I have sometimes expressed here my frustration over the disconnection between research and reality. The scientific process is about trying to explain some data. Any paper that just tries to see what could result from a model without looking at data or restrict parameter value in any way is useless. What do we learn if a model tells us anything could happen with some parameter values, but the authors are not capable of telling us in which range those parameter values lie? And what use is a model that is not, or worse cannot, be tested in any way?
Harald Uhlig opines on the relationship of Economics and reality from four different angles. The first is Economics as a science, where he has largely the same arguments as me above. Economics is also an art, as it is vain to find a great unified theory of Economics, as life is too complicated. One can only hope to explain observations in batches, and then the most elegant theory wins. But that theory still has to be tested against data, at least for the observations at hand.
Economics is also rhetoric. Indeed, convincing others is a big part of the success of a theory, and how that theory does with explaining the data is unfortunately not always central. This competition of ideas also spills over into politics, as Economics is in the end about giving policy advice. And this is where data and empirical relevance becomes the least important.
Uhlig then goes on to relate this to various aspects of macroeconomics. While there is much introspection in macroeconomics these days, I wish he would have defined Economics to be broader than that. Some areas of Economics need indeed to be much more closely related to data and trying to explain empirical observations. Game theory would be a prime example. public economic theory as well, if not much of microeconomic theory. Mainstream macroeconomic theory has always tried to relate to data, which is probably why it has changed over the decades: some crucial facts have changed.