Caplan on Keynesianism

Bryan Caplan recently defended the use of introspection in economic analysis:

But what is the source of Keynesians’ self-confidence?  Every Keynesian I’ve ever known has a stock answer: The empirics are on our side.  But when probed, they rarely deliver any details about these empirics.  It’s been three decades since Keynesians could merely point at the Phillips Curve and say, “See?  See?!”  And in terms of merely “fitting the data points,” Prescott infamously showed that a simple RBC model works rather well.

At this point, it’s tempting to dismiss Keynesianism as a dogmatic cult.  But in fact, there are key issues where their self-confidence is well-deserved.  The only problem: They’re too scared to admit why.  So let me answer for them: The source of Keynesian confidence is not “empirics,” but introspection.

When Keynesians study the Great Depression, for example, they don’t pore over “the data” to determine whether or not mass unemployment was voluntary.  They introspect – and correctly conclude that tens of millions of workers didn’t decide to take a ten-year vacation in 1929.  They reject the assumption of perfect wage flexibility on the same basis: Not by staring at “the data,” but by introspecting on the question, “How would human beings react if employers cut their nominal wages by 5%?”

Several things struck me about this post.  First, Keynes himself rejected wage rigidity as a cause of involuntary unemployment.  I agree with Caplan on wage rigidity, but it is interesting that introspection led Keynes in a very different direction.

This is the part of Bryan’s post that most intrigued me:

Of course, once Keynesians admit their real reasons, they do have a little problem: Some of their positions fail the introspective test!  Foremost examples:

1. Keynesians’ preference for fiscal over monetary stimulus contradicts introspection about the interest-sensitivity of money demand.

2. The Keynesian claim that wage cuts reduce Aggregate Demand seems introspectively plausible at first, but only if you neglect everyone but workers who already have jobs.

3. Above all, once you take introspection about nominal rigidity seriously, the obvious response is to swear eternal hostility against every government effort to boost labor costs.  Wages on the free market don’t rise and fall like the stock market, but labor market regulation merely amplifies this defect.  Consistent Keynesians should be championing radical labor market deregulation – not signing petitions to raise the minimum wage.

I’d be very interested in your reaction, but it seems to me that the Keynesian liquidity trap argument is extremely counter-intuitive.  That doesn’t make it wrong, but it suggests that Keynesians shouldn’t be mocking RBC explanations of cycles solely on the grounds that they seem implausible.

To make this example cleaner, I’d like to assume a strongly negative IOR is in effect.  People like Krugman argue that the liquidity trap is not just about banks hoarding reserves; in Japan so much cash was hoarded that safes became a popular consumer durable.  So let’s consider whether we could plausibly get a liquidity trap if we had a negative IOR, which assured that any new base money would go into currency held by the public.  To think about the issue, consider the following equation:

Nominal national income = k*Cash.    Or, k = (cash/nominal national income.)

Update:  Well algebra was never my strength.  Richard pointed out it should be Cash = k*NNI

Can the Fed increase nominal national income when the economy is at the zero bound?  If you don’t think so, then you must believe that people behave as follows:

Suppose the Fed increased cash in circulation by 17%.  If there is no effect on NNI, then the average person would react to the Fed’s action by increasing their “k ratio” by 17%.  The k ratio is the ratio of cash to income.  So if you had an income of $100,000, and you held cash of $1000, your k ratio would be 1%.  This is a variable that is obviously under your control.  You choose your k ratio.  Now we can see what is so amazing about the hardcore Keynesian view of liquidity traps.  They are essentially arguing that if the Fed increases the supply of cash by 17%, the average person will react to that action by increasing their k ratio by 17%.  But why?  Is that how you behave?  Do you watch how much the Fed increases the base, and increase your k ratio by that amount?  Would you respond to the Fed boosting the cash supply by 17%, by increasing your personal cash holdings from $1000 to precisely $1170?  Remember, if you only increase it to $1150, the liquidity trap no longer holds.  I simply can’t imagine any mechanism that would cause people to manage their cash that way.

Except . . . suppose the increase was expected to be temporary.  Then speculators would not expect any significant increase in the price level, because current prices are linked through inter-temporal arbitrage to future prices.  In that case if the public tried to get rid of excess cash balances by spending them on goods, speculators would supply more goods.  All the extra net demand would be for financial assets.  Nominal rates would fall until people were willing to hold this extra cash.  Maybe if T-bill yields fell slightly below zero, then people would hoard cash and put it in safes.  Most people wouldn’t increase their cash holdings by precisely 17%, but some people would hoard vast amounts of cash in response to lower T-bill yields.

Nothing new here, just another reason why the traditional explanation of the liquidity trap is inconsistent with common sense.  You really need to assume that currency injections are temporary.  You can’t get a liquidity trap by positing that people don’t want to borrow at zero rates, or that banks don’t want to lend.  That’s not enough—you need to show why people would hoard the extra cash.  And hoard exactly the amount of cash injected.  And you can’t do that in a way that’s consistent with common sense unless you assume the cash injection is perceived as being temporary.  If people think it’s permanent, the average guy won’t hoard it (Ratex or not) and the savvy investor won’t hoard what the average guy isn’t hoarding, as it’s not a good investment if prices are expected to be higher in the future.  Bernanke doesn’t have to convince the public, just the savvy investors.  And they are already bidding up asset prices fast in anticipation of November 3rd, even without any commitment to a higher nominal target, so it doesn’t look like much convincing is necessary.

PS:  In an earlier post I discussed how in the General Theory Keynes had mischaracterized people like Pigou.  Jim Glass sent me these interesting slides, which contain some evidence on that subject.

PPS.  This post is loosely related to recent Nick Rowe and David Beckworth posts on how the “paradox of thrift” is actually about the hoarding of cash.



25 Responses to “Caplan on Keynesianism”

  1. Gravatar of Felix Felix
    9. October 2010 at 08:36

    I think that thinking in terms of temporary vs. permanent money supply increase is silly. It’s the conditions that matter. If the fed is believed to hold the money supply constant at some level unconditional on the future than money will always have no value at all. Money has a certain value because the fed is expected to create demand for money (by selling assets on its balance sheet) whenever the value of money threatens to fall below that and vice versa.

  2. Gravatar of Richard A. Richard A.
    9. October 2010 at 10:17

    I think you meant
    Cash = k*(Nominal national income)

    Nominal national income = k*Cash

    I would tend to view k as a function of deflation expectations at the zero bound. The greater deflation expectations the greater k. The Fed in this simplistic example should offset k by increasing Cash. This should also have the effect of decreasing deflation expectations.

  3. Gravatar of Joe C Joe C
    9. October 2010 at 11:15

    I dont see how people would increase their cash holdings either. The fed using, say QE, should increase inflation expectations. In theory, people would want to reduce cash holding to spend more. Correct me if Im wrong but if the Fed action moves the AD curve left or upward, this results in higher prices. If so, it seems that folks couldnt increase their cash holdings because they need the extra cash to pay services and goods that now have higher prices.

    Generally, I dont think most people think about these things and if someone has more cash….they will spend it; especially with all the conspicuous consumption that goes on in our society.

  4. Gravatar of W. Peden W. Peden
    9. October 2010 at 11:41

    Joe C

    I agree. The Keynesian model of people’s behaviour in a “liquidity trap” situation is surely Homo Economicus-reasoning at its worst.

  5. Gravatar of W. Peden W. Peden
    9. October 2010 at 12:36

    The point about Keynesianism and labour market policy in Caplan’s article was particularly interesting.

  6. Gravatar of David Pearson David Pearson
    9. October 2010 at 12:37


    If the Fed doubles reserves at a zero or negative IOR, and “k” stays the same, would you expect those reserves to back 1/(reserve ratio) in loans? The effective reserve ratio is around 3% given sweep accounts.

  7. Gravatar of scott sumner scott sumner
    9. October 2010 at 14:51

    Felix, I’m afraid I don’t follow you.

    Richard A, Thanks, that was a silly error I made.

    If more cash reduced deflation expectations, then NNI would rise faster than cash. And that very well could happen.

    Joe and W. Peden, I agree.

    David Pearson. I’m not sure I follow your question. Are defining k in terms or reserves or cash?

    The effect of more reserves depends on many factors, most importantly whether the injection is perceived as permanent. The IOR is also important. If the IOR is negative, it is likely that loans will rise somewhat, but I have no idea how much.

  8. Gravatar of Chaitanya Chaitanya
    9. October 2010 at 15:29

    I’m not a total expert at monetary issues, and so had a couple questions:

    1) At the zero bound, how does the fed increase the amount of money in circulation, without arbitrarily affecting the distribution of income within the nation?

    Another, cruder way of asking the same thing is, who gets all the extra cash?

    2) Is an increase in NNI always going to deliver full employment?


  9. Gravatar of scott sumner scott sumner
    10. October 2010 at 05:04


    1. Easy, sell cash for bonds of equal value. The guy getting the new money earns no profit.

    2. No, not if there are aggregate supply problems.

  10. Gravatar of Bill Woolsey Bill Woolsey
    10. October 2010 at 05:33


    You seem to have a vision where currency leaves the banking system and stays out, and people begin to use currency to make payments for goods and services. Once output or prices rise, then currency demand rises (or k falls back to its desired level) and we are back to equilibrium.

    This is implausible.

    Those receiving currency payments will not spend them, but rather deposit them back into banks (or other institutions issuing checkable deposits.) Spending will continue to be largely by check or electronic payments.

    By far the most likely equilibrating process will be for interest rates and the quantity of checkable deposits to change so that people are willing to hold the quantity of currency that is outside the banking system. Now, to the degree there is more spending in the form of checkable deposits, and this raises income and prices, then this will also raise the demand for currency (more Y given k). But the excess currency to more expenditures of currency, with those receiving currency passing it on, is implausible. It will pass through banks.

    If some scheme is used to keep banks from accumulating reserves, then the result isn’t banks using shovels and wheelbarrows to carry currency to the street and give it away. They purchase earning assets (or make loans) and they don’t use currency to do it. They create checkable deposits for the borrowers or the sellers of those assets.
    We have been through this before assuming banks just want to hold T-bills rather than bank reserves. In reality, it is
    likely banks would expand credit on a variety of margins.

    The way currency actually leaves the banking system is that checkable deposits expand enough so that people want to hold more currency.

    This seems so natural if you assume ratios–the currency deposit ratio. (not cash!!!!!) But it is more checkable deposits, people have too much wealth in the form of checkable deposits and so shift some of it to currency. They want to hold more currency.

    In this odd scenario, however, it is almost certain that the currency deposit ratio would fall. Now, there is an interest rate path for this to work. The interest rate on deposits might fall. And this would reduce the demand for deposits relative to currency.

    And, of course, the most realistic scenario is that people spend more by writing checks, real output and prices rise, and so there are more payments of larger dollar amounts that are typically made by currency, and so the demand for currency rises and so, currency leaves the banking system.

    Your vision of dealing with a liquidity trap with open market operations with zero interest securities, and some fancy regulatory scheme to prohibit banks from accumulating reserves, is in error. It follows from an excessively mechanical view of banking (taking the money multiplier process way too literally) along with a variant of the neo-Keynesian promise to keep interest rates too low after recovery. You just shift that to keeping the quantity of money too high.

    A rule that limits the central bank to purchasing a certain set of securities (say T-bills) creates a problem if those securities have a zero yield. Personally, I could care less if it is possible to commit to permanently increase the quantity of money enough so that the long run price level will be high enough, that between now and this future time, there will be enough inflation that the real interest rate will be however low is necessary. It is much better to get rid of the T-bill rule and just have the central bank commit to purchase whatever type of security necessary, perhaps longer term and riskier ones, to keep or promptly return money expenditures to the target growth path. The only promised inflation should be what happens getting back to that path (if it has been left) and then whatever is implied by that path and output productivity.

  11. Gravatar of 123 123
    10. October 2010 at 06:37

    Bill Woolsey, I 100% agree.

  12. Gravatar of ssumner ssumner
    10. October 2010 at 06:45

    Bill, I think you misunderstood my point. I wasn’t trying to suggest that new base money would probably all go out as currency, I was trying to rebut the liquidity trap argument. I think we all agree that the Fed can charge a negative IOR, (the Swedes have already) so there is no zero lower bound for bank reserves. Hence the only potential cause of monetary policy ineffectiveness is hoarding of currency by the public. My example asked people to consider what would happen if all the new base money went out into circulation as currency.

    I agree that in the real world much of the new base money would flow into bank reserves, as deposits increased.

    Regarding temporary or permanent injections, I’d say that temporary injections have no effect on AD, regardless of what assets are bought. If people know that the Fed will soon sell off those assets, then asset prices won’t rise. Conversely a monetary injection that is perceived as permanent will be highly inflationary, regardless of what asset is purchased.

    123, See my response to Bill.

  13. Gravatar of Andy Harless Andy Harless
    10. October 2010 at 06:56

    “Suppose the Fed increased cash in circulation by 17%.”

    You haven’t given me enough information to judge what I would do in those circumstances. How did the Fed increase cash in circulation?

    If the Fed increased cash by buying T-bills, then yes, I would increase my k ratio by approximately the same factor. The yield on T-bills (already zero or negative by assumption, in the liquidity trap scenario) would go down slightly and induce me to sell T-bills and put the money in a safe instead. (I wouldn’t put it in the bank, because the bank would charge to hold it for me, if it faced negative IOR.)

  14. Gravatar of David Pearson David Pearson
    10. October 2010 at 08:36

    The Fed controls reserves, not cash in circulation. Your premise is that the Fed can determine bank excess reserve demand at the zero bound by lowering the IOR to zero or below. But banks, in aggregate, don’t lend out reserves — they just convert them from “Excess” to “Required”. The money multiplier involved in this conversion is quite large–with hyperinflationary implications assuming $1tr in ER’s and a 3% (effective) reserve ratio. I imagine you believe this multiplier is not operative now, and I would be interested in understanding why.

    Also, I wonder what happened in Sweeden under a negative IOR? Did ER’s plummet and RR’s spike? If not, then why would banks continue to hold ER’s under a penalty rate? Why would banks only decrease ER’s “somewhat”, or, “a little bit” if the premise is they only want to hold ER’s if they earn a return on them? What other factors intervened, and how could the Central Bank know how much inflation the negative IOR might produce? I know that the intervention can be viewed as “successful” since rates are now rising. So perhaps all these questions are moot. Still, as a case study, it would be useful to know how reserves responded, and what the Central Bank’s exit plan is now that inflation expectations (presumably) are rising. Can it control them without risking its victory over deflation expectations? What will the Kronor do if they start raising rates, and how will that impact the real economy?

  15. Gravatar of scott sumner scott sumner
    10. October 2010 at 08:42

    Andy, Your introspection is different from mine. I might hold a bit more cash, but I see no reason other than blind faith in the Keynesian model to conclude that I’d suddenly decide to hold 17% more cash. It seems to me that Keynesians simply assume that NGDP doesn’t rise, then assume that nominal rates have to fall enough so that people are willing to hold the extra cash as hoarding balances, not transactions balances. But why assume NGDP doesn’t rise? Normally when you increase the money supply NGDP does rise, even when interest rates fall. So why assume that in this case NGDP doesn’t rise, because the fall in rates will cause people to hoard ALL the extra cash? I must be missing something.

    BTW, my argument is consistent with Krugman’s 1998 liquidity trap paper, which also says temporary currency injections don’t work, and permanent one’s do. So it’s not just a monetarist argument I am making here, but also a new Keynesian argument.

  16. Gravatar of scott sumner scott sumner
    10. October 2010 at 08:48

    David, See my answer to Bill, I think you misunderstood the purpose of my post. I agree that in the real world some of the new base money would go into reserves, but there is no zero lower bound on IOR. I was trying to address the “worst case” liquidity trap argument, one that Krugman often makes when he talks about how the Japanese put money into safes. I wanted to address the zero lower bound by looking at the asset where it is almost impossible to pay interest–cash.

    I’m not very knowledgeable about the Swedish case, I’m told the penalty could be evaded fairly easily.

  17. Gravatar of Chaitanya Chaitanya
    10. October 2010 at 09:03


    Follow up q:

    You said that exchanging bonds for their cash equivalent would increase AD without affecting distribution.

    I have also heard the argument that exchanging bonds for cash isn’t effective with the ultra-low interest rates we have now.

    Also, if the fed decides to start buying up lots and lots of bonds, wouldn’t the price of bonds go up – meaning that bondholders would get the extra money being injected into the system, which would in fact alter the distribution of cash.


  18. Gravatar of David Pearson David Pearson
    10. October 2010 at 09:11


    One other question. Isn’t this question about how the Fed gets cash out into the economy actually moot? We cannot be in a “liquidity trap” if the Fed is financing fiscal deficit spending (unless total credit is contracting, which it is not). In other words, for all the talk of the Fed “pushing on a string”, we are in that special case — the one Bernanke cited back in 2002 — where the Fed creates dollars and the government puts them into circulation. Perhaps the multiplier on these new dollars is low, but the QE effect of a much larger base is high.

  19. Gravatar of scott sumner scott sumner
    11. October 2010 at 05:38

    Chaitanya, It’s a myth that monetary policy matters by changing rates.

    The bond price would go down, if the policy was expansionary enough to trigger strong growth in NGDP.

    David, In my view fiscal policy has little effect. I don’t think we are ever really in a “trap” regardless of whether we have a deficit or not. As a practical matter countries almost always have deficits at zero rates.

  20. Gravatar of David Pearson David Pearson
    11. October 2010 at 06:49

    I’m not saying spending has a Keynesian multiplier, but that it is a means of getting dollars created through OMO purchases into circulation. The 2002 Bernanke (he refers to a money-financed tax cut, which is the same, effectively, as a money-financed spending increase):

    “In practice, the effectiveness of anti-deflation policy could be significantly enhanced by cooperation between the monetary and fiscal authorities. A broad-based tax cut, for example, accommodated by a program of open-market purchases to alleviate any tendency for interest rates to increase, would almost certainly be an effective stimulant to consumption and hence to prices… A money-financed tax cut is essentially equivalent to Milton Friedman’s famous “helicopter drop” of money.18″

  21. Gravatar of Andy Harless Andy Harless
    11. October 2010 at 07:16


    “Normally when you increase the money supply NGDP does rise, even when interest rates fall.”

    Change “even when” to “because” and you’ve got something. When interest rates are at a floor, there is no transmission mechanism. I own T-bills, yielding zero. The Fed comes along and offers to buy them for par plus epsilon. I say, “Sure.” Now I’m epsilon richer, and I’m holding my money in a safe instead of a T-bill. What’s the difference? Why should anyone change their spending?

    Unless of course we believe the liquidity trap to be temporary and take the Fed’s purchase as a signal that it intends to allow a higher price level once the trap ends. Then the real interest rate has fallen, and there is an incentive to spend.

    In practice, I think our views may be observationally equivalent. Assuming the liquidity trap is expected to end eventually, a permanent OMO is equivalent to an increase in the price level (or NGDP) target (if by “permanent” we mean that it makes the entire future path of the money stock higher than it would otherwise have been). But in practice the only reasonable way to commit to making the operation “permanent” is by raising some kind of target. Philosophically, I think that raising the target is the cause of the short-run increase in NGDP, and you (if I interpret you correctly) think that increasing the money stock is the cause. But perhaps there is no real substance to the distinction.

  22. Gravatar of Morgan Warstler Morgan Warstler
    11. October 2010 at 07:39

    I think we need to spend more time celebrating this:

    “3. Above all, once you take introspection about nominal rigidity seriously, the obvious response is to swear eternal hostility against every government effort to boost labor costs. Wages on the free market don’t rise and fall like the stock market, but labor market regulation merely amplifies this defect. Consistent Keynesians should be championing radical labor market deregulation – not signing petitions to raise the minimum wage.”

    This is pure and clean and enough to chase Krugman out of his hidey-hole, and force him to eat it in big nasty bites.

    If you believe in sticky wages, then you should feel comfortable not mandating them…

    Otherwise you tacitly admit they aren’t real enough to bank on in your economic theory.

  23. Gravatar of scott sumner scott sumner
    12. October 2010 at 05:04

    David, I understand that argument, but I don’t agree. Even a helicopter drop is not effective if perceived to be temporary. But if the increase is perceived to be permanent, then even swapping cash for T-bills is highly effective.

    Andy, You said;

    “Change “even when” to “because” and you’ve got something. When interest rates are at a floor, there is no transmission mechanism.”

    As you know, I view the liquidity effect on interest rates as mostly an “epiphenomenon”, much less important than other transmissions mechanisms like changes in stock, commodity and real estate prices. When policy changes dramatically, as in 2001 and 2007, longer term interest rates move “the wrong way” in response to policy surprises. This tells me interest rates can’t be an important part of the transmission mechanism–at least nominal interest rates that is.

    When nominal rates hit zero, you still have the other 10 transmission mechanism mentioned in Mishkin’s Money textbook. And those are the powerful ones.

    You said;

    “Unless of course we believe the liquidity trap to be temporary and take the Fed’s purchase as a signal that it intends to allow a higher price level once the trap ends. Then the real interest rate has fallen, and there is an incentive to spend.”

    I think we all agree that it is future expected monetary policy that really matters. But that’s equally true of interest rate targeting. A change in the current interest rate has no effect if the Fed simultaneously announces it won’t change it’s price level target, rather it would simply create expectations that future interest rate changes would offset the current change. It’s always about communication, I certainly agree on that. If the Fed wants to communicate an easy policy, the best way to do so is raising their inflation target (or NGDP) If they won’t do that, they can permanently raise the money supply. To do that they must raise the current money supply. It won’t work as well because traders will have differing opinions on whether the increase is permanent. But it could certainly lead some traders to expect higher inflation, so it might have a modest effect. I read the stock market as indicating that the expected QE will have a positive effect, albeit quite small. How do you read the stock market reaction? I’d be surprised if we are that far apart.

    Morgan, I agree with that statement by Caplan.

  24. Gravatar of Ralph Musgrave Ralph Musgrave
    12. October 2010 at 06:45

    1. The original article at the top claims that “Keynes himself rejected wage rigidity as a cause of involuntary unemployment..” Keynes said that a fundamental cause of unemployment was that wages were “stick downwards”, didn’t he?

    2. Scott Sumner (12th Oct 05.04) claims “Even a helicopter drop is not effective if perceived to be temporary.” Two bits of research contradict that. First, this study of the allocation of windfall income showed that a significant proportion went fairly quickly to extra consumption:

    Plus I came across a similar study with similar results done in Singapore but haven’t go the URL for that.

  25. Gravatar of ssumner ssumner
    15. October 2010 at 15:03

    Ralph, Keynes said many things, but in the chapter of the GT on classical economics he argued that the classical model relied on wage rigidity, and that this meant the classical model assumed unemployment was voluntary–because workers could solve the unemployment problem by accepting lower wages.

    A windfall of income is not the same as temporary income in the sense I am using he term. You mean they get $X this year, and no more in the future. I mean they get $X this year, and the money is taken away next year. I’m not saying that has zero effect, but it would be quite small

Leave a Reply