Scott Sumner  

Adding a risk mandate: A 5 bumper shot in billiards?

PRINT
How Diabolical is Unz's Propos... Pillars of Economic Wisdom in ...

Jeremy Stein gave a speech that advocated adding financial risk to the Fed's traditional dual mandate. Or at least targeting risk in the hope that it makes it easier to fulfill its traditional mandate. I see lots of potential problems, and not much upside:

1. Can the Fed correctly judge the stance of monetary policy? My general fear has been that the Fed will wrongly associate easy money with low interest rates. And then assume that low interest rates lead to lax financial conditions. In fact, low rates are highly correlated with low NGDP, which means it usually reflects a tight monetary policy. The perception that money is easy may have caused the Fed to taper recently, when more monetary stimulus is probably needed to hit their inflation/employment targets. Now in fairness, Stein seems to have a better understanding of the stance of monetary policy than most Fed officials:

I am going to try to make the case that, all else being equal, monetary policy should be less accommodative--by which I mean that it should be willing to tolerate a larger forecast shortfall of the path of the unemployment rate from its full-employment level--when estimates of risk premiums in the bond market are abnormally low.
That's right, money is actually fairly tight right now as inflation and employment are both likely to be below target next year. However I worry the Fed will forget this perspective in a crisis, and revert to the usual "low rates = easy money."

2. Because low rates can reflect either easy money (liquidity effect) or tight money (income, price level, expected inflation effects) the stance of monetary policy is often misidentified in empirical studies. This gives me little confidence that we can accurately estimate the impact of monetary policy on risk premiums---a key precondition for Stein's proposed policy.

3. How much does risk premium instability lead to NGDP instability? Stein's proposal assumes the causation goes from risk instability to NGDP instability, but the reverse causation seems far more plausible to me. And note that this issue cannot be addressed via "Granger causality" studies, as financial markets are forward-looking.

4. Can monetary policy outperform regulatory fixes? Monetary policy would certainly seem to fall into the "second best" policy category. One of my professors complained that 2nd best policies are generally enacted by 3rd best policymakers and get 4th best results.

5. Even if monetary policy can impact the time path of risk premiums, can we be confident that the policy succeeds in reducing NGDP volatility? The last time I recall the Fed diverging from macro stabilization and focusing on risk was 1929. They did succeed in impacting the financial markets exactly as they hoped, but it did not have the NGDP stabilization effects that they expected from popping the stock market bubble. In fairness, Stein is not calling for bubble popping, but I still have doubts about the link between using policy to move the risk premium, and greater NGDP stability.

Common sense suggests that a stable path of NGDP will result in more financial market stability than a highly unstable path of NGDP. And many financial crises are associated with unstable NGDP (although causality can go either way.) It seems to me to be quite a leap of faith to assume that we can get more stable NGDP growth by (at times) deliberately aiming for less stable NGDP growth, in the hope that the policy will cause our economy to evolve in such a way as to make it easier in the future to use monetary policy to stabilize NGDP.

Obviously Stein might be correct. But the proposal is far from having the empirical and theoretical support necessary for policymakers to even consider such a policy today. It would be better to focus first on NGDP stabilization, and then see what other policies are needed once we have a track record with that policy regime.

Unfortunately, I believe that not only is the policy being considered, it is already (to some degree) in effect.

PS. Ryan Avent has a post making similar arguments, albeit much more effectively.


Comments and Sharing


CATEGORIES: Finance , Monetary Policy



COMMENTS (9 to date)
Ralph Musgrave writes:

Essentially Jeremy Stein is saying we should deny ourselves the benefits of a booming economy, in particular high employment levels, because a booming economy would cause banksters to behave irresponsibly.

I’m not surprised that someone with a well-paid job at the Fed, and who attends the same cocktail parties as Wall Street bankers says that. I suspect that everyone else (me certainly) will come to a different conclusion, namely that if banksters are a bar to full employment, then the rules governing banks need to be changed. Indeed they already are being changed, and in response to irresponsible behaviour by bankers.

Ebsim writes:

Arguably accounting for risk in financial sector is the cause of the great depression. No matter how many times they try to control financial systems all they do is ruin expectations by constantly combating the direction of the market rather giving it proper forward guidance.

Tyler Watts writes:

Isn't Stein falling for the classic blunder vis-a-vis massive failures of government bureaucracies? That is to say, the remedy for a massive bureaucratic failure is to give that bureaucracy more power and an expanded mandate. Indeed, it seems we've been through this before with the Fed itself. After the Fed's failure to prevent a massive negative money & spending shock in 1929-1933, its power was centralized and expanded with the Banking Act of 1935. Then, amid the inflationary debacle of the 1970s, the Humphrey Hawkins Act codified the "dual mandate" in 1978.

So once again, are we looking at a pattern of policy failure > expanded power... only to be repeated yet again after the next crisis?

Mark A. Sadowski writes:

"Because low rates can reflect either easy money (liquidity effect) or tight money (income, price level, expected inflation effects) the stance of monetary policy is often misidentified in empirical studies. This gives me little confidence that we can accurately estimate the impact of monetary policy on risk premiums---a key precondition for Stein's proposed policy."

Interestingly, I've been doing some estimates which seem to be highly relevant to this observation.

I recently came across the following paper by Daniel Thornton:

http://research.stlouisfed.org/publications/review/2014/q1/thornton.pdf

Thornton estimates the effect of long term one year or greater) Treasury supply on the term premium and seems to be largely an attempt to disprove prior estimates by Gagnon, Remache, Raskin and Sack (2010):

http://www.ijcb.org/journal/ijcb11q1a1.pdf

Gagnon et al regress the Kim-Wright measure of the term premium (one of the two measures discussed by Stein) on various measures of long-term Treasury supply and some control variables and conclude that Treasury supply is significantly correlated with the term premium. The implication is that QE reduces the term premium since it reduces the supply of long term Treasuries.

Thornton's main contribution is to show that the addition of trend variables renders these results statistically insignificant.

Both Gagnon et al and Thornton work with the same time period: January 1985 through June 2008. Personally I thought that it was odd that they used these results to form conclusions about the effects of QE. I wondered what the results might look like if we re-estimated them over the time period when there had actually been QE.

I use the same Kim-Wright measure of the term premium as used in both studies. I also use four control variables that both Gagnon et al and Thornton use: 1) the unemployment gap, 2) the year on year core CPI inflation rate, 3) the University of Michigan inflation expectations disagreement, and 4) the 6-month 10-year T-Note realized daily volatility.

I constructed several measures of long-term Treasury supply: 1) outstanding long term Treasuries less Fed holdings, 2) outstanding long term Treasuries less Fed holdings and less foreign official holdings, 3) outstanding long term Treasuries less Fed holdings of long term Treasuries and long term Agencies (one year or longer) and less foreign offical holdings of Treasuries, and 4) outstanding long term Treasuries less Fed and foreign official holdings of long term Treasuries and Agencies.

Consistent with Gagnon et al all supply measures were as a percent of GDP (I interpolated monthly GDP). The only thing I did different from Gagnon et al is that I did not subtract foreign official holdings of corporate bonds, which seemed minor, as well as irrelevant.

I included a trend variable like Thornton in alternate specifications. Consequently I estimated a total of eight specifications of the equation.

The period I looked at was December 2008 to August 2013.

In all of my estimates inflation disagreement and realized volatility failed to be statistically significant, in contrast to Gagnon et al and Thornton 's results. Core CPI was highly significant (at the 1% level) but was of the opposite sign as Gagnon et al and Thornton. Thus higher core inflation correlated with a lower term premium. The unemployment gap was not statistically significant in measures of Treasury supply that subtracted Fed or foreign official holdings of Agencies.

But the most interesting result was effect of the measures of supply themselves. All but one were statistically significant at the 1% level (one of the specifications including a trend variable was at the 5% level). But all were of the opposite sign as Gagnon et al. That is, a higher Treasury supply was correlated with a lower term premium.

The only specification in which the trend variable was statistically significant (at the 1% level) was in the measure of Treasury supply that only subtracted Fed holdings of long term Treasuries. Furthermore, the sign was positive, which was the opposite of Thornton's results.

The fact that Thornton's trend variable turned out not to be robust to choice of time period didn't surprise me at all. But the fact that the effect of Treasury supply on term premium is the complete opposite of Gagnon et al was a huge surprise.

Based on Granger causality tests I have previously conducted, QE raises 10-year T-Note yields. Obviously the term premium is not the same as the actual yield, but this did suggest to me that Gagnon at al's results concerning the effect of Treasury supply on the term premium might not be robust under actual QE conditions. But to obtain results that were statistically significantly and of the opposite sign was startling.

Consequently I am now very curious why no studies of the effect of QE on term premium seem to have been done during the time there has actually been QE.

dannyb2b writes:

"2. Because low rates can reflect either easy money (liquidity effect) or tight money (income, price level, expected inflation effects) the stance of monetary policy is often misidentified in empirical studies."

Instead of looking at interest rates to determine the stance of policy wouldn't it be best to just look at growth of base? NGDP growth or unemployment can show how effective monetary policy is but not really the stance. It seems to me as if there is a difference between stance and effectiveness of policy.

Scott Sumner writes:

Ralph, Well put.

Tyler, Even worse, they've misdiagnosed the problem.

Mark, Very interesting findings.

Danny, I don't like using the base, as it often reflects shifts in the demand for money. But it's not as bad as interest rates.


Andrew_FL writes:
In fact, low rates are highly correlated with low NGDP, which means it usually reflects a tight monetary policy.

Are you saying that lowering rates tightens monetary policy, or that monetary policy can be tight even if the rates are low?

Also, gauging the stance of monetary policy based on the rate of nominal GDP growth still strikes me as begging the question.

Frances Coppola writes:

So....We have denied ourselves full employment (by 1960 standards) for thirty years because of fear of inflation. Are we now to deny ourselves full employment for another thirty because of fear of financial instability? Crises leave long-term scars: folk memories of past disasters drive policy, to the detriment of current and future generations.

dannyb2b writes:

"Danny, I don't like using the base, as it often reflects shifts in the demand for money. But it's not as bad as interest rates."

The only reason the base reflects shifts in demand is because of rate targeting which is mistaken anyway.

Comments for this entry have been closed
Return to top