Some of the most interesting applications of econometric forecasting are in the political arena. Examples of regression analysis in politics range from part-time marketing consultants who help local candidates decide how best to use their advertising dollars to a fairly rich professional literature on U.S. presidential elections.13

In 2008, Haynes and Stone added to this literature with an article that specified (among others) the following equation:

Where:

VOTE_i = the Democratic share of the popular twoparty presidential vote

P_i = 1 if the incumbent is a Democrat and -1 if the incumbent is a Republican

DUR_i = the number of consecutive terms the incumbent party has held the presidency

DOW_i = the annual rate of change in the Dow Jones Industrial Average between January and October of the election year

GROWTH_i = the annual percent growth of real per capita GDP in the second and third quarters of the election year

INFLATION_i = the absolute value of the annualized inflation rate in the two-year period prior to the election

ARMY_i = the annualized percent change of the proportion of the population in the armed forces in the two-year period prior to the election

SPEND_i = the annualized percentage change in the proportion of government spending devoted to national security in the twoyear period prior to the election

a. What kind of variable is P? Is it a dummy variable? If not, what is it?

b. The authors specified their equation as a series of interaction variables between P and the other variables of interest. Look at the equation carefully. Why do you think that these interaction variables were required?

c. Using the data15 in Table 15.1 (datafile = ELECTION15) estimate Equation 15.20 for the years 1916–1996.

d. Create and test appropriate hypotheses on the coefficients of your estimated equation at the 5-percent level. Do any of the coefficients have unexpected signs? Which ones?

e. Create unconditional forecasts for the years 2000 and 2004 and compare your forecasts with the actual figures in Table 15.1. How did you do?

f. The authors wrote their article before the 2008 election. Create an unconditional forecast for that election using the data in Table 15.1. Who did the model predict would win? How did the model do?

Table 15.1 Data for the Presidential Election Exercise