Krugman vs. Eggertsson

I am getting whiplash trying to keep up with Krugman.  Four days after insisted in a NYT Op-ed that monetary policy was our only hope, and endorsed a major program of QE, he has again reverted to the view that monetary policy is ineffective once rates hit zero.  Seems my “vindication” was too good to be true.  We’ll get to monetary policy eventually, but first I’d like to examine the way that Krugman interprets a recent study by Eggertsson.

Gauti Eggertsson is in the process of presenting a new paper on fiscal policy; the paper is here. In his presentation “” though not in the paper “” he offers great phrase: the “paradox of toil.”

According to his paper, when you’re in the liquidity trap, certain kinds of tax cuts have perverse effects. Cutting taxes on capital income, for example, encourages more saving “” which is a bad thing, because we’re suffering from the paradox of thrift. In fact, reduced taxes on capital income actually end up reducing investment.

So what’s the paradox of toil? If you cut taxes on labor income, this expands labor supply “” which puts downward pressure on wages and leads to expectations of deflation, which increases the real interest rate, which leads to lower output and employment.

I am going to skip over the so-called paradox of thrift, which as all good monetary economists know is really about an increase in the demand for money, not saving.  Let’s focus on the assertion that a cut in the payroll tax can increase labor supply, and thus shift SRAS to the right, and yet still decrease employment.  How is that possible?  It is possible because Gauti Eggertsson assumes the AD curve is upward sloping when rates are stuck at zero.   Could something like this happen?  I suppose anything is possible.  But if you look closely at Eggertsson’s paper, it isn’t very likely.  And if you look at the evidence from when interest rates actually were near zero, there is overwhelming support for downward sloping AD curves.  But first, let’s see what makes Eggertsson’s model tick:

Figure 5 clarifies the intuition for why labor tax cuts become contractionary at zero interest rates while being expansionary under normal circumstances. The key is aggregate demand. At positive interest rates the AD curve is downward-sloping in inflation. The reason is that as inflation decreases, the central bank will cut the nominal interest rate more than 1 to 1 with inflation (i.e., φπ > 1, which is the Taylor principle; see equation 6). Similarly, if inflation increases, the central bank will increase the nominal interest rate more than 1 to 1 with inflation, thus causing an output contraction with higher inflation. As a consequence, the real interest rate will decrease with deflationary pressures and expanding output, because any reduction in inflation will be met by a more than proportional change in the nominal interest rate. This, however, is no longer the case at zero interest rates, because interest rates can no longer be cut. This means that the central bank will no longer be able to offset deflationary pressures with aggressive interest rate cuts, shifting the AD curve from downward-sloping to upward-sloping in (YL,πL) space, as shown in Figure 5. The reason is that lower inflation will now mean a higher real rate, because the reduction in inflation can no longer be offset by interest rate cuts.

One of the most important ideas of modern macro is that an interest rate peg is a horrible idea, as it can lead to hyperinflation or deflation.  As a result many economists advocate much more aggressive interest rate rules than anything Keynes contemplated.  One of those rules is the so-called Taylor Rule, which raises and lowers nominal rates more than one for one with any change in expected inflation.  Under this sort of policy there is no paradox of toil, indeed there is no paradox of thrift.  But let’s suppose the nominal rate has fallen to zero, and deflationary expectations increase.  What can the central bank do?  Eggertsson says the central bank “will no longer be able to offset deflationary pressures with aggressive interest rate cuts.”

Let’s stop right here.  This is what absolutely drives me nuts about Keynesians.  (And by the way, I hate to do this because I like Gauti Eggertsson; he was very kind to cite three of my papers in his recent AER piece—which is an excellent article.)  A Keynesian will keep insisting that you can’t be a realistic, sensible, modern monetary economist unless you think in terms of interest rate rules.  Monetary rules are too old-fashioned, and futures targeting is too, I don’t know, futuristic?  So it’s got to be interest rate rules.  Then they develop models that show interest rate rules don’t work very well when interest rates are at zero and further monetary ease is called for.  No kidding.

But is that our only option?  Fortunately it isn’t.  There is still inflation targeting, and elsewhere in the paper (which claims to provide theoretical support for fiscal stimulus through more government spending), Eggertsson offhandedly admits that inflation targeting can work, indeed admits that it is the first best policy, even better than fiscal policy.  And there are still some old-fashioned monetarists who insist that quantitative easing can work at zero rates:

So if we’re going to haveany real good news, someone has to take responsibility for creating a lot of additional jobs. And at this point, that someone almost has to be the Federal Reserve.

.   .   .

But while economic analysis says that we should have a large second stimulus, the political reality is that the president “” faced with total obstruction from Republicans, while receiving only lukewarm support from some in his own party “” probably can’t get enough votes in Congress to do more than tinker at the edges of the employment problem.

The Fed, however, can do more.

.   .   .

The most specific, persuasive case I’ve seen for more Fed action comes from Joseph Gagnon, a former Fed staffer now at the Peterson Institute for International Economics. Basing his analysis on the prior work of none other than Mr. Bernanke himself, in his previous incarnation as an economic researcher, Mr. Gagnon urges the Fed to expand credit by buying a further $2 trillion in assets. Such a program could do a lot to promote faster growth, while having hardly any downside.

Oops, my mistake, that’s actually from a Krugman Op-ed from a few days ago.

Before irate Krugman fans write in, let me admit that I am just having some fun here.  I realize that Krugman and Eggertsson actually have very similar views, and the apparent differences reflect differing assumptions about what sort of policies are politically feasible and/or credible.  But I do have a very serious point to make.  Krugman is using a double standard, basically accepting an argument made by Eggertsson that is far weaker than those he criticizes people like Mankiw and Barro for making:

All of this only applies in a situation of zero interest rates, which wouldn’t be interesting except that that’s the situation we’re in.

The general point is that we’re really through the looking glass, in a world in which lots of things have perverse effects “” and basing your policy ideas on intuition from “normal” times can lead you very much astray.

PS: Right at the beginning, Gauti made the point that empirical results from periods in which interest rates are not zero tell you little about this situation “” which is why most of what Barro and others, including Greg Mankiw, have been saying is besides the point.

OK, let’s see what else is “besides the point.”  Eggertsson says that his model applies only under very artificial conditions:

It is worth stressing that the way taxes are modeled here, although standard, is special in a number of respects. In particular, tax cuts do not have any “direct” effect on spending. The labor tax cut, for example, has an effect only through the incentive it creates for employment and thus “shifts aggregate supply,” lowering real wages and stimulating firms to hire more workers. One can envision various environments in which tax cuts stimulate spending, such as old-fashioned Keynesian models or models where people have limited access to financial markets. In those models, there will be a positive spending effect of tax cuts, even payroll tax cuts like the ones in the standard New Keynesian model.

Hmmm.  Aren’t those “old-fashioned Keynesian models,” the ones without Ricardian equivalence, also the ones that Krugman insists apply to the real world?  So now there are two strikes against Eggertsson’s model.  It assumes Ricardian equivalence, and it basically assumes monetary policy is impotent once rates hit zero.  But as we just saw, Krugman insists that the Fed “can do more” once rates hit zero.  They can engage in QE.  And both Krugman and Eggertsson insist the Fed can still engage in inflation targeting.  Indeed they both seem to agree that inflation targeting is a first best policy when faced with inadequate aggregate demand.

At this point the reader may be thinking “enough theory, anything is possible under the right theoretical assumptions.”  Let’s see how Eggertsson’s theory works in the real world.  Fortunately, FDR did a near perfect test of the theory.  He instituted a high wage policy in the midst of the Great Depression.  Indeed he didn’t just perform this natural experiment one time, he did it five times.  That’s right, on five different occasions between 1933 and 1940 the average nominal wage rate was increased sharply by government policy.  Four of the increases are easy to date, as they reflected minimum wage increases, and/or similar direct interventions into wage setting.  And in each case the aggregate hourly wage rose significantly.  The fifth was the Wagner Act, which led to greatly increased unionization in late 1936 and early 1937, and much higher wages in 1937.  The best experiment began in late July 1933, when nominal (and real) wages were increased by roughly 20% over two months!  I’m going to go out on a limb and argue that no respected macroeconomic theory has ever been so decisively refuted by the data as the theory that high wage policies can actually help the economy during a Depression.  This is from one of my first posts:

Table 12.2: Four month (nonannualized!) growth rates for industrial production

Before        After
July   1933 wage shock   +57.4%     -18.8%
May   1934 wage shock   +11.9%    -15.0%
Nov.  1938 wage shock   +15.8%     +2.5%
Nov.  1939 wage shock   +16.0%     -6.5%

You’ll notice that I left out the fifth wage shock, but it’s no better for Krugman’s view, just messier.  Historians argue that the huge union drives of late 1936 and 1937 were due to both the Wagner Act and FDR’s massive election victory.  Whatever the cause of the union gains, they led to rapid wage increases in late 1936 and much of 1937.  This time, monthly industrial production did not fall immediately, as prices were also rising fast in late 1936 and early 1937.  But when prices stopped rising industrial production began falling sharply under the burden of high wages.

Do you know of any other highly respected model that is so decisively refuted by the data?  In fairness to Eggertsson, he understands that there are a lot of qualifiers in his model.  This is from the first paragraph of his conclusion:

The main problem facing the model economy I have studied in this paper is insufficient demand.  In this light, the emphasis should be on policies that stimulate spending. Payroll tax cuts may not be the best way to get there. The model shows that they can even be contractionary. What should be done, according to the model? Traditional change in government spending is one approach.  Another is a commitment to inflate. Ideally, the two should go together. Government spending has the advantage over inflation policy in that it has no credibility problems associated with it.  Inflation policy, however, has the advantage of not requiring any public spending, which may be at its “first best level” in the steady state of the model studied here.  Any fiddling around with the tax code should take into account that deflation might be a problem.  (Italics added.)

Of course when Krugman discussed Eggertsson’s results, he presented them as facts, with none of the “can,” “may,” or “might” qualifiers used by Eggertsson.  Go reread the first few sentences I quoted from Krugman at the top of this post, to see how he described Eggerttsson’s findings.  So who’s got the most unrealistic model to use for policy evaluation?  Krugman?  Or Mankiw and Barro?  You know where I stand.

PS.  I am partially sympathetic to Ricardian equivalence, so it is possible that the demand-side boost from payroll tax cuts is not all that strong.  But I think these Keynesian models are way off base in their assumptions about monetary policy.  Modern central banks always have some sort of implicit nominal target.  Even the BOJ eventually did enough QE to stabilize their CPI between 2002-08.  And the BOJ is by far the most deflationary central bank in the world.  Under FDR the price of gold provided a nominal anchor, and was raised when deflation got too severe.


Tags:

 
 
 

63 Responses to “Krugman vs. Eggertsson”

  1. Gravatar of Bob Murphy Bob Murphy
    14. December 2009 at 20:06

    Great post Scott. I was getting nervous when you and Krugman agreed.

  2. Gravatar of rob rob
    14. December 2009 at 20:48

    I agree with Bob. Something was wrong with the space-time continuum when you agreed with him.

    Krugman seems hard to figure unless you view him as a pundit at heart. He may be a lot smarter than the average pundit, but he is still a pundit. And pundits don’t really believe anything.

    I can identify with pundits because I don’t really believe in anything either. There are too many convincing contradictory arguments. Currently, I am happy to call myself a Sumnerian (unless I am in polite company.)

    Seriously, it sounds to me like econ theorists aren’t very comfortable with econ theory. Maybe they are afraid, despite their theory, that the facts will conflict. So they hedge their bets. I think Krugman likes to hedge his bets, just like say, Cramer on CNBC.

    As I’ve said a million times: I’m no economist, but I try to follow the most convincing sounding arguments when it comes to my personal investments. So far that has worked for me, although I have no more than broken even for this year. (I made money last year listening to the cynics and shorting the market.)

    I think Krugman is no better than Cramer at this point. He hedges. A year from now he will be able to quote his own column and claim he was right, no matter what happened.

    I respect you more because you don’t have the same luxury.

  3. Gravatar of Don the libertarian Democrat Don the libertarian Democrat
    14. December 2009 at 21:00

    “The main problem facing the model economy I have studied in this paper is insufficient demand.In this light, the emphasis should be on policies that stimulate spending. Payroll tax cuts may not be the best way to get there. The model shows that they can even be contractionary. What should
    be done, according to the model? Traditional change in government spending ( I say Borrowing- Don ) is one approach.Another is a commitment to inflate. Ideally, the two should go together ( NB Don ). Government spending has the advantage over inflation policy in that it has no credibility problems associated with it. Inflation policy, however, has the advantage of not requiring any public spending, which may be at its “first best level” in the steady state of the model studied here. Any fiddling around with the tax code should take into account that deflation might be a problem. In that case, shifting out aggregate supply can make things worse.”

    I could be wrong, but this sounds like QE plus a Reinforcing Stimulus. In other words, the Chicago Plan of 1933, which I advocate.

  4. Gravatar of Mattyoung Mattyoung
    14. December 2009 at 21:17

    Lotta reading, I need more passes through. My first impression is that lowering wages causes the fixed rents on wages to rise in proportion. Hence the household cost of mainly government services seems high, fixed and hopeless.

  5. Gravatar of StatsGuy StatsGuy
    14. December 2009 at 21:23

    Keynes’ “paradox of thrift” was not really meant to describe a scenario with an actively managed non-gold-linked money supply. Your differentiation of it from “savings” is fair, but requires that we separate the purposes of savings into two components – delayed consumption and risk insulation. The Keynesian problem was really the management of risk, even though Keynes does not clearly separate this from the savings/investment function. Notably, individually rational mechanisms to manage risk (e.g. generating liquidity) caused extreme harm in aggregate due to a sort of coordination problem – that someone’s security is bought with someone else’s risk. (This is similar to the problem with national currency reserves, as I’ve argued.) The monetarists observed that central banks could solve that problem, largely by supplying the demand for liquidity with a medium that did not put individuals at risk, but rather put the collective at risk (and this worked, so long as the collective was not in turn exposed to international risk, as it now is via currency runs).

    Also, Krugman’s defense of Keynes’ dismissal of the monetary response rested on a set of monetary tools that was more constrained than we have available today. Yet even then, Keynes proved innovative – the depreciation of the dollar vs. gold was endorsed by Keynes’ crowd. Keynes was not, himself, a pure Keynesian.

    I ran across this link from a fellow at Baseline –

    http://www.nybooks.com/articles/23519

    I’m sure you disagree with much of the economic commentary (so ignore that), but it’s a thoughtful perspective on how the personal lives of individuals like Hayek and Keynes were shaped by local events. After reading this, I think we are rapidly moving toward a Hayekian world, in the sense that individuals will be confronted with massively greater risk and uncertainty that will penetrate into their individual lives (with little insulation from government).

    As much as Keynesianism (or Monetarism, which has been the dominant paradigm since 1980) has been a great experiment that yielded a not-too-bad stretch of prosperity (1945 to 2005), we may be about to engage on a great Hayekian experiment. Hayek’s _political_ beliefs, that strict limits on government power will in the medium/long term improve the stability of liberal (in the classical sense) regimes may finally be tested. And who knows, perhaps it will work.

    Or perhaps it will hasten the decay of the social/political infrastructure. It’s a grand scale historical experiment the likes of which is rarely run (on the same scale as Mao Tze Tung’s experiment), so it’s impossible to predict the outcome. But, we as a society seem to be intent on running the experiment with ourselves as test subjects. The fact that many of us are so eager to try out the Hayekian experiment represents an impressive triumph of pure ideas over pragmatism… We are quite eager to declare the Keynesian/Monetarist experiment a failure, even though (on many dimensions) seems formidable the past 60 years seemed not-too-bad.

  6. Gravatar of StatsGuy StatsGuy
    14. December 2009 at 21:27

    Finally, with regard to the wage-shock data, it would be nice to see the sequencing of monetary stimulus to compare this too. Parsing causality based on this data is hard.

    But I’m no expert on the Depression (like you); I try to stick to policy, trade, IO, and stats. Still I wonder – if you were to think hard to try to come up with a benefit of wage policy on the _heels_ of a 4 year depression that has already decimated production (assuming sufficient monetary action could accompany the wage policy), were there other reasons to favor a wage shock? Distributional, or pragmatic, perhaps?

    I know you will answer that everyone worries about distributional issues far too much – more than overall improvement (which helps everyone). Yet the social/political issues with excessive wealth concentration and the resultant concentration of political power are quite real. So, if you put your own faculties to work, is there anything redeemable about a wage policy in a Depression environment? And if there is something redeemable, how might that redeemable benefit be achieved through a less harmful policy?

  7. Gravatar of MIke Sandifer MIke Sandifer
    14. December 2009 at 21:45

    I wondered about this today when I read Krugman’s post. I wondered why he was talking about the zero rate bound. In my cmmments I asked if he thought penalties on deposits or negative interest rates could work in the current circumstances.

  8. Gravatar of marcus nunes marcus nunes
    15. December 2009 at 04:06

    Krugman “misrepresents” anyone to get his point across. He notes that in 1948, Samelson wrote:”Today few economists regard Federal Reserve monetary policy as a panacea for controlling the business cycle. Purely monetary factors are considered to be as much symptoms as causes, albeit symptoms with aggravating effects that should not be completely neglected”.
    But he neglets to mention that in later editions he wrote
    that: (1967 edition) “monetary policy had an important influence on total spending”. In the 1985 edition, Samuelson and co-author William Nordaus (of Yale) would write, “Money is the most powerful and useful tool that macroeconomic policymakers have,” and the Fed “is the most important factor” in making policy.

  9. Gravatar of Scott Sumner Scott Sumner
    15. December 2009 at 05:42

    Bob, You knew it couldn’t last for long.

    Thanks rob.

    Don, You are right that he does suggest the Chicago plan, which I know you favor. But he also mentions that creating inflation expectations is a first best solution that doesn’t require public spending.

    Mattyoung, Keep in mind that Eggertsson is discussing a payroll tax cut, which lowers the cost of hiring workers, but does not directly lower worker take home pay.

    Statsguy, You said;

    “Keynes’ “paradox of thrift” was not really meant to describe a scenario with an actively managed non-gold-linked money supply.”

    Bingo. I published a paper making that point in 1999, and I think I was the first modern economist to notice this. If you view the supply-side of money as being constrained by the gold standard, then changes in money demand loom very large. And saving can lower interest rates, which leads to more money demand. Hence deflation.

    I agree with the paragraph that follows this quotation. But then you said:

    “Yet even then, Keynes proved innovative – the depreciation of the dollar vs. gold was endorsed by Keynes’ crowd. Keynes was not, himself, a pure Keynesian.”

    Partly true, but this was more a Fisherian policy. Keynes wanted a modest one-time devaluation, and then a return to conventional policy tools. Fisher and Warren wanted to use the dollar price of gold as a policy instrument, and devalue much more sharply than what Keynes preferred. Actually the Fisher approach is in some ways more modern, and corresponds to Svensson’s foolproof escape from a liquidity trap.

    Regarding risk, my impression is that our society is increasingly smothered in insurance, and it is getting worse. Consider the likely health care “reform.” I think people in places like Singapore actually have more security than people in the US, because they save much more. And of course the Singapore government is much smaller than ours. On the other hand perhaps jobs are less secure than they once were, and 401ks are riskier than defined benefit plans. But the apparent security of defined benefit plans was partly a mirage, the risk was simply shifted on to society, where it creates others risks, like the need to bailout failing companies.

    Statsguy#2: You asked:

    “But I’m no expert on the Depression (like you); I try to stick to policy, trade, IO, and stats. Still I wonder – if you were to think hard to try to come up with a benefit of wage policy on the _heels_ of a 4 year depression that has already decimated production (assuming sufficient monetary action could accompany the wage policy), were there other reasons to favor a wage shock? Distributional, or pragmatic, perhaps?”

    I’d say no. Real hourly wages actually rose during the early 1930s, it was real total wages that declined. The NIRA further increased real hourly wages, at the expense of further sharp cutbacks in hours worked. So is was simply making the stylized facts of the GD even worse. Profits fell more than wages, so inequality was the the central problem of the Depression, it was falling total income.

    To answer your other question, I don’t think monetary shocks can explain these results, except that they contributed to the late 1937 recession. But not the 4 I showed in the table.

    Mike, Interestingly, he did mention the interest on reserves in a recent post, but not the idea of negative interest on reserves.

    Marcus, That’s a great point, and I wish I had included it in my post.

  10. Gravatar of Joe Joe
    15. December 2009 at 07:35

    How is inflation targeting any more flexible than interest rate targeting?

    If the Fed pays more than par for a short-term Treasury, regardless of what it’s trying to target, won’t that have the same effect of reducing the desire to lend in either case, causing more delevering and deflation?

    I don’t see how the zero interest rate bound has anything to do with what quantity is being targeted?

  11. Gravatar of Why Keynesians drive Scott Sumner nuts – Economics – Why Keynesians drive Scott Sumner nuts - Economics -
    15. December 2009 at 07:39

    […] December 2009 (268)November 2009 (369)October 2009 (36)September 2009 (397)August 2009 (439)July 2009 (553)June 2009 (503)May 2009 (468)April 2009 (522)March 2009 (178)February 2009 (1)January 2009 (2)December 2008 (1)November 2008 (3)October 2008 (1)September 2008 (3) Scott explains in a thought-provoking post. […]

  12. Gravatar of marcus nunes marcus nunes
    15. December 2009 at 08:53

    Scott
    To get Mankiw´s attention, just keep picking on Krugman!

  13. Gravatar of GS GS
    15. December 2009 at 08:57

    Economics has its head up its ass because it is still suffering from ‘aggregate demand’ retardation. Demand is actually infinite. It is supply that is limited. If you think about the world in barter terms you will understand this.

    Increasing people’s desire to work will lead to more output and more consumption. The ONLY argument for government spending is stickiness of prices. But in general better to remove the minimum wage, fight unions, etc.

  14. Gravatar of Michael Kogan Michael Kogan
    15. December 2009 at 08:59

    I think Krugman’s main point was that whatever happens at zero bound is different from what happens at other times, so you can’t just run a bunch of regressions on the whole sample and claim results. That I think was what he was trying to emphasize. As far as all the qualifiers in the paper – well he did link to it, so anyone curious enough would have seen the qualifiers

  15. Gravatar of Nick Rowe Nick Rowe
    15. December 2009 at 09:10

    Good post Scott. But I wish I understood more clearly *why* wage increases seemed to cause output declines in the 1930’s. And simply saying “labour demand curves slope down” isn’t enough, when aggregate demand may be the problem. Clower-constrained labour demand curves may well be vertical. And we’re back to the old question: what is the shape of the AD curve?

    GS: “Demand is actually infinite. It is supply that is limited. If you think about the world in barter terms you will understand this.”

    Yep, there can never be a shortage of aggregate demand in a barter economy. But we live in a monetary exchange economy. So thinking about the world in barter terms just won’t work, if there is monetary disequilibrium.

  16. Gravatar of Don the libertarian Democrat Don the libertarian Democrat
    15. December 2009 at 09:12

    “Don, You are right that he does suggest the Chicago plan, which I know you favor. But he also mentions that creating inflation expectations is a first best solution that doesn’t require public spending.”

    I actually agree with that. Facing what I thought was a Debt-Deflationary Spiral and for Pragmatic reasons, I favored the Reinforcing Stimulus. But I agree that the first solution should work and would be better in general.

  17. Gravatar of leo leo
    15. December 2009 at 09:13

    Scott,

    Can you expand on your view of Ricardian Equivalence and how it may differ from the mainstream? One heterdox view -the Fiscal Theory of the Price Level (FTPL)- assumes non-ricardian regimes either in the form of Woodford’s bond drop or a Bernanke-esque helicopter drop of money as a means of injecting “net wealth” during recessions. But I also know that one of your favorite economists, Bennett McCallum, rejects FTPL. Where do you stand? If covered in a prior post, please direct me there.

    Thanks, Leo

  18. Gravatar of Doc Merlin Doc Merlin
    15. December 2009 at 09:48

    Hrm, I’d go one step further and say that the “paradox of thrift” doesn’t work any time DD or credit are the main medium of exchange as opposed to paper cash.

  19. Gravatar of Ryan Ryan
    15. December 2009 at 10:08

    GS,

    Great point, and very Austrian. Mises was adamant that in order to understand the monetary system, one first has to understand that value can only relate to other forms of production, not the pass-through medium of exchange.

    Kudos.

  20. Gravatar of pgl pgl
    15. December 2009 at 10:26

    Two simple points: (1) liquidity traps and interest rate targetting by the FED are NOT the same thing; and (2) without even acknowledging the role of the debt deflation hypothesis, you seemingly dismiss this idea. A very long attempt at challenging this paper without ever noting that (1) and (2) above are essential elements.

  21. Gravatar of Nick Rowe Nick Rowe
    15. December 2009 at 10:49

    Doc: why?

  22. Gravatar of spencer spencer
    15. December 2009 at 11:41

    Boy, talk about cherry picking data points to demonstrate your points.

    I’m willing to bet you anything you want that you could not get this analysis published in a refereed journal.

    Why don’t you point out that you picked the absolutely bottom of the 1929-33 recession as your base.

    Moreover, over the four years of your so called shocks — March 1933 to March 1937 industrial production expanded some
    121.5% or at about a 30% annual rate. This is the strongest four year growth of industrial production in US economic history– even more than in WW II.

    So if you are saying wage shocks were responsible for the growth of industrial production under FDR I just wish we could have more wage shocks like this. When you look at the entire period rather than your few cherry picked observations it was the greatest expansion of industrial production ever.

  23. Gravatar of spencer spencer
    15. December 2009 at 11:49

    Boy, talk about cherry picking data points to demonstrate your points.

    I’m willing to bet you anything you want that you could not get this analysis published in a refereed journal.

    Why don’t you point out that you picked the absolutely bottom of the 1929-33 recession as your base.

    Moreover, over the four years of your so called shocks — March 1933 to March 1937 industrial production expanded some
    121.5% or at about a 30% annual rate. This is the strongest four year growth of industrial production in US economic history– even more than in WW II.

    So if you are saying wage shocks were responsible for the growth of industrial production under FDR I just wish we could have more wage shocks like this. When you look at the entire period rather than your few cherry picked observations it was the greatest expansion of industrial production ever.

    In the 12 months to 1938 Industrial output grew 4% and in the 12 months after your so called wage shock it expanded
    23%. In the 12 months after the so called November 1939 wage shock industrial production expanded 13%.

    What you have done is gone through and cherry picked some random noise that is not at all representative of the trend to prove a point. If you were a student in my class I would give you an F – for this paper.

  24. Gravatar of StatsGuy StatsGuy
    15. December 2009 at 12:08

    “Demand is actually infinite.”

    Yes, but we care about relative demand because we are economists. Notably, demand for risk insulation is also infinite, balanced only by our demand for other things. And since risk insulation is provided effectively through highly storable/fungible goods, like money, demand for liquidity grows.

    So the “demand” problem can be perceived as a shift in relative demand from goods/services to liquidity. But here’s the funny thing, in the real world, if the supply of liquidity is reduced under conditions of high contractual debt then the demand for liquidity actually grows. (And I do mean a shift OF the curve, not merely ALONG the curve – if it was merely a shift ALONG the curve then deflation would not be self-reinforcing and currency runs on banks would not happen.)

    Put a different way, in a deflationary economy with high debt and fixed supply of money, Liquidity is a Giffen Good. Once an economy is set upon that trajectory, it becomes self reinforcing – it’s not that the demand for other goods becomes less in an absolute sense (though it might due to psychological/behavioral changes). Rather, it’s that the relative demand for those goods (compared to the demand for liquidity) decreases.

    Telling the story from the Monetarist or Keynesian perspective depends entirely on whether you are talking about an excess RELATIVE demand for money or an insufficient RELATIVE demand for goods/services. The choice of viewpoints does, however, suggest a particular focus on the best instruments for intervention. The latter is entirely an empirical debate.

    But why do I suspect that so many people who favor hard money would strongly prefer the value of their own individual gold stockpile to increase even if that means the world economy implodes and governments collapse? (Indeed, it seems like many gold-lovers would take particular satisfaction in watching society collapse while they indulged in an illusion of personal security.)

    I doubt that Friedman would have endorsed monetary expansion in deflationary environments (a power-give to government) if he did not deem it utterly essential to avoid an even more terrifying risk of social calamity.

  25. Gravatar of ssumner ssumner
    15. December 2009 at 12:23

    Joe, Sorry, but I don’t follow your question. I am not arguing that inflation targeting is more flexible, nor am I advocating inflation targeting. I do think inflation targeting works better than interest rate targeting when you are at the zero bound.

    Marcus, Yes, I just noticed the Mankiw link. I can always “sense” a Mankiw link because lots of new commenters pour in.

    GS, I define AD as NGDP. I think you are referring to Say’s Law, which holds in the long run, and also holds in the short run if monetary policy is targeting expected future NGDP. But if you have bad monetary policy and sticky wages, then you have problems. And we have problems. I am not trying to contradict you, just clarifying what I believe.

    Michael, I agree with you as far as drawing attention to this issue. But I do have several problems with the direction Krugman takes this analysis. Even if he is correct about the econometrics problems at the zero bound, (and it is not at all clear he is, because Eggertsson’s model relies on implausible assumptions) but even if he is right, he provides not evidence that the AD curve is upward sloping, or that payroll tax cuts are deflationary, or that higher minimum wages are expansionary. And yet Krugman is making all these policy claims with:

    1. A model that is highly unrealistic in its assumptions, even at the zero rate.
    2. Massively contradicted by the empirical data.

    So he’s got a highly flawed model, and no evidence. In my book that is a much weaker argument that Mankiw and Barro, especially since I think it is unlikely that in the real world the AD curve slopes upward at zero rates. So I concede that Mankiw and Barro may me wrong, but I still think they are much more likely to be right than Krugman. He acted like they were living in an imaginary world, but just take a look at the assumptions required to make Eggertsson’s model work.

    Nick, You asked;

    “Good post Scott. But I wish I understood more clearly *why* wage increases seemed to cause output declines in the 1930’s. And simply saying “labour demand curves slope down” isn’t enough, when aggregate demand may be the problem. Clower-constrained labour demand curves may well be vertical. And we’re back to the old question: what is the shape of the AD curve?”

    If you read Eggertsson’s 2008 AER piece you will find the answer (or an answer) to your question. In Eggertsson’s paper he acknowledges that the upward sloping AD curve argument doesn’t hold if the Fed is targeting inflation. And starting in March 1933 FDR started targeting inflation by manipulating the price of gold. And this is (according to both Eggertsson and me) what triggered the recovery. So under this assumption you’d expect AD to be downward sloping, and any decrease in SRAS, such as a huge minimum wage increase, will reduce output. The intuition is that any tendency for the higher minimum wage raise AD is offset by monetary policy that is targeting the price level. BTW, even without FDR’s program, the gold standard itself provides a sort of nominal anchor. And this means that an upward sloping AD curve is rather unlikely even if the central bank is passive, but you still have a gold standard. (It depends on other factors like how well PPP works, how stable the real value of gold is, etc.)

    Don, OK, I’m fine with that.

    Leo, The FTPL isn’t my area, but my intuition agrees with McCallum. I think currency is very different from other financial assets, it is more like paper gold, and the market for currency determines the price level just as the market for gold would under the gold standard. The FTPL people accept this argument for gold, but think currency is just another type of bond. I think the FTPL only holds when the central bank is subserviant to the fiscal policymakers, which is not the case in most developed countries.

    Ricardian equivalence is different. It’s one of those theories that seems implausible, but has a surprising amount of empirical support. My hunch is that it is something between 1/2 and 3/4 true, but I am no expert. Maybe I’ll say more when I have more time.

    Doc Merlin, I agree with Nick’s final question–it’s not clear that the type of money matters. I really think the Paradox of Thrift is just a roundabout way of addressing money demand shocks. (Or velocity shocks, using monetarist terminology.)

    pgl, I wasn’t trying to compare the liquidity trap to interest rate targeting. I think that the liquidity trap shares some similarity with interest rate pegging, which is very different from interest rate targeting. Under interest rate pegging (say 1942-51) rates are kept fixed, which can be destabilizing. Targeting can include policies such as the Taylor Rule. My point about interest rate targeting is that it is not a very good policy tool once rates hit zero.

    I am not sure what your debt deflation comments refers to. Look, I agree that if we get deflation, we have major problems, debt deflation or not. Sticky wages are reason enough to oppose deflation, and I am one of the strongest opponents of deflation in the blogosphere. I have been advocating more monetary stimulus all year. But in this post I was arguing against the view that payroll tax cuts are deflationary. I did point out that Eggertsson is a good economist, and within the assumptions of his model he is right. But I also spelled out which asumptions I thought did not correspond to the real world, and presented some extremely relevent empirical evidence relating to the hypothesis. Neither Krugman nor Eggertsson presented any empirical evidence. So I think I was fair.

  26. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    15. December 2009 at 13:36

    Scott, pgl is Pro Growth Liberal from:

    http://econospeak.blogspot.com/

    He’s almost 100% left wing politics, and his knowledge of economics brings to mind that famous Dorothy Parker line about a flea’s navel and caraway seeds.

  27. Gravatar of Doc Merlin Doc Merlin
    15. December 2009 at 14:24

    Ok Nick, I’ll explain why I think so.

    First a term: I am using cash here, meaning what constitutes the majority of the monetary base. I don’t mean just paper currency.

    Lets assume you do have a gold backed currency and a banking system with a fairly high reserve requirement (so that cash dominates instead of DD or credit, and because of the high reserve requirement DD’s act more like cash).
    People for some reason start saving more. Naturally this increases demand for money which increases the “price” of money in terms of goods. This is textbook deflation.

    Now to make the economy dominated by DD and credit instead of cash by reducing the importance of cash in transactions. One easy way to do this is by removing reserve requirements, but . For this example, we keep the gold peg but “the paradox of saving” still disappears.
    Ok, we remove the reserve requirement so the only thing keeping banks from over/under-inflating is their own profit/loss calculations and their fear of default. This creates a system where a larger percent of the money is bank created credit and DD. If the demand for money rises (or the demand for credit falls), banks will either drop their own reserve ratio to sell more credit (and lower their rates) or will use their own DD creation to buy up currently low-price, non-gold assets. Either will lower the price of money again.

    I am not arguing that deflation and the business cycle wouldn’t still happen in either case (other things can cause it). Its just in an economy thats is heavily dominated by DD and bank credit creation, the cycle just wouldn’t be savings driven.

  28. Gravatar of spencer spencer
    15. December 2009 at 14:40

    I will repeat my earlier point in as simple a term as I can.

    Higher wages did NOT, I repeat higher wages did NOT cause lower output in the 1930s.

    I challenge any of you making that claim to show the results of a regression of wage growth against industrial production growth in the 1930s. Every regression I have ever tried, with all kinds of lags and leads always found a strong positive correlation between wage growth and output in the 1930s.

    I double dog dare a single one of you to show a single regression equation that demonstrates a negative correlation between wage growth and industrial production, employment or any other measure of output growth from 1929 to 1939.

    The evidence is overwhelming that your claim that higher wage growth had a negative impact on employment and/or growth is completely and utterly wrong.

    Again, I will repeat that if the original paper had been turned in to my class I would have given it an F minus and suggest the student change their major to art history.

    I will wait for a single one of you to display any decent econometric evidence that there was a negative relationship between wages growth and output growth in the 1930s.

    I am willing to bet any sum any of you wish to venture that you can not do it.

  29. Gravatar of David Beckworth David Beckworth
    15. December 2009 at 14:46

    Spencer,

    Where would one get data to do such a regression? Thanks.

  30. Gravatar of Doc Merlin Doc Merlin
    15. December 2009 at 14:56

    Ok, my last argument was somewhat convoluted, here is a better phrasing.
    Scott, another way to phrase my argument is that if bank DD and credit is the main form of money, then savings means giving the money to a bank. The bank then uses its multiplier it to create more money. This offsets a fall in V with an increase in M.

  31. Gravatar of faithkills faithkills
    15. December 2009 at 15:24

    Higher wages did NOT, I repeat higher wages did NOT cause lower output in the 1930s.

    Oh but they did, didn’t they?

    He didn’t ‘cherry pick’ dates, the wage shock policies picked the dates.

    So would you assert that IP would have been lower over the surrounding decade without those shocks and their associated dips?

    Because this would be great news for the current administration if so. We could just fix our IP slump by mandating wage increases!

  32. Gravatar of Scott Sumner Scott Sumner
    15. December 2009 at 17:42

    Patrick, Well I am a pro growth neoliberal, so we are just a “neo” apart!

    Doc Merlin, I think all your discussion of DDs, etc, misses a key point. If people save more interest rates will tend to fall. And that is true whether most money is in currency or checking balances. And if interest rates fall the demand for base money rises. And this is deflationary.

    Spencer, I’ll meet your “double dog dare.” Here is a regression:

    Table 2.1 The Relationship Between Industrial Production and Its Lagged Value, Nominal Wages, Wholesale Prices, 1920-39, monthly.

    Period
    Independent
    Variable 1920-39 1920-29 1930-39

    DLY-1 .509 .345 .543
    (9.09) (3.73) (7.43)

    DLW -.459 .124 -.697
    (-3.31) (0.60) (-3.84)

    DLP .624 .476 .898
    (5.59) (3.91) (4.56)

    Adj. R2 .453 .374 .518

    n 227 107 119 __________________________________________________________________
    Note: The variables are defined as follows: DLY and DLY-1 are the first differences of the natural log of industrial production, and its first lag. DLRW and DLW are the first differences of the natural logs of real and nominal wages of production-workers, respectively, and DLP is the first difference of the natural log of the wholesale price index. All equations included a constant term. T-statistics are in parentheses. A regression of the residuals on the lagged residuals, and other independent variables, showed no evidence of serial correlation.

    If the table get garbled up in posting, I’ll explain it in another post. The reason the 1930s are significant, but not the 1930s, is that in the 1930s there were governmental attempts to manipulate the wage rate, to move it away from its equilibrium value. These autonomous wage shocks had a strongly negative impact on output.

    Doc Merlin#2, I suppose that is possible, but you’d have to consider a lot of factors. Are DDs a very widespread form of saving?

  33. Gravatar of Scott Sumner Scott Sumner
    15. December 2009 at 17:47

    Spencer, Just as I feared, the table is hard to read. Each line shows a coefficient for 1920-39, then 1920-29, then 1930-39.

    The one to focus on is the third regression (1930-39) and the coefficient on wages, which is negative 0.697. Its T-stat is (-3.84) BTW, real wages were also highly countercyclical during the 1930s, whether deflated by the WPI or the COL (cost of living.) You must use monthly data; annual data are worthless.

  34. Gravatar of Scott Sumner Scott Sumner
    15. December 2009 at 18:08

    I missed the first Spencer comment:

    “Boy, talk about cherry picking data points to demonstrate your points.

    I’m willing to bet you anything you want that you could not get this analysis published in a refereed journal.”

    You lose. It was published here:

    Silver, Stephen and Sumner, Scott. 1995. “Nominal and Real Wage Cyclicality During the Interwar Period,” Southern Economic Journal, (January): 588-601.

    Since you are willing to bet me anything, I chose to bet your entire stock of wealth against mine. Please send the check to my office at Bentley University. Use certified mail. And since you also did a double dog dare, you owe me two dogs, preferably expensive purebreds.

    Statsguy, You said;

    I doubt that Friedman would have endorsed monetary expansion in deflationary environments (a power-give to government) if he did not deem it utterly essential to avoid an even more terrifying risk of social calamity.

    Very good point. And I would add that during calamities there is usually a rise in the power of government, so libertarians should strongly oppose the sorts of deflations that led to increased power of the state in the US after 1933, or in Argentina (after 2002), not to mention the infinitely more calamitous government that took power in Germany after four years of deflation (in 1933.)

  35. Gravatar of FT.com | Clive Crook’s blog | Further reading FT.com | Clive Crook's blog | Further reading
    15. December 2009 at 18:24

    […] Krugman v Eggertsson. Scott Sumner, The Money Illusion (via Greg Mankiw) […]

  36. Gravatar of marcus nunes marcus nunes
    15. December 2009 at 18:36

    Spencer
    I don´t think Scott was “cherry picking” data points at all.
    I have the pictures in front of me. Even if you check the level (not growth rate) of IP between 1929 and 1940, you will observe a peak at the moment of each of the wage shocks.
    Yes, IP went up by 120% between 3/33 and 3/37 but it would have likely performed much better if it weren´t for the wage shocks. For example, between july 33 and nov 34, a period encompassing 2 wage shocks, IP fell by 15%.
    By early 1937, IP had just surpassed the level seen before the 1929 crash. Another wage shock coupled with restrictive MP (increase in RR)hammered IP which fell by 32% between march 37 and june 38.
    MB rose 61% between june 38 and end 1940. IP thanked and increased by 71% during that period despite having to confront 2 wage shocks on the way! And over that 1 and 1/2 year period prices were flat (zero inflation). Talk about a flat SRAS curve!

  37. Gravatar of StatsGuy StatsGuy
    15. December 2009 at 19:56

    bah… I was really hoping you would tear apart my Giffen Good analogy for technical innaccuracy.

    As an aside on the regression – though I generally agree with your point – using industrial production probably magnifies the coefficients (as compared to other dependent variables). Per your earlier point, did you model the data with with total consumption (or a proxy) as the dependent variable? I suspect you’d get a similar result, but perhaps not quite as strong.

    Industrial producers could be using inventory accumulation/depletion to time-shift production (and exert bargaining pressure on wages). I’m not sure, but I don’t think the residual-on-lag-residual test would necessarily catch this if wages in subsequent periods were bargained down by low production in the prior period (e.g., reverse causation from the previous period). Imagine a strike/lockout/concession cycle. If you had that sort of endogeneity, you could get a magnified effect simply by tugging back and forth around a trend. The shortness of the time interval in which you define a data point can reinforce this, but would still not register as “serial correlation” due to the inclusion of a lag as an independent variable. This is more a concern about the magnitudes you’re estimating than the sign or significance.

    The natural experiment test is better, even with the timing and control issues, although it would be overly dramatic if producers anticipated a wage spike (allowing them to accumulate inventory beforehand), or anticipated concessions after the spike (allowing them to deplete inventories in expectation). I don’t know the era well, which is why I ask without making any assertions. I can only comment on the method.

  38. Gravatar of Jim Glass Jim Glass
    15. December 2009 at 21:16

    On a related note:

    Here’s a 1933 newsreel explaining to the public the departure from the gold standard — and need for inflation.

    “Inflation seen as the nation’s salvation! But what does it mean to you? And you?… ”

    At Youtube.

    Just in case anyone thinks inflating wasn’t overt policy in 1933.

    Compare how it’s explained to what’s in the textbooks today.

    Time marches on!

  39. Gravatar of TGGP TGGP
    15. December 2009 at 23:45

    Will Ambrosini discussed the upside-down world of depressions according to Eggertsson here.

  40. Gravatar of Doc Merlin Doc Merlin
    16. December 2009 at 02:29

    @Scott:
    Yes, DDs are a substantial form of short term savings. When you add up Checking accounts, savings accounts, sweeps, money market accounts such you end up with DD’s ~ MZM, so about 9.5T USD

    They are not as substantial as long term savings in real estate, or stocks + bonds, but those count as investment, not savings in economic models.

  41. Gravatar of marcus nunes marcus nunes
    16. December 2009 at 03:36

    Casey Mulligan deflates the Paradox of Toil:
    http://economix.blogs.nytimes.com/2009/12/16/no-new-paradox/#more-45083

  42. Gravatar of spencer spencer
    16. December 2009 at 05:39

    You can find the data in Historical Statistics of the US.

    % ch: %CH
    annual ind.
    earnings prod.
    dec/dec
    1929…16.5…. -2.5
    1930…-1.4…. -27.9
    1931…-8.0….-13.2
    1932…-12.1…-44.6
    1933…-6.7….+14.9
    1934….3.4….+5.2
    1935….5.1…..32.1
    1936….4.1…..25.0
    1937….6.2…..-20.0
    1938….-2.2….-24.1
    1939….4.4……5.3

    What wage shock. In 1933 wages fell 6.7% and
    in 1934 they rose 3.4%.

    You have a positive correlation and if you regress the data you get that output is positively related to wages.

    The data completely disproves the thesis that low wages leads to higher output.

  43. Gravatar of Scott Sumner Scott Sumner
    16. December 2009 at 05:42

    Statsguy. That’s a good point, but I simply don’t know whether consumption data was available monthly, or even quarterly. I might add that the pattern of inflation after these wage hikes also tends to conflict with the Eggertsson model, so that’s at least one more piece of evidence, albeit not as strong.

    Sorry I missed the “Giffen good” idea. That sounds like a really clever analogy to me. Do you know if anyone has discussed it before? If not, you should write a short paper on it.

    Jim Glass, I may have done a post on that newsreel a while back.

    Thanks TGGP.

    Doc Merlin. Two points:

    1. There are no reserve requirements on most of the assets you mention.

    2. I have some savings in DDs, but at the margin any increases go into higher yielding accounts. I just keep enough in DDs for transactions purposes. Am I typical?

    Marcus, Yes, the Mulligan evidence is relevant. Indeed there is so much evidence against high wage policies that one hardly knows where to start. France tried a high wage policy in the late 1930s, and it failed badly. Germany went with a low wage policy and it recovered faster than other European countries. (Of course the German government had some other policies I just as soon avoid.)

  44. Gravatar of spencer spencer
    16. December 2009 at 05:49

    P.S. My net worth is approximately -$50,000.

    I’ll expect a check in the mail for that.

  45. Gravatar of scott sumner scott sumner
    16. December 2009 at 06:20

    Spenser, Your data actually supports my argument. The right variable is hourly wages, not annual wages. When the hourly wage rises do to government policies, hours worked falls. Betweeen July and September 1933 the hourly wage rates were raised by about 20% by government fiat. The reason total earnings didn’t change much in 1933 and 1934, is that as hourly wages rose, hours worked fell. Which is exactly my point.

    Spenser#2, OK, but do I still get my two dogs?

  46. Gravatar of Doc Merlin Doc Merlin
    16. December 2009 at 08:13

    1. Scott, of course your economic behavior isn’t typical, you are an economist and I wager you make significantly above the median. Most americans (and roughly 1/3 in the above 100kpy income bracket) live paycheck to paycheck.

    2. Also, noobie mistake on my part, money in DD=MZM-AMB, not MZM. so closer to 7.4T than 9.5T

    3. DD’s do have reserve requirements. Depending on the net dollar worth of the institution’s DD’s. It is only zero % for institutions with less than 10.7M in net DD’s and for eurocurrency liabilities. http://www.federalreserve.gov/monetarypolicy/reservereq.htm
    It calls the accounts transactional accounts, but in a footnote it defines them the same way I did DD’s earlier in the thread.

  47. Gravatar of ssumner ssumner
    16. December 2009 at 13:23

    Doc Merlin, Is it true that most Americans have no financial assets other than checking accounts? That’s pretty pathetic given that we are essentially the richest country in the world (out side a few oil producers.) Even worse, we have much lower taxes than the Europeans. And our housing is cheaper. And the prices in our stores are lower. I’d like to see some data before I accept that.

  48. Gravatar of faithkills faithkills
    16. December 2009 at 14:20

    You have a positive correlation and if you regress the data you get that output is positively related to wages

    In an economy based on fiat inflation over any long period you can generally correlate increased wages with anything under the sun.

    I double dog dare you to show negative correlation of wages to global warming. You can’t! Therefore wages cause global warming. QED.

  49. Gravatar of happyjuggler0 happyjuggler0
    16. December 2009 at 17:14

    I like spencer, but like everyone else he has his normative biases that color his judgment. No doubt he found his way here, directly or indirectly, because Mankiw linked to the post. I think he has about year’s worth of post’s to read through though or he is likely to get in more trouble.

    He should probably also avoid the phrase “double dog dare you” lest people think he is a buffoon.

    faithkills,

    :)))

  50. Gravatar of Thorstein Veblen Thorstein Veblen
    16. December 2009 at 21:08

    A couple things — when you hit the zero lower bound, Monetary policy does become much less effective than it is in normal times. This is what Krugman and Eggertson believe. They both also believe that the Fed should still do large amounts of QE, just to do what a small rate cut would do in normal times.

    My biggest problem with the post was “There is still inflation targeting, and elsewhere in the paper (which claims to provide theoretical support for fiscal stimulus through more government spending), Eggertsson offhandedly admits that inflation targeting can work, indeed admits that it is the first best policy, even better than fiscal policy.”

    Inflation targeting? Hasn’t Bernanke and the Fed been targeting 2% inflation? Didn’t seem to prevent unemployment from rising to 10%. So, your idea is that the Fed announces it’s going to have more inflation, w/out doing anything, and, poof, the economy recovers? This is ridiculous. Don’t get me wrong, inflation should be targeted, but it’s only slightly better than twiddling ones thumbs. The correct policy solution is another $2 trillion of QE.

    A few years ago, when Japan announced an inflation target, Blanchard wrote, excitedly, in his undergrad textbook, that, due to the target “economists were very optimistic about Japan.”

    As we know, of course, that optimism was misplaced…

  51. Gravatar of Scott Sumner Scott Sumner
    17. December 2009 at 05:54

    Thorstein Veblen, As far as what Krugman “believes” that would seem to depend on the day of the week. Sometimes he says QE can work, other days he says it can’t “Period. End of Story.”

    The mistake that both Krugman and Eggertsson make is to assume that a liquidity trap is just some sort of unfortunate calamity that befalls a country, which prevents expansionary monetary policy, when it is actually a symptom of very tight monetary policy.

    Zero rates don’t really make monetary policy more difficult, they make interest rate-oriented monetary policy more difficult. That’s what the “What bugs me” line was all about. Permanent QE is just as effective as ever. Exchange rate depreciation is just as effective as ever, inflation targeting is just as effective as ever, NGDP targeting is just as effective as ever, commodity price targeting is just as effective as ever.

    You said;

    “Inflation targeting? Hasn’t Bernanke and the Fed been targeting 2% inflation?”

    Actually they haven’t been targeting inflation, as their mandate doesn’t allow it. (They’ve hinted that they prefer a rate of roughly 2%, which is a very different thing. It doesn’t require that they make up for past mistakes.) I should have said price level targeting, as you actually need some sort of level targeting regime (as Woodford recently emphasized.) If you aim for 2% inflation, and get 0%, then you have to aim for 4% next year. Policy needs a memory. Obviously the Fed is not doing anything remotely like that, although Bernanke did recommend that policy for Japan. BTW, Krugman also believes inflation targeting will work at the zero bound.

    [I should add that most economists who say they favor inflation targeting, don’t really favor strict inflation targeting, as they believe supply shocks should be accommodated. That’s why I favor NGDP targeting, level targeting.]

    Indeed not only is the Fed not trying to raise inflation expectations, but they have been actively trying to lower them, which is the worst possible thing you can do in a liquidity trap. So that fact refutes your argument that they are targeting inflation at 2%.

    I favor NGDP targeting, level targeting, which I believe is consistent with the Fed’s Congressional mandate.

    I had read that the Japanese had a target of stable prices. Is that wrong? Nobody in their right mind would think the BOJ is targeting positive inflation. Several times during the 2000s they tightened policy while the GDP deflator was still falling. Even if the BOJ had a stable GDP deflator target, they should be aiming for 12% inflation next year, because the GDP deflator has fallen that much since 1994. Obviously they aren’t aiming for 12% inflation. And surely I can’t be blamed for silly comments made by Blanchard. If the BOJ wanted higher prices, they would have never let the yen soar in value, so nobody who knows what’s really going on in Japan can take seriously the notion that the BOJ is in some sort of “trap.”

  52. Gravatar of Doc Merlin Doc Merlin
    17. December 2009 at 09:14

    1. It really depends on how you define “savings.” If you mean a 401k or a pension plan, then yes most americans don’t have one.
    This study talks about participation rates in employer provided plans, which is roughly 55% for salaried, full-time workers.
    http://www.ebri.org/pdf/briefspdf/EBRI_IB_11-2009_No336_Ret-Part.pdf

    If you mean paying down a mortgage as “saving” then yes a lot of americans have savings outside of DD. Oddly enough, mortgaging one’s home is an extremely common way to startup a small business.

    2. I said DD’s and defined them to include any account that was available on demand, including savings accounts, checking accounts, and money market accounts.

    3. Savings rate has been correlated to culturally be inversely proportional to that culture’s recent wealth. De Soto casually mentions in his book “The Mystery of Capital” that third world countries have a very, very high savings rate, and lots of money. What they lack is access to formal markets, and legal systems that allow them to legally exchange their property openly.

  53. Gravatar of Doc Merlin Doc Merlin
    17. December 2009 at 13:56

    On a non-physics point:
    This year we haven’t really had price deflation. That was the second half of last year. This year, CPIAUNCS has been growing pretty much the whole year, and PPI has been mostly growing since March.

    CPI is now about 3% above what it was at the start of the recession, and PPI is almost at the level it was at the start of the recession.
    Here is the plot:

    http://picasaweb.google.com/lh/photo/KhkqgfKdkHsxO73B3AL93w?authkey=Gv1sRgCOSj8aqKqty4lwE&feat=directlink

  54. Gravatar of Doc Merlin Doc Merlin
    17. December 2009 at 14:00

    DOH! I posted to the wrong, thread, this was supposed to go in the universe deflation thread.
    *Goes to post there.*

  55. Gravatar of ssumner ssumner
    18. December 2009 at 07:24

    Doc Merlin, Those are good points about saving. I answered the CPI question in the other thread.

  56. Gravatar of Felix Felix
    19. December 2009 at 09:36

    ssumner said:
    “Zero rates don’t really make monetary policy more difficult, they make interest rate-oriented monetary policy more difficult. That’s what the “What bugs me” line was all about. Permanent QE is just as effective as ever. Exchange rate depreciation is just as effective as ever, inflation targeting is just as effective as ever, NGDP targeting is just as effective as ever, commodity price targeting is just as effective as ever.”

    The real conflict at the zero lower bound is discretionary vs. time-consistent optimization. Suppose zero inflation is optimal. Then, when you are in the ZLB, you want to create higher expected inflation in order to end the ZLB. Once you are out, you want lower inflation.

    Monetary policy at the ZLB is more difficult because policymakers have to be aware of the conflict and be able to credibly commit to time-consistent optimization. At the ZLB, all policies that approximate time-consistency (like a level target) or that make it costly for the monetary authority to abolish its prior commitment (such as buying inflation linked bonds, which lose value upon lower inflation) can be helpful, while those that don’t (such as inflation-rate targeting) fail.

    If commitment to set future short-term interest rates is possible, then working with the interest rate in a ZLB is effective.

  57. Gravatar of ssumner ssumner
    19. December 2009 at 12:10

    Felix, Good point, and my new post replying to Ambrosini may help explain my views here.

    But if you are good at economics (which you seem to be) you might be better off reading Woodford’s recent defense of level targeting, as he is a more competent theorist than I am.

    The intuition is that you need level targeting, or what I call a nominal trajectory, or nominal target path. If you do a “memory-less” inflation targeting, then I think there may be exactly the problems you describe.

    But I think the so-called time inconsistency problem is overrated. Central banks care a lot about their reputation. They don’t like to break promises. That is why they are reluctant to set specific goals.

  58. Gravatar of Felix Felix
    19. December 2009 at 12:41

    You said:
    “But I think the so-called time inconsistency problem is overrated. Central banks care a lot about their reputation. They don’t like to break promises. That is why they are reluctant to set specific goals.”

    I don’t think that central banks being incredible is a problem, either. But in order to make a sensible commitment, central bankers first have to understand the problem. Central bankers today understand that it is generally important to promise to keep inflation low. But they don’t understand that, in a ZLB situation, it is optimal to promise (and deliver, because you can’t permanently fool the market) higher-than-optimal inflation.

  59. Gravatar of ssumner ssumner
    21. December 2009 at 13:44

    Felix, I basically agree, although once again I think the inflation target needs to be even further above optimal under a memoryless inflation target, than under a price level target.

  60. Gravatar of The Failure of Modern Macroeconomics « The Everyday Economist The Failure of Modern Macroeconomics « The Everyday Economist
    15. March 2010 at 20:34

    […] argue that it suggests that liquidity traps don’t exist. Put differently, as Scott Sumner has suggested: Zero rates don’t really make monetary policy more difficult, they make interest rate-oriented […]

  61. Gravatar of TheMoneyIllusion » Comparing the Great Contraction and the Great Recession (1932, pt 5 of 5) TheMoneyIllusion » Comparing the Great Contraction and the Great Recession (1932, pt 5 of 5)
    2. April 2010 at 13:33

    […] monetary policy (the Swedish kroner depreciation that Yglesias mentioned.)  In contrast, Krugman has argued that lower unemployment benefits and reduced labor costs may reduce AD.  And those effects […]

  62. Gravatar of Money as the Ultimate Giffen Good « The Baseline Scenario Money as the Ultimate Giffen Good « The Baseline Scenario
    15. August 2011 at 04:01

    […] (and by extension, US debt), is a Giffen good.  This is an argument I made to Scott Sumner in 2009 here and in emails.  Recently Eric Falkenstein made it here.  Simply, as the price goes up, […]

  63. Gravatar of TheMoneyIllusion » Aggregate supply really does matter TheMoneyIllusion » Aggregate supply really does matter
    9. February 2012 at 06:11

    […] couple years ago I did some posts strongly criticizing the Krugman/Eggertsson argument that the AD curve might slope upward when the […]

Leave a Reply