WARS AND THE NEED FOR REVENUE

Analyses of government spending shocks invariably point to the impact of wars (Barro Reference Barro1981; Burnside, Eichenbaum, and Fisher Reference Burnside, Eichenbaum and Fisher2003; Ramey Reference Ramey2011). The reason is that wars tend to be characterized by large, urgent revenue needs compared to the smoother expenditures experienced during peacetime (Brownlee Reference Brownlee1996). Figure 1 makes this point graphically, mapping periods of war onto U.S. government expenditures since 1789 and showing the often massive increases in spending that accompany war. Scholars have made two conflicting claims about how states will tend to meet these revenue needs. Schumpeter ([1918] Reference Schumpeter1954) and later Tilly (Reference Tilly1990), Levi (Reference Levi1997), and Reiter and Stam (Reference Reiter and Stam2002) suggest that, because of the magnitude of a war's revenue needs, states inevitably rely on taxation to generate the resources needed to fight the war. This view is at odds with Smith, who argues that leaders would be deterred from raising taxes “for fear of offending the people who, by so great and so sudden an increase of taxes, would soon be disgusted with the war. . .the facility of borrowing delivers them from the embarrassment which this fear and inability would otherwise occasion” ([1776] 1904, 37). Borrowing is a way to avoid the potential political and indeed administrative costs associated with war taxation, explaining why it is a staple in states’ war finance portfolios (Slantchev Reference Slantchev2012).

FIGURE 1.

U.S. Government Expenditures, 1789–2010

Source: Historical Statistics of the United States, Table Ea636–643: -Federal government expenditure, by major function: 1789–1970; Budget of the United States Government, Office of Management and Budget, 1962–2010.

The history of war finance in the United States paints a complex picture that incorporates elements of both perspectives. Although in many cases the country has financed wars through increased extraction, Table 1 shows that the United States has sometimes relied exclusively on a combination of borrowing and increasing the money supply (Cappella 2012; Goldin 1980). As we argue in the next sections, Smith's assertion is consistent with the general way in which different forms of war finance confront the populace with the costs of war, but it does not capture the specific distributional consequences of taxes that explain why some leaders resort to war taxes while others pursue alternative forms of war finance.

TABLE 1.

U.S. Wars and War Taxes, 1789–2010

Sources: See Online Appendix.

THE PARTISAN TURN TO WAR TAXES

As the previous section suggests, leaders can finance wars through taxes, which directly expose individuals to the costs of war and introduce political risk in terms of weakening public support for both the leader and the war. Alternatively, they can pursue finance options that impose costs indirectly, whether through borrowing or increasing the money supply. Although in general the directness of taxation is likely to be unpopular, tax policies have enormous distributional consequences, benefiting particular political constituencies and offsetting the political disincentives that Adam Smith attributed to war taxes. To the extent that political parties reflect the material preferences of their supporters (Alt Reference Alt1986; Hibbs Reference Hibbs1997), parties’ platforms will diverge based on the effects of different fiscal policies on their constituencies. Parties whose core constituencies benefit from the distributional impact of a particular tax arrangement tend to favor taxes, whereas those whose constituencies are adversely affected pursue revenue sources other than taxes (Dixit and Londregan Reference Dixit and Londregan1995; Ray Reference Ray1987). We suggest that this dynamic, in which partisan preferences on taxation emerge because they draw on different bases of support, drives decisions about whether to levy war taxes, with parties whose specific constituencies benefit from taxes favoring war taxes over other forms of war finance.

Although some scholars of the literature on domestic institutions and international conflict minimize the role of partisanship in crisis settings (Gowa Reference Gowa1998) or argue that outside threats unify domestic political elites (Coser Reference Coser1956; Kupchan and Trubowitz),Footnote ⁷ there are several reasons why politics-as-usual prevails when it comes to the question of war finance. One is that the meaning of bipartisan unity in the context of war finance is indeterminate, because neither taxation nor its alternatives are unambiguously more patriotic than the other. Second, wars involve enormous revenue needs, such that the impact of war finance policies is far greater than would be involved during peacetime (Brownlee Reference Brownlee1996; Hurst Reference Hurst1973). For example, the 1918 Revenue Act was designed to collect $6 billion in taxes (Blakey and Blakey Reference Blakey and Blakey1919), the equivalent of 8% of GDP, with a top tax rate of 77% and a lowest rate of 6%. The distributional consequences of wartime fiscal policies are considerable, and so, therefore, are the political stakes of debates about whether to issue a war tax.

A related reason why partisanship remains salient in terms of war finance is that, because of the magnitude of the fiscal needs, revenue policies implemented during wars can have durable consequences that outlast the wars and carry over well into peacetime (Mayhew Reference Mayhew2005). Civil War taxes established the terms of political discourse about the income tax that were eventually made constitutional (Ratner Reference Ratner1942, 134–35); the Spanish-American War telephone tax endured 108 years, ending in 2006 (Davidson Reference Davidson2006); and for World Wars I and II, “emergency-driven tax policies acquired a legitimacy and cultural force that helped keep them in place well after the emergency war was over” (Brownlee Reference Brownlee1996, 39).

Taken together, wars create high stakes in terms of the degree and longevity of their potential impacts. “Wars seem to be capable of generating whole new political universes,” (Mayhew Reference Mayhew2005), opening policy windows for durable change that might be closed during peacetime. Wars become opportunities to shift, sometimes permanently, the allocation of a large volume of resources. Recognizing this opportunity, Keynes (Reference Keynes1940, iii) urged the Labour government to finance World War II through progressive taxation, stating, “I have endeavoured to snatch from the exigency of war positive social improvements.” Henry Morgenthau, the U.S. secretary of the treasury, expressed a similar preference for American war finance: “The function of wartime taxation is to distribute the inevitable burden more equitably” (quoted in Sparrow Reference Sparrow1996, 109–10). War may have created the revenue needs behind progressive taxes (Scheve and Stasavage Reference Scheve and Stasavage2010), but the resort to taxation as the form of finance is as much about redistributive politics, in which partisan differences loom large, as about economics.

The specific effect of partisanship on war taxes has shifted over time as the prevailing type of taxation and its distributional consequences have changed. Before passage of the Sixteenth Amendment in 1913, taxes tended to mean ad valorem taxes and, in particular, protective tariffs. The tariff issue was a highly contentious source of division between the two major parties. As Bensel (Reference Bensel2000, 468–69) notes, “the tariff so starkly redistributed wealth between the sections and the socio-economic bases of the two parties almost exactly divided the electorate into winners and losers.” In the case of tariffs, the winners were the manufacturing sectors because protective tariffs made the products of domestic industries more attractive. The losers of protective tariffs were the export-producing agricultural sectors, with “the tax on imports in actuality working as a tax on exports” (James Reference James1978, 231). Republicans, drawing their support from the manufacturing sectors of the Northeast and Great Lakes regions, favored protective tariffs with an eye toward privileging their core constituencies. Democrats, supported by agrarian interests in the South and West, tended to oppose tariffs (Witte Reference Witte1985, 68). Sponsorship of and votes on tariff legislation during the nineteenth century were consistent with these partisan differences. Almost without exception, Republicans imposed taxation, whereas Democrats favored free trade (Bensel Reference Bensel2000, 468–72).

These preferences on taxation began to reverse after the economic depression of 1893 (Joseph Reference Joseph2004). Democrats, led by the Populists and Progressives, turned to a reformist agenda in the form of a progressive tax that would appeal to farmers and augment their support among constituents in the West and South (Baack and Ray Reference Baack and John Ray1985; Saldin Reference Saldin2010). Advancing the populist reform position as a member of the House Ways and Means Committee, William Jennings Bryan advocated the income tax as a more progressive and equitable form of revenue generation than taxes on trade (Reference Bryan1909, 163): “[W]hy should not those sections pay most which enjoy most?” he asked. In 1894, Congress passed an income tax of 2%, but the Supreme Court ruled it unconstitutional in 1895 (Barclay Reference Barclay1987; Dunbar Reference Dunbar1894). Progressives continued to advocate for the income tax on the basis of a “burden-redistributing argument” (King Reference King1983, 10), culminating in the Sixteenth Amendment, which authorized the income tax in 1913.

Before passage of that amendment, the U.S. government had collected almost none of its tax revenues from income. Throughout the nineteenth century, tariffs comprised more than 90% of federal revenue.Footnote ⁸ After 1913, taxation increasingly meant income tax. The contribution of ad valorem taxes dropped to less than 10% over the course of the twentieth century, with income taxes—individual, corporate, and payrollFootnote ⁹—coming to contribute close to 90% of federal tax revenues compared to 2% at the beginning of the twentieth century (Blakey and Blakey Reference Blakey and Blakey1940, 3; Witte Reference Witte1985, 80).

As income taxes replaced ad valorem taxes as the primary source of revenues, Republicans, supported by business interests with a preference for low taxes, generally became a party that opposed taxes (Quinn and Shapiro Reference Quinn and Shapiro1991; Ray Reference Ray1987). Over time Democrats became more likely to support higher taxes, because taxes came to mean income taxes, a progressive measure that their supporters backed (Cox and McCubbins Reference Cox, McCubbins and Cox1991; Snyder and Groseclose Reference Snyder and Groseclose2000).Footnote ¹⁰

In sum, general voter antipathy regarding taxation tends to mask the distributional effects of varying tax structures. Leaders from parties whose core constituencies benefit from changes to the tax structure tend to incorporate these preferences into their decisions about whether to generate revenues through taxes, which explains partisan fiscal differences. Because of the high stakes generated by the magnitude and durability of war's fiscal policies, we expect these partisan differences to carry over into wartime and play a key role in whether leaders seek to generate war revenues through taxes rather than alternative measures. In the following sections, we test the role of partisanship as a determinant of war taxation in three ways: a statistical analysis of presidential partisanship, a roll-call analysis of legislative partisanship, and a historical case study of the Civil War.

ASSESSING THE EFFECT OF PRESIDENTIAL PARTISANSHIP ON AMERICAN WAR TAXATION

As the first part of our empirical strategy, we created a new dataset of war taxes adopted in the United States between 1789 and 2010. Controlling for a number of political and economic factors that could affect the propensity to levy a war tax, we analyzed the independent influence of the president's party on the likelihood of adopting a war tax. Because the dependent variable—whether a war tax is levied in a particular year (discussed in detail later)—is dichotomous, we used logistic regression to estimate the effect of leaders’ partisan differences and to account probabilistically for the influence of other factors that could affect war finance.

Although the number of actual instances of war taxes (20) is not large, we were interested in the conditions that produced a war tax and therefore needed to consider instances in which leaders might have had an opportunity to levy a war tax but did not (Geddes Reference Geddes1990; Poast Reference Poast2012). U.S. history suggests that the opportunities that are “at risk” for a war tax arise not just during wartime but also in the years before, after, and altogether outside major wars. The United States has levied war taxes before its involvement in a war, such as in World War I and II; to mobilize for conflicts that never reach technical thresholds of war, such as the 1798 Quasi-War; and after the end of a war to pay retroactively for wars, such as the Revolutionary War and World War I. If we included only years in which the United States was engaged in a full-fledged war, we would fail to capture these war taxes. However, we also needed to guard against treating all years the same. Doing so would equate the decision not to adopt a war tax surrounding a conflict with the decision not to adopt a war tax when there is no conflict.

With these considerations in mind, we followed three approaches in structuring our analysis. The first approach restricted the number of observations to those years that were most proximate to major conflict and therefore most plausibly at risk for the adoption of a war tax. Accordingly, in Models 1 and 2 we restricted observations to years in which the United States was involved in a warFootnote ¹¹ plus five years after the war to take into account retroactive taxation, as seen after the Revolutionary War to pay off wartime debts, for example. We also included years in which the United States levied taxes in anticipation of a war, such as in 1914 in preparation for World War I or in 1940 for World War II.

A second approach was more inclusive and considered the Cold War years as at risk for a war tax. Before the Korean War, President Truman proposed a war tax that the chairman of the House Ways and Means Committee described as being able to help “build a better A-bomb than the one Joe Stalin's cooking up for us” (Paul Reference Paul1954, 599). Throughout the Cold War period, defense spending as a percentage of GDP remained high; for example, “over half of the current US debt was incurred in the 1980s for an unprecedented peacetime buildup of more than 50 percent in real terms” (Markusen Reference Markusen1993, 391). Given these high “peacetime” expenditures and the real possibility of a Cold War tax, Models 3 and 4 followed the restrictions in Models 1 and 2 but included the Cold War period as at risk for war taxation.

The third approach was the most inclusive and considered all years in which the United States engaged in the deployment of military force between 1789 and 2010. This approach has the virtue of hewing most closely to the historical record of war taxes, because the United States has been involved in some sort of war, occupation, or militarized dispute in nearly every year of its history (Grimmett Reference Grimmett2002).Footnote ¹² In models 5 and 6, we included all years in which hostilities reached mobilizational levels, measured as those reaching a hostility level higher than 3 in the Militarized Interstate Dispute database 3.10 (Ghosn, Palmer, and Bremer Reference Ghosn, Palmer and Bremer2004). Other ways to restrict the sample, including restricting it to a hostility level of 3 or higher and not excluding any observations, did not affect our findings.

Results

Table 2 shows the results from our analysis of the effect of presidential partisanship on war taxation.Footnote ¹⁸ Models 1, 3, and 5 represent baselines without control variables, whereas Models 2, 4, and 6 include controls. Across all models, there is strong support for the effect of partisanship. The severity of the conflict is also significant across models with controls, suggesting that a war tax becomes more likely as the level of hostility increases. Other controls find support in only some models or do not seem to play a significant role on average.

TABLE 2.

Logit Estimation of the Effect of Presidential Partisanship on War Taxes, 1789–2010

Note: Robust standard errors reported in parenthesis. * p < 0.10; ** p < 0.05; *** p < 0.01.

To interpret the substantive effect of these results, we used Clarify (King, Tomz, and Wittenberg Reference King, Tomz and Wittenberg2003) to estimate change in predicted probabilities of interest. Based on the model with controls that shows the most conservative results (Model 6) and setting all other variables to their means (and medians for dummy variables), the marginal effect of partisanship on the probability of adopting a war tax is 6.1 percentage points.

TABLE 3.

Marginal Effects on the Probability of a War Tax

Note: The marginal effects are estimated by setting all other variables to their medians and means. Dummy variables (i.e., Retaliate, Election Year, Divided, Type, and Party) are set to their medians, whereas the remaining variables are set to their means.

To test the sensitivity of these results, we carried out a number of robustness checks. In addition to following an alternative coding for our main independent variable—as an interaction term, as discussed earlier—we operationalized several control variables differently. We operationalized the cost variable in levels rather than change, with and without a lag and three- and five-year moving averages. We considered alternative measures of cost, such as military personnel as a percentage of total population, as well as a dummy for where the war was fought. Moreover, in case the adoption of war taxes was affected by the existence of a central bank or its independence of the executive branch, we included two additional controls: a dummy taking the value of 1 for the years when a central bank has existed in the Unites States (1791–1811, 1816–36, 1913–onward) and a dummy taking the value of 1 for the period of central bank independence (1951–onward). We also estimated the models including only those years in which severity of the conflict was greater than 2, and in an unrestricted sample including all years. None of these changes affected our findings.

We also probed whether political party influences war taxation or whether it is endogenous to war onset and whether war taxes are likely to follow once wars begin. At first glance, it seems plausible that the pro-tax party is more warlike and that with more wars come more opportunity for war taxes. Republicans were associated with many wars of the nineteenth century—the Civil War and Spanish-American Wars—when they were also the party more likely to tax. Democrats were associated with many wars of the twentieth century—both world wars, Korea, and Vietnam—at a time in which they were the party more sympathetic to taxation. To test for this possibility, we reestimated the models using war onset as the dependent variable instead of war taxes, along with standard control variables for war onset.Footnote ¹⁹ A leader's political party did not affect whether the United States became involved in a war (p value of 0.56), suggesting that the propensity to levy a war tax does not only reflect the fact that particular parties are more inclined to become involved in wars.

CONGRESSIONAL ROLL-CALL VOTES ON WAR TAX LEGISLATION

As a second component of our empirical strategy, we tested whether the partisanship argument applies to members of Congress by analyzing legislative roll-call votes. The dataset includes all legislative votes (6,814 votes) for every war tax bill voted in the House of Representatives since 1789.Footnote ²⁰ In addition to testing legislative partisanship, this approach has the advantage of not being contingent on identifying the years “at risk” for a war tax, because it focuses on war tax votes only.

As with the previous analysis, the dependent variable is dichotomous, with each legislator's vote coded as 1 when a legislator voted in favor of a proposed war tax and 0 otherwise—an alternative coding excluding abstentions and no shows is presented in the Online Appendix. The independent variable of interest, party, took on a value of 1 whenever the legislator belonged to the party with a pro-tax inclination and a 0 otherwise.Footnote ²¹ As with the analysis of the president's party, we addressed potential time dependence by including the variable numberyears along with three cubic splines (Beck, Katz, and Tucker Reference Beck, Katz and Tucker1998). Model 7 shows the effect of party without controls, and Model 8 includes controls.Footnote ²²

As shown in Table 4, our analysis finds that partisanship is a significant determinant of legislators’ wartime fiscal policy preferences, increasing our confidence in the findings of the previous section. As Table 5 illustrates, the marginal effect of changes in partisanship on legislators’ probability of voting for a war tax is 24 percentage points—which is larger than changes in every other factor.Footnote ²³

TABLE 4.

Logit Estimation of the Effect of Partisanship on Congressional War Tax Votes, 1789–2010

Note: Dependent variable is a dummy that takes the value of 1 whenever a legislator voted in favor of a war tax. Robust standard errors reported in parenthesis. * p < 0.10; ** p < 0.05; *** p < 0.01.

TABLE 5.

Marginal Effects on the Probability of Voting for a War Tax

Note: The marginal effects are estimated by setting all other variables to their medians and means. Dummy variables (i.e., Retaliate, Election Year, Divided, Type, and Party) are set to their medians, whereas the remaining variables are set to their means. The cubic splines are dropped because their high correlation (0.99) prevents the matrix from inverting when estimating the marginal effects.

The congressional debate and subsequent vote on legislation to finance the Spanish-American War illustrate this dynamic in which, rather than legislators setting aside partisan rancor during wartime, partisanship drives congressional behavior. Although legislators were supportive of the war and for the need to raise revenue, both parties used the “crisis of war” (Ratner Reference Ratner1942, 234) to try to shift resources toward their constituents. Republicans favored tariffs and excise taxes supported by their industry supporters. Democrats, who “indicted Republicans as agents of capitalists. . .saddling costs on the common people,” opposed the war tax proposal because of the view that such taxes favored industry at the expense of lower socioeconomic groups (Ratner Reference Ratner1942, 232). The San Francisco Star summarized the stakes of the war finance debate as follows: “The war against privilege is more important than the war against Spain, of which the privileged are taking advantage to increase their own emoluments and other people's burdens” (quoted in Ratner Reference Ratner1942, 238). Amidst this acrimonious debate, Republicans prepared the war tax legislation without the consultation of Democrats, who did not see the legislation until it was presented to the full committee. The Chicago Daily Tribune (“May Defeat the War Tax Bill” 1898), whose coverage of the conflict tended to support Republicans, concluded that Democrats “are inclined to think that they have been most unfairly treated” by these procedural slights. Ultimately, only 5 Democrats voted in favor and 81 were opposed, compared to 148 Republicans in favor and only 5 opposed, suggesting that political elites used the crisis as a vehicle for advancing partisan fiscal interests.

TRACING PARTISAN FISCAL DIFFERENCES: THE CASE OF THE CIVIL WAR

Our third test of partisan differences on war taxes is a case study of the American Civil War. As Schwab notes (1892, 38), “in many ways the North and the South were similarly situated: their organic laws closely resembled each other; their commercial and financial traditions were the same; their financiers had been trained in the same school in Washington; and the demand for revenue was equally pressing.”Footnote ²⁴ And yet the two sides generated revenues quite differently, with the North financing 20% of the war through war taxes, compared to the South's approach in which war taxes provided just 4% of the conflict's revenue needs (Ball Reference Ball1991; Schwab Reference Schwab1892). As such, the case of the Civil War helps illustrate our claims that the distributive interests behind partisanship influence war finance decisions.

With the secession of the South, the Confederate Congress became a one-party system comprised of individuals who had been Democrats. The Confederate constitution codified the Democrats’ abhorrence toward tariffs, stating that “no bounties shall be granted from the Treasury; nor shall any duties of taxes on importations from foreign nations be laid to promote or foster any branch of industry” (Davis [1881] Reference Davis1990, 565). In short, the constitution prevented the Confederacy from raising protective tariffs. This prohibition was motivated by the understanding that taxes in the antebellum period protected Northern manufacturing interests but were disadvantageous to the South for several reasons. Export duties were thought to “penalize the products of slave labor,” such as cotton, which constituted one-third of the South's exports (Ball Reference Ball1991, 42). Import duties were unpopular because, except for agricultural commodities, the South imported all its goods (Holcombe Reference Holcombe1992, 766). In short, the South “resented high tariffs as a tool of Northern interests” and sought methods other than taxation to finance the war (Weisman 2002, 60).

The result of the Confederate's aversion to taxation meant an initial reliance on loans and treasury notes to finance the war. Although the Confederacy did pass a war tax in the first two years of the conflict, it was a nominal tax; “the Government's policy was to make the burden of the war fall as lightly as possible upon the people” (Todd Reference Todd1954, 136). As the war continued, the South suffered a number of military setbacks, was unable to pay soldiers, was hurt by credibility losses in the bond market, and experienced skyrocketing levels of inflation (Ball Reference Ball1991 41, 119). Even reluctant Democratic congressmen came to the conclusion that the South had to generate additional sources of revenue and finally resorted to a more systematic approach to taxation. In 1863, the Confederacy passed an “act to lay taxes for the Common Defense and Carry on the Government of the Confederate States,” a set of taxes that reflected the interests of the Confederacy by exempting real estate, property, slaves, and religious institutions (Todd Reference Todd1954, 140); however, this set of taxes proved inadequate because of the large numbers of exemptions. Because “the necessities of war no longer permitted a strict adherence to the letter of the [Confederate] Constitution” (Todd Reference Todd1954, 148) the Confederacy had to seek more draconian tax measures in 1864; the most aggressive tax measure came only in 1865, but the Civil War ended before the Confederacy could collect the new revenues.

In contrast to the South's reluctance to raise taxes, the North more readily resorted to taxation as a way to finance the war's costs. One of the main reasons has to do with the composition of the Union Congress, which was more than three-quarters Republican. In the North, manufacturing sectors favored higher tariffs and formed the core constituency of Northeast Republicans, who in turn incorporated high tariffs into their fiscal platform (Bensel Reference Bensel2000; Stanwood Reference Stanwood1903). One of the first measures the Union Congress took after secession was to pass the Morrill Tariff of March 1861, which reintroduced protective tariffs that had been lowered through Democrat-sponsored tariff legislation of the 1840s and 1850s (Taussig Reference Taussig1910, 72).

Because the Republican-dominated Congress strongly favored tariffs, these taxes became an obvious way to finance the Civil War. During the summer of 1861, Congress passed emergency customs duties, followed by additional duties in December 1861; in almost every month between then and 1865 Congress increased the rate of almost every form of ad valorem tax (Mitchell Reference Mitchell1903, 120; Taussig Reference Taussig1910). The North did ultimately levy an income tax; Senator John Sherman (R-OH) attributed its willingness to do so only to the fact that “we are in the midst of a great war, which tasks the industry and wealth of this country. . .we are about to inaugurate a system of taxation that is unknown in this country” (Congressional Globe 1861–62). That the Republican-dominated Congress was uncomfortable with the income tax was apparent from its aggressive efforts to repeal it, which ultimately came about in 1872.

The case of the Civil War, in which we might expect to see similar fiscal policies in the North and South because of similar fiscal demands, among other reasons, lends further support to the partisan differences thesis. The Union, dominated by Republicans who drew their support from manufacturing sectors that favored protective taxes, readily resorted to taxation and financed a larger proportion of the war through war taxes. In contrast, the Confederacy, effectively a one-party system of Democrats supported by export-dependent agricultural sectors that favored free-trade policies, only belatedly raised taxes to finance a small share of the war's cost. Our emphasis on partisan differences between the North and South does not mean other factors did not play a role. For example, the South's trade relationship with cotton-dependent European countries led to the belief that the South could secure loans in Europe with the cotton as collateral, whereas the North did not expect such assistance. Nonetheless, taken together with the other two forms of evidence, the case study further strengthens our argument about partisan influences on war finance policies.

Supplementary materials

To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0003055413000476