Military Expenditures 1972-1990
The Reasons Behind the Post-1985 Fall in World Military Spending

World military expenditures have fallen by over 20 percent in proportion to GDP from 1985 to 1990. This study examines the determinants of military expenditures in 125 countries during 1972-90 to ascertain what factors may be behind the recent decreases. Economic decline among developing countries in the 1980s and among industrial countries in the later part of the decade emerges as one possible factor. A second is the move towards more democratic regimes, which could diminish support for the military. A third factor is the improved world security situation and the concomitant decrease in military aid by the former major cold war combatants.


World military expenditures have fallen by over 20 percent in proportion to GDP from 1985 to 1990. This study examines the determinants of military expenditures in 125 countries during 1972-90 to ascertain what factors may be behind the recent decreases. Economic decline among developing countries in the 1980s and among industrial countries in the later part of the decade emerges as one possible factor. A second is the move towards more democratic regimes, which could diminish support for the military. A third factor is the improved world security situation and the concomitant decrease in military aid by the former major cold war combatants.

I. Introduction

Data presented in this paper indicates that a substantial drop in world military expenditures has occurred in recent years. In proportion to GDP, military spending fell by about 23 percent from 1985 to 1990. In 1990, military expenditures are estimated to be about $950 billion for the over 120 countries considered in this study, however, if the same level of expenditures relative to GDP existed in 1990 as in 1986, expenditures would have been $1,225 billion. What is perhaps even more surprising is that this drop in expenditures for the world as a whole exceeds that of the U.S.A. and the Former Soviet Union (FSU), where expenditures fell by less than 20 percent as a proportion of GDP.

Questions remain as to whether this represents the beginning of a sustained downward trend in world military expenditures or a temporary aberration. In order to gain insights into future prospects and to ascertain the likely impact of key variables, a careful examination of past events could serve as a helpful guidepost. Thus, it is useful to examine the determinants of military expenditures in recent years.

This study offers a variety of econometric estimates of the determinants of military expenditures in 125 countries over the period 1972-90. The results are then used as a basis to investigate why military expenditures have fallen in recent years. To a large extent, this paper updates and improves on the analysis in Hewitt (1992). The differences are, first, the analysis is extended by two year, 1989 and 1990, and concentrates on the trends from 1985 to 1990 in more detail. Second, the present study provides estimates for disaggregated groups of countries. Third, the econometric analysis Is improved though introducing corrections for autocorrelation within each country and for heteroskedasticity between countries.

The paper is organized in the following manner. Section II is a discussion of the trends in world military expenditures. Section III discusses the econometric methodology and the main empirical results. Section IV examines factors that may have contributed to the decrease in military spending between 1985 and 1990. Section V concludes. Appendix 1 describes the data; Appendix 2 provides a more detailed description of the econometric model and methodology; and Appendix 3 discusses the disaggregated results. Since Hewitt (1992) offers a reasonably comprehensive review of the literature, this paper omits this section.

II. Trends in World Military Expenditures, 1985-1990

In the aftermath of the cold war, many analyst speculated that military expenditures in the U.S. and the FSU would plummet, while the impact on the rest of the world might go in either direction. In fact, considerable political resistance to decreases in the U.S. and the FSU has stalled major changes. Consequently, military expenditures in both have not lived up to expectations, falling by 15 percent in the U.S. and by 18 percent in the FSU in proportion to GDP. The extent to which military expenditures will fall in the future is uncertain.

Surprisingly, the average decrease in military expenditures in the rest of the world has exceeded that of the U.S. and the FSU. From 1985 to 1990, the world ratio of military expenditures to GDP fell by 23 percent from 5.6 percent of world GDP to 4.3 percent (Table 1). 1/ The reduction was widespread. Almost all regions on average decreased military expenditures significantly (by 10 percent or more). Among the 124 countries for which data was available, a total of 64 countries decreased military expenditures in proportion to GDP, 40 countries did not change military expenditure in proportion to GDP appreciably, and 20 increased military expenditures in proportion to GDP by more than 10 percent (Appendix Table 11).

Table 1.

World Military Expenditure as Percent of GDP, 1985-90

article image
Sources: SIPRI; and staff estimates.

Australia, Austria, Belgium. Canada, Denmark, Finland. France. Federal Republic of Germany, Greece, Ireland, Italy, Japan, Luxembourg, Netherlands, Norway, Portugal, Singapore, Spain, Sweden, Switzerland, United Kingdom, United States.

Bulgaria, Czechoslovakia, German Democratic Republic. Hungary, Poland, Romania. U.S.S.R.. Yugoslavia.

Bangladesh, China, Fiji, India, Indonesia, Korea, Malaysia, Myanmar, Nepal, Pakistan, Sri Lanka, Taiwan Province of China, Thailand.

Bahrain, Cyprus, Iran, Iraq, Israel, Jordan, Kuwait, Lebanon, Oman, Saudi Arabia, Syrian AR, Turkey, United Arab Emirates, Yemen AR, Yemen PDR.

Algeria, Egypt, Libya, Morocco, Tunisia.

Angola, Benin, Botswana, Burkina Faso, Burundi, Cameroon, Central African Republic, Chad, Congo, Cote d’Ivoire, Ethiopia, Gabon, Ghana, Guinea-Bissau, Kenya, Liberia, Madagascar, Malawi, Mali, Mauritania, Mauritus, Mozambique, Niger, Nigeria, Rwanda, Senegal, Sierra Leone, Somalia, South Africa, Sudan, Swaziland, Tanzania, Togo, Uganda, Zaire, Zambia, Zimbabwe.

Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Cuba, Dominican Republic, Ecuador, El Salvador, Guatemala, Guyana, Haiti, Honduras, Jamaica, Mexico, Nicaragua, Panama, Paraguay, Peru, Trinidad and Tobago, Uraguay, Venezuela.

Iran, Kuwait, Oman, Saudi Arabia, Taiwan Province of China, United Arab Emirates.

Argentina, Bolivia, Brazil, Chile, Congo, Costa Rica, Cote d’lvoire, Ecuador, Egypt, Mexico, Morocco, Nicaragua, Nigeria, Peru, Philippines, Venezueal.

Benin, Burkino Faso, Burundi, Cameroon, Central African Republic, Chad, Ethiopia, Ghana, Guinea-Bissau, Guyana, Haiti, Kenya, Liberia, Madagascar, Malawi, Mali, Mauritania, Moxambigue, Myanmar, Nepal, Niger, Rwanda, Sengeal, Sierra Leone. Somalia, Sri Lanka, Sucan, Tanzania, Togo, Uganda, Zaire, Zambia.

Results from Hewitt (1991b).

Total world military expenditures (in 1990 US dollars) rose from $842 billion in 1985 to $941 billion in 1990, remaining virtually stable from 1986 through 1990 (Table 2). If 1990 military expenditures had been 5.6 percent of GDP, they would have reached $1,225 billion. 2/ In contrast, between 1980 and 1985, total world military expenditures increased as a proportion of GDP from 5.2 to 5.6 percent. From 1980 to 1985, 48 countries decreased military expenditures significantly in proportion to GDP, 39 kept expenditures at approximately the same proportion of GDP, and 38 increased military expenditures in proportion to GDP by 10 percent or more.

Table 2.

World Military Expenditure in Billion 1990 Dollars, 1985-90

article image
Sources: SIPRI; and staff estimates.

Military expenditures in the industrial countries fell by 23 percent to 3.4 percent of GDP In 1990, 58 percent of total world military expenditures (Table 3). As percent of central government expenditures (CGE), military expenditures dropped from 15 percent in 1985 to 12 percent in 1990 and the overall level of CGE fell moderately in proportion to GDP. If the industrial countries had maintained the same proportion of military expenditures to GDP in 1990 as existed in 1985, they would have been $715 billion, $165 billion higher than they actually were. Military expenditures in proportion to GDP remained highest in Eastern Europe, 13.1 percent and accounted for 27 percent of total world military expenditures. However, military expenditures in Eastern Europe decreased by 1.5 percentage points of GDP.

Table 3.

World Military Expenditure Shares, 1985-90

article image
Sources: SIPRI; and staff estimates.

Among the developing countries, military expenditures fell as a proportion of GDP from 5 percent in 1985 to 3.8 percent in 1990, continuing a downward trend that started around 1980. If military expenditures in 1990 were at the same proportion of GDP as in 1985, they would have been $175 billion, $40 billion higher. Middle East military expenditures were 8.1 percent of GDP in 1990, 3 percentage points of GDP lower than in 1985. Military expenditures in proportion to GDP were lowest in the Western Hemisphere developing countries (1.8 percent). The Asian developing countries (3.5 percent), North Africa (3.9 percent), and Sub-Saharan Africa (3.2 percent) had ratios of military expenditures to GDP somewhat below the world average in 1990. Net creditor developing countries (6.1 percent) spent considerably more than average, heavily indebted (1.8 percent) and small low-income economies (2.7 percent) spent considerably less.

In relation to central government expenditures, military expenditures fell from 17.8 percent of central government expenditures in 1985 to 14.5 percent in 1990, a 19 percent drop (Table 4). The ratio fell in all but the Western Hemisphere. At the same time, central government expenditures as a share of GDP decreased from 31.3 percent in 1985 to 29.6 percent in 1990 and fell in all regions except Sub-Saharan Africa (Table 5). This pattern indicates a general tendency to reduce the share of resources used by government and at the same time placing an above average burden of adjustment on the military budget.

Table 4.

Military Expenditure in Percent of Central Government Expenditure, 1985-90

article image
Sources: SIPRI; and staff estimates.

Yugoslavia and East Germany are missing.

Central Government Expenditure statisites are not reliable for Iraq, Syria and Yemen.

Table 5.

Central Government Expenditure in Percent of GDP, 1985-90

article image
Sources: SIPRI; and staff estimates.

In the Asian Developing countries, central government expenditures (CGE) fell in proportion to GDP by 15 percent to 19.5 percent in 1990 while at the same time military spending in proportion to CGE fell by 6 percent to 18 percent. Among the North African countries, CGE/GDP fell by 21 percent to 35 percent of GDP while at the same time ME/CGE fell from 13 to 11 percent. Thus in these country groups, there was a general decrease in CGE in proportion to GDP and at the same time military expenditures in proportion to CGE fell; thus military spending did not keep pace with other categories of government expenditure. In Sub-Saharan Africa, CGE/GDP increased by 23 percent, at the same time the proportion of CGE allocated to military expenditures fell by 14 percent. In contrast, among the Western Hemisphere developing countries CGE/GDP fell over 21 percent and ME/CGE rose nearly 9 percent. This is not surprising since the ratio of military spending to both CGE and GDP is the lowest in this region.

The above trends indicate that substantial military expenditure decreases occurred in almost all parts of the world from 1986 to 1990. The decrease in international trade of military goods is even more pronounced (see SIPRI Yearbook). This study attempts to examine whether or not observable economic, financial, political, and geographic characteristics can explain these observed phenomena. An analysis of the determinants of military expenditures can provide insights into the reasons for the fall in expenditures and the factors that are likely to influence future levels.

III. The Determinants of Military Expenditures

1. The demand for military expenditures: econometric methodology

The empirical portion this study examines the determinants of military spending from 1972-90 in 125 countries. The econometric specification employed is based on a public choice model whereby the political leadership of a country is seen as choosing the overall level of government expenditures, and the share to allocate to the military simultaneously. The leadership maximizes its own utility. The leadership’s utility takes into account the preferences of its citizens to varying degrees, depending upon the political institutions in each country. The leadership faces an economy-wide budget constraint, a foreign financing constraint, and a revenue constraint.

The solution to this model is a simultaneous equations system, see Appendix 2 for the full derivation. In the first equation, military expenditures as a ratio of GDP (ME/GDP) is the dependent variable with the explanatory variables of central government expenditure as a ratio of GDP (CGE/GDP), GDP in US$ (purchasing power parity prices), population (POP), the level of foreign financing (FF), geographical variables and political variables.

article image

The expected signs follow from the rationale for the inclusion of the given variables. In general, such variables as central government expenditures and the financial variables are expected to have a positive effect on military expenditures since they increase the overall availability of resources. The impact of GDP and population is uncertain since offsetting tendencies exist. Because of the public good feature, one could expect ME/GDP could fall as GDP and population rise. However, because of the sharing the costs of the military and the fact that larger countries have the possibility of being major regional or global military powers, a positive association could exist.

The geographic variables are included as indicators of cost factors of defending a nation. The political variables are included as rough approximations of the political institutions in different countries, see Appendix 2 and Tables 8 and 9 below for further elaboration. The benchmark category Is a multiparty democracy not engaged in war. Obviously the presence of war will increase the demand for military expenditures. Furthermore it is hypothesized that a country run by a military government will place a higher emphasis on military security.

In the second equation. CGE/GDP is the dependent variable with the determining factors of ME/GDP, a development index (DI), the political variables, and the availability of external financing,


In this equation, military expenditures are expected to have a positive influence. The development index is meant to be a proxy for the ability of the government to raise revenues and therefore is expected to be positive. Availability of foreign financing is expected to have a positive influence for essentially the same reasons.

The specific equations estimated, derived by assuming a Cobb-Douglas utility function for the leadership, are as follows:

article image

The definition of each variable is available in Appendix 1 and the mean values are listed in Appendix Table 13. The various financial variables are included as proxies for the cost of foreign financing. In the first instance, the level of public and publicly guaranteed foreign financing is an indication of the level of concessional financing that a country receives. This level is likely to be highly correlated with the level of foreign grants–unfortunately the World Bank series on grant receipts begin with data for 1980. The dummy financial variables-HD70, HD80, and SLIE-are Included to indicate the probable cost of commercial borrowing to the country. The high debt countries have been unable to borrow in some cases or must pay very high rates. The small low income economies generally have a poor credit rating and are subject to similar financing constraints.

Three different techniques were used to estimate the above equations. In the first instance, Table 6, the three staged least squares (3SLS) method is used without adjusting the data. This method produces results which are almost identical to those obtained in Hewitt (1992) in which two staged least squares (2SLS) was used with a slightly smaller sample. For the second set of results, Table 7, prior to running the regressions the data was transformed via a method suggested in Kelejian and Oates (1981) in order to correct for autocorrelation and heteroskedasticity. 1/ The procedure requires separate data adjustments to each equation. Starting with equation (2A), first an OLS equation is estimated for each country. For those countries with a Durbin-Watson statistic of less than 1.57, it is concluded that autocorrelation is present and the appropriate correction is made. Then for each country, all the variable values are divided by the standard error of the appropriate country equation in order to correct for cross country heteroskedasticity. This was necessary because significant differences in the variance of regression for each country were found. Using the Goldfeld-Quant test (see Sayrs 1989), the ratio of the average variance of regression for one half the sample to the other half has an F Distribution. The ratio is found to be 196.3, which indicates that the variances are unequal with a 99 percent degree of confidence. Finally, the 2SLS method is used to obtain coefficient estimates.

Table 6.

Three Stage Least Square Estimations, 1972-90

article image
Table 7.

Two Stage Least Square Estimations, Adjusted Data, 1972-90

article image

The process is then repeated for equation (2B) in which central government expenditures is the dependent variable, see Appendix 2 for further details. The third set of estimates, Table 12a and 12b, uses the transformed data in a fixed effects framework and on the mean values for each country (see Section V). In each table there are five different estimates. The first is for the entire sample; the second for industrial countries, Eastern Europe and the FSU combined; the third is for developing countries; the developing countries are split between sub-Saharan Africa and other regions of the world in the final two sets of estimates.

2. Empirical results

Econometric specification are well known to have a significant impact on empirical results, Learner (1983). In the present case, the three sets of estimates are presented in order to check the robustness of various results through different econometric specifications. The introduction of the corrections for autocorrelation and heteroskedasticity produces results that are broadly similar, but by no means identical, to the original results. Normally, the corrections will only alter the degree of significance. However, when colinearity exists among the independent variables, it is possible that the signs of the estimated coefficient can switch. The switch in sign is an indication that the results obtained in any one equation are fragile. In some cases where this occurs, the linearity is obvious. For instance, there is clearly a relationship between the two GDP variables and population. Likewise, there is colinearity among the geographic variables. It is less clear why some of the signs of the form of government variables in the central government expenditure equation switch.

In general, the results in Table 7, which uses the 2SLS on the adjusted data, are taken as the primary results for theoretical reasons. If the coefficients are significant In Table 7 and the signs are the basically the same in Table 6, the result will be deemed robust. If however, the coefficients in Table 6 have significant opposite signs, the result will be deemed unreliable. In the in-between cases, for instance when opposite but insignificant signs are found in Table 6, the result In Table 7 will be accepted with reservations. The results in Table 12 are subject to different interpretations, see below.

a. The financial and demographic determinants of military spending

The results confirm that financial variables influence the level of military expenditures across time and across countries. In the first equation, the elasticity of military expenditures with respect to central government expenditures is positive and significant, though considerably less than unity throughout. For the entire sample the elasticity is found to be 0.7; in the subsamples, the range is about 0.5 to 0.8 (Table 7). This implies that increases or decreases in the share of resources allocated to government leads to a somewhat less than proportional change in military spending. This result could be interpreted to imply some degree of resiliency. However, when central government spending rises due to an external or internal shock, such as sharp increases in interest costs experienced in the 1980s, there is no reason that this sort of increase should induce higher military spending. In fact some crowding out of military expenditures would be expected. On the other hand, if central government expenditures increase or decrease due to a general rise or fall in the level of funding, this should have a direct positive impact on military spending.

In the second equation, military spending is found to lead to higher central government expenditures in most cases. However, for the industrial and Eastern European nations the sign varies between equations. Thus, it can be concluded that in the entire sample and In developing countries, increased military spending leads to increased central government spending while in the case of the industrial and Eastern European countries, the relationship is uncertain.

Among the coefficients that show the least stability are those associated with GDP and population, not a surprise since a linear relationship obviously exists between these variables. The level of ME/GDP is clearly effected by GDP level, however, in some the relationship is convex, the coefficient on ln(GDP) is negative and the coefficient on ln(GDP) squared is positive. 1/ In others, a concave relationship is found, military spending is a superior good at low levels of GDP and then becomes an inferior good at high levels of GDP. Since different specifications give differing results, it is more likely that military spending is a normal good, remaining at an approximately constant proportion of GDP throughout. There appears to be somewhat more evidence favoring a concave relationship, see below.

Unlike in the previous study, the coefficient on population is found to be positive and significant in most of the equations in Table 7. In Table 6, the population coefficient is positive and significant in two cases and insignificant in the other three cases. The positive elasticity with respect to population provides some insight into the motivation for military expenditures. There are two competing theories as to how military expenditures might vary across countries of different size and wealth. One theory emphasizes the defensive nature of military expenditures. Since security is a pure public good, each nation might have a threshold level of military expenditures that it deems necessary to achieve a minimum level of protection. This hypothesis predicts that military expenditures would not rise with GDP beyond a certain point and would be insensitive to population size and therefore is rejected by the findings. Military spending is found to grow with population and GDP. There are two possible explanations for why larger, higher income countries spend more. First, it may be nearly impossible for the small nations to reach the threshold level. Alternatively, since a larger population enables a country to field a larger army, population will enhance a country’s military potential. Larger nations could gain more from military expenditures because of the prospect of becoming a regional or global military power.

The elasticity of military spending with respect to per capita income can be derived from the coefficients on GDP and population. In Table 7, the range over the entire sample is between 0.6 for low income counties to 0.9 for high income countries. Thus, the elasticity is clearly less than unity. As per capita income rises, military spending also is found to rise, but by a lower percentage.

The econometric results associated with the other financial variables is somewhat different from the previous study. The net flow of public and publicly guaranteed foreign financing (PGFF) has positive coefficients in Table 7 which are not significant, while in Table 6 some of the coefficients are negative. This implies that PGFF does not directly effect the ratio of military expenditures to GDP. However, in the central government expenditure equation, PGFF is found to be positive and significant for all equations except Industrial and Eastern European countries. Thus, PGFF is still found to have an indirect positive effect on ME/GDP through increasing CGE/GDP. 1/

The high debt countries are found to spend less on the military than other countries and to have cut back military spending in the 1980s relative to the 1970s. This relationship holds for the entire sample and for the subsamples. In the second equation, high debt countries definitely had lower CGE/GDP than others in the 1970s and their CGE/GDP in the 1980s was definitely higher than In the 1970s. However, since the coefficients vary in different equations, it is uncertain as to how their CGE compared to others in the 1980s. In general, it appears that because of the higher interest cost that these countries had to pay, their central government expenditures rose in the 1980s to a level that approximates other countries.

The small low Income economies allocated a smaller share of GDP to the military than other countries. The results are mixed as to how their CGE/GDP stood in relationship to other countries.

In the central government expenditures equation, the development index (DI) is used as a proxy for the level of development, see Appendix 1 for the definition. For the entire sample, a positive association between DI and CGE is found. This was expected since higher development is generally associated with a greater ability to raise revenues and thus CGE tends to be higher. In the disaggregated results, this continues to apply to the industrial and Eastern European countries. However, the opposite sign emerges among the developing countries, thus, developing countries with higher development levels have lower CGE/GDP. A possible explanation for this could be reverse causality–numerous studies have found that higher CGE deters growth.

b. The political and geographic variables

The profile of the effect of the political variables on military spending is almost identical to that found in Hewitt (1992). The benchmark is a democratic government not engaged in conflict. All the coefficients associated with political variables are positive and therefore represent various tendencies to spend more on the military. The most significant coefficients are associated with international war and civil war. Next, in order of importance are monarchies, other forms of government, military governments, and socialist government in 1972-1985, and socialist governments in 1986-1990. socialist governments in 1986-1990 is a variable introduced to account for the possible effects of political changes in Eastern Europe, the FSU, and elsewhere since 1985. Indeed, it appears that these countries lowered their military spending somewhat in relation to other countries during this time period, but still spent more than democracies.

The effect of form of government on CGE/GDP varies somewhat between equations. Countries involved in civil war spent less than democratic governments, monarchies spent less, and socialist governments in the period 1973-1985 spent more. In most cases the countries with socialist governments prior to 1985 had a significantly lower CGE/GDP ratio in 1986-90. The signs on countries at war, military governments, and other governments vary across specifications leaving an uncertain conclusion.

The above results suggest that the leadership of nondemocratic countries tend to have a higher preference for military expenditures than the population at large. Presumably, the policies of countries ruled by democracy reflect the more closely the preferences of the population at large, while with other regimes the policies are more likely to reflect the preferences of the leadership. Therefore, a move towards democratic rule, or possibly more open forms of government, is likely to induce lower military spending.

The geographic variables display a somewhat unexpected pattern. The maintained hypothesis is that the geographic characteristics of a country would have an effect on the cost of defending a country and therefore on military spending. In Table 7, the strongest result is that a larger coastline induces greater military spending, and this conclusion is not contradicted by the results in Table 6 or 12. The same holds for land borders, but the results are slightly weaker. The evidence on the effect of land area is uncertain.

IV. Reasons Behind the Post-1985 Fall in Military Expenditures

The preceding econometric analysis suggests a number of factors that could explain the post-1985 decrease in military expenditures. Changes in economic and financial factors that effect the affordability of military expenditures is one possible explanation. A second explanation is changes in the priority attached to military spending–brought about either by a change in government or a shift in external security factors.

1. Financial/economic factors

The econometric analysis suggests that a change in the level of financial resources available to a country could alter the level of military spending and its ratio to GDP. Among the industrial countries, an economic slowdown is evident beginning in 1988 with a drop in the growth of real GDP from 4.3 percent in 1988 to an average of 2.4 percent, Table 8. Six of the G-7 countries experienced a significant fall in their rate of growth. 1/ Eastern Europe and the FSU also experienced a great decline in rates of growth. Among the developing countries, slow growth occurred in Sub-Saharan Africa, the Middle East, and the Western Hemisphere developing countries. These regions experienced negative real per capita GDP growth during 1985-90. Growth remained fairly brisk among the Asian developing countries.

Table 8.

Growth of Real GOP, 1985-90

(In percent)

article image
Sources: International Monetary Fund, World Economic Outlook.

Further evidence of the impact of a slowdown in growth on military spending is suggested from comparing military expenditures during 1985-1990 to movements in purchasing power parity per capita GDP (PPP). A strong correlation is found between ME/GDP and changes in the growth of PPP. 2/

2. Political changes and security factors

A great deal of political change occurred in the later part of the 1980s, In addition to the well publicized changes in the former socialist countries, many developing countries moved towards more open political institutions. Table 9 provides a summary listing of the distribution of countries between the political variables, see Appendix 1 for a definition of each category. 3/ From 1983 to 1989, in addition to the socialist countries, eleven countries moved between political categories. While some of the changes were offsetting, a net increase in eight democracies occurred while there were seven less military governments. Furthermore, immediately after the time period covered in this study, the pace of change accelerated. From 1990 to 1992, at least 15 countries switched political categories. 1/ Although such changes did not have time to effect the political variables in the econometric analysis, it is quite likely that the events leading up to these changes had an influence on government expenditure policies prior to the actual change.

Table 9.

Summary Statistics of Political Variables

article image
Sources: Sivard (1990), SIPRI, Europa World Yearbook.

Table 10 is a cross-tabulation of the political variables and changes in military spending from 1985 to 1990. The first column shows the number of countries in each category that reduced their military spending in proportion to GDP by more than 10 percent, column 2 lists the number of countries that did not change military spending as a proportion of GDP by more than 10 percent, and in column 3 the number of countries that increased military spending by more than 10 percent. In order to construct a summary statistic, the “rate”, column 4, is obtained by subtracting column 1 from column 3 and dividing by the total number of countries in the category.

Table 10.

Interaction Between Political Variables and Military Expenditures

article image

The no change designation means that the ratio of military expenditures to GDP increased or decreased by less than 10 percent.

Countries that were socialist prior to 1985.

Countires that changed categories from 1983-1989.

Countires that introduced democratic reforms from 1990-1992 or are in the process of introducing democratic reforms.

For the entire sample, the rating is negative 35 percent, which means that considerably more countries lowered their military expenditures as a ratio of GDP than raised them. All the categories of countries have a negative rating, the categories which had the greatest change include International war, democracies, and socialist governments, and post-1990 transition countries; their ratings ranged between negative 40 to 63 percent. Security improvements are likely to account for decreased military spending by countries at war, as well as decreased military aid resulting from the cessation of the cold war. The improved security environment brought about by the end of the cold war is also the likely explanation for the decrease in military expenditures among the democratic countries. Alternatively, the decreased military expenditures in the socialist/ transition economies probably represents a reaction to internal political changes. The post-1990 transition countries probably decreased military spending in reaction to internal political changes that preceded the change in regimes.

The rating for monarchies, military governments, countries engaged in civil war, and others forms of government varied between negative 14 to negative 27 percent. Thus these countries changed relatively less than average. The changes that did occur appear to reflect both internal political changes and external security improvements. Finally, countries that underwent a major political transition prior to 1990 had a rating of negative 27 percent, thus the number of countries that cut military spending was slightly below average. The reason for this relatively mixed reaction is in part that in a number of cases the change in political category occurred because of a deterioration in the political situation, such as the outbreak of civil war or a military overthrow of the government.

An additional security related financial factor is the level of military assistance. Clearly, both military and economic assistance from the FSU has fallen dramatically. Although no exact figures are available, the movement in arms exports, for which a large proportion are effectively aid financed, provides an indication. According to SIPRI estimates, U.S.S.R. arms exports fell from nearly $18 billion in 1987 to $4 billion in 1991. Furthermore, U.S. miliary assistance has fallen from $ 5.8 billion in 1985 to $ 4.8 billion in 1989.

V. Conclusion

A sizeable drop in military expenditures has occurred since 1985, the ratio of military expenditures to GDP has fallen by 23 percent. The foregoing analysis of the determinants of military expenditures provides some explanation for this observed drop. First, financial factors are found to have a significant effect on military spending. It Is likely that the poor growth performance in many developing countries in the 1980s contributed to a decrease in military spending in proportion to GDP. Furthermore, the economic decline in the late 1980s among industrialized nations could have contributed to a decline in military expenditures.

The second major category influencing military expenditures is the form of government. It can be surmised that the profound political changes that the world has witnessed in numerous countries in the recent past Is In part responsible for the fall in military spending. For instance, there is evidence that the political changes in the former socialist countries have induced lower spending. Furthermore, the evidence suggests that the moves towards more democratic forms of governments in developing countries is likely to have a retarding effect on military spending.

The final factor that could explain the drop in military spending is the improved security environment in the world. Although this analysis does not directly address this issue, there are indirect indications. The drop in military spending among democracies and countries at war is probably a result of a decrease in military tensions. Military assistance by the FSU fell as a result of the cessation of the cold war and internal economic problems. Additionally, the U.S. has decreased its military aid. It is likely other countries also cut back on military assistance. This study suggest that such factors would exert a strong downward effect on military spending.