Michael Woodford's Unjustified Skepticism on Portfolio Balance (A Seriously Wonky Rebuttal)
In his recent speech at the Federal Reserve’s annual Jackson Hole conference, Professor Michael Woodford of Columbia University attempted to pour cold water on the idea that the Fed’s purchases of long-term bonds (also known as quantitative easing) could lower bond yields.1 His contention was that the portfolio balance effect of such purchases would be minimal at best. I disagree, as do the bulk of central bankers. This debate matters because, if Woodford is right, the Fed’s only tool for delivering more stimulus now is to commit to future policy actions that may be viewed as undesirable when they occur—such as promising not to raise interest rates when inflation returns. If the market were to doubt such a commitment from the Fed, the Fed would lose its ability to steer the economy. In reality, the portfolio balance channel gives the Fed a tool to guide the economy without unduly restricting future policy choices. This is not to deny, however, that Fed statements about future policy intentions may have important effects.
Woodford devotes several pages in his Jackson Hole remarks to discussing a paper I wrote two years ago with former Fed colleagues Matthew Raskin, Julie Remache, and Brian Sack. We showed that Fed purchases of long-term agency and government bonds in 2008 and 2009 lowered a range of long-term interest rates. We argued that most of those declines appeared to reflect a reduction in term premiums rather than a reduction in expected future short-term interest rates. We posited that those reductions in term premiums were required to induce investors to accept the shift in their portfolios engendered by the Fed’s purchases; this is what Nobel Laureate James Tobin and others have long referred to as the portfolio balance effect. Woodford asserts instead that most or perhaps all of the declines in bond yields might have been caused by the market’s interpretation of the Fed’s statements and actions as indicating that the path of future short-term interest rates would be lower than previously expected.
Our paper, and more recent papers, presented evidence that counters most of Woodford’s empirical claims (D’Amico and King 2010, Neely 2010, D’Amico, English, Lopez-Salido, and Nelson 2011, Hamilton and Wu 2012, and Li and Wei 2012).1 The evidence shows that Fed purchases affected the prices of a broad range of assets at once, not just the prices of those bonds being purchased directly. Some Fed statements and actions clearly had no implications for the path of future short-term rates, and yet they still affected some asset prices. This evidence comes from a range of investigations, not solely from event studies, and the conclusion is not sensitive to changes in the underlying model used to identify movements in the term premium. These effects of Fed asset purchases are fully consistent with what would have been expected based on data from before the financial crisis, and with the portfolio balance view.
Readers are encouraged to check out the above papers for more on the empirical case for a portfolio balance effect as well as an Econbrowser post in response to Woodford by Jim Hamilton. In the remainder of this post, I focus on Woodford’s theoretical arguments against portfolio balance. I find them unpersuasive. For more theoretical arguments for and against portfolio balance, see a post by Cohen-Selton and Monnet on the Bruegel blog.
Woodford asserts that there can be no portfolio balance effect if two conditions hold: (1) the assets being bought and sold are valued only for their pecuniary returns, and (2) all investors can purchase and sell unlimited quantities of these assets. A third requirement he does not mention is that investors are rational, forward-looking, and fully informed about how the economy and political system operate.
These conditions are violated in clear and obvious ways in the real world. Indeed, Woodford acknowledges that money has a non-pecuniary return—transactions services—which is required for conventional monetary policy to work. There is no reason to assume that money is the only asset with a non-pecuniary return. The violation of Woodford’s second theoretical assumption is even clearer, as it requires that private agents have an unlimited ability to take short positions in the assets the Fed is buying. In fact, private agents seeking to borrow at any maturity face severe and binding collateral requirements and pay considerably higher interest rates than the Treasury.
It is common in economics to make the benchmark assumption of rational, forward-looking, fully informed agents. Woodford suggests that some of the empirical evidence cited in favor of portfolio balance—the large movements of bond yields in anticipation of future Fed purchases—are in fact evidence that investors are rational and fully informed. But it is one thing for investors to respond to a Fed announcement about imminent bond purchases and quite another for them to fully anticipate the implications of those purchases for their taxes and transfers in the distant future. Yet the case against portfolio balance requires precisely that knowledge.
Woodford further argues that even if there is a portfolio balance effect, quantitative easing might have greater costs than benefits because, for example, it could make a given class of assets scarce. But this is comparable to the effects of conventional monetary policy, which alters the relative returns on different assets, benefitting some households and hurting others. Such effects are secondary in importance and in magnitude to the broader goal of macroeconomic stabilization.
1. Woodford separates the expansion of the Fed’s balance sheet, which he terms quantitative easing, from purchases of unconventional assets, which he terms targeted asset purchases. But most observers analyze these policies jointly under the name of quantitative easing.
2. The paper by D’Amico, English, Lopez-Salido, and Nelson was presented at the Bank of England in November 2011 and is forthcoming in The Economic Journal but does not appear to be available online.