J. Bradford DeLong
October 15, 2016
The near-consensus policy rule for fiscal policy’s countercyclical role back in 2008 was simple: it had none. Automatic stabilizers were allowed to function, were even encouraged, and it seemed harmless to allow congress to make its overwhelming and very popular votes to extend the term of unemployment insurance in high unemployment states during recessions and in the early stages of recovery.
But Taylor (2000) serves as a powerful and eloquent marker of the crystallization of opinion that fiscal policy should be set according to “classical” principles: rightsizing the state, levying taxes efficiently, and achieving long run fiscal balance, with countercyclical fiscal policy was to be restricted to automatic stabilizers. Why? For three reasons:
Discretionary fiscal policy was unnecessary as a countercyclical policy tool, because monetary policy could do the job.
Discretionary fiscal policy was ineffective as a countercyclical policy tool, because decision and implementation lags were just too long.
Discretionary countercyclical policy was counterproductive as a countercyclical fiscal policy tool, because legislators and their staffs were not competent to choose appropriate policies even when they wished to do so.
Unnecessary, ineffective, and counterproductive—plus the excuse of needing to run deficits and have government buy things to boost employment provided legislators with an excuse to ignore the “classical” principles that they should be focusing on, especially in an environment in which rapidly rising medical care costs and the aging of America made achieving long run budget balance an extremely difficult political task. That legislators turned out not to be terribly good at providing the social insurance system with an efficiently-collected and adequate revenue stream to fund it over the long run provided no excuse for them to dabble in other policy areas, for only they could do anything to achieve that essential task. Hence as long as the countercyclical stabilization policy mission could be offloaded onto any other group—and it could be offloaded onto central banks—it should be.
The abandonment of discretionary fiscal policy did not lead to economists’ redirecting their energies towards thinking about fiscal automatic stabilizers. Indeed, only a few pieces like Blanchard (2000) and Fatas (2009) have kept automatic fiscal policy also from dropping to infinitesimal mindshare. I speculate that perhaps this was driven by intellectual imperatives within economics. Once the default macroeconomic model has a central bank using an interest rate tool to stabilize inflation driven by some Phillips Curve mechanism, save possibly at the zero lower bound the presumption is that the central bank will engage in what DeLong and Summers (2012) call “full monetary offset”.
In “full monetary offset”, the central bank will have a strong view of what point on the short run Phillips curve it should be aiming for. It will not allow the fiscal authorities to joggle its elbow: it will take steps to offset any impact of fiscal policy—discretionary or automatic—on spending that pushes spending away from the level the central bank considers appropriate.
Moreover, initially at least, the possibility of arriving at the zero lower bound did not lead to a complete reconsideration of the exclusive assignment of responsibility for countercyclical stabilization policy to monetary policy and the central bank. As Weinzerl and Mankiw (2011) demonstrate, even at the zero lower bound on short term safe nominal interest rates if a central bank can commit to the future money stock and thus to the future full-employment price level then there is no obvious theoretical need for fiscal policy as an additional stabilization policy tool for a central bank willing to use its policy tools. In any model in which there can be expected inflation that is neutral with respect to output, the central bank’s ability to set the rate of expected inflation is as good a stabilization policy tool as its ability to set the nominal interest rate.
Writing about policy at the zero lower bound, Friedman (1997) does not mention the possibility of fiscal policy. Bernanke (2002) speaks of how cooperation from the fiscal authorities “significantly enhances” the stimulative effects of monetary policies. But Krugman (1998) wonders whether fiscal policy would in fact be effective at the zero lower bound:
A recovery strategy based on fiscal expansion would have to continue the stimulus over an extended period. Which raises the quantitative question of how much stimulus is needed, for how long-and whether the consequences in terms of government debt are acceptable…
At the very least, it was not clear even within economics that fiscal policy would fully come back onto the countercyclical policy menu should interest rates hit their zero lower bound.
With the coming of the recession in 2008 the federal government’s tax and transfer automatic stabilizers followed their usual pattern, with perhaps congress voting, initially, slightly more generous expansions of unemployment insurance than had been typical (Stone and Chen (2014)). The green line in the figure below starts in the first quarter of 2008 when the two-year change in the unemployment rate turns positive, and shows in percentage points the two-year changes in the unemployment rate and the government purchases in GDP thereafter.
The initial stimulus in government purchases according to this framework starts out ahead of the curve: government purchases as the two-year change in the unemployment rate turns positive are already nearly one percentage point above e their value as of two years before. From then until the end of 2009 this measure of expansionary fiscal policy traces out the upper limit of the post-1954 scatter. By the end of 2009 the two-year changes in both the unemployment rate and the government purchases share of GDP are off the previous scale.
Thereafter things shift. As the two-year unemployment rate change first shrinks in 2010 and then turns negative in 2012, the two-year change in the government purchases share false and turns negative in 2011: government purchases in 2011:I are a smaller share of GDP than they were in 2009:I. The economy then heads for the lower left corner of the observed post-1954 scatter, where it sticks. From 2011:II until the present, the government purchases share of GDP has shrunk by an average of 0.6%-points/year for what is now more than five years. One-quarter of this shrinkage is federal. Three-quarters is state and local. And the cross-state shrinkage of the government purchases share is large, and follows the pattern you would expect: southern and prairie states cutting back on purchases; mid-Atlantic, northeast, and Pacific states not.
This is a big, and unprecedented deal. Previous excursions to the bottom left quadrant of this graph have been much more transitory, with the exception of the 1990s Bill Clinton-era of budget rebalancing. But monetary policy in the 1990s had the room to assist in the redeployment of productive resources to other categories of spending that monetary policy since 2008 has not.
In an environment of slack resources, stagnant wages, and exceptionally low interest rates, it does not fit any set of “classical” policy recommendations other than one that starts presuming that as of 2008 the marginal government purchase had very low value indeed. It might be rationalized by a belief that the shadow cost of accumulating government debt is extremely high because it sharply raises the risk of future financial crisis, or acts as a drag on growth as in Reinhart and Rogoff (2010). But such arguments are difficult to sustain with interest rates as low as they have been and still are with even small hysteresis effects (cf. Blanchard and Summers (1986), DeLong and Tyson (2013)).
Federal Reserve chairs have certainly not believed that this fiscal policy configuration has made their lives easier (cf. Bernanke (2015), Yellen (2016)). The canonical channel (Mankiw (2015)) through which a reduction in government purchases and in the government’s borrowing requirements avoids inflicting downward pressure on spending, production, and employment is if the reduction in purchases is accompanied by a sufficiently large fall in interest rates to induce offsetting increases in investment and consumer durables purchases, and to induce a fall in currency values from which will follow an increase in investment spending. But at the zero lower bound that channel cannot operate.