8
In chapter 7, I showed that war exerts opposite and lasting effects on fiscal capacity depending on access to international capital. Why do these effects persist? I advance two “channels” or mechanisms of transmission, which serve as the reasons that past war finance shapes fiscal capacity today. The first mechanism builds on the political repercussions of domestic resource mobilization. When rulers are excluded from external capital markets, incentives to strike deals with domestic taxpayers are stronger. Tax bargaining is more likely to materialize in power-sharing institutions in smaller and densely populated polities, where taxpayers face low coordination costs. Once in place, power-sharing institutions make taxation self-sustaining, carrying the fiscal effects of war into the future. The second mechanism of transmission puts emphasis on bureaucratic innovation and potentially applies to sovereign and nonsovereign countries alike. In other words, the bureaucracies created to fund war oppose disinvestment in fiscal capacity, securing their survival in the long run. Drawing from historical data on political and bureaucratic reform, I find empirical evidence consistent with both mechanisms.
8.1 The Ratchet Effect of War
The ratchet or “displacement” effect of war involves a regularity in public finance by which tax revenue gains during wartime do not return to prewar levels afterward—hence growing the size of government over time.1 The existence of a ratchet effect, however, is not immediately obvious. Rulers could raise tax rates during wartime and bring tax pressure down to prewar times once hostilities are over. Were this the case, one should expect no lasting impact of war on long-term state capacity.
Peacock and Wiseman suggest that the ratchet effect of war exists because it grows people’s tolerance of a high tax burden and because new obligations appear after war, including the provision of war pensions and nonmilitary expenses, for example, public employment for demobilized soldiers.2 Consistently, Scheve and Stasavage find that mass warfare can change perceptions of social fairness, making nonelites more demanding of redistribution via taxation and elites more welcoming to such demands.3 The rise of income tax progressivity coinciding with mass mobilization in World War I and its persistence afterward speak of a permanent change in preferences for taxation and social spending in advanced economies.
Mass warfare—namely, the mobilization of at least 2 percent of the adult male population—was rare before 1914 as were demands for a welfare state, particularly outside Western Europe.4 To shed light on the causes of the ratchet effect of war in the nineteenth century, I contemplate two alternative and mutually complementary mechanisms: one is political, the other bureaucratic, and both derive from the political economy of external finance advanced in chapter 2.
8.2 The Political Channel of Persistence
The political economy of war finance discussed in chapter 2 suggests that rulers may grant political rights to taxpayers to overcome credibility issues in tax policy and to minimize transaction costs in tax collection. Because both the ruler and the ruled benefit from power-sharing institutions, the new tax becomes self-enforcing, maintaining the fiscal effect of war in the long run.
Before I offer further specifics, three important clarifications are in order: First, war finance is not the only path to political reform. Political ideas,5 economic development,6 and political fragmentation7 are also key drivers of power-sharing institutions. Second, these institutions, which are not equivalent to democracy, refer to institutional configurations designed to address credibility problems on the ruler’s end—in our case, spending policy.8 A representative parliament is one (advanced) manifestation of power-sharing institutions, but not the only one—as will become clearer below.9 Third, fiscal capacity can be created in the absence of political reform. Great empires experienced sustained investment in capacity while not adopting power-sharing institutions. For instance, during the Warring States era in ancient China, the Qin dynasty (356–221 BCE) conducted self-strengthening reforms without adopting any recognizable form of power-sharing institutions.10 Likewise, the Umayyad caliphate initiated bureaucratic and military reform in the eighth century, spanning over three centuries while articulating no known form of executive constraint.11 The Chakri dynasty in Siam is a specific Bond Era example of ambitious bureaucratic reform that carried no political consequences, at least not in the short run. I return to it in chapter 9. Keeping these important considerations in mind, I next elaborate how tax-based war finance might lead to political reform and a ratchet effect of war, holding everything else constant.
8.2.1 WAR AND THE FISCAL CONTRACT
The connection between war finance and political reform has a long tradition in the literature of state formation.12 From ancient times to the Roman Empire, war expenses strained the state’s coffers and became the top preoccupation of sitting chiefs, emperors, and monarchs.13 Far from declining, war expenses grew larger in the last millennium. In the high and late Middle Ages, monarchs regularly summoned estates to raise funds to wage war.14 Beginning in the sixteenth century, the introduction of new war technologies—the longbow, the pike, and gunpowder—multiplied the financial needs of the Crown;15 however, elites feared that rulers would wage the wrong war—one that pursued only personal aggrandizement—or that they would spend tax yields foolishly.16
To credibly commit to avoid wasting subjects’ money, European rulers relinquished power over spending decisions to big taxpayers and government lenders—often the same individuals.17 Some power-sharing arrangements took the form of representative assemblies with extensive powers over fiscal policy. “The power of the purse” was conferred upon Parliament as early as 1572 in the Netherlands and 1688 in England with astonishing fiscal outcomes.18 This modern form of political representation extended to the rest of Western Europe in the nineteenth century.19
Power-sharing institutions other than representative assemblies were adopted in parts of continental Europe before 1800.20 The paramount of absolutism, Louis XIV (r. 1661–1715), was compelled to share fiscal powers with tax farmers in order to secure funds for ongoing wars.21 J. B. Colbert, his finance minister, promoted the creation of the Company of General Farms, which allowed an oligopoly of tax farmers, by then also government bankers, to keep the king’s fiscal policy in check.22 Because tax farmers secured tax revenue and public credit for the king, they were able to exclude him from obtaining fresh funds in case of default. There rested their capacity to constrain the French monarch’s power.23 For his part, by assuming the “significant political costs”24 of default, the king laid the foundations of cheaper, long-term sovereign debt.25
Louis XIV’s fiscal innovation exemplifies the notion of power-sharing institutions that might be put in place as a result of war.26 Tax bargaining between rulers and tax subjects does not have to result in a legislative assembly in which large groups of society are represented. Power-sharing institutions involve any institutional system designed to overcome the incumbent’s credibility issues in spending policy.27
8.2.2 THE FISCAL CONTRACT AND SUSTAINED TAXATION
The importance of power-sharing institutions for long-term taxation cannot be overstated. Executive constraints on fiscal policy grant credibility to promised returns for taxes and transform taxation into a self-sustaining nonzero-sum game: revenue is secured by the ruler, whom taxpayers hold fiscally accountable, facilitating sustained investment in tax capacity. Deviations from this equilibrium activate a sanctioning mechanism, by which the ruler is denied tax receipts (i.e., taxpayers withdraw tax payments) or excluded from domestic loans.
Besley and Persson formalize the opportunities of sustained cooperation in tax policy in “common-interest states.”28 These states are characterized by political institutions that impose checks and balances on the executive and constrain the policies of incumbent governments.29 When such institutions are in place, taxation becomes a win-win game: the ruler secures a constant stream of funds to produce public goods while taxpayers are given guarantees that contributions will be spent responsibly. Because gain is mutual, power-sharing institutions are conducive to sustained investment in fiscal capacity.
More recently, Acemoglu and Robinson speak of “the Red Queen effect” to characterize a similar self-enforcing equilibrium between elites, who want to maximize state capacity, and nonelites, who seek to maximize individual liberties.30 Power-sharing institutions (e.g., constitutional checks and balances) are institutionalized mechanisms that balance the state’s and society’s powers and enable mutually beneficial solutions for the state and society—the ruler and taxpayers, respectively.
Stasavage argues that, once in place, power-sharing institutions are self-reinforcing because collective action problems are more easily overcome a second time; that is, taxpayers learn to coordinate their actions to hold the ruler accountable.31 Levi has also theorized about the benefits of addressing credibility issues with power-sharing institutions.32 Once the ruler’s promises turn credible, the costs of enforcement of taxation decrease, expanding the capacity to tax and fund public goods. In light of the social benefits of higher fiscal pressure, taxpayers are willing to comply “quasi-voluntarily”33 with taxation. Consistently, Dincecco shows that limited government in Europe increased total tax revenue in the long term.34 This evidence draws from an original and precious dataset on tax revenue and institutional reform for 11 European countries as early as 1650. Within-country variation allows Dincecco to assess the marginal change in tax collection derived from the adoption of limited government while keeping time-invariant country characteristics constant (e.g., geography and cultural traits).35
8.2.3 THE FISCAL CONTRACT AND EXTERNAL FINANCE
To understand the conditions under which the political mechanism of war finance is set in motion, an examination of the ruler’s incentives to assume the political ramifications of taxation is necessary. Incentives to strike deals with taxpayers—namely, abiding by a fiscal contract—can be expected to be endogenous to the set of alternatives to taxation. European monarchs might not have shared fiscal powers with taxpayers had they had access to cheap external finance.36 Philip II of Spain (r. 1556–1598), who can be used as a reasonable counterfactual, had access to external finance from Genoese bankers, thanks to the silver that poured in from the Americas and was used as collateral. Despite the many wars waged, Philip II and his successors did not implement significant self-strengthening reform. After the defeat of the armada in 1588, the Castilian parliament, known as the Cortes, was summoned by the king and briefly gained a voice in fiscal policy in return for new tax concessions; however, this arrangement was unsustainable. Cities were too dispersed to effectively overcome collective action problems in monitoring the Crown.37 In addition, the greatest downside of silver
was that it weakened the bargaining position of the Cortes vis-à-vis the Crown. Because of silver revenues, Castile’s rulers could spend freely using borrowed funds and effectively present the Cortes with the bill.38
In other words, imperial rule and war were mainly funded by silver from the Americas, allowing the Spanish kings to escape the political cost of taxation: limited government. After 1663, the Cortes was summoned only on ceremonial occasions, and fiscal fragmentation became an endemic problem. Spain entered a long period of economic and political decay.39
Far from surprising, the behavior of Philip II and his successors is consistent with the political economy of external finance elaborated in chapter 2: A ruler will share fiscal powers with taxpayers only as a last resource; namely, when other sources of revenue are unavailable or fall short, key among them external finance. Genoese bankers were long gone in the nineteenth century, but emerging economies had access to British capital (and later French, German, and American too). Cheap money siphoned in from Europe helped rulers in large parts of the developing world finance war and public infrastructure while saving them the political costs of taxation. Economic historian Leland H. Jenks summarized this logic as follows:
There was endless preparation for war. And there was war itself—an expensive pastime—inspired by the general excitation and apologized for by the fashion of nationality. All these things were progress. They all meant profit for the fortunate contractors [the underwriters]. They meant money for which the taxpayers must not be burdened. They meant continuous appeals to the money market. Between the universal desire for progress and the equally universal desire for lower taxes there was a discongruity which could be bridged only by public borrowing.40
The idea that foreign credit allows incumbents to dodge political accountability is not new. Because taxation “constitutes the largest intervention of government[s] in their subjects’ private li[ves],”41 tax hikes are expected to make taxpayers (elites or the general public or both) more attentive to the way government spends tax receipts. This is even more compelling in wartime, when the lives and assets of taxpayers may be at stake. To minimize political contestation, rulers may finance war in other ways, including confiscation or inflation, alternatives that might secure funds but can also create new grievances: government confiscation of factories or tithes can easily escalate into open conflict with elites or masses or both. Printing money may solve the liquidity shortage but can rapidly derail the economy.42 In order to avoid political and economic problems, rulers may turn to a less invasive war-financing mechanism: borrowing.
Along these lines, Patrick Shea claims that democracies are more likely to win war when external credit is cheap, not simply because it allows for more military spending but also because it mitigates the societal and political pressures attached to higher taxation. “Rulers who do not have to depend on their citizens for tax revenue or other economic resources have a freer hand in enacting policy.”43 Kreps offers a fascinating historical account of how American leaders deflect public opposition to war by turning to credit markets instead of imposing more visible and onerous war taxes. Borrowing allows rulers to “diffuse and defer” the cost of war compared to the “direct and immediate impact of taxation”;44 that is, borrowing loosens public constraints today and shifts the responsibility of servicing debt to future leaders. Ironically, this strategy seems to do the trick: using experimental surveys in the United States and the United Kingdom, Flores-Macías and Kreps show that public support for war declines by 10 percent as soon as respondents learn that war expenses will result in a new war tax (relative to a baseline condition in which war is funded with ordinary receipts and debt).45
Results in Kreps’s single-authored work and jointly with Flores-Macías are illuminating because they draw from stable regimes with institutions that anticipate the long-term consequences of today’s actions. Most emerging economies in the nineteenth century lacked institutional stability that could infuse fiscal policy with a long-term perspective. Market-based constraints were no stronger. The modest size and high vulnerability of native financial institutions to fluctuations in international markets46 limited the ability of local bankers to monopolize public debt issue, and hence their capacity to discipline government.47
In sum, for roughly a century external capital remained a key source of government funding in the Global South, offering as many opportunities as perverse incentives to unconstrained rulers. Foreign creditors’ enthusiasm to lend beyond the reasonable in expectation of high margins did not help. From the viewpoint of a sitting ruler, high indebtedness and draconian clauses in case of default were a problem for the future and likely somebody else’s.
8.2.4 AN EMPIRICAL EXAMINATION OF THE POLITICAL CHANNEL
The foregoing discussion suggests that the likelihood of any given ruler setting in motion the political mechanism of transmission—namely, agreeing to power-sharing institutions—is inversely proportional to having access to external capital. In other words, rulers will agree to assume the political costs of taxation only if they run out of options.
To evaluate this proposition empirically, I examine whether a history of war and credit exclusion in the long nineteenth century increased the likelihood of having limited government on the eve of World War I. If power-sharing institutions transform taxation into a nonzero-sum game benefiting both the ruler and taxpayers, limited government in the long run may be predicted based on past forms of war finance. Following this logic, I also evaluate the strength of power-sharing institutions circa 2000 as a function of war and credit exclusion in the nineteenth century. The empirical design replicates expression 7.2 while substituting the original outcome variable, a measure of state capacity, for one of power-sharing institutions: executive constraints. This variable, drawn from the Polity IV Project,48 encapsulates an immediate consequence of relinquishing fiscal power to taxpayers.
One important result in the political economy of external finance in chapter 2 is that the political cost of funding government with taxes depends on the initial level of power-sharing institutions. The marginal cost of one additional tax dollar is greater for an autocrat than for an elected president. Empirically, this calls for a control for initial political institutions. Data on executive constraints for the early nineteenth century are limited and overrepresent countries internationally recognized by the 1820s. In order to maximize the number of cases, I compute the average level of executive constraints between 1800 and 1830. This variable ranges from 0 (minimal constraint) to 7 (maximal constraint).49
Figure 8.1a plots the marginal effect of waging interstate war with and without access to international capital from 1816 to 1913 on average executive constraints by 1913, holding everything else constant. Figure 8.1b plots these estimates for the 2000s. These models indicate that a one-standard-deviation increase in the number of years at war while credit is tight in the nineteenth century increases average executive constraints by 16 percent in 1913 and 4.5 percent in the 2000s. By contrast, war waged with access to external credit is not associated with political change in the short or long term. If any, that relationship is negative.
Although modest, these results suggest that political reform is more likely when incumbents cannot escape the political costs of domestic taxation, that is, when they are at war but lack external finance. But does taxation always lead to advances in power-sharing institutions? In chapter 2, I pointed out two scope conditions that students of democracy find important for the rise and persistence of representative institutions: geographic scale and capital mobility. In large polities, taxpayers find it difficult to coordinate and monitor the executive, particularly so when the means of transportation and communication are antiquated.50 The ability of taxpayers to escape taxation is also important to understand when power-sharing institutions are set in motion. Rulers are compelled to grant power-sharing institutions when they seek to tax owners of mobile capital, such as traders and financiers, and less so when taxpayers derive income from fixed capital, such as land.51
FIGURE 8.1. Effect of War Finance on Executive Constraints in the Short (1900–1913) and Long Run (1995–2005). This figure plots marginal effects of the number of years at war from 1816 to 1913 when international capital flowed and stopped on short- and long-term executive constraints. Short-run executive constraints take the average value of executive constraints in Marshall and Jaggers (2000) between 1900 and 1913. Long-term executive constraints take the average value from 1995 to 2005. Capital access (or lack thereof) is measured by global shocks in capital flows or “sudden stops” (more in chapter 7). Interstate war data are drawn from Wimmer and Min (2009). Models control for population density in 1820, oil production, sea access, colonial past, and initial executive constraints (average from 1800 to 1830). The latter control restricts the sample to N = 30, 90% CI reported. Countries in the sample are Argentina, Austria, Belgium, Bolivia, Brazil, Chile, China, Denmark, Ecuador, France, Greece, Iran, Japan, Mexico, Morocco, Nepal, Netherlands, Norway, Paraguay, Peru, Portugal, Russia, Spain, Sweden, Thailand, Turkey, United States, Uruguay, and Venezuela. Great Britain is excluded to maximize exogeneity of capital access.
I reexamine the association between war finance and executive constraints through the prism of these scope conditions. I interact the number of years at war waged without access to external capital with measures of geographic scale and capital mobility. I expect collective action problems of taxpayers to be milder in smaller-scale polities, and their bargaining power vis-à-vis the ruler to be stronger in urbanized and monetized societies, which I approximate by levels of population density in 1820. I expect the interaction term between the number of years at war while credit stops in 1816–1913 and the proxies of geographic scale and capital mobility to be negative and positive, respectively.52
In column 1 of table 8.1, we see that the effect of tax-funded war on executive constraints attenuates as the geographic scale of a country increases.53 In column 2, I report results with region fixed effects to account for unobserved heterogeneity between continents. Results hold, although attenuated. Figure 8.2a shows a visualization of this interaction. Under capital exclusion, the marginal effect of war on executive constraints is positive as long as the size of the polity is below ln(Area) = 3, the size of the United Kingdom. In larger polities, the effect of war on executive constraints is zero. In columns 3 and 4, I evaluate the effect of war and capital exclusion for different levels of urbanization in 1820, the proxy of capital mobility.54 There we observe that the effect of tax-funded war on executive constraints strengthens as population density increases. The relationship turns statistically different from zero at Population Density = 0.68, corresponding to the value of Italy in 1820.
TABLE 8.1. Scope Conditions for the Activation of the Political Mechanism of Persistence
|
(1) |
(2) |
(3) |
(4) |
|
|
# Years at war while credit stops |
−0.101*** |
−0.064* |
||
|
1816–1913 × ln(Area) |
(0.036) |
(0.036) |
||
|
# Years at war while credit stops |
0.460*** |
0.284* |
||
|
1816–1913 × population density |
(0.169) |
(0.159) |
||
|
# Years at war while credit stops |
0.451*** |
0.325** |
−0.189* |
−0.099 |
|
1816–1913 |
(0.156) |
(0.138) |
(0.102) |
(0.109) |
|
ln(Area) |
0.713* |
0.512 |
0.791** |
0.600* |
|
(0.417) |
(0.434) |
(0.314) |
(0.327) |
|
|
Population density |
2.559* |
3.430*** |
||
|
(1.397) |
(1.217) |
|||
|
# Years at war while credit flows |
−0.064 |
−0.025 |
−0.115 |
−0.035 |
|
1816–1913 |
(0.081) |
(0.091) |
(0.088) |
(0.075) |
|
Controls |
Yes |
Yes |
Yes |
Yes |
|
Region FE |
No |
Yes |
No |
Yes |
|
Observations |
49 |
49 |
49 |
49 |
|
R-squared |
0.284 |
0.529 |
0.358 |
0.612 |
Note: The outcome variable is the average value of executive constraints between 1900 and 1913 in Marshall and Jaggers (2000). Capital access (or lack thereof) is measured by global shocks in capital flows or “sudden stops” (more in chapter 7). Interstate war data, urban density by 1820, and land area are drawn from Wimmer and Min (2009), World Mapper (www.worldmapper.org), and Nunn and Puga (2012), respectively. Models control for oil production, sea access, and colonial past. Countries in the sample: Argentina, Australia, Austria, Belgium, Bhutan, Bolivia, Brazil, Bulgaria, Canada, Chile, China, Colombia, Costa Rica, Denmark, Dominican Republic, Ecuador, El Salvador, Ethiopia, France, Germany, Greece, Guatemala, Honduras, Hungary, Iran, Italy, Japan, Mexico, Morocco, Nepal, Netherlands, New Zealand, Nicaragua, Norway, Panama, Paraguay, Peru, Portugal, Romania, Russia, South Africa, Spain, Sweden, Switzerland, Thailand, Turkey, United States, Uruguay, and Venezuela. Intercept not reported. Great Britain is excluded to maximize exogeneity of capital access. *** p < 0.01, ** p < 0.05, * p < 0.1.
Results in table 8.1 and figure 8.2 suggest that the political mechanism of transmission of war finance activates under limited conditions: small geographic scale and high initial wealth. When these conditions are met, mobilization of domestic resources for war finance puts in motion political bargaining between rulers and taxpayers, transforming taxation into a nonzero-sum game and carrying the fiscal effects of war into the long run. Not coincidentally, small scale and relative wealth were conditions generally met in the formative period of state building in Europe.
FIGURE 8.2. War and Activation of the Political Mechanism. These figures plot marginal effects of the number of years at war from 1816 to 1913 when international capital stopped on executive constraints in 1900–1913 as a function of geographic scale and initial wealth. Estimates drawn from saturated models in columns 2 and 4 in table 8.1. Due to small sample size, I report confidence intervals at 90%. Model specification and sources shown in table 8.1.
8.3 The Bureaucratic Channel of Persistence
Countries that did not meet the scope conditions for the activation of the political mechanism—large and low-populated economies—or that were deprived of self-government by a foreign power55 might have capitalized on tax-financed warfare by articulating a stronger administrative apparatus. That was the state building path chosen by some European economies before the nineteenth century (e.g., Prussia56) and by China during the Warring States period, 475–221 BCE.57 Next, I study the transmission of war effects through this alternative bureaucratic channel.
8.3.1 YES, MINISTER
The modern Weberian tax administration was created for and by war.58 Tax bureaucracies were necessary to assess wealth and collect taxes as well as to resist the natural aversion to having one’s sources of income monitored.59 Once created, bureaucracies entrenched, grew larger, and became states within states.60 In Charles Tilly’s words,
The organizations that were necessary to amass revenue [for war] developed interest, rights, perquisites, needs, and demands requiring attention on their own.… Bureaucracies developed their own interests and power bases throughout Europe.61
More generally, institutions originally built to finance the means of war can give rise to a body of bureaucrats that organically develops a vested interest in safeguarding institutional survival.62 Based on this logic, we can expect tax bureaucrats to oppose disinvestment in administrative capacity, carrying the fiscal effects of past warfare into the future.63
The bureaucratic channel potentially operates in sovereign and colonial states. As I discussed in chapter 6, colonies were responsible for financing administrative, infrastructural, and defense expenses. To secure funds, colonies relied on a variety of methods, including tariffs, excises, and poll taxes.64 Although defense expense was heavily subsidized, colonies were expected to contribute to imperial war.
European colonialism in the nineteenth century relied on direct and indirect rule, and generally a mix of both.65 Direct rule implanted bureaucratically centralized states with a substantial presence of European administrators. Indirect (or customary) rule relied on precolonial leaders (or chiefs) to maintain political and legal power, requiring low investment in colonial administration. Indirect rule has been generally associated with worse economic and political outcomes;66 however, recent work by political scientist Kate Baldwin and others offers a more benign assessment of chieftaincy for public goods provision and state development in the modern day.67
Although the direct-indirect division is heuristically convenient, many scholars argue that both forms of rule were often combined within the same colony. Capital cities were predominantly under direct rule, and the hinterland relied on different degrees of indirect rule depending on the strength of preexisting institutions and geographic conditions. Mamdani employs the expression “bifurcated state” to characterize the unequal presence of colonial rule within African countries,68 and Boone and Ricart-Huguet show rich qualitative and quantitative evidence of it, respectively.69
Thandika Mkandawire claims that domestic resource mobilization during colonial times “left an institutional and infrastructural residue that still plays a major role in the determination of tax policies and the capacity to collect tax.”70 Dan Berger offers an illustration by exploiting a rare geographic discontinuity in northern Nigeria: his cleanly identified empirical analysis finds evidence of the long-run persistence of colonial tax efforts as late as the 2000s.71
Evidence of the bureaucratic persistence of colonial tax institutions is also found in Latin America72 and Asia,73 and is the subject of a recent collective monograph edited by Ewout Frankema and Anne Booth.74 Building on this evidence and keeping in mind that colonial rule was heterogeneous between and within colonies, and that military expenses were largely subsidized by the metropole, I expect the incentives of colonial administrators to mobilize local tax revenue to strengthen if only at the margins in times of war and low liquidity in international capital markets, relative to times of war and high liquidity.
8.3.2 AN EMPIRICAL EXAMINATION OF THE BUREAUCRATIC CHANNEL
Drawing from historical data for both sovereign countries and colonies, next I investigate whether a history of war and exclusion from external finance in the nineteenth century fostered bureaucratic advances by 1913 and whether early reform persisted until the current day. In the absence of systematic data on the size or composition of the tax bureaucracy in the early twentieth century, I measure bureaucratic capacity on the eve of WWI with two reasonable proxies: census capacity (a measure introduced in chapter 7) and primary school enrollment.
Censuses have been conducted since antiquity for tax and conscription purposes.75 Modern censuses, which cover the entire territory and population regardless of gender, race, or legal status, were first implemented in the second half of the eighteenth century (Sweden conducted the earliest modern census in 1751). Censuses were key instruments in tax capacity building in the Global South too. For instance, they allowed for an extractive head or capitation tax, which was “the mortar with which, block by block, the colonial state [in Africa] was built.”76 The implementation of modern censuses was challenging because it required a systematized collection of information by an army of trained or professional surveyors who had to travel the entire territory while standardizing data collection.77 These population surveys represented a major administrative and logistical achievement for the state.78
Along with census technology, I rely on primary school enrollment as an alternative measure of bureaucratic capacity in the early twentieth century. Public systems of education were meant to homogenize civil values,79 if only because a sense of belonging—national identification—made the population willing to fight war and pay tax.80 The nationalization of public education was a major administrative endeavor, requiring a solid bureaucratic structure to secure local funds, recruit instructors, standardize curricula, and enforce attendance. Modern states were built around a national system of public education.81
Public education was also an important feature of state building in the colonial world, particularly for the French. Huillery finds persistence in education investments in French West Africa: higher ratios of teachers to students in the early twentieth century predict higher school attendance by 1995.82 Similarly, Wantchekon, Klašnja, and Novta show positive effects of early colonial schooling on living standards in Benin today.83 The British externalized education provision to Christian missions but subsidized school infrastructure on a regular basis.84
Building on the empirical design of chapter 7, I model the two proposed proxies of bureaucratic strength by 1913 on the number of years at war with and without access to external capital (refer to expression 7.2 for details). Column 1 in table 8.2 reports a linear probability model in which having a modern census by 1913 is regressed on war making and exogenous credit access between 1816 and 1913 plus controls. With 90 percent confidence, the probability of having adopted a modern census by 1913 increases by 3 percentage points for each additional year of war waged without access to external credit. Column 2 runs the same specification, replacing census technology for primary school enrollment by 1913. The latter increased by approximately 1 percentage point for each additional year that a country was at war while excluded from credit markets between 1816 and 1913. By contrast, waging war with access to international capital markets had no effect on either proxy of bureaucratic strength.
TABLE 8.2. Effect of Past Warfare and External Capital Access on State Capacity on the Eve of World War I
|
(1) |
(2) |
(3) |
(4) |
|
|
# Years at war while credit |
0.030* |
0.935* |
−3.024*** |
−3.465** |
|
stops 1816–1913 |
(0.018) |
(0.508) |
(0.827) |
(1.591) |
|
# Years at war while credit |
−0.012 |
−0.135 |
2.233** |
−1.598 |
|
flows 1816–1913 |
(0.016) |
(0.577) |
(0.970) |
(3.897) |
|
Region FE |
Yes |
Yes |
Yes |
Yes |
|
Colonial origins FE |
Yes |
Yes |
Yes |
Yes |
|
Observations |
98 |
76 |
103 |
56 |
|
R-squared |
0.362 |
0.863 |
0.565 |
0.649 |
Note: Primary education enrollment is drawn from Lee and Lee (2016). Information of census capacity is coded by author from Goyer and Draaijer (1992a, b, c). External capital access is exogenized by global credit crunches. Column 1 fits a linear probability model. Columns 2–4 are OLS. Controls include population density by 1820, sea access, desert territory, and state antiquity index (columns 3 and 4). Column 4 model includes only countries with colonial past (N = 56). Intercept not reported. Great Britain is excluded to maximize exogeneity of capital access. Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.
In column 3, I take advantage of the escalated dates of census adoption by assessing whether capital exclusion accelerated adoption (regardless of whether it took place before or after 1913). In that column, higher values of the dependent variable imply delay in census adoption.85 The estimates indicate that fighting wars under market exclusion between 1816 and 1913 accelerated census adoption at the rate of three years per additional war year. Fighting wars with access to credit delayed adoption at the rate of two years per war year. Finally, in column 4, I repeat the analysis only for states under colonial rule in the nineteenth century. Results are roughly equivalent: waging war while being part of an empire in periods of tight capital markets accelerated census adoption (hence the negative sign in the first entry in column 4). By contrast, when capital was abundant, war did not spur improvements in legibility in the colonies (that estimate is negative but not statistically different from zero).
Results in table 8.2 suggest that war finance in the nineteenth century shaped the bureaucratic breadth of states on the eve of WWI. Sovereign and colonial authorities were seemingly compelled to mobilize local resources at times when they could not count on external funding. Now, if bureaucracies are meant to stay—namely, if the bureaucratic channel holds—similar results may be found in the longer run. To examine this possibility, I focus on the size and endowment of tax administrations circa 2000. One may argue that large bureaucracies signal extended patronage practices, not state capacity. Ertman and Geddes show that this was the case in parts of early-modern Europe and twentieth-century Latin America, respectively.86 In figure 8.3a, I evaluate this possibility with contemporary data. Specifically, I plot the size of the tax administration circa 2005, measured by the number of tax officials per thousand capita, against total tax revenue. These two variables correlate at 0.68. Arguably, deviations from the mean might be suggestive of some patronage, but on average more staffed tax bureaucracies seem to point to more fiscal capacity.
Next, I assess the extent to which tax administrations today are shaped by war finance in the long nineteenth century. For consistency with previous tests, I fit expression 7.2, replacing the personal income tax variable with tax staff per thousand capita circa 2005. In figure 8.3b, I plot the marginal effects of the coefficients of interest. Based on this model, the effect of a one-standard-deviation increase of the number of war years under exclusion is equivalent to a jump from the twenty-fifth to the seventy-fifth percentile of the tax staff distribution. By contrast, waging war with access to international capital does not contribute to long-term bureaucratic capacity.
Figure 8.3b might raise concerns of history compression. To address this point, I work with data collected by Tait and Heller for a few selected countries regarding the structure of tax administration in the late 1970s and early 1980s.87 These data include the size of the finance and planning administration per hundred capita, which I interpret as the extensive margin of the effect of war. In the absence of budget data, I approximate the intensive margin of war finance on bureaucratic development—namely, how many resources are put into the tax administration—by the wage premium of the finance administration employees relative to other branches of central government.
FIGURE 8.3. Current and Historical Correlates of Tax Administration and War Finance. Tax staff per thousand capita and total taxation to GDP are drawn from USAID (2012). The regression model in (b) includes the following controls: Region FE, colonial origins FE, population density in 1820, oil production, sea access, desert territory, Great Powers indicator, and state antiquity index. Great Britain is excluded to maximize exogeneity of capital access. N = 78, 90% CI.
Despite the very small sample size, figure 8.4 suggests that nineteenth-century war waged without access to external finance is associated with larger and better-funded finance administrations in the 1970s, whereas war waged with access to external finance is not. In particular, a one-standard-deviation increase in the number of years at war when credit is tight in the nineteenth century raises the average size and the wage premium of the finance administration in the late 1970s by 49 percent and 22 percent, respectively.88
Together, figures 8.3 and 8.4 suggest that nineteenth-century war exerted a differential and persistent effect on bureaucratic capacity depending on external capital access. In the absence of foreign loans, rulers were compelled to articulate state strengthening institutions to collect funds for war. Once created, these new bureaucracies were to stay and expand over time. The case of Siam in the next chapter offers a good example.
8.3.3 CAPACITY OR WILLINGNESS
Results in figures 8.3 and 8.4 are helpful in assessing a common concern with empirical measures of state building, namely, the debate between willingness and capacity. Performance measures like income tax ratios may confound the effect of institutions with that of preferences. Countries with a long history of warfare may forge a strong sense of national identity,89 key to creating reciprocity norms, including taxation compliance.90 Results above suggest that high tax ratios are not just the result of intrinsic preferences but of stronger bureaucratic capacity. Countries that funded war in the past (partially) with taxes articulated larger bureaucracies and filled them with public servants, who, subject to strict controls and relatively sheltered from the spurious fleeting interests of passing incumbents, carried the fiscal effect of past warfare into the present day. That is the bureaucratic mechanism of persistence.
FIGURE 8.4. Effect of War Finance on the Size and Wage Premium of the Finance Administration in the Late 1970s and Early 1980s. These figures plot 90% CI. Sample size is limited by Tait and Heller’s (1983) data and covariate availability. For the tax size administration sample, full data are available for 23 out of the 35 countries in Tait and Heller (1983): Argentina, Belgium, Congo, Cyprus, Ecuador, El Salvador, Germany, Guatemala, Iceland, Ireland, Japan, Netherlands, New Zealand, Panama, Senegal, South Africa, South Korea, Sri Lanka, Swaziland, Sweden, United States, Zambia, and Zimbabwe. For the wage premium sample, full data are available for 15 out of the 26 countries in Tait and Heller (1983): Argentina, Cyprus, Ecuador, El Salvador, Iceland, Japan, New Zealand, Panama, South Africa, South Korea, Sri Lanka, Swaziland, United States, Zambia, and Zimbabwe. Great Britain is excluded to maximize exogeneity of capital access. Controls include population density as of 1820, oil producer, desert territory, and Great Powers and former colony indicators.
8.4 Conclusion
Building from the political economy of external finance in chapter 2, I articulate two explanations for the ratchet effect of war finance from the nineteenth century onward. The first explanation builds from the history of limited government in Western Europe. Political reform resulting from war finance can transform taxation into a nonzero-sum game: rulers secure funds for war, whereas taxpayers hold them accountable for spending decisions. Mutual gain makes the fiscal contract self-enforcing, carrying the effects of warfare into the future. Similar forms of tax bargaining between rulers and taxpayers, I argue, were set in motion in the Bond Era, coinciding with periods of low international liquidity and war; that is, when rulers were compelled to mobilize resources domestically. Now, taxation did not always lead to advances in executive constraints. Scope conditions for the taxation-representation connection in Europe seem to apply to the larger world: smaller and more densely populated states in the Bond Era were more likely to couple domestic resource mobilization during wartime with (stronger) representative institutions. The next chapter offers one such example: Chile.
Second, I claim that efforts to raise taxation for war to substitute for external capital can trigger persistent improvements in the tax administration in sovereign and nonsovereign countries. Once bureaucratic reforms are set in motion, the new administrators are likely to safeguard organizational survival by pressing for larger and better endowed structures, channeling the effects of past warfare into the future. I tested the political and bureaucratic mechanisms with available historical quantitative data, often less complete and accurate than one would wish. To reinforce the plausibility of the argument, next I reexamine both mechanisms of transmission with case studies for Argentina, Chile, Ethiopia, Japan, and Siam. Together, the quantitative and qualitative analyses seek to provide coherent and compelling observable evidence of the theoretical argument of the book.
1. Peacock and Wiseman (1961). See Rasler and Thompson (1985) for evidence of this pattern in major Western economies.
2. Peacock and Wiseman (1961, pp. 26–27).
3. Scheve and Stasavage (2010, 2012, 2016).
4. Lindert (2004).
5. Pitkin (1967); Manin (1997).
6. Abramson and Boix (2019); Lipset (1959); Jha (2015).
7. Stasavage (2016).
8. Schultz and Weingast (1998).
9. Refer to Boix and Svolik (2013), Gandhi (2008), Gandhi and Przeworski (2007), Meng (2020), and Svolik (2012) for comprehensive accounts of power-sharing institutions in modern autocracies.
10. Hui (2004).
11. Kennedy (2015, pp. 398–401). At its height, the caliphate spread from the Iberian Peninsula and the Maghreb in the west to modern-day Pakistan in the east and the Arabian Peninsula in the south.
12. A cursory review of recent work includes Abramson and Boix (2019), Boix (2015), Boucoyannis (2015), Cox (2016), Dincecco (2011), Ferejohn and Rosenbluth (2016), Scheve and Stasavage (2012), Stasavage (2011), and Van Zanden, Buringh, and Bosker (2012).
13. Boix (2015); Diamond (1997); Scott (2017); Tan (2015).
14. Marongiu (1968); Stasavage (2011).
15. Rogers (1995).
16. Hoffman and Rosenthal (2000).
17. Bates and Lien (1985); Levi (1988); North and Weingast (1989); Stasavage (2011).
18. ’tHart (1999) and Tracy (1985) for the Netherlands; Dickson (1967), Cox (2016), and O’Brien and Hunt (1993) for England.
19. Dincecco (2009, 2011).
20. See Downing (1993) and Van Zanden, Buringh, and Bosker (2012) for the dissolution or loss of powers of national assemblies in Europe coinciding with the military revolution.
21. Balla and Johnson (2009); Mousnier (1974). See also Spruyt (1994, p. 106) for the origins of tax bargaining between the king and the town burghers in late medieval times, and Le Bris and Tallec (2019) for the fiscal constraints that provincial assemblies imposed on the king in the absolutist era.
22. Tax farmers collected indirect taxes. Under Louis XIV, tax farm revenue accounted for virtually half the ordinary revenues of the Crown (Balla and Johnson, 2009, p. 815).
23. Schultz and Weingast (1998, p. 34).
24. Potter (2000, p. 622).
25. Hoffman, Postel-Vinay, and Rosenthal (2000). Constrained by the same institutions that Louis XIV—his great-great-grandfather—put in place a hundred years earlier, Louis XVI was forced to call the estates general in 1779, which eventually led to fundamental fiscal and political reforms (Balla and Johnson, 2009, p. 825).
26. Another example of limited government in a so-called absolutist regime can be found in Austria in the second half of the eighteenth century (Godsey, 2018).
27. Schultz and Weingast (1998). See Boix and Svolik (2013) for the foundations of limited authoritarian government.
28. Besley and Persson (2011, chs. 2 and 3).
29. A common-interest state can also be achieved if the opposition is represented in the policymaking process.
30. Acemoglu and Robinson (2019).
31. Stasavage (2011, 2020).
32. Levi (1988).
33. Levi (1988, p. 52).
34. Dincecco (2009, 2011).
35. The effect of limited government is maximal when the country has achieved fiscal centralization.
36. Refer to chapter 3 for the globalization of external finance in the nineteenth century.
37. Stasavage (2011, pp. 147–150).
38. Drelichman and Voth (2014, p. 267).
39. Elliott (1963); Grafe (2011).
40. Jenks (1927, p. 264).
41. Tilly (2009, p. xiii).
42. For the negative consequences of inflation tax at wartime, see Sprague (1917); for the manner in which the perceived benefits of inflation tax can change in light of experience, see Fujihira (2000).
43. Shea (2013, p. 773).
44. Kreps (2018, p. 9).
45. Flores-Macías and Kreps (2017).
46. In 1873 alone, 20 banks in Latin America (22% of the total) went bankrupt because of the global financial crisis.
47. Marichal and Barragán (2017) for strengths and weaknesses of early public banking in Latin America, and Calomiris and Haber (2014) for a political economy theory of financial underdevelopment. Accounts reflective of financial underdevelopment in Asia and Africa before 1914 can be found in the edited volume by Austin and Sugihara (1993).
48. Marshall and Jaggers (2000).
49. Refer to appendix L in Queralt (2019) for alternative variables and time ranges to measure initial political conditions.
50. Stasavage (2011).
51. Bates and Lien (1985); Boix (2003).
52. These models have the same covariates as those in table 8.1. However, I do not include a control for initial executive constraints because it is potentially endogenous to the conditioning variables: geographic scale and capital mobility. Because the sample expands from N = 30 to N = 49 once I drop the control for initial executive constraints, I recover the region fixed effects to minimize unobserved heterogeneity.
53. Here I opted for a log transformation of the country size variable, but results are the same if a linear effect is assumed.
54. I stick to the original variable in this analysis, hence consistent with every other test. Results hold if I log-transform population density, as I do with country area.
55. Tax bargaining and power-sharing institutions occurred exceptionally outside white settler colonies. For instance, in Ghana local chiefs organized into a local legislative assembly as early as 1852 to negotiate the terms of direct taxation (the poll tax) with British authorities (Aboagye and Hillbom, 2020; Prichard, 2015). The legislative assembly never met again after 1852, but it was replaced by other forms of local representation, including the Aborigines’ Society, a group of native elites (barristers, teachers, merchants, chiefs), which effectively gained veto power over fiscal policy (Wight, 1947, pp. 25–26). Other examples of fiscal contracts under colonial rule can be found in Bräutigam (2008) and Makgala (2004), but these cases were hardly the norm.
56. Downing (1993); Ertman (1997).
57. Hui (2004).
58. See Brewer (1988) for Europe and Young (1994) for an application to colonial Africa.
59. Daunton (2001).
60. Fischer and Lundgreen (1975); Schumpeter (1991); Weber (1978).
61. Tilly (1990, p. 115).
62. Niskanen (1994, ch. 4).
63. See Porter (1994, ch. 7) for the growth of federal bureaucracy and (nonmilitary) agencies in the US following each major war waged by the US since independence. See Carpenter (2001) and Skowronek (1982) for earlier achievements in bureaucratic autonomy in the US, and Silberman (1993) for a comparative study of bureaucratic growth and development in advanced economies.
64. Refer to edited volume by Frankema and Booth (2019) and Gardner (2012) for an overview.
65. See Lange, Mahoney, and vom Hau (2006) for an excellent summary of colonial rule, including for ideal types: settler, direct, indirect, and hybrid rule colonialism.
66. Acemoglu, Reed, and Robinson (2014); Crowder (1964); Mamdani (1996); Lange (2009). See Iyer (2010) for a competing view.
67. Baldwin (2015); Baldwin and Holzinger (2019); Logan (2009); Von Trotha (1996); van der Windt, Humphreys, Medina, Timmons, and Voors (2019).
68. Mamdani (1996, p. 18). See also Berman (1984).
69. Boone (2003); Ricart-Huguet (2021).
70. Mkandawire (2010, p. 1648).
71. Berger (2009). Hassan (2020) offers a case-specific account of administrative continuity in Kenya. See Berman and Tettey (2001) and Lange (2004) for comparative findings.
72. Lange, Mahoney, and vom Hau (2006).
73. Booth (2007); Cheung (2005); Slater (2010).
74. Frankema and Booth (2019).
75. Scott (2017).
76. Young (1994, p. 127).
77. D’Arcy and Nistotskaya (2018); Lee and Zhang (2017).
78. Brambor, Goenaga, Lindvall, and Teorell (2020).
79. Bandiera, Mohnen, Rasul, and Viarengo (2019); Paglayan (2021); Ramirez and Boli (1987); Weber (1976).
80. Alesina, Reich, and Riboni (2017); Levi (1997).
81. Ansell and Lindvall (2020); Weber (1976); Soifer (2015).
82. Huillery (2009).
83. Wantchekon, Klašnja, and Novta (2015).
84. Frankema (2012).
85. To account for differences in initial state capacity, I include a control for the state antiquity index. Refer to chapter 7 for further details.
86. Ertman (1997); Geddes (1994).
87. Tait and Heller (1983).
88. The prediction for administration size is unusually high because both this variable and the key predictor are highly skewed.
89. Alesina, Reich, and Riboni (2017).
90. Besley (2020); Levi (1997).