Scott Sumner  

Is the economy becoming less interest sensitive? And does it matter?

PRINT
A Bet with Rick Moran... Slaying the Myths about Plasti...

Answers: Maybe, and no.

Tyler Cowen linked to a recent paper by Jonathan L. Willis and Guangye Cao, out of the Federal Reserve of Kansas City. Here's the opening paragraph:

Over the past three decades, the U.S. economy seems to have become less responsive to monetary policy. Slow recoveries followed recessions in 1990-91, 2001, and 2007-09, a contrast to the much more rapid recoveries that followed pre-1990 recessions. These slow recoveries occurred despite sizeable monetary accommodation from the Federal Reserve, primarily through reductions in short-term interest rates.

At first glance that looks like reasoning from a price change. But they later provide a standard 3-equation model (IS, Phillips Curve, monetary policy function) and thus I'll assume that what's really going on here is that the authors are assuming the economy responds to shocks to the gap between the policy rate (fed funds rate) and the Wicksellian equilibrium rate. That's a plausible model. And perhaps the economy is less responsive to that gap than before.

Even in that case, however, I have some reservations. There should be no surprise that the recovery from recent recessions has been slower than from earlier recessions such as 1981-82. If it does appear to be a surprise, that's probably because monetary policy is misidentified. Monetary policy has clearly been less expansionary during recent recessions, and that's why the recoveries have been slower. I use the Bernanke/Sumner technique for identifying the stance of monetary policy, looking at NGDP growth/inflation, not interest rates.

I see a couple potential problems with this sort of research. Those with a better understanding of modern macro should please correct me if I'm wrong:

1. The authors find a lower level of responsiveness to interest rates after 1985. Does that date ring a bell? One of the most important and surprising developments in modern macro has been the astounding decline in the Wicksellian equilibrium interest rate over the past three decades. We can't measure that rate with absolute precision, but it has clearly fallen very dramatically. Now think about how that might confuse policymakers. You cut rates the standard X% during a recession. You expect a certain response, assuming that you have not accounted for a long term decline in the Wicksellian equilibrium rate. But in fact, your policy is less expansionary than you assume, as the cut in interest rates is occurring against the backdrop of a declining Wicksellian equilibrium rate.

2. Let's suppose I'm right, and that expected future NGDP is the best measure of the stance of monetary policy. Then during a recession, when expected future NGDP is also falling, it's likely that the Wicksellian equilibrium rate is falling faster than the policy rate. So in that case money is actually getting tighter. The Fed sees its interest rate cuts in 2008 and 2009 as rescuing us from a recession caused by bad bankers. I see the Fed as having caused that recession through tight money. During 2008 the Fed cut rates more slowly than the Wicksellian equilibrium rate was falling, and hence money got tighter. BTW, as far as I know I've just described the Milton Friedman/Ben Bernanke interpretation of 1930-31. Is it so far-fetched to think that it might also apply to 2008?

On page 20 the authors say:

In comparison, the negative response of the 10-year yield to an unexpected decline in the federal funds rate is significant much earlier in the pre-1985 period than in the post-1984 period. This suggests that the transmission of monetary policy from short-term to longer-term interest rates occurs with a longer lag in the post-1984 period. This evidence is also in line with comments made by then-Chairman Greenspan in 2004 regarding the "conundrum" of longer-term yields not initially moving in the same direction as short-term rates.

Market monetarists would have a very different interpretation. First note that on some occasions long-term rates move in the opposite direction from a surprise move in short term rates. But why do they often move in the same direction? The authors imply that a cut in short-term rates gets "transmitted" to longer-term rates. I think that's wrong. A truly easy money policy would cause long-term rates to rise, as we saw in the 1960s and 1970s. So what explains the correlation? Suppose AD falls due to some previous error of monetary policy. We start to slide into recession. The Fed cuts rates, but not enough to keep monetary policy from tightening. So short-term rates fall because the Fed is cutting them because there is a recession. And long-term rates fall because there is a recession (because they didn't cut rates enough). The Wicksellian equilibrium rate is expected to be lower over the next few years, and that leads bond markets to reduce yields on the 10-year. Thus it wouldn't surprise me at all to see short and long-term rates move in the same direction, as both are responding to the same fundamentals. As a broad generalization, monetary policy is usually too tight when rates are falling, and too easy when rates are rising. Not always (and perhaps not this September) but on average.

What makes the market monetarist explanation better is that we can also explain why long rates sometimes move perversely, the so-called "conundrum." In January 2001, and September 2007, and December 2007 the long rate moved in the opposite direction from what conventional economists would have expected, given the policy surprise. Easier than expected policy made long-term rates rise (Jan. 2001 and Sept. 2007) and tighter than expected money made long-term rates fall (Dec. 2007.)

To summarize, it's possible that the economy is becoming less responsive to gaps between the market and Wicksellian equilibrium rates, but it's very difficult to prove that. Fortunately, even if the economy is becoming less sensitive to interest rate gaps, it's not becoming less sensitive to monetary policy. Shocks to expected future NGDP impact RGDP just as much as in the old days. The recession of 2009 was about as deep as one would expect, given the amount by which monetary policy reduced NGDP. In that most important respect, monetary policy is just as effective as it's ever been.


Comments and Sharing






COMMENTS (10 to date)
Baconbacon writes:
If it does appear to be a surprise, that's probably because monetary policy is misidentified. Monetary policy has clearly been less expansionary during recent recessions, and that's why the recoveries have been slower.

Q: How do we know monetary policy was less expansionary?

A: Because the recovery was slower than usual.

Q: Why was the recovery slower than usual?

A: Because monetary policy was less expansionary.

Q: How could we show that the relationship didn't hold?

A: By finding a period of economic expansion that wasn't economic expansion.

Q: What statistics are acceptable for describing such a scenario?

A: The UE rate, specifically U3

Q: Nothing else? What about U6? LFPR?

A: No, those won't do.

Q: So as long as U3 is near its "natural" rate then it doesn't matter that U6 has not come down at the same rate U3 did? It doesn't matter that disabled worker status has increased by 80% since 2000? That applications are up an average of 5-600,000 a year since 2007 despite ~300,000 people being added per year to the rolls? It doesn't matter that LFPR has declined by 3 percentage points during that time? It doesn't matter that three other measures of the labor force are indicating slack?

Kevin Erdmann writes:

It seems as though the mm argument would be greatly enhanced by regularly using forward interest rates instead of Treasury yields. The 10 year bond includes a bunch of short term rates that confuse the matter when they are moving apart from each other.

bill writes:

I was always disturbed when Greenspan would call it a conundrum. Even without knowing MM it just seemed obvious that if the Fed was operating correctly, long term Treasury rates should be somewhat stable.

One large concern of mine now, and why this quarter point matters so much, is that the Fed seems hyper concerned with appearing perfect. If it raises a quarter and the economy starts slowing, they won't loosen again until it's way too obvious (and late) that they made a mistake. They seem more concerned with providing a stable path of short term interest rates than they are with a stable NGDP path.

Clearly an alternative path for the Fed would be to stay loose until NGDP growth exceeded their (as yet unmentioned) NGDP target and then move to tighten faster. But they'd be too embarrassed to move 50 or 100 bps at a time and have someone say "the Fed is behind the curve".

Lastly, since we don't have NGDP futures prices, we can look at other market prices. And as long as the yield on 10 year Treasuries is this low, it's clear that policy is still too tight. At least if the Fed wants 4% NGDP growth or more. Since they won't say, maybe policy is "just right".

Scott Sumner writes:

Bacon, Not sure what that's all about . . .

Kevin, Even better, skip interest rates entirely and focus on NGDP growth.

Bill, Excellent comment.

baconbacon writes:

What type of evidence would it take to convince you that attempting to target nGDP is a bad idea?

Jeff writes:

@Scott,

Implicit in what you're saying is a rule that could be used right now. The Fed should cut both the funds rate and the IOER rate to zero and then start unsterilized buying of foreign currencies until the nominal long rate starts to rise. (No need for quantitative easing when the FX market exists.)

Unless, of course, the Fed is happy with what the market is currently expecting.


E. Harding writes:

I think long-term nominal (and real) interest rates have declined due to a smaller risk of unanticipated inflation, and short-term rates have little effect on long-term rates in the short run due to monetary policy becoming easier to anticipate. So if the economy is becoming less interest-sensitive, it's because interest rates have less room to move due to low inflation/NGDP growth.

baconbacon writes:

Perhaps this is a better track (an abbreviated and not really accurate history)

The Fed started with a gold standard (or a "gold standard")- which "worked" for a while and then suddenly didn't work at all. Then they targeted the idea of the Phillips curve, which looked good and then suddenly didn't work at all. Then they decided interest rates were the way to go, which looked good and then suddenly didn't work at all.

The baseline hypothesis for basically everything else in the economy is if the government (or a large monopoly agent if you prefer) targets X that it will cause massive distortions, and that you generally expect the governments job to get harder over time. If they target banana production and their target is high then more and more marginal land must be turned to producing bananas and the government will have to pay more and more in subsidies to produce those bananas, if they set it to low they will have to pay more to keep bananas off the black market. Either way each year enforcement becomes more costly than the last under most conditions.

This paper is describing a situation in which managing the economy by targeting interest rates has become harder and harder as the policy went on. The strongest argument against nGDP targeting is that once you start the relationship between nGDP and rGDP will fall apart. The relationship between U3 and other employment statistics (LFPR, disability claims, U6) has arguably weakened while the Fed has been focused on returning the economy back to appropriate levels of U3.

Is this paper correct? Maybe. Does it have implications for nGDP targeting? If true it definitely carries concerns for monetarists.

John Hall writes:

This isn't exactly related to what you've written.

I did two regressions. The variables were the same, just over two different time periods (1940s to 1985 and 1985 to present). The regressions were log changes in private fixed investment against four of its lags, four lags of log changes in GDP, and some interest rate variables (10 year and 3 month currently, and a variety of lagged changes). I decided to regress investment instead of GDP because it is more obviously the component of GDP that should be interest rate sensitive.

The interesting thing is that there is some fairly different behavior in terms of the coefficients. However, the R^2 really isn't that different. It's like 50% for the first period and 49% in the second.

In the first period, the sum of the coefficients on lagged investment is about -0.1, while in the second it is 0.8. The sum of the coefficients on the GDP lags is 1.35 in the first one and -0.4 in the second. These are wildly different. So while the first period might have higher interest rate sensitivity, the second period has offset this with much higher autoregressive behavior.

Roger McKinney writes:

The economy has never been as sensitive to interest rates as mainstream economists assume. Hayek pointed out in 1938 in his Profits, Interest and Investment that businessmen mostly ignore interest rates and fixate on profits to guide investment decisions. He also showed that recessions are inevitable even if the Fed pigs out on ZIRP!

Comments for this entry have been closed
Return to top