Helping Countries Develop
Chapter

18 Fiscal Consequences of Armed Conflict and Terrorism in Low- and Middle-Income Countries

Author(s):
Benedict Clements, Sanjeev Gupta, and Gabriela Inchauste
Published Date:
September 2004
Share
  • ShareShare
Show Summary Details
Author(s)
Sanjeev Gupta, Benedict Clements, Rina Bhattacharya and Shamit Chakravarti 

1. Introduction

Contrary to expectations, the end of the Cold War has not been a harbinger of peace. There has been a proliferation of armed conflicts around the world over the past dozen years. In particular, terrorist groups have become increasingly sophisticated, daring, and destructive. More than 4 million people are estimated to have perished in violent conflicts between 1989 and 2000, and 37 million people have been displaced as refugees, either inside or outside their countries (World Bank, 2000). In 2000, there were 25 major armed conflicts around the world, of which 23 were intrastate conflicts (SIPRI Yearbook 2001).1 International terrorist attacks increased from an average of about 342 a year between 1995 and 1999 to 387 a year between 2000 and 2001.2 Most of the armed conflicts and terrorist activities have taken place in low- and middle-income countries. Between 1996 and 2000, almost 70% of the major armed conflicts, more than 20% of all international terrorist attacks, and over 70% of all casualties due to such attacks took place in Asia and Africa.

While the literature has documented the economic costs of armed conflict and terrorism, a cross-country examination of their fiscal consequences is yet to be undertaken. Armed conflict and prolonged terrorist activities can strongly influence the revenues and expenditures of countries, and in turn affect their economic growth. Although armed conflict and terrorism are often treated as distinct phenomena, experience from different parts of the world shows that there is a close link between the two. This paper analyzes the effects of armed conflict and terrorism on fiscal balances and economic growth in low- and middle-income countries.

The remainder of this paper is structured as follows. Section 2 provides a brief overview of the literature, followed in Section 3 by a description of the channels through which armed conflict and terrorism can affect the fiscal accounts and economic growth. Section 4 sets out the methodology for the empirical analyses presented in the paper. Section 5 compares the evolution of various macroeconomic variables and socioeconomic indicators before, during, and after 22 episodes of armed conflict in a number of low- and middle-income countries. Section 6 estimates an integrated system of equations for real per capita income growth, government revenue, and government spending, to highlight the main channels through which armed conflict and terrorism affect the fiscal accounts. Section 7 concludes.

2. Review of the Literature

Several studies have assessed the economic costs of armed conflicts.3Richardson and Samarasinghe (1991) estimate that the total accumulated economic cost of the armed conflict in Sri Lanka in the five years between 1983 and 1988 was about U.S.$4.2 billion, or 68% of Sri Lanka’s GDP in 1988. Arunatilake et al. (2001) perform a similar exercise for a longer period and estimate that the conflict between 1983 and 1996 cost Sri Lanka about twice the country’s 1996 GDP. In a similar vein, several empirical-studies, based on different techniques, approaches, and data, have found an inverse relationship between different measures of political instability and violence on the one hand, and growth or investment on the other (Veneiris and Gupta, 1986; Barro, 1991; Alesina and Perotti, 1993 and 1996; Alesina et al., 1996; and Rodrik, 1999).

Armed conflict impacts on a country’s financial development. Addison et al. (2002) conclude that conflict can (1) adversely affect the process of financial deepening by undermining confidence in the domestic currency due to fear of inflation and depreciation; (2) encourage the movement of funds away from productive assets (bank deposits, capital) to nonproductive assets (gold); and (3) affect the regulation and supervision of the financial system. Their model, applied to 79 countries, shows that conflict significantly reduces financial development, and that the negative effect increases as conflict intensifies.

Prolonged terrorist activities, like armed conflict, also lower growth, both directly and indirectly. Abadie and Gardeazabal (2003) find that after the outbreak of terrorism in the 1970s, per capita GDP in the Basque region of Spain declined by about 10% relative to a “synthetic” control region, and that this gap widened in response to spikes in terrorist activity. Some studies have empirically assessed the impact of terrorism on tourism, both domestic and regional, and have found the expected negative effect (Drakos and Kutan, 2001; Enders and Sandler, 1991; and Enders, Sandler, and Parise, 1992). For example, in a study covering Greece, Israel, and Turkey, and using Italy as a “control variable,” Drakos and Kutan (2001) found that the intensity (measured by number of casualties) of terrorist incidents has significant domestic and cross-country effects on the market shares of the affected countries, and that there are significant contagion effects from terrorism within the region.

Terrorist threats raise the transaction costs of doing business and trade. Nitsch and Schumacher (2002) show that terrorist acts and large-scale violence adversely affected bilateral trade flows for more than 200 countries for the period 1960-93. A doubling of the number of terrorist incidents is associated with a decrease in bilateral trade by about 6%. Moreover, additional security measures put in place to deter terrorist attacks can impede the flow of goods and services. Walkenhorst and Dihel (2002) estimate the global welfare losses due to tighter security precautions that have been put in place following the attacks of September 11, 2001 at about US$75 billion.

As noted earlier, a cross-country examination of the fiscal consequences of armed conflicts is yet to be undertaken. However, recent case studies and related empirical studies of military spending and growth suggest channels through which armed conflict and terrorism can have an effect on fiscal accounts and economic growth.

3. Fiscal Effects of Armed Conflict and Terrorism: Potential Channels

Armed conflict and terrorism can affect the fiscal accounts by disrupting economic activities, eroding the tax base, lowering the efficiency of tax administration, and distorting the composition of public spending. Tax receipts, for example, vary with the health of the economy. Economic downturns due to insecurity and violence can lead to a decline in tax revenues. Beyond their effects on real activity, armed conflict and terrorism (especially if prolonged) can destroy part of the tax base (e.g., through the destruction of business firms,) and weaken the efficiency of tax administration. For example, Ndikumana (2001) notes that, following the outbreak of armed conflict in two countries in Africa, not only did the tax base collapse, but tax administration was also hampered. With the return of peace and the resumption of normal production in one of the two countries, tax revenues recovered progressively, and by 1998 exceeded the preconflict level.

Military expenditures typically increase in response to conflict and terrorism, and tend to remain high even after cessation of violence.4 Higher spending for security can also affect the composition of public spending by decreasing outlays for education, health, and other productive items. Moreover, the destruction of physical infrastructure and human capital due to violence, and the indirect effects on trade, tourism, and business confidence, all weaken the fiscal position and adversely affect economic growth, as noted earlier.

Defense spending can affect the long-run sustainable growth rate both negatively and positively (Shieh et al., 2002). First, there is a “crowding out effect,” whereby an increase in defense expenditures by the government reduces the resources available to the economy for private investment and for public spending on sectors that have a strong and positive impact on growth. Second, there is a “spin-off” effect from the positive supply-side spillover effects of defense expenditure on the nondefense sectors of the economy. This effect is likely to be small in low- and middle-income conflict-affected countries, since the majority of defense spending tends to be on imported armaments. Third, there is a “resource mobilization” effect on savings and investment: defense spending provides both internal and external security, and hence, boosts private savings and investment and attracts foreign investment. This has a positive effect on growth.5

Earlier studies have suggested that defense spending has a positive effect on economic growth in less-developed countries (Benoit, 1978). However, more recent empirical research shows that cutting military spending fosters economic growth (Arora and Bayoumi, 1993; Bayoumi et al., 1993; Knight et al., 1996). These papers argue that lower military spending can encourage growth by increasing capital formation and improving the efficiency with which resources are utilized in the economy. Cessation of conflict and terrorism can result in a “peace dividend,” releasing fiscal resources to be used for deficit reduction, lowering taxes, or raising the allocation for spending in social sectors.6

4. Empirical Methodology

The empirical analysis in this paper is based on two approaches. The first approach assesses the impact of armed conflict within conflict-affected countries, by examining the evolution of macrofiscal and socioeconomic variables before, during, and after 22 episodes of conflict in 20 low- and middle-income countries.7 The sample includes those episodes of armed conflict that either began or were ongoing in 1985 or later, and which ended by 1999, based on SIPRP’s definition of major armed conflicts.8

SIPRI draws data on armed conflicts from the Uppsala Conflict Data Project of the Department of Peace and Conflict Research, Uppsala University, Sweden. The Uppsala Conflict Data Project divides armed conflicts into the following three categories based on the level of casualties:

  • Minor armed conflict: At least 25 battle-related deaths a year and fewer than 1,000 battle-related deaths during the course of the conflict.
  • Intermediate armed conflict: At least 25 battle-related deaths a year and an accumulated total of at least 1,000 deaths but fewer than 1,000 in any given year.
  • War: At least 1,000 battle-related deaths a year.

SIPRP’s characterization of a major armed conflict covers the two most severe levels of conflict, i.e., “intermediate” armed conflict and war (Gleditsch et al., 2001). This paper does not include “minor” armed conflicts, since these are unlikely to have measurable effects on the fiscal accounts and the economic growth of the affected countries.

One shortcoming of the SIPRI index is that it applies an absolute criterion for the number of battle-related deaths. Thus, a country with a large population will be classified as being in conflict even though the number of deaths may be small relative to its population. Moreover, the number of battle-related deaths may not adequately capture the economic impact of armed conflict; it is possible that a number of sporadic, low-intensity incidents affecting mainly the local population will have a different impact on business and consumer confidence and international perception of risk in the country concerned than a single dramatic event affecting mainly the tourist sector or key sectors linked to foreign trade. Despite these drawbacks, the SIPRI index is broadly consistent with the conflict index produced by the Heidelberg Institute for International Conflict Research (HIIK).9

The second approach followed compares the economic consequences of armed conflict and terrorism across countries by estimating an integrated system of equations for real per capita income growth, government revenue, and government spending. The International Country Risk Guide (ICRG) ratings on internal conflict are used as a proxy for the combined risk from terrorism and conflict.10 The ICRG ratings provide an overall assessment of violence in a country due to civil war, terrorism, and civil disorder, and the actual or potential impact on governance. The highest rating is given to those countries “… where there is no armed opposition to the government and the government does not indulge in arbitrary violence, direct or indirect, against its own people.” The lowest rating is given to a country embroiled in an ongoing civil war and/or facing terrorist attacks. Given the difficulty of reaching a consensus on a universally acceptable definition of terrorism as well as of measuring terrorist activities, separate risk ratings for terrorism are not available. One advantage of the ICRG ratings is that they provide ratings of risk due to internal conflict and terrorism for a wide range of countries, and not just for those that have had major armed conflicts as defined by SIPRI.11 The SIPRI index of armed conflicts (proportion of each 5-year period during which there were armed conflicts) is used to check the robustness of the results. The SIPRI index has been used in other empirical studies, such as Davoodi et al. (2001).

5. Macroeconomic and Fiscal Variables and Socioeconomic Indicators: Preconflict, Conflict, and Postconflict Periods

The results from comparing the conflict, preconflict, and postconflict phases of 22 episodes of armed conflicts in lower- and middle-income countries are presented in Figures 1-5 and Table 1. The data on real GDP are consistent with the hypothesis of a significant pickup in growth in the immediate postconflict years. There is a dramatic pickup in inflation during the conflict period, followed by a significant decline in the immediate postconflict period (see Figures 1 and 2). The data show a notable increase in the share of gross fixed-capital formation to GDP in the immediate postconflict years, particularly in the private sector (see Figure 3).

Figure 1.Real GDP Growth in Conflict Countries1

(Average annual percentage change)

Sources: IMF, World Economic Outlook; World Bank, World Development Indicators 2001; and IMF staff calculations.

1 Based on a sample of 12 countries. The real GDP per capita growth corresponding to the preconflict, conflict, and postconflict periods are −1.1, −1.6, and 3.3 percent a year, respectively.

Figure 2.Consumer Price Inflation in Conflict Countries1

(Average annual percentage change)

Sources: IMF, World Economic Outlook; World Bank, World Development Indicators 2001; and IMF staff calculations.

1 Based on a sample of 9 countries.

Figure 3.Capital Formation in Conflict Countries1

(Percent of GDP)

Sources: IMF, World Economic Outlook; World Bank, World Development Indicators 2001; and IMF staff calculations.

1 Based on a sample of 17 countries for gross fixed capital formation, and on 11 countries each for gross public and private capital formation.

Figure 4.Fiscal Aggregates in Conflict Countries1

(Percent of GDP)

Sources: IMF, World Economic Outlook; World Bank, World Development Indicators 2001; and IMF staff calculations.

1 Based on a sample of 14 countries.

Figure 5.Composition of Government Spending in Conflict Countries1

(Percent of GDP)

Sources: IMF, World Economic Outlook; World Bank, World Development Indicators 2001; and IMF staff calculations.

1 Based on a sample of 12 countries for defense expenditure and on 6 countries each for education and health spending.

Table 1.Selected Social Indicators in Countries Experiencing Armed Conflictsa(Average annual rates of change, percent)
PreconflictbConflictbPostconflictbNumber

of Countries

for Which Data

Are Available
Life expectancy at birth, total (years)0.4−0.50.45
Mortality rate, infant (per 1,000 live births)c3.80.60.07
Gross primary enrollment rate2.63.29
Gross secondary enrollment rate1.12.19
Gross tertiary enrollment rate−1.52.19
Sources: IMF, World Economic Outlook; World Bank, World Development Indicators 2001; and authors’ calculations.

Combines all the low-income, lower-middle-income, and upper-middle-income countries affected by armed conflict as discussed in the paper. Countries are classified into income categories based on the World Bank’s criteria in terms of level of 1998 GNP per capita—low-income, US$760 or less; lower-middle-income, US$761 to US$3,030; and upper-middle-income, US$3,031 to US$9,360.

Conflict period refers to the period over which a country experienced armed conflict (as defined by SIPRI); preconflict refers to the average of three years preceding the conflict, and postconflict refers to the average of three years following the conflict (depending upon availability of data).

Positive rates of growth signify an improvement in the variable.

Sources: IMF, World Economic Outlook; World Bank, World Development Indicators 2001; and authors’ calculations.

Combines all the low-income, lower-middle-income, and upper-middle-income countries affected by armed conflict as discussed in the paper. Countries are classified into income categories based on the World Bank’s criteria in terms of level of 1998 GNP per capita—low-income, US$760 or less; lower-middle-income, US$761 to US$3,030; and upper-middle-income, US$3,031 to US$9,360.

Conflict period refers to the period over which a country experienced armed conflict (as defined by SIPRI); preconflict refers to the average of three years preceding the conflict, and postconflict refers to the average of three years following the conflict (depending upon availability of data).

Positive rates of growth signify an improvement in the variable.

Figures 4-5 show the evolution of fiscal variables over the preconflict, conflict, and postconflict periods. Due to data constraints, government revenue and foreign grants are used as a proxy for government revenue.12 The available data for the sample of countries show that the share of government revenue in GDP tends to fall during the conflict period, and to pick up somewhat in the immediate postconflict period. On the expenditure side, there appears to be a significant increase in government expenditure and net lending as a percent of GDP during the conflict period compared with the preconflict period, followed by a notable decline in the immediate postconflict period. In particular, the available data suggest high government spending on defense during the conflict period and in the period immediately preceding it, followed by a significant fall in the immediate postconflict period. However, high defense spending during the conflict period and in the years immediately preceding it tends to be at the expense of macroeconomic stability (as reflected for example in higher budget deficits and a pickup in inflation) rather than at the cost of lower spending on education and health as a share of GDP. Nevertheless, since conflict is associated with lower real GDP growth, the implication is lower growth in real per capita government spending on education and health during conflict periods.

Turning now to the socioeconomic indicators, Table 1 shows a significant decline in the rate of improvement of life expectancy at birth during the conflict period, but the trend for improvement in life expectancy picks up again in the immediate postconflict period. There is also a significant deterioration in the rate of improvement of infant mortality during conflict years, but the deterioration continues into the immediate postconflict period. The available data also show a marked improvement in gross enrollment rates (at all three levels—primary, secondary, and tertiary) following the end of armed conflict.

While a useful exercise, the conclusions drawn from the before-during-after analysis should be interpreted with caution. This analysis does not control for other factors that affect macroeconomic and fiscal outcomes, independent of armed conflict and terrorism, which may have also changed over the periods of violence. To isolate more rigorously the effects of conflict and terrorism, the following section presents the econometric estimation of a system of interlinked equations covering a wider range of countries, including those not affected by conflict and terrorism.

6. Econometric Estimates

As mentioned earlier, there are three main ways in which armed conflict and terrorism can affect the fiscal accounts: by influencing real economic activity (GDP) and therefore, government revenues; by adversely affecting both the tax base and the efficiency of the tax administration; and by changing the composition of government spending. These fiscal consequences can have repercussions on economic growth, which would further affect the public finances. To capture all these effects, a structural model with three equations is specified: the first for economic growth, the second for the ratio of government revenue to GDP, and the third for the composition of government spending measured by the share of defense spending in total government expenditure.

In the structural model, the equations for per capita income growth (equation (1)), government revenue to GDP (equation (2)), and defense expenditure as a share of total government spending (equation (3)) are specified as follows:

where

GRPCY = growth of real per capita income (GDP).

PCYINI = real per capita income (GDP, in US$) in the initial year of the sample period.

GSECINI = gross secondary school enrollment rate in the initial year of the sample period.

DEFEXPD = share of defense expenditure in total government spending.

AGEDEP = age-dependency ratio.

CONF = a conflict variable (discussed below).

INVGDP = total investment in percent of GDP.

GREVGDP = government revenue as a ratio of GDP.

PCY = real per capita income (in US$).

NONAGRX = share of nonagricultural exports in GDP.

AGRVA = agriculture value added in percent of GDP.

URBPOP = urban population as a share of total population.

DEFGDPN = (unweighted) average of neighboring countries’ ratio of defense spending to GDP.

αr, βr and λr are region-specific factors, and μ1i,t, μ2i,t and μ3i,t are the usual error terms. The subscript (i,t) for the main explanatory variables refers to country and time period, respectively. The endogenous variables in the system are the three dependent variables and the investment ratio. The model is estimated using 5-year averages of annual data for each country over four time periods: 1980–84, 1985–89, 1990–94, and 1995–99. Region and time dummies were included in the estimated equations.

Some authors have argued that conflict and terrorism are, in a sense, endogenous due to the possibility of reverse causation, i.e., that prolonged poor growth performance may help engender conflict. Violence and unrest may not only be a cause but may also arise from fluctuations in economic variables. Indeed, instrumental variable techniques have been used in some of the studies to correct for reverse causation, but the validity of instruments in cross-country regressions has been questioned by some authors (Abadie and Gardeazabal, 2003). However, given the difficulty of empirically modeling conflict and terrorism, and in finding suitable instruments, they are taken to be exogenous in line with a number of other studies (e.g., Davoodi et al., 2001; Gupta et al., 2001; Hess, 2003).

The above structural model was estimated using the Generalized Method of Moments (GMM) estimation technique so as to address the underlying problems of autocorrelation and heteroscedasticity that typically arise in estimating a structural panel model with endogenous variables. Following the standard approach, the instruments used in the estimation were all the exogenous variables of the structural model—i.e., all the variables in the system except for the three dependent variables and the investment ratio—plus an ICRG corruption index and inflation volatility. The latter two are used as proxies for the investment climate to instrument for the investment ratio.13 All of the results presented below pass the Sargan test for validity of the instrument set.

The data used in estimation of the structural model were taken from the IMF’s World Economic Outlook, the World Bank’s World Development Indicators 2001, Yearbooks of the Stockholm International Peace Research Institute, and the International Country Risk Guide. Due to the limited availability of time series data on tax revenues, data on revenues and foreign grants are used as a proxy for domestic government revenues.

Baseline Regressions

Model la (Table 2) uses the ICRG measure of internal conflict and terrorism. Note that a higher value of the ICRG conflict rating implies a lower risk of internal conflict and terrorism.14 As in the standard Barro growth equations (Barro, 1991), the coefficient on the initial level of per capita income is negative and statistically significant at the 5% level. However, the coefficient on the initial stock of human capital (proxied by the gross secondary school enrollment rate) is not statistically significant. The implication is that, at least for the sample of countries included in this study, convergence toward a common level of real per capita income is not dependent on the initial stock of human capital. The age dependency ratio is also not statistically significant, and neither is the ICRG rating for internal conflict and terrorism. Consistent with our hypothesis, the ratio of defense spending in total government expenditure has a negative and statistically significant effect on growth.

Table 2.Regression Results
Growth of Real Per Capita Income
Dependent VariableModel 1aModel 1b
Per capita income, initial−0.0004 (−2.33)**−0.0005 (−3.20)***
Gross secondary enrollment, initial−0.017 (−1.22)−0.02 (−1.40)
Ratio of defense spending to government expenditure−0.37 (−7.01)***−0.28 (−5.65)***
SIPRI rating for major armed conflicts−1.72 (−1.90)
ICRG internal conflict rating (civil wars and terrorism)−0.15 (−0.91)
Age dependency ratio−4.04 (−1.30)−6.84 (−2.22)**
Total investment0.17 (1.10)−0.01 (−0.05)
R-squared0.180.24
Revenue (in percent of GDP)
Dependent VariableModel 1aModel 1b
Real per capita income0.001 (2.76)***0.001 (2.70)***
Ratio of nonagricultural exports to GDP0.23 (4.85)***0.22 (4.75)***
SIPRI rating for major armed conflicts−0.34 (−0.30)
ICRG internal conflict rating (civil wars and terrorism)0.06 (0.30)
Agriculture value added−0.07 (−1.09)−0.09 (−1.38)
Urbanization0.02 (0.41)0.02 (0.42)
R-squared0.580.58
Defense Spending

(in percent of government spending)
Dependent VariableModel 1aModel 1b
Average defense spending of neighbors (in percent of GDP)0.97 (2.61)***1.26 (3.38)***
SIPRI rating for major armed conflicts2.68 (1.31)
ICRG internal conflict rating (civil wars and terrorism)−0.80 (−3.87)***
R-squared0.460.43
Number of observations137127
p-valuesa0.700.78
White’s heteroscedastic consistent t-statistics are in parentheses; (***), (**), and (*) denote significance at the 1%, 5%, and 10% levels, respectively.

The p-values refer to the test of overidentifying restrictions implied by the exogeneity of instruments. The instruments used are: corruption, inflation volatility, and all the exogenous variables in the system (i.e., all variables except for the three dependent variables and the ratio of total investment to GDP).

White’s heteroscedastic consistent t-statistics are in parentheses; (***), (**), and (*) denote significance at the 1%, 5%, and 10% levels, respectively.

The p-values refer to the test of overidentifying restrictions implied by the exogeneity of instruments. The instruments used are: corruption, inflation volatility, and all the exogenous variables in the system (i.e., all variables except for the three dependent variables and the ratio of total investment to GDP).

The structural equation for the government revenue-to-GDP ratio is based on studies such as Bahl (1971), Tanzi (1992), and Ebrill et al. (2001). The estimates are consistent with their findings that the share of government revenue in GDP in developing countries is a function of the level of development (proxied by real per capita income) and of the openness of the economy (proxied by the ratio of nonagricultural exports to GDP). However, the internal conflict and terrorism variable does not have any significant effect on the government revenue-to-GDP ratio, and neither does the structure of the economy (proxied by the ratio of value added in agriculture to GDP). One reason why stronger results are not obtained for this equation could be the inclusion of foreign grants in the measure of revenues; some of the structural variables explaining government tax revenues, for example, may not have an impact on grants in the same way.

The third equation for the share of defense in government expenditure is consistent with the finding in Davoodi et al. (2001) that higher spending on defense by neighboring countries—which could be interpreted as a measure of regional tensions—is associated with a significantly higher share of defense in total government spending. Moreover, the coefficient for internal conflict and terrorism is positive and statistically significant at the 1% level.15

In summary, the empirical results using the ICRG rating for internal conflict and terrorism suggest that violence and insecurity raise the share of defense spending in total government expenditure, which in turn has a negative effect on growth by diverting resources away from spending on sectors (education, health, infrastructure) that promote economic growth over the long term. The risk from conflict and terrorism does not seem to have any additional negative impact on growth, over and above its impact on the composition of government spending. Moreover, conflict and terrorism do not seem to have any impact on government revenue.

Robustness Tests

To assess the robustness of the results, the above model is reestimated using a different measure of conflict: the proportion of years during each 5-year period when the country was in conflict according to the SIPRI index. The results (Model 1b) tell a somewhat different story from the Model la estimates; the ratio of defense spending to government expenditure still has a statistically significant and negative effect on growth, but (unlike in Model 1a) the SIPRI-based conflict variable does not have a statistically significant impact on the composition of government spending. However, the SIPRI-based measure of armed conflict has a direct negative effect on growth which is statistically significant at the 10% level. In addition, the age dependency ratio now becomes statistically significant as well. In short, the results using the SIPRI-based conflict variable suggest that conflict has a direct negative impact on growth, rather than an indirect effect through the composition of government spending.

The results using the SIPRI-based conflict index may differ from those using the ICRG rating for internal conflict and terrorism because the former is discrete for any given year (either 0 or 1). By contrast the ICRG rating for internal conflict and terrorism is a more continuous variable, and varies from 0 to 12 with changes in the perceived risk from violence and insecurity.

Some authors have argued that ethnic fractionalization also has an impact on growth (e.g., Easterly and Levine, 1997). A variable measuring fragmentation, however, is not found to have a statistically significant effect (Table 3). Furthermore, the SIPRI-based measure of armed conflict remains a statistically significant determinant of growth, while the results using the ICRG measure of internal conflict and terrorism are broadly unchanged.

Table 3.Regression Results: Robustness Test
Growth of Real Per Capita Income
Dependent VariableModel 1aModel 1b
Per capita income, initial−0.0004−0.0006
(−3.52)***(−4.76)***
Gross secondary enrollment, initial−0.001−0.008
(−0.07)(−0.51)
Ratio of defense spending to government expenditure−0.36−0.27
(−8.74)***(−5.73)***
SIPRI rating for major armed conflicts−2.17
(−2.16)**
ICRG internal conflict rating (civil wars and terrorism)−0.13
(−0.86)
Age dependency ratio−3.58−6.79
(−0.92)(−1.80)*
Total investment−0.005−0.08
(−0.042)(−0.51)
Ethnic fragmentation0.010.01
−0.99−1.10
R-squared0.190.24
Revenue (in percent of GDP)
Dependent VariableModel 1aModel 1b
Real per capita income0.00080.0008
(1.81)*(1.91)*
Ratio of nonagricultural exports to GDP0.180.18
(4.74)***(4.48)***
SIPRI rating for major armed conflicts0.60
(0.60)
ICRG internal conflict rating (civil wars and terrorism)−0.29
(−1.75)*
Agriculture value added−0.08−0.09
(−1.33)(−1.49)
Urbanization0.080.05
(1.70)*(1.24)
R-squared0.510.49
Defense Spending
(in percent of government spending)
Dependent VariableModel 1aModel 1b
Average defense spending of neighbors (in percent of GDP)0.791.12
(2.30)**(2.79)***
SIPRI rating for major armed conflicts1.56
(0.57)
ICRG internal conflict rating (civil wars and terrorism)−0.67
(−2.58)**
R-squared0.440.34
Number of observations126127
p-valuesa0.740.96
White’s heteroscedastic consistent t-statistics are in parentheses; (***), (**), and (*) denote significance at the 1%, 5%, and 10% levels, respectively.

The p-values refer to the test of overidentifying restrictions implied by the exogeneity of instruments. The instruments used are: corruption, inflation volatility, and all the exogenous variables in the system (i.e., all variables except for the three dependent variables and the ratio of total investment to GDP).

White’s heteroscedastic consistent t-statistics are in parentheses; (***), (**), and (*) denote significance at the 1%, 5%, and 10% levels, respectively.

The p-values refer to the test of overidentifying restrictions implied by the exogeneity of instruments. The instruments used are: corruption, inflation volatility, and all the exogenous variables in the system (i.e., all variables except for the three dependent variables and the ratio of total investment to GDP).

7. Conclusions

The empirical literature on economic costs of armed conflicts and terrorism has yet to provide a comprehensive, cross-country examination of their fiscal consequences. This study provides a cross-country examination using two approaches. First, the evolution of various macroeconomic and fiscal variables and socioeconomic indicators during 22 episodes of conflict, and in the years immediately preceding and following the conflicts, was analyzed. Second, an integrated system of equations for real per capita income growth, government revenue, and government spending was estimated to examine the main channels through which armed conflict and terrorism affect the fiscal accounts.

The empirical results using the ICRG measure for internal conflict and terrorism are consistent with the hypothesis that armed conflict and terrorism lead to a higher share of defense spending in total government expenditure, which has a negative effect on growth by diverting resources away from spending on socially and economically productive sectors that promote economic growth. The results using the SIPRI-based conflict measure, however, suggest that conflict has a direct and significant negative impact on growth, rather than an indirect effect through its impact on the composition of government spending. The results using the SIPRI-based conflict index may differ from those using the ICRG rating for internal conflict and terrorism because the former is discrete for any given year (either 0 or 1). By contrast, the ICRG rating for internal conflict and terrorism is a more continuous variable, and varies from 0 to 12 with changes in the perceived risk from violence and insecurity.

The findings from the econometric estimation are generally consistent with the conclusions of the before-during-after conflict analysis. The share of government revenue in percent of GDP tends to fall during the conflict period, and to pick up somewhat in the immediate post-conflict period. This analysis also suggests that armed conflict leads to higher government spending on defense, but this tends to be at the expense of macroeconomic stability (reflected, for example, in significantly higher budget deficits and a pickup in inflation) rather than at the cost of lower spending on education and health—at least when measured as a percent of GDP. However, since conflict is associated with lower real GDP growth, the result is lower growth in real per capita government spending on education and health during conflict periods. Not surprisingly, the data are consistent with an increase in the share of investment in GDP in the immediate postconflict period, and in the share of private sector investment. The available data also show a dramatic pickup in inflation during the conflict period, followed by a significant decline in the immediate postconflict period.

The results suggest sizable economic gains in terms of economic growth, macroeconomic stability, and the generation of tax revenues to support poverty-reducing spending, for countries that end conflicts and tackle terrorism. Ending violence and restoring security can be expected to lower the share of the budget allocated to military spending. These results confirm those of earlier studies, underscoring the potential for the “peace dividend” to contribute to economic development. For example, a recent study by Hess (2003) finds that the pure economic welfare losses from conflict are quite large. The authors estimate that these losses are typically four times larger than the welfare costs of business cycles as calculated by Lucas (1987), and that, on average, individuals would give up over 6% of their current annual level of consumption as a one-time payment in order to live in a world of perpetual peace.

Successful reconstruction after conflict involves rebuilding damaged institutions and infrastructure, renewing the social contract, generating a sense of trust among the warring parties, and ensuring that grievances due to economic disparities or perceived biases in fiscal policies are addressed. All this takes time. The continued involvement (and not just one-shot assistance) of the donors and the international community is therefore critical, especially in countries that have experienced prolonged conflicts.16

International institutions (such as the IMF) have been involved in lending for reconstruction to postconflict countries. As part of its emergency assistance facility to help members emerging from conflicts rebuild capacity and recover economic stability, the IMF, for example, has provided US$300 million over the period 1995-2000 to seven postconflict countries. The findings of this paper have implications for the design of macroeconomic and fiscal policies for countries emerging from conflicts. In particular, the results suggest that conflict- and terrorism-affected countries are likely to experience a pickup in government tax revenues and a reduction in military spending (albeit with a lag) following the cessation of violence, and this would help in restoring macroeconomic stability.

Appendix A. Sample Countries

For the preconflict, conflict, and postconflict analysis, a sample of 20 countries (22 episodes of major armed conflicts based on SIPRI data) where conflict began or was ongoing after 1985, but ended by 1999, is used. The sample includes 15 low-income countries (Armenia, Azerbaijan, Bangladesh, Cambodia, Chad, the Republic of Congo, Georgia, Guinea-Bissau, the Lao People’s Democratic Republic, Mozambique, Nicaragua, Senegal, Tajikistan, Uganda, and the Republic of Yemen), 3 lower-middle-income countries (Albania, El Salvador, and Guatemala) and 2 upper-middle-income countries (Croatia and Lebanon).

Where ICRG country ratings on internal conflict are also available for the corresponding episodes (for 14 of the 20 countries), there is a broad match between low ICRG ratings of 8 or less (the lower the ICRG rating, the higher the risk of internal conflict) and countries that have been classified as conflict-affected by SIPRI.17 The average ICRG internal conflict score (where available) for these 20 countries is 3.7 between 1984 and 1989, 6.4 between 1990 and 1994, and 8.2 between 1995 and 1999. This is a reflection of the fact that in most of these 20 countries, the conflicts took place mainly during the 1980s (or before) and during the first half of the 1990s.

For the econometric analysis, a larger set of 66 countries including conflict and nonconflict, low- and middle-income countries is used (see list below). Of these 66, the following countries—Armenia, Azerbaijan, Croatia, El Salvador, Guatemala, Mozambique, Nicaragua, Senegal, and Uganda—were classified by SIPRI as countries experiencing major armed conflicts. Problems of data availability, which are particularly severe for countries affected by armed conflict, constrained the sample considerably.

Countries Used in the Econometric Analysis

Albania, Angola, Argentina, Armenia, Azerbaijan, Belarus, Bolivia, Brazil, Bulgaria, Cameroon, Chile, Colombia, Côte d’Ivoire, Croatia, Czech Republic, Ecuador, Egypt, El Salvador, Estonia, Ethiopia, Gabon, The Gambia, Guatemala, Honduras, Hungary, India, Indonesia, Iran, Jamaica, Jordan, Kazakhstan, Kenya, Latvia, Lithuania, Madagascar, Malawi, Malaysia, Mali, Mexico, Moldova, Morocco, Mozambique, Nicaragua, Niger, Nigeria, Oman, Pakistan, Paraguay, Philippines, Poland, Romania, Saudi Arabia, Senegal, Slovak Republic, South Africa, Sri Lanka, Syrian Arab Republic, Tanzania, Thailand, Tunisia, Turkey, Uganda, Uruguay, Venezuela, Zambia, and Zimbabwe.

References

    Abadie, A.,Gardeazabal, J.,2003. The economic costs of conflict: a case study of the Basque country. American Economic Review 93, 113–132.

    • Search Google Scholar
    • Export Citation

    Addison, T.,Murshed, S.M.,2001. The fiscal dimensions of conflict and reconstruction. WIDER Discussion Paper No. 2001/49. World Institute for Development Economics Research, United Nations University, Helsinki.

    • Search Google Scholar
    • Export Citation

    Addison, T.,Chowdhury, A.R.,Murshed, S.M.,2002. By how much does conflict reduce financial development?WIDER Discussion Paper No. 2002/48. World Institute for Development Economic Research, United Nations University, Helsinki.

    • Search Google Scholar
    • Export Citation

    Alesina, A.,Perotti, R.,1993. Income distribution, political instability, and investment. NBER Working Paper No. 4486. National Bureau of Economic Research, Cambridge, MA.

    • Search Google Scholar
    • Export Citation

    Alesina, A.,Perotti, R.,1996. Income distribution, political instability, and investment. European Economic Review 40, 1203–1228.

    Alesina, A.,Ozler, S.,Roubini, N.,Swagel, P.,1996. Political instability and economic growth. Journal of Economic Growth 1, 189–212.

    Arora, V.,Bayoumi, T.,1993. Economic benefits of reducing military expenditure, Annex II in World Economic Outlook. International Monetary Fund, Washington, DC.

    • Search Google Scholar
    • Export Citation

    Arunatilake, N.,Jayasuriya, S.,Kelegama, S.,2001. The economic cost of the war in Sri Lanka. Research Studies: Macroeconomic Policy and Planning Series No. 13. Institute of Policy Studies of Sri Lanka, Colombo.

    • Search Google Scholar
    • Export Citation

    Bahl, R.W.,1971. A regression approach to tax effort and tax ratio analysis. IMF Staff Papers 18, 570–612.

    Barro, R.J.,1991. Economic growth in a cross section of countries. Quarterly Journal of Economics 106, 407–443.

    Bayoumi, T.,Hewitt, D.,Schiff, J.,1993. Economic consequences of lower military spending: some simulation results. IMF Working Paper No. 93/17. International Monetary Fund, Washington, DC.

    • Search Google Scholar
    • Export Citation

    Benoit, E.,1978. Growth and defense in developing countries. Economic Development and Cultural Change 26, 271–280.

    Brunetti, A.,Weder, B.,1998. Investment and institutional uncertainty: a comparative study of different uncertainty measures. Weltwirtschaftliches Archiv/Review of World Economics 134, 513–533.

    • Search Google Scholar
    • Export Citation

    Collier, P.,Hoeffler, A.,2002a. Aid, policy and growth in post-conflict economies. Paper presented at a Joint World Bank-IMF Seminar, Washington, DC.

    • Search Google Scholar
    • Export Citation

    Collier, P.,Hoeffler, A.,2002b. Military expenditures: threats, aid, and arms races. Paper presented at a Joint World Bank-IMF Seminar, Washington, DC.

    • Search Google Scholar
    • Export Citation

    Davoodi, H.,Clements, B.,Debaere, P.,Schiff, J.,2001. Military spending, the peace dividend, and fiscal adjustment. IMF Staff Papers48 (2), 290–316.

    • Search Google Scholar
    • Export Citation

    Drakos, K.,Kutan, A.M.,2001. Regional effects of terrorism on tourism: Evidence from three Mediterranean countries. Center for European Integration Studies (ZEI), Working Paper No. 26. Rheinische Friedrich-Wilhelms-Universität, Bonn.

    • Search Google Scholar
    • Export Citation

    Easterly, W.,Levine, R.,1997. Africa’s growth tragedy: policies and ethnic divisions. Quarterly Journal of Economics 112, 1203–1250.

    • Search Google Scholar
    • Export Citation

    Ebrill, L.,Keen, M.,Bodin, J.P.,Summers, V.,2001. The Modern VAT. International Monetary Fund, Washington, DC.

    Enders, W.,Sandler, T.,1991. Causality between transnational terrorism and tourism: the case of Spain. Terrorism 14, 49–58.

    Enders, W.,Sandler, T.,Parise, G.F.,1992. An econometric analysis of the impact of terrorism on tourism. Kyklos 45, 531–554.

    Gleditsch, N.P.,Strand, H.,Eriksson, M.,Sollenberg, M.,Wallensteen, P.,2001. Armed conflict 1946-1999: a new dataset. Paper presented at the Euroconference on: Identifying Wars: Systematic Conflict Research and Its Utility in Conflict Resolution and Prevention. Uppsala University, Uppsala, Sweden.

    • Search Google Scholar
    • Export Citation

    Gupta, S.,de Mello, L.,Sharan, R.,2001. Corruption and military spending. European Journal of Political Economy 17, 748–777.

    Gupta, S.,Plant, M.,Dorsey, T.,Clements, B.,2002. Is the PRGF living up to expectations?Finance and Development 39, 17–20.

    Hess, G.D.,2003. The economic welfare cost of conflict: an empirical assessment. Working Paper, Department of Economics, Claremont McKenna College, CA.

    • Search Google Scholar
    • Export Citation

    Knight, M.,Loayza, N.,Villanueva, D.,1996. The peace dividend: military spending cuts and economic growth. IMF Staff Papers 43, 1–37.

    Lucas Jr., R.E.,1987. Models of Business Cycles. Blackwell, Oxford.

    Ndikumana, L.,2001. Fiscal policy, conflict, and reconstruction in Burundi and Rwanda. WIDER Discussion Paper No. 2001/62. World Institute for Development Economics Research. United Nations University, Helsinki.

    • Search Google Scholar
    • Export Citation

    Nitsch, V.,Schumacher, D.,2002. Terrorism and trade. Paper presented at the German Institute for Economic Research (DIW) workshop, The Economic Consequences of Global Terrorism. Available via the Internet: http://www.diw.de/deutsch/service/veranstaltungen/ws_consequences.

    • Search Google Scholar
    • Export Citation

    Richardson Jr., J.M.,Samarasinghe, de A.S.W.R.,1991. Measuring the economic dimensions of Sri Lanka’s ethnic conflict. In:Samarasinghe, de A.,Coughlan, R. (Eds.), Economic Dimensions of Ethnic Conflict. St. Martin’s Press, New York.

    • Search Google Scholar
    • Export Citation

    Rodrik, D.,1999. Where did all the growth go? external shocks, social conflict, and growth collapses. Journal of Economic Growth 4, 385–412.

    • Search Google Scholar
    • Export Citation

    Shieh, J.-Y.,Ching-Chong, L.,Wen-Ya, C.,2002. Endogenous growth and defense expenditures: a new explanation of the Benoit hypothesis. Defense and Peace Economics 13, 179–186.

    • Search Google Scholar
    • Export Citation

    Stockholm International Peace Research Institute, SIPRI Yearbook, 2001. Armaments, Disarmament and International Security. Oxford University Press, Oxford.

    • Search Google Scholar
    • Export Citation

    Tanzi, V.,1992. Structural factors and tax revenue in developing countries: a decade of evidence. In:Goldin, I.,Winters, L.A. (Eds.), Open Economies: Structural Adjustment and Agriculture. Cambridge University Press, Cambridge, UK.

    • Search Google Scholar
    • Export Citation

    U.S. Department of State, 2002. Patterns of global terrorism—2001 (Washington). Available via the Internet: http://www.state.gov/s/ct/rls/pgtrpt/2001/html.

    • Search Google Scholar
    • Export Citation

    Venieris, Y.P.,Gupta, D.K.,1986. Income distribution and sociopolitical instability as determinants of savings: a cross-sectional model. Journal of Political Economy 94, 873–883.

    • Search Google Scholar
    • Export Citation

    Walkenhorst, P.,Dihel, N.,2002. Trade impacts of the terrorist attacks of September 11, 2001: a quantitative assessment. Paper presented at the German Institute for Economic Research (DIW) workshop, The Economic Consequences of Global Terrorism. Available via the Internet: http://www.diw.de/deutsch/service/veranstaltungen/ws_consequences.

    • Search Google Scholar
    • Export Citation

    World Bank, 2000. Economic causes of civil conflict and their implications for policy. Press Briefing, June 15, Washington, DC. Available via the Internet: http://www.worldbank.org/html/extdr/extme/pr061500.htm.

    • Search Google Scholar
    • Export Citation

Reprinted from the European Journal of Political Economy, Vol. 20, Sanjeev Gupta, Benedict Clements, Rina Bhattacharya, and Shamit Chakravarti, “Fiscal Consequences of Armed Conflict and Terrorism in Low- and Middle-Income Countries,” © 2004, with permission from Elsevier.

The authors would like to thank Emanuele Baldacci, Hamid Davoodi, Stefano Fassina, Hong-Sang Jung, Mansoob Murshed, Erwin Tiongson, an anonymous referee, and participants of the DIW workshop for useful comments and suggestions.

1

The Stockholm International Peace Research Institute (SIPRI) publishes a yearly review of armaments, disarmament, and international security. A major armed conflict is defined in the SIPRI Yearbook 2000 as “a contested incompatibility that concerns government and/or territory over which the use of armed force between the military forces of two parties, of which at least one is the government of a state, has resulted in at least 1000 battle-related deaths over the duration of the conflict.”

2

Data on terrorist activities and casualties are drawn from a report prepared by the U.S. Department of State (2002). There is no consensus regarding how terrorism should be defined. Title 22 of the United States Code, Section 2656f(d) defines terrorism as “premeditated, politically motivated violence perpetrated against noncombatant targets by sub-national groups or clandestine agents, usually intended to influence an audience.” The Columbia Encyclopedia, 6th Edition, 2001, defines terrorism as “the threat or use of violence, often against the civilian population to achieve political ends. Terrorism involves activities such as assassinations, bombings, random killings, hijackings, and skyjackings. It is used for political, not military purposes, and by groups too weak to mount open assaults.”

3

Over and above the economic costs, prolonged armed conflicts can impose significant social and political costs that are difficult to estimate. For example, it is not possible to quantify the intangible costs of violence and insecurity, the human suffering and trauma, the breakdown in law and order, the animosity and mistrust that are created among warring parties, and the adverse effects of the reduced stock of health and education endowments on the long-run growth prospects of a country.

4

In Sri Lanka, for example, between 1983 and 1996, defense spending increased from 1.4% to 6% as a share of GDP, and from 4.4% to 21.6% as a share of total government spending (Arunatilake et al., 2001).

5

This effect is likely to be highly nonlinear: up to a certain basic level of spending on defense, there is a positive impact on savings and investment, but after this threshold is passed, higher government spending on defense is unlikely to promote further private sector savings and investment.

6

Conflict and violence can itself be affected by the perceived inequities in the distribution of the tax burden and in the pattern of public spending (Addison and Murshed, 2001).

7

Because of the problems of defining terrorism and of the sensitivity involved in classifying countries as victims or as perpetrators of terrorism, the preconflict, conflict and postconflict analysis is restricted only to countries that have experienced armed conflicts as defined by SIPRI. See Appendix A.xs

8

See footnote 1 for the definition of armed conflict used in this paper. Appendix A lists the sample countries for this as well as for the subsequent econometric analysis.

9

Unlike SIPRI, the Heidelberg Institute does not consider a cutoff level of 1,000 conflict-related deaths to classify a country as being affected by conflict. It defines conflict broadly as “the clashing of interests (positional differences) on national values and issues (territory, independence, self-determination, autonomy, ideology, power, resources) of some duration and magnitude between at least two parties (states, groups of states, organizations, or organized groups) that are determined to pursue their interests and win their case.”

10

Because more than 90% of all major armed conflicts since 1990 have been internal (SIPRI Annual Yearbooks), only the ICRG internal conflict rating is used in the econometric estimation. The ICRG ratings are compiled by a U.S.-based consultancy service, the Political Risk Services Group. Details are available via the Internet: http://www.prsgroup.com/index.html.

11

Where ICRG internal conflict ratings are available for the corresponding episodes of the conflict, preconflict, and postconflict analysis (for 14 of the 20 countries), there is a broad match between low ICRG ratings (of about 8 or less) and countries that have been classified as conflict-affected by SIPRI and HIIK. The average ICRG internal conflict score for these 14 countries is 3.7 between 1984 and 1989, 6.4 between 1990 and 1994, and 8.2 between 1995 and 1999.

12

Grants, on average, are much lower than revenue. For example, for a sample of 31 low-income countries with programs supported by the IMF since 1999, grants were only 3.5% of GDP, compared with revenue of about 18% of GDP (Gupta et al., 2002).

13

See, for example, Brunetti and Weder (1998), who find that economic volatility and corruption are detrimental to investment.

14

For the sake of brevity, the estimates of the time dummy and regional dummy coefficients are not presented in Table 2.

15

Collier and Hoeffler (2002b) find, based on data for the period 1960-1999, that military expenditure by a country is strongly influenced by the level of military expenditure of its neighbors. They estimate that an initial exogenous increase in military expenditure by one country is more than doubled in both the originating country and its neighbors. Potentially, there is an offsetting public good effect if rebellions are deterred by military expenditure. However, instrumenting for military expenditure, Collier and Hoeffler find no deterrence effect of military spending on the risk of internal conflict. Hence, there appears to be no regional public good effect offsetting the public bad arising from a neighborhood arms race.

16

This is emphasized by Collier and Hoeffler (2002a), who find that during the first three postconflict years, absorptive capacity on average is no greater than normal, but that in the rest of the first postconflict decade, it is approximately double its normal level. Thus, ideally, aid and donor involvement should be phased over several years following the end of the conflict. Collier and Hoeffler find that historically, aid has not been higher on average in postconflict societies, and indeed it has tended to taper off over the course of the decade following the cessation of conflict.

17

On the 0-12 ICRG scale, 0 denotes Very High Risk of Conflict and 12 denotes Very Low Risk. For example, Liberia had an average ICRG (internal conflict) rating of 2.1 between 1990 and 1994.

    Other Resources Citing This Publication