A Manuscript Submitted to the University of Michigan Political Science Program in 1993 in consideration for an "honors" classification on the bachelors degree.

By Chetly Zarko, 
Copyright and all rights, 1993-2005.
Project complete April 1993. 

Due to unusual characters in the early (1993) MSW version, this version may contain inaccuracies.
Bibliography.
Supporting graphs (no longer available).
Main Text, at Chapter 1.

Table of Contents

Chapter 1 Introduction
Chapter 2 Review of the Literature
Chapter 3 The Development Context
Chapter 4 A Model of Military Expenditure
Chapter 5 Testing the Model
Chapter 6 Conclusions and Policy Implications
Footnotes

INTRODUCTION
Chapter 1

Every political good carried to the extreme must be productive of evil. Mary Wollstonecraft, The French Revolution, 1794

Just as the French Revolution taught us about the problems of excessive 'justice,' the second Soviet revolution nearly two-hundred years later has taught us about the problems resulting from excessive commitments of resources to the military.  It can be argued that the surprising collapse of the Soviet Union was precipitated, at least in part, by the excessive spending and the extremes of its military.

The collapse of the Soviet Union and end of the Cold War have heralded a new era in which dramatic cutbacks in superpower military expenditure seem certain.  The pace at which events have proceeded has left many optimistic about the future of international arms control.  Even in the wake of such historic changes however, cuts in military expenditure have come more slowly than expected.  Changes in the military expenditure patterns of most Third World states are even harder to discern than in the United States, the European Community, and the former Soviet Union.  In some regions, such as the Middle-East and Latin America, there is little sign that the three decade trend of growth in military outlays will be reversed. (1)

Ironically, as a result of the demobilization by the superpowers, there may be a temptation in the Third World to increase military outlays. This could be a political consequence of the declining hegemonic power of both the United States and the former Soviet states.  Third World states may perceive their military expenditure as proportionally more significant relative to the superpowers, which are declining rapidly, and therefore perceive themselves as more capable of challenging the status quo or pursuing their own hegemonic interests. (2) It is in this context that understanding and predicting the effects of military expenditure on Third World states will become increasingly important.

Unfortunately, at least for the Third World, most of the theoretical and empirical research exploring the way military expenditure  affects political and economic development has looked at either the First and Second World states. (3)  Even when focusing on the Third World, discussion and research have been framed in the context of the strategic realities of the Cold War, rather than the economic and political realities of development.
 
Furthermore, the little work that has dealt with the effects of military expenditure on the development process has often been dominated by purely econometric analysis, to the exclusion of attempting to explain the otherwise equally important political effects that can have a bearing the economy.  Military expenditure may open or close some of these political channels of development.  Although in the past it was notoriously difficult to quantify such political linkages, some excellent recent efforts could potentially be applied to the military expenditure debate. (4)

More importantly, the empirical results of the many econometric studies conducted to date are highly contradictory and divergent. Benoit's seminal study in 1974 indicated that military expenditure actually improves economic performance, while a host of other proceeding studies leaned towards the conclusion that it has no significant effect or a negative effect.  It is the variation in these research findings which pose the greatest puzzle to researchers in the field.  From the apparent patterns in this variation, I will propose a model that I believe largely explains why previous studies have diverged in their results.

Since the variation in previous research results is relatively ordered and has a discernible pattern, one way of deriving at least an informal understanding of the military expenditure relationship with economic growth is by reviewing the literature. Chapter 2 begins therefore, as the logical first step of this paper, by examining the literature which attempts to empirically determine the impact of military expenditure on economic growth.  Chapter 3 attempts to frame that debate within the larger context of both economic and political development theory.  These chapters will be the largest part of the paper since they form the basis for the hypothesis I will suggest.

In Chapter 4, I shall propose an informal theoretical model of the underlying causes of economic growth and contraction that result from military expenditure.  This model will then form the basis for some crude tests (5) that should determine whether further research is warranted.  Theresults of those tests will be detailed in Chapter 5. The model was derived from my observations of the structure of variation among the various studies in the field and my understanding of the development process. Chapter 6 will suggest some of the conclusions and resulting policy implications.

The purpose of this paper is primarily to propose some important directions for research, and secondarily to offer some evidence that might support the conclusion that further, more detailed research is warranted. It must be noted that I do not seek to demonstrate that any particular factor dominates as a cause of economic growth or retardation, but only that some possible explanations may have been overlooked or paid little attention.  Further, although I believe that the model I will propose is compelling based on the results of previous literature, the data manipulations I have used to test the model with suffer from many, if not more, of the defects that previous studies have been criticized for.  A formal multivariate econometric model would be necessary, but is impossible within the constraints of my current resources.  Therefore only simple statistical bi-variate tests will be conducted.  This obviously means that one shouldn't take the results too seriously.

The basic thrust of the model is that the variation in the economic effects of military expenditure is dependent upon the level of development of the state in question.  The hypothesized relationship is curvilinear with respect to development, as depicted in the chart below. (6)

In the very least developed LDCs (primarily agricultural states like those in sub-Saharan Africa), military expenditure has a decidedly negative effect on growth rates.  In moderately developed LDCs (primarily Argentina, Brazil, India, Korea, and other similar industrializing states), military spending may have a 'more positive' effect (see footnote 5).  Finally, the curve turns back downward in the advanced industrial states (OECD states).

What causes this curvilinear relationship?  What factors might exaggerate or flatten the curve?  Since there are many similar curvilinear relationships that have been verified by development theorists, an immediate solution comes to mind.  The relationship between military expenditure and economic growth is non-existent and in reality spurious. Another possibility is that there exists an intervening or confounding variable (or set of reinforcing variables) which affects both the development process and the economic outcome of military expenditure.

Given the assumption that their are other relevant variables, three additional alternatives theories exist.  Either the militaryexpenditure relationship is purely spurious and there is no causal effect on economic growth; external variables change the way military expenditure affects economic growth; or different 'kinds' of military expenditure (e.g. labor as opposed to capital intensive militaries, arms production as opposed to importation, or investment versus consumption oriented militaries) change the way external variables affect economic growth. Additionally complicating the matter is the fact that the latter two possibilities would imply the possibility of feedback effects, as well.

In essence, military expenditure would have a magnifying effect on other variables (or the other variables would have a magnifying effect on military spending) that affect the development process Such a magnifying effect would reinforce growth or slow it further, depending on the stage of development and the pre-existing conditions that drive an economy.  The relevant factors could be determined by qualitative aspects of the expenditure itself, or by the presence or absence of an alternative intervening variable. (7)

Likely candidates for intervening or confounding variables include differences in savings-investment patterns, capacity utilization and employment, labor-capital ratios, and the impact of arms imports and exports.  Although the many empirical findings that there is no significant correlation among these variables and growth would at first glance support a purely spurious relationship; they do not disprove other multiple-variable models.  It is possible, even likely, as I will argue, that the mix of relevant variables would usually counteract each other so as to produce the appearance that there is no significant relationship.

A properly specified model would be able to highlight the effects of different individual component variables on the outcome.  In the real world, since all of the involved variables are known to have feedback effects on each other in general, and since military expenditure can have an effect on each variable, it could be argued that a purely spurious alternative is the most unlikely.

Regardless, it would be important to determine which variables played the most significant role in the military-economic-development process so that one might most appropriately match policy to them.  Since military expenditure itself can be broken down in component qualitative aspects, such as the ratio of soldiers to money spent (somewhat analogous to a capital to labor ratio in the civilian economy), it becomes reasonable to postulate that the military and civilian sectors should break down in a similar way. Even then, both sectors would have to matches the levels (of labor to capital inputs)  that would be most "most appropriate" for a state at a given stage of development, then it is most likely to have a positive impact (or pessimistically, the least negative impact) on the economy.

If, on the other hand, the military expenditure becomes disjointed to the 'appropriate' levels of each factor input, then it would be more likely to have a negative effect.  In essence, the qualitative inputs of military expenditure should be similar to the 'appropriate' inputs in the rest of the civilian economy, otherwise bottlenecks and resource-tradeoffs will occur.

The conclusion that the effects of military expenditure are curvilinearly correlated with growth would seem to be born out by intuition as well.  It would certainly seem that some minimum level of military expenditure may be necessary to preserve social order and forprotection from external threats, and that are such services are required preconditions for economic growth. On the other hand, although militaryexpenditure is obviously an important political good, it should also be obvious that a state can carry this 'good' out to an extreme, resulting in tradeoffs with other essential developmental inputs and therefore negative economic results.

REVIEWING THE LITERATURE
Chapter 2
PROGRESS AND FAILURE IN UNDERSTANDING THE
EFFECTS OF MILITARY EXPENDITURE ON ECONOMIC GROWTH IN THE THIRD WORLD

Research in the field has expanded rapidly since Emile Benoit's pioneering study in 1973, (8) which surprised the research community with its conclusion that military expenditure actually increases economic growth in LDCs.  Until Benoit's seminal study, it was generally assumed that military purchases trade off with civilian growth in a simple 'guns for butter' fashion.  This assumption led to research focusing on the political and strategic aspects of military expenditure to the exclusion of economic factors.  Although Benoit's results are almost unanimously rejected today, the re-examination of economic-military linkages that his study forced will be his enduring contribution.

Benoit's study utilized a cross-sectional sampling of 44 LDCs between 1950 and 1965 to determine the relationship between military expenditures and 'growth rates, investment rates, foreign aid receipts, and certain other variables.' (9) Using correlation results and regression analysis to arrive at his initial conclusions, Benoit showed that military expenditure was positively and significantly correlated with economic growth. (10)

Benoit was genuinely surprised by these results since his model had predicted that a one percent increase in military expenditure would cause a .25 percent decrease in economic growth.  Since his model included only negative linkages between the two variables, he speculated that the negative effects predicted by the model must be more than offset by some unknown or unmeasured positive contributions of military programs.  He suggests the military might succeed in 'feeding, clothing, and housing a number of people who would otherwise have to be fed, housed, and clothed by the civilian economy....  Military forces also engage in certain R & D and production activities which diffuse skills to the civilian economy and engage in or finance self-help projects producing certain manufactured items for combined civilian and military use which might not be economically produced solely for civilian demand.' (11) Benoit's description, though insightful, is flawed because, as he freely admits, "most of the evidence for the growth-stimulative benefits of defense programs is imprecise, anecdotal, and difficult to evaluate (Benoit, 1978:  277)." In fact, the large amount of research since Benoit has been focused on identifying and measuring these effects as well as improving on the relatively simple research design Benoit relied on.

Benoit also logically questioned whether the military expenditure was causing the economic growth, or whether it was the economic growth which was allowing for greater military spending.  Wayne Joerding conclusively solved this problem in 1986 by applying a Granger Causality test to several studies. (12) The test, which analyzes the directionality of the impact of several variables, concluded that the only way to explain the data for several studies would be for military expenditure to precede economic growth in the system. (13)
       
Among the first and most obvious improvements to be made to Benoit's work were those of Basudeb Biswas and Rati Ram.  They realized that Benoit's 'sample seemed to be biased against poorer LDCs,' (14) and that 'of the total sample of 44, only seven states would probably be classified as low- income LDCs, although World Bank documents indicate that about 40 percent of LDCs fall in the low-income category.' (15) By adding several least developed LDCs and then dividing the new group into separate subgroups, Biswas and Ram were able to identify a significant difference between the two groups.  In the least developed LDCs, the correlation between military spending and growth is consistently negative.

Further corroboration of this split between rich and poorer LDCs is offered by the findings of Kwabena Gyimah-Brempong (16) and Oumar Nabe, (17) which strongly support this division of LDCs into sub-groups.  Both Nabe and Gyimah-Brempong's studies focus on the impact of military spending in the least developed states of Africa.  Each finds that growth is stunted as a result of increased military burden.  The studies that focus on the poorest states in the world contrast sharply with Benoit's findings over a more affluent sample of states where improved economic performance results from high military burdens.

Perhaps the most illustrative study of this nature was done by Lance Taylor in 1981. (18) Taylor did a cross-sectional analysis of sixty-nine states and found a negative relationship between military expenditure and economic growth when the data was aggregated.  When the data was disaggregated state by state in multiple time-series analyses, however, some interesting results occur.  In 72 percent, or 50 cases, no significant relationship was discernible; in 14 cases, a clear negative relationship was identified; and in 5 cases a positive relationship was identified.

Taylor used the 72 percent figure for no significant relationship to justify his overall conclusion that there was no significant relationship between military expenditure and economic growth.  Taylor seemed to miss an opportunity to explain at a greater level of detail what was going on by aggregating the figures even after he had done the work to disaggregate them and run multiple independent time-series analyses.  The fact that nearly 30 percent of the states did show a significant relationship in time-series analysis, which offer far greater detail and accuracy than aggregated cross-sectional designs (at the expense of additional complexity), indicates that something curious is going on, at least in certain states.

As was argued in the introduction, multiple interpretations of these kinds of data are often possible.  Taylor could be correct in suggesting that there is no significant relationship, but there also exists the possibility that multiple variables have offsetting effects which makes it difficult to measure significant aggregate effects.

When two or more variables act in opposite directions, the cancellation effect would make results appear near zero, and therefore no different than truly non-significant results.  It would be when the variables reinforce rather than offset, that we would see the cases of actual significant positive and negative relationships.  The research puzzle should then become a question of when do the variables reinforce and why?

As we shall see later, there do exist several variables which simultaneously affect the economic outcome of military expenditure.  This, coupled with the partial success of some mutli-variate models, is evidence that the relationship is truly spurious or insignificant as some would argue; only more complex.

The conclusion at this point then, is that there exists some variable that changes as development occurs which alters the way military expenditure effects economic growth.  Identifying exactly what that variable is would give economists and policy analysts a powerful tool for predicting and explaining the impact of the military on the economy.

By 1986, Saadet Deger had formulated a detailed study aimed at identifying specific factors involved in the relationship between military spending and growth.  Deger argues that military burden, savings and investment, and overall growth are interdependent factors and require a "simultaneous-equations system based on a properly specified theory." (19) This is a significant advance in methodology over Benoit and others in that it identifies a broader range of variables and examines a broader spectrum of causality directions.

This was also the first major study to compare supply and demand effects.  Deger argues that one possible source of growth is the creation of additional aggregate demand. This is important because if aggregate demand is a determining intervening variable for economic growth, then one could explain why lower income countries differ from moderate income LDCs. Deger argues:

In Deger's estimation, one possible avenue for growth would be simple Keynesian demand stimulation.  In an economy that had excess capacity as a result of high productivity and low consumer demand, (21) the government could be a useful vehicle for filling the demand gap.

Deger proceeds with the simultaneous-equations model including all three variables. Her conclusion is that the growth equation, independent of savings, is consistent with Benoit's findings that military spending does accelerate economic growth through aggregate demand stimulation. However, when this effect is combined with the negative crowd-out effect on aggregate savings (which results from a resource tradeoff with the civilian economy) in a simultaneous model identifying all three variables, the sum of the positive and negative impacts yields an overall negative effect on growth.

Although Deger accurately identifies the need for a simultaneous regression of all of the variables independently, a key problem with the study is her failure to divide the sample into distinguishable sub-groups; or, at minimum, to attempt to use the involved variables to explain differences among individual countries.  From the data she presents, isolation of the effects of level of development, surplus productive capacity, or differing labor intensities is impossible.  Finally, even though Deger realizes and suggests that the presence of surplus productive capacity would promote economic growth (through a Keynesian stimulation), she fails to include any specific variables in her model which might account for differing capacity utilization.

Analyzing the supply side effects of military spending, Robert Looney argues that the existence of an indigenous arms production industry positively influences the effect of military spending on economic growth relationship.  Looney finds that when producers and non-producers are separated, the effects of military spending on consumption and investment are quite different.  In producing states, increased consumption is accompanied by a commensurate increase in investment. The opposite is true of investment in non-producers.  It is possible that greater arms spending in producing countries stimulates demand specifically in those heavier industries which require greater numbers "of relatively skilled and managerial workers at high incomes." (22) A multiplier effect occurs as these newly employed workers make purchases that give a secondary boost in demand to other sectors of the economy.  Furthermore, acquisition of a domestic arms industry represents a form of investment by the government in expanding productive capacity and in increasing the skilled labor pool through training.  It may be this increase in productive ability that at least partially offsets any inflationary consequences of increased demand. In non- producing states, decreased investment and increased consumption are a natural result of the fact that such weapons must be purchased from abroad.  Capital that would otherwise be used in the civilian economy for investment is lost forever.

Looney's analysis also extends to the possible inflationary impacts of military spending.  Looney suggests two possible sources of greater inflation in an arms producing state.  First, military spending could result in cost-push inflation (inflation resulting from insufficient supply inputs) because the military bureaucracy continues to reward contractors that sustain substantial cost over-runs.  Second, demand related inflation (inflation resulting from excessive demand) could result in an economy, already operating at full capacity, from the increased aggregate demand associated with increased government spending. (23) Cost-push inflation could occur only in the producing states while demand related inflation would occur in either economy only if it was at full capacity.

Empirically, producers do not experience significant inflationary pressures while non-producers do.  Seeing inflation only in non-producers supports a capacity driven model and therefore the "suggestion that 'defense' spending encourages fuller utilization of the existing productive facilities" for producers.  Non-producers "are likely to be more constrained by supply." 24 Looney's final regression analysis shows that military spending restricts growth in non-producing states while enhancing it in producing states.

Although Looney's study does not account for the variables simultaneously, it goes a long way toward demonstrating that non-producers and producers are affected differently by capacity and resource constraints.  Since indigenous arms production typically coincides with higher levels of economic development and occurs in economies with more idle capacity, it should come as no surprise that the net effect is to promote growth in moderately developed LDCs.

Finally, Erich Weede contributes to the literature by focusing on the impact of military-participation ratios (25), rather than total military expenditures.  Weede justifies this focus by arguing that "military participation ratios are indicators of some kind of human capital formation, and one should certainly control for more obvious or conventional human capital formation as well." 54 Essentially, Weede attempts to identify the amount and effect of skilled labor formation as a result of education and training.  Weede's results show that, controlling for total military expenditure, higher military participation ratios are positively correlated with higher levels of economic growth.  Korean and Taiwanese ratios were between four and five times as high as Brazilian and Peruvian ratios, respectively.  Such results indicate that the best military policies are labor intensive rather than capital intensive.

Among Weede's major methodological advances, which have not been duplicated for other any of the other studies to date, is the fact that he uses a polynomial equation to control for the natural curvilinear developmental pattern. (26) As we shall see in Chapter 3, moderately developed states have so-called 'advantages of backwardness' which creates a natural tendency for them to grow faster than either the most advanced or least developed economies.  Factoring this background relationship into the equation is extremely important, and it is this developmental relationship which leads me to believe that the causes behind the military's impact on the economy shouldn't be much different.

Weede even went so far as to suggest a mechanism by which varying capital-intensities and savings and investment effects might be linked.He suggested that 'low interest rates provide an incentive for capital-intensive rather than labor-intensive industrialization.' (27) Given the link between low interest rates and higher savings-investment, it is conceivable that both Weede and Deger could be correct.  Low interest rates might drive both the savings related growth that Deger identifies while simultaneously changes the labor-utilization ratios, which Weede focuses upon.  Weede also suggests that government revenues might change during the course of development and therefore have an effect on both military expenditure and growth (see footnote 4 and Chapter 3).

Although Weede does an excellent job at identifying these additional possibilities, the particular scope of his research was unable to examine them empirically. Furthermore, although Weede's introduction of labor participation ratios fits nicely into a general developmental framework, his variables aren't compared to any of the other variables in the developmental process to determine if the relationship is spurious.

The model that follows predicts that if Weede's study were expanded, one could show that military participation ratios are or should be highest in moderately developed LDCs and are correlated with modestly large skilled labor pools and high idle capacity.  These changes should closely parallel the changes endemic to the 'normal' developmental process, and if they don't then negative effects would most likely result. Those militaries that have higher military labor participation ratios or have ratios that most closely match the ratio for the general economy, will have the most positive effects.

MILITARY EXPENDITURE IN THE DEVELOPMENT CONTEXT
Chapter 3

Central to any discussion of economic or political development theory is the demographic transition theory.  The central thrust behind the demographic transition theory are empirical facts.  The theory was originally (28) conceived to explain the series of developmental changes witnessed in Europe in the late 19th and early 20th centuries.

Quite simply, the theory postulates that there are three stages of fertility and mortality levels, each of which society must progress through as it develops from an agricultural (29) to industrial to post industrial society (see footnote graphs). (30) The first stage, in which society is almost 100 percent agricultural in orientation, is characterized by high fertility which is offset by high mortality rates. Mortality is high because sanitation and medical technology are not well developed, and because agriculture itself is often inconsistent and prone to catastrophes such as drought and floods.  Human fertility is high in order to compensate for the losses in mortality.  Population growth is at near zero a long-run equilibrium (in the short-run wide fluctuations occur as disease and famines strike) condition.

The second stage follows as agricultural efficiency improves and progress is made in sanitation and medical technology.  These advances cause mortality rates to fall at a relatively fast pace.  Fertility is slow to respond however, since the cultural values that shape reproductive habits are deeply ingrained psychological beliefs which are extremely resistant to change.  With the fertility rate now higher than that necessary solely to replace the population, population growth occurs.

Fertility does begin to slowly fall however, as individuals change their lifestyles and have fewer children, and will eventually, in a fully developed society, reach a new and lower equilibrium with mortality.  It is during this transition period, as agricultural efficiency improves, that some surplus goods are created and some labor is freed up for other tasks.  This surplus build-up is the essence of savings and capital accumulation. Further, labor that is no longer required on the farm is used in various manufacturing tasks.  As larger pools of under employed labor are created, factories begin to form.

Manufacturing has a feedback effect on agricultural efficiency through the invention and improvement of machines and techniques (tractors, fertilizers, pesticides, etc.) designed to garner greater output from the land.  This feedback frees up more labor, and the process begins to accelerate.  The ongoing and continuous trend however is very clear. The percentage employed in agriculture continuously declines fromits starting point at near a 100 percent to nearly nothing, and is replaced by employment in the manufacturing, and later the services sector.

As society moves into the industrial phase of development, society moves from a predominantly rural setting to a highly urban one. Economic pressures and a weakened need for the nuclear family in the urban setting are primary motivators in the third stage's fertility declines.  As women are introduced into the labor force, the opportunity costs of having children, measured in both time and earnings, rise.  Furthermore, as universal education becomes the norm, children no longer enter the labor force early and the advantages once associated with having additional productive family members are reversed by the disadvantages of having to pay the educational and other costs associated with having children.

These factors combine to slowly counteract the dramatic decline in mortality rates. There are difficulties with trying to apply the demographic transition theory, which was developed primarily to explain Western development, directly to the Third World.  The first and most obvious is that modern developing nations must undergo the process at a greatly accelerated rate compared to 18th and 19th century Europe.  The faster declines in mortality rates are much harder to match with fast declines in birth rates, if for no other reason than the fact that it is difficult to change human nature.  What Europe had centuries to do, the Third World must do in decades.

More importantly however, the demographic transition theory is a purely descriptive theory.  For a causal explanation we must look to the political and economic changes that occur during the development process. A.F.K. Organski, in Birth, Death, and Taxes, offers perhaps the most unified causal explanation of development.  His primary argument is that it is the political capacity of government that is the key factor omitted from the demographic transition explanation.  Greater governmental capacity implies a greater ability to provide the collective goods which no individual can afford, but which are so essential to development. Within such a framework, it would be easy to see how increased military expenditure, by increasing the power and capacity of a government (often through simple coercion), could lead to further development and greater economic growth.

Following the process specifically, agricultural societies, because they have few resources to begin with, and perhaps more importantly, because they have little surplus or accumulated capital, exist in a very constrained political environment.  The states that do exist at this level of development are very decentralized, and the elite that lead them command few resources because of the inherent difficulties involved in collecting them (through taxes).  As technical development slowly occurs, and surplus labor and capital slowly develop, central government elites slowly gather more resources and become more powerful.

Eventually, new institutions (factories, banks, and new bureaucratic agencies) are created to make the capital accumulation process more efficient and in turn make government revenue collection more efficient.  The government, often through military force financed by its new revenues, is further able to consolidate its power.  As this process continues, the power of the central ruling elite rises.  Many of the resources collected through this process are, of course, reinvested in public goods such as infrastructure, health, and sanitation; all of which contribute to the continued growth of the economy (and as a side-benefit the well-being of the masses).

Such reinvestment by elites is self-motivated.  Economic growth increases their power and resource pools, although to actually use these resources the state must tax.  Taxation naturally creates resistance among the masses, which must be broken and therefore requires the creation of a bureaucracy or the redistribution of some of the taxes in the form of payoffs to maintain a sufficiently strong coalition that keeps them in power.

The political development process eventually reaches a peak though, since its primary engine is the falling 'cost' of political action. (31) 'Political cost' essentially represents the difficulty for a government in exerting its influence and collecting revenue (such is the measure of penetration referred to in footnote 4).  Somewhere in the midst of the development process, diminishing returns takes its toll on the innovative new technical and managerial techniques.  Simultaneously, members of the ruling coalition (32) begin demand more in return for their support. This support must be maintained if a leader is to remain in power. Power becomes more diffuse as the size of bureaucracies and the red tape they produce grow to complex proportions, as the legal system becomes a form of resistance, and as universal education brings larger numbers of competing elites into the arena.  This represents a new form of political cost; one which contributes to the slowing of the growth rate.

Aside from the theories of political development, much work has been done within the economics discipline in identifying the various specific economic variables involved in this process. Currently, the most accepted view of economic development is that as states move from least developed to moderately developed, their economies become more capital-intensive.  This follows naturally from the demographic transition theory's demonstration that labor moves from agriculture, which is naturally a more labor intensive endeavor, to manufacturing or industrial jobs.

As a result of the demographic transition theory, some early economists in the 1950s and 60s advocated that Third World states industrialize as quickly as possible and therefore use the most capital intensive technologies available.  This strategy however created problems in the Third World since there were few laborers skilled or educated enough to employ such technology, and more importantly, since moving the laborers off the farm was not preceded by correspondent or sufficient increases in agricultural productivity.  The shift away from agriculture was therefore followed by shortages of food, which hindered growth since food had to be imported (or people starved).  Large numbers of unemployed and unskilled farmers were created leaving a pool of people which represented a tremendous drag on the economy.  This often repeated sequence of events in the Third World highlighted the importance of surplus capital-accumulation models (Deger's savings-investment model works on this surplus drives growth assumption) and the importance of matching the capital-intensity of technology usage to a level appropriate for the population.

A more appropriate model, which has been accepted by economists and policy makers in the 1980s, is that the most efficient and most capital intensive technologies need not be adopted immediately.  In fact, older, more labor intensive technologies should be adopted initially since it employs more labor. (33) Although the labor is employed at less than maximum efficiency, it is nonetheless employed and producing output and earning wages which get pumped back into other sectors of the economy. The essence of Keynesian economics is to maintain full employment, and thereby allow wage-earning workers to create a secondary feedback in demand for consumer goods.  Unemployed labor however, clearly does nothing but drag on government welfare resources, and therefore hinders growth.

Given this economic framework, then, one can imagine that military expenditure could have a positive effect by employing labor and thereby stimulating aggregate demand, or a negative effect by investing too heavily in capital intensive resources that aren't appropriate given a low level of development.  Since military expenditure is often biased towards the latter, especially in a competitive international environment that demands the most sophisticated weapons for security reasons, it is likely that military expenditure will distort capital to labor ratios from their appropriate level (Faini, Annez, and Taylor, 1984; find that militaries typically shift economic activity from agriculture to manufacturing). According to Nicole Ball, who is perhaps the only author to really frame military expenditure in the context of the larger development debate, 'one possible negative effect of military-led industrialization would, for example, be the orientation of a country's industrial sector toward capital-intensive production processes and products not most urgently required by the large majority of the country's population.' (34)

Only in the moderately developed states like Brazil, Korea, or India, which are already industrially oriented, would one expect to find positive effects of militarily led industrialization.  In the least developed states, the military is going to draw capital and labor resources from vital agricultural activities, while in advanced industrial states, the military is likely to draw resources from efficient but labor-intensive service and information sectors.

As can be seen from this analysis, the process of growth and development is curvilinear.  Slow growth builds momentum until it reaches a critical threshold at which point it accelerates rapidly. Eventually though, even the fastest growth is brought into check by diminishing returns and coalition-related constraints, and growth slows again.  It is in the midst of the process the growth rate is at its peak.

The similarity between the theoretical-empirical constructs of the various developmental theories, all of which hypothesize different types of curvilinear relationship, and the observed curvilinear relationship between military expenditure and economic growth, leads me to propose the following model.

A MODEL OF MILITARY EXPENDITURE
Chapter 4
MATCHING THE STRUCTURE OF THE MILITARY APPROPRIATELY TO THE NEEDS OF THE ECONOMY

As stated in the last chapter, the relationship between military expenditure and economic growth should be curvilinear. Furthermore, given the economic arguments that nations should, in general, use technologies which match the capital to labor ratio appropriate to their level of development, it seems reasonable to postulate that militaries should match their own technology usage and labor employment to that same appropriate level.

How would one properly operationalize these ideas?  The following variables are defined as: (35)

Given these variables, it is possible to argue that CGD divided by POP is a somewhat crude measure of the government's labor to capital expenditures (military plus non-military).  In reality it measures the amount of money the government spends on each citizen.  A more appropriate measure (I will be constrained to the CGD/POP version because the subtraction process creates some negatives that I can't adequately run through statistics program.  This will create a small bias in the results, but since CGD is much higher than MXD, the problem is small) would be to subtract out the military to obtain a civilian CGD (CGD Ð MXD) and a civilian population (POP - FRC) then to divide the civilian CGD by thecivilian POP.  This will measure the amount of dollars spent on each civilian.
       
I will take it as an assumption that this figure is an indicator of the labor to capital ratio. (36) My reasoning is that given the fact that capital intensity rises over the course of development and the fact that government expenditure rises as well, the two should be highly correlated (when tested the correlation is about .85) (along the same lines, GNP/capita would be a good predictor of capital intensity, but that variable is being used elsewhere, and isn't comparable to government military expenditure anyway).  The lower the figure, the more labor intensive an economy, the higher the more capital intensive.  We shall call the resulting variable CIVRAT.
       
An approximation of the military labor participation ratio (as Weede refers to it) is much simpler.  It should simply be the total military expenditure (MXD)(also referred to as RATIO in the appendix) divided by total armed forces personnel (FRC)(AFP in some graphs in Chapter 5).  A nation with a high ratio, like the U.S.'s 140, is heavily capital-intensive and investing $140,000 for every soldier in it army. Nations with low ratios are extremely labor-intensive and derive their military power not from technology, but from sheer manpower.  Theresulting variable will be called MPART. (37)
 
When comparing MPART and CIVRAT (or CGD/POP), one would expect to find that they most closely match each other in the moderately developed LDCs, where civilian production is concentrated in the industrial sector. To test the two variables against level of development, it would be necessary to subtract them and plot the result against level of development (an absolute value of this would be most convenient and could be obtained by [MPART Ð CIVRAT]2 ).
       
Several factors are going to work against the prediction that a smooth curvilinear relationship would result, although one would expect it to be close.  First, there are other variables which act in combination with appropriate capital-labor ratios.  Finding any significant positive relationship would be important.  Second, every military is going to try to use the most capital-intensive systems it can afford for genuine security reasons.  Therefore, one would expect that MPART capital-intensity will increase in a linear fashion regardless of the stage of development (we see this in the limited OECD data chart in Chapter 5 and the larger ICPSR data results in Appendix 1).  This should mean that the MPART capital-intensity curve won't turn back downward in advanced industrial countries (although the correlation between expenditure and growth, which unfortunately can't be examined here, will be turn negative because the there would no longer be a match between CIVRAT and MPART).
       
An ideal study would be to perform a time-series analysis on many countries individually to determine the correlation between military expenditure and economic growth, and then aggregate those results cross-nationally and compare them to level of development and the match ([MPART Ð CIVRAT]2) between military labor participation and civilian capital-labor ratios; then an adequate answer could be given to whether the relationship was curvilinear and to whether it was caused (partially) by the traditional developmental pattern of rising capital-intensity. Unfortunately, such an answer is beyond the scope of my tests.

TESTING THE MODEL
Chapter 5
THE RESULTS OF THE DATA

Several rudimentary tests were conducted to determine if capital-labor ratios were a worthwhile avenue for further research.  Initially, using a relatively small and manageable data set to derive a rough picture (the larger data sets require far more difficult MTS mainframe operations). Using data from OECD nations, the chart below, and the correlation and regression statistics, were generated. (38)