Friday , April 10 2020
Home / Econbrowser - James Hamilton / How Plausible Is the Administration’s GDP Forecast?

# How Plausible Is the Administration’s GDP Forecast?

Summary:
As has been noted, the Administration’s forecast is about a percentage point higher than CBO’s. This seems like a large economic difference; as I’m teaching econometrics this semester, how does this seem in terms of statistical significance. Figure 1 below summarizes. Figure 1: GDP in bn Ch.2012\$ SAAR as reported (black), Administration forecast (red square), CBO January 2020 projection (blue), ARIMA(1,1,0) on log GDP 2009Q2-2019Q4 (teal), and 67% prediction interval (gray lines). Source: BEA, 2019Q4 advance release, CBO, Budget and Economic Outlook, January 2020, OMB, Budget for FY2021, February 2020, Table S-9, and author’s calculations. Notice that an ARIMA(1,1,0) on post-recession data (which includes the relative rapid recovery in the first year of expansion) still predicts

Topics:
Menzie Chinn considers the following as important:

This could be interesting, too:

Menzie Chinn writes Where Did All the Stimulus Go?

Jeffrey Frankel writes It’s Finally Time for German Fiscal Stimulus

Eric Crampton writes Around the budget traps

As has been noted, the Administration’s forecast is about a percentage point higher than CBO’s. This seems like a large economic difference; as I’m teaching econometrics this semester, how does this seem in terms of statistical significance. Figure 1 below summarizes.

Figure 1: GDP in bn Ch.2012\$ SAAR as reported (black), Administration forecast (red square), CBO January 2020 projection (blue), ARIMA(1,1,0) on log GDP 2009Q2-2019Q4 (teal), and 67% prediction interval (gray lines). Source: BEA, 2019Q4 advance release, CBO, Budget and Economic Outlook, January 2020, OMB, Budget for FY2021, February 2020, Table S-9, and author’s calculations.

Notice that an ARIMA(1,1,0) on post-recession data (which includes the relative rapid recovery in the first year of expansion) still predicts noticeably slower growth than does the Administration. A plus/minus one standard error band suggests that the difference is also statistically significantly different at the 33% msl.

Table S-9 shows the economic assumptions in the budget.

It’s important to recall that we are comparing different types of forecasts here. The CBO’s projection is conditional on current law. The Administration’s forecast is conditional on Mr. Trump’s policies being implemented, with the Administration’s assumptions regarding effects (like how they assessed the impact of the Tax Cut and Jobs Act), and extension of tax cuts. In other words, the two forecasts condition on different things. Finally, the time series model is unconditional in the sense that all that is assumed is that the future is like the past.

Two things that might be particularly relevant, then, is that the Administration forecast implies extension in tax cuts, and interest rates that are in line with CBO’s estimates, even though the administration predicts growth roughly a percentage point higher than CBO. In other words, the Taylor rule has (seemingly) been assumed away. More from the Committee for a Responsible Federal Budget.

It might be of interest to see how the Administration forecasts fared in the FY2019 Mid Session Review.

Figure 2: Q4/Q4 real GDP growth (blue bar), Trump-Pence campaign promise (red dashed line), FY2019 MidSession Review (orange line), FY2021 Budget (green line). Source: BEA, OMB, author’s calculations.

The Administration is not revising downward its out-years forecasts, despite the big misses in 2018 and 2019.

Discussion of the historical record of CBO vs. Administration vs. Fed forecasting in the CBO analysis from December of last year.

He is Professor of Public Affairs and Economics at the University of Wisconsin, Madison