Remarks on quantitative easing

The following post is a bit long and can be downloaded in pdf here.
The issues discussed in this text are, quite naturally, the subject of a lively online debate among US economists. See here for an entry point into the discussion.

The so-called “Great Recession” has led, almost overnight, to a new conventional wisdom in macroeconomics. This new conventional wisdom roughly amounts to the view that monetary and fiscal stimulus are costless and should be used without restraint. One justification is that interest rates are stuck at a zero lower bound and the economy runs the risk of a deflationary spiral: As inflation becomes negative, real interest rates go up and the central bank is unable to reduce them because its policy rate cannot go below zero. What should be done to lift the economy off this deflationary spiral, the conventional wisdom goes, is to raise inflationary expectations. This will prompt people to get rid of the money they are hoarding by spending more. How do you raise inflationary expectations, since this is not a policy variable? The answer is that it is not easy. But the dominant view, it seems, is that “Quantitative Easing” — increasing the money stock by massive purchases of government securities — is likely to raise inflationary expectations and get us out of the dreaded liquidity trap. The argument is fuzzy because it is rational to believe that once one gets out of the liquidity trap, the central bank will resume its usual monetary policy, and will not let inflation go beyond its target. The central bank’s supposed virtue and consistency is running against its attempt to raise inflation expectations. For it to work, we would need that people irrationally believe that the central bank has shifted to (for example) a constant money growth policy starting from a new inflated money stock. Conversely, appointing an inflation-prone central banker (as Russia has in the 1990s) should be enough to raise inflationary expectations without having to resort to QE.
Besides the fact that the government simply does not control them, the idea that one should raise inflationary expectations, even in a liquidity trap, is highly dubious.
For one thing, inflationary expectations is the worst way to raise aggregate demand, because at the same time it reduces aggregate supply. The effect may be weak if in a recession the aggregate supply curve is locally very flat, as would be the case if, for example, there is also a zero lower bound on nominal wage increases; that is, there could be so much slack in the economy that a rise in inflationary expectations would not be reflected in increased wage pressure. Even so, it is always a better idea to move along the aggregate supply curve than to shift it in the wrong direction. Thus using fiscal policy looks like a better idea than QE (my preferred option being public infrastructures, since they do not commit future expenditures as compared to, say, hiring government workers). Indeed, mainstream economists believe that the fiscal policy multiplier is quite large at a zero lower bound (a view that was present in macro 101 textbooks already in 1970). So why insist on pushing monetary stimulus further when it has reached its “physical” limit?
Furthermore, QE may even fail to raise aggregate demand, because it is likely to raise future uncertainty about inflation and therefore activity. If consumers make precautionary savings, then they will react to that uncertainty by reducing current consumption, and the stimulus policy will in fact contract aggregate demand. This criticism applies, of course, to any stimulus strategy which raises uncertainty about the future; for this reason, transparency, predictability, and credibility are a crucial component of any successful stabilization policy.
The purported motivation for QE (the risk of a deflationary trap) is also dubious and vastly exaggerated . In the monthly inflation series shown below, inflation has been negative only three times. It is around 2 % since April 2011 and the most recent data indicate an acceleration.

fed funds

Despite these developments, the policy rate remains stuck at zero.

fed funds

As a result, short-term real rates remain largely negative. This is not unusual, as this was also the case in the preceding recession, although they are even lower since then the Fed did not indulge in zero rates. The following graph shows how volatile real interest rates are due to the Fed’s active stabilization policy.

Real short rate

From those data I was tempted to conclude that the monetary policy regime has changed and that the Fed has shifted from an inflation targeting rule to a debt monetization rule. But it is too early to draw such a conclusion. The reality is that the Fed is willing to engineer enormous swings in real interest rates in order to fulfill its targets — which of course makes it more likely to hit the zero lower bound, compared to the ECB which is less reactive. Despite the steady fall in unemployment, the Fed has maintained its zero-rate policy. But the unemployment rate is still relatively high (6.3 %); this is about its peak level during the preceding recession, during which real rates were reduced to similar negative levels, and the zero lower bound was not hit because inflation was a bit higher than now.

civilian unemployment

An interesting way to look at the Fed’s monetary rule is to correlate the short-term real rate with the unemployment rate. We see a steep negative relationship, which confirms the very strong reaction of the Fed; we may ask whether it is a good idea to have so wide fluctuations in real rates, an important allocative price. We also note that the relationship shifts down at some point; the shift takes place somewhere around mid-1997, after which a lower real rate is tolerated. In my view, this captures a more expansionary monetary policy rule (there was a view at the time that potential output growth has improved, but inflation crawls up after this shift until the collapse of the Internet bubble).
The circled zone represents the crisis period during which monetary policy hits the zero lower bound and interest rates cannot be reduced further; the Fed is then compelled to deviate from its real rate/unemployment schedule. The graph makes clear that we have exited this zone; the last available point is not far from the usual rule. Hence my conclusion that we can’t conclude that the rule has changed, although we may ask whether it is not too expansionary (is -2 % at 6 % unemployment sending the right signals to the economy?). Another reason to believe so is the sharp acceleration of inflation in the run-up to the crisis, which suggests that 4.5 % is below the natural rate of unemployment, and therefore that the U.S. economy may not be that far from it.

unemployment vs real rate

Some people believe that the unemployment rate is a poor indicator of slack because participation has fallen during the crisis. According to this argument, there are discouraged workers out there who do not bother to register as unemployed but would be happy to hold a job should they find one. Therefore, these people say, the Fed is justified in pursuing the zero interest rate policy despite that the unemployment rate is only one point above the natural rate.
Just because something evolves during a crisis does not make it cyclical. The distinction between trend and cycle may be misleading. A crisis is a situation where actions are highly correlated across agents, and these actions may be permanent and indicate structural change. For example, the Spanish construction sector collapsed during the crisis, but will probably never recover to the same levels. The structural change in the sectoral composition of Spanish economic activity is taking place during, and being facilitated by, the crisis. This is the (painful) way the economy “learns” its sectoral allocation of activity is inadequate.
Looking at participation rates separately for men and women confirms that the evolution of participation is mostly structural. The female participation rate is flat and has been so (at 75 %) since 1996. There is no surge in female discouraged workers during the crisis. The male participation rate is trending downwards from the mid-sixties. The trend was indeed slowing down before the crisis, and has sharpened since the onset of the crisis. Interestingly, the process continues despite the steady recovery of the American economy.

female participation

male participation

The downward trend in male participation is due to the decomposition of family structures. Statistically, half of households will be dissolved by divorce and custody will be awarded to the mother in the vast majority of cases. The return to labor market participation for men is much lower than in the past; the need for their earnings is reduced both due to marital instability and rising female participation. On the other hand, divorce is associated with a substantial tax on those earnings, with little counterparts. The contrary holds for women who can expect to end up as a sole provider in a single-headed household with substantial probability. For those reasons we may even think that female participation will end up being higher than male participation. At present this is not the case, though.
Hence, the case for a purely cyclical downturn in participation rates seems rather weak to me.
How effective is the Fed at raising inflationary expectations (remember it is not a good idea)? The following graph shows some consumer survey data. The median inflationary expectation is remarkably stable around 3 %. Despite a little blip in 2011, agents return to that level; if the goal of quantitative easing is to raise those expectations above their normal secular value, this goal has not been achieved. On the other hand, these numbers have been above realized inflation since the beginning of the crisis, so perhaps QE has indeed raised inflation expectations a little bit (and we don’t really know why).
Finally we may believe that the zero lower bound rhetoric is just an ideological justification for monetary financing of the deficit. I am quite tempted to believe that, but if the economy is in a liquidity trap, then money and bonds are perfect substitutes. Therefore, in principle the public is willing to hold government bonds at the going, zero, interest rates, and it is not clear why the central bank should purchase them instead. While one may dispute the merits of monetary policy, at the end of the day what matter is the equilibrium interest rate, not the way in which money is injected into the economy. If, say, the zero lower bound were reached by massive injections of liquidity into banks instead of bond purchases, the government would not find it difficult to float its debt on the market at the going rate.
Things are different, of course, in the case of purchases of troubled member states debt by the ECB. While part of these purchases act by killing the expectation that the country might leave the Euro, the other reason why they act is by indirectly mutualizing the debt between member states. The seignoriage revenues paid by a member state to the ECB are more or less proportional to the share of the monetary base that is present in that state. But if the proceeds are used to purchase Portuguese debt, those revenues accrue disproportionately to the Portuguese state (and this indirect transfer, and the associated implicit guarantee, naturally reduces the cost of borrowing for the Portuguese government).

Inflation expectations

Advertisements