The trouble with economics

It isn't exactly science, as noted at the economist.com:

First of all, I know that we all consider economics a science, but as sub-fields go, macroeconomics is one of the least science-y. Among the reasons—too many variables, too small samples, no repeatable experiments, and so on. Consider the paper Mr Cowen would have us consider. It examines the American economy from 1955 to 2000, and it excludes all fiscal shocks but those that are orthogonal to the business cycle. What that leaves is, well, not very much. There are similar methodologies in other key papers on the subject, including that by the family Romer, and given the range of multipliers presented I don't know how one could conclude, definitively, that the science isn't there.

Which is why it's important to have a good, qualitative model of the mechanisms involved to supplement the data analysis. Greg Mankiw gave us a potential model for a way in which tax cuts might boost private investment, but it's not clear that his narrative is superior to those explaining just how deficit-funded government investment might work. In short, the data, on its own, isn't compelling enough in such cases to justify policy.

with Mark Thoma also noting:

We have very little U.S. historical data for time periods when the economy is in a depression, so we don't know a lot about the effectiveness of policy in this framework. It's hard to find decent data about the economy prior to 1947 (and make that 1959 for data on money), and we haven't had that many recessions in that time period. And more importantly, we haven't had the deep kind of recession that depression economics is intended to address. When most of your data (half in any case) is from good times, it is not surprising that the empirical evidence finds that crowding out is an important consideration. If we had lots of episodes like the current one to look at, then I would have more confidence in these results, but we don't. Parameters such as the responsiveness of investment and money demand to changes in the interest rate, the marginal propensity to save, etc., can all change drastically in deep recessions, and that means that the results from empirical investigations covering other time periods won't be very informative. I don't think we know much at all from the econometric evidence about the success of fiscal policy in deep downturns. We'll know more in the future because we'll be able to look back at this one, but for now policymakers are flying pretty blind. What we can examine is the experience of the Great Depression, and when you do, the case for fiscal policy is strong.

Go read the rest of Marks Post Depression Economics: Normal Rules Don't Apply, as it takes apart a lot of what conservatives are saying right now... and has an awesome example for answering the question about government spending crowding out private investment.