Download War and the Growth of Government

War and the Growth of Government
Colin O’Reilly
Assistant Professor of Economics
University of Wisconsin Stout
[email protected]
and
Dr. Benjamin Powell
Professor of Economics
Rawls College of Business
Texas Tech University
[email protected]
1/16/2015
Abstract:
This paper empirically examines how wars impact the size and scope of government using a
panel of all wars from 1965 to 2010 to test the hypothesis that wars are crises that cause an
expansion of government that persists after the wars are over. The results indicate that wars
permanently expand the scope of government regulation. Wars also increase the size of
government but only in highly autocratic regimes.

We thank the participants at the Association for Private Enterprise Education annual conference for comments on a
prior draft. Powell thanks the John Templeton Foundation for financial support.
JEL Codes: H11, H12, N40, P48
Key Words: War, Economic Freedom, Size of Government, Institutional Change
1
I.
Introduction
Wars destroy lives and capital while they are fought. But their impact on human
suffering could last into the long run if they change a country’s institutional environment.
Institutions are an important fundamental cause of economic development (Rodrik,
Subramanian, & Trebbi, 2004) and a growing empirical literature supports the importance of an
institutional environment of strong private property rights, a rule of law, and an environment of
economic freedom for promoting long run growth (for surveys see: De Haan, Lundström, and
Sturm (2006), Hall and Lawson (2014)). This paper examines how wars impact aspects of these
institutions.
There is a large political economy literature that explains the existence of government
policies as the result of the power of vested interests.1 The role of ideas, though less formally
modeled has also long been assumed to play a role. Keynes even claimed the primacy of ideas,
“The ideas of economics and political philosophers, both when they are right and when they are
wrong, are more powerful than is commonly understood. Indeed the world is ruled by little else.
Practical men, who believe themselves to be quite exempt from any intellectual influence, are
usually the slaves of some defunct economist” (1936: 383).
Crises impact both interests and ideas. Crises can lead to a change in policies “because
prevailing interests may lose some of their legitimacy and because incumbents may be open to
trying new remedies”(Rodrik, 2014: 203). In other words, crises can break the political
equilibrium by changing the power of vested interests or changing peoples’ ideas about how the
world works and what policies a government should adopt.
War is one major form of a crisis that could impact the strength of interest groups and
peoples’ ideas about the proper role of government. We empirically investigate how wars have
changed the size and scope of governments over the last 40 years.
There are theoretical reasons to believe that war could either increase or decrease the size
and scope of government. Higgs (1987) argued that the U.S. government’s growth in the 20th
century was caused by a “ratchet effect” stemming from crises. In times of crisis people
demanded increased government services and interventions as their faith in the efficiency of
1
See Mueller (2003) for a survey.
2
markets is shaken or a larger government was needed to fight wars. According to Higgs, the
scope of government is expanded during the crisis, but then after the crisis passes government
does not revert all the way back to its pre-crisis size and scope (the ratchet effect). Higgs
theorizes that both ideas and interests play a role in locking in the growth of government.
Interests who benefit from the growth of government lobby to keep the new policies and
individuals do not demand that the government revert to its original scope because they are more
accustomed to the new role of the state.
Alternatively, Olson (1982, 2000) argued that major disruptions and crises, such as wars,
remove entrenched interest groups and clear the way for productive reforms. Olson theorizes
that interest groups form over time, as they solve collective action problems, and demand more
public goods and pursue the cartelization of the economy. This implies that over time rent
seeking will expand the size and scope of government. A war can shatter these interest groups
allowing reforms that shrink the size and scope of government.
Both Higgs and Olson point to World War II for evidence in support of their theories.
The case of the United States during WWII fits the narrative of the ratchet effect well, with
government spending increasing drastically during the crisis of WWII and the Great Depression
and contracting after the war ended. However, government spending and the size of the military
never contracted to pre-WWII levels (Higgs, 1987).2 Conversely, the cases of post-war Japan and
Germany are described better by Olson’s theory. Both Japan and Germany were plagued by
economic stagnation and high inflation before WWII, but after massive destruction and the
breakup of powerful interests, both countries had smaller governments and experienced
impressive economic expansion (Coyne, 2008).
A number of countries that have undertaken major liberalizations in the last twenty years,
such as Ireland, New Zealand, and India, all experienced some form of financial crisis
immediately prior to liberalization (Powell, 2008). Pitlik and Wirth go so far as to say, “A
commonly shared wisdom among economists and political scientists is that crises promote the
adoption of market-oriented reforms” (2003: 565). There are a few papers that have empirically
examined whether financial crises impact the size and scope of government across a cross
2
Holcombe(1993) argues that after taking account of the trend toward government growth through the 20 th century
that the depression/WWII ratchet is not statistically significant.
3
section of countries.3 Pitlik and Wirth (2003) examined how GDP growth crises and inflation
crises impacted the economic freedom scores of 57 countries over five year periods from 1970
through 2000 and found that deep growth and inflation crises were significantly correlated with
increases in economic freedom. Similarly, De Haan et al. (2009) find that banking crises are
correlated with decreases in the size of government, as measured by the economic freedom
index, over five year periods.
However, Bologna and Young (2014) give us reason to doubt financial crises are
correlated with increased economic freedom. They examine a panel of 70 countries from 1966
through 2010 using five different types of financial crises (banking, currency, inflation, internal
debt, external debt) to see how they impact all five areas of the economic freedom index (size of
government, legal structure and property rights, sound currency, freedom to trade internationally,
and regulation) over 5, 10, and 40 year time periods. They also employ several weighting
schemes to account for the timing and severity of the crises. Most of their results were
statistically insignificant. However, they did find that financial crises tended to be associated
with a decrease in that country’s government consumption to GDP ratio but they find crises are
largely uncorrelated with other measures of economic freedom. However, over a 40 year period
they do find that countries that spend more time in crisis tend to have less sound legal systems
and protections for property rights.4
Although some work has been done on how various types of economic crises impact the
size and scope of government, cross country statistical analysis of the effect of war on
institutions is more limited. A World Bank study of civil conflict finds that institutional measures
of economic policy, democracy, and civil liberties are, on average, worse than their pre-conflict
level in the decade following the conflict (World Bank, 2003: 22). In a more rigorous study using
control groups, Chen, Loayza, and Reynal-Querol (2008) study the pre- and postwar levels of
macroeconomic variables for a large cross section of countries that experienced civil wars.
3
A few other studies Bruno and Easterly (1996), Drazen and Easterly (2001), and Lora and Olivera (2004) examine
whether crises are associated with improved economic performance which may indirectly measure a change in the
size and scope of government or policy.
4
Bologna and Young include intra-state conflicts (civil wars) as a contemporaneous control variable when
investigating whether financial crises cause changes in freedom in subsequent periods. Thus, they do not investigate
whether wars themselves, as crises, lead to changes in freedom in subsequent periods as our study does not do they
examine other more numerous categories of war that we examine.
4
Although their focus is not the scope of government, they find that “government expenditure (as
a percentage of GDP) increases 1 percentage point between the pre- and post-war periods;” (75)
however, their results are not statistically significant leading them to conclude that there is “no
discernable level change,” (71) in government expenditure. Though they do suggest a declining
trend in government expenditure after civil war in absolute terms. Assessing these studies in their
survey of the civil war literature Blattman and Miguel (2010) acknowledge that “The social and
institutional legacies of conflict are arguably the most important but least understood of all war
impacts.” In this study we contribute to understanding the institutional legacies of war.
We also add to the literature that examines the causes of economic freedom. There is
evidence that economic freedom is enhanced by fiscal decentralization (Cassette & Paty, 2010),
more educated politicians (Dreher, Lamla, Lein, & Somogyi, 2009), and by the competitiveness
of the political environment (Leonida, Maimone Ansaldo Patti, & Navarra, 2007). Djankov,
McLiesh, Nenova, and Shleifer (2003), and Bjørnskov (2010) examined the determinants of
legal institutions consistent with the economic freedom. This paper contributes to this literature
by addressing how wars impact two specific components of economic freedom: the size of
government and regulation.
This paper tests whether war impacts the size and scope of government on a large cross
section of countries using data on conflict events from 1965-2005. We test the impact of a war
on the size and scope of government, comparing pre-war and post-war size and scope, while
controlling for government’s size and scope prior to the war. A variety different types of wars
are used as crisis events, including civil wars, international wars, internationalized wars and extra
state wars. The next section describes our data and methodology. Section III contains our
results. The final section concludes.
II.
Data and Methodology
A panel spanning from 1965-2010 is constructed from two main data sources: the
Economic Freedom of the World Annual Report (J. Gwartney, Lawson, & Hall, 2013) and the
Correlates of War dataset (Sarkees, 2010). The full sample considers 124 countries over the
period though the panel is unbalanced due to missing data on the size and scope of government
for the early periods. All analysis is conducted on a panel broken into five year periods.
5
We use two components of the Economic Freedom of the World index (EFW) to capture
the size and scope of government across countries: the “size of government” and “regulation.”
The size of government component measures government consumption as a percent of total
consumption; transfers and subsidies as a percent of GDP; the extent of private investment and
business compared to government investment and state-owned enterprises; and the top marginal
income and payroll tax rates. The regulation component includes three measures of credit market
regulation, six measures of labor market regulation, and six measures of business regulation. The
EFW data is based on a 0 to 10 point scale where 10 indicates the greatest economic freedom
(smallest size/scope of government). The EFW data is available back to 1970 in five year
intervals. As an additional measure of the size of government we use government expenditure as
a percentage of GDP from the World Development Indicators (World Bank, 2014). We
acknowledge that these variables measure only some aspects of the size and scope of
government, and that many other dimensions of the size and scope of government are of interest
but do not lend themselves to quantitative analysis.
Data on violent conflicts including international wars, civil wars, and extra state wars is
taken from the Correlates of War (COW) dataset. The COW dataset includes all wars from 1816
to 2007. A war is defined as “sustained combat, involving organized armed forces, resulting in a
minimum of 1,000 battle related deaths” (Sarkees, 2010: 40). This definition, with some
modifications, has become standard in the empirical economic and political science literature on
violent conflict and does not substantially differ from the violence thresholds used in the Peace
Research Institute Oslo dataset of wars (Gleditsch, Wallensteen, Eriksson, Sollenberg, & Strand,
2002).
The war event data is reshaped from event data to annual data and then paired with the
EFW data on five year intervals. We use a dummy variable equal to 1 if a country experienced a
war during a five year period. The four main war categories and their corresponding indicator
variables are as follows: civil wars or intrastate wars (civil wars), international wars, extra state
wars, and internationalized wars. Extra state wars corresponds to wars “in which a member of
the interstate system is engaged in sustained combat outside its borders against the forces
(however irregular or disorganized) of a political entity that is not a member of the state system”
(Sarkees, 2010: 63). The COW dataset most often codes wars as extra state in cases of colonial
6
or imperial wars.5As an example of an extra state war, in 1991 Turkey engaged in combat with
the Kurdish PKK group which was located inside of Iraq. This is not coded as an international
war since Turkey was not engaged in conflict with state of Iraq, and it is not coded as an
intrastate conflict since the conflict did not occur within the borders of Turkey (Sarkees, 2010:
327).
International wars are conflicts between two members of the interstate system, whereas a
war is coded as internationalized when an outside state intervenes in an intrastate conflict. For
example, the Third Somalia War is classified as an intrastate conflict because the Somali
government was fighting rebel groups. However, in 2007 the United States and Ethiopia
intervened in this war, causing the war to be “internationalized” (Sarkees, 2010: 480). For the
purposes of this study, states conducting the intervention, the United States and Ethiopia in this
example, are coded as being engaged in an international conflict, as well as, an internationalized
conflict for the period; thus internationalized conflicts are subset of international conflicts. The
coding for Somalia is not effected by the intervention, Somalia is coded as experiencing an
intrastate war in 2007.
Note that there are many cases when a country is experiencing multiple types of war
during the same five year period, in these circumstances dummy variables are equal to one for
multiple war categories. Finally, the variable War is an indicator equal to one if a country
experiences any type of war during the corresponding period. Of the 124 countries in our sample
60 experienced wars of some type. Of these 60 countries, 34 experienced civil wars and 45
experienced international wars. Many countries experienced multiple wars and even multiple
types of wars over the sample period. For instance, Nigeria experienced a civil war in the 1985
and 1990 periods and then was engaged in an international war during the 1995, 2000 and 2005
periods.
Control variables including Net ODA as a percentage of GDP, the logarithm of real GDP
per capita and military expenditure as a percentage of GDP are collected from the World
Development Indicators (World Bank, 2014). The polity index of democracy and autocracy from
5
A more in-depth discussion of the typology of extra state wars can be found in Sarkees (2010).
7
the Polity IV dataset is also used a control variable (Marshall, Gurr & Jaggers, 2013).
Descriptive statistics are presented in Table 1.
(Table 1)
To test the effect of wars on the size and scope of government we estimate the following
equation:
𝐺𝑜𝑣𝑡𝑡+1 − 𝐺𝑜𝑣𝑡𝑡−1 = 𝛽1 𝑊𝑎𝑟𝑡−1 + 𝛽2 𝑊𝑎𝑟𝑖,𝑡 + +𝛽3 𝑊𝑎𝑟𝑖, 𝑡+1 + 𝛽4 𝑊𝑎𝑟𝑖, 𝑡 ∗ 𝑊𝑎𝑟𝑖,𝑡+1 + 𝛽5 𝜏
+𝛽6 𝑓𝑖 + 𝛽7 𝐺𝑜𝑣𝑡𝑖,𝑡−1 + 𝛽8 𝑋𝑖,𝑡 + 𝜀𝑖,𝑡
Where i indicates countries and t indicates five year time periods. The variable Govt indicates the
size or scope of government as measured by the components of the EFW index or government
spending as a percentage of GDP and 𝑊𝑎𝑟𝑡 indicates a dummy variable equal to one if a country
is engaged in war during the period t. War in the prior period and in the following period are
controlled for with 𝑊𝑎𝑟𝑡−1 and 𝑊𝑎𝑟𝑡+1 . An interaction between wars in period t and wars in
period t+1 is included to account for wars that extend beyond period t. Period fixed effects, 𝜏, are
included to account for any long term trends in the size or scope of government and country
fixed effects, 𝑓𝑖 , are included to control for any omitted country characteristics6. The inclusion of
time and countries fixed effects constitutes a difference-in-difference methodology. Finally, X is
a vector of controls that vary over time.
The inclusion of 𝐺𝑜𝑣𝑡𝑡−1controls for the initial size or scope of government at the outset
of the time period when a war later occurred. The evolution of institutions is complex and not
well understood. By controlling for initial levels of the size and scope of government we are
implicitly controlling for these long run factors and enabling us to examine only the impact of
having a war.
The parameter of most interest is 𝛽2which measures the change in the size or scope of
government from before the war began to five years after the war has ended. A negative
6
A hausman test rejects the random effects model.
8
estimated coefficient on the war variable, 𝛽2 < 0, indicates that a war in period t (1985-1990, for
example) caused the size or scope of government to increase (the change from 1985 to 1995).
III.
Results
a. The Size of Government
To assess the impact of war on the size of government we estimate fixed effects models
with Area 1 of the Economic Freedom of the World and government spending as a percentage of
GDP as measures of the size of government, these results are presented in Table 2 and Table 3
respectively. All specifications include controls for war in the prior period, t-1, controls for war
in the following period, t+1, as well as an interaction between war in period t and t+1. These
variables are included to account for countries that experience many wars, or have an ongoing
war. The coefficient on the main variable of interest, war, indicates the impact that a war has had
on the size of government in a country that has experienced a war during the five year period, but
is not experiencing war during the subsequent five year period.
(Table 2 and Table 3)
The baseline specification in Column 1 includes country fixed effects to account for all
time invariant country characteristics, such as unobserved cultural an institutional factors. In
addition the baseline specification includes time effects, to account for any long run trends in the
size of government. Therefore, the baseline specification is a difference-in-difference approach
to estimation.
The coefficient corresponding to the war variable is negative when the Area 1 index is
the outcome variable (indicating an increase in the size of government). Similarly the coefficient
is positive when government spending as a percentage of GDP is the outcome. Though both
results are consistent with the ratchet effect, a decrease in the index or an increase in government
spending, neither estimate is statistically significant. This finding constitutes a null result of a
ratchet effect regarding these specific measures of the size of government.
Variables that changes over time, including the pre-war Polity IV score, Net ODA as a
percentage of GDP and the log of pre-war GDP per capita are included in Columns 2 through 4.
The inclusion of these controls does little to alter the conclusions from the baseline specification;
we do not find a significant relationship between war and the size of government. When the
outcome variable is the size of government index the inclusion of control variables reduces the
size of standard errors, though the point estimate on war is still statistically insignificant. The
9
final Column adds pre-war military expenditure as a percentage of GDP as a control which
reduces the sample size by about half since it is only available since 1980. Estimates are still
insignificant when military expenditure is included.
Despite point estimates of the expected sign, we find no significant evidence that the size
of government is affected by experiencing war. We will investigate the possibility of a
conditional relationship in subsection e. Certainly there are many aspects to the size of the
government that we are not able to measure, thus the lack of a significant result should be viewed
as a null result, not as a rejection of the Higgsian ratchet for the size of government.
b. The Scope of Government
Fixed effects estimates of the effect war on the scope of government are presented in
Table 4, with the baseline specification in Column 1 and control variables are added in
subsequent columns. The index of the scope of government measures the absence of regulation,
thus larger values mean a smaller scope of government. In the baseline specification the
coefficient on war is -.29, and is statistically significant at the five percent level. Therefore, war
is associated with an increase in the scope of government. The estimated effect is stable when
control variables are added in columns 2 through 4, increasing slightly to -.30. The results are
even robust, though less precisely estimated, to the inclusion of military expenditure as a
percentage of GDP, which restricts the sample to wars occurring after 1980.
The effect of -.30 is an economically meaningful change in the scope of government
given that the standard deviation of the regulation index is 1.36. Further, the dependent variable
is the change in the scope of government over a period of 10 years. The average change in the
scope of government regulation over a 10 year period is .57 with a standard deviation of .93.
Thus a coefficient of -.30 represents of about a third of a standard deviation decline in
improvement of the scope of regulation index.
(Table 4)
c. Outcome of the Conflict
The ratchet effect may be contingent on the outcome of the conflict. In Tables 5 through
7 we re-estimate the most complete specification, with controls of every variable except military
expenditure, but replace war with a variable that indicates if the war ended in a decisive loss or
10
decisive win. For each table Column 1 presents the results for a decisive loss and Column 2 for a
decisive win7.
(Table 5 and Table 6)
Again, the results that correspond to the size of government yield the expected sign, but
lack statistical significance. However, for the scope of government outcome the coefficient on
loss in Table 7 is -.48 and highly statistically significant. The coefficient on win is -.11, but is not
statistically significant. This is some indication that decisive losses of war are driving the main
result of a war induced ratchet effect. However, the results for win and loss should be interpreted
with care. Modern wars a rarely clear cut events, and defining a war as a win or a loss is
inherently subjective.
(Table 7)
d. Type of Conflict
The final three columns of Table 5 through 7 present results for different types of wars:
interstate wars, civil wars and internationalized wars. Consistent with the results from Table 2,
when considering the size of government the estimates are statistically insignificant. However,
the coefficient is the largest and most precisely estimated for internationalized conflicts. Table 7
presents the results for the scope of government. Consistent with the previous results the
coefficients are all negative and of a similar magnitude to the results in Table 4, however all
estimates statistically insignificant regardless of the type of war chosen. The loss of statistical
significance appears to be due to larger standard errors when considering only one type of war.
Breaking the results down by the type of war does not provide any compelling evidence that a
particular type of war is driving the main results.
e. Democracy Interaction
To investigate the possibility that the growth in government caused by wars is dependent
on the political decision rules in a country we include an interaction between the instance of war
7
Many wars to not end in a decisive win or loss and are not coded as either.
11
and the Polity index of democracy and autocracy. Calculated from the interaction between the
Polity IV index and the instance war, the marginal effects of war are presented in Table 8.
Polity scores are listed down the left side of the table, with a -10 indicating the most
autocratic and 10 indicating the most democratic decision rules. The effect of a war on the size
of government as measured by Area 1 of EFW is statistically significant for the most autocratic
counties. For countries with more democratic institutions the impact of war is positive but not
significant, the effect of war becomes negative at a polity score of zero and becomes larger in
magnitude as counties become more autocratic. The size of the effect of war ranges from -.30 to
-.46 for polity scores ranging from -6 to -10. Therefore, the effect of war on the size of
government is conditional on the degree to which decision rules are democratic. This explains
the absence of a significant relationship when considering the unconditional effect of war in
Table 1.
The marginal effects of government spending over different measures of the polity index
are presented in column 2. Though the marginal effects are not statistically significant for any
level of the polity index, the point estimates indicate the largest increase in government spending
for the most autocratic countries; this is consistent with the results from column 1.
The conditional relationship between democratic institutions and the scope of
government is the most compelling. The effect of war on the scope of government ranges from .12 for the most democratic countries to -.45 for the most autocratic countries. These results are
significant that the 5% level for all polity scores below 2, which indicates institutions that have
very few democratic elements. For the most autocratic counties, engaging in war is associated
with half of a standard deviation lower improvement in the scope of regulation in the economy.
The executive constraints built into democratic institutions appear to limit the extent of the
ratchet effect.
(Table 8)
IV.
Conclusion
Crises can break existing political equilibria and allow for institutional change that was
not previously possible.
The existing cross-country empirical work examining the role of crises
on institutional change has focused on economic crises. Although results of those studies vary,
12
in general they tend to indicate that if economic crises have an impact on institutions, it is the
direction of liberalization and greater economic freedom.
This study finds that, unlike economic crises, wars tend to be associated with an increase
in the scope of government (decrease in economic freedom). We also find some evidence that
wars expand the size of government in the most autocratic regimes.
Paul Collier has described civil war as “development in reverse” (World Bank, 2003).
This same description may apply to war in general. The scope of regulation score has been
improving around the world at a rate of .57 over a span of two decades. Yet, if a country engages
in war during that period the expected improvement in the index is only .27. Conflict is slowing
the process of developing more free and open regulatory environment.
Collier identified a cycle of low income leading to conflict and conflict leading to lower
incomes; he dubbed this cycle “the conflict trap” (World Bank, 2003). Our results indicate that
the deterioration of economic freedom may play a role in the conflict trap. If, as our results
indicate, engaging in war actually erodes a dimension of economic freedom, we may have
identified another channel through which Collier’s conflict trap is perpetuated.
13
References:
Bjørnskov, C. (2010). How does social trust lead to better governance? An attempt to separate
electoral and bureaucratic mechanisms. Public Choice, 144(1-2), 323–346.
Blattman, C., & Miguel, E. (2010). Civil War. Journal of Economic Literature, 48(1), 3–57.
Bologna, J. and A. Young (2014). Crises and Government: Some Empirical Evidence. Working
Paper. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2313846
Bruno, M., & Easterly, W. (1996). Inflation’s Children: Tales of Crises That Beget Reforms. The
American Economic Review, 86(2), 213–217.
Cassette, A., & Paty, S. (2010). Fiscal decentralization and the size of government: a European
country empirical analysis. Public Choice, 143(1-2), 173–189.
Chen, S., Loayza, N. V., & Reynal-Querol, M. (2008). The Aftermath of Civil War. The World
Bank Economic Review, 22(1), 63–85.
Coyne, C. J. (2008). After war: the political economy of exporting democracy. Stanford, Calif:
Stanford Economics and Finance.
De Haan, J., Lundström, S., & Sturm, J.-E. (2006). Market-oriented institutions and policies and
economic growth: A critical survey. Journal of Economic Surveys, 20(2), 157–191.
De Haan, J., J. Strum, and E. Zandberg (2009). The Impact of Financial and Economic Crises on
Economic Freedom. In Gwartney & Lawson (Eds.), Economic Freedom of the World
2009 Annual Report. Vancouver: Fraser Institute.
Djankov, S., McLiesh, C., Nenova, T., & Shleifer, A. (2003). Who Owns the Media? Journal of
Law and Economics, 46(2), 341–382.
Drazen, A., & Easterly, W. (2001). Do Crises Induce Reform? Simple Empirical Tests of
Conventional Wisdom. Economics and Politics, 13(2), 129–157.
14
Dreher, A., Lamla, M. J., Lein, S. M., & Somogyi, F. (2009). The impact of political leaders’
profession and education on reforms. Journal of Comparative Economics, 37(1), 169–
Gleditsch, N. P., Wallensteen, P., Eriksson, M., Sollenberg, M., & Strand, H. (2002). Armed
Conflict 1946-2001: A New Dataset. Journal of Peace Research, 39(5), 615–637.
Gwartney, J., Lawson, R., & Hall, J. (2013). Economic Freedom of the World: 2013 Annual
Report. Vancouver: Fraiser Institute.
Gwartney, J., Lawson, R., & Holcombe, R. (1999). Economic Freedom and the Environment for
Economic Growth. Journal of Institutional and Theoretical Economics (JITE) /
Zeitschrift Für Die Gesamte Staatswissenschaft, 155(4), 643–663.
Hall, J., & Lawson, R. (2014). Economic Freedom of the World: An Accounting of the
Literature. Contemporary Economic Policy, 32(1), 1–19.
Higgs, R. (1987). Crisis and leviathan: critical episodes in the growth of American government.
New York: Oxford University Press.
Holcombe, R. G. (1993). Are There Ratchets in the Growth of Federal Government Spending?
Public Finance Review, 21(1), 33–47.
Keynes, J. M. (1936). The general theory of employment, interest and money. United States:
Macmillan Cambridge University Press.
Leonida, L., Maimone Ansaldo Patti, D., & Navarra, P. (2007). Towards an Equilibrium Level of
Market Reform: How Politics Affects the Dynamics of Policy Change. Applied
Economics, 39(13-15), 1627–1634.
Lora, E., & Olivera, M. (2004). What Makes Reforms Likely: Political Economy Determinants
of Reforms in Latin America. Journal of Applied Economics, 7(1), 99–135.
15
Marshall, M., Gurr, T. & Jaggers, K. (2013). Polity IV Project: Political Regime Characteristics
and Transitions, 1800-2012. Centre for Systemic Peace.
Mueller, D. C. (2003). Public Choice III. Cambridge University Press.
Olson, M. (1982). The rise and decline of nations: economic growth, stagflation, and social
rigidities. New Haven: Yale University Press.
Olson, M. (2000). Power and prosperity: outgrowing communist and capitalist dictatorships.
New York: Basic Books.
Pitlik, H., & Wirth, S. (2003). Do crises promote the extent of economic liberalization?: an
empirical test. European Journal of Political Economy, 19(3), 565–581.
Powell, B. (2008). Making poor nations rich: entrepreneurship and the process of economic
development. Stanford, Calif.: Stanford Economics and Finance, Stanford University
Press.
Rodrik, D. (2014). When Ideas Trump Interests: Preferences, Worldviews, and Policy
Innovations. Journal of Economic Perspectives, 28(1), 189–208.
Rodrik, D., Subramanian, A., & Trebbi, F. (2004). Institutions Rule: The Primacy of Institutions
Over Geography and Integration in Economic Development. Journal of Economic
Growth, 9(2), 131–165.
Sarkees, M. R. (2010). Resort to war: a data guide to inter-state, extra-state, intra-state, and
non-state wars, 1816-2007. Washington, D.C: CQ Press.
World Bank. (2003). Breaking the conflict trap: civil war and development policy. (P. Collier,
Ed.). Washington, DC : [New York]: World Bank ; Oxford University Press.
World Bank. (2014). World Development Indicators. Retrieved February 22, 2014, from
http://www.worldbank.org
16
Table 1: Descriptive Statistics
Variable
Change in Size of
Government-Area 1
Government
Spending/GDP
Change in Scope of
Government-Area 5
Net ODA/GDP
Source
Gwartney, Lawson, &
Hall (2013)
World Bank (2014)
Polity IV
Marshall, Gurr & Jaggers,
(2013)
World Bank (2014)
Ln GPD per capita
Gwartney, Lawson, &
Hall (2013)
World Bank (2014)
Obs.
Mean/
Standard
Proportion Deviation
771
0.34
1.50
689
0.57
0.93
856
0.27
6.89
2130
4.03
8.23
1296
0.22
7.31
1516
8.35
1.23
Military
World Bank (2014)
719
2.73
Expenditure/GDP
Note: Each observation represents a five year period.
3.05
17
Table 2: Size of Government- Area 1 of EFW
(1)
(2)
(3)
(4)
(5)
-0.0354
(0.1464)
0.1277
(0.1712)
0.2339
(0.2047)
-0.0471
(0.1096)
-0.9855***
(0.0469)
0.0161
(0.0144)
-0.0113
(0.0154)
-0.5808***
(0.1693)
0.3264
(0.2808)
0.6223
(0.3753)
-0.2260
(0.4120)
0.1033
(0.1938)
-1.2455***
(0.0758)
0.0587**
(0.0243)
0.0136
(0.0180)
-0.2835
(0.3685)
-0.0451*
(0.0267)
VARIABLES
War(t)
War(t+1)
War(t)*War(t+1)
War(t-1)
Gov. Size (t-1)
Polity(t-1)
Net oda
Lngdpcap(t-1)
Milexpcap(t-1)
-0.0965
(0.1405)
0.0262
(0.1847)
0.1766
(0.2057)
-0.0056
(0.1078)
-0.9636***
(0.0478)
-0.0933
(0.1373)
0.0424
(0.1795)
0.1694
(0.2004)
0.0127
(0.1082)
-0.9616***
(0.0475)
0.0245
(0.0151)
-0.1047
(0.1400)
0.0389
(0.1793)
0.1750
(0.2003)
-0.0024
(0.1126)
-0.9591***
(0.0486)
0.0236
(0.0154)
0.0111
(0.0170)
Observations
R-squared
Number of countries
771
729
722
682
298
0.635
0.6398
0.640
0.654
0.709
124
118
117
110
108
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Note: These are the results of a fixed effects regression with county and period fixed effects are
included.
18
Table 3: Size of Government- Gov. Spending/GDP
(1)
(2)
(3)
(4)
(5)
VARIABLES
War(t)
War(t+1)
War(t)*War(t+1)
War(t-1)
Gov/GDP(t-1)
Polity(t-1)
Net oda
Lngdpcap(t-1)
Milexpcap(t-1)
0.0818
(0.7174)
0.9150
(1.2390)
0.8444
(1.1694)
-0.9697***
(0.0887)
0.2280
(0.4365)
0.3697
(0.7249)
1.3789
(1.2811)
0.3276
(1.1889)
-0.9670***
(0.0923)
0.2374
(0.4275)
0.0757
(0.0604)
0.4961
(0.7448)
1.4127
(1.2740)
0.2578
(1.1940)
-0.9594***
(0.0884)
0.3490
(0.4251)
0.0730
(0.0597)
-0.1017
(0.0799)
0.2706
(0.7328)
0.8521
(1.2988)
0.0945
(1.2926)
-0.9290***
(0.0948)
0.4507
(0.4158)
0.1230***
(0.0469)
-0.0107
(0.0617)
1.6141*
(0.8740)
0.3029
(0.6397)
0.1003
(0.7104)
-0.2166
(0.8750)
-0.9533***
(0.0786)
0.3472
(0.3861)
0.1408**
(0.0672)
-0.0202
(0.0604)
0.0063
(0.8330)
0.2947***
(0.0990)
Observations
R-squared
Number of countries
755
709
709
684
305
0.552
0.554
0.559
0.539
0.657
137
129
129
125
119
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Note: These are the results of a fixed effects regression with county and period fixed effects are
included.
19
Table 4: Scope of Government (Regulation) - Area 5 of EFW
(1)
(2)
(3)
(4)
(5)
VARIABLES
War(t)
War(t+1)
War(t)*War(t+1)
War(t-1)
Regulation(t-1)
Polity(t-1)
Net oda
Lngdpcap(t-1)
Milexpcap(t-1)
-0.2934**
(0.1188)
-0.3400**
(0.1644)
0.2412
(0.2015)
0.0262
(0.0866)
-0.9657***
(0.0465)
-0.2966**
(0.1149)
-0.3453**
(0.1608)
0.2387
(0.1963)
0.0385
(0.0917)
-0.9814***
(0.0462)
0.0209
(0.0142)
-0.3046***
(0.1146)
-0.3478**
(0.1609)
0.2415
(0.1961)
0.0309
(0.0934)
-0.9819***
(0.0476)
0.0212
(0.0146)
0.0057
(0.0067)
-0.2961**
(0.1261)
-0.3203*
(0.1784)
0.2666
(0.2051)
0.0018
(0.1040)
-0.9763***
(0.0479)
0.0197
(0.0147)
0.0061
(0.0070)
-0.1248
(0.1375)
-0.2841*
(0.1473)
-0.3338*
(0.1911)
0.4422**
(0.2170)
-0.1272
(0.1004)
-1.1658***
(0.0578)
0.0208
(0.0160)
-0.0054
(0.0124)
-0.6762***
(0.2449)
-0.0465***
(0.0110)
Observations
R-squared
Number of countries
689
653
646
614
298
0.535
0.547
0.549
0.550
0.739
123
117
116
109
107
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Note: These are the results of a fixed effects regression with county and period fixed effects are
included.
20
VARIABLES
War(t)
War(t+1)
War(t)*War(t+1)
War(t-1)
Gov. Size(t-1)
Polity(t-1)
Net oda
Lngdpcap(t-1)
Observations
R-squared
Number of countries
Table 5: Size of Government- Area 1 of EFW (War Types)
(1)
(2)
(3)
(4)
Loss
Win
Inter
Civil
(5)
Intervention
-0.1834
(0.1845)
0.3781
(0.2435)
0.7165
(0.5193)
-0.1693
(0.1302)
-0.9847***
(0.0464)
0.0191
(0.0147)
-0.0077
(0.0155)
-0.5558***
(0.1563)
-0.2214
(0.1502)
-0.0932
(0.1719)
0.4370
(0.2666)
0.0235
(0.1673)
-0.9866***
(0.0473)
0.0170
(0.0146)
-0.0080
(0.0159)
-0.5161***
(0.1647)
-0.1558
(0.1249)
-0.2341
(0.1586)
0.5393**
(0.2525)
-0.1788
(0.1227)
-0.9870***
(0.0477)
0.0158
(0.0147)
-0.0076
(0.0153)
-0.5162***
(0.1640)
-0.0324
(0.1250)
-0.0181
(0.1563)
0.1270
(0.2354)
-0.0973
(0.1147)
-0.9848***
(0.0471)
0.0165
(0.0147)
-0.0086
(0.0156)
-0.5398***
(0.1683)
-0.1830
(0.2000)
0.1637
(0.2848)
0.5149
(0.3141)
-0.0698
(0.1448)
-0.9852***
(0.0467)
0.0187
(0.0144)
-0.0110
(0.0152)
-0.5415***
(0.1596)
682
0.655
110
682
682
682
682
0.653
0.650
0.656
0.651
110
110
110
110
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Note: These are the results of a fixed effects regression with county and period fixed effects are
included.
21
VARIABLES
Table 6: Size of Government – Gov. Spending/GDP (War Types)
(1)
(2)
(3)
(4)
Loss
Win
Inter
Civil
War(t)
War(t+1)
War(t)*War(t+1)
War(t-1)
Gov/GDP(t-1)
Polity(t-1)
Net oda
Lngdpcap(t-1)
Observations
R-squared
Number of countries
0.3719
(0.6967)
-0.5035
(1.1358)
-0.8738
(1.9662)
0.7963
(0.8010)
-0.9227***
(0.0952)
0.1138**
(0.0452)
-0.0025
(0.0627)
1.7782*
(0.9054)
0.3048
(0.7131)
1.7500
(1.4291)
-1.9049
(1.4617)
-0.0072
(0.5812)
-0.9199***
(0.0934)
0.1211***
(0.0462)
-0.0047
(0.0625)
1.6855*
(0.8672)
0.5109
(0.7509)
1.8601
(1.4782)
-1.0011
(1.4846)
0.4247
(0.5423)
-0.9317***
(0.0956)
0.1182**
(0.0472)
-0.0096
(0.0590)
1.6340*
(0.8918)
-0.4986
(0.6495)
0.0450
(0.9106)
0.8362
(0.9613)
1.1733*
(0.6204)
-0.9290***
(0.0957)
0.1211***
(0.0460)
0.0019
(0.0618)
1.7478*
(0.9102)
(5)
Intervention
1.0831
(0.9838)
1.3703
(0.9677)
-1.9562*
(1.1775)
0.3098
(0.6458)
-0.9271***
(0.0959)
0.1229***
(0.0468)
-0.0096
(0.0591)
1.6990*
(0.9211)
684
0.537
125
684
684
684
684
0.540
0.542
0.538
0.537
125
125
125
125
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Note: These are the results of a fixed effects regression with county and period fixed effects are
included. The title for each column represents a different type of war.
22
VARIABLES
War(t)
War(t+1)
War(t)*War(t+1)
War(t-1)
Regulation(t-1)
Polity(t-1)
Net oda
Lngdpcap(t-1)
Observations
R-squared
Number of countries
Table 7: Scope of Government – Area 5 of EFW (War Types)
(1)
(2)
(3)
(4)
Loss
Win
Inter
Civil
-0.4745***
(0.1649)
-0.5356**
(0.2193)
0.2139
(0.4680)
0.1161
(0.1411)
-0.9732***
(0.0459)
0.0176
(0.0137)
0.0046
(0.0077)
-0.1362
(0.1430)
-0.1062
(0.1366)
-0.2337
(0.1615)
0.2867
(0.2365)
-0.1071
(0.1307)
-0.9684***
(0.0465)
0.0188
(0.0145)
0.0038
(0.0075)
-0.1680
(0.1409)
-0.2025
(0.1675)
-0.3571*
(0.1831)
0.1466
(0.2501)
-0.0296
(0.0994)
-0.9769***
(0.0452)
0.0200
(0.0147)
0.0043
(0.0071)
-0.1262
(0.1410)
-0.2561
(0.1605)
-0.4448**
(0.2004)
0.5128*
(0.2735)
0.0974
(0.1607)
-0.9775***
(0.0448)
0.0204
(0.0145)
0.0048
(0.0072)
-0.1369
(0.1392)
(5)
Inter
-0.0779
(0.1882)
-0.4794***
(0.1284)
0.4853*
(0.2487)
0.1109
(0.1494)
-0.9628***
(0.0447)
0.0199
(0.0145)
0.0048
(0.0069)
-0.1231
(0.1366)
614
0.55
109
614
614
614
614
0.54
0.55
0.55
0.55
109
109
109
109
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Note: These are the results of a fixed effects regression with county and period fixed effects are
included.
23
Table 8: Interaction between War a Polity IV
(1)
(2)
(3)
Polity IV Score
Gov. Size
Gov/GDP
Regulation
-10
-8
-6
-4
-2
0
2
4
6
8
10
-0.4562**
(0.2293)
-0.3775*
(0.1979)
-0.2987*
(0.1709)
-0.2199
(0.1506)
-0.1411
(0.1399)
-0.0624
(0.1410)
0.0164
(0.1538)
0.0952
(0.1756)
0.1740
(0.2036)
0.2527
(0.2356)
0.3315
(0.2701)
0.6392
(1.1957)
0.5446
(1.0781)
0.4500
(0.9644)
0.3555
(0.8562)
0.2609
(0.7560)
0.1664
(0.6674)
0.0718
(0.5954)
-0.0228
(0.5467)
-0.1173
(0.5279)
-0.2119
(0.5420)
-0.3064
(0.5866)
-0.4548**
(0.2320)
-0.4220**
(0.2043)
-0.3893**
(0.1787)
-0.3566**
(0.1560)
-0.3238**
(0.1377)
-0.2911**
(0.1257)
-0.2584**
(0.1219)
-0.2256*
(0.1271)
-0.1929
(0.1402)
-0.1602
(0.1594)
-0.1274
(0.1827)
Observations
682
684
614
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Note: This table reports the marginal effect of war across different values of the Polity IV index.
Each column represents a different dependent variable.
24