I. Introduction

The purpose of this paper is to review recent developments in military expenditures worldwide and for different country groups, present a preliminary analysis of the effect of changes in military spending on other expenditures and fiscal stance, and assess the resilience of military spending in the face of fiscal adjustment. Earlier studies, ^2/ which analyzed worldwide trends in spending from 1972 to 1990, based on data from Stockholm International Peace Research Institute (SIPRI), had concluded that military expenditures absorbed some 5 percent of the world’s output between 1972 and 1990. It was also found that military spending had declined steadily worldwide between 1985 and 1990, from 5.6 percent of GDP to 4.3 percent of GDP, with declines in all regions and for both industrial and developing countries. This decline contrasts with the experience of the previous decade, when there was a significant rise in military spending, ^3/ and was attributed to a combination of factors, including a slowdown in economic activity in both industrial and developing countries in the period 1985-90, increased democratization, improvements in the world-wide security environment, and a fall in bilateral military assistance from industrial countries.

There are a number of sources of data on military spending, all of which have shortcomings, resulting in large part from the confidential nature of much military activity. ^4/ In addition, different data bases vary mainly by the definition of military spending employed, coverage, and treatment of calendar and fiscal year data. We have attempted to address data problems in several ways. First, the paper presents results from several sources, including the IMF’s World Economic Outlook (WEO), SIPRI, and the International Institute for Strategic Studies (IISS). ^{5/ 6/} The countries included in each data base differ, as do results for individual countries and groups of countries. However, the broad trends in military spending are similar. We also compare results obtained from the three data bases, as well as Government Finance Statistics Yearbook (GFS), for a smaller group of countries for which complete data are available for the years 1985-92 for all four data bases. ^7/ Again, while differences between the data sets exist, the basic picture that emerges from each is similar. Nevertheless, while one can have confidence in broad trends that emerge, particular point estimates, or results for individual country groupings, should be viewed with caution.

We find, utilizing any of the three data sets, that the decline in military spending that began in the mid-1980s has continued through 1995 and that this decline may have had an important effect on nonmilitary spending and the ability of countries to undertake fiscal adjustment. Also, in contrast to findings for previous periods, ^8/ military spending was seen to be not resilient to fiscal adjustment; that is, rather than increasing its share of total spending when total spending falls, military spending has declined more than proportionately in those countries that have reduced total spending. Countries with Fund programs are seen to have reduced military spending more sharply than developing countries as a whole, although this largely reflects outcomes in the transition economies.

In Section II, we examine worldwide trends in military spending, and provide an estimate of the size of the “peace dividend” associated with the worldwide reductions in military spending. In Section III, we look at military spending by specific country groups. Following that, in Section IV, we examine trends in military spending for countries with Fund-supported stabilization and structural reform programs. Section V considers the impact of changes in military spending on other aspects of fiscal policy, in particular nonmilitary spending and the fiscal deficit, in order to address the issue of how the peace dividend has been used. Section VI examines the issue of whether military spending has been resilient in periods of fiscal adjustment. Finally, Section IV briefly summarizes, discusses potential policy implications, and identifies issues for future research.

II. Trends in Aggregate Military Spending

Our analysis of data for the years between 1991 and 1995 indicates that the downward trend in military spending that began in the mid-1980s has continued. Utilizing WEO data for 130 member countries, ^9/ weighted by country GDP, we find that worldwide military spending has declined from 3.7 percent of GDP in 1990 to an average of 2.8 percent of GDP over the period 1991-95 and to a low of 2.4 percent of GDP in 1995 (Table 1 and Appendix Table 1). ^10/ Results for unweighted WEO data (Appendix Table 2) show a similar decline, from 4.1 percent of GDP to 2.7 percent of GDP over the period 1990-95. It is possible that we are observing, over the last decade or so, a “virtuous cycle” in which military spending reductions in some countries lead to cuts in such spending by other countries, as perceived external risks decline.

Table 1.

Military Expenditures, 1990-95 ^1/

Source: VEO database.

^1/Data weighted by country GDP.

^2/For a list of countries contained in each country group, see Appendix Table 2.

Table 1.

Military Expenditures, 1990-95 ^1/

				Change 1990-95
		As percent of GDP		(In billions
		1990	1995	of USS)
All countries		3.6	2.4	−120.7
Industrial countries		3.2	2.4	−14.2
Developing countries		4.9	2.6	−106.5
	Africa	3.0	2.1	−2.8
	Asia	2.8	2.3	15.3
	Middle East and Europe	8.8	7.0	−0.2
	Western Hemisphere	1.3	1.2	6.4
	Countries in transition	7.9	3.0	−125.2
Memorandum:
	IMF program countries	5.1	2.0	−119.4

Source: VEO database.

^1/Data weighted by country GDP.

^2/For a list of countries contained in each country group, see Appendix Table 2.

View Table

Table 1.

Military Expenditures, 1990-95 ^1/

				Change 1990-95
		As percent of GDP		(In billions
		1990	1995	of USS)
All countries		3.6	2.4	−120.7
Industrial countries		3.2	2.4	−14.2
Developing countries		4.9	2.6	−106.5
	Africa	3.0	2.1	−2.8
	Asia	2.8	2.3	15.3
	Middle East and Europe	8.8	7.0	−0.2
	Western Hemisphere	1.3	1.2	6.4
	Countries in transition	7.9	3.0	−125.2
Memorandum:
	IMF program countries	5.1	2.0	−119.4

Source: VEO database.

^1/Data weighted by country GDP.

^2/For a list of countries contained in each country group, see Appendix Table 2.

The results obtained from SIPRI data are qualitatively similar to those from the WEO data. We examine the SIPRI data in two ways—first by including in our sample any country for which data are available for any year during the period 1990-94 and, second, by limiting the sample to those 68 countries for which data are available for all years 1990-94. The first, larger sample, which varies between 73 and 107 countries shows a decline in military spending from 4.2 to 2.7 percent of GDP between 1990 and 1994 (Appendix Table 3), while the second, smaller sample—which excludes the former U.S.S.R.—shows a somewhat smaller decline, from 3.3 percent of GDP to 2.7 percent of GDP (Appendix Table 4), Data from IISS, covering 90 countries show a decline in military spending worldwide from 3.3 percent of GDP to 2.9 percent of GDP for the years 1990-94 (Appendix Table 5). ^11/ Thus, there is unambiguous evidence that worldwide military spending has been declining. ^12/ Furthermore, combining these results with those of previous studies, it appears that, in the decade since 1985, military spending may have declined by as much as 3 percentage points of GDP.

In order to further assess the comparability of data from different sources, an evaluation was made of military expenditures for a common set of 43 countries shared by these different data sources for 1985 and 1992. As indicated in Appendix Table 6, it appears that measured global and regional spending are fairly similar in all three data sources, although there are important deviations for some regions with relatively few countries. For example, all four data sources indicate a decline in worldwide military spending of between 1.0 percent of GDP and 1.3 percent of GDP between 1985 and 1992. For industrial countries, the reduction over the period varies from 1.1 percent of GDP for the GFS data to 1-6 percent of GDP for IISS data. For developing countries, three of the four data sources show a decline over the period, but I1SS data indicate an increase from 3.6 percent of GDP to 4.1 percent of GDP. This difference largely reflects differences in results for the developing countries of the Middle East and Europe. ^13/

The reduction in worldwide military spending evident during 1990-95 implies a large and growing world-wide resource savings–the so-called peace dividend. According to WEO data, if military expenditure as a share of GDP had been maintained at 1990 levels, spending in 1995 would have been some US$345 billion higher than observed, and if the expenditure-to-GDP ratio had continued at its 1985 level, military spending would have been some US$720 billion higher in 1995. The nominal level of military spending in the WEO sample declined by US$121 billion between 1990 and 1995, which reflects almost entirely the decline in spending by transition economies (see Chart 1).

View Full Size

Military Expenditure, 1990-95

(In billions of U.S. dollars)

Citation: IMF Working Papers 1996, 064; 10.5089/9781451848519.001.A001

Source IMF, World Economic Outlook.

• Download Figure
• Download figure as PowerPoint slide

A number of papers, ^14/ indicate that cuts in military spending encourage economic growth by increasing capital formation and by improving resource allocation more generally. This raises the issue of how the peace dividend of 1985-95 has, in fact, been used. It may be the case, for instance, that countries have shifted some of these saved resources into more productive areas of government spending, such as basic social services, or returned the peace dividend to the private sector, through lower deficits or tax cuts. We address this issue in greater detail in Section V.

III. Military Spending by Country Group

Analysis of the WEO data between 19 90 and 1995 suggests that the decline in military spending as a share of GDP has been widespread both geographically and by level of development (Chart 2). ^15/ Note, in particular:

• Military spending by industrial countries declined consistently over the period, from 3.2 percent of GDP in 1990 to 2.4 percent of GDP by 1995, continuing the decline seen since 1985. In nominal terms, industrial countries reduced their military spending by some US$14 billion in 1995 compared with 1990, implying a large real decline.

• Developing countries as a group (including transition economies) cut their military spending nearly in half as a share of GDP, from 4.9 to 2.6 percent between 1990 and 1995. Small low-income countries reduced military spending sharply over the last three years, from 4.2 percent of GDP in 1993 to 3.2 percent of GDP last year. This is particularly important as these poorest countries can least afford to spend scarce resources on the military. ^16/

• The most dramatic decline was seen among the transition economies, including the former U.S.S.R., ^17/ for which military spending declined by nearly 5 percentage points of GDP over 1990-95, from 7.9 percent of GDP to 3 percent of GDP. ^18/ Military expenditures in 1995 were US$125 billion lower than 1990, with more than US$120 billion of this decline coming from the countries of the former U.S.S.R.

• African countries steadily reduced their spending from 3 percent of GDP in 1990 to 2.1 percent of GDP by 1995, implying a savings of about US$3 billion last year. Sub-Saharan Africa—which had increased military spending during 1985-90—reduced such spending sharply during the last five years, from 3.2 to 2 percent of GDP. ^19/

• Asian countries continued their reduction in military spending from the previous period, lowering these expenditures by 0.5 percentage points of GDP between 1990 and 1995, reaching 2.3 percent of GDP by last year. ^20/ However, military spending in the region increased in nominal terms, by US$15 billion, with US$9 billion of that reflecting higher spending by the newly industrializing Asian economies.

• Latin American countries, which spend on average the lowest share of GDP on the military of any region, maintained their level of military spending at 1.2 percent of GDP, implying a rise in nominal expenditures of US$6 billion over 1990-95. ^21/

• Developing countries of the Middle East and Europe, which historically have had the highest levels of military spending, reduced such expenditures by nearly 2 percentage points of GDP over the years 1990-95, reaching 7 percent of GDP by the end of the period. Military spending in the region was virtually unchanged in nominal terms.

View Full Size

Reduction In Military Expenditure, 1990-95

(Percent of GDP)

Citation: IMF Working Papers 1996, 064; 10.5089/9781451848519.001.A001

Source IMF, World Ecomonic Outlook.

• Download Figure
• Download figure as PowerPoint slide

The extent to which military spending reductions are indeed a world-wide phenomenon can be seen by examining the expenditure trends for individual countries. Comparing 1990 with average spending for the years 1991-95, we find that 90 countries (69 percent of the sample) reduced military spending as a share of GDP, by an average of 1.6 percentage points of GDP, while only 25 countries (19 percent) increased military spending, by an average of 0.8 percentage points of GDP (Appendix Table 7). ^22/ Among industrial countries, 17 countries reduced spending, by an average of 0.3 percentage points of GDP, while only one increased spending. Among the transition economies, only one country, which was engaged in armed conflict for much of the period, increased its spending. In Africa, however, the picture is mixed; 13 countries increased military spending while 23 reduced such spending. Interestingly, increases in military spending were relatively more prevalent among heavily indebted countries, for which 6 out of 13 countries in the sample raised spending. ^23/

IV. Fund-Supported Programs and Military Spending

One issue of interest is the experience of countries with Fund-supported programs. While there is no conditionality with respect to military spending in Fund-supported programs, the implementation of a stabilization and structural adjustment program might be expected to lead to changes in a country’s expenditure level and priorities. We first address this issue by examining the experience of those countries with a Fund-supported program in at least one year between 1991 and 1995. It is found that program countries reduced their military spending from 5.1 percent of GDP in 1990 to a low of 2.0 percent of GDP by the end of the period (Appendix Tables 1 and 9). This represents a larger decline than for nonprogram developing countries, which cut their military spending from 4.5 percent of GDP to 3.2 percent of GDP over the same period. ^24/ In addition, 48 program countries (66 percent of sample) experienced a decline in military spending as a share of GDP between 1990 and the average of 1991-95, while only 17 (23 percent) increased their military spending, a similar proportion as that for all developing countries (Appendix Table 7).

Military spending also declined as a share of total government spending for program countries, from 19.7 percent in 1990 to 9.9 percent in 1995, indicating that military spending has assumed a lover priority for governments with Fund-supported programs. This compares with a smaller decline, from 18.9 percent of total spending to 15 percent of total spending for nonprogram developing countries over the same period (Appendix Table 9).

One should, however, exercise caution in interpreting this result; program and nonprogram countries may differ in other systematic ways that might independently influence military spending. Countries in different regions may have responded differently to the need for fiscal consolidation and structural reform. In addition, different types of Fund facilities may have a different impact on fiscal policy, and specifically military spending. Further, the possibility exists that any impact that Fund-supported programs have on a country’s military spending is influenced by the duration of the program. In what follows, we attempt to analyze the country data with these issues in mind.

A closer examination of the data suggest that differences between program and nonprogram countries reflect, in large part, the behavior of countries in transition, 19 of which entered into Fund-supported programs during the period 1991-95, and which generally reduced military spending sharply from relatively high levels (Appendix Tables 9 and 10).

It is also interesting to note that the size of the reduction in military spending is correlated with the initial level of military spending; countries with higher military spending as a share of GDP in 1990 tended to make larger reductions, and this may partly explain the results for program and nonprogram countries. Program countries, other than the transition economies, tended to have substantially lower military spending at the start of the period—2 percent of GDP compared with 4.6 percent of GDP for nonprogram countries—and thus may have had less scope for further reductions. Among countries in transition, the reverse was true; program countries began the period with higher military spending as a share of GDP, and were able to make much sharper cuts over the five years.

Country behavior with respect to military spending appears to differ by type of program. Countries with stand-by arrangements (SBA) or programs supported by the Systemic Transformation Facility (STF) showed a substantially larger decline in military spending as a share of GDP, 3.7 percentage points, than countries either with programs supported by the Structural Adjustment Facility (SAF) or Enhanced Structural Adjustment Facility (ESAF), 1.3 percentage points of GDP, or with extended arrangements (EFF), 0.5 percentage points of GDP. This again reflects the experience of many countries in transition with programs supported by the STF.

The persistence of the impact of Fund programs is addressed by comparing countries with programs in three or more of the years 1991-95 with those countries with programs during two or fewer of those years. It appears that those countries with longer program experience have tended to reduce military spending more sharply, by 3.5 percentage points of GDP compared with 2.2 percentage points of GDP for those countries with shorter experience.

In Section VI, we examine a different aspect of the comparative behavior of program and nonprogram countries, finding that, over a somewhat longer time period than considered above, program countries appear to have relied more heavily than nonprogram countries on cuts in military spending to affect fiscal adjustment.

V. Military and Nonmilitary Spending: Crowding-Out and the Peace Dividend

Sections II and III of this paper provided evidence on the worldwide decline in military spending. In this context, the question arises of how the resources saved through these cutbacks have been used. One might expect that at least some portion of the peace dividend would be used to increase nonmilitary spending, but a part of the savings could be returned to the private sector, either through a reduced fiscal deficit or lower taxes. ^25/ For countries in which military spending has increased, the relevant issue is the extent to which higher military spending crowds out other, perhaps more productive, spending, increases the fiscal deficit, or leads the country to raise taxes.

In this analytical framework, causality is assumed to run from military spending to other spending or taxes. At the same time, however, changes in nonmilitary expenditure levels would likely influence military spending. Thus, one might ask to what extent countries that have reduced overall spending, for example as part of overall fiscal adjustment, have reduced military spending. In addition, exogenous changes, such as in the demand for certain public goods, level of revenue available to a government may simultaneously influence both military and other spending.

These factors suggest that it would be quite difficult to observe directly the links between military and other spending. We examine these links in a preliminary way by looking at the experience, on average, of groups of countries that have increased and decreased military spending as well as individual countries with particularly large increases and decreases. These results should, however, be viewed simply as suggestive of the potential interactions between military spending and other aspects of fiscal policy, which could perhaps be examined more rigorously in the context of an econometric model.

1. Analysis by country group

In Table 2, we compare the experience of those countries increasing and decreasing their military spending over the period 1985-92. ^26/ It appears that many countries undertaking fiscal adjustment during this period looked to cuts in military spending as a primary component of this adjustment. Countries that reduced military spending as a share of GDP also, on average, cut nonmilitary and total spending as a share of GDP. In addition, these countries were able to reduce the central government deficit despite falling revenue-to-GDP ratio, with cuts in military spending averaging 56 percent of the decline in total expenditures.

Table 2.

Changes In Military and Other Expenditure for Selected Countries, 1985-92 ^1/

(Changes in spending as percent of GDP)

Source: WEO database.

^1/Unweighted average, for those countries in WEO sample without missing data.

^2/Change in total revenue less change in total expenditure. Negative sign implies increase in deficit.

^3/Change in military spending of less than. 05 percent of GDP.

Table 2.

Changes In Military and Other Expenditure for Selected Countries, 1985-92 ^1/

(Changes in spending as percent of GDP)

	Expenditure					Total		No. of
	Military	Nonmilitary	Total	Current	Capital	Revenue	Deficit 1/	Countries
Countries increasing military spending	1.1	1.8	2.9	3.2	−0.4	−3.3	−6.2	29
Countries with no change in military spending 1/	—	−0.1	−0.1	0.9	−1.0	0.2	0.3	6
Countries decreasing military spending	−2.0	−1.6	−3.6	−2.1	−1.5	−1.8	1.8	51

Source: WEO database.

^1/Unweighted average, for those countries in WEO sample without missing data.

^2/Change in total revenue less change in total expenditure. Negative sign implies increase in deficit.

^3/Change in military spending of less than. 05 percent of GDP.

View Table

Table 2.

Changes In Military and Other Expenditure for Selected Countries, 1985-92 ^1/

(Changes in spending as percent of GDP)

	Expenditure					Total		No. of
	Military	Nonmilitary	Total	Current	Capital	Revenue	Deficit 1/	Countries
Countries increasing military spending	1.1	1.8	2.9	3.2	−0.4	−3.3	−6.2	29
Countries with no change in military spending 1/	—	−0.1	−0.1	0.9	−1.0	0.2	0.3	6
Countries decreasing military spending	−2.0	−1.6	−3.6	−2.1	−1.5	−1.8	1.8	51

Source: WEO database.

^1/Unweighted average, for those countries in WEO sample without missing data.

^2/Change in total revenue less change in total expenditure. Negative sign implies increase in deficit.

^3/Change in military spending of less than. 05 percent of GDP.

Thus, it appears that reductions in military spending have not been associated with increases in other spending. This is not to deny that there exists a peace dividend, but simply that its use is difficult to observe directly. In fact, If military spending had not declined in these countries, other spending would presumably have had to decline even more than it did, and/or there would have been additional crowding-out of private sector activity, either through higher taxes or increased fiscal deficits.

Countries increasing military spending have behaved in a symmetric way to those cutting military expenditures. Increases in military spending have tended to come in the context of higher nonmilitary and total spending, as well as sharp increases in the deficit. This last point suggests that military spending may have crowded out private sector investment, although a direct empirical investigation of this relationship is beyond the scope of this paper. Also, despite higher total spending, capital spending declined, suggesting that higher military spending may have crowded out public investment as well.

VI. Resilience of Military and Social Spending

The previous section attempted to examine the impact of military spending on other fiscal variables, including total and social spending, by looking at the experiences of countries with large increases and decreases in military spending. We now turn the relationship around, to examine the impact of changes in total expenditure on military and other spending, by focusing on the experiences of those countries with relatively large increases and decreases in total spending. Again, this represents a preliminary analysis of a complex simultaneous relationship between military spending, nonmilltary spending, and a number of other variables. We look at the extent to which military and social spending have been resilient in the face of overall budget cuts, and how military and social spending are affected by increases in overall spending.

This analysis follows DeMasi and Lorie (1989) and Hewitt (1991a). The former find that, during the period 1978-84, military spending in developing countries tended to exhibit resilience during those IMF-supported programs that emphasized fiscal tightening, but that in programs accompanied by fiscal accommodation, the nonmilitary sector tended to be given priority. Hewitt (1991a) finds, similarly, that for the period 1972-88 military spending exhibited an elasticity of less than unity with respect to total government spending, implying that military spending was not resilient during fiscal expansion, but was during periods of fiscal contraction. ^32/ Our results indicate that military spending during the more recent period of 1985-92, has not been resilient either to fiscal tightening or expansion. In particular, countries that implemented fiscal adjustment appear to have largely protected the social sector, while military spending bore a substantial part of the burden of adjustment.

In Table 3, we see that of the 51 countries for which complete data on total, ^33/ military and social spending are available, total spending declined in 29 countries (57 percent) over the period 1985-92. For these countries, military spending declined, on average, by 1.7 percentage points of GDP, contributing some 28 percent of the total budget cut. Military spending declined more than proportionally to total spending, falling by an average of 1.5 percentage points of total spending over the period. This result contrasts with that for social spending, which increased its share of total spending by over 6 percentage points of total spending.

Table 3.

Resilience of Military Spending, 1985-92

^1/Unweighted data.

Sources: WEO database; and GFS.

Table 3.

Resilience of Military Spending, 1985-92

		Change in Expenditure as Share of GDP 1/		Change in Expenditure as Share of Total Spending 1/
						No. of
		Total	Military	Military	Social	Countries
Countries increasing
	total spending	6.3	0.1	−1.9	0.4	22
	IMF Program Countries	3.7	−0.2	−3.0	2.3	8
Countries decreasing
	total spending	−6.1	−1.7	−1.5	6.4	29
	IMF Program Countries	−5.8	−1.7	−2.8	5.4	12
Total		−0.8	−0.9	−1.7	3.8	51
	IMF Program Countries	−1.9	−1.1	−3.0	4.1	20

^1/Unweighted data.

Sources: WEO database; and GFS.

View Table

Table 3.

Resilience of Military Spending, 1985-92

		Change in Expenditure as Share of GDP 1/		Change in Expenditure as Share of Total Spending 1/
						No. of
		Total	Military	Military	Social	Countries
Countries increasing
	total spending	6.3	0.1	−1.9	0.4	22
	IMF Program Countries	3.7	−0.2	−3.0	2.3	8
Countries decreasing
	total spending	−6.1	−1.7	−1.5	6.4	29
	IMF Program Countries	−5.8	−1.7	−2.8	5.4	12
Total		−0.8	−0.9	−1.7	3.8	51
	IMF Program Countries	−1.9	−1.1	−3.0	4.1	20

^1/Unweighted data.

Sources: WEO database; and GFS.

If we limit the analysis to countries with Fund-supported programs, this result is, if anything, strengthened. Of the 19 program countries in the sample, 11 (58 percent) reduced total spending. Of these 11, military spending declined by an average of 2.8 percentage points of total spending, while social spending increased by 5.4 percentage points of total spending.

Our results for countries increasing total spending are qualitatively similar to those in DeMasi and Lorie (1989) and Hewitt (1991a), with increases in spending going disproportionately to nonmilitary and, in particular, social spending. For the 22 countries increasing total expenditures, military spending was virtually unchanged as a share of GDF, so that the share of military spending in total spending declined by almost 2 percentage points. Social spending rose as a share of total spending by 0.4 percentage points. Again, results in Fund program countries were stronger, with military spending declining by 3 percentage points of total spending and social spending rising by 2.3 percentage points of total spending.

In short, during the years 1985-92, military spending became a somewhat lower priority for countries, on average, declining by 1.7 percent of total spending, while social spending increased its average importance in the budget, by 3.8 percentage points of total spending. Countries increasing total spending as a share of GDP only maintained military spending as a share of GDP, while those cutting total spending as a share of GDP tended to reduce military spending relatively sharply.

This was particularly the case for countries with Fund-supported programs. Thus, while our comparison of program and nonprogram countries in Section IV was inconclusive, this analysis of resilience suggests that, over a longer period of time, the existence of a Fund program may have led countries to rely somewhat more heavily on cuts in military spending to implement fiscal adjustment, and to focus more on nonmilitary, and in particular social, spending when increasing total spending. That is, military spending seems to have been less resilient in countries with Fund-supported programs than in those without such programs.

VII. Summary and Conclusions

Our analysis of both WEO and other data sources indicates that the decline in military spending which began in the mid-1980s has continued through 1994/95. This decline, in terms of share of GDP, is broad-based by region and for developing and industrial countries alike. In terms of cuts in nominal spending, however, most of the decline has taken place in the former U.S.S.R. The peace dividend in 1995 was some $345 billion dollars compared with 1990 spending levels and $720 billion compared with 1985.

Results suggest that countries that made sharp cuts in military spending typically also reduced nonmilitary spending as well as the fiscal deficit, thereby potentially encouraging private investment. There is indirect evidence that military spending cuts also allowed countries to maintain or increase social spending in the face of total spending cuts. On the other hand, increases in military spending were associated with increases in other spending and sharp increases in the deficit. Further, higher military spending may have crowded out private investment and, for some countries, public investment. Military spending, in contrast to previous periods, does not appear to have been resilient in the face of spending cuts during 1990-95, bearing a more than proportionate share of the burden of these cuts. Countries with Fund-supported programs made larger reductions in military spending than other developing countries during the period, although this appears to reflect, in large part, the behavior of economies in transition. There is some evidence that, over a longer period of time, military spending has been less resilient in program countries than nonprogram countries.

This paper points to a number of potential areas for future research. A more complete analysis of the use of the peace dividend would require a rather complex structural model, based on explicit behavioral assumptions regarding government decision-making, and linking military spending to other spending as well as revenues. In addition, a more complete understanding of the role of Fund-supported programs in the determination of military spending levels would also benefit from a fully-developed model of government spending behavior, which would allow one to separate the impact of a program per se from other differences among countries.

Finally, the question arises of likely future trends in military spending. This of course will depend on a great many factors, including the future state of world security, that are difficult to predict. Nevertheless, some minimum level of spending for adequate defense is likely to remain necessary for the foreseeable future; as our results suggest, cuts in military spending have tended to be smaller for those countries with lower levels of such spending. Thus, reductions in military spending relative to GDP are unlikely to continue, at least at the same pace witnessed since 1985. Further, the decline in military spending over the last five years reflects, in part, a sharp cut in nominal spending by countries of the former Soviet Union, which are not expected to be repeated. This makes it all the more imperative that the savings from cuts already made be used in an efficient and equitable manner.