The subtleties of the liquidity traps

One of the older definitions of the liquidity trap is that excess cash balances get absorbed into hoards.  That clearly isn’t going on today because sales are booming.   The unwillingness of people with liquid assets to invest is another common aspect of the traditional definition; that’s not the case today either, even though we might wish investment were stronger.

Is there a “trap” at one portfolio margin or at all margins fanning out from cash balances?  It makes a big, big difference.  Today there is a trap at the money-bonds margin (and even then not all bonds), but not at most other margins.  Not the money-goods margin, for instance.

Whether we are in a liquidity trap is not always an either/or proposition.  An economy could face some zero bound constraints on particular kinds of monetary policy without a full liquidity trap model applying across the board.

It makes a big difference how many margins are covered by “traps.”  When there is a bottomless demand for money hoards, and all margins encounter traps, the correct model does seem to invert the slope of its AD and AS curves, as Krugman sometimes suggests.  Empirically, that is not the case today and probably never has been the case.

When there is a trap only at the money-bonds margin, it is a mistake — both empirically and theoretically — to invert the slopes of the AD and AS curves.

Like most econbloggers, I tend to favor simple models.  Yet a more complex model — with more disaggregation, with explicit distinctions between local and global properties, and with explicit stability conditions, may be needed to show this mistake.  (This is one example of where “high theory” really does come in handy.  In its place, offering casual remarks about the weakness of the real balance effect is asking the wrong question and does not fill the gap.)  In the meantime common sense and empirics suffice to reject it.

When there is a trap only at the money-bonds margin, positive aggregate supply shocks tend to be expansionary.  They generate money income, jobs,  and boost real rates of return.  It is a slow and painful way to recover from a recession, but indeed it is a way to recover and it’s what we have been doing.

Any claim about the duration of a liquidity trap has an implicit dual in real business cycle theory, about the enduring weakness of real rates of return.  For all the brickbats thrown at rbc, the liquidity trap proponents are relying on it, while keeping real rates of return in the background of their arguments.  A booming economy on the real side gets out of a one-margin liquidity trap much more quickly than does a stagnating economy.

If one wishes to do exegesis of Krugman’s liquidity trap posts, he often makes the common sense and indeed indisputable observation about a liquidity trap at one margin, money-bonds.  Very often he then draws sensible conclusions from this starting point, such as the need to analyze the credibility of monetary policy.  Other times (pdf) he slips into writing as if all margins were covered by traps and the AD and AS curves invert their slopes.  As I’ve mentioned, that is simply a mistake and the extreme properties of those models do not follow from a world where the trap is at a small number of margins only.

Arnold Kling makes some related observations and also serves up the requisite links.


Unfortunately, I am one of the those young economists who has never learned Keynesian economics, and as such I'm unable to interpret comments about "AD" and "AS" diagrams. (I looked them up, and while they seem to be able to capture certain glimmers of intuition, they're mostly just confusing.)

I have long been a little bewildered by your comments on the "money-bonds" margin. Tautologically it's true that this is "just one" margin, but it's a very, very important one: holding real factors constant, it determines the baseline, short-term nominal interest rate in the economy. A sufficiently rich policy rule for the nominal interest rate, meanwhile, captures almost everything we need to know about conventional monetary policy. I agree that the Fed can act along other margins -- that's what we've seen in QE1 and QE2 -- but to a first approximation, its policy rule (both present and forward-looking) is what pins down nominal variables. QE1 and QE2 are best interpreted as either (1) signals of future monetary policy, (2) attempts to lower long-term yields slightly through portfolio balance effects, or (3) in the case of QE1, a program to increase the public supply of liquidity in markets.

I have the lingering sense that you're trying to make some very complex point that I just can't understand. Perhaps, then, we should make this discussion more concrete. Is there some policy that the Fed could follow where you and Krugman would make strikingly different predictions, thanks to this disagreement?

Is your point that the Fed can "always" manufacture inflation, by printing enough money? (Of course, the key question here is how to define "enough".) Do you dispute the New Keynesian notion that the Fed needs to "commit to be irresponsible" -- at least if "irresponsible" is defined by departure from its implicit long-term policy rule -- or do you simply think that committing to be irresponsible isn't so hard?

Young Economist, whether "boosting wages" is a good idea will draw the policy split you are looking for.

Paul Krugman wrote about this in 1936.

By the way, could you please explain what the "money-goods margin" is? I am not quite sure what it is supposed to mean. If it's literally the margin between "money" and "goods", I assume that you're asking how the Fed can alter the relative price between money and goods -- in other words, you're just talking about the price level. If so, I can't say I understand your analysis: New Keynesian economics already has a fully-fledged theory of price determination that implicitly rests on the "money/goods" margin. Where are you departing from that?

Or, perhaps, you're talking about how the demand for goods changes when the supply of money increases? Even there, however, we shouldn't be talking about a "money/goods margin"; rather, since we've exhausted the money/bonds margin and those two assets have become equivalent we should be discussing a "money + bonds/goods" margin, or just a "liquid wealth/goods" margin. I agree that increasing the supply of liquid wealth, as in QE1 (and perhaps QE2, depending on your definition of liquidity), can have real effects, but this is essentially fiscal policy rather than monetary policy, even if it's undertaken by the Fed. And if so, it rests on a very different class of models: rather than talking about AD and AS diagrams, we should be poring through Holmstrom and Tirole.

Perhaps you're trying to say that real money balances and consumption are complements, so that (1) there isn't really a liquidity trap and (2) monetary policy matters for the real economy through more than its effect on price-setting and the real interest rate. This would be a distinct and interesting channel through which monetary policy operates. But if this is really your claim, I urge you to think about the relevant wedges. Real money balances enter the consumer's optimization decision with a price: the nominal interest rate. That price governs the extent to which consumers are (inefficiently, at least in a simple model) allocating their wealth to bonds rather than money that would be complementary to their consumption. But if the nominal interest rate is near-zero, the size of this wedge is near-zero as well. Maybe we'd get *some* effect through this channel from bringing the nominal interest rate even lower, but it doesn't strike me as being of first-order importance.

"Young Economist, whether “boosting wages” is a good idea will draw the policy split you are looking for."

Thanks. To make a concrete policy question, perhaps we should think about "boosting wages" as "raising the payroll tax rate and then rebating the proceeds lump-sum to households". If so, I agree that there is probably some disagreement between you and Krugman, though I am still not sure where exactly the difference in models arises.

My interpretation is that there are two effects of such a policy in a liquidity trap, one positive and one negative. The "positive" effect is that an increase in real wages (from firms' perspective) would raise inflation expectations, shrinking the gap between the real interest rate and its natural level and helping to relieve the liquidity trap. The negative effect, of course, is that it would increase unemployment.

I'm not sure how exactly these two effects compare, quantitatively. The great weakness of modern macroeconomics is the lack of any truly convincing model of an extensive margin in the labor market -- at least one that has been well integrated into the main framework. From a qualitative perspective, it's possible that either effect would dominate. Should I interpret your difference of view with Krugman here as (1) a belief that the extensive margin response to a change in real wages is quite strong, so that the negative effects here outweigh the positive, or (2) a belief that the "liquidity trap" logic is fundamentally flawed, and that the "positive" effect doesn't really exist?

I might subscribe to (1); for that matter, so might Krugman. (His arguments against a payroll tax cut seem to imply that he believes that he views the liquidity-trap effects matter more than the real-wage effects, or else that there is some unknown asymmetry between payroll tax cuts and tax increases. But it's possible that a reasonable model of the extensive margin of the labor market *would* contain such an asymmetry, and I don't think this position can be dismissed so easily.)

Since you don't like the idea of a "liquidity trap" in general, however, I think you might believe in (2). And then my question still stands: Where is the difference in mental model between you and Krugman that makes you skeptical of the (beneficial) inflationary effects of a payroll tax hike in a liquidity trap? Do you believe, a la Sumner, that the Fed can credibly promise to do just about anything if it tries hard enough (and, therefore, that a payroll tax hike is a needlessly painful way to accomplish something that the Fed could do anyway)? Or, the Fed's credibility aside, do you think that Krugman's model is fundamentally incorrect in some other way?

What's with the obsession with "liquidity traps"? It seems overrepresented in MR posts recently. Is it objectively that important an economic concept or just a pet obsession for Tyler?

I'd say it's Krugman's pet obsession not Tyler's and that Tyler is offering a response. If it's as real or meaningful as Krugman claims, it's a case for more stimulus.

"That clearly isn’t going on today because sales are booming. "

My head exploded at this point.

Why? Retail sales are at an all-time high.

I was also surprised. Maybe my memory is hazy, but I had the opposite impression from reading posts at Modeled Behavior. Someone should give a link on sales numbers.

Tyler and Young Economist,
I will not spend time going through all the historical and analytical details of the idea of a liquidity trap. My intention is only to make clear why this idea is irrelevant. Today central banks increase the supply of currency either by buying tradable bonds or by financing directly government expenditures (in the latter the central bank may receive an IOU but it's not tradable). In the first case, a central bank must define which bonds will buy --say bonds X-- and indeed anytime that it buys bonds X their price increases. In some extreme cases, the increase in the price of X may be negligible and therefore its rate of return does not decline. In these extreme cases one may say that currency and X have become close substitutes but it is irrelevant because we know that it is due to some special circumstances that we are hardly able to determine. They may be so special that even if the central bank buys a large amount of X (say equivalent to 50% of the outstanding stock of currency), the price of X will increase little. Indeed, in these special circumstances, we can suggest the central bank to buy bonds Y and the price of Y may increase, but we cannot know whether the increase will be significantly larger than the increase in the price of X. The central bank can experiment with different types of bonds and it may find one type for which an increase in the supply of currency leads to a significant increase in its price in the special circumstances in which the prices of X and Y increase little. The only problem is that this type of bond most likely will make no difference in terms of increasing expenditure on current output --neither theory nor evidence can help the central bank to identify a type of bond that may bring about a significant increase in expenditure. Thus, we may conclude that there are extraordinary circumstances in which monetary policy is not effective but we don't know which they are.
That is why the second mechanism to increase the supply of currency, that is, to finance directly an increase in government expenditure on current output, is more interesting. The only problem is that in this case there is no difference between fiscal and monetary policies. You can claim that fiscal policy is effective but you cannot say that monetary policy is not effective. Indeed, the effectiveness of fiscal policy depends on how the money is spent --for example, the more is spent in increasing the incomes of public-sector employees, pensioners and welfare recipients, the lower the effect may be. This example is directly related to Friedman's helicopter, that is, to increase the supply of currency by giving it away to people that may spend all or part of it on current output. To argue that they will hoard all the money means that the central bank distributed confetti --regardless of how extraordinary the circumstances are.

Since the recent bond purchase program began bond prices have moved down not up. How should that be interpreted?

" Indeed, the effectiveness of fiscal policy depends on how the money is spent –for example, the more is spent in increasing the incomes of public-sector employees, pensioners and welfare recipients, the lower the effect may be. "

And perhaps, Krugman would say that the most effective policy is for government to spend the money when there are clear and easily definable infrastructure needs that would most likely add a high marginal social benefit vis-a-vis the likelihood of confetti hoarding by banks and households.

"In these extreme cases one may say that currency and X have become close substitutes but it is irrelevant because we know that it is due to some special circumstances that we are hardly able to determine."

Special circumstances? They seem pretty straightforward to me. Once a bank has more than enough reserves to cover the reserve requirement on its transactions accounts, there is virtually nothing that reserve balances can do that T-bills cannot. (True, you can't send a T-bill over Fedwire, but liquidity there has never been a problem.) How are these *not* perfect substitutes?

Young Economist,
The amount of bank reserves on all their deposits (demand, savings and time deposits and perhaps on some other liabilities) is determined by both regulation and bank behavior. Also they are invested in accordance with regulations but banks have some discretion on at least the amounts of the particular assets they choose (unfortunately it has been some years since the last time I did work on the details of all these issues, but for 40 years I worked on central bank issues in several countries, starting with my Ph.D. thesis on the post-Peron Argentina's system of 1957-1970). I hope you are able to find some detailed study of bank reserves in the U.S. banking system before and after September 2008 (I live in Santiago Chile and I don't have time to look after these issues). If you look at what happened in several financial crises --starting with the crises of Argentina and Chile in mid 1982-- you are going to find that to bail out banks governments have relied on their central banks to improve the quality (risk) of bank portfolios and to do it they have fabricated transactions that are accounted in banks' balance sheets in ways that generate income. You cannot understand bank statistics if you don't know the relevant policies and accounting rules. Thus, to answer your question you have to look carefully at those policies and rules.

In those 40 years I learnt that the concept of the monetary base (currency + bank reserves with the central bank) should not be used for any analytical purpose --the two components are not substitute because of those policies and accounting rules. If you read again my earlier post you will see that I refer only to currency not to the monetary base. More important, if you look at the experience of many countries since the Great Depression, you will see that bank reserves have played different roles --the extreme case was that of Argentina between 1949 and 1957 when banks had to make a reserve of a 100% on all deposits and the central bank allocated the funds at its discretion but through the banks, a situation that was not different from what I saw in PR China in the mid 1990s when I moved there to advise on the reform of the state banking system.

There is a huge amount of corporate hoarding going on. Even as investment is beginning to pick up, it has been great drag on the economy.

A bigger drag would be to deploy capital on negative expectancy projects and decrease their profitability. Corporations are profit seeking entities. They don't hoard capital to irritate economists.

Liquidity trap!

even as investment is picking up?

I find a lot of modern thinking on liquidity traps (and the zero lower bound) to be confused at best, and when people are confused they fall back onto their priors. The best analysis I've seen is the stuff that Mertens and Ravn are working on--they're explicit about the role of expectations in pushing us into these kinds of situations, and their policy recommendations are pretty "mainstream". I find shocks to expectations much more palatable than black-box shocks to intertemporal conditions--if there's some intertemporal distortion, why don't policymakers attack that distortion?

Of course, standard New Keynesian models don't have a liquidity effect, so I take them with a grain of salt anyhow.

Comments for this post are closed