Explaining the Three Scenarios in "Primary Numbers"

Last week, we released our in-depth analysis of the debt impact of the four Republican candidates’ budget proposals, a report that received a significant amount of media attention and brought some attention to the fiscal side of the campaign. The report has been covered in most major publications and on other media across the Internet. While the headlines have been highlighting candidates' deficit impacts in the intermediate-debt scenario, we also scored low-debt and high-debt scenarios based on different assumptions about the candidates' policies. By using these three estimates, similar to a confidence interval, we can provide a range of the likely budget impact based on varying assumptions.

One of the difficulties of scoring a candidate’s budget plan is that some proposals are highly specific, while others contain less detail. We account for the variation in specificity using the three estimates. A specific policy to reduce the deficit is included in all three scenarios. However, when a candidate proposes to reduce spending in an area more generally, such as reducing discretionary spending by a certain amount, it is included in the low- and intermediate-debt scenarios, but not in the high debt scenario. Finally, a top line reduction, such as reducing overall spending by a certain amount, is credited only in the low debt scenario.


Often, the campaign has different numbers from official source estimates of the impact of a candidate’s policies. In these cases, we use the campaign estimates for our low debt estimate, and the official estimate for the other two. For example, a few candidates who proposed repealing the Davis-Bacon Act gave much higher savings estimates than CBO had scored. In these cases, we gave the candidates credit for their numbers in the low-debt scenario, but used the CBO score in the other scenarios.

While a candidate’s policy may be detailed, the time horizon for the policy to be implemented may be unspecified. Consequently we assume a rapid phase-in for our low debt estimate, and a slow phase-in for our high debt estimate.

There are also policies for which we had to produce our own estimates that relied on varying assumptions about its parameters. For example, with Rep. Ron Paul's (R-TX) tax subsidies for medical expenses, we calculated the cost of the subsidy based on out-of-pocket health spending and assumed that 50 percent of the tax subsidy would be utilized in our low- and intermediate-debt scenarios. In the high-debt scenario, we assumed that the entire available tax subsidy would be utilized.

Projections often use a confidence interval to improve precision, and our three different scenarios provide a range of the likely effect of a candidate’s current plan on the budget. As predictions are only as valid as the assumptions that they make, the three curves ensure that the estimates produce a fair and objective analysis to the best of our ability.