Federal Reserve officials like to say their policy course is “data-dependent.” That sounds very cautious and intelligent, but what does it actually mean? Which data and who’s interpreting it? Let’s ask a few questions.

Photo: Federal Reserve Board of Governors

First, how could their policy choices not be data-dependent? The only alternatives would be that they made decisions randomly or that there was an a priori path already determined by previous Fed policymakers that they were forced to comply with. A predetermined path would, of course, eventually be leaked, and then everybody would know the future of Fed policy. Until they changed it.

Of course, they do depend on data, and lots of it, but are they looking at the right data? If it is the right data, theoretically speaking, is it accurate? As we will see, more often than not they are basing their decisions on data created by models that rely on potentially biased assumptions derived from past performance, etc. Often the data they look at is actually a sort of metadata, a kind of second-derivative model, with all sorts of built-in assumptions, quite removed from the actual data.

That approach is actually reasonable when you realize that the amount of data that must be managed is simply too large for any human being to process in a coherent manner. The data has to be massaged, and that means making assumptions that create the models in the programs. Those are assumptions are not made by computers, they are made by human beings who are doing the best they can – using models based on assumptions they build in to guide them as to what assumptions they should make about the data. Convoluted? Yes.

…click on the above link to read the rest of the article…