A few remarks on Kling’s business cycle theory

Arnold Kling recently made this criticism of my argument:

I think that my least favorite Sumnerian proposition is that the Fed can affect the economy by announcing a long-term target for a nominal variable. I think that in order for this to work, you have to assume that people are forward-looking and focused on future monetary policy. On the other hand, his mechanism by which monetary policy works is the conventional story in which nominal wages are sticky, so that with higher aggregate demand you get lower real wages and more real output. But for nominal wages to be sticky, workers cannot be forward-looking and focused on monetary policy. So, on the one hand, for targets to matter, people have to be forward-looking. On the other hand, for monetary policy to matter, people cannot be forward-looking. I cannot past what I see as a basic contradiction.

Commenters frequently ask me what would be the point of higher inflation expectations.  Doesn’t the rational expectations model imply that only unanticipated inflation has real effects?  Yes and no.  As far back as the late 1970s people like Fisher had already shown that in a new Keynesian ratex model with sticky prices, monetary policy could still influence real variables.  For this to happen, the monetary shock must occur after some wage or price decisions have already been made, but before those wages and prices are scheduled to be re-adjusted.

Kling probably knows this, which is why he focuses on long run inflation targets.  Suppose we target NGDP growth over the next 5 years, surely all wages and prices would have adjusted by then?  Yes, but the point of changing future expected NGDP is to change current AD, and hence current NGDP.  And if we do that, and if wages and prices are sticky, then monetary policy can have real effects in the short run.

Here is an analogy.  Suppose than the US government announced that starting three years from today it would begin buying unlimited quantities of gold at $2000/ounce.  But for now it would not intervene in the gold market, leaving prices at their current $1250.  Obviously the policy, if credible, would cause gold prices to immediately rise to about $1900, and employment in the gold mining industry would rise.  Even a promise to intervene to hit a future target can have powerful current effects on output.

[BTW, this experiment actually occurred in 1933, when the government hinted that once it returned to the gold standard it would buy unlimited gold at the new price point.  That’s why higher gold prices in 1933 mattered, even though the implicit promise wasn’t implemented until 1934.]

Here is Arnold Kling on recalculation:

I do not think of the prosperity we experienced a few years ago as inherently false. It happened to be false because it relied on unsustainable beliefs. If we had not had a housing bubble, I think it is possible that instead firms and households could have created patterns of specialization and trade that resulted in full employment and affluence that were more sustainable. Ultimately, I think that the Recalculation will result in such new patterns of specialization and trade.

As it is, a lot of people’s plans have been disrupted by the collision between prior habits and reality. I do not think that there is much that government can do about it.

I agree that the prosperity of mid-2007 was partly false.  But the so-so economy of mid-2008 (with 6% unemployment) was not false.  By mid-2008 the housing crash had occurred, but the rest of the economy was doing OK.  Then AD fell sharply, and the economy went from so-so to horrible.  Not only can government do something about this, but government caused the problem with its tight money policy that depressed NGDP 8% relative to trend, and caused the financial crisis to worsen dramatically.

Here is Karl Smith:

The simple fact of the matter is that the structure of the American economy hasn’t changed that much in that last 24 months. We were building houses at an unsustainable rate, sure. But, at its peak the US employed about 7.7 million construction workers. It now employees about 5.8 million. That’s a difference of a little less than 2 million workers or about 1.5% of the American workforce.

He doesn’t say this, but the vast majority of those jobs had been lost by mid-2008, when unemployment was still at fairly reasonable levels.  Here is Matt Yglesias commenting on Smith:

In October of 2007, the unemployment rate was at 4.4 percent. By January of 2008, it had risen to 5.4 percent. Then it dipped down a bit, but by July of 2008 it was at 6 percent, about the level you’d expect from 2.2 million construction workers losing their jobs. But then by Election Day it was all the way up to 6.5 percent. Between Election Day and Barack Obama’s inauguration, it increased two more percentage points. Then in the following year, it went up another 1.1 percentage points before very slowly starting to tumble.

This is, as Smith argues, not about the difficulty of shifting those construction workers into other segments of the economy. It’s about system-wide excess demand for money, and a failure of the policy department to respond appropriately.

Now I know that some of my commenters will respond that “it wasn’t just houses, we were making too much of almost everything.”  Too many cars, too many drapes, too many washing machines, etc.  Funny how these generalized “overproduction” problems affecting almost all sectors just happen to show up at precisely the moment when NGDP begins falling.  Funny how the cars and houses produced in the 1920s found buyers, but suddenly in the early 1930s when NGDP fell in half we found out that all along we had been producing too much of almost everything.

PS.  Speaking of Yglesias, he issues the following challenge to right-of-center economists:

That’s because Congress is unwilling to extent UI eligibility beyond 99 weeks. Which means that soon enough we should see . . . something. But what? My guess is not much. My guess is that basically nobody wants to hire the vast majority of the current long-term unemployed. So my guess is that the labor market prospects of people who’ve been unemployed for 98 weeks will look an awful lot like people who’ve been unemployed for “only” 70 weeks. I wonder what people with a more right-of-center approach would predict? Are we going to see a surge in employment among people around the 99 week mark? Will they uncover hidden secret jobs that are stashed away somewhere?

A few comments.  I am surprised that this in an open question.  We had extended benefits in other recessions—surely we must already know the answer?  But if not, then yes, I would expect a big increase in job finding success for those whose benefits have run out.  Not because they are deadbeats, but because unemployed factory workers would get desperate and take jobs at McDonalds.  Obviously I differ from most economists on the right in that I favor more stimulus, whatever the answer to this question.  And I favor Singapore-style self-funded UI, whatever the answer.  But I do accept Yglesias’ implicit argument that if there is no surge of job-finding, then (in the absence of an optimal self-funded UI scheme) it would strengthen the case for extended benefits during recessions.



27 Responses to “A few remarks on Kling’s business cycle theory”

  1. Gravatar of Jeff Jeff
    16. July 2010 at 06:57

    Kling is wrong. Total private and public debt in the US alone is over $50 trillion, much of it long term, and almost none of it indexed to inflation. Nor are many other prices indexed. Couple that with the fact that no one predicts inflation or monetary policy itself very well, and you arrive at the conclusion that whatever the central bank does has enormous repercussions.

  2. Gravatar of scott sumner scott sumner
    16. July 2010 at 07:12

    Jeff, In fairness to Kling, I believe his specific criticism of me related to the sticky wage/price argument, not the ability of the Fed to influence debt. But if he made that argument somewhere in his piece, then yes, it is wrong.

  3. Gravatar of Jeff Jeff
    16. July 2010 at 07:50


    But long term nominal debt is a sticky price, and it persists in a real world where people are forward-looking. And this is an industry (finance) where you would think menu costs would be minimal.

    Even if we didn’t have sticky prices, the Lucas island model showed way back in 1972 that unanticipated changes in monetary policy have real effects. Kling doesn’t seem to believe even this.

  4. Gravatar of jj jj
    16. July 2010 at 08:04

    Kling is right that it doesn’t make sense to say one market is forward-looking and another is not, but I see why you need to believe that’s happening. “Sticky prices” can exist in a forward-looking environment. I put it in quotes because sticky prices is a misleading term for the many phenomena that it tries to model.

    I see it as the mismatch between the “long runs” for different markets. The long run for inflation expectations is overnight: that’s how fast the Fed could change them. The long run for the job market is months, or years. The long run for debt is overnight, for some debt, and 30 years for other debt. And that’s with everybody involved being forward-looking.

    So when NGDP drops, some markets adjust instantly and some don’t.

    There have been some good posts on Worthwhile Canadian Initiative about sticky prices recently.

  5. Gravatar of scott sumner scott sumner
    16. July 2010 at 08:11

    Jeff, Nominal debt is not a price (sticky or flexible.) The price of long term nominal debt is flexible, and changes minute by minute. Monetary policy redistributes money between debtors and borrowers. But these are deadweight losses and gains, and don’t affect the incentive to produce, at the margin.

    jj, Those are good points.

  6. Gravatar of jj jj
    16. July 2010 at 08:32

    oops, make that “…I DON’T see why you would need…”

  7. Gravatar of Morgan Warstler Morgan Warstler
    16. July 2010 at 09:05

    Scott, again you show your bias. There was NO housing crash in 2008. Prices were STILL above trend (2% per year since 2000).

    Raise interest rates and liquidate MBS – foreclose and auction everything 90 days late. Require 30% down and do not let banks buy the assets.

    Increase balance sheets of those with dry powder (so they’ll use it elsewhere). We’ll get lower rents for those without savings. Increase worker mobility and competition amongst states.

    Best of all, we’ll find out which banks are insolvent. Do not underestimate the value of killing zombies.


    WE CAN ALWAYS PRINT MONEY, before we do that, let’s raise rates and dwindle the Fed’s hard asset balance sheet. Its only fair.

    The Fed eating a massive loss on MBS is printing money.

  8. Gravatar of Lord Lord
    16. July 2010 at 09:05

    I expect we will see more leave the workforce as we have already started seeing. McDonald’s doesn’t need more workers.

  9. Gravatar of Wonks Anonymous Wonks Anonymous
    16. July 2010 at 10:31

    Casey Mulligan presented some numbers from another time+place with high unemployment:

  10. Gravatar of david glasner david glasner
    16. July 2010 at 10:33

    Scott, When the real rate is less than the expected rate of deflation, rational expectations and perfectly flexible prices cannot save you from disaster. So, if the real rate is, oh 0.5 percent and the expected rate of deflation is 1 percent, adding up 0.5 and negative one gives you negative 0.5. Slight problem with that. The nominal rate can’t fall below zero. So what happens? Asset prices fall and investment tanks because the rate of return on cash exceeds the rate of return on real assets. Which means that we have gotten ourselves into a downward spiral in which the real rate starts to rise as asset prices fall, but as asset prices fall the expected rate of deflation also starts to rise and there is nothing in this model that guarantees that you get to a solution before the economy has completely imploded. So invoking rational expectations and price flexibility does not accomplish as much as Kling wants it to.

    Until about April, the expected rate of deflation was probably less than the real rate, so the equilibrium nominal rate was probably just above zero. After the Greek crisis, deflation expectations started to increase which put the equilibrium nominal rate into negative territory, so that we are not again flirting with the danger of cumulative deflation. Yikes!

  11. Gravatar of Jon Jon
    16. July 2010 at 11:32

    Mostly agree, but Kling has a point to question what is ‘M’ in the quantity relationship. Isn’t this what we learned during the 70s and 80s? Its hard to find a money aggregate with a stable V.

    I think the evidence favors ‘M’ being a broader measure of money than the monetary base. In particular, ‘V’ is more volatile in countries with more financial product innovation.

    The trouble becomes then that the MB is small relative to M. So the MB becomes useful as a constraint on the growth of M (because of statutory relationships between base money and deposit creation which is a much larger component of the true M) but not very effective at replace an absence of M from other sources.

    So its possible that V is stable and not so sensitive to i if M is selected appropriately but if M is identified as the MB then V appears sensitive to i because the relationship between between the MB and the true M is sensitive to i.

    I don’t think gold targeting will be so successful as you suggest in modern times. The basic problem is that gold has been decoupled from contract prices. In a gold-standard regime, changing the gold price has the effect, inter alia, of changing the value of all monetary instruments because gold is money and is exchangeable for goods. In your model, the effect depends on the government trading money for gold and depends on the quantity of gold exchanged. Under a gold standard, all gold held becomes worth more nominally and all contracts are devalued immediately versus gold-holdings, but, because almost no gold is privately held these days, this policy will be attenuated.

  12. Gravatar of Jeff Jeff
    17. July 2010 at 03:15


    Sure, nominal debt is not a price, but for the borrower, it is a long term contract with fixed nominal payments. Like a sticky price, that rules out some kinds of period-by-period optimizations.

  13. Gravatar of Morgan Warstler Morgan Warstler
    17. July 2010 at 07:21

    @Jeff, which is another reason we want lower rents… and not inflated housing prices. Prices can be less sticky when a larger percentage of our income is spent in non-fixed ways.

  14. Gravatar of Doc Merlin Doc Merlin
    17. July 2010 at 16:47

    ‘Sure, nominal debt is not a price, but for the borrower, it is a long term contract with fixed nominal payments. Like a sticky price, that rules out some kinds of period-by-period optimizations.’

    Thanks Jeff, I’ve been trying to make this point to Scott for some time. You just did it a lot better than I could have.
    What is interesting is that default and refi are still possible. This means that when interest rates drop, debt prices fall (and readjustment doesn’t destroy the value of principal), but when interest rates rise the way to readjust is default (which does destroy value of principal.

    If this is the mechanism for stickiness, it explains why crashes are shorter and more steep than booms. Something that standard Neo-Keynesian wage and price stickiness doesn’t explain well.

  15. Gravatar of scott sumner scott sumner
    17. July 2010 at 18:11

    Morgan, Who cares about trends? Markets are unpredictable.

    House prices rose above trend in Boston 25 years ago, and they are still above trend. 25 years from now they’ll still be above trend.

    Lord, A professor once told me never to use the word “need” in economics, it always means your argument is flawed. Need depends on price.

    wonks anonymous, I need to do a post on that. It is a good article but I have something to add. I’ll try to do it early next week.

    David, I mostly agree, but I have a couple reservations:

    1. If you really did have perfect price flexibility and ratex, I’m not sure you’d have unemployment even with a deflationary shock. But I honestly don’t know exactly why not. My hunch is that a shock normally expected to produce say 2% a year deflation, would instead cause prices to adjust downward discontinuously. And then fall at a rate consistent with macro equilibrium from that point forward.

    My other concern is the question of how to measure the real interest rate. Let’s suppose that there is a deflationary shock that everyone thinks will cause the price level to eventually fall by 10%. But what happens is that flexible asset prices immediately fall by 10%, whereas the overall CPI is expected to fall by 2% per year for 5 years. If nominal rates are 1%, what is the real rate? Well the real rate in terms of the CPI is 3%, but the real rate in terms of the expected change in asset prices is 1%. So which is the right real interest rate? I don’t know the answer, but I am also not satisfied that monetary models have paid enough attention to this distinction.

    Jon, You said;

    “Mostly agree, but Kling has a point to question what is ‘M’ in the quantity relationship. Isn’t this what we learned during the 70s and 80s? Its hard to find a money aggregate with a stable V.”

    Yes, I completely agree; V is not constant for any definition of M. But that doesn’t undermine any of my arguments, as I’m not claiming V is stable.

    I also completely agree that gold targeting would not be a good idea in modern times. I just used the gold example as an analogy of how a policy intended to affect prices in the future, would also double back and affect prices today.

    Jeff, Yes, I can accept that. Ironically, I once published a paper making a similar point to you. I argued that debt affects money demand long term, as debt repayment was one of the things that money is used for. So if nominal debt is sticky for decades (due to long term nominal contracts) then the demand for money needed to pay off those debts is also sticky. I don’t know how important this effect is, however. I still think wage stickiness is the bigger problem, as it directly impacts the marginal cost of production.

    Doc Merlin, I think I rejected it as a prime business cycle factor, as much of what occurs is redistribution from lenders to borrowers, or vice verse. On the other hand I obviously can’t deny that this has SOME effects, and maybe I need to keep an open mind as to how important those effects are.

  16. Gravatar of John Papola John Papola
    18. July 2010 at 04:45

    “Who cares about trends? Markets are unpredictable.”

    This is something I trip over, Scott. Why should the Fed try to bring nominal GDP back to some prior growth path? Why not stabilize nominal GDP or target 3%? I realize that you don’t fully buy the Austrian story of cumulative malinvestments when productivity-driven deflation is offset by monetary policy… but what if it’s right? Why should nominal spending grow? Is it all about the Fed’s institutional inability to act at the zero-bound?

    The co-called “liquidity trap” seems to be a mental block at central banks, not an natural element of economic life.

    Or is it simply that you want the Fed to inflate away enough private debt to free up the perception among so many that they are stuck in a long slog to pay things down in a period of austerity?

  17. Gravatar of ssumner ssumner
    18. July 2010 at 05:43

    1. I am not arguing that the Fed can still get us back to the old 5% growth path. I’d be thrilled if they got us back to the three percent NGDP growth path that Bill Woolsey favors.

    2. You are partly right about the liquidity trap being a mental block.

    3. I have never favored having the Fed inflate away any debt. I am opposed to the Fed adopting a policy that favors eith lenders or borrowers, I want a neutral policy of steady NGDP growth.
    If the Fed goes with price level targeting rather than NGDP, then I want a steady inflation rate which would then get factored into debt contracts, and would result in inflation not favoring either side.

  18. Gravatar of John Papola John Papola
    18. July 2010 at 07:42


    But a steady rate of inflation, as measured by CPI or some other aggregate, means that productivity gains in one sector must necessarily (and in my opinion arbitrarily) be offset by inflated prices of another.

    Why should this happen? Let’s pretend there is no liquidity trap-inducing Fed hangups about doing whatever it takes to stabilize NGDP. Why should it grow at all?

    I don’t get it. Target NGDP to be flat. Let falling prices of iPads simply be that and not necessitate an increase in the price heavy iron to offset it for some arbitrary aggregate “price level”. Didn’t Keynes warn that the price level can be a tyranny? Seems like it to me.

    Where am I wrong?

  19. Gravatar of Morgan Warstler Morgan Warstler
    18. July 2010 at 15:20

    Yes, yes, don’t say “need”… housing prices NEED to go down BECAUSE previous Fed actions have artificially raised prices.

    If we just happened upon a system where the Fed hadn’t been buying up MBS, so Fannie and Freddie were sitting stuck with them, unable to backstop more loans…

    We’d have lower home prices. And more insolvent banks.

    And if Fannie and Freddie, didn’t happen – we’d have even lower real estate prices.

    And if there wasn’t a mortgage tax incentive… well you get the idea.

    The point is, after the fact, we have this monster owned by banks overseeing our money supply – who’s members have a vested interest in seeing housing prices go up.

    We have in front of us a unique opportunity to really set finance back on its heels, and you want to inflate our way out of it instead.

    I’d urge you to start your next response with, “Even though home prices are comparatively too high…”

    And see how the rest of your argument sounds.

  20. Gravatar of Jon Jon
    18. July 2010 at 22:33

    Scott writes:

    Yes, I completely agree; V is not constant for any definition of M. But that doesn’t undermine any of my arguments, as I’m not claiming V is stable.

    Yes, sure you don’t depend on V being stable per-se, but you depend on V !~ 1/M. I.e., if every change in the MB was offset, how would the Fed transmit its medicine. Clearly there are many points in between. The practical point Kling was getting to though is whether the proportionality NGDP ~ M dominates.

    I also completely agree that gold targeting would not be a good idea in modern times. I just used the gold example as an analogy of how a policy intended to affect prices in the future, would also double back and affect prices today.

    Personally, I think we would better off with gold convertibility, but ironically I think this because the cross of gold is a myth not because gold is some buttress against inflation, a point that FDR demonstrated for us quite well. So, if there was gold-convertibility, and gold was accepted as money, then I think the Fed would have a much more powerful tool-chest to use.

  21. Gravatar of scott sumner scott sumner
    19. July 2010 at 05:45

    John, I also prefer targeting NGDP. But I’d prefer a slow rate of increase. I don’t see why a stable NGDP is special.

    Morgan, Again, I don’t want to inflate out of anything. I want steady growth in NGDP.

    Jon, That’s a non-sequitor. It would be like arguing that an increase in the supply of apples will have zero effect on the price of apples. And then when someone challenges this assertion, you say, “well statistical studies showed that there is no precise inverse relationship between the price of apples and the quantity of apples.” Yes, but it doesn’t mean that changes in supply have ZERO effect on price.

    Fluctuating velocity provides zero support for Kling’s views.

    I interpret FDR’s action as showing that gold does protect against inflation, and only by leaving the gold standard can you get inflation.

  22. Gravatar of Jon Jon
    19. July 2010 at 07:24

    Scott: you have to be will and enable to move the gold price peg. This is of course what monarchs have done for centuries to create inflation.

    It works very well. If you call that leaving the gold-standard, fine, that matches with people’s common understanding of the term. My point is that convertibility is really an independent matter.

  23. Gravatar of John Papola John Papola
    19. July 2010 at 20:22


    Have you seen William White’s 2006 paper “Is Price Stability Enough?”


    I fear that growing NGDP leads to cumulative mal-investments and the business cycle. That’s what makes stable NGDP better.

  24. Gravatar of scott sumner scott sumner
    20. July 2010 at 04:21

    Jon, If the price of gold moves then you don’t have a gold standard, so I misunderstood what you were talking about.

    The dollar is currently convertible into gold, it’s just that the rate of conversion varies every single day.

    John, There is no good theoretical reason why growing NGDP is more likely to lead to business cycles than stable NGDP. The people that make those arguments (as far as I can see) don’t seem to understand the basics of the neutrality of money. They think rising house prices will cause a housing bubble, for instance, which is nonsense.

    The Fisher effect means that real interest rates are not affected by the trend rate of inflation.

    I can’t say I care for that White paper. He seems to think low interest rates mean easy money.

  25. Gravatar of Jon Jon
    20. July 2010 at 07:38

    No, there is a difference, today holders of gold bear the risk of goldbeing speculatively valued. I’m implicitly assuming that since the monetary authority will always avoid deflation, the would only move the gold peg monotocally.

    The problems of gold arise from never moving the gold peg.

  26. Gravatar of John Papola John Papola
    20. July 2010 at 16:06

    I find the argument against the neutrality of money in the short run pretty convincing. If the Fed creates new money through the credit markets and lowers the rate of interest as a result, the immediate effect of that seems to be an increase in the quantity demanded for interest elastic goods (like capital goods and durable goods). That shift between capital goods and consumer goods is a real change, not a nominal one.

    New money enters the economy at a particular point, through a particular policy instrument. Do you reject the Cantillon effect?

    I think your dismissal of the austrian business cycle theory, and your disagreement about macro-stability effects with George Selgin, may hinge on this money neutrality.

  27. Gravatar of scott sumner scott sumner
    21. July 2010 at 06:57

    Jon, OK, but I’ve forgotten the original context of this issue.

    John, I agree that money is not neutral in the short run, but we are talking about different long run inflation rate targets. I am saying that a long run inflation rate of 0% or 3% has about the same effects. The only difference is that 3% inflation discourages investment, because it raises the real tax rates on investment.

    There is nothing in all of economics that I reject more than the Cantillon effect. Monetary policy shocks, even big ones, usually involve only tiny injections of money. It makes no difference whether money enters one bond market or another. Some of my commenters have talked about money “going into” the stock market or the housing market, but this confuses money and credit, two completely different concepts. Money doesn’t “go into” markets. To the extent that there is a Cantillon effect, it is merely the effect of the asset side of one Fed balance sheet differing from another. And that is trivial.

    Your point about Selgin may be correct.

Leave a Reply