The incredible shrinking president and the economy

Abstract

This paper employs regression analysis to investigate alternative explanations of presidential approval ratings. Since perhaps the greatest determinant, actual performance, is not readily measurable, the analysis is restricted to the first and fourth month of each term, a period sufficiently short that performance might not be expected to be clearly established. Strong empirical evidence is found to support: (1) the importance of economic conditions, especially those projected for the upcoming quarter; (2) a handicap in approval ratings today of about 11 points when compared to 1948; (3) an advantage (of about 5 to 13 points) enjoyed by Republican Presidents; and (4) the existence of a honeymoon (of about 6 points) in the first month of a presidential term that dissipates by the fourth month. Perhaps most important, strong evidence is found to discount the value of one and four month approval ratings as an accurate reflection of public evaluation of a President’s abilities or performance. Instead, those ratings are almost entirely explained by factors outside the control of the individual Presidents.

Notes

1 “That Sinking Feeling,” Time, June 7, 1993, pp. 23–29.

2 Several other potential explanatory variables were investigated but were not found to contribute to the explanatory power of the regression equations. These include the popular vote (%) won by the President, the popular vote (%) of any third party candidate, and dummy variables for the Vietnam War, the Watergate scandal, and the Reagan assassination attempt.

3 To preserve sufficient degrees of freedom, variables with coefficients not significantly different from zero at a 10% level (using a one-tail test) were omitted from each estimated equation.

4 From the regression results it can be concluded that while the four month approval rating for any given President would be 3 points lower for a Republican in 1992 compared to 1948, it would be approximately 11 points lower for a Democrat.

5 One previously cited explanation for the lack of significance of the coefficient of YEAR in R-4 was that there may be an 11 point honeymoon in the first month in 1948 (relative to 1992) that dissipated in the first few months. Combining this conjecture with the implication arising from DMONTH, it can be concluded that there may have been a three month honeymoon enjoyed by President Truman in the neighborhood of 17 points, compared to that of President Clinton at about 6 points.

6 For example, the error in the predicted one month rating for R-1 is greater than 5 points in only one of twelve years while that for R-4 in four of 12 years. The error never exceeds 8 points.Council of Economic Advisors 1994, Miller 1980, Nelson 1989, Newport 1989, Newport 1993, United States Bureau of the Census 1992 United States Department of Commerce,Economics and Statistic Administration, Bureau of Economic Analysis 1999^????