In marked contrast to what Stein, Burns, and Brown once said about Herbert Hoover’s suicidal tax increases in 1932, Cole and Ohanian (1999: 12) recently wrote that “tax rates on both labor and capital changed very little in 1929–33, which implies that they were not important for the decline.” Yet marginal personal tax rates on both labor and capital (e.g., small business and partnership income, dividends, interest, rent, and capital gains) were greatly increased in 1932 on everyone who was not exempt and exemptions were reduced. Marginal excise tax rates were greatly increased in 1932 and 1936. The effective corporate tax rate was increased from 11 percent in 1929 to 13.75 percent in 1932, 17.6 percent in 1937, and 19 percent in 1938 (Seater 1982: 363).
The 1932 tax law involved raising the top tax rate from 25 percent to 63 percent, quadrupling the lowest rate from 1.1 percent to 4 percent, and increasing corporate and estate tax rates. It is important to realize that federal excise and state sales tax rates also affect marginal rates on factor incomes, and both rose in the 1930s. Hoover’s 1932 tax was mainly about broader and higher excise taxes on goods and services. Those increased excise taxes (as well as the infamous tariffs of 1930) ultimately fell on factor incomes. Taxes on the uses of income, like taxes on the sources of income, reduce after‐tax rewards to people who supply the complementary factors of labor and capital.
To focus exclusively on marginal tax rates on only personal income and at only the federal level, clearly understates the increases in marginal tax rates on income and sales at the federal, state, and local levels. In an understatement, Romer and Romer (2012: 15–16) briefly mentioned in their background notes that “the majority of the [estimated] revenue effects were due to a vast increase in excise taxes. The most notable of these was an across‐the‐board tax of 2¼ percent on all manufactured articles.”
The revenue‐maximizing top tax rate formula that Romer and Romer borrow from Diamond and Saez (2011: 168) was never intended to be a top rate for the federal income tax alone as they imply, but to include all taxes on income, payrolls, and sales at the federal, state, and local levels. Even for the narrow purpose of estimating revenue‐maximizing tax rates, it is a serious omission for Romer and Romer and Goolsbee to ignore federal and state excise and sales taxes, as well as state income and sales taxes. In fact, to focus only on the economic impact of the federal tax on individual incomes in 1932–40 is to leave out 92.5 percent of all federal, state, and local tax receipts in those years (BEA 2020: Tables 3.4 and 3.5).
In 1932, federal income tax rates were roughly doubled at all income levels, with larger increases at the lowest and highest incomes. In 1934–36, there were additional increases in income taxes on capital gains and dividends, and marginal tax rates were again increased on net incomes of $50,000 or more. In 1929, the federal income tax on individuals accounted for 50.9 percent of all federal taxes and 14.1 percent of all federal, state, and local taxes. From 1932 to 1940, by contrast, the federal income tax accounted for an average of 25.8 percent of all federal taxes and only 7.5 percent of total federal, state, and local taxes (BEA 2020: Tables 3.4 and 3.5). If the objective of high marginal tax rates on personal income from labor and capital was to maximize revenue, regardless of any adverse impact on growth of the economy (and therefore growth of taxable income), the Hoover and Roosevelt plans to fix the Depression with higher taxes in 1932–36 were a total failure.
To point out that high and rising federal income tax rates in 1932–36 reduced revenue from $1,179 million in 1929 to $855 million in 1939, certainly does not imply that steep marginal tax rates were small in terms of distortions, disincentives, or tax avoidance. On the contrary, the decade‐long inverse relationship between tax rates and revenue suggests high tax rates were extremely inefficient, causing large deadweight costs with no lasting revenue gain. The greatly increased tax rates on realized capital gains from 1934 to mid‐1938 are a good example. By discouraging asset trading, high tax rates on realized gains (47 percent after two years) reduced realizations and tax receipts. Falling tax receipts, in turn, makes those higher tax rates look unimportant in terms of static bookkeeping and Keynesian macroeconomics. Yet falling revenue from higher tax rates illustrates the high cost of punitive marginal tax rates with no offsetting benefits.
Despite large increases in many such distortive federal and state marginal tax rates on what people earn and on how they spend those earnings, Mulligan (2002) nonetheless relies only the federal tax on individual labor income to agree with Cole and Ohanian that new taxes on labor and capital were trivial and insignificant in 1932–33 (and Mulligan includes 1936–38). He remarks that
Cole and Ohanian (1999) suggest that … taxes on factor incomes might help explain some of the Depression economy.… Of course, taxes on labor income create such a wedge, but Barro and Sahasakul’s study suggests that federal taxes on payroll and individual income were trivial, and unchanging.… [The] vast majority of the population did not file individual income tax returns during the 1930s, so that any IRS‐induced tax wedge affected very few people (not to mention small effects for the few affected) [Mulligan 2002: 27].
Mulligan emphasizes only the labor tax wedge (without mentioning the 1937 payroll tax) while neglecting marginal tax rates on capital incomes: higher corporate tax rates, surtaxes on undistributed profits, higher individual tax rates on partnerships and small enterprises, higher tax rates on capital gains in 1934–37, on dividends in 1936, and raising the top tax on estates from 20 percent in 1931 to 70 percent 1936.
Wright (1969: 300) estimates the average U.S. marginal tax rate on dividends rose from 11.1 percent in 1928 to 27.1 percent in 1932 and 29.9 percent in 1936, and the marginal tax on interest income rose from 9.2 percent in 1928 to 14.4 percent in 1932 and 17.6 percent in 1936. Federal taxes on gifts and estates rose from $35 million in 1932 to $379 million in 1936—nearly half as large as the $732 million collected that year from individual income taxes (Joulfaian 2007: Table 6).
The only reason both Mulligan and Cole and Ohanian imagine nothing significant happened to tax policy in the 1930s is that both rely entirely on inappropriate or irrelevant estimates of an income‐weighted “average” of marginal tax rates from the federal individual income tax alone.
To explain why they suppose “tax rates on both labor and capital changed very little 1929–33,” Cole and Ohanian cite a 1981 study by Joines, which shows average marginal rates on labor little changed until 1937 when the Social Security tax was added. Mulligan (2002) cites similar 1983 estimates from Barro and Sahasakul (1983). Yet, Barro and Sahasakul were highly critical of the income tax rate increases of 1932 and 1936, which they explained in detail. They wryly concluded: “Apparently the tax increases between 1932 and 1936 reflect the Hoover‐Roosevelt program for fighting the Depression!” (Barro and Sahasakul 1983: 21).
Barro and Sahasakul (ibid.: 1) “argue that the explicit rate from the schedule is the right concept for many purposes.” And those rates, they explicitly explained, were greatly increased between 1932 and 1936. Yet the same authors’ income‐weighted “average marginal tax rate” fell from 4.1 percent in 1928 to 2.9 percent in 1932 and 3.1 percent in 1933. That is why Mulligan (2002: 27) concludes, “Barro and Sahasakul’s study suggests that federal taxes on payroll and individual income were trivial, and unchanging.” On the contrary, Barro and Sahasakul (1983: 1) derided higher income tax rates in 1932 and 1936 and apologized for not including payroll taxes (until later). They also apologized for excluding excise taxes. “A full measure of marginal tax rates,” they wrote, “would incorporate other levies, some of which are based on property or expenditures” (ibid: 2). Increased federal expenditure (excise) taxes on dozens of good and services directly affected millions more people than the individual income tax, and excise taxes brought in over twice as much revenue (an average of $1,536 million per year from 1932 to 1939 compared with $728 million for the income tax).
The seemingly paradoxical decline from 1928 to 1932 in the Barro and Sahasakul “average marginal tax rate” does not show that marginal tax rates fell in 1932 or were either trivial or unchanging. What it does show is that the 1928–29 income‐weighted average was heavily weighted toward a rising number of high incomes above $50,000 taxed at rates of 18–25 percent, while the 1932–35 average was instead heavily weighted toward many more low incomes taxed at 4–10 percent (up from 1.5–5 percent in 1931).
As Seater (1982: 354) explained, an “income‐weighted average marginal rate” is computed by “using for weights the fraction of income that fell within each income class.” Because very little income still fell into the high‐income tax brackets after 1932, the weight of high‐tax brackets within the average was diluted. This is an entirely different concept of “average marginal rate” than I used in Figure 1, which is an unweighted average of statutory rates.
Income‐weighted averages of marginal tax rates in this period are a roundabout way of describing the same phenomenon as Figure 1—namely, that many high incomes dropped out of those averages after 1932. Reported net incomes above $50,000 collapsed from $6.3 billion in 1928 to $800 million in 1932 and never again even hit $2 billion before WWII. As billions of dollars of high taxable incomes vanished when faced with the high 1932–35 marginal tax rates, all the income‐weighted “average marginal tax rates” became dominated by much larger weights then attached to numerous taxpayers with relatively inelastic lower incomes.
Mulligan focuses on the fact that few people paid federal income tax after 1932—especially fewer high‐income people—than before. Yet it is also true that relatively few people start new businesses, employ many people, or supply capital to those who do. As McGrattan (2012: 2) notes, “Although few households paid income taxes, those who did earned almost all of the income distributed by corporations and unincorporated businesses. If the increases in rates were not completely unexpected [Hoover announced them in December 1931], these households would have foreseen large declines in future gross returns on investments … even before 1932 when major changes were enacted.”
When it comes to blaming the Depression’s severity and length partly on bad tax policy, McGrattan and others remain somewhat closer to the early postwar consensus of Stein, Burns, and Brown. McGrattan assigns considerable blame for the stubborn severity of the post‐1932 depression in general, and 1937–38 recession in particular, on high marginal corporate and individual tax rates on capital. That includes new taxes on retained earnings and higher taxes on dividends in 1936, plus high tax rates on most realized capital gains from 1934 until early 1938.
Calomiris and Hubbard (1995) also find the 1936–37 surtax on undistributed corporate profits was particularly harmful to the working capital and plant and equipment outlays of smaller, rapidly growing firms. Large corporations could avoid the tax by paying more dividends (reported as taxable income on individual income tax returns), but this was not a viable option for smaller growth companies with limited access to external funds and therefore dependent on retained earnings to reinvest for expansion.
Christina Romer (2009) blames the 1937–38 economic collapse partly on the new payroll tax for Social Security, which raised even more revenue that year ($1,473 million) than the briefly engorged personal income tax ($1,305 million), although not as much as newly increased excise taxes ($1,771 million). Cole and Ohanian (1999: 12) estimate that increases in the marginal tax on labor between 1929 and 1939 (notably the payroll tax) reduced the steady‐state labor supply by 4 percent.
Mulligan (2002) does acknowledge that excise taxes fall on the income received from supplying labor and capital and could therefore raise marginal tax rates on factor income.3 However, he mistakenly compares today’s minor excise taxes (0.4 percent of GDP) with those of 1932–40 (2.1 percent of GDP). He writes, “The federal government did not have a general sales tax, although it does have (and has had) excise taxes on goods such as cigarettes, gasoline, and imports. However, the revenues from these taxes are [today] too few, and not changing enough over time, to drive much a of wedge” (ibid.: 25). On the contrary, excise taxes that were added and increased in 1932 and 1936 were neither few in number, small in size, nor modestly changed (see Figure 3).
In 1932, the secretary of the Treasury reported that the Revenue Act of 1932 “effected one of the largest increases in taxes ever imposed by the federal government.” The report boasted of “manufacturers’ excise taxes on numerous articles” and “other miscellaneous taxes, including new and increased stamp taxes, increased taxes on admissions, and new taxes on telephone, telegraph, cable, and radio messages, checks, leases of safe deposit boxes, transportation of oil by pipe line, and the use of a boat” (U.S. Treasury 1932: 21).
Excise tax receipts in 1933 were $1.58 billion—up from $532 million in 1931 and more than four times the $390 million collected from high personal income taxes. From 1932 to 1939, individual income taxes averaged less than 1 percent of GDP, but excise taxes were 2.1 percent of GDP (OMB 2020: Table 2.3).