An important body of literature on civil war has examined its economic consequences. Most of these studies have focused on the adverse influence of civil war on economic growth. These studies emphasize “the double loss,” meaning that civil war depresses economic growth by causing destruction and diverting valuable resources from production to violence (Hoeffler and Querol 2005). Specifically, they have found that an increase in military spending, a decrease in domestic investments, rapid capital flight, the destruction of infrastructure, and the human losses from civil war generally lead to an economic downturn (Collier 1999, Hoeffler and Querol 2005; Imai and Weinstein 2000; Kang and Meernik 2005). Though these studies have made a substantial contribution to our understanding of the negative growth effects of civil war, there has been little systematic analysis of how civil war affects other important macroeconomic conditions, such as inflation.

In fact, soaring inflation is one of the most important features of economies engaged in civil war. According to Reinhart and Rogoff (2011), an inflation crisis is defined as an annual inflation rate above 20%, while hyperinflation is defined as an annual inflation rate of 500% or higher. The first recorded episode of hyperinflation in world history occurred during the French Revolution. Moreover, the two times the United States ever experienced hyperinflation were during the Revolutionary War and the American Civil War. More recently, war-torn South Sudan saw a 109.9% increase in its consumer price index in 2015, and during the 2013 civil war in Syria, the Syrian economy suffered a catastrophic increase in its annual inflation rate, to 292%. More generally, using civil war data from Escriba-Folch (2010) and inflation data from the World Bank, I found that the average inflation rate in countries engaged in civil war is approximately four times higher than that of countries at peace (see Table A-2 in the Appendix). I also found that the probability of an inflation crisis is 15 percentage points higher in countries experiencing civil war than in countries at peace (see Table A-3 in the Appendix). Furthermore, the probability of an episode of hyperinflation is 3 percentage points higher in wartime economies than in peacetime economies (see Table A-4 in the Appendix).

Despite this close relationship between civil war and inflation in general, there is remarkable cross-national variation in the levels of inflation experienced by countries engaged in civil war. That is, some countries have seen even more devastating and frequent inflation episodes than others during civil war. Using the same data above, I found that more than a third of countries engaged in civil war (35.8%) experienced inflation crises, while others were able to avoid these disastrous events. This finding provokes important questions: Why are some countries more prone to extraordinarily high inflation than others during civil war? More specifically, under what conditions are these inflation crises more or less likely? This article seeks to systematically examine these questions, with a focus on the political and institutional conditions of countries engaged in civil war.

To theorize the factors that significantly affect the likelihood of an inflation crisis during civil war, I start with a basic premise of Friedman. He explains that “inflation is always and everywhere a monetary phenomenon in the sense that it is and can be produced only by a more rapid increase in the quantity of money than in output” (Friedman 1996, 17). Based on this fundamental premise, I contend that weak government capacity is the most important cause of the abnormally high money supply that produces inflation crises during civil war. More specifically, I suggest that governments’ inability to collect taxes during civil war plays an important role in inflation crises by increasing their need to resort to an inflation tax to finance the war.

Moreover, given that people’s expectations of future inflation in their country are another crucial factor that can determine the actual level of inflation in a self-fulfilling prophecy, I explore the conditions that cause high levels of expected inflation during civil war (Blinder 2000). I suggest that the credibility of the government’s commitment to low inflation can substantially influence the market's expectation of future inflation during civil war. In particular, I explain that when a stable democratic regime and a fixed exchange rate system are maintained during civil war, the government can maintain strong credibility in terms of price stability, thereby reducing the likelihood of an inflation crisis. Lastly, I suggest that certain characteristics of civil war are directly or indirectly related to the likelihood of an inflation crisis. I focus on a civil war’s destructiveness and its duration. I expect that the more violent the civil war, the more likely that a country will experience an inflation crisis. This is because a more violent civil war requires governments to make greater military expenditures and at the same time weakens their ability to collect taxes, increasing their need to rely on an inflation tax as a source of revenue. In the third section of this article, I also discuss some mixed theoretical expectations as to how the duration of civil war affects the likelihood of inflation crises.

My empirical analysis, which examines 33 countries that experienced 35 civil war episodes from 1975 to 1999 using civil war data from Escriba-Folch (2010), demonstrates that governments’ inability to tax is positively associated with the probability of an inflation crisis. Moreover, it shows that both the collapse of state authority and the breakdown of a democratic regime significantly increase the likelihood of an inflation crisis. Finally, it shows that a low level of violence and a long duration are negatively correlated with the occurrence of an inflation crisis during civil war.

This article makes a few important contributions to the literature. First, it addresses an important gap in the literature by systematically illuminating the determinants of devastating inflation during civil war, a topic to which the existing literature has paid little attention. Explaining the causes of inflation crises during civil war is important, given the significant harms they cause to wartime economies. As Reinhart and Savastano (2003) have suggested, inflation crises seriously undermine the credibility of national currencies. When a national currency cannot function properly as a store of value, a unit of account, and a medium of exchange due to high inflation, the market economy that operates according to the price system cannot work efficiently. Moreover, when the values of national currencies plummet due to inflation crises, capital flight and depletion of bank deposits are more likely to occur, in turn raising the likelihood of other collateral damage such as banking crises or exchange rate crises. When countries experience civil war, all these negative economic consequences resulting from inflation crises worsen their suffering. Moreover, I argue that the existing literature’s accounts of civil war’s adverse effects on economic growth are incomplete without a systematic explanation of the mechanism underlying high inflation during civil war because as Kang and Meernik (2005) suggest, it is one of the most important factors contributing to dismal economic growth during civil war.

Second, this article offers important policy implications that may be used to prevent the economic disruption caused by serious price instability during civil war, in that it identifies the political and institutional conditions under which inflation crises are likely. The main findings of this article are that maintaining a democratic regime and governments’ ability to tax can significantly reduce the likelihood of an inflation crisis. Moreover, this article suggests that adopting a fixed exchange rate system and sustaining it during civil war are crucial to maintaining price stability. Other than Kang and Meernik (2005) and Flores and Nooruddin (2009), most studies have focused exclusively on the economic costs that accrue from the destructiveness of civil war, an aspect of such a conflict that is usually beyond governments’ direct control. This article sheds light on the ways government policies can ameliorate these economic harms.

The article is organized as follows. First, I review previous studies that have examined the impacts of civil wars on economic growth and identify their limitations. Second, I establish my own theories as to the causes of inflation crises during civil war and propose hypotheses based on my theoretical expectations. Then, I test these hypotheses using civil war data collected by Escriba-Folch (2010). Finally, I draw conclusions and suggest a few important policy implications.

Most studies that have examined the economic consequences of civil wars have focused on such wars’ impacts on economic growth. The pioneering work in this literature is a study by Knight et al. (1996). They suggest that a sharp increase in military spending during civil war negatively affects economic growth by reducing government spending on more economically productive uses (i.e. investments). Through a simulation analysis, they found that a 1% increase in military expenditures during civil war is associated with around a 1% permanent loss of GDP. Given that Collier and Hoeffler (2002) found that on average governments' military spending during civil war increases by 2.2% of GDP, a simple estimation suggests that countries experiencing civil war tend to permanently lose about 2% of their GDP just due to this increase in military spending. Furthermore, considering that increased military spending often persists even a decade after the end of civil war, such war’s negative influence on economic growth can continue well into the postwar period (Collier, Hoeffler and Pattillo 2001).

Moreover, Imai and Weinstein (2000) examine another causal mechanism connecting civil war and economic depression. They argue that civil war is negatively associated with economic growth because it causes a reduction in capital stock. Specifically, they explain that civil war reduces capital stock by lowering its value and facilitating capital flight. There is a substantial amount of evidence that civil war is devastating to both immobile and mobile capital stock. For example, Mozambique’s civil war destroyed nearly 40% of the country’s immobile capital stock in the agricultural, communications and administrative sectors (Hoeffler and Querol 2005). Based on survey data, Stewart and Matovu reported that Ugandan households lost around two-thirds of their assets during the country’s civil war (Stewart and Matovu 2000). Collier, Hoeffler, and Pattillo (2001) found that civil war, on average, causes approximately 10% of a country's total capital stock to move abroad, thereby seriously undermining the country’s productive capacity.

While the above studies examine the overall effects of civil war on economic growth, Kang and Meernik (2005) and Flores and Nooruddin (2009) focus on cross-national variation in the negative growth effects of civil war. Kang and Meernik (2005) point out that the characteristics of civil war and the varying political conditions of countries engaged in civil war can significantly interact with a civil war’s impact on economic growth. First, with regard to civil war’s characteristics, they suggest that the economic consequences of civil war may differ according to its level of destructiveness and who wins. Kang and Meernik also found that political conditions such as regime type, the degree of ethnic fractionalization in a country, and the extent of the government’s fiscal space can shape the degree to which civil war dampens economic growth. On the other hand, Flores and Nooruddin (2009) emphasize countries’ varying levels of economic recovery after civil war, focusing on the role of postconflict political institutions. They found that successful economic recovery after civil war depends on the government’s ability to credibly commit to postconflict peace. Specifically, they suggest that due to institutional immaturity and the political fragility of postconflict democratization, countries may be prone to a recurrence of domestic conflicts, leading to a slow and weak economic recovery after civil war.

Overall, these existing studies have made a significant contribution to the civil war literature by shedding light on the negative relationship between civil war and economic growth and the specific mechanisms underlying it. However, little attention has been paid to civil war’s effects on other crucial aspects of the economy, such as price stability. With the exception of a few studies that have focused exclusively on inflation crises in the United States resulting from the unusually expansionary macroeconomic policy that occurred during its civil war, no study, to my knowledge, has attempted to systematically examine the causal mechanism of the relationship between civil war and inflation. Given the fundamental role of stable prices in sustaining the market economy, I suggest that this gap in the literature warrants a systematic exploration.

In this section, I establish a theoretical framework that links various political, economic, and institutional factors to inflation crises during civil war. I also explain how civil wars' characteristics are associated with these crises.

Government capacity

1) Tax capacity

I argue that a government’s inability to raise and collect taxes during civil war increases the likelihood of an inflation crisis. Governments can raise revenue in two ways: collecting more taxes or printing more currency. Seigniorage is the revenue a government generates by printing money. Under normal circumstances, it contributes only a small portion to a government’s budget. However, when highly unusual circumstances such as wars compel governments to spend more resources than they can afford based on taxation or access to capital markets, they must resort to using their central banks’ printing presses. After World War I, for example, the German central bank used “more than 30 dedicated paper mills, 29 plate manufacturers, and 132 printing plants” to finance the government's budget deficits (Frieden 2007, 135). It issued 100 trillion D-marks in one day, equivalent to \$312.50. During the American Civil War, both the Union and the Confederacy printed money to finance their war expenses.

When governments dramatically increase the quantity of currency in circulation, the prices of goods and services increase, leading to a decrease in the amount of goods or services that a unit of currency can buy. That is, when the supply of currency in the economy increases, the real value of those currencies decreases, and with it people’s purchasing power. This phenomenon is called an inflation tax, that is, “the reduction in the value of the money held by the public, by printing money to cover governments’ budget deficit and creating inflation” (Krugman 2021, 489).

The level of the inflation tax is directly related to the extent of the seigniorage a government intends to generate. The more money the government prints to finance wartime spending that is not covered by tax revenue, the greater the inflation tax the people have to pay and the higher the level of inflation in the economy. Therefore, in the context of civil war, I expect that governments that lack the ability to raise and collect taxes to finance their military expenditures are more likely to rely on an inflation tax, thereby increasing the likelihood of an inflation crisis.

H1 [Government Capacity]: The weaker a government’s ability to collect taxes during civil war, the greater the likelihood of an inflation crisis.

Government credibility

I also argue that political and institutional conditions that can determine a government’s ability to sustain a credible commitment to low inflation during civil war can significantly affect the price stability of the wartime economy. Inflation levels depend not only on actual levels of money growth but also on the market’s expectations about future inflation. Expected inflation is the rate of inflation that market participants expect in the near future. The standard macroeconomic theory includes expected inflation as one of the most important determinants of actual inflation, as suggested by the following equation:

π = Eπ − β(μ − μⁿ) + ν

In this equation, π is actual inflation, Eπ is the market’s expected inflation, μ is actual unemployment, is the natural rate of unemployment, is cyclical unemployment, β is a parameter “measuring the response of inflation to cyclical unemployment,” and ν is a supply shock (Mankiw 2010, 418). This equation clearly shows that expected inflation is directly related to actual inflation. Therefore, when higher inflation is expected by the market, it is more likely to actually occur. This is the reason inflation is usually considered a self-fulfilling prophecy (Gali 2015, Heymann and Leijonhufvud 1995).

During civil war, it is highly likely that market participants will expect extraordinarily high inflation in the near future due to the a dramatic increase in military spending (a demand shock) and the severe shortage of goods and services (a supply shock). The expectation of unusually high inflation then leads to unusually high actual inflation. For example, during civil war, most firms can expect that the prices of raw materials or intermediate goods will increase due to a shortage of those productive factors caused by disruptions to production and transportation. Because most firms expect that the costs of production will increase, they raise the prices of their goods to compensate for the increase in those costs. When most firms in a wartime economy behave this way, unusually high inflation will follow. Therefore, during civil war, it is particularly important for governments to effectively manage the market’s inflation expectations. To keep inflation low, governments must credibly commit to maintaining price stability in the eyes of the market. I expect that under conditions of civil war, there may be significant variations in governments’ abilities to credibly commit to keeping inflation low and thus to stably manage the public’s inflation expectations. In the following paragraphs, I explain institutional and political conditions that can affect governments' abilities to make credible commitments to maintaining price stability during civil war and ultimately to influence the likelihood of an inflation crisis.

1) Collapse of the state and democratic regime breakdown

I argue that the collapse of state authority or the fall of a democratic regime during civil war dramatically increase the probability of an inflation crisis by undermining the government’s ability to send a credible signal to the market that it will maintain low inflation. Marshall, Gurr, and Harff (2019) describe the collapse of the state (or failure of the state) due to civil war as the central state losing its ability to maintain authority or political order in most of its territory and the failure of its institutions to function properly. On the other hand, the breakdown of a democratic regime is defined as the replacement of a democratic or quasi-democratic regime with autocratic political institutions through the use or the threat of force.

First, unstable political environments created by either the collapse of the state or a democratic regime breakdown during civil war can make the government’s commitment to low inflation untrustworthy in the eyes of the market since disciplined monetary policy and effective enforcement of tax laws can hardly be expected under these situations. As Alesina and Tabellini (1987) suggest, extreme political instability can shorten the time horizon of governments, leading them to embrace expansionary macroeconomic policy that may help achieve their short-term political goals (e.g. greater military spending or higher economic growth) but may also produce undesirable economic outcomes such as high inflation or currency depreciation in the long term. Moreover, when a government’s authority and control are seriously weakened due to civil war, engaging in tax collection and preventing tax evasion are fairly demanding. As the collapse of state authority weakens a government’s tax capacity, I expect that governments will be more likely to use inflation tax in order to finance their military, thereby increasing the likelihood of an inflation crisis. Finally, the fact that under such extreme political turmoil, governments with a short time horizon and weakened tax capacity are more likely to generate high inflation will naturally lead people to expect particularly high inflation in the near future. This concern about higher inflation in the future will lead to a self-fulfilling prophecy, high actual inflation. Thus, both the collapse of the state and the breakdown of a democratic regime can lead to high inflation purely via people’s expectations.

Second, the collapse of the state can make it more difficult for governments to provide credible information from which market participants can derive a common expectation of low inflation. That is, when a government’s seriously weakened bureaucratic system cannot offer even the most basic information about the direction of their macroeconomic policy or economic forecasts, the market will not develop homogeneous expectations as to price stability. Morris and Shin (1998) have demonstrated that even when a country’s actual economic fundamentals are sound, the market may lose trust in the value of the national currency and thus future price stability if significant informational discord emerges among market participants in terms of their beliefs about macroeconomic conditions. When stable market expectations of future inflation cannot form due to a disarray in government institutions, and uncertainty regarding future inflation emerges, wage setters in the economy are incentivized to competitively raise their nominal wages to avoid a decrease in the real value of wages due to unexpectedly high inflation in the future. In such situations, the pure expectation of high inflation may lead to actual high inflation even if the actual economic conditions are relatively sound. Moreover, the lack of credible information about future inflation may unreasonably increase expected inflation since in the midst of the uncertainty created by civil war, the market may overestimate the negative effects of an increase in military expenditures and the government’s incentive to use an inflation tax on future inflation.

Third, when democratic regimes fall and authoritarian governments take over by force during civil war, those governments will have great difficulty making credible commitments to low inflation since authoritarian regimes do not provide transparent information to the market. As Broz (2002) explains, democracies are generally superior in transparency since decisions are made openly in an environment where plural interests compete. Under democratic regimes where multiple actors with conflicting preferences have strong incentives to verify the information offered by governments, information about economic conditions and the economic forecast will be highly transparent and, thus, credible. Persson, Roland, and Tabellini (1997) also emphasize that the separation of powers among government branches in democracies plays an important role in providing voters with better information.

Civil war generates unusually high uncertainty regarding the direction of macroeconomic policy and the overall conditions of the economy. Governments are likely to engage in unpredictably expansive macroeconomic policy as war expenses fluctuate according to new developments (e.g., which side is winning the war). Moreover, the levels of production and consumption in the economy may change rapidly as a war progresses. Given the strong advantages of democratic regimes in terms of their provision of transparent economic information, I expect democratic regime breakdown to seriously increase economic uncertainty created by civil war. Moreover, given the opaque policymaking of authoritarian regimes, a violent transition to an authoritarian regime would destroy confidence in the government’s economic policies and information provision. Therefore, when a democratic regime is broken down by force during civil war, the government’s commitment to low inflation cannot be sufficiently maintained. The market’s expectations about future inflation are destabilized, and the probability of an inflation crisis increases. Finally, as Bodea and Hicks (2015) suggest, because authoritarian governments usually lack checks and balances and do not operate according to the rule of law, they are more likely to interfere with the monetary policy of independent central banks whose primary objective is to maintain low inflation. Thus, I expect that an inflation crisis is more likely when a democratic or a quasi-democratic regime is replaced with an authoritarian regime. Furthermore, because the market can reasonably expect that an authoritarian government will manipulate the central bank’s monetary policy for their short-term political interests (seigniorage or economic growth via money creation), the levels of inflation expected by the market under an authoritarian regime will likely be higher than under a democratic regime. To summarize, I expect that both the collapse of the state and the fall of a democratic regime during civil war are likely to increase the probability of an inflation crisis by seriously undermining the government’s ability to send a credible signal to the public that it will maintain low inflation.

H2 [Government Credibility]: The collapse of the state during civil war makes the occurrence of an inflation crisis more likely.

H3 [Government Credibility]: A democratic breakdown during civil war makes the occurrence of an inflation crisis more likely.

2) Fixed exchange rate system

I suggest that a de jure fixed exchange rate regime that is maintained during civil war reduces the likelihood of a wartime inflation crisis. A de jure fixed exchange rate regime, where it is legally required that the currency value, as determined by the IMF, is sustained within ±1 percent (up to 2%) of a fixed rate, has been widely acknowledged as one of the most effective ways to demonstrate a commitment to low inflation (Bernhard and Leblang 1999; Ghosh et al. 1997; Leblang 1999). This type of system tends to lower inflation levels in the economy by disciplining a government’s macroeconomic policies (i.e. restraining money growth) and sending a clear signal to the market that the currency value is and will remain stable (Simmons 2020, Eichengreen 1992).

Two important aspects of fixed exchange rate systems provide credibility. First, they are the most transparent way for governments to commit to price stability, since a fixed exchange rate is the most visible indicator of the stable value of money, thus keeping inflation low (Broz 2002). The public can easily monitor whether a government’s promise of prudent macroeconomic policies and low inflation is being kept just by observing whether the currency peg is sustained or not. Second, the collapse of a fixed exchange rate is politically costly (Bernhard 1998; Edwards 1996). Because the transparency of a fixed exchange rate system makes it easy for the public to monitor whether the government is maintaining its commitment to price stability, the public can easily hold the government accountable when the commitment is broken. Given the crucial role of the exchange rate in the economy, the abrupt collapse of a fixed exchange rate system can generate the public perception that the government is incompetent or untrustworthy, damaging it politically. In particular, a disruption in trade or a sharp decrease in the real value of financial assets created by the sudden breakdown of a fixed exchange rate system can seriously undermine electoral support for incumbents.

The political cost of breaking a commitment to a fixed exchange rate has been well demonstrated empirically. According to Cooper (1993), the likelihood that an incumbent government will be defeated in an election doubles when there is a substantial reduction in the value of the national currency. He also found that a sudden depreciation also increases the probability that a finance minister will lose their job. Because the collapse of a fixed exchange rate regime is so politically costly, a government’s decision to fix the exchange rate is considered a costly signal to the public that the government will strive to maintain the exchange rate at all costs. Given that the more costly the signal, the more credible it seems to its audience, a fixed exchange rate system can provide governments an effective way to make a credible commitment to low inflation (Fearon 1994). I suggest that maintaining a fixed exchange rate system is particularly important when it comes to preventing an inflation crisis during civil war. Most importantly, its simplicity and clarity as a monetary anchor can provide an informational focal point on which the market’s expectation of low inflation can converge during wartime when it is difficult for the government to deliver reliable information to the market.

H4 [Government Credibility]: When a country maintains a de jure fixed exchange rate system, an inflation crisis is less likely to occur.

Characteristics of civil war

1) Destructiveness of civil war and its duration

I suggest that when civil war is highly destructive, an inflation crisis is more likely to occur. Both intense violence and large-scale civil war can directly or indirectly cause an inflation crisis, in the latter case by making the government incapable or unstable. First, I start with the simple assumption that the more destructive the civil war, the more difficulty the government will have financing its military expenditures. Governments facing more violent and powerful rebel groups would be expected to spend more heavily on the military, giving them a stronger incentive to use an inflation tax to finance these military expenditures. Moreover, as devastating and widespread rebel attacks dramatically reduce economic production, thus decreasing household income and the value of assets, a country’s tax base would shrink. This would also increase the attractiveness of an inflation tax as an alternative source of revenue. Finally, if a government’s administrative system is weakened by a highly destructive war, its efforts to collect taxes and to prevent tax evasion would be limited, significantly decreasing its tax capacity and again, creating more incentive for those governments to rely on an inflation tax. To summarize, I suggest that the necessity for substantial military spending combined with a weak tax capacity resulting from intense domestic conflict is likely to generate an inflation crisis by strongly incentivizing governments to rely on an inflation tax.

Second, I argue that both the level of violence and the scale of civil war can directly affect the probability of an inflation crisis via impacts on public perceptions of expected inflation. It is likely that the more destructive the civil war, the greater future inflation expected by the public, since a more violent war affecting a larger territory implies a greater government incentive to use an inflation tax to increase military expenditures and weaker tax capacity, both of which can be reasonably expected as the public observes that the civil war has increased in intensity or size. A highly destructive civil war can also cause the public to lose trust in the national currency since large-scale rebel violence and the inability of an incumbent government to suppress it might increase the expectation that the incumbents may eventually be defeated by the rebels. Because a rebel victory and the total collapse of the state might lead to a situation in which the current national currency stops functioning as the official legal tender, people in countries experiencing a destructive civil war might consider their currency less valuable. When a national currency loses significant value, the price of goods and services will dramatically increase, increasing the probability of an inflation crisis.

In terms of the effects of a civil war’s duration, I offer mixed theoretical expectations. On one hand, I expect that as a civil war endures, the accumulation of wartime money supply will increase the probability of an inflation crisis. Thus, the longer the civil war, the higher the probability of an inflation crisis. On the other hand, as Kang and Meernik (2005) argue, a lengthy war might provide “more time and opportunity for an economy to adjust” (93). That is, in the long term, market participants in acivil war economy may have sufficient time and experience to adjust to or settle into a ‘new normal’ in which they can develop more consistent and stable expectations of the economy and future inflation levels. As a result, it may be more likely that expectations of low inflation emerge and converge. Moreover, according to Flores and Nooruddin (2009), it is likely that during persistent domestic conflicts, market participants may “resume activity even while the conflict continues” (13). I suggest that if the long duration of a civil war allows production and transportation of goods and services to recover, even to an imperfect extent, thereby increasing the supply of those goods and services, their prices will decrease. Thus, the resumption of normal economic activities is expected to reduce the probability of an inflation crisis. Furthermore, as civil war comes to an end, governments may be able to normalize their macroeconomic policy in a way that prioritizes price stability. A gradual decrease in military spending would not only allow a government to achieve a more sound fiscal policy but would also reduce its incentive to use an inflation tax. In summary, a more destructive civil war is highly likely to increase the specter of an inflation crisis, while a longer duration of the conflict may either increase or decrease the probability of an inflation crisis.

H5 [Civil War Characteristics]: The more destructive a civil war, the greater the likelihood of an inflation crisis.

H6 [Civil War Characteristics]: The longer a civil war, the greater the likelihood of an inflation crisis.

Data and Method

The objective of my empirical analysis is to test my theoretical arguments as to the relationship between the varying political and institutional conditions of countries engaged in civil war and the likelihood of an inflation crisis. To test my hypotheses, I use civil war data collected by Escriba-Folch (2010). His data were compiled from Fearon and Latin (2003), Humphreys (2005), and Fearon (2004, 2005). In his data, a domestic conflict is considered a civil war if it satisfies three major criteria: (i) the conflict should be between state agents and organized non-state groups; (ii) the number of deaths in the conflict must exceed at least 1,000 (with a yearly average of at least 100); and (iii) both sides must have suffered at least 100 deaths from the conflict. His dataset contains 87 civil war episodes and 63 countries. However, the number of countries used in my empirical analysis was limited based on the availability of various economic variables, most importantly inflation rates. My dataset consists of 33 countries and 35 episodes of civil war from 1975 to 1999 (my dataset lost 42 episodes of civil war from the total of 77 during this particular period). The country-years reflected in my dataset total 348. In the Appendix, Table A-1 lists the sample countries used in my empirical analyses.

The dependent variable in my analysis is a dichotomous variable that measures the presence or absence of an inflation crisis in a particular year of a civil war. I coded an episode as an inflation crisis if a country’s annual inflation rate exceeded 20%, following the definition of Reinhart and Rogoff (2011). The inflation rate data are from the World Bank Database. I have six independent variables. First, as a measure of a government’s ability to tax, I use its total tax revenues as a percentage of GDP. I assume that the lower the total tax revenues collected by a government, the lower its ability to tax. Thus, I expect that this variable (Tax Rev. in Model 1) will have a negative coefficient because I hypothesize that reduced ability to tax is associated with a higher probability of an inflation crisis.

In order to measure the degree to which there is a state collapse or a democratic regime breakdown, I use State Failure Problem Set data provided by the Political Instability Task Force (Marshall et al. 2019). According to Marshall et al., the collapse of the state (Failure of State in Model 2) is defined as “situations in which the institutions of the central state are so weakened that they can no longer maintain authority or political order in significant parts of the country.” Democratic regime breakdown (Democratic Breakdown in Model 3) is defined as “situations in which democratic or quasi-democratic institutions are weakened or replaced, through the use or threat of force, by autocratic political institutions.” The values of both variables range from 0 to 4. The higher the values, the greater the seriousness of the failure. Therefore, I expect that both variables will have positive coefficients.

I use a dichotomous variable to indicate the presence or absence of a fixed exchange rate system. If a country's exchange rate system features either a currency board or a traditional currency peg as classified by the IMF, it is categorized as having a fixed exchange rate system and coded one. Otherwise, it is coded zero. These data are from the IMF annual report on exchange arrangements and exchange restrictions. I expect that this variable (Fixed Exchange Rate in Model 4) will have a negative coefficient since my hypothesis suggests that the existence of a fixed exchange rate system decreases the probability of an inflation crisis. As measures of the destructiveness of civil war and its duration, I use the logarithm of the annual number of deaths during civil war (Log death in Model 5) and the number of years since the onset of civil war (Duration in Model 6). The data for both variables are from Escriba-Folch (2010). While I expect that the former will have a positive coefficient, the sign of the coefficient of the latter may be either positive or negative as I have offered competing theoretical expectations as to the effects of civil war’s duration on the incidence of inflation crises.

In addition to the six independent variables, I use several variables to control for other factors that could affect the probability of an inflation crisis. To select the control variables, I followed the model specifications of the empirical analysis by Bodea and Hicks (2015), which examines the determinants of cross-country variation in inflation rates. First, I include the log of GDP per capita (Log GDPPC) since the existing literature on inflation suggests that countries that are less economically developed tend to have higher inflation rates (Argy 1970). Second, I add the GDP growth rate to my models since higher GDP growth means higher income growth and thus higher demand for goods and services, which usually leads to higher inflation (Mankiw 2010). Third, I include trade openness (Trade) measured in terms of the sum of a country’s imports and exports as a percentage of GDP because free trade tends to reduce the overall prices of goods and services in the economy (Triffin and Grubel 1962). Finally, I included countries' logged GDP (Log of GDP) in the models.

Given that the main dependent variable of this analysis is a dichotomous variable, I use logit regressions to estimate the effects of the six independent variables on the occurrence of an inflation crisis. I lag all the independent variables and control variables by one year to reduce the possibility of reverse causality. I estimate the effects of each independent variable in separate models because some of my independent variables share a common causal mechanism. For example, my theory suggests that the collapse of state authority can increase the likelihood of an inflation crisis by weakening a government’s tax capacity and its ability to make a credible commitment to price stability. Therefore, including the Tax Rev. variable measuring a government’s tax capacity and the Failure of State variable measuring the collapse of state authority that is directly related to the tax capacity in a single model will lead to an incorrect estimation of the influence of the two variables. Moreover, including all the independent variables in a single regression model significantly reduces the average number of observations (approximately by 16%). Finally, although some independent variables (e.g. Failure of State and Democratic Breakdown) measure quite different political and institutional aspects of a government, they are often highly correlated (the correlation coefficient between the two variables is 0.70). Thus, I expect that using separate models helps prevent the problem of severe multicollinearity.

Results and Analysis

Table 1 presents the results of the six logit regression models. Overall, the outcomes are consistent with my six hypotheses. In Model 1, the statistically negative sign on Tax Rev. shows that as I hypothesized, when governments’ ability to tax is weak and thus their incentive to use an inflation tax increases during civil war, an inflation crisis is more likely. Figure 1 depicts the substantive influence of governments’ ability to tax on the predicted probability of an inflation crisis during civil war. It suggests that a change in tax revenues as percentage of GDP from one standard deviation below its average to one standard deviation above its average reduces the probability of an inflation crisis by 26% points, which demonstrates the strong influence of governments’ tax capacity on the occurrence of an inflation crisis.

In Model 2 and 3, the statistically significant and positive signs on State Failure and Democratic Breakdown suggest that when governments’ ability to tax and to credibly commit to low inflation seriously deteriorates during civil war, the likelihood of an inflation crisis increases. Figure 2 and 3 show the substantive effects of the two variables on the predicted probabilities of an inflation crisis. Figure 2 suggests that a change from the minimum to the maximum level of State Failure increases the probability of an inflation crisis by 60% points while Figure 3 demonstrates that the same change in Democratic Breakdown raises the probability of an inflation crisis by 70% points. The statistically significant negative sign on Fixed Exchange Rate in Model 4 supports my argument that a fixed exchange rate system maintained during civil war plays an important role in reducing the probability of an inflation crisis. According to Figure 4 that shows the substantive effects of a fixed exchange rate on the predicted probability of an inflation crisis, compared to countries without this system, those with a fixed exchange rate system are 28% points less likely to suffer from an inflation crisis during civil war.

In addition, the statistically significant coefficients on Log Deaths and Duration in Model 5 and 6 show that while a highly destructive civil war significantly raises the probability of an inflation crisis, long-lasting civil war reduces the likelihood of the event. Particularly, the latter finding suggests that even if a civil war destabilizes the national economy and creates great uncertainty regarding the prospect of the economic conditions in the near future, as times goes by and the market participants adjust to ‘the new normal’, the war’s devastating effects tend to subside. Finally, Table 2 shows that though I replace an episode of an inflation crisis with that of hyperinflation as the dependent variable, the coefficients of all the independent variables except Fixed Exchange Rate still have statistically significant and hypothesized signs. Because Fixed Exchange Rate perfectly predicts the occurrence of hyperinflation, a logit regression model could not estimate the effects of this variable. It means that all of the countries that have a fixed exchange rate regime experienced no episode of hyperinflation.

In order to check the robustness of my findings, I included additional regression models in my empirical analysis. First, I estimated two regression models including all the independent variables in the same models except for Failure of State and Democratic Breakdown. I again included these two variables in the two separate models due to severe multicollinearity. Despite the changed model specification, I found that my main findings remain the same (see Table A-9 in the Appendix). Then, I estimated another model including all the independent variables without any exception. I found that the coefficients of all the independent variables except for Failure of State and Democratic Breakdown have statistically significant and hypothesized signs (see Table A-9 in the Appendix). This outcome seems to demonstrate the problem of strong multicollinearity that I explained above.

Second, I estimated other models using three different measures of the dependent variable. One measure defines an inflation crisis using a threshold of 15% per annum while another measure defines the crisis using a threshold of 25% per annum (see Table A-5 and A-6 in the Appendix). The third alternative measure of the dependent variable is growth rates of inflation. Specifically, I used the logarithm of annual inflation rates and estimated Ordinary Least Squares models (see Table A-7 in the Appendix).¹ The outcomes of these models suggest that my findings are robust to the use of varying measures of the dependent variables.

Third, I also use a different measure of a government’s ability to tax. O’Reilly and Murphy (2022) constructed an index that measures “fiscal capacity (the ability to raise revenue)” of governments by using V-Dem’s “state fiscal source of revenue.” I included their measure of fiscal capacity in my empirical model and found that the replacement of the measure does not change the main outcome (see Table A-8 in the Appendix). Fourth, I added more control variables to my main model (see Table A-10 in the Appendix). Because the two oil crises in 1970s may have contributed to extraordinarily high inflation in countries engaged in civil war, I included a dummy variable that indicates the 1970s. Moreover, since an inflation crisis may be attributable to a shortage of goods and services caused by economic sanctions imposed by other countries during civil war, I included a dummy that measures the presence of economic sanctions in the model. Furthermore, because an increase in oil production may have increased rebels’ incentive to take over a government and raised inflation at the same time, I included the amount of oil exports in the model. Again, I found that the inclusion of these control variables does not change the outcome. Finally, I included country- and year fixed effects separately in my main model to control for both unobserved country heterogeneity and year-specific shocks. In the model using country fixed effects, I found that Tax Rev., Fixed Exchange Rate, and Duration are not statistically significant anymore while Failure of State, Democratic Breakdown and Log Death still remain statistically significant (see Table A-11 in the Appendix). In the model using year-fixed effects, the main findings remain unchanged (see Table A-12 in the Appendix).

This article explored several important causes of inflation crises during civil war. It found that weak government capacity, characterized by a limited ability to collect taxes, may increase the prospect of abnormally high inflation by increasing the government’s dependence on an inflation tax as an alternative source of revenue. Moreover, it suggests that a government’s fragile credibility, caused by the collapse of the state, a democratic regime breakdown, or the absence of a fixed exchange rate system, may heighten the probability of an inflation crisis during civil war. Finally, it shows that a civil war’s negative influence on a country’s price stability may significantly vary according to the destructiveness and the duration of the war.

This article addresses an important gap in the existing literature on the economic costs of civil war by systematically examining the determinants of devastating inflation during civil war, a subject to which scholars have paid little attention. The economic costs of extraordinarily high inflation can be as damaging to the welfare of people suffering from civil war as economic contraction. High inflation can seriously disrupt even the most basic market exchanges in a civil war economy. It can thus intensify the economic pain of people who may be in desperate need of essential goods to survive. Moreover, the depletion of bank deposits and capital flight induced by inflation crises can raise the likelihood of collateral damage such as banking crises or exchange rate crises. This suggests that the costs of inflation can extend beyond deterioration of market efficiency to seriously threaten countries’ overall economic security. Given the economic significance of price stability, I expect that this systematic study of the causes of high inflation during civil war may contribute to a more comprehensive understanding of the economic consequences of civil war.

The findings of this article also suggest a few important policy implications for preventing costly inflation crises during civil war. First, sustaining a government’s ability to collect tax during civil war is one of the most important ways of preventing an inflation crisis. The use of inflation tax as a source of government revenue, which is a primary cause of inflation crises, can only be avoided when a government’s administrative system can continue to enforce tax laws during civil war. Second, maintaining democratic governance is also a key factor in maintaining price stability during civil war. Only when people see that the government is strictly following the rule of law and maintaining transparency about its macroeconomic policymaking will they trust the government’s commitment to low inflation, thus maintaining stable market expectations of future inflation.