Alan Blinder and Mark Zandi’s Keynesian black box

In a recent Wall Street Journal column, Princeton economist Alan Blinder wonders why 64 percent of Americans do not believe the $849 billion “fiscal stimulus” bill “saved or created” many jobs.  “The main reason,” he explains, “appears to be that the White House’s January 2009 forecast was too optimistic—projecting, for example, an unemployment rate around 8% by the end of 2009 if the stimulus passed.”  He thinks that’s unfair.

Ironically, Blinder’s next column relied on the same failed forecasting model that lead the White House astray in January 2009.   Specifically, he asked us to take seriously the fact that:

Mark Zandi of Moody’s Analytics [economy.com] used his model to estimate that extending unemployment insurance benefits has almost five times as much “bang for the buck” as making the Bush tax cuts permanent.”

We should have learned from the White House’s reliance on Mr. Zandi’s forecasts in January 2009 that magical “multipliers” yanked from some forecaster’s black box are evidence of nothing.  They simply reveal dubious assumptions built into that forecasting model.   Yet the White House and Congressional Budget Office, as well as Professor Blinder, keep citing such models as if they were evidence the “stimulus” (spending) was effective.   On the contrary, recent academic studies of real world events have been unable to find a multiplier effects even half as large as Zandi’s model assumes. They find the addition to GDP is significantly smaller than the addition to the national debt – a bad bargain indeed.

In 2009 one of the new studies appeared in The Journal of Applied Econometrics by Andrew Mountford (University of London) and Harald Uhlig (University of Chicago).  They found “the best fiscal policy to stimulate the economy is a deficit-financed tax cut.”  Moreover, “fiscal expansion through government spending” will soon begin to “crowd out both residential and non-residential investment” resulting in a bigger government but a smaller private economy.

The economy.com model, by contrast, simply assumes-away any positive incentive effects of lower tax rates, or disincentives from higher tax rates, because “the level of resources and technology available for production is taken as given.”

If the economy.com model could predict anything, why was Zandi so upbeat about housing in January2008?   And why was he likewise so upbeat about the jobs alleged created by President Obama’s stimulus bill in January 2009?  In reality, no macroeconomic model has proven more accurate than judgmental guesswork even for short-term forecasting.   Attempting to use such models to predict the effects of higher tax rates or larger transfer payments has been considered disreputable since 1976, when University of Chicago Nobel Laureate Robert Lucas penned his famous “Lucas critique” of the whole idea.

In 1984, in OECD Economic Studies, James H. Chan-Lee and Hiromi Kato conducted a rigorous study of such economic models in 14 countries.  They concluded:

“Most [national economic] models remain broadly income/expenditure systems of a fundamentally Keynesian inspiration . . . However, few embody fully-specified stock or wealth effects in expenditure functions. And none seems to embody the latest theoretical thinking on expectations or supply-side effects. Considerations such as these may serve to limit the contribution that economic models can make to policy analysis.”