« Affordable Housing | Main | Should the Fed Focus on Core Inflation or Headline Inflation? »

Sunday, August 27, 2006

Kudlow Needs Help with the Phillips Curve

Cactus, writing at Angry Bear, takes on Kudlow's latest at the NRO and his claims about tax cuts and the budget deficit. It's the usual from the NRO crowd, cut taxes and the world will be wonderful. Cactus notes

Kudlow ends his post with the words: “I don’t get it.” Well, I agree with him about something.

Separately in an email PGL, who is busy today (update: he found some time), notes the following statement from Kudlow and encouraged me to comment:

Let's start with the Fed's goofy sacrifice ratio, which basically refers to how much unemployment has to go up in order to bring inflation down. I call this economic garbage the "Phillips curve in drag," because over the last 25 years, unemployment and inflation have actually moved in tandem and they have both moved down. In other words, as inflation slows, unemployment comes down because the economy is strong. (If you look at their relationship during the 1970s, you would see unemployment and inflation both moving higher.) The fact is, strong growth coexists rather nicely with low inflation. And since inflation is too much money chasing too few goods, then if you're producing more goods that absorb the money supply, especially with low tax rates to produce more goods, then why should we fear growth?"

Okay, I'll bite.

There is a lot of confusion over short-run and long-run Phillips curves, and whether the short-run tradeoff is exploitable. First, there's little doubt that inflation and unemployment were negatively correlated in the 1970s and in other time periods as well. It requires going past simplistic raw correlations to see that clearly (e.g. to separate out supply shocks where correlations are positive and isolate demand driven fluctuations), but there's really little doubt on this point. Most of the debate currently is over whether a hybrid New Keynesian Phillips curve is a better description than the traditional Phillips curve, not over the existence of the Phillips curve itself (e.g. see Rudd and Whelan and the response by Gali, Gertler, and Lopez-Salido for recent estimates of various specifications of the Phillips curve and discussion on this point).

The current war against the Phillips curve from a small section of the right, and a few economists, is amusing. One of the main motivations for building the New Classical model and the models that followed was to try to explain the positive short-run correlation between money and output (which is just another way of stating that inflation and unemployment are negatively correlated, i.e. the goal was to explain the Phillips curve correlations) without admitting that systematic policy intervention could affect real variables such as output and employment.

The traditional classical model had nothing useful to say about short-run fluctuations, and the traditional Keynesian model, as it existed at the time, could explain the correlations but was easy to criticize due to its lack of microfoundations to support assumptions about rigid prices or other imperfections in the adjustment mechanism. The poor treatment of inflation and expectations in the traditional Keynesian framework led to further problems (the New Keynesian model has resolved most of there issues). The formulation the New Classical economists came up with, incomplete information as the source of fluctuations, was a clever method of explaining the Phillips curve correlations in an equilibrium setting where expectations are rational and the natural rate hypothesis holds while still denying that policy can affect real variables. The model had trouble explaining both the magnitude and duration of actual cycles, and there were other problems, so it has been replaced by subsequent theoretical constructs such as the New Keynesian model, the current dominant paradigm.

There is a group, the real business cycle theorists, who would argue that the short-run Phillips curve is not a useful description of the adjustment process (others, such as Robert Hall, argue that the output gap is not a relevant or useful measure of economic activity, and strictly interpreted there is no gap at all, but see Robert Bean's response to Hall as well as the cautions from Jean-Claude Trichet on the use of the gap). But even those who argue against the existence of an exploitable short-run Phillips curve don't generally deny that money and output are correlated, instead they argue that the money-output correlation is due to to factors such as reverse causality - i.e. they argue that productivity shocks increase the natural rate of output which increases money demand. If the Fed accommodates the increased money demand, then output and money will co-vary positively. I haven't heard this argument lately, but in any case most models take as given that output and money or inflation and unemployment are correlated. The goal is to explain why the are correlated and, in particular, whether the correlation can be exploited to stabilize economic fluctuations. According to the New Keynesian model, it can. There is nothing inconsistent about strong output growth and falling inflation in the standard Phillips curve framework as would happen, for example, as the long-run Phillips curve shifts outward due to productivity driven growth while the short-run Phillips curve moves downward along the long-run Phillips curve at the same time due to falling inflationary expectations or adjutments to sluggish prices.

While we're on this topic, let me help Kudlow with his inflation and output growth argument. Staying within the classical framework, it's all very simple in the long-run. Start with the quantity equation, MV=PY, and take percentage changes:

%ΔM + %ΔV = %ΔP + %ΔY


(Money growth) + (growth in velocity) =  (inflation) + (output growth)

and finally

inflation = money growth - output growth

assuming that velocity changes are close enough to zero to ignore (if not, just add them in). Thus, so long as money growth equals output growth, i.e. money grows to accommodate increased money demand from expanding output, there will be no inflation. But, any money growth in excess of output growth will cause both output and inflation to increase in the short-run as described, like it or not, by the short-run Phillips curve, and will be purely inflationary in the long-run as descried by the long-run Phillips curve. There is a lot of confusion between the short-run and long-run Phillips curve in what has been written. I think what these writers are really trying to argue (or at least should be trying to argue) is that the short-run tradeoff is not exploitable. There are those who deny any significant role for demand driven output and employment fluctuations, but I don't think the empirical evidence supports that position.

Bottom line: There is a Phillips curve. In the short-run, it has a negative slope, in the long-run it is vertical. That the short-run tradeoff is exploitble by policymakers is doubted by some, but the evidence for an exploitable short-run tradeoff is fairly strong.

Update: KNZN offers more help, with pictures.

    Posted by on Sunday, August 27, 2006 at 01:44 PM in Economics, Macroeconomics | Permalink  TrackBack (0)  Comments (8)


    TrackBack URL for this entry:

    Listed below are links to weblogs that reference Kudlow Needs Help with the Phillips Curve:


    Feed You can follow this conversation by subscribing to the comment feed for this post.