Saturday, November 15, 2008

Some nontraditional observations on the latest Turkish data

Below is my take on Turkish data of the last week. I know it is not the usual data turned out to be X; implications are Y type of analysis, but if you are interested in that, I'll be writing a blog on my favorite Turkish economy reports soon.

First, the obvious: As many columnists market economists have reported, the latest data point to a slowdown and growth and several analysts have cut their 2008 and 2009 growth forecasts in the last few days. Moreover, while it is always problematic to read into one-month of data, the September current account as well as other data released in November hint that the economy might be adjusting faster than you can solve a Calvo model (one of the more challenging ones, at least).

Second, the less obvious: Forecasts of market economists have started to really miss the target. Forecasts are bound to to be different from actual realizations, and anyone who follows such forecasts knows to take them with a grain of salt- in fact, Ferhan Salman shows that market forecasts tend to be outperformed by simple time series models. But what is more intriguing that the accuracy of the forecasts has increased considerably over the last couple of months. When I simply compare the mean of the big three market forecasts (inflation, current account and industrial production), as reported by CNBC-E, I see the absolute value of the difference between the forecasts and realizations widening considerably in the last two months. Therefore, the strong form (see below) of the unskillful analyst explanation can be rejected unless all research houses fired their economists and hired less skillful ones at the end of summer.

Now, there is a catch: Leaving aside the issue that my data is only from 2008 (9 observations!), it could be that a couple of outliers could be driving the results. For example, it could be that the less skillful analysts could be driving the mean away from the realizations. Therefore, I looked at the forecasts of each bank in the CNBC-E sample separately, and I still get the same result. Therefore, I am able to reject the weak form of the unskillful analyst explanation as well.

In fact, it is rational for each analyst to follow the herd. While I have not seen a similar study for economists recently ( but there are quite a few papers from Makriadis in the last two decades), there are many papers that explain why equity analysts flock. Basically, if you are wrong (and so is everyone else), it won't look that bad, but if you make a radical forecast different from everyone else's and you are wrong, the traders who depend on your input won't be very happy, as I can personally tell from my experience as a bank economist. However, if this is the case, then the all the forecasts should be bundled together. And they were, until the end of the summer.

This brings me to my third and least obvious point: Not only are the forecasts being less accurate, they are also getting increasingly disperse as measured by their standard deviation (again, I am talking about the big three for 2008).

What do I make all of this? I think we are entering a very uncertain environment and because of the unpredictability of certain key assumptions, even professional forecasters are having a hard time seeing where the Turkish economy is headed. I am not trying to say they are clueless. Think of it it as a ship sailing south somewhere in the Pacific before longitude could be measured at sea; they all know that we are heading south, but they are just not sure we'll end up landing in Australia or New Zealand. The scary part is that we could also be heading to Antarctica and not be aware of it yet...

No comments: