|]DEC]ig ifserch fndoings Measuring the peace dividend Will the end of the Cold War and the resulting decline in military spending mean higher economic growth? New research shows that countries that cut military spending do reap long-term benefits-and the bigger the cuts, the greater the growth High levels of military spending are widely Casual empiricism assumed to retard long-run economic RIGOROUS growth. In an insecure region, the argument The view that low levels of military spending THEORETICAL AND goes, countries must devote a disproportion- are associated with strong growth, and vice ate share of their resources to "unproductive" versa, is usually supported by casual empiri- EMPIRICAL military spending. As each country tries to cism. For example, the strict limits on mili- ANALYSES ARE outspend its neighbors to ensure its own tary expenditures imposed on Germany and security, military spending rises higher and Japan after World War 1I-combined with REQUIRED TO higher, yielding no increase-and perhaps the Allies' guarantee of their security- DETERMINE even a decrease-in overall security. High allowed these countries to devote large shares military spending is thought to slow long-run of their total factor endowments to produc- WHETHER MILITARY growth in two ways. First, it may absorb tive capital formation, contributing to their SPENDING S resources that would otherwise have been impressive economic performance during the available for growth-enhancing investments, succeeding five decades. The success of these ECONOMICALLY such as productive capital, education, and countries has led most economists to presume UNPRODUCTIVE market-oriented technological innovation. that, on average, a country with a relatively Second, high military spending may exacer- low ratio of military expenditure to GDP is bate economic distortions that reduce the likely to enjoy relatively strong long-run efficiency of resource allocation, thereby low- growth. ering total factor productivity. Yet not all military spending is unambigu- If high military spending does affect ously counterproductive, or even unproduc- growth in these ways, the converse is also tive, in an economic sense. Military spending likely to be true: sustained cuts in military may benefit the economy by improving spending will lead to higher long-run national security, thereby facilitating growth-the "peace dividend." The end of increased private investment and growth. In the Cold War-and of proxy conflicts in developing countries military training may Asia, Africa, and Latin America-and the improve the level of education and discipline move toward a comprehensive peace in the of the labor force and act as a stabilizing Middle East have stirred fresh interest in influence in society (see, for example, determining whether a peace dividend exists Thompson 1974). Military capital expendi- and how significant it may be. ture can also have productive civilian uses: FROM THE DEVELOPMENT ECONOMICS VICE PRESIDENCY OF THE WORLD-B"NK. NO. 6 MAY 1996 _W- - I many countries continue to benefit from trans- data for a large sample of industrial and devel- portation networks originally constructed for oping countries. military purposes. The results show clearly that military spend- But the question of whether, and to what ing is economically unproductive. When the extent, military spending is economically military spending ratio is added to a growth unproductive cannot be resolved by historical equation that already includes the determinants generalizations and anecdotes; rigorous theoret- suggested by standard theory, the direct effect of ical and empirical analysis are needed. The esti- higher military spending on per capita output mation problems inherent in such analysis are growth is unambiguously negative. The indirect daunting: because increases in growth resulting impact of military spending on economic from cuts in military spending would be expect- growth by way of its negative effect on produc- ed to appear only after a lag of at least several tive investment is also negative-and statistical- years, the beneficial effects of large military ly significant. The model indicates that, all other spending cuts may be hard to disentangle from things being equal, a 10 percent cut in the mili- other factors. These problems have made it dif- tary spending ratio (say, from 10 percent of ficult to measure the size of the peace divi- GDP to 9 percent) would result in an increase in In Eastern Europe dend-or to establish unequivocally that it the per capita GDP growth rate of 0.07 per- and the Middle exists. Uncertainty about the effect of military centage points (say, from 3.00 percent to 3.07 East, where military spending-in particular the lack of robust, percent). quantified estimates of the improvements in liv- Using the estimated parameters on the effect spending ratios ing standards that could result from sustained of military spending on investment and output have been high, lower levels of military spending-makes it dif- growth, the study presents two simulations that cuts in military ficult to persuade national governments that suggest the rough magnitude of the peace divi- military spending cuts will improve people's dend in every region of the world. The first sim- spending could lives-that fewer guns will indeed mean more ulation assumes that over the long run military increase capacity butter. spending is kept at the levels maintained during output in the long the late 1980s (figure 1). The second simulation run by nearly 50 Quantitative answers assumes that as a result of lasting global peace military spending is reduced below those levels. percent "The Peace Dividend: Military Spending Cuts The first simulation indicates that, if sus- and Economic Growth," by Malcolm Knight tained, the improved security conditions and (IMF), Norman Loayza (World Bank), and leveling off of military expenditures would lead Delano Villanueva (IMF), provides quantitative answers to the question of whether cuts in mili- Figure 1. Average military spending as a share of gross domestic product, by region rary spending enhance growth by showing how Percentage of GDP much productive investment is likely to increase 16 in response to cuts in military spending and how 14 H 972-850 much the associated improvements in the effi- 12 ciency of resource allocation will increase long- io run output. The paper extends a standard neo- 8 classical growth model to take account of impor- 6 tant links between military spending, productive 4 investment, and the long-run growth of per 2 0 capita output (box 1). It uses an econometric Latin Sub- Industrial Asia North Middle Eastem technique to consistently estimate the model Arica 5fHn countries Afraca East Europe from a panel of time-series and cross-section Source: Knight, Loayza, and Villanueva 1996. Box 1. Model, data, and methodology The regression model is based on a version of the account data are not available. For the time-series Solow neoclassical growth model, extended in two dimension the data set is divided into three ways. First, in addition to the investment ratio, the nonoverlapping five-year intervals. This approach population growth rate, and the initial level of per provides a simple way of averaging out short-run capita GDP as determinants of per capita GDP cyclical variations in the rate of capacity utilization, growth, the model includes a proxy for human cap- thereby helping to ensure that this variable approx- ital investment based on secondary school enroll- imates long-run output growth. Since panel data are ments, a proxy for international trade restrictions available for most of the variables of interest, it is based on average tariffs on intermediate and capital possible to account for both country-specific and goods, a proxy for the incidence of wars based on time specific unobserved factors. Country-specific the number of years of international wars over the factors are especially important. A number of factors tme period, and a measure of military spending as that are unique to each country (government poli- a share of GDP Second, an investment regression is cies, resource endowments, social institutions, cul- estimated in which the ratio of investment to GDP tural traits) may well be correlated with the regres- is determined by, among other factors, the share of sors considered in the model. Failing to account for GDP devoted to military spending. (The military them would lead to inconsistent estimates of the spending data are from the 1992 Yearbook of the parameters. Country specific factors are controlled The policy Stockholm International Peace Research Institute). for using the methodology proposed by The sample consists of pooled time-series and Chamberlain (1982, 1983), commonly known as implication is cross-sectional data for 79 countries for 1970-85. the 7x-matrix technique, and time-specific factors are The countries come from every region in the world controlled for by removing the time mean from straightforward: except Eastern Europe, where reliable national each variable. fewer guns do indeed imply more to substantial gains in per capita output over the lower military spending ratios would result in a butter long run. The geographic regions that experi- gain in per capita output relative to the baseline enced the largest reductions in military spending level that would have prevailed had military ratios during the late 1980s would eventually spending not been cut of 16 percent in North benefit the most from these cuts (figure 2). The Africa, nearly 14 percent in Asia, and 3.6 per- simulation indicates that in the long run the cent in the Middle East. Even for industrial and Latin American countries, where cuts in military Figure 2. Long-run change in per capita gross spending were more modest during the late domestic product resulting from changes in 1980s, per capita output levels would be 2.0 the military spending ratio percent and 2.9 percent higher than in the base- Percentage change in per capita GDP line paths. By contrast, military spending ratios so | rose in Eastern Europe and Sub-Saharan Africa 40 during the second half of the 1980s. If these 30 higher levels of military spending had been 20 maintained, per capita GDP would have been lo 0 | - - | | lower in the long run by about 5.3 percent in _ | ] * * J | Eastern Europe and 2.3 percent in Sub-Saharan o m Africa. Military spending ratios began to decline IC Latin Sub- Industrial Asia North Middle Eastem in both regions toward the end of the 1980s, America Saharan countries Afrca East Europe r d sts Africa however, and recent data suggest that thiS a Projected impact on GDP of changes in miltary declining trend has continued. spending levels made during the late The second simulation inicates that deeper Projected impact on GDP of cutting military spending *in all regionsto the level n Latin Aeeca cuts in military spending would result in an even Source: Knight, Loayza, and Villanueva 1996. larger peace dividend. An improvement in secu- rity that allowed military spending ratios in all political insecurity has long limited educational regions to fall to the levels that exist in Latin opportunities. America (where no major armed conflicts have The central policy implication of this study is occurred in the past fifty years) would result in straightforward: fewer guns do indeed imply very large long-run gains in capacity output in more butter. In a world in which international most regions (see figure 2). In Eastern Europe tensions are diminishing, reductions in military and the Middle East, where military spending spending should be viewed as attractive struc- ratios have been high, cuts in military spending tural policy elements of macroeconomic pack- could increase capacity output in the long run by ages designed to enhance growth. nearly 50 percent over the levels that would have prevailed if military spending ratios had -by Malcolm Knight remained at 1972-85 levels. Though less spec- and Norman Loayza tacular than in these cases, the peace dividend for other regions would be very large in the long run. Further reading These simulation results may actually under- state the positive effects of enhanced interna- Chamberlain, Gary. 1982. "Multivariate Regression Models tional security on growth. The simulations for Panel Data." Journal ofEconometrics 18: 5-46. explicitly assume that all determinants of . 1983. "Panel Data." In Zvi Grilliches and Michael investment and growth other than military Intrilligator, eds., Handbook of Econometrics, vol. 3. Amsterdam: North-Holland. spending would remain unchanged, even if a generalized peace were achieved; in fact, Hewitt, Daniel. 1993. "Military Expenditures, 1972-1990: The Reasons Behind the Post-1985 Fall in improvements In international security would World Military Spending." IMF Working Paper 93/18. almost certainly result in improvements in the International Monetary Fund, Washington, D.C. other economic variables that affect economic Knight, Malcolm, Norman Loayza, and Delano growth. As political tensions subsided, more Villanueva. 1996. "The Peace Dividend: Military and more countries would be able to dismantle Spending Cuts and Economic Growth." Policy Research barriers to free international exchange of goods, Working Paper 1577. World Bank, Policy Research services, and financial assets. In this way, a gen- Department, Macroeconomics and Growth Division, eralized peace would foster economic interde- Washington, D.C. pendence, more open trading systems, and asso- Landau, Daniel. 1993. "The Economic Impact of Military ciated gains fromspecializa . ImExpenditures." Policy Research Working Paper 1138. ciated gains from specialization. Improved World Bank, Policy Research Department, Public international security would also allow national Economics Division, Washington, D.C. education programs to concentrate on produc- Thompson, Earl A. 1974. "Taxation and National tive skills, and participation in education sys- Defense." Journal ofPolitical Economy 82 (July/August): tems could rise markedly in countries in which 755-82. This DECnote was prepared by Malcolm Knight of the International Monetary Fund and Norman Loayza in the Policy Research Department of the World Bank. DECnotes transmit key research findings to Bank Group managers and staff. They are drawn from the work of individual Bank researchers and do not necessarily represent the views of the World Bank and its member countries-and there- fore should not be attributed to the World Bank or its affiliates. DECnotes are produced by the Research Advisory Staff. We welcome your questions and comments; please e-mail them to the authors or to Evelyn Alfaro, RAD. Prepared for World Bank staff