I have been arguing that highly discretionary (versus rules-based) macro policy provides the best explanation for generally poor macroeconomic performance, such as recent years, in comparison with periods of good performance, such as the 1980s and 1990s.  This was the theme of my recent Journal of Money Credit and Banking (JMCB) lecture “Monetary Policy Rules Work and Discretion Doesn’t: A Tale of Two Eras,”  which was essentially an update of Milton Friedman’s JMCB lecture 30 years earlier.

Using charts and other simple representations of rules-based policy, I identified the late 1960s and 1970s as a heavily discretionary period, the years from 1985 to 2003 as more rules-based, and the period from 2003 through the present as another discretionary period.  Clearly average economic performance in the discretionary policy periods, such as the past few years, has been much worse than during the rules-based period, and the timing helps support the idea that this is causal from policy to performance.

But as Alex Nikolsko-Rzhevskyy, David Papell, and Ruxandra Prodan pointed out in a recent paper and guest blog post on Econbrowser, there is a danger with this approach.  The procedure for choosing and defining rules-based policy periods and discretionary periods, if not done sufficiently objectively and rigorously, can be affected by the performance.

But Nikolsko-Rzhevskyy, Papell, and Prodan did not stop at criticism; they went ahead and applied more objective statistical methods to the problem.  They set out to find and detect, using these methods, which periods were rules-based and which were not. As they explain in their Econbrowser post, their “structural change tests” detected a monetary policy that was largely discretionary from 1974 to 1984, rules-based from 1985 to 2000, and discretionary from 2001 to 2008.  These periods are quite close, though not exactly the same, as the less formal approach I used.  They also examined a Markov switching model with roughly similar results.

As the authors put it “These results both accord with and reinforce previous work that identifies monetary policy eras less formally, although our dating the start of the most recent extended discretionary era in 2001 is earlier than most accounts. In contrast to previous work, however, our results are not subject to the criticism that the choice of eras was influenced by subsequent outcomes. They therefore provide a better basis to evaluate whether ‘monetary policy rules work and discretion doesn’t.’”