Friday, September 30, 2011

How Big is the Fiscal Multiplier?

Scott Sumner once compared arm wrestling with his daughter to the relationship between monetary and fiscal policy.  Scott explained that no matter how hard his daughter tried to win the arm-wrestling contest he would always apply just enough pressure to offset her efforts and keep her in check.  Likewise, no matter how hard fiscal policy may attempt to stimulate aggregate spending the Fed has the ability to offset such actions and place aggregate demand where it so chooses.  In other words, the size of the fiscal multiplier ultimately depends on the stance of monetary policy.

Recent studies by Eric Leeper, Nora Traum and Todd Walker, Lawrence Christiano, Martin Eichenbaum, and Sergio Rebelo, and Michael Woodford all lend support to this understanding.  They show using formal models that in a world where nominal and real rigidities exist the impact of fiscal policy on economic activity is muted when the central bank follows something like a Taylor Rule.1  Eric Leeper et al. explains why:
Active monetary policy reacts to a persistent fiscal expansion and the attendant increase in inflation by sharply raising the nominal policy interest rate. This raises the real interest rate, which reduces consumption and investment demand to attenuate the stimulative effects of the fiscal expansion. It is not too surprising that in this monetary-fiscal regime, very large fiscal multipliers are unlikely.
These studies also show, however, that the fiscal policy multiplier can be large if monetary policy is passive and does not raise the policy interest rate to check expansionary fiscal policy.  These studies say this is particularly true when monetary policy hits the zero bound since at that point the short-term interest is effectively pegged.  Here again is Leeper et al.:
Since the 2007–2009 recession, many central banks have shifted their emphasis from stabilizing inflation to stimulating demand by maintaining low and constant policy interest rates—near zero in some economies. When monetary policy makes the interest rate unresponsive to inflation, a “passive” stance, it amplifies fiscal policy’s impacts. By fixing the interest rate, monetary policy allows higher current and expected inflation to transmit into lower real interest rates. Instead of attenuating the demand stimulus of a fiscal expansion, monetary policy amplifies it: lower real rates encourage greater consumption and investment demand. Lower real rates induce a positive substitution effect from higher government spending, substantially raising output, consumption, and investment multipliers.
All the papers make this point that the fiscal multiplier is large and above 1 at the zero bound.  I, however, see several problems with this argument. First, just because the short-run nominal interest rate hits zero percent it does not follow that stimulus will be added by lowering the real interest rates via fiscally-induced higher inflation expectations.  What if, because of worsening economic conditions, the natural interest rate was falling faster than the real policy interest rate? There would be no stimulus.

Second, in the scenario above it is possible that the reason the economy is getting worse and causing the natural interest rate to fall in the first place is because the central bank is falling to sufficiently respond to adverse aggregate demand shocks. For example, if the zero bound is already binding and then a spate of bad news causes the public to become even more pessimistic about future economic growth the central bank would need to aggressively respond to these developments just to maintain the existing stance of monetary policy.  If the central bank failed to respond, this unchecked worsening of expectations would amount to a passive tightening of monetary policy.  In other words, even at the zero bound a central bank can passively tighten monetary policy and offset any attempt at fiscal expansion.   This point is particularly relevant for the Fed today.  Even though the federal funds rate is more or less at the zero bound, the Fed is still concerned about stabilizing inflation and unwilling by its own admission to use all of its monetary tools even though the U.S. economic outlook is worsening.  How possibly could a fiscal multiplier be above 1 in this environment?  Fiscal policy at this point looks to me like a fool's errand.

Third, even if fiscal expansion were to work as well at the zero bound as these studies claim, the expected and actual improvement in economic activity from the fiscal expansion should quickly lead to higher interest rates.  But if there are higher interest rates then this zero bound channel through which fiscal policy works disappears.  So there is a paradox here.
So where does that leaves us?  The aggregate demand slump problem is and has always been the Fed's.  It  has been shaping the path of nominal spending during this crisis and therefore has been determining the effectiveness of fiscal policy.  Yes, monetary policy to some extent becomes fiscal policy if Fed did something like a helicopter drop, but this is not the type of fiscal policy  being considered above and does not seem to be a politically viable option.  My takeaway from this is that our time is better spent encouraging the Fed to act more aggressively than pushing for more fiscal policy stimulus. 

1Leeper et al. show that these rigidities are essential for fiscal policy to have any expansionary effect at all in these models. See Robin Harding's discussion of this point.


  1. Extremely good blogging by David Beckworth.

    Yes, I would like to print money and dump it low-income neighborhoods, as they would spend it. I would run a national lottery with larger pay-outs than receipts, and both tickets and winnings in smallish numbers. But these are not politically viable.

    So, Beckworth is 100 percent right--we have to convince the Fed to target NGDP, and run a QE program, and perhaps cut IOR and force feed the economy until it revs up.

    The Fed needs to appear confident and aggressive in seeking higher NGDP numbers and employment. Inflation can take a beak seat for a few years.

    Please Mr Bernanke (and Dallas Fed Chief Fisher) read Beckworth.

    Boy, Beckworth is down in Texas in Rick Perry and Fisher. I wonder if Beckworth travels around incognito. Well, not yet, but if Perry wins....

  2. I suggest David Beckworth’s and Leeper’s arguments have weaknesses.

    Leeper claims that fiscal stimulus will be amplified if it results in higher inflation which in turn leads to lower real interest rates. My answer to that is that given excess unemployment, fiscal stimulus WONT result in much extra inflation. Put another way, if fiscal DOES result in significant extra inflation, the economy is at NAIRU, and stimulus (monetary or fiscal) should not be being implemented.

    As to David’s answers to Leeper’s point, David’s first two objections seem to me to be based on breaking the “other things being equal” implicit assumption in Leeper’s point. To illustrate, if I claim that increased US exports boost the value of the dollar, most people would take it as read that I’m assuming the Chinese don’t simultaneously sell big chunks of US debt (which would obviously counter the above dollar boosting effect).

    That is, David’s first two objections all rest on the phrase “What if, because of worsening economic conditions….”

    In his third criticism of Leeper, David says that if fiscal is successful, it leads to higher interest rates, which messes up the Leeper transmission mechanism mentioned above. My answer that is that the Leeper transmission mechanism is flawed for reasons I gave above. And secondly, the main transmission mechanism via which fiscal works strikes me as being that it increases private sector net financial assets (i.e. government borrows $X, gives the private sector $X of bonds and spends $X back into the private sector, which leaves the private sector $X up). That mechanism is still there even if the Leeper mechanism is useless.

    Finally, I object to EITHER monetary OR fiscal policies being used in isolation. Reason is that implementing either of them alone is ALWAYS distortionary. E.g. interest rate adjustments work only via entities that rely on variable rate loans. That makes as much sense as boosting an economy via entities whose names begin with the letters A-L while ignoring the “M-Zs”.