Janet Yellen gave a speach which has received a lot of attention and not just because she is one of the most powerful people in the world. I agree with Lorcan Roche Kelly that the best parts are footnotes 28 and 26.
I got into a twitter debate with Andy Harless who was unconvinced by footnote 28
28. My interpretation of the historical evidence is that long-run inflation expectations become anchored at a particular level only after a central bank succeeds in keeping actual inflation near some target level for many years. For that reason, I am somewhat skeptical about the actual effectiveness of any monetary policy that relies primarily on the central bank’s theoretical ability to influence the public’s inflation expectations directly by simply announcing that it will pursue a different inflation goal in the future. Although such announcements might potentially persuade some financial market participants and professional forecasters to shift their expectations, other members of the public are probably much less likely to do so. Hence, actual inflation would probably be affected only after the central bank has had sufficient time to concretely demonstrate its sustained commitment and ability to generate a new norm for the average level of inflation and the behavior of monetary policy–a process that might take years, based on U.S. experience. Consistent with my assessment that announcements alone are not enough, Bernanke and others (1999) found no evidence across a number of countries that the initial disinflation which follows the adoption of inflation targeting is any less costly than disinflations carried out under alternative monetary regimes.
This is my view expressed by my snarky comments about the expected inflation imp. I didn’t know about Bernanke et al 1999 and just looked at the Volcker deflation (which is alleged to show the power of expecations management but doesn’t).
Harless disagreed, Brad DeLong retweeted his tweets, and I promised to write a blog post, so here it is.
Andy Harless @AndyHarless Sep 25
Disagree with Yellen’s note 28. For inflation expectations to have an effect, you don’t need to convince the public, just the smart money.
I disagreed with Harless and will reproduce the tweet debate after the jump. What I should have written is that Yellen’s footnote 28 must be read in the context of footnote 26
26. Another complication is that we do not know whose expectations “matter” for determining inflation. Inflation expectations of professional forecasters (such as those collected in the Blue Chip Economic Indicators, the Survey of Professional Forecasters, or the Survey of Primary Dealers) or inflation expectations derived from asset prices probably capture the views of participants in financial markets but need not reflect the views of households and firms more broadly. As an empirical matter, the little information available on the longer-term inflation expectations of firms from the Atlanta Federal Reserve Business Inflation Expectations survey suggests that firms’ expectations more closely resemble expectations from the University of Michigan Surveys of Consumers than the expectations of professional forecasters. Similarly, preliminary data from a New Zealand study suggests that inflation forecasts of firms are much more similar to those of households than those of professional forecasters (see Coibion and Gorodnichenko, 2015).
The “other members of the public” mentioned in footnote 28 include the managers of firms. Assuming Yellen’s claims of fact are correct and the “smart money” consists of bond traders and professional forecasters but not ordinary people or managers of non-financial firms, how can the “smart money” affect aggregate demand ?
The Harless & Waldmann debate after the jump (but Waldmann gets almost all the pixels as he knows the Angrybear blogger password (or at least his hard disk does)).
By the way, while the quotes aren’t scare quote (I am quoting Harless) I don’t think the smart money is smart at all — I think they grossly over-react to recent inflation.
robertwaldmann @robertwaldmann Sep 27
@AndyHarless How does the smart money cause a boom in real output ?
I can see how the expected increase in house prices has had a large effect on aggregate demand. But I don’t think that it is based on Fed announcements. In fact, while Shiller argues that the general price index is an excellent predictor of house prices, it is clear that ordinary people believe that changes in the relative price of houses are extremely important. I don’t see any way to manage these expectations. I firmly think that these are the price expectations that matter if one wishes to calculate an economically useful real interest rate. I think that interest rates (real and nominal) affect aggregate demand through residential investment.
If the inflation expectations of managers of firms are similar to those of ordinary people and not of “smart money” then it is hard to affect teir investment through their beliefs about the price for which they can sell their products. I think they interact with the “smart money” through nominal interest rates (on bonds they issue and on loans) not expected inflation. Higher long term inflation expected by bond traders but not managers would cause higher nominal interest rates and higher real rates as expected by managers.
Harless argues for two channels
Andy Harless @AndyHarless Sep 28
@robertwaldmann Bid up prices of real assets (or forex) => produce more real assets (or exports) => output boom bcuz static Phillips curve
I grant the (forex) and (exports) arguments. I agree that monetary authorities can beggar their neighbors by driving down exchange rates (although it is easier to do this by declaring a target exchange rate and buying foreign denominated assets than by influencing inflation expectations). Living in Europe, I don’t consider this an acceptable approach for the FOMC — the US has less of an aggregate demand problem than other rich countries and its neighbors bet it not to beggar us.
But what is this about “real assets” ? The only ones I can think of are stock prices. I think the evidence that stock prices have essentially no effect on investment decisions is overwhelming. The failure of the Q theory of investment makes the permanent income hypothesis and the Lucas supply function look good. Did you notice the effects of the huge shift in stock prices in the 1987 crash ? Neither did I.
Harless goes on
Andy Harless @AndyHarless Sep 28
@robertwaldmann @Frances_Coppola But even if you think it’s only houses, smart money inflation expectations => they will finance builders
What ?!? is the idea that smart money investors and builders will build houses that ordinary people don’t want to buy ?!? I think this just doesn’t happen. Empty houses are a huge waste. for one thing they lose value as they are vandalized and stuff. But for another required returns on risky investments are huge compared to the sort of shifts in expected inflation which are being discussed. I think this is crazy.
I don’t know where to put this, but I don’t think that interest rates have much effect on non residential investment. The interest rate term in estimated accelerators is, as far as I can tell with 60s style econometrics entirely due to residential investment. This actually makes sense as about two thirds of non-residential fixed capital investment is in equipment and software and, well, I’ll just hand the microphone to Krugman.
I’ll only add that, I guess (without data) that, by value, most construction of non residential structures is construction of office buildings and shopping centers not factory buildings and warehouses. I think it tracks housing investment, because the nonresidential capital complements residential capital. I think even in the case of factory buildings, interest rates probably don’t matter much. I think in that case the key issue is capacity utilization. I can see how even if a firm has enough equipment to meet demand it might buy new equipment to reduce other costs, but I don’t see why it will build space it doesn’t need immediately. I also think that the future resale value of a factory building is always almost exactly zero no matter how much inflation there is.
Oh finally, I think that interest rates wether nominal or real have a negligible effect on consumption — there is essentially no evidence of an inter-temporal substitution effect in the data.
*Grammar error corrected thanks to Kaleberg in comments