The relationship of US federal marginal tax rates to US per capita real GDP, 1959-2003: a note.

Kamerschen, David R.

It is often argued--especially in a US presidential election year--how much the highest marginal tax rate (HMTR) on personal income influences a nation's real or inflation-adjusted output, i.e., an economy's real gross domestic product (GDP). Most economists would agree, ceteris paribus, that the long-run impact of cutting personal taxes is that people have a greater incentive to work which, in turn, produces more output and less inflation. But in the short run, a personal tax cut could, ceteris paribus, increase spending, overheat the economy, and cause inflationary pressures that the country's central bank, such as the US Federal Reserve, would have to prevent. (1) One reason that a cut in the marginal tax rate or HMTR is not necessarily a stimulus to the economy is that this HMTR applies to only a small number of taxpayers and is not necessarily a reduction in the average tax rate for all taxpayers. Another reason that there is some doubt about the impact of a tax cut is that the shift in aggregate demand (AD) can be greater or smaller than the fiscal change. The multiplier effect amplifies the effects on AD, and the crowdingout effect diminishes the effects on AD. (2)

President John F. Kennedy cut taxes significantly from an HMTR of 91% in 1963 to 77% in 1964 to stimulate the US economy. Similarly, Ronald Reagan emphasized supply-side economics in his election bid in 1980. The principal method that supply siders advocated to increase output and decrease inflation was to cut taxes. Table 1 shows how dramatically President Reagan did this, as the HMTR went from 70% in 1980, when he went into office, to 28%, when he left in 1988. (33) Yet this historical Reagan example represents a case when in fact the significant cut in the marginal tax rate did not translate into a reduction in the average tax rate, as the fraction of income subject to tax was increased for many taxpayers (see Bruce, 1998, p. 682).

In this paper, the real GDP is calculated on a per capita basis--i.e., real GDP is divided by the population. The reason for this metric is that an increased population is likely, ceteris paribus, to increase real GDP just because the economy has more workers to contribute to the real GDP. Thus, population must be controlled for in assessing the pure impact of HMTR on real GDP. From 1913 to 2002 the US marginal tax rate ranged from 7% in 1913-1915 to 94.0% in 1944-1945 (and 92.0% in 1952-1953). Since the HMTR data are rather difficult to locate and may be useful to future researchers, I show the history of HMTR from 1913 to 2006 in Table 1. In addition, I show the per capita real GDP in the US from 1959 to 2003 in Table 2.

I test the relationship between HMTR and per capita real GDP in the US and find, using OLS, that for the 45 years from 1959 to 2003 the adjusted coefficient of determination ([R.sup.2]) = .80 between these two variables. (4) This means that 80 percent of the variance in per capita real GDP is statistically explained by the HMTR. (5) Put differently, the t-value for the regression coefficient on HTMR on per capita real GDP is 13.29 with N = 45. (6) Clearly, the HMTR has a statistically significant and negative linear influence on per capita real GDP in the US at the 1% or better level. (7) While there are obviously numerous other factors at work, the convential wisdom that HMTR has a statistically significant and profound influence on per capita real GDP appears undeniable.

REFERENCES

Economic Report of the President (2004), US Government Printing Office, Washington DC.

Bruce, N. (1998), Public Finance and the American Economy, Reading, MA: Addison-Wesley.

Hall, R.E. and M. Lieberman (2005), Economics: Principles and Applications, Third Edition, United States: Thomson Learning South-Western.

Opalka, E. (2003), The history of federal individual income tax rates, How to Protect and Manage Your 401K. The Career Press, Franklin, NJ, Figure 3-3, 43-44. Opalka uses tax data from the Congressional Joint Committee on Taxation and the Internal Revenue Code.

Rym, E. and F. Koray (2004), Average Marginal Tax Rates in the UK Economy, Applied Economics, Vol. 36, No. 21, pp. 2369-2372.

NOTES

(1.) See, e.g., Hall and Lieberman (2005), pp. 692-693.

(2.) The multiplier effect is the additional positive shift in AD when an expansionary tax cut increases income and therefore increases consumer spending. The crowding-out effect is the additional negative effect on AD that occurs when an expansionary tax cut increases the interest rate and thereby decreases investment spending.

(3.) Yet this historical Reagan example represents a case when in fact the significant cut in the marginal tax rate did not translate into a reduction in the average tax rate, as the fraction of income subject to tax was increased for many taxpayers (see Bruce, 1998, p. 682).

(4.) The HMTR data are from Opalka (2003). They show the highest individual tax rates in Figure 3-3, "The History of Federal Income Tax Rates," pp. 43-44. The real GDP data are from the Economic Report of the President (2004), Table B-2, "Real Gross Domestic Product, 1959-2003." The population data are from the Economic Report of the President (2004), Table B-34, "Population by Age Group, 1929-2003."

(5.) Of course, the usual pitfalls and limitations associated with correlation analysis holds (see e.g., any statistics or econometrics textbook).

(6.) The t-statistic shows the significance of our explanatory variable, HMTR, in predicting the dependent variable, per capita real GDP. The results are virtually identical if population is not controlled for in the regression. The adjusted [R.sup.2] and t-value between HMTR and real GDP (not per capita GDP) are .76 and -11.7, respectively.

(7.) The issue of HMTR is of current interest in many countries, see e.g., Rym and Koray (2004) on HMTR in the UK.

David R. Kamerschen *

David R. Kamerschen, Distinguished Professor of Economics, Jasper N. Dorsey Chairholder, The University of Georgia, The Terry College of Business, Economics Department, Brooks Hall, Athens, Georgia 30602-6254, USA, E-mail: davidk@terry.uga.edu.

Table 1
The US History of the Highest Marginal Tax Rates on
Federal Personal Income, 1915-2006

Year Highest Year Highest Year
 marginal marginal marginal
 tax rate * tax rate * tax rate *

1913 7% 1937 79% 1961
1914 7% 1938 79% 1962
1915 7% 1939 79% 1963
1916 15% 1940 81.1% 1964
1917 67% 1941 81% 1965
1918 77% 1942 88% 1966
1919 73% 1943 88% 1967
1920 73% 1944 94% 1968
1921 73% 1945 94% 1969
1922 56% 1946 86.45% 1970
1923 56% 1947 86.45% 1971
1924 46% 1948 82.13% 1972
1925 25% 1949 82.13% 1973
1926 25% 1950 91% 1974
1927 25% 1951 91% 1975
1928 25% 1952 92% 1976
1929 24% 1953 92% 1977
1930 25% 1954 91% 1978
1931 25% 1955 91% 1979
1932 63% 1956 91% 1980
1933 63% 1957 91% 1981
1934 63% 1958 91% 1982
1935 63% 1959 91% 1983
1936 79% 1960 91% 1984

Year Highest Year Highest
 marginal tax
 rate *

1913 91% 1985 50%
1914 91% 1986 50%
1915 91% 1987 38.5%
1916 77% 1988 28%
1917 70% 1989 28%
1918 70% 1990 28%
1919 70% 1991 31%
1920 75.25% 1992 31%
1921 77% 1993 39.6%
1922 71.75% 1994 39.6%
1923 70% 1995 39.6%
1924 70% 1996 39.6%
1925 T0% 1997 39.6%
1926 70% 1998 39.6%
1927 70% 1999 39.6%
1928 70% 2000 39.6%
1929 70% 2001 38.6%
1930 70% 2002 38.6%
1931 70% 2003 38.6%
1932 70% 2004 37.6%
1933 69.13% 2005 37.6%
1934 50% 2006 35%
1935 50%
1936 50%

Source: Congressional Joint Committee on Taxation and the Internal
Revenue Code as cited in Opalka (2003).

Table 2
US Real Gross Domestic Product (GDP), Population, and
Per Capita Real GDP, 1959-2008

(1) (2) (3) (4)
Year Real GDP Population Per Capita Real
 (Billions of chained (Thousands GDP = (2)/(3)
 2000 dollars) of Persons)

1959 $2,441.3 177,830 $0.013728
1060 2,501.8 180,671 0.013847
1961 2,560.0 183,691 0.013936
1962 2,715.2 186,538 0.014556
1963 2,834.0 189,242 0.014976
1964 2,998.6 191,889 0.015627
1965 3,191.1 194,303 0.016423
1966 3,399.1 196,560 0.017293
1967 3,484.6 198,712 0.017536
1968 3,652.7 200,706 0.018199
1969 3,765.4 202,677 0.018578
1970 3,771.9 205,052 0.018395
1971 3,898.6 207,661 0.018774
1972 4,105.0 209,896 0.019557
1973 4,341.5 211,909 0.020488
1974 4,319.6 213,854 0.020199
1975 4,311.2 215,973 0.019962
1976 4,540.9 218,035 0.020826
1977 4,750.5 220,239 0.021570
1978 5,015.0 222,585 0.022531
1979 5,173.4 225,055 0.022987
1980 5,161.7 227,726 0.022666
1981 5,291.7 229,966 0.023011
1982 5,189.3 232,188 0.022350
1983 5,423.8 234,307 0.023148
1984 5,813.6 236,348 0.024598
1985 6,053.7 237,466 0.025493
1986 6,263.6 240,651 0.026028
1987 6,475.1 242,804 0.026668
1988 6,742.7 245,021 0.027519
1989 6,981.4 247,342 0.028226
1990 7,112.5 250,132 0.028435
1991 7,100.5 253,493 0.028011
1992 7,336.6 256,894 0.028559
1993 7,532.7 260,255 0.028944
1994 7,835.5 263,436 0.029743
1995 8,031.7 266,557 0.030131
1996 8,328.9 269,667 0.030886
1997 8,703.5 272,912 0.031891
1998 9,066.9 276,115 0.032837
1999 9,470.3 279,295 0.033908
2000 9,817.0 282,434 0.034759
2001 9,866.6 285,545 0.034554
2002 10,083.0 288,600 0.034938
2003 10,493.1 * 291,049 0.036053

* The data for real GDP for 2003 are for the third quarter of 2003.
The full year data are not available at this time. Source: Economic
Report of the President (2004), Table B-2, "Real Gross Domestic
Product, 1959-2003," and Table B-34, "Population by Age Group,
1929-2003."