I’ve been asked to summarize my views on liquidity traps in one place, so brace yourself for a long post. (Longtime readers will definitely want to skip this one.)
For simplicity, I’ll define the term ‘liquidity trap’ as a situation where a fiat money central bank with a freely floating currency is unable to boost nominal spending because nominal interest rates have fallen to zero. There may be some cases where central banks are limited by laws regulating the sorts of assets they are allowed to purchase, but I know of no real world cases where that was a determining factor. Indeed I know of no case where a central bank that wished to boost inflation and/or NGDP was unable to do so. Nor do I think we need ever worry about that scenario actually occurring.
On the other hand, I do think the zero rate bound is a real problem for real world central banks. Because central banks are used to using short term rates as their primary policy tool, policy may well become sub-optimal once rates hit zero. But that would not be because an economy is “trapped” at a zero bound, rather it is because central banks are reluctant to aggressively use alternative policy tools, including tools that would be much superior to fed funds targeting even if the economy were not up against the zero bound.
Part 1. Basic monetary framework
Unlike most economists, I don’t believe that changes in short term interest rates play an important role in the monetary transmission mechanism. The liquidity effect is an epiphenomenon, having little impact on investment. Woodford argues that what really matters is changes in the expected future path of interest rates. I agree that policy expectations are a key, but find it more useful to think in terms of changes in the expected path of the supply and demand for base money. Simply put, I believe that current and expected future increases in base supply relative demand cause expected future NGDP to increase. This is because even if we are at the zero bound, and cash and T-bills are perfect substitutes, we are not expected to be there forever.
Increases in expected future NGDP (my preferred policy indicator) raise current asset prices (foreign exchange, stocks, commodities, commercial real estate, etc.) Because wages are sticky in the short run, production of corporate fixed assets, exports, commercial and residential real estate, commodities, etc, increase as their prices increase. The resulting higher real incomes also boost consumption. The reverse is true during tight money, as in late 2008.
I don’t like the interest rate transmission mechanism because interest rates often move in the “wrong” direction in response to monetary surprises. An unexpectedly small cut in the fed funds target in December 2007 sharply depressed equity prices at 2:15pm. The fed funds futures market confirmed that the decrease was smaller than expected. Keynesian theory says T-bond yields should have risen on the news. Instead, yields fell from 3 months to 30 years, as investors (correctly) understood that the Fed’s pathetic response to the sub-prime crisis would slow economic growth, and hence future fed funds rates would have to be cut sharply. (And they were in January 2008.) The action slowed the economy, but not because interest rates rose.
The powerful monetary stimulus of 1933 (dollar depreciation) had little effect on interest rates or the current money supply, but sharply raised future expected NGDP. This sharply raised current asset prices, and led to rapid growth in output.
If you buy my argument that changes in expected future NGDP (what Keynes probably meant by “business confidence”) is driving current asset prices and aggregate demand, then the next question is whether monetary policy can influence future expected NGDP at the zero bound.
Part 2. Unconventional policy tools.
My favorite example of an unconventional policy tool is the 1933 dollar devaluation. In 1932 the Fed had tried open market purchases to boost the money supply, but the policy failed as fears the US would be forced to devalue led to gold outflows, which negated most of the effect of the asset purchases. This is the only example of a liquidity trap cited in the General Theory. Unfortunately, Keynes confused two closely related problems. A liquidity trap is where an increase in the money supply fails to boost NGDP. In 1932 the constraints of the international gold standard meant that purchases of assets failed to substantially increase the money supply. That’s gold standard economics 101, having nothing to do with liquidity traps. As soon as we left the gold standard in March-April 1933, FDR was able to easily create rapid inflation despite 25% unemployment, near zero T-bill yields, and much of the banking system shutdown for many months.
FDR’s policy of raising the price of gold can be seen in one of two ways. In one sense it was a devaluation of the dollar in the forex markets, as most countries did not follow his action by raising their purchase price of gold. But even if the US had been a closed economy the Fed could have depreciated our currency by reducing the weight of gold in one dollar. In ancient times this was called “debasing coinage” and no one worried about the zero rate bound preventing central banks from inflating. As far back as 1694 John Locke used a reductio ad absurdum argument to criticize monetary ineffectiveness claims.
Another approach is to do quantitative easing. But printing money (even under a fiat money regime) will not be very effective unless the currency injections are expected to be permanent. Why would people bid up asset prices if the central bank was expected to pull the money out of circulation in the near future? This is why the QE done in Japan around 2003 did not do much, and it is why Paul Krugman is skeptical about QE. The other problem is that it is hard to make a credible promise to permanently increase the money supply by X%, because once you exit the zero rate bound the velocity of base money will rise sharply, and unless the base is reduced you will get hyperinflation. Markets know this, and hence don’t expect QE to be permanent.
The solution is to adopt an explicit nominal target such as the price level, or better yet NGDP, and then do level targeting. This is essentially a promise by the central bank to leave enough base money in circulation long term to allow for a modestly higher price level of NGDP. For instance, they might want to target 5% NGDP growth. Even if we are at the zero bound and monetary policy appears to be spinning its wheels, a commitment to higher future NGDP will tend to raise current AD.
The Fed made two mistakes. First, they did not engage in level targeting. It has long been understood that once nominal rates hit zero the central bank must adopt a level target. Indeed Bernanke lectured the Japanese on exactly this point back in 2003. So in September 2008 the Fed should have switched to level targeting, indicating that they wanted core inflation to grow along a 2% path until we were out of the recession, promising to later make up for any near-term shortfalls. Instead they allowed core inflation to fall well below 2%, and then (implicitly) indicated that they were going to continue inflation targeting, allowing bygones to be bygones. There was to be no above 2% inflation to catch up for the shortfall. This actually made their job much more difficult, as it resulted in a more severe recession than necessary in 2009, and then plunging investment pushed the Wicksellian equilibrium nominal rate below zero. They could no longer use their traditional policy instrument and they were reluctant to aggressively employ alternative measures, because they didn’t know how strong the effect would be. For instance, when banks needed more liquidity in late 2008 the Fed neutralized the effects of the large monetary base injections by paying interest on reserves at a rate higher than T-bill yields. Only when the recession drove stock prices to extremely low levels in March 2009, and deflation appeared on the horizon, was the Fed willing to do QE1. And only when the recovery faltered during mid-2010 (as European troubles increased the value of the dollar) was the Fed willing to do QE2.
The best way to avoid the zero rate bound is to create and subsidize trading in a price level or NGDP futures market, target the futures price, and let the money supply and interest rates respond endogenously. The Fed should be willing to supply an unlimited amount of reserves in order to keep NGDP futures prices rising along a 5% growth trajectory. Because there is no zero bound for NGDP futures prices, the Fed will always be able to keep NGDP expectations on target.
The flaw in the Keynesian model is that it assumes sticky wage and prices, whereas only T-bond prices are flexible. But there are lots of other asset prices that are also flexible, and that don’t have a zero lower bound. These include commodities like gold and silver, stocks, and foreign exchange. Unfortunately, all of those asset prices have drawbacks as targets for a major central bank like the Fed. And the asset price that would work best (NGDP futures prices) doesn’t yet exist.
[BTW, it's a disgrace that the government has not yet set up and subsidized trading in a NGDP futures market. Contrary to popular impression the Fed isn't trying to create more inflation; they are trying to create more NGDP. For any given increase in NGDP, the Fed would actually prefer less inflation and more RGDP growth. We desperately need a real time measure of market NGDP growth expectations in order to know whether AD is likely to exceed or fall short of the target.]
In this imperfect world the best the Fed can do is to focus on TIPS spreads as a crude measure of expected inflation, and a whole range of indicators for expected RGDP growth, such as the relative prices of stocks and commodities, as well as other indicators or real output trends. The Fed needs to do enough QE to increase expected NGDP growth (using all these imperfect indicators) up to the desired level. Since we are below trend, they should probably target slightly above 5% NGDP growth for a few years, then 5% thereafter.
Part 3. Fallacious arguments in favor of the liquidity trap
There are so many, I hardly know where to begin. One common argument is that swapping cash for zero interest T-bills is useless, because they are perfect substitutes. I don’t view them as perfect substitutes at all. When I get in the car to go shopping at Walmart I don’t think “Hmmm, should I take cash or T-bills.” At this point people will say “Yes, but zero interest bank reserves and T-bills are near perfect substitutes. And all the recent base injections are going into excess reserves.” Yes, but there is no zero lower bound on interest paid on reserves (and yes I’m including vault cash in “reserves.”)
But let’s suppose cash and T-bills were perfect substitutes. Even in that case a permanent injection of new base money would still be expected to raise the future level of NGDP, as liquidity traps don’t last forever. (If they did we should legalize counterfeiting.) Yes, a temporary currency injection wouldn’t do anything, but that’s almost equally true when rates on T-bills are positive. Temporary currency injections don’t matter, permanent ones do. It makes little difference whether rates are at zero or not.
The second fallacious argument is that monetary policy was ineffective in the Great Depression. Actually, when the government got serious about inflating they left the gold standard, and then they had no difficulty in raising prices sharply.
The third fallacious argument is that monetary stimulus would not be effective at the zero bound because central banks are conservative and no one would believe their promises to inflate. In fact, no one can provide an example of a central bank that tried to inflate but failed because they were stuck in a liquidity trap. Some cite Japan, but that example doesn’t meet any of the criteria for a liquidity trap:
1. The Bank of Japan has frequently expressed opposition to a positive inflation target. Because they are not trying to produce inflation, it’s no surprise they have failed to produce inflation.
2. It’s true that they pay lip service to avoiding deflation, but every time the inflation rate rises above zero percent they tighten monetary policy and go right back into deflation.
3. Some point to the large QE the BOJ did around 2003. But their promise to keep prices stable meant the QE was going to be temporary. The public knew this and quite rationally refused to bid up prices. Sure enough, when 1% inflation threatened to rear its ugly head in 2006, they promptly reduced the monetary base by 20%.
4. They passively sat by and allowed the yen to appreciate strongly, even as deflation was accelerating in recent years.
5. If it walks like a duck . . .
Some point to the alleged failure of the Fed to inflate, despite trying hard. Yet a few months back when Brad DeLong asked Bernanke why the Fed didn’t adopt a 3% inflation target, Bernanke said that would be a horrible idea. The Fed had worked so hard to bring inflation down to low levels. If you heard an answer like that, would you expect the Fed to produce higher inflation? It’s no surprise inflation expectations have remained low. Admittedly the Fed doesn’t want deflation either. My view is that when core inflation falls to about 1%, warning bells go off and the Fed grudgingly does some QE to boost inflation a bit closer to 2%. If that’s not what they are trying to do, I’d love to know their policy goal.
There’s another problem with the view that QE is ineffective at the zero bound. Even if the Fed couldn’t reduce nominal rates, they could always reduce real rates. Indeed Mishkin’s textbook suggests about 10 different transmission mechanisms other than nominal rates. Yet liquidity trap proponents ignore them all. Even worse, when the mechanisms are shown to work they go into denial, asserting that it just can’t be true because their theory says it’s impossible. So for instance during September and October there were more and more rumors of aggressive QE (and possibly even level targeting) emanating from various Fed officials. Mishkin’s text says this should boost stock prices, it should depreciate the dollar, it should raise commodity prices, it should raise inflation expectations in the TIPS markets, and it should lower real interest rates. And all of those things happened. For years Paul Krugman has been arguing that what the Fed really needed to do was to raise inflation expectation. Well they did it. And his response seemed to be incredulity, as if the markets were nuts in thinking QE could actually raise inflation expectations.
Although Krugman and Robert Barro are poles apart ideologically, they both suffer from one weakness–relying too much on what their models tell them. Both expressed skepticism about whether QE would do very much, because they looked at QE from a mechanistic perspective. But QE is much more than that; it is an implicit commitment by the Fed to seek (slightly) higher inflation. Of course they need to do much more, but they did succeed in terms of their very conservative goals. They did generate about 0.5% higher inflation expectations over 5 years. The reason QE worked was not the operation itself (which I agree does little or nothing) but rather because it tells the market that the Fed is now more serious about boosting long term prices and NGDP. In other words they are now willing to leave that base money out there long enough to get closer to their implicit inflation target. The Fed was afraid to directly call for higher inflation (something Krugman thinks could work) so they spoke in code, hoping the markets would understand them but Sarah Palin would not. Unfortunately for Bernanke, Sarah Palin has advisers who know exactly what the Fed was up to.
Part 4. Reductio ad absurdum arguments
I can’t take anyone serious who actually believes in a complete liquidity trap–i.e. that no amount of monetary base injections would be inflationary. Taken literally, that would imply the Fed could buy up all of Planet Earth without creating any inflation. Among serious economists the debate is over magnitudes. The skeptics will say that the amount of QE required would be unacceptable, it would expose the Fed to excessive risks if they later had to resell assets in order to prevent runaway inflation.
In fact, there are all sorts of reasons why this “risk” argument is bogus. First of all, the high base demand is itself a product of the Fed’s contractionary policies, which allowed NGDP to fall in 2009 at the sharpest rate since 1938. That’s why banks hoard reserves. A much more aggressive monetary policy would mean less real demand for base money. Second, the demand for base money has been artificially bloated by the IOR policy; the public is not hoarding much cash and they certainly would not do so if the Fed set a much higher inflation or NGDP target path. Most importantly, any capital losses suffered by the Fed would be tiny compared to the gains the Treasury would get from much faster NGDP growth. Remember that the big drop in NGDP is the number one reason the deficit ballooned in 2009—more important than even the fiscal stimulus. Furthermore, many of the Fed’s purchases have been medium term T-notes, for which price risk is not that significant. If people are actually worried about this issue, the Fed could buy equities and foreign bonds, which would appreciate with an expansionary monetary policy. But in my view those (controversial) steps would not be necessary, as the risks are greatly overblown.
Maybe I should stop there—I feel like I am beating a dead horse. Does anyone still believe in liquidity traps? Is there even anyone still reading this post?