I spent Friday at the NBER Asset Pricing conference in Palo Alto. All the papers were really good, and the discussions were especially thoughtful. Here are a few highlights that blog readers might like.
The Pre-FOMC Announcement Drift.(If these links don't work for you, most papers can be found with google.)
Here are average cumulative returns on the S&P 500 in the day preceding scheduled FOMC announcements (when the Fed says what it will do with interest rates). The grey shaded areas are 2 standard error confidence intervals. The S&P500 drifts up half a percent in the day before FOMC announcements! In fact, 80% of the total return on the S&P500 over this period was earned on these days.
So what the heck is this? Obviously, there was some disssection that it is spurious. Stocks are so volatile that it's easy to find 3 days that account for 80% of its total return, let alone a few hundred. But they did not fish.
I am still a little worried -- there are two big positive outliers in the distribution of returns (Figure 2) and a missing left tail. If we had two more such outliers on the negative side, would those confidence intervals get bigger? Did options markets know that the left tail is missing? The pattern is there for international stocks before US announcements, but not in bond markets. But Annette Vissing-Jorgenson, setting the style for the day, dissected it every which way and didn't get rid of it, so discussion moved on.
This observation generated what I'll call the consensus of the room: Lots of equity traders sit out or hedge their positions in advance of this day. (Confirmed by people in the room who talk to such traders.) Anyone trading on the morning before an FOMC announcement is suspected of being "informed," which makes markets less liquid. So the risk is concentrated and held by a narrower group.
Emanuel answered that they did regressions including these and all sorts of liquidity measures, which didn't get rid of the puzzle. But when you do that, you assess how much expected return premium corresponds to illiquidity by the correlation of returns with liquidity on other days, and there are all sorts of reasons to think this measurement underestimates the effect. Anyway, as a fan of facts linking trading to pricing, it's a great paper. (And a good hint to PhD students: make sexy graphs like these.)
Annette also brought up the issue, should journals publish papers that just pose well-documented puzzles, without offering (usually lame -- my view) theory or explanation? I think this paper makes a hearty case for "yes!"
Xiaoji Lin presented his paper with Jack Favilukis Wage Rigidity: A Solution to Several Asset Pricing Puzzles. How can I make a general equilibrium model with adjustment costs and wage rigidity sexy for a blog?
Well, this one is. "Standard" real-business cycle models drive the economy with productivity shocks. When there is a good such shock, investment and output go up, and people work harder. But, the marginal product of labor goes up (that's why they work harder), so wages go up. Since wages go up, profits don't go up that much, and equity isn't that risky. By putting sticky wages in the model, now wages are like a bond payment, so the firms profits are leveraged, making them more risky. This helps to fit a broad range of asset pricing facts. I'm especially impressed that the model generates a spread of value vs. growth firms (hard to do) and a value premium.
Lots of discussion here. Of course "stickiness" is an abstraction for all the interesting things that labor/macro people put in their models. One good comment, wages are not "smoothed," they're "screwed" -- workers don't get the present value of the marginal product increase (or feel it if a decrease) as they might under an intertemporal smoothing contract. That likely has a big effect on the value of stocks.
The last one I'll mention (they were all great, just running out of steam here) Erkko Etula and Tyler Muir presented their paper with Tobias Adrian on "Financial Intermediaries and the Cross Section of Asset Returns"
They construct shocks to broker-dealer leverage from the flow of funds, and then construct a single-factor asset pricing model, expected excess return = beta on broker-dealer leverage shocks times lambda. Here it prices the size and value portfolios, momentum portfolios, and bond portfolios! All with a single, economically motivated factor!
Much discussion (of course). One possibility, which I called the "AQR theory of asset pricing." Suppose you look at the portfolio of one trader, who is invested in value, momentum, small, and term risk. The the wealth, and (if borrowing is pretty constant) leverage of that agent will be a good pricing factor for those anomalies. So just because leverage works well does not necessarily prove the usual causal story, that these broker-dealers are "marginal," they get in to trouble sometimes and then start selling securities in "fire sales," etc. If they sell, after all, someone else must buy, so they're "marginal" too. Much good discussion on the facts too, with great graphs by Bryan Kelly showing that it is a bit unstable over different samples.
Lubos Pastor presented his paper with Pietro Veronesi "Political Uncertainty and Risk Premia" with a great discussion by Nick Bloom showing us the latest of his uncertainty index. Welcome to the Krugman-thinks-you're-a-moron club, Nick.
Snehal Banerjee presented his paper with Jeremy Graveline, "Trading in Derivatives When the Underlying is Scarce", really interesting (especially to me, given writing on the 3 com / palm issue) with Nicolae Garleanu discussing.
Chris Polk, presented his paper Dong Lou "Comomentum: Inferring Arbitrage Capital from Return Correlations" with a great discussion by Robert Novy-Marx. They find that momentum works when the pairwise correlations of momentum stocks are low, indicating the trade is "less crowded," and conversely.
All cool stuff, but but the plane is landing. (Thanks to glamorous Southwest airlines for onboard wifi.)