Justin Wolfers just sent me the following tweet from Alex Tabarrok:
Hypothesis: In past 48 hours there is some time frame such that combination of short and long i rates and stock prices fits any theory.
I sympathize with this argument, but I think it’s too pessimistic. Those of us who closely follow the market responses to Fed announcements see some very clear patterns:
1. In the vast majority of cases it’s easy to see if the announcement was more or less expansionary that expected. That’s because we know the expected fed funds setting from the futures market, even 5 minutes before the announcement. And the Fed usually relies on changes in the fed funds target. Hence a lower than expected fed funds target setting almost always leads to a sharp stock market rally at 2:15, and vice versa. Interestingly, the response of bond yields is less consistent. This sort of response in equities is especially likely when the market is worried about a near term recession, or low output during a recession. Check out the September and December 2007 market responses, for instance. My assertion here isn’t controversial; I think all Fed watchers on Wall Street would agree. When the Fed cuts rates more than expected, Wall Street rallies, they don’t say; “Hmm, the Fed must be really worried, maybe we should sell stocks.” In fact, stocks fall if the Fed disappoints, and the market thinks “Hmm, maybe the Fed doesn’t see what we see.” I’m not claiming this is always true, but in my experience it is almost always true.
2. Because the Fed is no longer using current changes in the fed funds rate as a policy instrument, it’s now harder to ascertain whether a particular Fed announcement is an expansionary or contractionary surprise.
3. In recent weeks both short and long term rates have been highly correlated, and both have been highly correlated with stock prices. That’s especially true using intraday data.
4. If you combine items 1, 2, and 3, then it seems very likely that if Fed announcements are moving stock prices, they are doing so through some mechanism other than changes in short and long term rates on T-securities. If that was the mechanism, then both stock prices and bond yields should move in opposite directions at high frequencies–but they clearly aren’t doing so.
Of course none of this proves my hypothesis (outlined in previous posts) is correct. But I do think our vast database of market reactions to Fed surprises means there are many fewer degrees of freedom in developing hypotheses than some people might assume–especially if they were merely focusing on the admittedly confusing response last Tuesday.
Now of course if one wants to throw away lots of data, and argue that Tuesday supports the view that Fed rates cuts boosted the market because stocks and bond yields moved in opposite directions, then yes, Alex is right. But why would we want to throw away the intraday data that strongly supports the alternative hypothesis? As an analogy, suppose a researcher found variable X and Y were negative correlated for the total change between the two data points 1960 and 2010, but were strongly positively correlated using 600 monthly changes between those two dates. Which observation would be more meaningful?