The circularity problem is making me dizzy

Here is a recent story from

NEW YORK (Reuters) – Stocks rose on Tuesday after opening lower on weak economic data, with investors saying the data bolsters expectations the Fed will pump more money into the economy, which would support equities.

And what sort of economic data was weak?

September data showed U.S. consumer confidence fell to its lowest level since February, underscoring lingering worries about the strength of the economic recovery, while home prices dipped in July.

Let’s suppose consumers react with a lag to economic data, or suppose the survey was done early in the month.  In that case the survey might have reflected the very weak economic data coming out in August (revised GDP at 1.6%, etc) and also a weak stock market, which was partly a response to the weak data. 

So let me get this straight:

1.  The markets were weak in August, causing a low consumer confidence number in September

2.  This leads investors to expect more easing by the Fed

3.  This leads to a stronger stock market

4.  This will lead to a better consumer confidence number in October

5.  Which will lead investors to fear the Fed won’t ease

6. Which will cause stock prices to fall in October

7.  Which will lead to a weaker consumer confidence number in November.

8.  And so on

Are you getting dizzy yet?  This is the so-called “circularity problem,” which occurs when the Fed tries to target market expectations.  It was discussed in 1997 in a pair of JMCB papers by Garrison and White, and also Bernanke and Woodford.

The Fed needs to be careful here.  It’s easy to say the Fed doesn’t respond to the stock market; but let’s face it, they do.  They cut rates after the 1987 crash, even though there was no sign of recession or deflation, and they announced a bond purchase program in March 2009, right after a sickening plunge in equity prices.  Make all the jokes about the stock market you want, people do see it as an important indicator of which way the economy is headed.  Even if only subconsciously.

So if the Fed were to meet in November and decide not to do QE because the market was looking up, and if the market was looking up because they expected QE in response to weak economic data, then the Fed could end up with a nasty surprise.  Something like what occurred in December 2007 and January 2008, or again in September 2008 and October 2008.  Using Wall Street lingo, they could “fall behind the curve.”

Of course none of this would be a problem if the Fed used the sort of futures targeting idea I proposed back in 1987 (or similar ideas by people like Dowd, Woolsey, Jackson, etc.)  But that’s not going to happen, so they’re going to need to be very careful in evaluating market signals.  I almost broke out laughing when I read the first paragraph of that Yahoo story–it’s a near perfect example of the circularity problem.

I worry that the Fed does not fully appreciate the circularity problem.  From the WSJ:

Under the alternative approach gaining favor inside the Fed, it would announce purchases of a much smaller amount for some brief period and leave open the question of whether it would do more, a decision that would turn on how the economy is doing. This would give officials more flexibility in the face of an uncertain recovery.

.  .  .

Markets anticipate the Fed will pull the trigger, barring some surprise turn in the economy. Economists at Goldman Sachs Group Inc. estimate the Fed will end up purchasing at least another $1 trillion in securities, and estimate that would push long-term interest rates down by a further 0.25 percentage point.

A leading public proponent of a baby-step approach is James Bullard, a 20-year Fed veteran who has been president of the St. Louis Federal Reserve Bank since 2008. He says he has made progress convincing his other colleagues to seriously consider that path.

“The shock and awe approach is rarely the optimal way to conduct monetary policy,” he says. “I really do not think it is the right way to go except in really exceptional circumstances.”

These are exceptional circumstances; we’re in a Great Recession.  Aggregate demand is expected to remain far too low to allow for a robust recovery.  The only way this dynamic can be changed is if the Fed does much more than the markets currently expect.  That means shock and awe.  It’s a pity it won’t happen.

PS. I notice that changed the wording of the article I linked to.

Update:  Here is an old post that explains why the circularity problem doesn’t occur under NGDP futures targeting.



12 Responses to “The circularity problem is making me dizzy”

  1. Gravatar of OGT OGT
    30. September 2010 at 06:49

    If traders expect NGDP to rise in the future because of expected Fed actions the market rises today. If a futures market expects NGDP to be higher in the future based on expected Fed action it will rise today too. So, how does the futures market change this circularity dynamic?

  2. Gravatar of Ryan Ryan
    30. September 2010 at 06:53

    How does this theory jive with the efficient markets hypothesis? Shouldn’t the expectations of the circularity problem smooth out the price?

  3. Gravatar of ssumner ssumner
    30. September 2010 at 07:09

    OGT, You have the futures market participants determine the setting of monetary policy that produces on-target growth expectations. Check out this post:

    I should have linked to it in the post because you are right, the logic is not at all obvious.

  4. Gravatar of Leigh Caldwell Leigh Caldwell
    30. September 2010 at 07:23

    While the circularity problem is of course genuine, there is also another factor that allows the Yahoo article to make sense.

    Stock prices are not affected just by the performance of the economy but also by monetary factors. If long-term interest rates are low, the required return on equity falls, and equity prices increase.

    Because this effect provides a stable (local) equilibrium point for stock prices, it should dampen the cycles you describe and resolve the indeterminacy problem. Real effects unrelated to the Fed’s actions should also provide a similar dampening effect.

  5. Gravatar of Nick Rowe Nick Rowe
    30. September 2010 at 07:41

    With a consistent inflation targeting central bank, it used to make sense that weak data causes stocky prices to rise. Weak data is news that the natural rate of interest is lower than we thought it was. The central bank will adjust monetary policy to keep AD and earnings on target, but the lower natural rate means a higher PV of those earnings.

  6. Gravatar of Gregor Bush Gregor Bush
    30. September 2010 at 07:57

    What if the Fed used a variation of Hatzel’s idea – they announced QE and then conditioned it by saying they would continue to accelerate their purchases until 5-years TIPS spreads moved into the 2%-2.5% range? I don’t think you would have the same Bernanke-Woodford circularity problem because there is no lag between actual changes in expectations and measured changes in expectations (as there is with consumer confidence or the Consensus forecast).

    The intermediate target would essentially be a market price – the price of expected inflation. So, in principal, it would be no different than using the fed funds rate or the 10-year bond (as was the case in the 1950s) as the intermediate target.
    It may not be perfect but this idea has one clear advantage over your NGDP futures market: the market already exists and is already closely watched by the Fed. And even if they didn’t want to take the “radical” step of explicitly targeting TIPS spreads, they could signal to the market that this was effectively what they were doing by saying it in Fedspeak:

    “Medium-term expectations of inflation are below those the committee considers consistent with its dual mandate of price stability and maximum sustainable employment and are also below those that prevailed prior to the recent financial crisis. The Federal reserve will continue to provide support for the economy through asset purchases until such time as expectations of inflation over the medium term have reached levels that the committee believes are consistent with it mandate.”

    I think the market would quickly catch on.

  7. Gravatar of Liberal Roman Liberal Roman
    30. September 2010 at 13:51

    I think we had a little bit of the circularity problem today in the stock market.

    Better than expected Chicago PMI numbers caused a decline in the market.

  8. Gravatar of StatsGuy StatsGuy
    30. September 2010 at 14:32


    “Because this effect provides a stable (local) equilibrium point for stock prices, it should dampen the cycles you describe and resolve the indeterminacy problem.”

    You’re describing two effects: 1) A boost to net present value of profit flows from a firm due to a decrease in interest rates. 2) A drop in profit flows due to due to AD decline. Generally, 2) dominates 1). Among other things, profit flows from firms are leveraged (with credit and fixed assets) and highly volatile.

    In our current bizzarro world, we’ve observed negative news _lift_ stocks. That is 1) would be dominating 2)?

    Moreover, over the past 3 weeks we’ve seen stocks and treasury prices (and gold) move together. This strongly implies the market is responding to expectations of Fed action.

  9. Gravatar of scott sumner scott sumner
    30. September 2010 at 18:12

    Ryan, The EMH will give you a definitive market response, but doesn’t solve the circularity problem. The problem isn’t really markets bouncing around (although it seemed that way in my example), the problem is that the Fed doesn’t know how to control its policy instrument.

    Leigh, That’s a fair point, but I don’t see how it resolves the indeterminacy problem. The Fed still doesn’t know where to set their policy instrument. On the other hand I may well be wrong, as this isn’t my area of expertise.

    Nick Rowe, Yes, that’s a very interesting observation. I wonder if stock prices were as procyclical under the gold standard. In those days both real and nominal rates were procyclical, and also highly correlated with each other, as expected inflation was near-zero.

    Gregor, Hetzel’s idea is especially subject to the circularity problem. On the other hand this does not mean it wouldn’t work, as no one really knows how much of a problem the circularity problem really is. It might not be much of a problem in practice.

    Here’s the problem with injecting money until TIPS reach the right level. If the policy is credible, then TIPS spreads should immediately move to that level, and in that case you don’t know how much reserves to inject.

    Liberal Roman, I came in late, are you sure the PMI number didn’t cause the initial increase in the morning? You might well be right, I simply wasn’t paying enough attention.

    Statsguy, The point about the correlation with gold is a good one.

  10. Gravatar of Gregor Bush Gregor Bush
    1. October 2010 at 05:28


    But the credibility of the policy would be contingent of the Fed doing the “right” amount of QE. So the announcement would be met with an expectation that the Fed would do a certain amount of QE and TIPS spreads would move immediately to target. But if the Fed saw TIPS at the “correct” level and then failed to inject the reserves TIPS spreads would fall back down again. But this time, spreads would not move immediately back to target when the Fed tried to talk them back up. The moderate erosion in credibility would force the Fed to have to back up its words with action. This would continue until the Fed had injected the “correct” amount of reserves.

    And if the Fed’s credibility were infallible and never eroded following it failure to inject the reserves, TIPS spreads would remain on target and asset prices would reflect expectations of a higher price level and therefore as stronger recovery. So the Fed would never need to inject the reserves – the market would always know that the Fed could inject the reserves so there is no point in ever expecting below-target inflation.

    So I’m not sure I see the problem. What am I missing?

  11. Gravatar of marcus nunes marcus nunes
    1. October 2010 at 11:16

    Krugman posts a nice illustration of the Circularity Problem

  12. Gravatar of ssumner ssumner
    1. October 2010 at 18:21

    Gregor, You seem to assume it can all be done with smoke and mirrors, that you can raise NGDP w/o limit without putting any new base money into circulation. I doubt that.

    I’m not saying you are wrong, but I’d suggest you look and the Bernanke-Woodford paper, as I’m sure they could defend their hypothesis much better than I can. My sense is that they are probably technically correct, but that the issue is much less of a problem than it might seem.

    Marcus, Thanks, that’s a funny example.

Leave a Reply