Fiscal Implications of Trade Liberalization
  • 1, International Monetary Fund

This paper examines the relationship between trade liberalization and the budget deficit, which depends on the specifics of country’s economic structure, and the trade regime which is being liberalized. It relates some popular but incomplete approaches to assessing this issue (such as analysis of the foreign exchange budget) to a more comprehensive approach using an applied general equilibrium model. The argument is illustrated using data from the most recent of a sequence of abortive planned liberalizations in Kenya, as well as a number of stylized illustrations. The conclusions are not only that liberalization may be budget enhancing, but that in certain circumstances it may be strongly so.


This paper examines the relationship between trade liberalization and the budget deficit, which depends on the specifics of country’s economic structure, and the trade regime which is being liberalized. It relates some popular but incomplete approaches to assessing this issue (such as analysis of the foreign exchange budget) to a more comprehensive approach using an applied general equilibrium model. The argument is illustrated using data from the most recent of a sequence of abortive planned liberalizations in Kenya, as well as a number of stylized illustrations. The conclusions are not only that liberalization may be budget enhancing, but that in certain circumstances it may be strongly so.

I. Introduction

There is now a very considerable literature devoted to the theory and practice of trade liberalization 1/, and a rapidly growing literature devoted to fiscal reform in developing countries 2/. To a surprising extent, these two types of policy reform either have been considered in isolation from each other, or as two separate types of exercise that might interfere with each other, and hence pose a sequencing problem. Relatively few studies 3/ have focused on the extent to which trade liberalization exacerbates the government’s existing budgetary problems, or the converse. Yet this is a consideration which often makes developing country governments wary of embarking on a liberalization programme, and which can lead to the collapse of a programme once adopted.

The problem is that liberalization typically involves wholesale movements in both relative and nominal prices, including the exchange rate itself, and consequential shifts in outputs and expenditures. It is very much a general equilibrium type of problem for which partial equilibrium reasoning may be highly misleading. In addition, liberalization in practice is always partial, rather than complete, so the usual second best types of ambiguity abound. The consequence is that the relationship between the trade policy regime and the budget is complex and not well understood, depending in a detailed way on the fiscal structure, substitution possibilities in production, and the pattern of preferences. The budgetary impact of liberalization cannot be reliably signed without specifying this structure in some detail.

The purpose of this paper is to throw some light on this relationship. To the extent that trade liberalization has budgetary implications, these can be absorbed in three different ways, possibly in combination. First, the fiscal structure may itself be altered to accommodate the changes. Second, there may be a change in net capital inflow to the government, that is a change in the foreign financing of the deficit. Third, there may be a change in the domestic financing of the deficit; in the present context, where domestic capital markets are supposed to be undeveloped, this involves changes in the amount of the deficit which has to be monetized and hence changes in the inflation tax. In practice, the size of fiscal impact that can be absorbed via the inflation tax is fairly limited, so the compatibility of trade liberalization must be ensured either by being part of a self financing package, or by accommodating inflows.

The paper examines the issue under each of these mechanisms. In both cases use is made of a simple CGE model whose structure is given in Annex I. While simplified, the model attempts to capture the salient features of the economy, with a sufficiently rich characterization of the structure of production, preferences and particularly, government, to permit computation of the budgetary impact of the relative price changes involved in liberalization. It can be used to compare the relative (stationary) equilibria under alternative (compatible) trade regimes, and can also track the path between them since it captures the main macroeconomic channels of impact (monetization of the deficit, exchange rate changes, inflation, as well as real changes) other than transitional unemployment. Section 2 attempts to estimate the actual impact of a given liberalization package, using Kenya as an illustration. (Annex 2 details how the model was calibrated on Kenyan data). It turns out that simple quota removal is substantially budget improving in this case, so that compatible liberalization can combine this with tariff reduction. The section focuses on a number of packages, including an aid assisted package, which are compatible in a comparative static sense (that is, they generate steady state inflation rates roughly comparable to the pre-liberalization rate); in particular, it examines the characteristics of the dynamic adjustment under these reforms. Even when the reform is compatible and welfare improving in a comparative static sense, the transition path may be unacceptable to government. For example, it may involve unacceptably high transitional rates of inflation; in these circumstances there may indeed be a role for temporary additional aid to accompany the liberalization, even though this will ultimately be at least self financing.

Section 3 further examines the relationship between fiscal structure, economic structure and liberalization. It again provides illustrative calculations, but in this case of stylized economies. It is concerned with the relative (stationary) equilibrium under different trade regimes, rather than the path between them, so it suppresses the dynamic and monetary features which figure importantly in Section 2. The budgetary impact of various changes is now assessed directly by calculating the change in net capital inflow that would be required to accommodate it. This provides a first cut at the scale of the fiscal adjustment that would be required in the absence of variable capital inflow. In other words, it takes into account the relative price changes consequent directly on the liberalization, in isolation from any further changes that may be induced by the associated financing requirements. Section 4 concludes.

II. Liberalization and the Kenyan Budget

In this section, the impact on the Kenyan budget of a hypothetical liberalization package is considered, starting with a simple foreign exchange budget calculation, then importing relative price changes from the CGE exercise, and concluding with the CGE simulations themselves. It should be stressed that the assumption is maintained throughout that the pre-liberalization regime is sustainable though inefficient; the liberalization package is not embedded within a stabilization programme. Whether or not it is practicable to combine liberalization with stabilization, as opposed to deferring the liberalization until stabilization is achieved, it is clearly inappropriate to attribute fiscal changes required for stabilization to the liberalization programme itself.

In practice, different components of the budget will be affected by the liberalization/devaluation package in different ways. To obtain a rough idea of the overall impact, it is helpful to disaggregate both the revenue and expenditure sides of the budget into tradable, non-tradable and wage components. The procedure is illustrated in Table 1, where data for 1988/89 and 1989/90 are averaged to provide an estimate of calendar 1989 values. The disaggregation between tradable and non-tradable expenditures is fairly crude, based on the relatively highly aggregated breakdown provided in the Economic Analysis of Expenditure tables in the Statistical Abstract: it could be considerably refined using the Appropriation Accounts. However, given the impossibility of accurate calculation of these general equilibrium effects, such refinement is probably not worthwhile.

Table 1.

Central Government Revenue and Expenditure, 1989

(K£ million)

article image
Source: Economic Survey 1991, Tables 6.1, 6.4, and 6.9, from figures averaging for 1988/89 and 1989/90 by simple averaging.Notes:

Labor costs from Table 6.9 plus cost of teachers salaries from forward budget figures in Table 23 of Kenya: Public expenditure Issues, World Bank 1989.

Capital transfers plus current transfers less cost of teachers salaries as obtained in note 1.

Tradable is share of transport operating expenses, purchase of stores, food rations, uniforms, other consumable stores, maintenance, military construction and equipment in ‘Other Goods and Services’ in Current Expenditure plus share of plant, machinery, and equipment and transport equipment in Gross Fixed Capital Formation in Capital Expenditure. Non-tradable is residual of ‘Other Goods and Services’ and Gross Fixed Capital Formation’. Shares calculated from breakdowns in Statistical Abstract 1990, Table 195(b) and 196(b), using averages for 1984/5-1987/8. Share of tradables is computed at 71 percent for ‘Other Goods and Services’ and 20 percent for ‘Capital Formation’.

Business and trading licenses, licenses and fees under the traffic act, other Licenses and duties on production or sale, other revenues and income.

First, consider the fiscal implications of devaluation. Discussions of this often focus heavily on the ‘foreign exchange budget’ to ascertain whether the government is a net seller or purchaser of foreign exchange. From the table, direct transactions in foreign exchange involve inflows from external grants (K202 million) and net foreign financing (K251 million); and outflows for direct imports (K142 million) and foreign interest payments (K159 million). For 1989, the government was a net seller of foreign exchange to the tune of K152 million. Evidently, these components of the budget would be revalued by the full amount of any devaluation, and to that extent would improve the net budgetary position.

There are two important caveats to this conclusion. First, the underlying flows are very volatile so that, for a country like Kenya, not only the magnitude but also the sign of the foreign exchange budget varies from year to year. Table 2 gives the main components of the foreign exchange budget for the five years before 1989. For three of the five years the government was a net buyer of foreign exchange, for two of them a net seller. The move into surplus in the later years was partly a consequence of increased external grants and loans which were directly associated with the liberalization programme.

Table 2.

Foreign Exchange Budget, 1984-88

(K£ million)

article image
Source: Research Department of Central Bank, except for foreign interest payments, which were obtained from simple averaging of data in Economic Survey, Table 6.9.

The second reason for caution in using the foreign exchange budget is more general. It is, at best, only an impact measure which uses a partial equilibrium approach to address a general equilibrium question. A liberalizing devaluation will result in wholesale shifts in relative prices. For a small country, the cif price of imports and the fob price of exports will move, in domestic currency terms, by the full amount of the devaluation. The movement in the implicit-tariff inclusive domestic price of importables, the price of non-tradables, and the wage rate is much harder to estimate, depending on the precise nature of the package, and the ease of substitution in production and consumption.

The obvious way to generate numbers for these relative price changes is via a CGE simulation, which also produces estimates of the associated volume shifts and the composite fiscal consequences of both. Since appropriate CGE models are not routinely available, however, it is of some interest to consider price and quantity effects separately. While the net consequences depend on both, and they are intimately related, it is possible to draw some inferences from the likely relative price movements, even when the quantitative responses to these are unknown. Here we focus on the impact of changes in four prices, as faced by the government (that is net of various explicit and implicit taxes), denominated in domestic currency terms. They are the price of foreign exchange (1/e), the wage rate (w), the producer price of non-tradables (pn) and the producer price of domestically produced importables (pm).

The first step is to recompose the numbers in Table 1 into groups whose values depend on movements in these four prices. We assume 1/ that expenditure on imports and foreign interest, revenue from sales tax on imports, import duties, and export duties, and receipts of external grants and net foreign financing all move with the cost of foreign exchange. Labour costs and transfers, income tax 2/ and other revenue are assumed to move with the wage rate. Expenditure on non-tradables, and 80 percent of the sum of domestic sales tax and excise taxes 3/ are assumed to move with the price of non-tradables. Expenditure on domestically manufactured tradables, and 20 percent of revenues from domestic sales tax and excise taxes are assumed to move with the price of importables. Finally, the domestic currency cost of domestic interest payments is assumed to move with the wage rate, as a proxy for inflation. The alternative would be to assume this component invariant to the various liberalization packages. However this seems an extreme assumption when liberalization markedly increases the inflation rate. The assumption made here is conservative, that is it probably overestimates the adverse impact of liberalization. The results of this exercise are presented in Table 3.

Table 3.

Central Government Revenue and Expenditure, by Price Category

article image
Source: Table 1, and discussion in text.

The attributions in this table are to some extent arbitrary, but the general picture conveyed in the net column of panel (b) appears to be fairly robust for the Kenya case. It is that the directly foreign exchange sensitive budgetary operations run at a substantial surplus, largely offset by deficits in the other three categories. An immediate implication of this is that an augmented foreign exchange budget calculation is considerably more favorable than the direct exercise reported earlier. That is, if the exchange rate alone were to change, leaving the wage rate and price of non-tradables invariant, and the importable price to rise by the full amount of the devaluation, then the base for the budgetary gain would not be net direct foreign exchange sales of K152 million, but K491 million (728-237). This reflects the fact that the government obtains substantially greater (ad valorem) revenue on tradable goods than its expenditure on them. The surplus on tradable activities in general is much greater than that on direct foreign exchange transactions.

More generally, the combination of trade liberalization and depreciation is likely to have still more positive budgetary effects. Broadly, we would expect the price of foreign exchange to rise and the domestic price of importables to fall, relative to the wage rate and the price of nontradables. This is borne out by price paths generated in the simulations reported later in the paper. Table 4 gives a sample of these, together with the net budgetary impact when they are confronted with the budgetary structure of Table 3.

Table 4.

Prices Following Liberalization, Percent

(Relative to non-liberalization prices)

article image
Source: Simulation exercises reported later in the paper.

Takes enhanced capital inflow in years 1 and 2 into account.

While the actual paths tabulated are very different, they all share the property discussed above, (1/e > w, pn > pm). Hence in all cases, the budgetary impact is favourable, ranging from the elimination of the domestically financed component of the deficit (panel (a)) to its reversal from K£118 million to a surplus of similar magnitude (panels (b), (c) and (d)). These latter figures are misleading, however, since they take no account of the tax and tariff changes in the package. For example, in panel (b) the tariff cut would reduce revenue (on unchanged volumes) by around K£130 Million. Overall, the pure price plus tax effects of these packages would imply a very substantial reduction, but not reversal of the domestic deficit.

This discussion has suppressed all volume responses. Tables 5, 6, 7, 8 and 9 present the main quantity, price and budgetary effects of a number of liberalization packages, obtained by simulation using the model whose structure is given in Annex 1 and whose calibration to Kenyan data is described in Annex 2. Results are presented for the first four years of the adjustment path, and for the new steady state to which the system is converging. The model is calibrated for the pre-liberalization equilibrium to involve a level of deficit financing sufficient to sustain steady state inflation at 9 per cent per annum, the historical average. Four types of package are considered, in 3 variants of the model. The four packages are as follows.

Table 5.

Liberalization-cum-Harmonization 1/

article image
Source: Simulation using model detailed in the Annexes (post equilibrium exchange rate is arbitrary, depending on adjustment path, so given as *)Note:

Abolition of quota, tariffs harmonized at low uniform rate of 6.4 percent.

Table 6.

Complete Liberalization 1/

article image
Source: Simulation using model detailed in the Annexes (post equilibrium exchange rate is arbitrary, depending on adjustment path, so given as *).Note:

Abolition of quotas and tariffs: sales taxes raised to uniform rate of 18 percent.

Table 7.

Aid Assisted Liberalization-cum-Harmonization 1/

article image
Source: As Table 5.Note:

As Table 5, but with increased capital inflow to government in first 2 years of liberalization programme.

Table 8.

Liberalization-cum-Harmonization: Rapid Depreciation Case 1/

article image
Source: As Table 5.Note:

As Table 5, but with capital depreciating at 20 percent per annum, rather than 5 percent per annum.

Table 9.

Quota Abolition with Counterfactual Representation of Initial Trade Regime 1/

article image
Source: As Table 5.Note:

Simple quota abolition; counterfactual specification of initial trade regime has importable production protected solely by quotas and all tariffs on non-competing imports.

First, quantitative restrictions are abolished in Year 1, but no accompanying fiscal changes are made. This is a large and abrupt trade reform: the implicit tariff on competing imports falls by 55 percent (from 1.20 to 0.54), amounting to a 30 percent fall in the domestic price before we consider induced exchange rate changes. Subsequently trade policy is constant. It turns out, given the fiscal structure described in Table 1 and Annex 2, that this is powerfully budget surplus enhancing. This is partly because of the relative price effects discussed earlier (Table 4 panel (a) refers). It is also because the quantity constrained imports also carry very high tariff rates. Hence the expenditure shifts into competing imports are strongly revenue enhancing. These effects are strong enough to drive the domestic budget into substantial surplus unless foreign capital inflows are reduced. Indeed, restoration of a domestic deficit of the pre-liberalization magnitude (so that the system converges back to a 9 percent inflation rate) requires the capital inflow to be eliminated. (K£230 million net inflow to K£30 million net outflow).

While this outcome is clearly sustainable, we assume here that the government would wish to maintain net capital inflow at something like its historic level. In consequence, this first liberalization package would in practice be accompanied by other fiscal changes to take up the budgetary slack. We consider three such packages on the base model, and the figures are given in Tables 5, 6, 7. Each package is designed to have the property of long run fiscal neutrality, in the sense that the new equilibrium has a steady state inflation rate in the broad vicinity of 9 percent. The liberalization cum harmonization package (Table 5) supplements the abolition of quotas with a massive cut in the tariff rate on competing imports to equal the much lower rate on non competing imports (a cut from 54 percent to around 6 percent). This cut also takes place in year one; it generates a steady state inflation rate of 8.2 percent per annum. The complete liberalization package (Table 6) abolishes quotas and all tariffs, compensating for this by raising sales taxes from 16 percent (non-tradables, competing imports and domestically produced importables) and 11 percent (non competing imports) to a uniform rate of 18 percent. Once again, these changes all take place in year one and are maintained thereafter; in this case the new steady state inflation rate is 9.7 percent per annum. Governments of developing countries are prone to negotiate for additional transitional aid to “finance” the liberalization process. The last package considered, the aid assisted liberalization (Table 7), replicates the liberalization cum harmonization package of Table 5, with the addition of an enhanced capital inflow of K£50 million in each of the first two years.

All these exercises are conducted within the confines of the base model calibration of Annex 2. This is characterized by relatively slow depreciation and hence rather slow adjustment of the capital stock. Table 8 reworks the case of Table 5 under the assumption of a much faster potential transition (depreciation is at 20 percent per annum instead of 5 percent) The new steady state inflation rate is 7.0 percent per annum. A more radical alternative is depicted in Table 9, which utilizes the counterfactual assumption that all protection of domestic importable production takes the form of quotas, so that all tariff revenue is levied on non competing imports. This turns the simple quota abolition from being powerfully revenue enhancing to being mildly detrimental (the steady state inflation rate rises to 11.6 percent per annum).

The model is described in Annex 1: in brief, it has four goods; three produced using a fixed factor, labor (mobile between sectors, fixed in aggregate) and capital (mobile between sectors, variable in aggregate) and three consumed. There is an importable good and a non-tradable good (produced and consumed); an exportable good (produced but not consumed); and a foreign good (consumed but not produced). These are indexed by m, n, x and f. The distinction between competing and non competing imports is particularly important in capturing the structure of trade policy in many developing countries. Production sectors share the same elasticity of substitution between capital and labour, but differ in their capital intensity (and in reliance on the fixed factor). In consumption, any pair of the three goods may be specially related (either as substitutes or complements) viz à viz the third. Government volume purchases are taken to be independent of relative prices: hence relative price changes translate directly into public expenditure changes. Taxes are ad valorem so revenues respond to private expenditure changes, whether emanating from price or quantity changes. The government may also make discretionary tax changes as part of its liberalization programme. Except for the “aid assisted” case, it is assumed that capital inflows are fixed in foreign currency terms. The residual budget deficit is monetized, and via a simple transactions demand for money this generates a variable inflation rate. Gross investment in each sector is linearly related to the gap between the immediately previous sectoral rate of return and a target rate, subject to it being non negative. Two further assumptions are made about investment. First, it has the same sectoral composition as consumption expenditure. Hence the split of private spending between consumption and investment has no impact on the composition of output. The second, and closely related, assumption is that the same function that determines the substitutability of goods in producing utility also determines their substitutability in producing units of capital. It is a consequence of the assumed fixity of financial capital inflows that changes in the private domestic investment rate must be domestically financed. The target rate of return reflects domestic time preference, not any assumed perfect international capital market.

Each table gives data for the common pre liberalization year, assumed to be in steady state, the first four years of liberalization, and the steady state to which the economy converges following liberalization. Data are given for output volumes (Qi), private expenditure at constant prices (Ei), the volume of competing imports of the private sector (Em-Qm), sectoral labour force (Li) and capital stock (Ki) aggregate gross and net real investment (Ig, In), and sectoral (rij) and average (riav) real rates of return. The wage rate (w), producer price of non-tradables (pn), and consumer prices (pic) are all given in terms of their foreign exchange values using the current exchange rate (foreign currency obtained per unit of domestic currency, e). The same convention is adopted in reporting the budgetary magnitudes (gexp, gwag, drev irev aid the domestically financed deficit, def); the value of GDP at factor cost (gdp where importable production is valued at distorted domestic relative prices, gdpw where it is valued directly at world prices); the real value of the money stock (m1) and the value of private expenditure at producer prices (ex); and the prices of a composite unit of capital and of consumption (pkap, pcon). u2 is the real value of private expenditure, and uc the representative level of utility generated by private consumption. (No allowance is made for the benefits of government expenditure, but this is modelled to be stationary). Finally, the true rate of inflation of consumer prices is given as the rate of change of the domestic currency value of pcon (infln). For the pre and post liberalization equilibrium, the steady state inflation value is given; for years 1-4, the year on year value.

All the cases tabulated have an immediate abolition of quantitative restrictions as one component. As earlier noted this is a large and abrupt trade reform; the exchange rate falls by around 22 percent compared to its usual depreciation of 9 percent. Most dramatically, the sector producing importables sheds two thirds of its labour force in the first year of transition. This is partly a reflection of the fact that in the model as calibrated to Kenyan data there is really no future for a domestic sector producing importables in the absence of substantial protection. In the model, all this released labour is immediately absorbed, entirely in the exportable sector, since non-tradable employment is stable or falls slightly. In practice this rate of reabsorption would be extraordinarily difficult to achieve, even in those African economies with very high rates of circular migration between rural and urban sectors. In consequence abrupt liberalization is likely to be more disruptive and slower to achieve beneficial resource reallocation than the present simulations portray.

Once this substantial initial switch takes place, the transition appears to become somewhat becalmed on a plateau, even when the system is some way from the final equilibrium. For example, in Table 5 Lm falls in year 1 to 35 percent of its initial value; by year 4 it has fallen only to 28 percent, or one fifth of the remaining adjustment. This plateau effect is common to all the main quantity variables other than those involving capital.

The second relatively systematic feature to note is that aggregate investment falls, and hence so does the capital stock. This is a feature of adjustment programmes which has been much remarked recently 1/. In the present context it arises simply because the contracting sector is far more capital intensive than the expanding sector. Although the real price of capital falls, and also falls relative to the wage, so that sectoral capital intensity actually rises, this is not enough to offset the sectoral switch, and aggregate desired capital and gross investment fall. In the low depreciation case, the fall in net investment is fairly modest, roughly to its new equilibrium value (Table 5). In the high depreciation case, the effect is much more dramatic, with net investment being nearly eliminated during the four transitional years depicted (Table 8). One consequence of this is that welfare actually overshoots its new equilibrium value during the transition. (Of course, transitional unemployment and other costs of the transition which are neglected here might overturn this result).

The third feature is the time path of the budget deficit and associated inflation. In each of the five cases tabulated, the reform package is calculated to generate an equilibrium domestic deficit which can be financed by a roughly similar rate of inflation tax as the pre reform deficit. This requires the foreign exchange denominated value of the deficit to fall, since the value of expenditure and hence required real money balances (also denominated in foreign exchange) are lower in the new equilibrium. However, in the absence of increased capital inflow, in each case the deficit initially rises before it falls. This has the effect of accelerating the inflation rate temporarily, after the initial period of grace due to the substantial fall in the price of importables. Thus in Table 5, the consumption inflation rate drops to 4.2 percent in the first year of reform, and then jumps to 17.1 percent, 14.7 percent, and 13.0 percent during years 2, 3 and 4. The fully employed representative agent in the model presumably would be mollified by the welfare gain that accompanied this inflationary surge. However, in practice, with substantial redistributions taking place and very incomplete information about what is happening, high rates of inflation may be very damaging both to a reform’s credibility and to its political acceptability. In consequence, it is not unreasonable for governments to wish to minimize the inflationary consequences of reform.

To the extent that the pattern of Tables 5, 6, 8, and 9 is typical, there are two options available. Either the government can engage in a temporary tightening of fiscal policy, or it must obtain increased capital inflows. Since trade policy reform may already be a politically fragile option--because it threatens the interests of the elite--fiscal measures which impact on the population at large may be very unattractive. It is natural therefore that a government pressurized to implement a trade reform should seek additional transitional aid.

Table 7 repeats the simulation of Table 5, with the addition of a temporary injection of aid during the first two years. It has sometimes been suggested that aid assisted liberalization is very problematic because it slows the resource reallocation 1/, However, in the present case it appears to be remarkably successful. The disruption it causes to the necessary process of resource reallocation is minimal--the real figures for year 4 are almost indistinguishable for Tables 5 and 7, and the earlier years differ very little. On the other hand, the inflation sequence is radically altered. The unproblematic first year is marginally reduced to 3.3 percent but the rates for years 2 and 3 are more than halved to 8.1 percent and 6.4 percent. The consequence is that the inflation rate falls below the norm for the first three years of the programme, rising to only a little above it temporarily thereafter.

The fourth feature of the tables to be stressed is how remarkably consistent they are in respect of these major features. If a trade and fiscal reform were competently designed for long run compatibility, then, in the circumstances portrayed by the model of Annexes 1 and 2, the transitional path would very likely involve very substantial labour movements, increased budget deficits, accelerating inflation and an investment collapse. It is hardly surprising that many liberalizations are aborted in practice well before completion.

III. Fiscal Structure, Economic Structure and Liberalization

The previous section analyzed the dynamic aspects of liberalization, concentrating on packages which were compatible in comparative static terms, and using a stylized model of the Kenyan economy as a vehicle for the analysis. In this section we turn briefly to consider the circumstances in which liberalization packages might or might not have the property of comparative static compatibility. For this purpose, we continue to use the model of Annex 1, but now concentrate on the static features, and use arbitrary coefficients for the comparison.

The deterioration or improvement in the budgetary position is fully covered by a change in the level of foreign financing, so no fiscal adjustment is required. While this is normally a poor way in which to close this type of model, it is appropriate in the present case, since it enables us to track the extent of the ‘pure’ fiscal consequences of liberalization. The alternative assumptions of exogenous foreign finance or limited accommodation would force endogenous budget adjustment, obscuring the pure effects.

Before providing illustrative calculations, it is helpful to consider the following highly stylized liberalization. Trade policy in the initial equilibrium consists entirely of quota restrictions. Liberalization takes the extreme form of abolishing quotas without introducing tariffs, and allowing the exchange rate to equilibrate the balance of payments. The question is whether this package needs to be accompanied by fiscal reform. In other words, were the quotas in any way substituting for a tighter fiscal stance? Economies often end in a quota ridden equilibrium having started with an excessive fiscal deficit that spilled over into a balance of payments problem which was countered by the adoption of quantitative restrictions on imports. However, other policy changes, including fiscal ones, will also have been undertaken during this process.

In a very simple model 1/, where government spending on importables and non-tradables exhibits the same (constant) elasticity of substitution as private spending, it turns out that for equilibrium to have been achieved, it is analytically as plausible that the fiscal regime has become tighter than would be required for free trade as that it remains too lax to be consistent with it. The reason is that quotas do not remove the excess purchasing power in the hands of the private sector, given the government’s spending. They simply re-route it away from imports towards import substitutes and non-tradables, drawing resources out of the export sector. Whether this process helps to make the original fiscal laxity sustainable, or whether it exacerbates it, depends on the elasticity of substitution in consumption between importables and non-tradables, and on the pattern of the budget. If government revenues are more concentrated on tradables than government expenditures, and if tradables and non-tradables are net complements in consumption, then some fiscal tightening (an increase in taxes or a reduction in spending) must accompany the removal of quotas. The same is true if revenue is relatively more concentrated on non-tradables, and the goods are net substitutes. On the other hand, if revenue is derived relatively more from tradables (non-tradables) and the goods are net substitutes (net complements), the removal of quotas could be accompanied by fiscal relaxation.

While it is relatively common for governments to obtain a high proportion of their revenue from the tradable sector, and to devote a high proportion of their expenditure to non-tradables, there does not appear to be any obvious presumption as to the substitutability of those goods in consumption. In consequence, there is no strong a priori presumption that a liberalizing government will need to generate additional revenue to make the liberalization feasible.

This argument has been constructed for a very stylized case. In practice, the conclusion is likely to be reinforced. Typically there are tariffs in place and these are usually higher on the most quantitatively restricted goods as in the Kenya case of the previous section. Hence, additional revenue is generated automatically as the volume of imports expands, the composition shifts towards more heavily taxed goods and the domestic price of imports rises. However, the actual budgetary impact of a liberalization will depend on the broad structure of the budget and of the economy in question.

1. Illustrative calculations

In all the calculations reported here, unless explicitly noted otherwise, the capital intensities (measured by the profit share in profit and wage income) of the three production sectors are proportional to 0.4 (importables), 0.3 (non-tradables), and 0.2 (exportables) with equality holding in the undistorted equilibrium. Returns to capital and labour combined are 0.8 and the elasticity of substitution between them is 0.5. The distribution of private expenditure in the undistorted state is 36 percent on importables, 54 percent on non-tradables and 10 percent on the foreign good. The elasticity of substitution in consumption is 1 between importables and non-tradables, and 0.2 between this composite and the foreign good. There are no direct taxes on wages, profits or rents. The government purchases equal quantities of the three consumption goods, and 15 percent of labor services. The output of these services does not enter into production or private consumption. The indirect tax regime is stipulated in each set of calculations.

In Table 10, column (1) gives the equilibrium of the economy under non-distortionary taxation, provided by tm = tn = ttf which is tantamount to a uniform sales tax. This provides a benchmark against which equilibria can be measured. At the given parameter values the budget deficit is 0.034 or 2 percent of GDP at factor cost, so this might be regarded as a sustainable equilibrium.

Table 10.

Trade Liberalization: Illustration with Exports Highly Elastic 1/

article image

Gm = Gn = Gf = 0.05.

def(%) is deficit as percent of GDP.

Column (2) portrays a highly illiberal fiscal/trade regime. Apart from the 20 percent tariff on the foreign good (ttf) there is now a 20 percent tax on exports (tx), a 30 percent tariff on importables (ttm) and, in addition, a fairly savage quota on importables having a tariff equivalent of 80 percent. Since this is over and above the tariff of 30 percent, the full tariff equivalence of the two combined is 1.34 (1.8 × 1.3 - 1). All revenues are now generated by trade taxes: because of the high elasticities involved, revenues are very low (competitive imports have fallen from 0.383 to 0.048 and exports from 0.582 to 0.043, offset partly by increased imports of the foreign good from 0.133 to 0.228); however expenditures have risen, reflecting the rise in prices of importables and non-tradables, and of wages. The deficit, in consequence, is 0.333 or more than 12 percent of GDP.

We now consider a variety of liberalization measures. In column (3), the quota is simply abolished, leaving all other taxes in place. Total private imports are substantially up (0.203 + 0.132 in place of 0.048 + 0.228), as is the tariff yield. There is a large gain in export volume (up from 0.043 to 0.295). Coupled with the very substantial decline in prices, the budget improves to a deficit of 0.140. In this case the abolition of quotas is substantially beneficial to the budget as well as to resource allocation.

Column (4) illustrates the consequences of replacing the quota with an equivalent tariff, rate 1.34. Interestingly, this hardly improves the budgetary position from the illiberal regime of column (2) and, if anything, worsens the allocative outcome. (Resources move yet further into the importables sector). In this case, the budgetary consequences of replacing a quota with a tariff are unambiguously worse than simple abolition.

Column (5) combines quota replacement with a tariff and abolition of the export tax. In the present case this has remarkably little revenue impact, since exports are already so small but nor does it lead to any substantial recovery of exports.

It is apparent from the table that there are at least four different meanings of tariff equivalence of a quota. The usual meaning is the one underlying column (4), that yields an equivalent elevation of domestic producer prices over the world price. However, it does not necessarily imply the same level of imports, the same scale of domestic output of importables, or the same level of employment in that sector. In the present case, the ‘equivalent tariff leads to a substantially more severe contraction of imports (0.029 against 0.048). Column (6) gives the tariff rate (absent quotas) which yields a similar degree of import compression; namely a tariff rate of only 98 percent. Similarly column (7) gives the rate (112 percent) which yields the same level of production of importables: and column (8) the rate (110 percent) which yields the same level of employment in the importable sector. Column (9) shows that abolition of all taxes as well as quotas yields a smaller absolute value of the deficit than the quota-ridden equilibrium of column (2).

How dependent are these results on the high degree of elasticity of export supply? Table 11 gives comparable calculations when this is very low. In particular, the elasticity of output with respect to the variable factors in the exportable sector is now set at 0.1 instead of 0.8. The same tax variations are considered.

Table 11.

Trade Liberalization: Illustration with Exports Highly Inelastic 1/

article image

Gm = Gn = Gf = 0.1.

def(%) is deficit as percent of GDP.

Once again, column (1) gives the non-distortionary benchmark, and column (2) the illiberal equilibrium. As can be clearly seen, the quantity shifts are much more muted, with the fall in export volume being 10 percent instead of 93 percent. Despite this dramatically less responsive economy, the rest of the story is remarkably similar. In particular, abolition of the quota (column 3), dramatically improves the budget deficit (0.250 to 0.046). In this case, however, replacement with an equivalent tariff is still more beneficial, (column 4), yielding a budget surplus of 0.132. The cost of this further budgetary gain is a dramatic worsening in resource allocation, with a very large further increase in the level of resources going to the importables sector and a very sharp reduction in imports. Indeed, the tariff having an equivalent compressive effect on imports would only be a little over the existing 30 percent rate at 36 percent, while that sustaining an equivalent level of domestic output would be around 43 percent (column 7). Finally column (8) shows the reduced uniform sales tax rate (9.7 percent) which would maintain the budget deficit at a level equivalent to that generated by the illiberal trade regime of column (2).

In both cases, the abolition of quotas markedly improves the budget deficit. This is not, in the case of Table 11, because tariffs and quotas combined have reduced the level of imports below the revenue maximizing level. It is due partially to the gain on export duties, but very substantially to the fall in the public sector wage bill associated with the large fall in pn and w.

These calculations have two main implications. First, they show that the large potential fiscal gains from liberalization discussed in the previous section depend neither on very high tariff rates within the protected sector, nor on the presence of very elastic export supply. Second, they raise difficult questions concerning tariff equivalence. Replacing a composite of tariffs and quotas by a tariff regime which is equivalent in the sense of elevating domestic prices to the same extent [(1+ttm1) = (1+ttm) (1+ttq)] is far more severe than the alternative definitions, which include (1) maintenance of a similar degree of import compression, (2) maintenance of the same level of domestic production of the importable, (3) maintenance of the same level of employment in domestic production of the importable, (4) revenue neutrality. In particular, any attempt to preserve equivalence defined by price neutrality is likely to involve significant overshooting by the other criteria. This is likely to be inefficient and may even threaten the implementation of the reforms.

IV. Conclusion

Developing country governments are often resistant to pressures from international agencies and others that they should liberalize their trade regimes. There are two distinct types of reason for this. First, the government may be associated with or dependant on elites who benefit directly from the rents associated with illiberal trade regimes. Reduction or removal of these rents poses a real problem in political economy. Second, even if the government is able and willing to confront these political difficulties, it may be concerned that liberalization will be problematic in more narrowly economic terms.

The particular problem on which this paper focused is that liberalization may exacerbate the government’s already difficult fiscal circumstances, and lead to larger deficits and accelerating inflation. One reason for the prevalence of this concern lies in the historical evolution of the trade regime itself. A quota ridden equilibrium has often arisen not so much from concerns to protect domestic industry as from some initial fiscal laxity spilling over into a balance of payments disequilibrium. Since the gradual imposition of tighter quotas was originally a response to a fiscal problem, it is natural to worry that quota relaxation will lead to the re-emergence of the deficit. However, it is clear from the earlier discussion in the paper that quota relaxation may as well ease the government’s fiscal situation as worsen it. This conclusion can still hold even when the liberalization package includes tariff reductions as well as quota reductions, and when the government is a net purchaser of foreign exchange. In short, there is no presumption that liberalization requires an associated fiscal tightening to be rendered compatible in the long run.

However, it also emerges from the simulations reported in Section 2 that the dynamic path of adjustment may be more problematic. This has two aspects. First, there is likely to be a temporary deterioration in the budget deficit and hence the foreign financing requirement or domestic inflation rate even when the long run consequences of the programme are sustainable. This may lead to severe political difficulties for a reforming government, and there may be a substantial role for an aid assisted package. Second, in an uncertain world, these temporary phenomena may call into question the credibility of the entire programme. It may wrongly be assumed that the adverse transitional characteristics imply that the programme is not in fact compatible in the long run. Alternatively, it may be assumed--perhaps correctly--that they imply the government will be unable politically to sustain the programme, so that even though feasible, it becomes incredible.

This leaves government facing a real dilemma. It may be necessary to overshoot the long run equilibrium, by tightening the fiscal system more that is economically justified, to ensure political implementation. This involves an inefficient set of signals, and resource allocations, and may make the whole reform programme unacceptable. On the other hand, a feasible (and politically acceptable) programme may have transitional characteristics that soon undermine its credibility, or that of the government’s continued resolution.