Friday, January 16, 2009

The Failure of Deflation Orthodoxy

As readers of this blog know, I am someone who believes that the loose U.S. monetary policy of 2003-2005 was an important contributor--though not the only one--to the buildup of economic imbalances that are the source of the current economic crisis. I have also argued that a key a reason for the highly accommodative monetary policy was that the Fed, following the conventional wisdom on deflation, viewed the deflationary pressures at the time as a sign of weakened aggregate demand and acted to offset it. Though well intended, the Fed's response was highly distortionary since the deflationary pressures turned out to be driven by rapid productivity gains rather than weakened aggregate demand. The Fed, therefore, was adding significant stimulus to the economy at the same time it was being buffeted by rapid productivity gains, a surefire way to push the U.S. economy past its speed limit. What all this means is that had the Fed been better able to distinguish between malign and benign deflationary pressures some, maybe much, of the economic imbalance buildup could have been avoided. This is an important lesson from this economic crisis. It is also one that I more fully discuss in a recently published article that can be found here.

Update: In addition to my article cited above, other recent papers making a similar case include William White's article "Is Price Stability Enough?" and this interesting article by The Economist magazine.


  1. David: I have not read your paper as carefully as I should (does one ever?) but I will give you some comments anyway, and hope you will forgive me if my comments are off-base.

    Figure 3:
    (1) But why did they move the AD curve more than the AS curve? If they moved the AD curve the same amount, the subsequent short-lived boom and bust would not have happened.
    (2) If they moved the AD curve more than the AS curve because they were targeting (say) 2% inflation, then expected inflation would mean the SRAS curve would be also moving upwards, relative to the LRAS curve (i.e. so that the intersection of SRAS and LRAS curves would be moving vertically at the expected rate of inflation of 2%), and so the AD/SRAS intersection would stay on the LRAS curve.

    More generally, I don't like the expression "good deflation", because it is the growing SRAS which is good, and that would be good regardless of whether the central bank let it translate into deflation or inflation.

    Start with long-run super-neutrality of money (the growth rate of the money supply has no real effects). The rate of inflation or deflation in the long-run does not matter. Unexpected deflation (i.e. where the actual rate of inflation is less than the actual rate of inflation) will be a bad thing, but it will be a bad thing regardless of the long-run rate of inflation.
    Long-run growth (LRAS moving right) is also a good thing, regardless of the rate of inflation/deflation chosen by the central bank.

    Now, there are two things wrong with long-run superneutrality: the zero lower bound on nominal interest rates, and the (possible) downward rigidity of nominal wages/input prices. These are arguments against targeting long run deflation (but as you say, and what I learned from reading your paper, these arguments will matter less when productivity is growing, because those constraints are less likely ever to bind).

    (There are other reasons why super-neutrality does not hold exactly of course, like shoe-leather, but these are the only ones which seem relevant to your paper.)

  2. Fair questions Nick. Below are my answers.

    Regarding Figure 3:
    (1) The amount the AD curve is moved to the right is based on the need to maintain price level stability, whether that means stabilizing the actual price level or some inflation target. ( i.e. The monetary authority has to increase nominal spending or aggregate demand enough to offset the downward price level pressures generated by the positive aggregate supply shock.)

    (2) Since the monetary authority is simply maintaining the existing inflation target or price level target that existed before the positive AS shock there is no change in price level expectations that would shift left the the SRAS.

    You are right with long-run super-neutrality that deflation vs. inflation issue is moot. However, we both agree that in the short run money does have real effects and that is when I argue the distinction between aggregate demand-driven deflationary pressures and aggregate supply-driven deflationary pressures can become important.

  3. Thanks David. I think I follow you. The key question is how the SRAS moves relative to the LRAS. If it moves as you show it moving in Figure 3, then they will have to move the AD curve the way you show it moving to keep the price level constant. So we are now on the same page!

    Suppose there is a supply shock, moving LRAS to the right (as in your Figure 3). In new classical (misperceptions) models, the SRAS always moves horizontally right by the same amount as the LRAS (for a given expected price level). The new classical SRAS always intersects the LRAS at the expected price level. In new keynesian (sticky nominal wage) models, it may be different. Holding the expected price level constant, the SRAS may move either more or less horizontally than the LRAS, depending on the exact type of supply shock, and whether it affects labour supply, demand, or just the production function, or some combination.

    So it depends on the exact theory of the underlying difference between LRAS and SRAS curves.

    But if what is happening is not an unexpected shock to AS, but part of an ongoing, slow, steady, and hence predictable improvement in productive capacity, such as occurs over a century, I would have thought that wages, prices, expectations, whatever, always adjust to keep us on the LRAS. In other words, if the LRAS is moving rightwards slowly, steadily, and hence gives time for everything to adjust, and if the AD cuve is also doing the same thing, the SRAS will always adjust to wherever the AD cuts the LRAS.

  4. Nick--If I may move away from the model for a moment, let me explain what it is I am trying to show. Assume there is a productivity shock. All else equal, this should lead to an increase in the real interest rate and put downward pressure on the inflation rate. It should also increase output. [Gali (1999) shows something like this empirically in a VAR with technology shocks; see his figure 4.] Now assume also that the central bank at the same time provides a positive monetary policy shock. This policy should offset the downward price pressure and temporarily push output beyond its natural rate level. If so, one would find this unsustainable increase in output would occur without the normal warning signs of increased inflation. The policy move would also keep the real interest rate below its temporarily higher level implied by the productivity shock. In short, price level stability--the result of the monetary policy shock offsetting the productivity shock--would not ensure that there would be no output gap. That is what I was attempting to demonstrate with the AD-AS model. Really what I need is a full blown IS-LM-AS model to show everything, but I was trying to make the point without too much confusion.

    [Instead of a positive monetary policy shock, one could just assume the central bank eases to offset the deflationary pressures. If there were sufficient rigidities then one could still tell a similar story... nominal spending stimulus by central bank in presence of nominal rigidities = temporary real effect]