Thursday, February 19, 2009

Why determining causation is important

Yet economists have been shirking away from this task and have instead opted to go with correlation (and then adding the caveat that "correlation does not imply causation").

In the 'golden age' of economics structural equation modeling exploded and while these models were large complex, it allowed economists to say clearly the impact of a change in an exogenous variable. (Whether the variable is truly exogenous is another subject altogether.) Then came the "Lucas Critique" which not only destroyed the foundations of structural equation models but led economists to abandon making causal statements and instead resort to using "reduced form" estimates that illuminate little and say even less.

Two recent posts in the blogosphere point to why it is important to be able to make a causal statement in light of the current financial crisis:
1. Megan McArdle's Did World War 2 End the Great Depression or in another form in various blogs as Did the New Deal End the Great Depression - economists should be able to say that the impact of WW2 reduced the output gap by y percent, or alternatively, taken together, New Deal policies increased the output gap by x percent. If the output gap is too vague a measure then the statement can be: Because of WW2 (or New Deal policies) output was y percent higher than it would have been without WW2 (or New Deal policies).

2. Mankiw, commenting on Obama's recent plan states the following:
The expression "create or save," which has been used regularly by the President and his economic team, is an act of political genius. You can measure how many jobs are created between two points in time. But there is no way to measure how many jobs are saved.

While there is no way to measure how many jobs are saved, it is possible to think of the counter factual: Of what would happen to jobs if the fiscal stimulus (or whatever policy that is under consideration). In fact economists do this all the time - for instance, CGE models are used to make claims about what the gains from trade will be. (Gains from trade itself sounds like an immeasurable concept.) Likewise, DSGE or VAR models are used to make claims about the effects of monetary policy or tax cuts.

Yet today when they are needed most to make causal statements about the impact of fiscal stimulus versus tax cuts and instead of dealing with the causal impacts with modern economics tools such as VAR or DSGE or even a return to structural equation models, they have chosen to wage a war of ideology and name calling over the Internet or media.

The amount of time spent slinging mud could have been spent building models and examining and reexamining the data and if economists do not know the price of the trade off between name calling and making model based impact statements then no one really does and it all degenerates into muck raking. The more ominous trend is that economists have actually abandoned all hope of making causal statements and have made the conscious trade off that it is more beneficial to be profiled as being partisan and peddling policy snake oil than being a real economist.

Is the profession in such a mess that instead of battling it out with new models and new analysis of data that it is more welfare enhancing to revisit the Romer studies and making Great Depression analogies?

No comments: