Menzie Chinn has done us all a service with his review of recent exchange rate theory, which I also skimmed in this. I was particularly intrigued by his reference to Frydman and Goldberg’s “Imperfect Knowledge Economics” approach. I read their article but not their book, and at the risk of thereby embarrassing myself offer these thoughts.
1. F&G are certainly right that a single model should not be expected to explain xrate movements over long periods of time because the market determinants are changing. This fits to a Kuhnian view: there are periods of “normal” trading, where movements respond predictably to economic news, and paradigm shifts—discontinuities in trading behavior.
2. PPP is a weak attractor at best. This is because the vast majority of transactions in international markets concern stocks, not flows. Forex markets are more like stamp or coin markets than markets in toothpaste, but PPP is based on the toothpaste template. Self-fulfilling prophecies can persist until the effects of currency misalignment are so disruptive that macro events force a correction; arbitrage doesn’t regulate. Any intermediate xrate would appear to be a statistical attractor due to mean reversion; is there any evidence that PPP outperforms other values in that respect?
3. F&G frame their argument in terms of forecasting, which on a practical level is certainly the test. The larger question, however, is whether their approach, or any of the alternatives, is consistent with the role assigned to xrates in micro models of international trade. This was the issue I raised in Challenge. The general answer, I still think, is that they don’t. F&G in particular present a view that, in theory, cannot be reconciled with strict comparative advantage. If a shifting bundle of macro fundamentals determine xrates, and if their weights change from one period to the next, how then can international prices be relied on to settle at levels that balance trade at the margin?
F&G end with the habitual sop toward trade orthodoxy, continuing the Hayekian tradition of deep insight into the process of markets combined with an inability or unwillingness to see that the normative view of markets has been eviscerated.
Peter,
ReplyDeleteNice post. Regarding the first point, in effect the markets go back and forth as to whether they are thinking in terms of stocks or flows, each of which has its own potential equilibria positions (and there is an old ratex monetarist argument due to Wallace, that any forex rate can be an equilibrium, holy indeterminacy!). And, of course, there are speculative bubble, or "self-fulfilling prophecies" as you call them, just to really mess things up.
Regarding the stock-flow distinction, one can see the problem with one of the most basic and important variables, relative real interest rates. So, on a flow basis, one would expect a currency to appreciate when its interest rate rises, as people will want to get those higher yields in the currency. OTOH, the higher interest rate devalues assets in the currency, with any expectation that there will be further such devaluation driving people out of the currency.
On the second point, well, PPP is a trade-based equilibrium that essentially says capital accounts (and the non-trade current account) are in balance, with trade equilibrium being the ultimate fundamental. Yes, it is weak attractor, if one that does seem to operate a little bit, and more strongly the further currencies get away from it.
And, regarding forecasting, well, Menzie Chinn has more recently defended the strength of the now quarter-century old Meese-Rogoff finding that it is very hard to beat a random walk with any model when forecasting forex rates. That certainly sours the standard story on the trade side, as you note. I would simply add that the random walk does not do a very good job of forecasting either. Forex rates are just plain hard to forecast, probably the hardest of all macro variables to forecast. They do the darnedest things, as the late Art Linletter used to say about kids...
Barkley
One minor (but not too minor) point, Barkley: has anyone tested to see whether measured PPP is the strongest attractor in its neighborhood? In other words, is its status as attractor really just a statistical artifact (it's an intermediate value) or does it offer something in addition?
ReplyDeleteIf no one has, this is an appallingly easy research project.
Peter,
ReplyDeleteWell, I am not aware of such a test or search per se. Of course, from time to time this or that model is claimed to do better than the random walk, although there has been no consistency in these results or clear pattern. The ongoing strength of that result suggests that there is probably not anything noticeably stronger than the pathetically weak PPP one.