Thursday, June 30, 2011

The Lucas Critique

One of the most powerful ideas in economics is the Lucas Critique. The notion is that statistical relationships estimated in historical data do not necessarily represent causal relationships manipulable by policymakers. It applied in particular force to the Phillips Curve, a connection that some researchers found between unemployment and inflation. According to this critique; simply observing that unemployment and inflation tend to move opposite one another does not mean that central bankers can push inflation higher at will and gain lower employment.

This was a powerful critique at the time, and has yet to penetrate a lot of economic talk. Here, for instance, is Christina Romer, former Chair of the CEA:
The real division is not about the acceptable level of inflation, but about its causes, and the dispute is limiting the Fed’s aid to the economic recovery. The debate is between what I would describe as empiricists and theorists.

Empiricists, as the name suggests, put most weight on the evidence. Empirical analysis shows that the main determinants of inflation are past inflation and unemployment. Inflation rises when unemployment is below normal and falls when it is above normal.
It seems that Romer is accepting not only the empirical Phillips Curve relationship, but also its causal ability to be used by central bankers. Not only is this her own opinion, but the view that ought to be held by "empiricists" who are rigorous in the way they face data, as opposed to those woolley-headed theorists.

Yet even the empirical relationship between unemployment and inflation has broken down in the past few decades. And even if one such relationship did exist, that would not necessarily provide a guide for monetary policy. I do agree that more monetary easing would be worthwhile, but this is a bad way argument in its favor.

On the other side, you have other individuals arguing against monetary easing on the grounds that higher structural unemployment makes monetary easing futile. Scott Sumner neatly addresses that argument:
For similar reasons there is no hard and fast distinction between cyclical and structural unemployment. For instance, if structural unemployment in American has risen closer to European levels, it may be partly due to the decision to extend unemployment insurance from 26 weeks to 99 weeks, and to increase the minimum wage by over 40% right before the recession. Does that mean that demand stimulus cannot lower unemployment? No, because the maximum length of unemployment insurance is itself an endogenous variable. If stimulus were to sharply boost aggregate demand it is quite likely that Congress would return the UI limit to 26 weeks, as it has during previous recoveries. For similar reasons, the real minimum wage would decline with more rapid growth in demand. Aggregate supply and demand are hopelessly entangled, a problem that many economists haven’t fully recognised.
Once again, some relationship we observe today ("there is more structural unemployment now") doesn't provide an unambiguous guide on what to do in the future.

Finally, here’s a recent example from India’s Economist Prime Minister, Manmohan Singh:
We are committed to a growth rate of 9 to 10 % per annum. Our savings rate is about 34 to 35 % of our GDP with an investment rate of 36 to 37 %. And with a capital output ratio of 4:1 we can manage to have a growth rate of 9%.
On the face of it, this is a reasonable statement. Savings do translate into investment closely enough (give or take foreign direct investment, cash stuffed under the mattress, etc.), and accumulated capital in the form of investment aids in future output growth. And, at any point in time, one can compute the ratio between capital and output as a ratio. That’s all true enough.

What’s disturbing is the manner in which Manmohan Singh apparently relies on the crutch of capital/output as a solid parameter to be manipulable by policy. This has been a consistent factor in his economic thinking for quite some time, and goes back to the assumptions of early Indian planners that the accumulation of capital alone would suffice for growth.

Well that wasn’t necessarily true in Nehru’s time and it's not true now. For instance, note as Yasheng Huang does that China is far less effective than India at translating savings into growth (the country also saves more in general). So this is not some fixed parameter set by immutable laws. Instead the capital/output ratio is highly responsive to the general policy environment and incentives faced by economic actors. In China, presumably what you have going on is a lower degree of allocative efficiency. Yet one might equally have concerns in the Indian context about the role of policy; with microeconomic problems of labor quality, health, land acquisition, general governance, labor laws, taxation, and so on and so forth. It is exactly in order to evade the government’s abysmal failure to tackle these existing and tangible problems that Manmohan Singh suggests that the problem of growth can be reduced to an arithmetic question of savings and investment. Yet that relationship may not hold up in the absence of additional reforms to improve governance and tackle the various other binding constraints that hold back growth. (oh, to be sure he mentions other steps to make sure this will happen; but if his proposals haven't taken off in the last seven years why would they take effect now?).

Let me put it this way — Tim Pawlenty recently received a lot of flack for suggesting that his economic policy would simply demand 5% growth for a decade. What if he had said instead; “I will push savings up to 20%; then since there is a 4:1 relationship between investment and growth we can expect 5% growth.” I think most people would find the idea a little nuts. Many misadventures in development economics have included failures where simply pumping in more capital didn't necessarily get you more growth in an arithmetic fashion. The root problem is that any current relationship between savings and growth doesn’t represent a causal relationship manipulable by policymakers. That’s the Lucas Critique in action.

1 comment:

Michelle said...

Virtually no evidence exists that empirically substantiates the Lucas critique. Empirical refutation of the Lucas critique by using tests of super exogeneity is illustrated with U.K. money demand.

Relationship Banking