Thursday, March 5, 2009

Probability of depression is 20% Really?

Barro-nomics
Robert Barro Harvard professor and frequent story teller in the Journal Op-Ed section is at it again. To refresh your memory we have seen Mr Barro offer opinions before. He 'discovered' that the US economy works 'best' when the office of the CEA is vacant. He did this when trying to demonstrate that office of the Council of Economic Aadvisors is unimportant because the economist of his choice would not be occupying it that year. We have somewhat hilariously heard him tell us us that the multiplier for government spending is small and perhaps below unity using data from the war years when policy was trying to squeeze everything it could out of the (private) economy to aid the war effort as it expanded its own spending. Barro used that as evidence that Keynesian stimulus does not work. He bridged war-time policy results into a more general conclusion despite the obvious bias. Now he has discovered the market for pessimism.

'See what are the odds of a depression' (link below) in teh WSJ
Barro's answer 20%!

http://online.wsj.com/article/SB123612575524423967.html

Depression: want some? Read his article
Depressions are unusual events- thankfully. Of course the idea here is that this is conditional probability and his calculation turns on that. The associative event is the sharp drop in the stock market. Since we already know that the stock market has fallen so much, Barro has arrayed data with other periods of sharp declines in peacetime stock markets. He has calculated probabilities of 'depression' using an international data set. The data surely support Barro's probablitiy number but that is not the issue: The issue is this: do you believe their relevance? Mr Barro has a way of getting data to say things he wants them to say whether or not its true.

Peer review?
In this case Barro has lumped the US, a huge economy, in with a group of much smaller countries and more vulerable economies, right off that seems inappropriate. Barro seems to have taken no account of the activist role government has played in this epidsode. He places a zero weight on the unprecedented actions of the Fed and the Treasury and their multi-trillion dollar support effort. That harldy seems appropriate either, does it?

Odds of depression
Frankly the idea that there is a one in five chance of depression right now in the US seems one of the more irresponsible things for an economist to write. Bridging the conclusion to the US from an international data set where other economies would not have had the resources of the US and in circumstances in which their economies would have been starved of capital fleeing from a stock market collapse, hardly seem to place the US in a data set of peers (The US continues to get huge capital inflows). What relevance is there to the US in 2009? I do not doubt that Mr Barro has a supporting set of criteria for his statistical statement. I just wonder if the data have any relevance to the use he has made of them. Often in empirical research it is what you do not control for that gets you in trouble. This would not be the first time that statistics were used improperly. You can go back to most of the pricing of mortgage-backed securities in this cycle for other examples.

Ignore it!
I'd recommend not following the link to Barro's article. It will only frustrate you, worry you, and in the end confuse you, because I would bet dollars to donuts that the odds of the results of this study applying to the US right now are miniscule.

Time to kick the pessimism, if not the pessimists.

No comments: