تغییر در ترکیب هزینه های دولت در پاسخ به بار بدهی خارجی

This paper examines the effects of external public debt burden on the composition of public spending in a sample of 47 developing countries for 1972–2001. We specify a system of equations for the relative spending shares of six economic categories and estimate it using several estimation methods and samples. We find support for the adverse effect of the debt burden on “capital expenditure” in the full sample and in the Middle East-North Africa and sub-Saharan Africa subsamples where the debt burden was relatively high. Among components of “current expenditure,” increased debt burden shifted the shares against “nonwage goods services” and “subsidy and transfers” while leaving the share of politically sensitive category of “wages and salaries” unscathed in most cases.

In the past three decades or so, a number of developing countries (DCs) experienced major episodes of financial crisis that were characterized by unsustainable fiscal deficits and rising external debt. The deficits raise the level of external public (and publicly guaranteed debt) through two channels: Directly, they cause more external public borrowing for financing purposes. Indirectly, when coupled with overvalued exchange rates, they encourage capital flight. This, in turn, drains national foreign exchange reserves and leads to “flight-driven external borrowing.”1 Not surprisingly, since the “debt crisis” of the early 1980s, fiscal deficit reduction became a key component of the macroeconomic adjustment programs that many indebted countries had to undertake as a condition for their inclusion in debt restructuring and debt relief initiatives and/or eligibility for the IMF's “rescue packages.” Two main issues arise in relation to efforts aimed at containing the fiscal deficits. The first issue is the distribution of total reduction in the size of the deficit between spending cuts and higher (tax) revenues. It may be argued that, in general, the spending side of the budget is likely to bear a heavier burden of deficit containment than its revenue side. For one reason, spending cuts can be more quickly implemented than measures that can generate more tax receipts. Moreover, raising taxes may be perceived as a policy with distortionary effects and in conflict with reducing the role of government (Woodward, 1992).2 The second issue is the distribution of total spending cuts among various categories of public spending. Since interest payments on the public debt is a relatively substantial and inflexible component of total public spending, such distribution presents a particularly challenging problem: If spending cuts mainly fall on expenditure categories that affect current income and consumption levels of large segments of population, they will have adverse welfare (poverty) implications. The latter may, in turn, lead to a rise in the levels of public discontent and political instability. On the other hand, if the brunt of spending reductions is borne by categories that maintain and enhance the economy's productive capacity, it may well endanger the long-term economic growth.3 This is particularly true in countries where the public sector is the source of much of total investment in infrastructure and fixed capital. Concerns regarding the growth and welfare implications of public spending cuts, particularly under IMF-supported adjustment programs, generated many debates. Attempts to shed light on various aspects of these debates resulted in a large body of empirical studies. A relatively small branch of these studies focused on the distribution of the burden of spending adjustment among the relative shares of various functional and economic categories of public spending.4Hicks and Kubisch (1984), for example, noted that unlike capital expenditures, social and defense expenditures tended to be well protected during periods of fiscal retrenchment over the period 1972–80. A later study by Hicks (1989), essentially confirmed this finding. After examining the data from a sample of 11 “high-debt” countries during 1978–84, Hicks concluded that the major burden of expenditure reductions fell on the heavily capital-intensive sectors (particularly infrastructure) as well as real wages and salaries of government employees. At the same time, social and defense sectors were sheltered. In a more formal and extensive study, Heller and Diamond (1990) analyzed the data for a large number of DCs over 1975–86. An important finding of their study was that, among the economic categories of spending, the shift in spending patterns turned out to be most pronounced against “fixed assets and capital transfers and toward interest, subsidy and transfer payments, given the functional structures of expenditures over the period” (p. 2). Papagapitos (1992) attempted to address widespread concerns regarding stabilization programs by estimating the impact of various proxy variables for the programs on the share of “developmental expenditure.” Using a sample of 25 DCs over 1972–88, he estimated the relative share equations for various functional categories of government expenditures. Interestingly enough, he found that “for the most part, stabilization programs do not appear to consistently shift public expenditures away from growth-augmenting areas.” (p. 85). Habibi's (1994) study focused on the political dimension of the allocation of total spending among competing functional and economic expenditure categories. His analysis of the data for a cross-section of 67 countries (with 1984 as the reference year) suggested that a higher level of political democratization was associated with a higher relative budget share of “subsidies and other current transfers” and a lower share of “nonwage goods and services.” As its main objective, the present study seeks to investigate empirically the way an increase in the external public debt––directly (through changes in spending priorities) or indirectly (through higher interest payments)––shifted the relative shares of economic categories of public spending in a sample of indebted DCs. Our investigation is motivated by the fact that: (a) Debt-induced changes in public spending priorities and their social and economic implications continue to generate debates among professional economists and draw criticisms from nongovernmental organizations (NGOs).5 (b) The nexus between public debt and the mix of public spending is an important link in the “causal chain” running from public debt to economic growth or welfare. As such, our empirical examination of this nexus nicely complements the results of several recent studies in which public debt or spending composition was directly related to economic growth rate, private investment, or measures of poverty.6 Some more specific features of our empirical analysis are as follows: First, we use an updated pooled data set that contains observations for 47 developing countries collected over a time period spanning roughly three decades (1972–2001). The extended time-series dimension of the pooled data set employed here should yield a more accurate assessment of the impact of public debt burden and other variables on the composition of public spending than those reported by Heller and Diamond (1990) for 1978–86. Second, unlike many previous studies, we define various economic expenditure categories in terms of their relative shares in total government expenditure. This definition is intuitively more appealing for, as noted by Habibi (1994), fiscal decision makers are likely to be more concerned about funds allocated to a particular spending category as a percentage of total spending than GDP. Third, we test the sensitivity of our results to various ways in which the (unobservable) country- and/or time-specific effects in the pooled data set may be captured and to the sample composition.7 Finally, as an extension of our analysis, we also estimate the impact of external debt on the size of the public sector. In view of the features noted above, it would be interesting to see whether our results are consistent with those obtained from earlier studies. Our findings should also be of interest in relation to sustainability of fiscal adjustment programs and debt relief, for both depend on the reorientation of the composition of public spending toward “productive” and away from “unproductive” spending categories. In Section 2 of what follows, we specify a system of relative share equations. Section 3 provides a brief description of the data set. Section 4 presents our empirical results. The last section contains a summary of the major findings and concluding remarks.

High public debt levels are usually accompanied by high interest payments, fiscal austerity, and changes in spending priorities that may require a reshuffling of the budget shares of various spending categories. The way this reshuffling changes the mix of spending has important implications regarding the effect of debt on future economic and current welfare. In this connection, we explored the relationship between a measure of external public debt burden and the budget shares of six economic categories of public spending in a sample of 47 indebted developing countries over the period 1972–2001 and four subsamples. Our main empirical results may be summarized as follows: (a) Controlling for the effects of several relevant factors (including the presence of IMF-supported programs), we found evidence of a significant statistical relationship between the public debt burden and one or more spending shares in all samples. But, the debt burden turned out to be a more important factor in altering the mix of spending in the Middle East and North Africa and sub-Saharan Africa regions where the burden of debt was relatively heavy. An extension of our analysis further suggested that a higher debt burden was significantly associated with a larger size of the government in the full sample. This relationship, however, did not hold in any of the subsamples. (b) The pattern of adjustment in spending mix with respect to changes in the debt burden showed some similarities and differences across various samples. The burden of debt invariably changed the composition of spending in favor of interest payments (INT) and displaced the share of nonwage goods and services (OGS) category in most cases. The share of public capital expenditure (CEX) category, however, fell as the debt burden rose in the full sample and the MENA and SSAF subsamples. (c) In relation to the current expenditure part of the budget, increased debt burden reallocated relative shares against the subsidy and other current transfers (STR) in the full sample and the ASIA subsample in a statistically significant manner. With the exception of the CSAMC subsample, the share of politically sensitive category of wages and salaries of public employees (WGS) was not adversely affected by the debt burden. In fact, the public debt burden had a significant positive impact on this category in the full sample and the SSAF subsample. Two composite pictures seem to emerge based on the frequency of statistically significant coefficients of the debt burden in various equations and samples: (i) Among the relative shares, the share of nonwage goods and services was the least protected category from direct and indirect effects of the debt burden, while the share of wages and salaries was well sheltered. (ii) Among the subsamples, the pattern adjustment in the spending shares in the SSAF subsample was most favorable toward sheltering current expenditures (mainly wages and subsidies) and against spending that raise and maintain physical capital stock. These observations perhaps reflect relatively lower political costs of shifting the burden of debt related adjustments to a spending category whose benefits are less direct and concentrated, and higher priorities attached to sheltering direct sources of current income and consumption especially where they are low to begin with. The implications of our results regarding debt relief initiatives and reorientation of the mix of public spending crucially depend on the extent to which each of the aggregate spending categories can be considered as “productive.” If one simply considers WGS and STR as (mainly) “unproductive” and OGS and CEX as (mainly) “productive” spending categories, then fiscal reforms that reorient spending mix away from the former and toward the latter categories would be desirable especially in the MENA and SSAF regions. In this case, debt relief initiatives that tie debt reductions to such reforms should raise growth.31 The implications of our results may however, be less straightforward in view of the findings of several recent studies suggesting that, depending on a number of factors, increases in some components of current expenditure may promote growth while further capital expenditure may actually retard it.32 If so, changes in the distribution of shares within the current expenditure portion of total spending matter; for they affect not only the current welfare (poverty) of different groups but also the long-term growth. Therefore, especially in low-income countries, it is even more imperative to adopt a careful and selective approach to cutting components of current expenditure as a condition for debt relief during periods of fiscal retrenchment.