Tuesday, March 29, 2011

Talk about consistency

I had another frustrating day, with a couple of suppliers who were swearing they were selling at a loss (come on, try to a bit more convincing), so it is time for the occasional "taking my anger out" email...

I am often asked about my forecasting methodology; therefore, I sketched it out some time ago. I think the crucial part of any forecast is consistency; financial programming and its derivatives ensure that. Otherwise, you get some inconsistent results as in CBT expectations surveys...

The aggregation in CBT surveys could explain some of the inconsistency, but there is no excuse if a major NGO puts out a projections table on their front page full of such inconsistencies:
I managed to replicate some of the TUSIAD numbers with weird assumptions on oil prices, but there is no other way I can reconcile the figures, especially the current account and growth projections. I tried to find a background paper explaining the methodology, but all they have is a footnote saying that the 2001 scenarios have been developed with the "TUSIAD time series model", which is a complete black box to me. It is normal for the IMF and CBT not to reveal their models, but I would have expected a bit of transparency from TUSIAD, especially because they seem a bit unconventional....

Speaking of transparency.... No I will not post a nude picture with my eagle tattoo, that one is already in my Facebook and Twitter pages, but I was offered a position as chief economist of TUSIAD a few months ago and had to decline a few days before my agreed starting date because of personal/health reasons. Because of this fact, I thought it would have been unethical for me to criticize them, and so I held on to this post for a while. I have decided that since I was the one who rejected, there should not be any issue of ethics here... But as I a said, I just wanted to share this fact anyway for full transparency:)...

1 comment:

Anonymous said...

Very good point! I do not believe any of the forecasters in Turkey have the guts to reveal their methodology. Some of them claim that they refrain from boring the reader with technical stuff(?!). I think, they just pretend and make us believe that they use appropriate techniques to generate those numbers. How the hell would we know? Hiding behind the so called "credibility" is not enough anymore. Of course, you could still be wrong despite all your efforts. But if you stick to the scientific metholodogy and do your part properly, then you can blame the data and nobody will accuse you of anything. Respect is an earned gift!