The Base Run
Ask to solve the October 30, 2011, version of the US model. Work with the US model. Pick a name like NAMEBASE, where "NAMEBASE" is unique to you. Ask to copy XAXBASE to NAMEBASE. Ask to change the exogenous variables, and change D4G to be .06212 in 2012:1 rather than .07159 as in XAXBASE. (XAXBASE assumed that there was no payroll tax cut extension, and using .06212 in place of .07159 incorporates the two-month extension.) Also, add 6.2 to TRGHQ for 2012:1. This incorporates the unemployment benefits extension. TRGHQ is in real terms, and 6.2 is roughly the real term equivalent of the 7.4 nominal value mentioned above ($3.7 billion per month). Make no other changes and solve the model. You can then exit---NAMEBASE has been created.

The New Run
Ask to solve the October 30, 2011, version of the US model. Pick a name like NAMENEW, where "NAMENEW" is unique to you. Ask to copy NAMEBASE to NAMENEW. This means you are running off of your base, not XAXBASE. Ask to change the exogenous variables, and change D4G to be .05739 in each of the four quarters of 2012. This is assuming that the extension is for all of 2012. Also, add 3.1 to TRGHQ for 2012:1 and add 9.3 to TRGHQ for each of the remaining three quarters of 2012. This is assuming that unemployment benefits are extended through 2012. Make no other changes and solve the model. Examine the output using NAMEBASE as the base dataset.

Comparing New to Base
You can use the results in NAMEBASE to compute G, P, and Z, the three economic variables that are needed in the vote equation. GDPR is real GDP at a quarterly rate, and PCGDPR is the percentage change in real GDP for the particular quarter at an annual rate. GDPD is the GDP deflator, and PCGDPD is the percentage change in the GDP deflator for the particular quarter at an annual rate. If you compute the percentage change in GDPD between 2008:4 and 2012:3 at an annual rate, you will get 1.89. This is P. If you compute the percentage change in GDPR between 2011:4 and 2012:3 at an annual rate, you get 4.02. This is G except you need to subtract population growth to get per capita growth. If you take population growth to be 1.0 percent, then G is 3.02. Regarding PCGDPR for the 15 quarters from 2009:1 through 2012:3, there is only one quarter greater than 4.2 percent, which is 2012:3 at 4.27. If you take population growth for this quarter to be 1.0, this is a strong growth quarter, and so Z is 1. Using P = 1.89, G = 3.02, and Z = 1 in the vote equation, the predicted value of VP, Obama's share of the two-party vote, is 50.17. Compute your own predictions for 2012

You can also use the results in NAMENEW to compute G, P, and Z. If you do this, you will get 1.89 for P, the same as for NAMEBASE (subject to rounding). The percentage change in GDPR between 2011:4 and 2012:3 is 4.94. If you take population growth to be 1.0, then G is 3.94. There are three quarters in which PCGDPR is greater than 4.2 percent, which are 2012:1 at 4.24, 2012:2 at 5.02, and 2012:3 at 5.57. The latter two quarters are clearly strong growth quarters, so Z is at least 2. The question is what to do about 2012:1. PCGDPR increased from 3.90 in NAMEBASE to 4.24 in NAMENEW. If one takes population growth for this quarter to be 1.0, then this is a strong growth quarter since the cutoff is 3.2 percent. On the other hand, the 0,1 nature of Z makes this somewhat problematic. Z has a coefficient of 0.99 in the vote equation, and does one really believe that increasing PCGDPR by 0.34 percentage points increases Obama's vote share by 0.99 percentage points? If so, then Z is 3; otherwise it is 2. Using P = 1.89, G = 3.94, and Z = 2 in the vote equation, the predicted value of VP is 51.78. If Z is 3, then the predicted value is 52.77. The present estimate is thus that extending the payroll tax cut and unemplyment benefits for all of 2012 adds between 1.61 and 2.60 percentage points to Obama's vote share. The more trustworthy estimate is probably 1.61.

Levels versus Differences
The forecast of real GDP from the US model for 2012 is more optimistic than the consensus forecast at the time of this writing. Fortunately, the above analysis is to some extent independent of the level of the GDP forecasts and thus independent of the US model's optimism. What matters is the difference between G in the base run and G in the new run, not the level of either. If, say, the extension through 2012 increases G by 0.92 percentage points, as estimated by the US model, this will add 0.62 percentage points to Obama's vote share (the coefficient of G in the vote equation is 0.672). This calculation does not depend on the base forecast of G.

The change in Z, on the other hand, may depend on the base forecast. If, for example, a base forecast has no quarters close to a strong growth quarter, then the new forecast may not either. In this case the change in Z would be zero, and the change in the vote share would be just the 0.62 percentage points. If the change in Z is 1, we have 1.61, and if the change in Z is 2, we have 2.60. Even 0.62 percentage points could, of course, be important in a close election, which the vote equation suggests is likely.