The standard view is that the Fed pursued an excessively easy money policy in 2002-03, which drove interest rates down to 1% and blew up the sub-prime housing bubble.  Everything about this view is wrong.  The Fed’s policy wasn’t excessively easy, it did not cause the low rates, and the low rates did not blow up the sub-prime housing bubble.

Sometimes it is necessary to go back to the basics of classical economics.  Imagine a production possibilities frontier for two goods, business investment and residential investment.  Now imagine that there was massive over-investment in some types of business investment during 1996-2000.  In 2001 it becomes apparent that over-investment in tech and communications has occurred, the floor drops out of the business investment sector, and business investment falls sharply.  What happens to the economy?  In the classical model a well-functioning economy should reallocate resources into other sectors, like housing construction.  This will happen automatically, without any government help.   All that is required is that monetary policy be stabilizing, that it keep NGDP growing at a fairly steady rate.

Is there any way to prevent reallocation into housing?  Sure, the Fed could adopt a very tight monetary policy (as in 2008 or 1930) and NGDP would fall, reducing output in almost all sectors.  But why would they want to do that?  And if they did do that then nominal interest rates would not rise, they would fall close to zero, just as in the 1930s, or as in Japan in the 1990s, or the US in 2009.

Or the Fed could run a more stable monetary policy, keeping NGDP growth up near 5%, and the economy would avoid a recession.  There would merely be a period of economic sluggishness as resources got reallocated from business investment to residential investment.  Interest rates would be slightly below normal, due to the weak business investment sector.

In fact the Fed did something in between these two extremes.  Easy enough money to keep NGDP growing a bit in 2001-02, but not easy enough to maintain 5% NGDP growth.  So we had a very mild recession, and interest rates fell, but not all the way to zero.  If money had been much tighter, rates would have fallen to zero.  If money had been much easier, rates would have been higher, as in the 1970s.  Indeed money was much easier by 2006, and rates were much higher.

So a reallocation from business investment into housing was entirely appropriate in the early 2000s if we were to avoid a depression.  Fed policy worked pretty well.  So what went wrong?  The problem was not growth, we need the economy to grow; the problem was foolish sub-prime loans.

But isn’t that a failure of the free market?  Not entirely.  Housing is not a free market.  The money being loaned out was essentially government funds (due to FDIC insurance), and thus the government needed to regulate the use of those funds to insure that banks were not taking excessive risks.  When I bought my house in 1991 you needed to put 20% down or else buy mortgage insurance.  I believe that requirement was phased out during the 1990s.  That was the cause of the sub-prime bubble, not easy money.  Money wasn’t easy.  The job of monetary policymakers is to keep NGDP growing at a low and steady rate.  The job of regulators is to correct for market failures, which includes other government policies that distort economic decision-making.  The regulators failed in the early 2000s, not the Fed.  (Of course the Fed is one of the regulators, so they are partly to blame.  When I refer to ‘the Fed’ I mean the monetary policy unit within the Fed.)

PS.  Real rates were also pretty low in the early 2000s, but the Fed has very limited control over real rates.  The low real rates reflected some combination of low business investment (post-tech bubble), or high savings rates (in Asia?)  I don’t have a theory as to what caused the low real interest rates, other than that it definitely wasn’t the Fed.  A better term would be ‘easy credit,’ not easy money.

PPS.  If it is hard to visualize how low rates might not be easy money, consider this counter-factual.  Instead of cutting rates to 1%, the Fed only cut them to 3%.  The economy does much worse, and then the Fed reacts to that much weaker economy by eventually cutting rates to near 0%.  A mere hypothetical?  No, I’ve pretty much described the 4 years after 1929, which is why the 2000 tech stock crash was not followed by a Great Depression.

PPPS.  If someone has pneumonia, and is running a fever of 102, I don’t deny that putting them in a freezer will lower the fever.  And if business investment is tanking, and resources are flowing into housing, I don’t deny that really tight money will prevent a housing boom.  I simply question the wisdom of that policy.

Recently I’ve wanted to comment on almost everything Paul Krugman posts.  Here is a recent post where he discusses nominal GDP targeting:

But it would be a big mistake to count on monetary policy alone. The zero lower bound on short rates really does matter, even if longer-term rates are positive. The Fed can control short-term interest rates, it can influence long rates “” there’s a world of difference between those two statements. So it’s not safe to assume that the Fed can, for example, hit any target for nominal GDP that it chooses.

I think Krugman is confusing two issues here:

1.  Can the Fed control actual NGDP perfectly?

2.  Can the Fed create any desired expected rate of NGDP growth?

The answer to the first question is obviously no, due to policy lags.  On the other hand there are no policy lags between changes in monetary policy and changes in expected NGDP growth.  So in the reductio ad absurdum case where the Fed is willing to buy up all the world’s assets, and incur all sorts of price risk, then I think the answer is pretty clearly yes.  And I would add that the answer is yes whether we are talking about internal Fed forecasts, or forecasts from an artificially created and subsidized NGDP futures market, or some sort of hybrid, such as having the Fed use TIPS spreads and estimates of the slope of the SRAS to generate the implied NGDP growth forecasts embedded in the market’s current inflation forecast.

The real question is whether such a policy would be so risky that it would not be politically feasible.  I say no.  If the Fed stops paying interest on reserves, and targets NGDP growth at a much higher rate than currently expected, then the real demand for base money would almost certainly be lower than today, not higher.  When you target expectations, the monetary base becomes endogenous.  So the question is not “How much money do we have to create to raise NGDP growth expectations up to the desired level?”  Rather the question is: “What is the real demand for base money if the Fed does target a much higher NGDP growth rate?”

Krugman is too pessimistic on two different levels.  In other posts he sees the currently bloated monetary base doing little to boost AD, and assumes that a much larger monetary base would be required to generate the appropriate level of nominal spending.  He is an expert on the importance of central bank credibility, but doubts whether the market would find a more aggressive NGDP target to be credible.  But if we target the forecast, then that’s not a problem.  The Fed needs to do whatever it takes to make the policy credible.  There may be some indeterminacy problems here, but they can be circumvented if you let the market forecast the instrument setting that will hit the target.  And level targeting can greatly reduce any indeterminacy.

Krugman is also too pessimistic about long term interest rates.  He wonders whether the Fed can reduce long term rates enough to get the desired nominal growth.  But that is the wrong question to ask.  The real question is “What is the equilibrium long term interest rate if NGDP is expected to grow at the target rate?”  It is very likely that much more rapid expected NGDP growth would be associated with higher nominal long term rates, and there are plausible forward-looking models where even the real long term rate would rise with monetary stimulus (due to higher expected real growth resulting from reflation.)

Krugman might argue that all of my ideas are pie in the sky, and that the Fed won’t be willing to take these sorts of radical steps.  And of course he’d be right.  The longest journey begins with a single step.  I am just trying to get people to think about these issues from a Svenssonian perspective—TARGET THE FORECAST!

Part 2:  Did Keynesian stimulus work in Asia?

Maybe, but how would we know?  Krugman says we know from this table:

In early 2009, the IMF estimated the size of stimulus programs (pdf) in G20 countries:

Ben Bernanke doesn’t think the economy needs any more monetary stimulus, even though many types of monetary stimulus are essentially costless.  Oddly, however, Bernanke does seem to think we need fiscal stimulus, even though fiscal stimulus is extremely costly, imposing huge dead-weight costs on the economy from future tax increases:

Federal Reserve chief Ben Bernanke told Congress on Thursday that the economy needs continued government stimulus spending to strengthen the recovery and reduce unemployment. But more stimulus spending would be a tough sell with congressional Republicans, who say the first round hasn’t helped enough.

This has never made any sense to me.  In the comment section of a recent post, Andy Harless provided the most plausible explanation that I have seen thus far:

Under today’s circumstances, fiscal policy is more precise than monetary policy. (It’s easy to get a vaguely reasonable estimate of the effect of a fiscal policy move, e.g. extending unemployment benefits, on AD. Much harder to estimate the effect of, say, a $200 billion purchase of 5-year T-notes by the Fed.) I think this is somehow related to the conception that most people seem to have that fiscal policy affects real output and monetary policy affects the price level. I’m not sure exactly how.

The term ‘today’s circumstances’ refers to the current monetary base setting and near-zero interest rates.  Harless is arguing that this fulfills the “other things equal” assumption required to get relatively accurate fiscal stimulus multiplier estimates.  Elsewhere I have criticized that view, noting that Fed signals about future policy intentions (exit strategies, etc) are its most powerful tool.  But as this is a minority view, I’d like to focus on some other issues.

1.  The first thing I’d say is that if the Fed believes it doesn’t have any precise tools to use, then it has no one but itself to blame.  In 1988 or 1989 I presented a paper at the New York Fed advocating the targeting of NGDP futures.  The audience assured me that the Fed already had all the tools it needed to target NGDP, and didn’t need any help from futures markets.  So how’s their interest rate targeting working out now?  Even worse, if you run out of interest rate “ammunition,” you’d presumably want to turn to QE.  Unfortunately the Fed began paying interest on reserves in October 2008, which essentially sterilized the effects of QE.  The Fed is like someone who puts on boxing gloves, and then complains that they are having trouble sewing on a button.  Take off your gloves!

2.  The Fed definitely needs to move toward level targeting, and give up on the inflation targeting approach, which has obviously failed.  Under level targeting the economy is much less likely to overshoot toward high inflation, because commodity speculators will know that if inflation rises above target, tight money will later bring it back down to the target.  Even better, since markets are forward-looking, the mere expectation of future corrective action will make overshooting much less likely (as with a credible currency peg.)

3.  But even if the Fed sticks with inflation targeting, they ought to be able to prevent overshooting.  I think many economists are still over-reacting to a very painful episode in American monetary history, 1965-1981.  As you may recall (if you are an old guy like me), the economics profession was continually surprised by unexpected increases in inflation during that period, as the trend rate rose from about 1% to over 10%.  But there are two very good reasons why that won’t happen today.  Unlike during 1965-81, we have TIPS markets which allow us to measure inflation expectations in real time.  And second, the Fed has a much greater understanding of the risks associated with letting inflation expectations drift to higher levels.

You may ask why I focus on inflation expectations, and not inflation.  After all, market expectations can be wrong.  It turns out that it isn’t just me; the Fed also focuses on inflation expectations.  The reason is that transitory movements in actual inflation, which don’t get embedded into expectations, are not very harmful to the economy.  For instance, inflation might rise temporarily because of an oil shock.  But if inflation expectations don’t rise then the temporary rise in actual inflation won’t become embedded in core inflation and wages (which respond to inflation expectations.)

If inflation starts to creep up to unacceptable levels the Fed can raise rates.  There is nothing equivalent to the zero rate bound on the high side.  And the mere fact that markets know the Fed can do this will tend to keep inflation expectations well anchored.  Of course all of this would be much easier if the Fed came up with an explicit price level, or better yet, an NGDP target.

Since I’ve responded to Harless’s comment, I’ll give him the last word.  In a different post’s comment section he expressed skepticism about whether TIPS spreads can be counted on to accurately measure inflation expectations:

Also, it risks getting whipsawed by changes in risk/liquidity premia associated with TIPS vs. nominal bonds. (I’m guessing this might be an even bigger problem with long-horizon NGDP futures, since my impression is that futures with distant settlement dates usually have low liquidity, and I can’t see a way to do NGDP futures with a settlement date before the actual horizon date.)

Once Paul Krugman has reaching a firmly held policy view, it can be costly to express a contrary opinion.  Just ask Ken Rogoff or Tyler Cowen.  On the other hand he can be very supportive to those who toe the party line:

Joe Gagnon Is Right

He calls on the Fed to implement a plan based on the ideas of someone the central bank seems to have been ignoring “” a macroeconomist by the name of Ben Bernanke.

Of course Krugman has recently been calling on the Fed to ease monetary policy, having basically given up on Congress providing more fiscal stimulus.  Gagnon called for eliminating the interest rate on reserves, and suggested that the Fed buy Treasury bonds.  There was no mention of Krugman’s favorite idea, higher inflation targets.

But what happens if you call for monetary stimulus before Krugman is ready, while he is still pinning his hopes on fiscal stimulus?  On March 1, 2009, I wrote an open letter to Krugman calling for a three-pronged attack on deflation; elimination of IOR, having the Fed buy Treasury bonds, and Krugman’s favorite idea, inflation targeting.  So he must have liked my proposal even more that Gagnon’s, right?   Ummm . . .  actually not:

A quick response to Scott Sumner OK, I see that Scott Sumner has written an open letter to me. But I’m puzzled. He writes:

“I think you have acknowledged that there is some level of quantitative easing that would boost demand. If I am not mistaken you are concerned that if such a policy boosted inflation expectations sharply, the Fed would have to quickly sell off these assets, suffering massive capital losses.”

Um, you are mistaken. I’ve never said such a thing. Did you mean to address this letter to someone else?

Very funny.  Why the negative reaction?  Here was my concluding paragraph:

To conclude, I ask you to reconsider your position on monetary policy.  If you did change your view, some people might accuse you of inconsistency. But remember what your hero once said:

“When the facts change, I change my mind “” what
do you do, sir?”

The Obama administration is obviously struggling in coming up with an effective solution to the banking crisis.  The stimulus package seems inadequate, either because (as you believe) it is too small, or (as I believe) the multiplier may be less than we think.  The economic data seems to be consistently worse than expected.  The facts have changed.

In early March, 2009, Krugman wasn’t ready to hear this message, especially from a right wing economist.   Now he is.

Part 2.  The Fedfail index

I suppose I should include something a bit less petty and self-indulgent in this post, so let’s consider Krugman’s Fedfail index:

But forecasts aside, we really have to bear in mind that the Fed is failing in fulfilling its dual mandate, price stability and full employment. I thought it might be convenient to have a simple measure of just how big the failure is; let’s call it the Fedfail Index. It’s related to the Taylor rule, but instead of offering a rule of thumb for the Fed funds rate, it measures how far unemployment and inflation are from their presumed targets.

The rule I’ve chosen takes its coefficients from the Rudebusch version of the Taylor rule; 1.3 times the deviation of unemployment from 5 percent + 2 times the deviation of core inflation (CPI) from 2 percent. So it’s 1.3* ABS(unemployment – 5) + 2* ABS(core inflation – 2). You can make up your own version; I don’t think it will look very different.

I don’t have any big problem with this index, but I prefer to use NGDP deviations from trend for two reasons.  First, I’d rather not try to estimate the natural rate of unemployment.  So I prefer NGDP (which includes real output) rather than unemployment.  Of course the closest parallel would be to use deviations from the natural rate of real output, which is equally hard to estimate.  But I favor simply targeting NGDP, as most of the problems (wrongly) attributed to inflation instability are actually associated with instability in NGDP growth.

More importantly, my Fedfail index treats supply shocks very differently from Krugman’s index.  In Krugman’s index each term is calculated as an absolute value.  Thus if a supply shock causes both unemployment and inflation to rise above normal, the Fedfail index gets worse.  But in my view that’s not really a monetary policy failure, as the Fed can’t do much about supply shocks.  So I’d prefer to use a steady 5% NGDP growth track as my benchmark for success, even if a supply shock causes inflation to rise above 2% and real growth to fall below 3%.  In a sense, Krugman’s index shows how poorly the economy is doing, not how much AD is deviating from its appropriate level.

[Note, please let me know if I misinterpreted Krugman’s index, especially the absolute value expression.]

PS.  The snarky “party line” comment wasn’t aimed at Gagnon, he’s been saying these things for a long time—indeed arguably helped created the “party line.”  Instead, it was aimed at Krugman.  And a note to Krugman-lovers:  I hope it goes without saying that the whole thing was meant to be humorous, I am quite aware that Krugman has held some of these views for quite a long time, it’s just that recently he is giving them more emphasis.

It’s been fascinating to see the topic of interest on reserves (IOR) all over the news.  I was watching a CNBC video sent to me by Mike Sandifer, and they discussed two interesting tidbits around the 5 minute mark.  First, that rumors the Fed would eliminate IOR triggered a big stock market rally the day before.  And second, they actually discussed the idea of negative interest rates on reserves.  I was gratified by the stock market reaction to the rumor that IOR would be eliminated (assuming the market was in fact reacting to this rumor, and I don’t know if it was) because market responses to news are the gold standard of causality.  Even if the stock market is wrong, it might nevertheless be right for the usual “Tinkerbell” reasons.  (If I think I can fly, then I can fly.)  If the stock market thinks a Fed action is very bullish, and rises sharply in response, that market reaction can trigger more investment even if based on an erroneous theory of monetary economics.

I was also pleased to see the elimination of interest on reserves mentioned in these two excellent posts by Joe Gagnon and Bruce Bartlett.

And finally, I few days ago I mentioned how Jim Hamilton and Robert Hall had noticed the contractionary nature of IOR quite early on.  Yesterday I ran across a David Beckworth post that was one of the earliest and best posts on the subject.  Not only did David notice the contractionary impact, but he also saw the analogy to 1937, at a time (October 2008) when the vast majority of macroeconomists were still completely oblivious to what was going on.

Monetary policy can affect either the supply or demand for money.  In the short run, monetary policy tools are generally used to change the supply of base money.  Indeed most textbooks are only able to come up with a single example of the Fed targeting the demand for base money in order to change the stance of monetary policy;  the 1936-37 decision to double reserve requirements.  Futures textbooks will almost certainly mention the October 6, 2008 IOR decision as a second, and even more foolish, example of raising the demand for base money at an inappropriate time.  Shame on the economics profession for not calling out the Fed at the time.

I like to point out that 2009 saw the biggest drop in NGDP since 1938.  But as of October 2008, the decline was still not expected to be that large.  So David’s reference to the 1937 example proved to be not only accurate, but also prophetic.  Here’s how he modestly concluded his post:

Pardon me for being skeptical, but from what I can see this approach has only encouraged banks to hoard more excess reserves. I may be missing something here–and please let me know if I am–but it seems like we are making the same policy mistakes that were made in 1936-1937.

No David, you didn’t miss anything, but the rest of the profession sure did.