-wide mean income (see last row). Some striking patterns emerge; for the manufacturing exporter, regardless of the adjustment mechanism, farmers and agricultural capitalists always gain in absolute terms relative to the Base Run. In the other two economies, no group gains significantly when adjustment is by devaluation. However, when t;iere is adjustment by premium rationing, industrial capitalists gain absolutely along with service sector capitalists and, to a lesser extent, service sector labor. What emerges from these figures is that starting from a reasonable spread in mean incomes among groups in the base run, the distribution of premia, amounting to close to 10 percent of GDP, raises the mean income of some recipient groups by up to 25 percent. No wonder that the choice of policy instruments is such a politically sensitive topic. Consider, for example, service sector capitalists. They are the largest and least well-off of the capitalist groups. With adjustment by devaluation, even their relative position in society is likely to deteriorate as a result of the increase in the relative price of traded goods. However, when they receive the premia associated with the licenses they are granted for importing consumer goods, their mean income rises by almost 10 perceklt in the PE economy and by 7 percent in the CL economy. But in the ME economy, where import substitution in the primary sector is assumed achievable in the medium run, the farmers and agricultural capitalists appropriate these rents through higher domestic agricultural prices. In that economy, service sector capitalists gain little from a system of import licenses since imports of consumer goods become effectively replaced by domestic substitutes. Finally, one can take a more aggregate view of the distribution of income along broad functional lines and lump all capitalists together in one group. Table 3 indicates that, taken as a group, capitalists always improve their relative position when they get premia, but in the case of devaluation they only improve their position in the SM economy. ?t would thus appear that devaluation is not H politically easy adjustment IX&C~to implement. The relative position of marginal labor, the largest labor group in each of the economies, declines in virtually all instances. As is typical in developing countries, 1Neassume that the enlployed support the unemployed by direct income ttransfers.2l Struggling to maintain real wages leads the marginal labor group as a whole (including tht: unemployed) to mitigate their relative income loss (4.4-2.2 “In this model, this is done through transfers so that the unemployed receive half the mean income of the employed.

J. de M&o

and S. Robinson,

Trudv adjtattnent

poiicirs

81

100.0 0.51 0.59

Total Gini 7’heil index

“In percentage points.

9.4 21.8 13.7 12.1 9.9 35.7

-

17.0 16.1

Farmers Marginal labor Unemployed Org. labor Serv. labor .fIg. capital Ind. cap. Serv. cap. AI1

CL

8.9 16.5 17.8 11.5 7.6 36.9 100.0 0.52 0.62

-

21.8 15.9

PE

0.0 -0.01 - 0.02

loo.0 0.52 0.63

13.2 17.5 11.7 17.7 8.0 37.4

0.1 -0.2 --. 0.0 1.3 -0.9 -0.5 0.2 - 1.2

CL

14.4 17.5 --

ME PE

0.0 -0.01 - 0.03

1.3 0.4 0.2 -0.4 -0.6 - 0.2 -0.7 -1.5

Devaluation

0.0 0.02 0.07 ---

- G -2.7 4.3 - 1.1 -0.6 2.6

3.1 -1.7

ME

Change from base run” -

Table 7 Group shares in totai income.

0.0 0.02 0.07

- 1.8 - 1.1 -0.8 0.9 - 1.7 3.4 1.1 2.8

CL

0.0 0.01 0.04

1.4 -3.2 3.4 1.1 1.3

Go

-2.6 -0.1

PE

Premium rationing

13.0 0.02 0.07

0.0 0.01 0.04

3.6 1.0 1.2

3.7 1.0 2.6 0.0 0.03 0.12

-2.8 -0.5 0.6 02 1.3 -3.4 -2.1 -1.9 1.1 -0.5 0.8 -2.1

2.2 -2.6 - 2.8 - 1.8 3.7 0.8 0.5 5.0

PC:

CL ME

*YE

0.0 003 0.1 I

1.2 0.4 4.7

1.9 -4.4 2.2 -25 - 1.9 3.1

Premium rationing with fixed real wage

-.

& s=’ -z 3 g -z 3cs !$+

9 Y %

z Ts 3 h h % $ 2

= 2.2 percent loss) compared to premium rationing (2.C percent loss). With or without fixed real wages, the alternatives facing the poorest are gloomy indeed.22 5. Chss conflict and policy choice The picture that emerges from comparing the impact of alternative adjustment mechanisms on the distribution of income in different economic environments is complex. There are certainly significant changes in the relative distribution among socioeconomic groups, with some groups gaining in absolute tetms in spite of an overall decline in real income due to the external shock. Differences in economic environment alone suffice to make different groups gain or lose. even with the same adjustment policy. Different adjustment policies also have quite different effects on the distribution. The fact that the choice of policies has a significant impact on the distribution of income leads one to wonder how the struggle between the gainers and losers decides the final choice of policy. Assuming that one can analyze how policies are selected on the basis of conflicting group interests, is it likely that the same policy will be selected in different economic environments? To explore these questions, we make a number of drastic simplifying assumptions. First, assume that each group realizes how its income would be affected under each policy regime. Second, assume that socioeconomic groups are distinct and care only about their own interests. We thus assume that we have partitioned the society into the important or politically relevant power groups. each pursuing its own interests with perfect knowledge. Third, assume that the intensity with which any group cares about a given policy change is measured by the relative difference in its income between the two situations. Define I:(P) as the mean income of group i under policy regime P. In our experiments, P can be the following: B (Base Run), D (devaluation), P (premium rationing), and P, I%’ (premium rationing with fixed real wage). For example, the intensity with which group i is for ( +) or against ( -) devaluation compared to premium rationing as a function of its relative change in income is defined by (1) Next consider the problem of determining the weight of each socioeconomic group in influencing policy choice. There is (certainly little agreement about how to derive such weights, so we have specified three different schemes which are consistent with different views about the sources of power or influence and which ‘“En these experiments, aggregate measures of income inequality such as the Gini coefficient hide more than they reveal. Even when the share of marginal labor falls over 4 percentage points relative to capitalists, the Gini coeffkient never changes by more than 3 percentage points. More revealing from a distributional standpoint is an analysis of the composition of poverty. The results show that a large fraction of society can be adversely affected, even when aggregate measures of the relative distribution show little variation. See de Melo and Robinson (1980b).

J. de Melo and S. Robinson, 7hde adjustment policies

84

should

reasonable range of values. It should be stressed that the analysis is not concerned with more fundamental and difficult issues relating to the theory of how the political process works. Rather, given a political system, we are trying to capture in a stylized manner some of the elements which are likely to determine the course of policy selection by those who are already in power. The mechanisms by which political power is exercised or altered are not considered. The first scheme assumes that relative influence is given by shares in total population - ‘one person, one vote’. The second scheme assumes that influence is measured by relative shares in total income - ‘one rupee, one vote’. Finally, the third scheme assumes that the political system is dominated by the economic elite and hence that the influence of different groups is measured by their share in the elite. In this third case, we assume that the elite is defined as the top 5 percent of the overall personal income distribution and that influence in decision-making is measured by the socioeconomic composition of this group. The first scheme is rather naive but appealing to utilitarian welfare economists, if not to cynical political scientists. The second scheme is a kind of ‘effective demand’ theory of the distribution of influence with groups able to buy influence in proportion to their shares in total income. Large groups are still important. even if they have a relatively lcw mean income. The third scheme is really a simple oligarchy that might charactc. :e a Daive Marxist view of how a capitalist society works. This last measure, though somewhat extreme, is intended to provide a first approximation of how policy decisions are made in countries where power is concentrated in the hands of the capitalists class.2” With the ‘political power’ shares and the information on changes in the mean incomes of the socioeconomic groups, we can undertake an analysis of the selection of different policy regimes. Eq. (1) defined Ri(Pi ; P2) which measures the relative gains or loss for group i between policy regimes P, and P2. Expressions for the average gain of the gainers and the average loss of the losers, weighting by influence shares, are given by provide

a

G(Pi; P,)=C KXi(Pi; P2) C I+( for all Ri>O, i

i

i

L(F,; &I) = - 1 lvRir Pr; P2) ‘T, I4( for all Ri
I

i

where P,, P2 are policy regimes, M/iis the relative influence of group i, and Ri is the relative gain of group i from eq. (1). 23The population shares are given in table 4 except for the capitalist groups whose shares are very small (from 1.1 to 2.2 percent). Income shares are in table 6. For elite shares, the capitalists are the most important (from 20.1 to 39.5 percent). The choice of the top 5 percent is completely arbitrary. In general. the srnaller the elite, the more relative power is ascribed to capitalists. See de Melo and Robinson (1980b) for a more detailed disrussion.

J. de Meio

and S. Robinson.

Trade adjustment

policies

x5

The sverage gain and loss functions reflect the intensity with which differeni groups :upport or reject one policy compared to another. The politics1 feasibility of one policy compared to another depends, in addition, on the relative influence of the gainers compared to the losers. Given that we used relative influence weights i;n defining the average gains and losses, G and L, the political feasibility of a particular policy choice depends on whether the weighted difference between G and L,is positive or ne ative. We can thus define an index of political feasibility, F, which is equal to this weighted difference. It is also just equal to the weighted average of the relative gains, R,, over ail groups, gainers and losers:

F(P,; P,)==

i

bf(Ri(P 1: P,).

(3)

Note that since we have three different weighting schemes. there are three different feasibility indices for pairwise comparison of policy regimes. Table 8 presents the average gains and losses of the gainers and losers, G’and L, as well as the index of feasibility, F, for all pairwise compar sons of the three different policy regimes. In each pairwise comparison of policies, the first column gives the relative influence of the gainers. WC have already discussed which groups gain and lose and it is important to remember here that the group composition of the gainers and losers does vary across archetype economies and policy regimes. However, withrn each archetype and pairwise comparison, the differences in G, I, and F are due only to the different weighting schemes used. Consider. for example, the comparison of devaluation with premium rationing. As the weighting scheme goes from population shares PO elite shares -democracy to oligr*chy -- the relative power of the gainers declines substantially. In the comparison of premium rationing with and without a fixed real wage (the last four columns), the effect is reversed, with the weight of the gainers increasing with the more oligarchic weights. In the first comparison, the gainers tend to be the small farmers and laborers whose population share is high but whose representation in the top 5 percent is small. In the last comparison, the gainers are everybody but marginal and organized labor, so their weight is high under all three schemes, but especially so for elite shares. The average gains and losses, C and L, measure the intensity with which the gainers and losers care about the outcome. For the first two policy comparisons, with a few exceptions (six out of eighteen, given the different INcighting schemes and archetypes), the feasibility index indicates that those who care the most will determine the policy choice. In the third pairwisc corjparison, between premium rationing with and without a fixed real wage, the opposite is the case in all three archetypes for the first two weighting SC emes. Even though the losers lose more than the gainers gem, their weight is so at the feasibility index is

56.8

43.4

Income shares

ME

ME Elite shares ‘Set text for definition of G, L and F.

64.4 55.9 71.1

74.7

Population shares

income shares Elite shares Population shares

68.5 56.1 46.8

PE PE ME

Population shares Income shares Elite shares

CL CL CL

5.3

5.3

10.2 11.8 5.2

10.0

4t 3.8 15.6

12.1

9.4

20.0 26.0 6.3

12.6

4.6 14.2 14.0

1.4

-4.5

- I.1

-0.6 -5.1 1.9

4.2

-4.1 -0.2

F

Feasibihty index

43.2

39.3

48.6 55.6 39.6

49.0

39.9 40.0 46.6

11.8

8.9

5.4

11.0

_

13.1 24.3 2.9

5.8

2.9 10.0 18.8

L

Average loss

14.1 13.3 10.4

15.2

8.1 7.1 6.3

*.verage gain G

0.0

- 1.0

- 0.1 3.4 2.3

4.6

1.4 - 3.0 - 7.2

F

Feasibrhty index

Political weight of gainers

Average loss L

Political Average weight of gain ga lers G

PE

Wfzightmg scheme

Archetype ecoI.dmy

Devaluation vs yt=mia with Exed wage -

Devaluation vs. premium rationing -.-

Table 8 Policy compansorxa

99.8

82.5

73.2 90.6 68.5

65.7

6: 9 74.6 90.9

Political weight of gainers

4.6

4.2

1.6 1.7 4.2

1.8

2.9 2.7 2.8

Average gain G

7.6

7.6

2.4 0.5 7.6

2.7

4.5 4.0 1.3

L

Average i0SS

Premia vb. premia with tixed W.-J-p-

4.6

2.2

0.6 1.5 0.5

0.3

0.1 1.0 2.5

F

Feasibility index

s -z. 2

:

& E. 2 f 3

&

3

*g

% -. 2

B

s & E/,

J. de Melo

and S. Rothsou.

Trade udjustmerrt

policies

87

Table 9 Policy rankings. Archetype economy - --CL CL

Policy ranking

CL

Population shares Income shares Elite shares

PE PE PE

Population shares Income shares Elite shares

ME Population shares ME Income shares ME Elite shares -VI): devaluation. (P): premium rationing, (P. Cvi):premium >: preferred to, and = : indifferent to.

(L)>V)>(P.

W)

(P)>(P* W)>ID)

(WV,

W(W

(WW>(R W (Pj>(D)>(P, W) (W(R m>(D) (D)>tP)>(P, IV) (Y)>IP, W)>(D) (P)>rP, W)=(D) rationing

with fixed real wage,

Table 9 summarizes the ranking of different policies based on the pairwise comparisons in table 8. It is interesting to note that the various pairwise comparisons never violate transitivity. That is, if policy P, is preferred to P2 and P2 is preferred to P,, it is always true that the pairwise comparison shows that P, is preferred to P3, as would be predicted by assuming transitivity. Devaluation, which skews the distribution of income the least, always prevails in a more democratic political system. Noting that historically more democratic political processes tend to emerge in the more developed countries, one might speculate that the Kuznets U hypothesis (that the distribution of income first becomes worse and then improves as the process of development unfolds) is at least partly related to changing policy choices as political power shifts over time. There is remarkable agreement in policy rankings, given the choice of weights, across the three archetype economies. This is true in spite of the fact that the particular groups which gain or lose, and the magnitudes of their gains and losses, vary widely. Although our particular choices of policy regimes to compare were purposely made to enhance their differences, it is tempting to conclude that there are geueralizations about the feasibility of implementing different policies that are valid for a wide variety of economic environments. The universal dominance of devaluation under a democratic weighting scheme and of premium rationing alnder the other schemes is certainly consistent with casual observation of a wide variety of countries, although our three simple weighting schemes do not capture adequately the wide variety of political systems represented ill the world. 6. conclwions

By constructing three archetype economies that cover a broad range of initial conditions - resource endowments, past trade policies, and economic structure - this paper has examined the quantitative importance of different initial

88

J. de Melo and S. Robinson,

Trade adjustnwlt

poli~*ies

conditions in adjusting to an identical external shock. While all three economtes suffered, for the particular external shock we simulate (a decline in export markets and an adverse movement in the international terms of trade), the closed, rather than the open, economy is the most severely affected. This result is due to the relative importance of imported intermediate inputs in the closed economy. These inputs are crucial and cannot be easily replaced by domestically produced intermediates, so the economy has to bear the full brunt of their rising costs. The composition of imports determines to a large extent whether the burden of adjustment falls on consumption or investment. For the manufacturing exporter (a food importer), the consumption loss tends to be high. On the other hand, for the closed economy and for the primary exporter (which both import capital goods), the decline in GDP falls most ileavily on investment. The rate of real investment, which is around 20 percent of GDP prior to the shock, falls to 17 percent in the closed and primary exporting economies. How the distribution of income among socioeconomic groups is affected by the external shocks depends both on initial conditions and on the choice of adjustment policy. In general, the distributional shifts are most pronounced for the manufacturing exporter where farmers and agricultural capitalists gain regardless of the selected policy while all categories of labor lose. In the other economies the response is more varied with industrial capitalists losing under devaluation but gaining under a policy of premium rationing of imports. Finally. when real wages are fixed, the contractionary effect of the external shock is more pronounced for the manufacturing exporter, which experiences a 10 percent fall in mean income along with an unemployment rate of 7.1 percent. The picture that emerges from comparing the impact of alternative adjustment mechanisms on the distribution of income in different economic environments is complex. In particular, the distribution of premia, amounting to close to 10 percent of GDP, raises the mean income of some recipient groups by up to 25 percent, in spite of a sizable decline in overall mean income. With so much at stake, it is no surprise that the choice of policy instrument is such a politically sensitive topic. The fact that the choiLe of policies has a significant impact on the distribution of income leads to a set of political questions that go beyond the usual economic analysis of policy choice. By making quite strong assumptions about the way socioeconomic groups operate in the political process, one can determine the policy most hkely to be chosen. The analysis proceeds with three different measures of relative influence of socioeconomic groups, reflecting decision making under political regimes ranging from democratic to oligarchic. The simple assumptions we have made certainly do not do justice to the relevant theories of political and economic conflict, yet they roughly capture how vested interests are likely to influence policy making. The model-based results indicate that the likely outcome is a selection of adjustment policies resulting in a deterioration of the distribut:on of income.

Appendix: The eqoations of the model of the flexible exchange rate produced goods are assumed to be imperfect substitutes in use and all demands are for a composite good defined for eacil sector. Since all demanders are assumed to minimize costs, the price of this composite good in eq. (3) is given by the cost function dual to the CES trade aggregation function. Exports in each sector are determined by fixed shares of sectoral supply - eq. (10). In table A.1, the income and flow-of-funds equations are not explicitl:y written out. The average incomes of all socioeconomic groups _ different types of wage earners, farmers and recipients of non-wage income __ are determined endogenously by mapping from factor incomes. The distribution of income within each group is given by a two-parameter lognormal distribution function. The logvariance is specified exogenously and the logmean is a function of the logvariance and the endogenous group mean income. All aggregate distributional statistics are derived numerically by summing the within-group distributions. Note that cost-of-living indices are computed for each group separately so that real incomes for socioeconomic groups are defined using appropriate separate deflators. Aggregate savings and investment is determined by applying fixed savings rates to the income of the various institutions in the economy: enterprises, households and government. Foreign capita1 inflow is also assumed to be saved. The allocation of total investment to sectors is given by exogenous shares - eq. ( 18) - --and the demand for invc stment goods by sector of destination is calculated by assuming a fixed composition of aggregate capital which differs across sectors -- eq. (19). In the flexible exchange rate version of the model, the exchange rate (ER) is determined endogenously to clear the market for foreign exchange, satisfying eq. (12). In the p.cemium rationing version, the exchange rate is fixed and a uniform premium is added to sectoral tariffs in eq. (l), thus affecting the domestic price of imports PM). The premium rate is solved endogenously to satisfy eq. (12). Except for the use of adjustment by premium rationing with fixed real wages, the labor market is specified so that all wages are solved endogenously to clear the labor markets. In the version with a fixed wage, the real wage for unskilled labor is set exogenously and the excess demand functon for the corresponding labor category in eq. (9) is dropped. For further discussion of this type of model, see Dervis, de Melo and Robinson ( 1982). Table A.1 provides a summary

of the equations

version of the mcdel. Note that imports and domestically

(I) Pricq

(I j PfMi = PTJ 1 + tmi)ER (2) PEi=PWEi(l + teJER (?I Pi =f(PIlD,, PM,) (4) f’Ni= PDi-LjPiaii-tJiPDi _ (5) CQPi=P ER : exchange rate, tmi : tarX rate, PT : world price of imports, PMi : domestic price of imports, tei 1 export subsidy rate, PWEi z world price of exports, P, : composite good price, tdi : indirect tax rate, . input-output coe&ients, a. d;vi I net price or value added, 4 : price index weights, P : exogenous level of price index. : : : : :

exports, export shares, imports, domestic demand for production. exogenous net inflow of foreign exchange.

Ei = EiXf M,lDi=f (PDJPMJ ~PW;M~-~P~~Ei-P=O

Foreign Thade

sectoral production, exogenous sectoral capital stock, labor of category k in sector i, wage of labor category k, total demand for Iabor category k, exogenous labor supply for category k.

Table A.1 Equation of the flexible exchange rent model.”

lnconle andfIow

offinds

<=za14ij.X$

Zi =z S&D, SC, = i7iGC/P, CiP=~WP.Plr---,Pn)

ZDi=OiTZ

F)

groups.

bi

:

-- _.-

_

._______Ip___.

6.. : capital composition coeficients, if ’ investment by sector of origin, GCi : government demand by sectors, expenditure share parameters, Cig : private demand by groups and sectors, v, : intermediate demand, di : domestic demand ratio, x; : total demand for domestic production.

ZDl : investment by sector of destination. Oi : sectoral investment allocation shares,

(V) Product #markets, cont.

_

“Endogenous variables are denoted by capital letters. Lower case letters (except n), Greek letters, and letters with a bar are exogenous variables or parameters. There are n sectors and m labor categories. The letter f denotes a function. In eq. (6) it is a Cobb-Douglas production function. In eq. (3) it is the cost function dual of the CES trade aggregation function. In eqs. (11) and (24) it is derived from the first-order conditions associated with the trade aggregation functions. In eq. (21) it is the linear expenditure system.

Di=dAZi+GCi+C,C,g+ (24) d =f( PDJPM,) (25) Xy=Di+E* (26) Xf-X;=d

(18) (19) (20) (21) (22) (23)

(V) Product markets

Variables calculated: (13) GY : total government revenue, (14) ye : income of socioeconomic groups, (15) TZ : total investment, (16) GC : total government consumption, (17) Cs : total consumption by socioeconomic

(Iv)

92

J. de Melo and S. Robinson, Trade adjustment policies

References Adelman, I. and S. Robinson, 1978, Income distribution policy in developing countries: A case study of Korea (Stanford University Press, Stanford, CA). Brecher, R., 1974, Minimum wage rates and the price theol y of international trade, Quarterly Journal of Economics 88,98-l 16. Chenery, H. a& M. Syrquin, 1975, Patterns of development: 1950-1970 (Oxford University Press, London). Cooper, R.N., 197 1, Currency devaluation in developing countries, Princeton Essays in International Finance, no. 86 (Princeton, NJ). Corden, W.M. and P. Neary, 1980, Booming sector and de-industrialization in a small open economy, M imeo. Dervis, K., J. de Melo and S. Robinson, 1982, General equilibeium models for development policy (Cambridge University Press, Cambridge). Findlay, R., 1979, International trade and development theory (Columbia University Press, New York). Hirschman, A.. 1949, Devaluation and trade balance: A note, Review of Economics and Statistics 31, 50-53. Jones, R., 1971, A three factor model in theory, trade and history, in: J. Bhagwati et al., eds., Trade, balance of payments and growt,,.i-. Essays in honor of C.P. Kindleberger (North-Holland, Amsterdam). Ku bo, Y. and S. Robinson, 1979, Sources of growth and structural change: A comparative analysis of eight countries, Mimeo (World Bank, Washington, DC). Lluch, C., A. Powell and R. Williams, 1977, Patterns of household demand and saving (Oxford University Press, London). MMsrtin,R. and R. Selowsky, 1981, Energy prices, substitution, and optimal borrowing in the short run: An analysis of adjustment in oil importing developing countries, Mimeo. (World Bank, Washington, DC). Melo, J. de and S. Robinson, 1980a, The impact of trade policies on income distributions in a planning model for Colombia, Journal of Policy Modeling 2, no. 1, 81-100. ILfelo, J. de and S. Robinson, !980b, Trade adjustment policies and income distribution in three archetype developing economies, World Bank Staff Working Paper no. 442 (Washington, DC). %lussa. M., 1974, Tariffs and the distribution of income: The importance of factor specificity, substitutability ?y< intensity in the short and long-run, Journal of Political Economy 82, no. 6 1191-1203. Rodriguez, C., 1978, A stylized model of the devaluation-inflation spiral, Staff Papers, 76-89. Sa!ter, W., 1959, International and external balance: The role of price and expenditure effects, Economic Record 35,226238. ‘Taylor, L., 1980, Macro models for developing countries (McGraw-Hill, New York).