Turkish Potential Output: Just Keeping My Promise
I did not blog yesterday, as I was working on Turkish potential output, in the little time I have from my moonlighting activities, as part of my promise to Yapi Kredi, or YKB, economics research team.
Anyway, I have reread the YKB potential output research note, where they use what I call an augmented production function approach. Let me tell you my general take on production function approaches for estimating potential output:
First of all, the simple univariate filters, such as the famous (or infamous if you were to ask YKB economists) HP filter assumes that the economy moves around the trend, so by definition, you’ll converge to full-capacity output during a business cycle horizon. I think this is an advantage of this approach. But full-capacity labor and capital demand are exogenous to the model, which introduces an element of subjectivity. For example, who is to know what trend unemployment is? Yet another problem is the technology, another area of subjective interpretation. The way “A” is constructed will heavily influence the resulting potential output estimates. Finally, estimation of the parameters of the production function will heavily influence the results.
Having made these general comments, let me turn to the YKB paper. They constrain the sum of the labor and capital parameters to be 1, but Eren Ocakverdi, one of the authors of the report, told me that they get a total larger than 1 of they don’t (this is probably because of technology). As for their production function, since the share of labor in GDP has been decreasing for Turkey, I would have opted for labor-augmenting technological change. Then, the easiest approach would be to use a constant elasticity of substitution (between capital and labor, called sigma), or CES, production function. But then the problem becomes how to choose sigma, and you’d have to depend on previous research that has done that for Turkey. In any case, the sigma you’ll choose will heavily influence your results, as the Hungarian case illustrates (page 42).
In their defense, Eren told me that they have estimated capital very conservatively by keeping lifetime and depreciation of capital very high. They have also chosen not to add a trend to the technological change; if they did, their potential output would be much higher, meaning that the economy is even more below potential output than their estimates. The fact that you are getting away from reality as you make more realistic assumptions tells me that there are problems in the model in the first place. Or it could mean that there is a serious problem with GDP measures, as Eren was suggesting.
I am not saying the production function approach is the worst method of estimating potential output. Other methods, such as univariate/multivariate filters and VARs, have their flaws as well, and in fact, production function approaches are better ones. But at the end of the day, unless I am presented with a paper that goes through at least half a dozen methods of estimating potential output, with detailed robustness tests, as in the Hungary paper, I will not be convinced by any methodology on whether the economy is overheated or not. That’s why I try to form my own subjective opinion on the overheating debate by looking at indicators of overheating, i.e. symptoms. And that’s why I said “well, that is your conclusion and this is mine and neither is conclusive”, which was criticized by YKB economists. In fact, if I had a grudge against them, or the Central Bank for that matter, I could have easily come up with a production function estimate that would have shown the economy as above potential output as their findings that show it below. That would have been economic hitmanship, but nevertheless…
But I agree with YKB economists that it would have been much better if the overheating camp countered their claims with more intellectual weapons. But given all the issues regarding measuring potential output, the economy would have cooled down before that debate reached a successful conclusion in any case:):):)
One Response to “Turkish Potential Output: Just Keeping My Promise”
edeliveli • July 21st, 2011 at 12:07 pm
Uppps, I mistakenly said long life-cycle for capital rather than short, which YKB economists immediately noticed and let me know. Apologies…. Also, please note that they do have a dummy in the model, which in effect causes technology ti change through time. But as I said, they don't have a time-trend for A…
