7

War, Credit, and Fiscal Capacity

The political economy of public finance advanced in chapter 2 suggests that funding military expenses with taxation should contribute to state building because doing so stimulates self-strengthening reform. By contrast, financing war with foreign capital might not necessarily lead to stronger states if countries default on war debt and surrender national assets and entire revenue lines to foreign bondholders as part of debt readjustment agreements. In this chapter, I investigate the short- and long-term effects of external capital access on fiscal capacity, namely, the ability to raise taxes. I first show that access to external finance during wartime in the Bond Era decreased the likelihood of strengthening fiscal capacity, measured by direct tax ratios. Then I investigate whether early tax efforts (or their absence) had long-term repercussions. I show that war finance in the Bond Era shaped fiscal capacity all the way into the early twenty-first century. Countries that disproportionally financed war externally in the nineteenth century show lower tax capacity circa 2000 than countries that relied on domestic resource mobilization. To come to this conclusion, I address various threats to inference and decompress history by showing intermediate effects. This chapter sheds light on the mixed evidence that often surrounds empirical accounts of the bellicist hypothesis after 1800. The means to finance war are key to understanding the conditions under which war makes states.

7.1 Theoretical Expectations

A key implication of the theoretical discussion in chapter 2 is that fiscal shocks strengthen state capacity when rulers are compelled to undertake tax reform to mobilize domestic resources, a course of action that cannot be taken for granted. The “transaction costs”¹ of enforcing a new tax unilaterally are high: obliging the populace to pay taxes by intimidation and force is expensive and inefficient. To induce “quasi-voluntary compliance” with new taxation, the ruler might grant taxpayers some political say over fiscal policy—namely, power-sharing institutions, which convey information to taxpayers about who owns what (i.e., what the tax base is) and how tax yields are to be spent, limiting the ruler’s discretion over government funds (thus the expression limited government).

Power-sharing institutions are second-best solutions for rulers who seek to maximize wealth and power. Besides reducing the ruler’s discretion over fiscal policy today, in the future political rivals might choose to use the enhanced tax capacity against the best interests of today’s ruler if they achieve office. Together, current and future costs of taxation explain rulers’ hesitancy to rely on taxation.²

Rulers will consider policy alternatives to meet fiscal shocks while not sharing power with taxpayers or having to spend resources in the tax bureaucracy. External finance is one such policy alternative. Foreign capital inflows can be put to work immediately. In case of war, they can be used to purchase new military equipment, mobilize a larger army, or feed the troops; alternatively, foreign loans can subsidize government consumption in other areas, freeing domestic resources to pay for war. Either way, external finance allows rulers to meet an expenditure surge while not having to strengthen fiscal capacity, postponing the (unwelcome) political ramifications of higher taxation.

In the remainder of this chapter, I test several aspects of the political economy of war finance. First, I show that the disproportional use of external finance during wartime did not build tax capacity during the Bond Era—the short term. Then I document its negative consequences for long-term fiscal capacity. Overall, results suggest that the average ruler reshuffled the tax administration and raised new taxes when pressed by circumstances, that is, if excluded from international credit markets. Once enhanced, however, fiscal capacity did not revert to prewar years, a phenomenon commonly known as the ratchet or “displacement” effect of war.³ In chapter 8, I elaborate on the causes of persistence, the so-called mechanisms of transmission.

7.2 War Finance and Short-Run Gains in Fiscal Capacity

To test the first part of the argument—rulers prioritize external finance over taxation to minimize political and administrative costs—I examine the evolution of fiscal capacity as a joint function of war (the paramount fiscal shock before 1914) and access to international capital markets. I expect tax capacity of country i at year t to evolve in the following form:

where X denotes a vector of country-level controls, and η_i and γ_t full batteries of country and year fixed effects, respectively. If rulers seek to minimize the political ramifications of taxation, they will only strengthen fiscal capacity when they run out of options. Building from this premise, I expect tax capacity to increase when countries wage war—the fiscal shock—while excluded from capital markets, therefore β₃ > 0. By contrast, waging war with access to external capital should relax incentives to increase tax capacity if only to avoid sharing fiscal powers with taxpayers and new administrative expenses, hence β₁ = 0.

One could use the experience of Spain with war finance as a heuristic. This country participated in three waves of war in the second half of the nineteenth century: In the first wave, Spain waged war in Cochin China (1858–1862), Morocco (1859–1860), Santo Domingo (1863–1865), and Chile (1864–1866). Two years later, it waged war against Cuban rebels (1868–1878)—the second wave—and lastly the Philippines (1896–1898) and Cuba again (1895–1898)—the third wave. Based on the magnitude of military expenses, the first wave of war was the least costly of the three,⁴ but it was also the only one in which Spain was excluded from international capital markets. Domestic resources were mobilized: the tax-to-GDP ratio increased by 1.7 points within seven years, from 7.1 percent to 8.8 percent (a 23 percent increase). Small as it was by modern standards, the tax pressure remained at that level for roughly 25 years, until the early 1890s. The other two military campaigns were financed with a mix of external debt, colonial debt, and inflation tax.⁵ Despite being significantly more expensive, the second and third waves of war increased tax ratios by 0.40 and 0.88 points of GDP, respectively, a quarter and a half of the increase in the first wave. The Spaniards, like many other nations at the time, exerted meaningful fiscal effort when they could not rely on borrowed money overseas.

7.2.1 DESIGN AND MEASUREMENT

The period between 1816 and 1913 was one of sustained military conflict (see chapter 6) and coincided with the advent of global finance (see chapter 3), thus my focus on the long nineteenth century. Before 1914, fiscal efforts were largely driven by military need. The boom in welfare spending following World War I makes isolating the effect of war on fiscal capacity thereafter more difficult because the newly created social programs also required higher taxation. In addition, because the financial costs of the two world wars were unprecedented, the most active participants were countries with high fiscal capacity to begin with. Expanding the analysis beyond 1913 would exacerbate problems of selection.

For the dependent variable, I rely on an important dataset on public finance recently assembled by Andersson and Brambor, who put together an unbalanced panel of tax ratios for 31 countries in Europe, Latin America, Oceania, and Japan as early as 1800.⁶ To examine advances in tax capacity before 1914, I follow Dincecco and Prado and my earlier work,⁷ employing the share of direct taxes as a percentage of total taxation. Direct taxes (income and property taxes) require a sophisticated bureaucratic apparatus to assess wealth and monitor compliance of an atomized tax base.⁸ Efforts to increase direct taxation in the nineteenth century signal rulers’ resolve to strengthen fiscal capacity.

Table 7.1 lists the effective sample for the empirical analysis, reduced to 23 countries because of limited tax data before 1914. The other two key covariates are war and access to international finance. War data are drawn from Wimmer and Min for reasons provided in chapter 6: their data are representative of the universe of warfare waged by sovereign and nonsovereign countries in the nineteenth century.⁹ For each year, I compute the total number of interstate and independence (or secessionist) wars fought by any given country, or the intensive margin of war. For robustness purposes, I also consider the Correlates of War dataset, but the number of wars decreases because it lists only interstate wars waged by internationally recognized states, that is, those with diplomatic relations with both the French and the British.

TABLE 7.1. Sample Coverage of Direct Tax Ratios Prior to 1914

Argentina	1895	1913		Mexico	1867	1895
Australia	1910	1913		Netherlands	1816	1913
Austria	1816	1913		New Zealand	1879	1913
Belgium	1830	1912		Norway	1851	1913
Chile	1817	1913		Peru	1899	1913
Denmark	1820	1913		Portugal	1833	1913
Ecuador	1830	1913		Spain	1845	1913
Finland	1882	1913		Sweden	1850	1913
France	1816	1913		Switzerland	1875	1913
Germany	1906	1913		United Kingdom	1816	1913
Italy	1862	1913		Uruguay	1903	1913
Japan	1868	1913

Source: Andersson and Brambor (2019). Note: First and last observation per country (N = 23).

To determine access to international capital, I resort to episodes of external default as early as 1800 coded by Reinhart and Rogoff, who define sovereign default as the failure of a government to meet a principal or interest payment on the due date (or within the specified grace period).¹⁰ Consistent with the Wimmer and Min war data, Reinhart and Rogoff define states based on international borders as of the early 2000s. The median duration of default episodes in the period under consideration is six years.¹¹ Critically, while in default, countries are excluded from the international lending market. I expect war waged during episodes of default to stimulate improvements in wealth assessment, tax enforcement, and institution building.

Tax capacity is a slow-moving variable. To account for serial correlation of the outcome variable, I include its first lag as an additional control. To adjust for time-invariant unobserved characteristics between countries, I add a battery of country fixed effects. Any secular trend that affects, for instance, access to credit and tax capacity is accounted for with a full battery of year fixed effects. To minimize selection bias, I control for the cumulative number of wars and external defaults, accounting for differences in military and financial trajectories that could affect the likelihood of being at war at any particular time (i.e., the more war a country has waged in the past, the more likely that country is to wage war in the future¹²) and of being in default (i.e., past default might increase the chances of future exclusion¹³).

The point estimates for β₁ and β₃ in expression 7.1 are plotted in figure 7.1. This first set of results suggests that the effect of war on tax capacity hinges on external capital access. When a country is in default, hence excluded from international capital markets, war exerts a positive effect on direct tax ratios. The effect is the opposite and half its size when a country has access to external capital. The magnitude of the estimates is substantial given that it speaks to immediate fiscal effects: for each additional war year excluded from external capital, the direct tax ratio increases by 0.7 percentage points, all else held constant. Based on these estimates, fighting an entire war (four years on average) while in default would increase direct tax ratios by 2.8 percentage points, an 11-point increase relative to the sample average. Against expectation, waging war with access to capital does not have a null effect but, in fact, decreases tax capacity. For reasons of popularity, rulers might be willing to relax the tax pressure while at war but at the same time secure sovereign loans to meet military expenses. If so, bringing tax ratios back to prewar levels after hostilities cease might require additional efforts, increasing the chances of default.¹⁴

7.2.2 GLOBAL CREDIT CRUNCHES

Although intuitive, using a country-specific measure of external default to establish external capital access is problematic. Countries that wage war in default might differ from those that postpone war until foreign capital is available in ways that also affect long-term fiscal capacity. In this section, I address endogeneity in capital access by exploiting global credit crunches. In the period under observation, European countries experienced economic and banking crises that rapidly reverberated in sovereign debt markets.¹⁵ I take advantage of unanticipated global credit crunches initiated in Europe, also known as “sudden stops,”¹⁶ to identify periods in which countries were precluded from financing war externally irrespective of their (un)observed characteristics. In the words of Reinhart and Rogoff,

banking crises in global financial centers (and the credit crunches that accompany them) produce a “sudden stop” of lending to countries at the periphery…. Essentially, capital flows from the “north” dry up in a manner unrelated to the underlying economic fundamentals in emerging markets.¹⁷

FIGURE 7.1. Effect of War on Direct Taxation as a Function of Default Episodes. The outcome variable is the annual share of direct taxes to total tax revenue between 1816 and 1913. Regression estimates for expression 7.1 using external default episodes to measure external capital access. Covariates (unreported) include a first lag of the outcome variable, cumulative war and default episodes, and country and year fixed effects. Sample size (N) = 1,225. Errors clustered at country levels with 90% and 95% CI.

TABLE 7.2. Banking Crises and Stock Market Crashes in London, 1816–1913

Banking crises		Stock market crises
1825	1849	1865
1837	1850	1866
1838	1857	1867
1839	1866	1910
1840	1873	1911
1847	1890	1912
1848		1913

Note: Dates drawn from Reinhart and Rogoff (2009). 1873 banking panic added by author.

Building on this intuition, I examine whether incentives to finance war with tax money grew stronger in periods in which international capital dried up for reasons exogenous to the borrower’s characteristics. To identify the onset and duration of global credit crunches, I draw on banking and stock market crises in Britain, the world’s banker, before 1914 (table 7.2). Crashes in London rapidly spread to Paris, Berlin, and New York. Contagion took different routes, including arbitrage in commodities and securities and movement of money in various forms (specie, bank deposits, bills of exchange), cooperation among monetary authorities, and pure psychology.¹⁸ One way or another, financial crashes in London dried up international lending on a global scale.¹⁹

Important for exogeneity purposes, the causes of the British financial collapses in the nineteenth century are domestic—certainly the case for the major crises of 1825, 1847, 1857, and 1866 but less true for the 1890 panic, in which a large financial imbalance in Argentina halted British lending.²⁰ More importantly, British panics did not respond to defaults by borrowers, which would cast doubt on the exogeneity of these shocks. Most of the countries that defaulted in the nineteenth century were in the periphery. Although the defaulted quantities were significant relative to their home economies, from a global perspective they were merely a “sideshow”²¹ for the British economy. All things considered, the periods of sudden stops can be safely treated as exogenous to every country except Great Britain and 1890 Argentina.

FIGURE 7.2. British Capital Exports from 1865 to 1914. In light gray: Banking panics of 1865, 1873, and 1890. In dark gray: The stock crisis of 1907. Sources: Reinhart and Rogoff (2009); Stone (1992).

For the purposes of illustration, figure 7.2 shows the evolution of British capital exports since the earliest date, while indicating the years of banking panics and stock crashes as dated by Reinhart and Rogoff.²² Figure 7.2 reflects the boom-and-bust cycles preceding and following a banking crisis, as exemplified by those occurring in 1873 and 1890. Prior to each bust, lending was ferocious, but once the debt bubble burst, international capital flows temporarily dried up across the board. Precisely, during periods of sudden stop, I expect rulers to have stronger incentives to finance military campaigns by means other than external borrowing; namely, taxes.²³

TABLE 7.3. Frequency and Duration of War as a Function of Endogenous and Exogenous Credit Access in the Short-Run Test Sample

	Endogenous measure			Exogenous measure
	Not in default	In default		Credit flows	Credit stops
Frequency	94%	6%		43%	57%
Duration in years	3.06	1.89		2.01	2.42
	(0.21)	(0.35)		(0.18)	(0.17)
Duration difference	1.17**			−0.41

Note: The total number of war years is 162. Credit stops refer to periods of sudden stop. Standard error in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.

Sudden stops in the nineteenth century lasted four years on average.²⁴ Accordingly, I establish four-year windows following the onset of each sudden stop and assume that within these windows countries had no access to external loans. If sudden stops are predictable, countries may go to war before their onset, casting doubt on the exogeneity of this historical accident. To assess the unanticipated nature of sudden stops, table 7.3 shows the frequency and duration of war when the endogenous measure of capital access is considered—being in default or not—and when the exogenous measure is used instead. The first two columns show that few war years took place while participants were in default, confirming concerns about endogenity in using this intuitive measure. By contrast, the breakdown for the exogenous measure is fairly balanced: 57 percent of war years coincide with periods in which the international lending market is down. This distribution speaks favorably to the unpredictability of global credit crunches.

Now I consider the decision to end war, a second threat to inference. A weak state that finances war externally may be more prone to surrender during sudden stops. If that is the case, weak states would eventually experience a higher proportion of war years when credit flows and a lower proportion of war years during sudden stops. This would bias the estimation results toward finding a negative effect of war for years when credit flows. If this pattern were systematic, on average shorter wars during periods of sudden stops should be observed. However, table 7.3 suggests that the duration of war in and outside sudden stops is fairly balanced: 2.42 years in periods of sudden stops compared to 2.01 years when credit flows, the difference not being statistically significant. These numbers contrast with the breakdown for the endogenous measure: wars in times of default are shorter than those in which participants have access to capital. If war is judged by its frequency and duration, table 7.3 suggests a comparison of apples to apples when tackling war waged during periods in which international lending flows and war waged during sudden stops of credit.

Results with Global Credit Crunches

Next, I reestimate expression 7.1, substituting the endogenous measure of external capital access (i.e., default episodes) for the exogenous measure (i.e., sudden stops). To maximize exogeneity, I drop Great Britain (the world’s banker), as well as France and Germany, which by 1880—the average year in the effective sample—had already assumed an important role in international financial networks. I plot the point estimates of β₁ and β₃ in figure 7.3 and report results in regression format in table 7.4.

The first model, reported in column 1 of the table, uses four-year windows to approximate the average duration of a credit crunch. Results confirm the opposite effect of war on fiscal capacity with reliance upon an exogenous measure of capital access. When countries wage war with access to external capital, the effect of war on fiscal capacity is negative. By contrast, waging war excluded from external credit increases short-run fiscal capacity. Based on the new set of estimates, an additional war year excluded from capital access increases tax ratios by 7.6 percentage points relative to the sample mean.

For robustness purposes, models 2 and 3 set global credit crunch windows to three and five years, respectively. Because shorter and larger windows add noise to the exclusion measure, coefficients in models 2 and 3 attenuate relative to those in model 1. In model 4, I include a control of per capita GDP drawn from Bolt, Inklaar, de Jong, and van Zanden,²⁵ arguably a relevant control: direct taxation requires high levels of monetization, which correlates with income levels. Likewise, wealthier countries are expected to access international capital at better terms and build stronger armies, increasing the likelihood of going to war. The downside of this control is that GDP relies heavily on linear interpolation and reduces the sample size. As expected, this control attenuates the point estimates and relative to model 1; however, the substantive interpretation remains the same.

FIGURE 7.3. Effect of War on Direct Taxation as a Function of Exogenous Access to External Capital. The outcome variable is the annual share of direct taxes to total tax revenue between 1816 and 1913. Regression estimates for Expression 7.1 using global credit crunches as an exogenous shock for external capital access. Great Britain, France, and Germany are dropped from the sample to maximize exogeneity. All models include the following covariates: first lag of the outcome variable, cumulative war and default episodes, and country and year fixed effects. Additional controls are included as indicated in the figure legend. Errors clustered at country levels with 90% and 95% CI.

The relationship between the number of wars in which a country is involved in any given year and the intensity of war might be inversely correlated—a country can wage only so many major wars at once. Model 5 maintains the income control while replacing the intensive measure of warfare used so far by its extensive version, measured by an indicator variable equal to 1 when a country is waging war in any given year regardless of the number of wars in which it is involved. Results across specifications are virtually identical: waging war with access to external finance decreases short-run taxation, whereas fighting war excluded from capital increases it.

TABLE 7.4. Effects of War and External Capital Access on Short-Term Fiscal Capacity

	Figure 7.1	Figure 7.3
		(1)	(2)	(3)	(4)	(5)	(6)
: War	−0.316**	−1.146*	−0.787	−1.085	−0.825*	−1.540**	−1.244
	(0.124)	(0.576)	(0.493)	(0.663)	(0.469)	(0.689)	(0.866)
: External capital exclusion	−0.309	−11.336***	−11.252***	−11.228***	−11.816***	−11.985***	−11.022***
	(0.389)	(1.483)	(1.453)	(1.470)	(1.675)	(1.720)	(1.253)
: War× external capital exclusion	0.684*	1.936**	1.356*	1.455*	1.195*	2.174**	2.336*
	(0.339)	(0.736)	(0.721)	(0.779)	(0.656)	(0.832)	(1.326)
External capital measurea	Default	Crisis	Crisis	Crisis	Crisis	Crisis	Crisis
War marginb	Intensive	Intensive	Intensive	Intensive	Intensive	Extensive	Extensive
Global financial centersc	Included	Excluded	Excluded	Excluded	Excluded	Excluded	Excluded
Credit crunch window	–	4-year	3-year	5-year	4-year	4-year	4-year
Lagged outcome variable	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Cumulative default years	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Cumulative wars	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Country FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,225	1,024	1,024	1,024	1,001	1,001	1,001
R-squared	0.831	0.812	0.811	0.811	0.820	0.820	0.822
Countries	23	20	20	20	20	20	20

Note: This table reports results in figures 7.1 and 7.3 in regression format. Fiscal capacity is measured by the share of direct taxes to total tax revenue. Robust standard errors clustered at the country level in parentheses. Intercept not reported. *** p < 0.01, ** p < 0.05, * p < 0.1.

a External capital access is measured by country-specific default episodes (endogenous measure) or a global financial crisis (exogenous measure).

b War is measured at the intensive margin (total wars per year) or extensive margin (war or no war in any given year).

c Great Britain, France, and Germany, the three global financial centers, are excluded in models in figure 7.3 to maximize exogeneity of the financial shock in the last decades of the nineteenth century.

Models 1–5 rely on Wimmer and Min’s war data.²⁶ Model 6 reruns the same specification as in column 5 while using war data from the Correlates of War (COW) project,²⁷ a broadly used dataset in the historical study of warfare. As I explained in chapter 6, the COW dataset includes fewer interstate wars than Wimmer and Min’s because it lists only wars waged by internationally recognized countries—namely, those having diplomatic relations with Britain and France. Likewise, the main COW dataset lists only interstate wars, excluding wars of independence.²⁸ The merits of COW clearly outnumber its limitations and it is worth confirming results once COW war data are used. Results for that test are reported in model 6. The point estimates are virtually identical to those in model 5. The confidence intervals are slightly larger because of statistical power issues derived from having fewer wars listed in COW relative to Wimmer and Min.

Altogether, figure 7.3 suggests that waging war during the period from 1816 to 1913 yielded heterogeneous consequences on short-term taxation depending on whether or not a country had access to external capital. War waged while excluded from capital markets increased tax pressure, whereas war waged while having access to capital did not and sometimes decreased it (statistical significance is borderline with 90 percent confidence intervals). Together, figures 7.1 and 7.3 shed light on the mixed results that scholars have found for the relationship between war and state making in the era of international finance. The globalization of capital arguably distorted the rulers’ incentives to mobilize domestic resources when they were most needed.

7.3 War Finance and Long-Run Gains in Fiscal Capacity

Financing war with external funds might not damage state capacity in the long run. If the debt-tax equivalence of public finance holds, rulers will assume responsibility for war debt and enhance taxation to service debt after military conflict, thus contributing to state making. Generalized debt relief and the exchange of war debt for nontax revenue in case of default may, however, preempt self-strengthening reform—as I argued in chapter 5—unraveling the equivalence of taxes and loans for the purpose of state building. In light of the (at best) uncertain effects of external finance on capacity building, I expect long-term fiscal capacity to be stronger among countries that disproportionally relied on taxation to finance war relative to foreign loans, holding everything else constant.

7.3.1 DESIGN

To test for the long-term effects of war finance on fiscal capacity, longitudinal conflict-specific data regarding the manner in which war was financed in the past (i.e., tax and loan shares) can be exploited and used to model tax capacity in the long run. As I mentioned in chapter 6, the ideal dataset on war finance does not exist beyond a select number of cases, most of them in Western Europe. To move the analysis beyond advanced economies, I propose comparing the frequency with which war was waged with and without access to external capital in the Bond Era, and using that information to model tax capacity circa 2000. Drawing from the political economy of war in chapter 2 plus results in the first part of this chapter, I assume that if external capital were available, rulers would disproportionally resort to it to fund war.

For each country and war in the nineteenth century, I establish whether war was waged while having access to international credit markets. To establish access, I rely on episodes of default (endogenous but intuitive) and international financial crunches (unanticipated common shocks). With that information in hand, I regress tax ratios today on the number of years at war in the long nineteenth century with and without access to external finance while controlling for a battery of confounders, including war duration, casualties, and war outcome. Formally,

where the baseline category is fighting no war in the nineteenth century. I expect war making to strengthen rulers’ incentives to invest in fiscal capacity whenever the country cannot finance externally, contributing to long-term fiscal capacity, β₁ > 0. The effect of war financed externally is ambiguous: countries may exert a fiscal effort after war to service debt and thus expand tax capacity (path C in figure 1.3), yet other countries may suspend debt service. Default settlements may include debt-equity swaps and foreign international control, which potentially shrink the tax base of the borrowing government (path E in figure 1.3). Less stringent adjustments of external debt might involve debt relief and extension of maturities, which relax the need to build capacity. Together, I expect a null (if not negative) effect of war making the more countries wage war while having access to external credit, β₂ ≤ 0.

Two clarifications are in order: First, the expectation β₂ ≤ 0 works against the debt-tax equivalence of war finance for the purpose of state building. Note that if the equivalence holds, borrowing and taxes should be indistinguishable for long-term state capacity, implying β₁ ≈ β₂ > 0, all else being constant. Second, in the absence of external credit, rulers might resort to printing money, seeking domestic loans, trading slaves, or engaging in financial repression to finance the means of war.²⁹ If any, these alternatives introduce a downward bias on β₁ because they relax the incentives to enhance taxation in times when external credit dries up.

To measure fiscal capacity today, I rely on the personal income tax (PIT) as a percentage of GDP. Implementing a PIT requires a sophisticated bureaucratic apparatus capable of assessing a highly atomized tax base, enforcing compliance, and sanctioning evaders. In light of its administrative challenges, this tax is considered to be the endpoint of fiscal capacity building.³⁰ For the same reason, it sets a compelling benchmark to establish how far each country has gone in building tax capacity since 1914. To minimize the influence of anomalous observations (for instance, following a temporal economic shock), I work with average PIT-to-GDP ratios between 1995 and 2005.³¹

Note an important change in the data structure with respect to the first part of the chapter. In the short-run models, I exploited longitudinal cross-sectional data; here the variation is purely cross-sectional. The sample is no longer restricted by nineteenth-century tax data, which is relatively scarce; as a result, the long-term models include over four times the number of countries, covering a wider spectrum of developed and developing nations, sovereign and dependent. Dependent countries, I argue in chapter 6, resorted to a mix of international loans and subsidies from the metropole to finance local and imperial war, conditions under which I expect few gains in long-term state building. War data again are drawn from Wimmer and Min.

Because the data are cross-sectional, instead of country fixed effects, I consider a series of potential confounders, X, affecting fiscal capacity today, as well as war participation or credit access or both in the nineteenth century. First, I add an indicator to identify Great Powers.³² These countries had strong capacity and access to finance and were likely to wage war. Second, I consider a measure of initial wealth because wealthier countries are more likely to go to war and have stronger fiscal capacity in the first place.³³ In the absence of systematic GDP data for developing countries in the early nineteenth century, I include a measure of population density as of 1820, a standard proxy of economic prosperity before the Industrial Revolution.³⁴ Third, I include two geographic characteristics that could affect both sides of expression 7.2. The first one, sea access, is defined as the percentage of the land surface area of each country that is within 100 kilometers of the nearest ice-free coast. I expect sea access to correlate with trade activity, hence the geostrategic value of the country as well as its integration in international capital markets.³⁵ The second geographic control is the percentage of territory that is desert. I expect deserts to inhibit industrial growth and preempt monetization, but desert territory might also work as a natural barrier to foreign invasion, thus reducing the frequency of war.³⁶ Finally, I control for an important source of nontax revenue that could shape incentives to go to war (or suffer attack): being an oil producer.³⁷ Arguably, this variable gains relevance for the later years of the period under consideration.

Finally, all models below include a battery of region fixed effects,γ, that account for continent-specific characteristics in the frequency of war, access to credit, and statehood timing; and a battery of colonial origins indicators,ρ, because I expect opportunities of the colonies to go to war, the tax structure that they build up, and the terms of external credit to be conditioned by the metropole.³⁸

7.3.2 NAIVE ESTIMATES

The first set of results are plotted in figure 7.4. To establish a meaningful benchmark, I first plot the estimate for the unconditional version of the bellicist hypothesis—namely, more war, more state—for all the states sampled in Reinhart and Rogoff for which I can gather full information (63 out of 68).³⁹ The specification for the unconditional model is as follows:

including the same covariates, X, and region and colony fixed effects, γ and ρ, respectively, as above. The coefficient for number of years at war between 1816 and 1913, represented by a circle in the top portion of figure 7.4, is positive but not significant at 90 percent, consistent with the mixed findings mentioned in the literature review on the bellicist hypothesis in chapters 1 and 6.

Model 1’s unconditional estimate in figure 7.4 should be compared to the six that follow, which distinguish the effect of war fought in default from that of war fought with access to international credit markets, β₁ and β₂, respectively, in expression 7.2. The first pair of point estimates (model 2, represented by squares) set a baseline. The signs of the coefficients are consistent with expectations: is positive and statistically different from zero; is centered around zero. Substantively, the benchmark specification means that a one-standard-deviation increase in the number of years at war while in default expands the PIT-GDP ratio today by 0.41 points. This is equivalent to a 15 percent increase with respect to the mean PIT. The estimate of β₂, by contrast, suggests that wars waged with access to international markets exert no lasting effect on fiscal capacity.

FIGURE 7.4. Long-Run Fiscal Capacity as a Function of War and Endogenous Credit Access. This figure plots marginal effects of war and credit access on personal income tax as a percentage of GDP circa 2000. N = 63. OLS, 90% and 95% CI. Credit access and exclusion are given by episodes of default. All four models include the following covariates: region and colonial origins fixed effects, total years in default, population density as of 1820, being an oil producer, sea access, desert territory, and a Great Powers indicator. Estimates in models 3 and 4 are drawn from models that include baseline controls plus war location and war casualties, respectively.

War causes destruction, but damage may vary depending on the location of military engagement. The tax base can be badly hurt when military conflict takes place within national boundaries, thus inhibiting investment in fiscal capacity. The location of war is thus likely to be a confounding variable. To address this logic, in model 3 I rerun the benchmark model while adding a control for the location of the conflict.⁴⁰ The new estimates for β₁ and β₂ (represented by diamonds) remain virtually identical to those in the benchmark model.

Not all wars are created equal. Bloodier and longer wars might overcome resistance to taxation while maximizing rulers’ incentives to invest in fiscal capacity. To address this possibility, in model 4 I add a control for the intensity of warfare, measured by the total number of battle deaths between 1816 and 1913.⁴¹ The inclusion of a measure of war casualties does not substantively affect the point estimates of the coefficients of interest, β₁ and β₂, now represented by triangles.

This first analysis yields results consistent with theoretical expectations: if countries wage war while excluded from international credit markets, the incentives to invest in tax capacity are strong and long-term state capacity follows. By contrast, when states have access to foreign capital, war (on average) does not translate into state making.

7.3.3 THREATS TO INFERENCE

Although intuitive, the analysis above is arguably plagued by endogeneity concerns. Being in default is not exogenous nor is the decision to wage war or which war to fight. Next, I address stepwise both sets of issues.

Global Credit Crunches and Long-Term Capacity

As discussed in section 7.2.2, one can replace country-specific default episodes for global credit crunches and use them as exogenous shocks in external capital access. These sudden stops of credit originated in the financial capitals of the world, went unannounced, and lasted, on average, four years. Conveniently, by relying on global credit crunches, I create a measure of external capital access that is also unconstrained from country-level default data availability in Reinhart and Rogoff,⁴² effectively expanding the sample size to 100+ countries in the world. For each of them, I count the number of years at war inside and outside periods of global capital freeze.

TABLE 7.5. Frequency and Duration of War as a Function of Exogenous Credit Access in the Long-Run Test Sample

	Interstate war			Interstate and secessionist war
	Credit flows	Credit stops		Credit flows	Credit stops
Frequency	47.74%	52.25%		50.89%	49.11%
Duration in years	2.32	2.25		2.23	2.29
	(1.87)	(1.51)		(1.73)	(1.58)
War-year-country	465			615
Countries	107			107

Note: Credit stops refer to periods of sudden stop. Standard deviation in parentheses. Countries in sample by regional breakdown are as follows: Africa: Burundi, Chad, Congo, Democratic Republic of the Congo, Egypt, Ethiopia, Guinea, Ivory Coast, Kenya, Lesotho, Madagascar, Mali, Morocco, Namibia, Nigeria, Rwanda, Senegal, South Africa, Swaziland, Tunisia, Zambia, Zimbabwe; Asia: Bangladesh, Bhutan, Cambodia, China, Cyprus, India, Indonesia, Iran, Israel, Japan, Lebanon, Malaysia, Mongolia, Myanmar, Nepal, Pakistan, Philippines, South Korea, Sri Lanka, Thailand, Turkey, Vietnam, Yemen; Europe: Albania, Armenia, Austria, Azerbaijan, Belarus, Belgium, Bosnia and Herzegovina, Bulgaria, Croatia, Czech Republic, Denmark, Estonia, Finland, France, Georgia, Germany, Greece, Hungary, Iceland, Ireland, Italy, Kazakhstan, Latvia, Lithuania, Macedonia, Moldova, Netherlands, Norway, Poland, Portugal, Romania, Russia, Slovakia, Slovenia, Spain, Sweden, Switzerland, Tajikistan, Ukraine, United Kingdom; Latin America: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, Venezuela; North America: Canada, United States; Oceania: Australia, New Zealand.

Table 7.5 illustrates the exogenous character of these shocks: half of total war years in the sample were waged during global credit crunches, and the average duration of wars was balanced across periods. The balanced distributions grant credibility to the unanticipated nature of sudden stops.⁴³

In figure 7.5, I plot the point estimates and in expression 7.2 net of the influence of other controls, together with the scatterplot between tax capacity today and the distribution of the number of years at war with and without access to external capital, respectively. Figure 7.5a suggests that the more war was waged without access to capital markets in the nineteenth century, the higher the PIT as a percentage of GDP today, holding everything else constant. Conversely, that relationship turns negative when focusing on the number of years at war waged while having access to external credit, figure 7.5b. Before drawing further conclusions, I address the second threat of endogeneity: selection into war.

The Decision to Go to War

Countries that wage war when international credit is tight (even for exogenous reasons) might be more capable than those that wait for lending to resume. To account for differences in initial state capacity, I consider two covariates: Bockstette, Chanda, and Putterman’s state antiquity index⁴⁴ and census capacity. If older states exist because they won wars in the past, the state antiquity index should reflect cumulative military and administrative capacity.⁴⁵ Census technology, initially adopted to establish the tax base and conscription potential within a given territory, should likewise reflect the administrative capacity of the state. To control for initial administrative capacity, I create the indicator variable “modern census by 1820,” which equals 1 if country i has conducted a modern census by 1820.⁴⁶

Initial controls are recommended to account for intrinsic characteristics that make a country more likely to go to war, access external credit, and build a tax administration. They are, so to speak, a good substitute for the absence of country fixed effects; however, following the onset of a sudden stop, a ruler can still choose whether to wage war or what kind of war to fight. I address this form of selection bias by considering only wars that are initiated while the market is still lending and eventually dries up as a result of a financial crisis. These wars are initiated without the expectation of a sudden stop. Thus, the decision to go to war or what type of war to fight is disconnected from external credit access.⁴⁷

FIGURE 7.5. Long-Run Fiscal Capacity as a Function of War and Exogenous Credit Access. Estimates reflect partial correlations of a full model of personal income tax as a percentage of GDP circa 2000 as a function of the number of years at war with and without access to external finance (established by international credit crunches) between 1816 and 1913 with the following controls: population density as of 1820, oil producer, sea access, desert territory, colonial origins fixed effects, and region fixed effects. N = 106 (Great Britain is excluded). Only a few country names are shown to avoid cluttering. Refer to appendix E in Queralt (2019) for models that drop Russia, Georgia, and France, potential influential outliers. Results hold.

FIGURE 7.6. Long-Run Fiscal Capacity as a Function of Exogenous Credit Access and Ongoing War Plus Initial State Capacity Controls. This figure plots marginal effects of war and credit access on personal income tax as a percentage of GDP circa 2000. N = 106. I run separate analyses for interstate war and interstate plus secessionist war. Access to international credit markets is exogenized by global credit crunches. Also, only “ongoing wars” are considered: namely, only wars initiated while the market is still lending and eventually dries up as a result of a financial crisis. Great Britain is excluded to maximize exogeneity. All models control for region and colonial origins fixed effects, population density as of 1820, being an oil producer, sea access, desert territory, and Great Powers status. In addition, I control for initial state conditions, including the use of a modern census by 1820 and the state antiquity index of Bockstette, Chanda and Putterman (2002).

Models 1 and 2 in figure 7.6 show point estimates for β₁ and β₂ in expression 7.2 when access to credit is exogenized by sudden stops, and selection into war is addressed by including controls for initial capacity (census, diamonds; state antiquity, squares) and subsetting for ongoing wars.⁴⁸ Based on the new estimates, a one-standard-deviation increase in the number of ongoing wars increases long-term average PIT by 12.5 percent, a sizable long-term effect. By contrast, is no longer negative (as it appeared to be before selection into war was considered in figure 7.5(b) but zero, which is still inconsistent with the unconditional interpretation of the bellicist hypothesis—more war, more state—and the debt-tax equivalence of war finance for the purpose of state building, by which loans should behave as deferred taxes.

To conclude this section, I also consider secession (or independence) wars. Leaders in secessionist territories might engage in calculations similar to those of sovereign rulers. For instance, in the 1820s Greek rebels negotiated external loans to finance the war of independence against Turkey. Default followed soon after, and independence loans were a matter of dispute for decades to come.⁴⁹ Models 3 and 4 in figure 7.6 show results once the ongoing war criterion is implemented to both interstate and secessionist war.⁵⁰ The point estimates are substantively identical, suggesting that secessionist wars have long-term effects on tax capacity similar to those of interstate war by sovereign nations.

Effects on the Periphery

The distinct effect of war on state building and on long-term tax capacity is robust to sample changes and additional controls, including military alliances or exclusion from war in which the British or its colonies participated. I refer the reader to Queralt (2019) for a full battery of robustness checks. Here I focus attention on the sample composition. So far, I have considered Great Powers and other wealthy countries (US and the Netherlands) in the sample. These were all militarily powerful states capable of substituting external for internal credit in times of crisis and deeply embedded in international capital networks.

Coping with financial shocks may be qualitatively different for developing nations, which are arguably more exposed to exogenous changes in global credit liquidity. Based on this premise, in table 7.6, I reestimate β₁ and β₂ after dropping from the sample all Great Powers plus all the foundational members of the OECD—namely, current advanced economies. For consistency, these models incorporate the various strategies to tackle endogeneity in credit access and war participation I elaborated above. The point estimates for developing nations in table 7.6 are of the same magnitude as those for the entire sample (figure 7.6), but they are more efficiently estimated. If anything, the periphery-specific analysis suggests that developing nations had more to gain from not having easy access to international credit markets at early stages of state building. Arguably, too much capital access too soon distorted incentives to build capacity in the developing world.

TABLE 7.6. Models of Personal Income Tax as Percentage of GDP circa 2000 in the Global Periphery as a Function of War and Exogenous Credit Access in the Long Nineteenth Century

	(1)	(2)	(3)
# Years at war while credit stops 1816–1913	0.116**	0.108**	0.117**
	(0.056)	(0.054)	(0.058)
# Years at war while credit flows 1816–1913	0.048	0.057	0.056
	(0.109)	(0.108)	(0.120)
Baseline controls	Yes	Yes	Yes
Census 1820 control	No	Yes	No
State antiquity control	No	No	Yes
Region FE	Yes	Yes	Yes
Colonial origins FE	Yes	Yes	Yes
Observations	96	96	93
R-squared	0.538	0.553	0.580

Note: Great Powers and foundational OECD countries are excluded from this analysis. Only “ongoing wars” are considered. Credit access exogenized by credit crunches. Baseline controls include population density, oil producer, sea access, and desert territory. Intercept not reported. Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.

War Noninitiators in COW

An alternative route to tackle selection into war is to study the effect of war making and credit access for states that did not choose to go to war but were dragged into it. The identification assumption for this test is that states did not strike first in anticipation of a likely attack.

To implement this test, I rely on the Correlates of War (COW) dataset, which identifies the initiator of each military conflict.⁵¹ Although COW reduces the representativeness of the universe of war and war participants—as discussed above—it guarantees that countries are sovereign, hence in full command of their military and fiscal policy.⁵² Conveniently, COW also facilitates information to control for war outcomes. This is substantively compelling because military outcomes potentially affect the incentives to invest in fiscal capacity; for example, winners might extract from losers, reducing the need to build capacity to pay for war.

FIGURE 7.7. Long-Run Fiscal Capacity as a Function of Exogenous Credit Access and Correlates of War Data, Subsetting for Noninitiators and Controlling for War Outcomes. This figure plots marginal effects of war and credit market access on personal income tax as a percentage of GDP circa 2000. N = 106 (Great Britain is excluded). War data are drawn from COW. One of the models uses the entire sample and the other focuses on war noninitiators as defined by COW. Access to credit markets is exogenized by global credit crunches. All models control for region and colonial origins fixed effects, population density as of 1820, oil producer status, sea access, desert territory, net victory, and Great Powers status. The models include one of the two initial state capacity measures: census capacity by 1820.

Figure 7.7 reports the coefficients for the full COW sample and for the noninitiator subsample. Both models control for the history of military victories and losses in the period from 1816 to 1913.⁵³ Results are similar across subsamples: waging war with access to external finance exerts null effects regardless of whether a country initiates war or is dragged into it. By contrast, going to war, voluntarily or forcibly, without access to external finance leads to long-term tax capacity. Based on these estimates, a one-standard-deviation increase in the number of years at war waged without access to credit increases average PIT ratios today by 25.7 percent, a stronger effect than I found in figure 7.6, likely because COW data overrepresents wealthier states.

Combined, results in table 7.6 and figures 7.4, 7.6, and 7.7 suggest that foreign loans potentially unraveled the connection between war and state making, shedding light on the reasons that war did not make stronger states in the last two hundred years in large parts of the developing world.

7.4 Addressing History Compression

So far I showed evidence of short- and long-term effects of war finance on taxation. What happened in between? I address concerns of “history compression”⁵⁴ by evaluating war effects at intermediate points in time. Specifically, I show persistence of the effects of war from 1816 to 1913 on tax ratios between 1945 and 1995, the post—World War II era. Given data constraints, I approximate fiscal capacity with the share of total tax revenue not accrued from trade taxes. This share measures the effort to raise revenue through sophisticated taxes (e.g., income tax, value-added tax) instead of tariffs, a tax handle that low-capacity countries often use.⁵⁵

To conduct this test, first I compute decennial averages of nontrade taxes as a percentage of total taxation from 1945 to 1995; then I regress those ratios on the number of years at war with and without access to external loans in the nineteenth century plus controls, or expression 7.2, for each decade in the second half of the twentieth century. Data for nontrade tax revenue are limited. To minimize unobserved heterogeneity across units while maximizing degrees of freedom, I include a former colonial status indicator, which collapses the three previous dummy variables (British, Iberian, and Other Colonies) into one; and a Great Powers indicator, which adjusts for the systematic difference in European core powers. In addition, I include a control for initial wealth (population density in 1820), as well as controls for oil production and sea access. Because the sample size is small, particularly in the earlier decades, I report 90 percent confidence intervals. Results are reported in figure 7.8.

FIGURE 7.8. Effect of Past Warfare and Exogenous Credit Access on Fiscal Capacity from 1945 to 1995. These figures plot the marginal effects of the number of years at interstate war with and without access to external credit between 1816 and 1913 on nontrade tax revenue from 1945 to 1995 (decennial averages centered at the first year of the decade). OLS, 90% CI. Data on nontrade tax revenue are drawn from Cagé and Gadenne (2018). Sample sizes are 34 (1946–1955), 37 (1956–1965), 55 (1966–1975), 71 (1976–1985), 84 (1986–1995). Great Britain is excluded. Controls are former colonial status indicator, Great Powers indicator, population density in 1820, oil producer, and sea access.

The estimates for the 1945–1995 decennial tax ratios are generally consistent with previous findings: the plot on the left in figure 7.8 suggests that waging war in the nineteenth century with access to external credit is not associated with post-WWII fiscal capacity, whereas waging war without access to international finance is (right plot). Approximately, an additional year at war in the nineteenth century without external finance increases post-WWII nontrade tax revenue by 1 percent, all else held constant. More importantly, figure 7.8 suggests that the different types of war finance can have lasting effects because they push countries onto different paths of state building, consistent with figure 1.3.

7.5 Conclusion

This chapter is data intensive because it seeks to establish short-, medium-, and long-term effects of war and external finance on fiscal capacity. Despite usual limitations of historical data, results suggest that incentives to expand tax capacity strengthen when external capital is unavailable, and that those efforts are capitalized on in the medium and long run. The findings also draw attention to the potentially negative consequences of financing war externally at a time when international bondholders were allowed to impose severe conditions on loan contracts, growing the share of debt service in the national budget at best and enabling asset seizure and foreign control at worst. Now that the short- and long-run relationship among war, international credit, and state making are documented, I devote chapter 8 to elaborating on the mechanisms of persistence, namely, the reasons that past war finance exerts lasting effects. In chapter 9, I reexamine the effects of war on fiscal capacity and their transmission with a series of case studies.

1. Levi (1988).

2. Acemoglu (2003) and Besley and Persson (2011) formalize the ruler’s fear of future extraction if the opposition takes over and uses enhanced tax capacity against the former ruler.

3. Peacock and Wiseman (1961).

4. In temporal order, the three waves of war consumed 35.1, 48.7, and 43.7 percent of the budget, respectively (Comín, 2015).

5. Comín (2012).

6. Andersson and Brambor (2019).

7. Dincecco and Prado (2012); Queralt (2015).

8. Daunton (2001); Tilly (1990).

9. Wimmer and Min (2009).

10. Reinhart and Rogoff (2009).

11. Reinhart and Rogoff (2009, p. 81).

12. Gennaioli and Voth (2015).

13. Tomz (2007).

14. The interested reader can refer to table 7.4 for results in regression format.

15. Eichengreen (1990); Neal (2015).

16. Calvo (1988).

17. Reinhart and Rogoff (2009, p. 74); emphasis added.

18. Kindleberger (1996, p. 109).

19. Bordo (2006).

20. For the domestic origins of the 1825, 1847, 1866, and 1890 crises, see Neal (1998), Dornbusch and Frenkel (1982), Mahate (1994), and Kindleberger (1996), respectively.

21. Eichengreen (1991, p. 151).

22. Reinhart and Rogoff (2009).

23. In light of figure 7.2, banking crises might be more damaging than stock market crashes (e.g., 1907). In appendix I in Queralt (2019), I show that results hold when stock market crashes are disregarded.

24. Catão (2006).

25. Bolt, Inklaar, de Jong, and van Zanden (2018).

26. Wimmer and Min (2009).

27. Sarkees and Wayman (2010).

28. The auxiliary COW datasets list wars of independence but do not map them into current state borders.

29. Cappella Zielinski (2016).

30. Besley and Persson (2011); Tilly (1990).

31. Note that the denominator in this ratio is GDP, not total taxation. GDP is a better choice because it conveys the extractive capacity relative to the tax base. PIT-to-GDP ratios are unavailable in the historical data. For data sources and the use of value-added taxes (or VATs) as a proxy of tax capacity today, refer to appendix M in Queralt (2019).

32. Great Powers were Austria-Hungary (treated as two different states), France, Germany, Italy, Russia, and Great Britain.

33. Gennaioli and Voth (2015).

34. Tilly (1990).

35. Salvucci (2006) for the strong ties between commerce and finance.

36. Data for sea access and desert territory are drawn from Nunn and Puga (2012).

37. Oil production data are drawn from Wimmer and Min (2009).

38. Accominotti, Flandreau, and Rezzik (2011); Ferguson and Schularick (2006).

39. Reinhart and Rogoff (2009). Breakdown of countries in my analysis by region: Africa: Egypt, Ivory Coast, Kenya, Morocco, Nigeria, South Africa, Tunisia, Zambia, Zimbabwe; Asia: China, India, Indonesia, Japan, Malaysia, Myanmar, Philippines, South Korea, Sri Lanka, Thailand, Turkey; Europe: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Netherlands, Norway, Poland, Portugal, Romania, Russia, Spain, Sweden, Switzerland, United Kingdom; South America: Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Peru, Uruguay, Venezuela; North America: Canada, United States; Oceania: Australia, New Zealand.

40. War location is the sum of the years at war fought abroad minus the years at war fought at home from 1816 to 1913. This variable is positive when a country fights more wars abroad than at home, negative when military disputes at home are more frequent than abroad, and zero when countries never go to war. Data for war location are drawn from Wimmer and Min (2009).

41. Data for war casualties from 1816 to 1913 are drawn from Dincecco and Prado (2012).

42. Reinhart and Rogoff (2009).

43. See text explanation of table 7.3 for the importance of balance in terms of duration. Refer to appendix W in Queralt (2019) for evidence of war not being waged in anticipation of sudden stops.

44. Bockstette, Chanda, and Putterman (2002).

45. Note that in the short-run models, this property is measured by the cumulative number of wars in the nineteenth century.

46. To create this variable, I manually coded the date of the first modern census ever implemented in all 107 countries in the sample. This information is retrieved from Goyer and Draaijer (1992a, b, c).

47. The 222 country-year wars taking place during sudden stops fall to 72 once I consider only wars that are ongoing by the onset of a sudden stop.

48. The interested reader may refer to table 6 in Queralt (2019) for separate analyses.

49. Reinhart and Trebesch (2015).

50. Substantially fewer secessionist wars than interstate wars appear in Wimmer and Min. Hence, I pool them together instead of analyzing them separately.

51. Sarkees and Wayman (2010).

52. The sample of interstate wars now comprises 37 conflicts and 172 war years in total; 78 were fought when credit flowed, and 96 when credit had suddenly stopped. Average war duration was 1.57 (SD = 1.04) and 1.76 (SD = 1.23) years, respectively.

53. This is measured by net victory, which indicates the number of wars won between 1816 and 1913 by country i net of wars lost during the same period.

54. Austin (2008).

55. Cagé and Gadenne (2018); Queralt (2015); Soifer (2015).