How did Friedman and Schwartz persuade us?

This is ambitious post.  Quite frequently people ask me to provide data supporting my hypothesis that the Fed (and ECB and BOJ) caused the crash of 2008.  Whenever I get that question I always have a sinking feeling, a feeling that people don’t really understand what I am driving at.  In the first part of this post I will try to explain the success of Friedman and Schwartz’s Monetary History of the U.S. But the real goal will be to use that explanation to help you see why if you are asking for data, you probably don’t see the problem as I do.

At the time F&S’s Monetary History was published in 1963, very few people thought the Fed was responsible for the Great Depression.  It was the heyday of Keynesian economics, when investment instability was viewed as a more significant problem than central bank incompetence.  And the Fed had dramatically increased the monetary base and cut the discount rate to a very low level during the early 1930s, so monetary policy certainly didn’t seem contractionary.  Yet within two decades F&S had convinced most of the profession that the Fed, or at least monetary policy broadly defined, was to blame for the Great Contraction of 1929-33, and to a lesser extent the 1938 depression as well.  How did this happen?

Like most economic histories, F&S had an ulterior motive—their goal was not merely to re-evaluate the Depression, but also to provide persuasive evidence for their (monetarist) approach to contemporary policy issues.  I think most people would agree that they failed in their attempt to displace Keynesianism with monetarism.  Today, relatively few elite macroeconomists call themselves monetarists.  On the other hand (as Brad DeLong pointed out) new Keynesian economics did incorporate some monetarist insights.  But one insight that did not get included was the view that the monetary aggregates were the proper indicator of the stance of monetary policy—which is the single most important argument in the Monetary History.  So we still don’t know why the book was so influential.

Just to be clear, I am not arguing that the profession accepted all of the F&S explanation, but merely the core idea that monetary policy failure lay at the heart of the Great Contraction.  There are still important differences over whether the Fed committed errors of omission or commission, and the extent to which the international gold standard limited the Fed’s ability to act.  But even the decision to adhere to the gold standard was in a broad sense “monetary,” and by the early 1980s almost all debate about the Great Contraction focused on the role of the Fed (and other major central banks.)  Even that was a major victory for F&S.

The best way to understand why F&S’s book became the most influential economic history ever written; is to look at the following random selection of data from an old Robert Barro textbook:

Country  Money growth  Real GDP growth   Inflation  Time period

Brazil               77.4%                     5.6%                    77.8%        1963-90

Argentina        72.8%                     2.1%                    76.0%        1952-90

Chile               47.3%                     3.1%                    42.2%        1960-90

Israel               31.0%                     6.7%                    29.4%        1950-90

S. Korea           22.1%                     7.6%                    12.8%        1953-90

Iceland            18.4%                     4.3%                    18.8%        1950-90

Portugal          11.5%                     4.7%                      9.9%        1953-86

Britain               6.4%                     2.4%                      6.5%        1951-90

U.S.                   5.7%                     3.1%                      4.2%        1950-90

Switzerland       4.6%                     3.1%                      3.2%        1950-90

The period from 1950-90 is actually two separate periods.  When the Monetary History was published in 1963, most developed countries still had relatively low and stable inflation rates.  By the late 1970s, however, even developed countries were beginning to show dramatic variation in inflation rates.  From 1974-79 Germany averaged 4.3% inflation, France averaged 10.4%, whereas Britain averaged 16.3%. At the other extreme, between 1975-79 inflation in Switzerland averaged only about 2%.

It is very easy to look at this sort of data and come to the conclusion that monetary growth drives inflation.  I don’t have interest rate data, but even if I did it wouldn’t show much about monetary policy, as the Fisher effect completely dominates the liquidity effect when looking at long run differences in trend inflation.  Thus economists began to think of trend inflation in a way that Keynes himself had never considered, as something that could be selected by central banks as casually as one might order an entree off a restaurant menu.  By the early 1980s the major central banks looked around the world and said to themselves (to paraphrase a line from When Harry Met Sally) “I’ll have what Switzerland’s having.”

This fundamental change in the way that people thought about monetary policy was a very lucky break for F&S.  Now the relevant policy counterfactual was not an alternative path for nominal interest rates, but rather an alternative inflation target.  By the 1970s we were in a fiat money world and almost everyone just sort of assumed (if only subliminally) that the central bank could have whatever money growth they preferred.  In addition, since variations in real GDP growth were small compared to variations in inflation rates, money growth also seemed to explain the path of NGDP growth.  If NGDP fell roughly in half during the early 1930s, then obviously monetary policy was too contractionary (or so it seemed.)

When F&S’s very carefully constructed time series for M1 and M2 showed precipitous declines between 1929 and 1933, it was essentially game over.  By the late 1970s (when I was in grad school) even new classical economists such as Robert Lucas would simply refer to Friedman and Schwartz’s study, if anyone questioned whether monetary shocks had real effects.

Here I’d like to argue that there are two Monetary Histories; the one in our imagination, and the (much different) one written by F&S.  The one in our imagination is in some way’s much stronger, and in other ways much weaker, than the actual Monetary History.  I’ll begin with Krugman’s view that the Monetary History doesn’t really show what it purports to show, and especially what Friedman later would claim it shows.  Then I will partially defend F&S from Krugman’s assertions.  Finally I will argue that my account of how the Monetary History became accepted, makes my view of the Fed policy during the recent crash seem much more plausible, indeed almost inevitable.

Krugman’s Critique of the Monetary History

Right after Friedman died Krugman wrote a relatively harsh appraisal of his career in the New York Review of Books.  Here is his most important accusation:

Although A Monetary History is a vast work of extraordinary scholarship, covering a century of monetary developments, its most influential and controversial discussion concerned the Great Depression. Friedman and Schwartz claimed to have refuted Keynes’s pessimism about the effectiveness of monetary policy in depression conditions. “The contraction” of the economy, they declared, “is in fact a tragic testimonial to the importance of monetary forces.”

But what did they mean by that? From the beginning, the Friedman-Schwartz position seemed a bit slippery. And over time Friedman’s presentation of the story grew cruder, not subtler, and eventually began to seem””there’s no other way to say this””intellectually dishonest.

In interpreting the origins of the Depression, the distinction between the monetary base (currency plus bank reserves), which the Fed controls directly, and the money supply (currency plus bank deposits) is crucial. The monetary base went up during the early years of the Great Depression, rising from an average of $6.05 billion in 1929 to an average of $7.02 billion in 1933. But the money supply fell sharply, from $26.6 billion to $19.9 billion. This divergence mainly reflected the fallout from the wave of bank failures in 1930-1931: as the public lost faith in banks, people began holding their wealth in cash rather than bank deposits, and those banks that survived began keeping large quantities of cash on hand rather than lending it out, to avert the danger of a bank run. The result was much less lending, and hence much less spending, than there would have been if the public had continued to deposit cash into banks, and banks had continued to lend deposits out to businesses. And since a collapse of spending was the proximate cause of the Depression, the sudden desire of both individuals and banks to hold more cash undoubtedly made the slump worse.

Friedman and Schwartz claimed that the fall in the money supply turned what might have been an ordinary recession into a catastrophic depression, itself an arguable point. But even if we grant that point for the sake of argument, one has to ask whether the Federal Reserve, which after all did increase the monetary base, can be said to have caused the fall in the overall money supply. At least initially, Friedman and Schwartz didn’t say that. What they said instead was that the Fed could have prevented the fall in the money supply, in particular by riding to the rescue of the failing banks during the crisis of 1930-1931. If the Fed had rushed to lend money to banks in trouble, the wave of bank failures might have been prevented, which in turn might have avoided both the public’s decision to hold cash rather than bank deposits, and the preference of the surviving banks for stashing deposits in their vaults rather than lending the funds out. And this, in turn, might have staved off the worst of the Depression.

Intellectual dishonesty?  Hmmm.  But let’s move on.  The first thing that should be said is that it is not clear that Krugman’s accusation is correct.  F&S do clearly argue that had the Fed never been created, the initial banking crisis of late 1930 might well have been handled like the 1907-08 crisis, through temporary currency restriction.  As F&S put it:

“As it was, the existence of the Reserve System prevented concerted restriction, both directly and indirectly.”  (p. 311)

A more fundamental problem with Krugman’s argument is that he grossly oversimplifies the concept of ‘causation.’  Readers of my blog probably noticed that I believe the only coherent way of thinking about causation is in terms of plausible policy counterfactuals.  There is no single “baseline” monetary policy to use in making inferences about causality.  Under the gold standard the price of gold was a sort of policy decision.  The only other purely discretionary tool available to central banks was the gold reserve ratio—and if one used that ratio then the Fed did seem to “cause” the onset of the Great Contraction, as the gold ratio rose at a double digit rate between October 1929 and October 1930.  Indeed nearly half of the 9.6% rise in the world gold reserve ratio during that period occurred in the U.S.  Others might argue that the discount rate was the relevant indicator of monetary policy, or M1, or M2.  Krugman chooses the monetary base, which is endogenous under a gold standard, and which actually fell between late 1929 and October 1930 (before it began rising in response to the banking panics.)

Nevertheless, I suppose there is probably a bit of truth in Krugman’s argument about Friedman shifting his position.  Like everyone else, Friedman responded to the events of the 1970s by increasingly seeing the rate of inflation as a policy choice of the central bank.  I’ve already argued that this change in the way monetary policy was perceived largely explains the increasing acceptance of the Monetary History; it would be very strange if Friedman himself had not gotten swept up in that change.

Krugman’s much more important charge is that F&S never really showed what they claimed to show, that a more aggressive Fed could have prevented the Great Contraction.  I have some sympathy for Krugman’s accusation, as I think many people who casually examine the Monetary History simply assume that F&S have assembled a mountain of empirical evidence relevant to their hypothesis.  And I think when people ask me for data in support of my hypothesis that the Fed caused the crash of 2008; they implicitly seek a similarly data-driven argument.

The data do show a big drop in M1 and M2, but that doesn’t really prove that a more expansionary policy would have helped.  Perhaps banks and the public would have simply hoarded any extra injections of base money.  Here Krugman draws an analogy with the situation in Japan during the past 15 years:

And under those [zero rate] conditions, monetary policy proved just as ineffective as Keynes had said it was in the 1930s. The Bank of Japan, Japan’s equivalent of the Fed, could and did increase the monetary base. But the extra yen were hoarded, not spent. The only consumer durable goods selling well, some Japanese economists told me at the time, were safes. In fact, the Bank of Japan found itself unable even to increase the money supply as much as it wanted. It pushed vast quantities of cash into circulation, but broader measures of the money supply grew very little. An economic recovery finally began a couple of years ago, driven by a revival of business investment to take advantage of new technological opportunities. But monetary policy never was able to get any traction.

In effect, Japan in the Nineties offered a fresh opportunity to test the views of Friedman and Keynes regarding the effectiveness of monetary policy in depression conditions. And the results clearly supported Keynes’s pessimism rather than Friedman’s optimism.

So the Bank of Japan valiantly tried to end Japan’s deflation, but failed year after year after year.  No explanation of why the BOJ let the yen appreciate, no explanation of why they raised rates not once but twice when deflation threatened to become price stability.  But if you accept Krugman’s view of monetary ineffectiveness in a liquidity trap, and that view has certainly become much more plausible with the recent events in the U.S., then Krugman does have a point.  F&S assume that a more expansionary Fed policy could have turned things around, but they never really showed that Keynes was wrong.

When I first read the Monetary History in the 1970s, I recall being impressed by the severity of the decline in NGDP.  It seemed obvious that if the Fed had not let the monetary aggregates decline, the decline in nominal spending would have been far milder.  I found myself nodding in silent agreement each time F&S would say something to the effect of “if the Fed had adopted a more aggressive policy at this point, the contraction might have been much milder.”  They made those arguments at various stages of their narrative, although if one bought their basic argument that the Fed controlled the monetary aggregates, the specific policy counterfactuals seemed almost superfluous.

If one’s vision of monetary policy is the 1970s cross-sectional data presented above, then F&S’s argument seems like a slam dunk.  One doesn’t really even need to see much data, all you need to do is think about the fact that NGDP fell in half, and then think about what would had happened if the Fed had printed money like the Bank of Argentina.  By the late 1970s nominal spending seemed obviously a monetary phenomenon.  On the other hand if one’s vision of monetary policy is ultra-Keynesian, then the 1930s seem just like Japan circa 1998, or America in 2008-09, just another case of where monetary policy had become ineffective due to the zero nominal rate trap.  And this is a point that I cannot make forcefully enough; these two visions are about much more than “data.”  They are about how one processes a great deal of theoretical, historical, empirical, institutional, and political information.

I am struck by how people cite the Monetary History more frequently than any other monetary study.  Friedman and Schwartz did not do sophisticated VAR studies; their “econometrics” is basically just descriptive statistics.  Why haven’t we totally forgotten about this book, and moved on to other, more convincing studies of monetary policy in the Great Depression?  Perhaps it’s because there are no more convincing studies out there.

The actual Monetary History is much more complicated than most of us remember, and even a bit more complicated than Krugman assumes.  The complex and relatively sophisticated narrative (with more good analysis in its footnotes than most entire books contain) are F&S’s way of addressing issues such as exogeniety.  I’m not saying they got everything right; indeed I redid the entire exercise from a gold market perspective partly because I thought they missed a lot of what was important.  But their analysis is much better than what most supporters and opponents likely remember, especially if all that they recall is that F&S showed a correlation between the monetary aggregates and NGDP.

The difference between my view and Krugman’s view cannot be resolved through econometrics, it is a complex question of interpretation, including how one thinks about the relevant counterfactuals.  For instance, for me the relevant counterfactual is what would have happened if the Fed had abandoned the gold standard as soon as it began to constrain their policy.  We know that after they did so in 1933, NGDP rose very rapidly.  So what Krugman views as evidence of monetary policy ineffectiveness (say the failed 1932 open market purchase program), I view as merely reflecting the constraints of the international gold standard.  On the other hand Krugman could point to recent liquidity traps in Japan and the U.S., which occurred under fiat money regimes.  And you already know my views on those episodes.

One of the things that has most puzzled me about recent events is that many of the very same economists who were persuaded by F&S’s evidence, most notably Ben Bernanke, now seem strangely passive in their evaluation of current Fed policy options.  And almost none blame the Fed for the rapid decline in NGDP that began last September.  So does that mean Krugman wins the argument?  More likely it means both Krugman and I will lose in the short run.  I expect the “official narrative” to go something like this:

In 1929-33 a passive Fed allowed banking troubles to snowball into a systemic breakdown, causing NGDP to fall by roughly 50%.  In contrast, under the active leadership of Ben Bernanke the Fed was able to address an even more severe financial crisis, limiting the damage to a small decline in NGDP during 2008-09, and a recovery thereafter.

Why then do I think that this episode provides indirect support for my view of the crash of 2008?  Go back to my explanation for why the Monetary History became so widely respected after the 1970s.  I argued that once economists began to think of the money supply growth rate and inflation as being controllable, once they began to think of monetary policy as inflation targeting, then the F&S argument became retrospectively much more persuasive.

By the 1980s a new generation of economists was used to thinking of gold as a minor inconvenience, a “barbarous relic.”  They had no memory of just how powerful gold loomed in the interwar imagination.  How a fiat money regime with no gold backing was seen as the road to hyperinflation.  Thus even if economists like Eichengreen and Temin pointed out that the Fed may have been constrained by gold standard considerations, mainstream macro types would simply brush aside that issue, noting (correctly) that a Great Depression was much too high a price to pay for adherence to the gold standard.  And we were still a decade away from Krugman’s attempt to revive liquidity trap economics.  When the major economic problem facing the world is high inflation, and when tight money countries like Germany and Switzerland mostly avoid the high inflation, it’s pretty hard to argue for monetary policy ineffectiveness.

Forecast Targeting

I’ve been advocating NGDP futures targeting since 1986.  A few years ago I had almost given up hope, and then came across a paper by Robin Hanson entitled “Shall We Vote on Values, But Bet on Beliefs?”  (What a great title, and what a sad comment on our profession that this paper takes 9 years to get published.)  I immediately realized that my monetary policy idea was part of something much bigger, and as I began to look into the prediction market literature, and Svensson’s work on targeting the forecast, it suddenly dawned on me that there was something inevitable about futures targeting.

Using Robin’s framework the choice of policy target can be seen as a value judgment.  (In fact, the value judgment is much deeper, but we aren’t easily able to establish the link between macro aggregates and unobservable social welfare, so in practice the choice between inflation and NGDP targeting becomes a value judgment.)  Once an aggregate is selected; investors bet on which instrument setting is most likely to achieve the policy goal.  The FOMC would increasingly be seen as a political institution, not a bunch of technocrats.

In 1933 no one could foresee a world where central banks would simply choose among an infinite number of potential trend rates of inflation.  Even 30 years later when the Monetary History was first published, we were still under the Bretton Woods regime, and most developed countries did not have independent monetary policies.  But after 20 more years inflation targeting had become not only thinkable, but almost the only way that economists thought about monetary policy.

I am asking you to do a similar thought experiment.  Go out 50 years into the future and look back on current events from the perspective of the likely future policy regime.  Assume I am right and that we do end up with a futures targeting regime, where people naturally think of a neutral policy as one that equates the target and the forecast.  From that vantage point economists will look back on the events of 2008 and scratch their heads, wondering why the Fed let NGDP growth expectations go from plus 5% to sharply negative, and why they began paying a positive interest rate on bank reserves, and why they did all these things in the midst of a financial crisis.

Nobody in 1933 thought of Friedman’s 4% M2 growth rule as the baseline policy by which to judge the actions of the interwar central banks.  But 50 years later exactly that sort of policy counterfactual had become the standard way of thinking about the Great Depression, and the major reason why the 1930s Fed is held in such low repute.  They let the monetary aggregates decline by more than 30%; and it’s pretty hard to justify that level of passivity from any perspective.

Are we being unfair to the Fed in this sort of exercise?  Maybe slightly unfair.  But there were prominent economists such as Irving Fisher who warned that policy was too deflationary.  And during the 1920s hundreds of economists signed a petition favoring an active Fed policy of stabilizing the price level.  New York Fed Governor Strong had engaged in some fairly aggressive countercylical policy before his untimely death in 1928.  So it’s not like the ideas were not floating around.

One can make a similar argument today.  It’s not just me; highly respected economists such as Lars Svensson have advocated targeting the forecast.  Ben Bernanke has spoken of the need to equate the forecast and the longer term inflation goal.  Policymakers are very aware of these ideas.  Lot’s of proposals for “foolproof escapes” from liquidity traps have been published.  Furthermore, last year the same sort of mistakes were being made in Europe, where no one could seriously argue that policy was constrained by a liquidity trap (the ECB target rate was still 4.25% last October, when the dimensions of the crisis were already clear.)

If in the late 1970s or early 1980s someone had asked F&S for some “data” to support their hypothesis that monetary policy errors caused the Great Contraction, all they really would have had to say is “NGDP fell in half, QED.”  Similarly, the only data relevant to my argument is that NGDP growth expectations plummeted in the fall of 2008.  And anyone paying attention already knows that.

Of course you also need to buy my argument that monetary policy should be forward looking, and that policy can be highly effective even when rates are near zero.  But those latter two assumptions cannot be established with more data, they revolve around the question of how best to interpret existing data, and how to make sense out of previous episodes like Japan in 1998, or the U.S. in 1933.  To summarize, although only a handful of people seem to accept my view that the Fed caused the crash of 2008, it is actually the logical implication of three relatively widely believed propositions:

1.  NGDP expectations fell far below the Fed’s implicit target in October 2008.  (I find that almost all economists accept this view.)

2.  Monetary policy instruments should always be set at the level most likely to achieve the policy goal.  (Which is just common sense, isn’t it?)

3.  Monetary policy is highly effective even when nominal rates are near zero.  (Which I can find all sorts of prominent economists saying before this crisis hit, including Ben Bernanke.)

So if you still think my theory that tight money caused the crash is far-fetched, tell me which of these three assertions are not just false, but obviously false.  Or else explain why my hypothesis is not the logical implication of these three assumptions.  I’ve thrown down the gauntlet, and look forward to hearing how others respond.

I feel I’ve done all I can on this blog, and in the future will mostly respond to what others have to say.



66 Responses to “How did Friedman and Schwartz persuade us?”

  1. Gravatar of septizoniom2 septizoniom2
    20. April 2009 at 12:43

    another wonderfully clear and compelling post.
    i would be most grateful if you would post on your views on why your conclusions are not yet widely accepted at the fed.

  2. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    20. April 2009 at 14:32

    ‘This fundamental change in the way that people thought about monetary policy was a very lucky break for F&S.’

    To steal an old line, the harder Friedman (at least) worked, to promote his ideas–Newsweek columns, TV appearances, speeches, rewriting Capitalism and Freedom, PBS television series–the luckier he got.

  3. Gravatar of Dilip Dilip
    20. April 2009 at 15:02

    Yeah… I would be extremely interested to know why people like Dr. Mankiw don’t buy your theory (as you mentioned elsewhere). There must be some point of disagreement you can put your finger on, can’t you?

  4. Gravatar of Larry Larry
    20. April 2009 at 15:56

    Human nature compels us to stick to our preconceptions even when new information or in this case, new concepts, appear.

    How easy would it be for Bernanke to say “I’m reversing my policy of paying interest on reserves. Now we’re going to charge interest on those same reserves, and we should have done so from the beginning. Sorry.”

    How easy would it be for Krugman to say “I now realize that there is no such thing as a liquidity trap. We should have been targeting the forecast all along. Sorry.”

    That’s why I like what Mankiw appears to be trying to do, which is to take a baby step towards the new notions. He can help make seem less radical ideas that otherwise will take so long to become mainstream that they will only be able to help with the next crisis.

    And in case, as it seems, Professor Sumner has other things to do than blog, I’d like to thank him for this amazing series of calm, lucid, and immensely stimulating posts.

  5. Gravatar of Nick Rowe Nick Rowe
    20. April 2009 at 16:09

    There is a view that monetary policy (interpreted as setting the Fed funds rate) cannot or should not move too quickly (for fear of upsetting markets?). And events did seem to move much more quickly than they normally do, so it was harder for monetary policy to get ahead of the curve than it normally is. In such cases, we often say road conditions, not the driver, are responsible.

    Though that does not explain the apparent loss of economists’ confidence that monetary policy can still work now, even if it is belatedly.

    “I feel I’ve done all I can on this blog, and in the future will mostly respond to what others have to say.”

    Sometimes, even if you are just saying the same thing in different ways, it is worth doing, because it makes it clearer, and helps different people, who have different ways of understanding things. What you are doing is worthwhile. And I don’t think your ideas will suddenly dry up.

    You could always try a road trip 😉

    Off topic (sort of): during the 1930’s, when various countries went off gold (or raised the price of gold), what happened to interest rates? Not knowing the history, my guess would be that short safe rates increased, but that spreads between safe and risky rates decreased, and that risky rates may even have decreased. Do you know if that’s what happened? Because I think that’s what would happen now with QE (or something similar).

  6. Gravatar of Devin Finbarr Devin Finbarr
    20. April 2009 at 17:49

    And under those [zero rate] conditions, monetary policy proved just as ineffective as Keynes had said it was in the 1930s. But the extra yen were hoarded, not spent.

    I simply do not understand Krugman’s argument. If people are hoarding the money, then the bank should print more. In fact, the entire rule that the bank should follow is that it should keep printing money until people start trading those dollars for goods again. Does Krugman believe that no amount of printing money will get people to spend again? I simply do not get how he could miss such an obvious point.

    One problem applying Friedman’s analysis to 2009, M2 is not the money supply. The money supply consists of all notes that are convertible into dollars backed by the USG. Unfortunately, in the world of banana republic finance that we live in, that changes by the day as USG decides which banks are too big to fail and which will be backed by the Fed and the Treasury.

    To summarize, although only a handful of people seem to accept my view that the Fed caused the crash of 2008,

    The Fed did not directly cause the crisis by its actions. It did not actively tighten, the tightening happened because the infinite money machine shut off. But the Fed could have prevented the crisis, but it failed to respond as it should have.

  7. Gravatar of Josh Josh
    20. April 2009 at 19:39

    This is an interesting perspective on F&S.

    Regarding Krugman:

    1. Schwartz and Nelson’s response to Krugman in the JME is excellent and I would recommend that anyone interested in Krugman’s critique, as discussed above, consult that exchange.

    2. The trouble with looking at the monetary base (as Krugman does) is that it assumes that the money multiplier is constant. This, of course, is not the case in the wake of bank runs, etc. In other words, the Fed can appear to be following a loose monetary policy, but in reality merely offset changes in the money multiplier. I think this is important to your analysis Scott as it implies that increasing the monetary base to the point that it only stabilizes the money supply (or in the case of the early 1930s fails to even do that) is insufficient to raise NGDP.

    3. The evidence regarding Japan is not as clear cut as Krugman would make it seem. If you listen to Allan Meltzer (who was an honorary advisor to the BOJ) he would tell you that monetary policy was simply too loose and that the BOJ made significant mistakes regarding not only the policy mechanisms, but also its influence on expectations. Specifically, the BOJ was overly concerned with fears of inflation from the perpetually low interest rates.

  8. Gravatar of Jeremy Goodridge Jeremy Goodridge
    20. April 2009 at 21:03

    Thanks again Scott for another GREAT post. I would encourage you to keep posting: as Nick above says, even saying the same things in different ways is helpful in persuading new people. It’s also helpful in persuading converts like me that you haven’t given up!!

    I think your assessment of how people will look back on this episode is exactly right. In the end, people will see Bernanke as aggressive and preventing a great depression. Well, in a way, he did (even if he could have prevented more than just a Great Depression). In addition, that’s all Friedman claimed could be done in the 1930’s. In fact, he thought there would have been a sharp recession, but just not a Great Depression if the Fed had done the right things. But I also think we will be INCREASING our reliance on the Fed. And expanding the range of things the Fed can do. I don’t think people are going to adopt the Keynesian view that fiscal policy is the way to go. So, I’m not sure there’s much of an argument here.

    The truth is Krugman himself is a believer in the importance of the Fed too. There are some oddities in Krugman’s whole presentation that I think many just don’t understand. He just DEFINES liquidity trap in a way that makes the Fed seem inept. So, in the end, he’s just communicating his ideas on this topic poorly. He DEFINES a liquidity trap to occur whenever short-term treasury rates hit zero. OK. Well, that’s fine, but it’s pretty limited. Monetarists never focused on short-term interest rates in the first place. (And nor do you). So, that doesn’t seem like a limit to them. Krugman could be right that central bankers DO fixate on short-term rates (if they do, they certainly aren’t monetarists — ironically they’re actually closer to Keynesians!). In fact, what Krugman proposes is explicit inflation targeting to get out of a liquidity trap. OK. So, that seems pretty easy to me. But maybe it’s the case that the fixation on short-term rates is equivalent to the gold standard fixation of the 1930’s — an idea that is so powerful, you can’t get out of it. But then again, doesn’t Bernanke believe in inflation targeting??!! And the ECB? In the end, the liquidity trap, as Krugman defines it, is more of a POLITICAL trap then it is something genuinely economic. As further evidence of this, he says somethign like, “the Fed has to promise to be irresponsible…” as a way to get out of a liquidity trap (see his articles on Japan). Well that’s a really strange way to phrase things. Is it irresponsible to do the right thing and fix the economy?!! No. But what Krugman is probably implying is that central bankers think it is irresponsible to promise any inflation, ever. So, what must be required is a REDEFINITION of irresponsibility. That sounds very political.

    But the case of Japan, I think, did push Krugman to be more pessimistic about the ability of central banks. “Look at all that money they printed and still no effect”, he might say. A monetarist would say, if it doesn’t increase M2, it didn’t print enough. In fact M2 growth in Japan was very sluggish — as Krugman knows. To a monetarist, it just means print MUCH more and get it into more people’s hands. To Krugman, it seems like there is some inherent limit on what the Fed can do. So, now Krugman might say, “depressions are harder to prevent than we thought, but there’s still hope …” and I’ll bet one of the main instruments he would want to change is the Fed’s fixation on short-term rates — and focus them on an explicit inflation target. So, to persuade a Krugmanite, one would have to show that when money does really end up in people’s hands, it gets spent and that there are tools to achieve that. And most likely this would mean bypassing the banking system. One issue in both Japan and the US now is that the true hoarders aren’t regular consumers but the banks THEMSELVES.

    My view is that the monetarists do under-estimate how hard it is to get money into people’s hands. Buying short-term Treasuries is a crappy way to do it, in some cases. And on this the Keynesians are right. But, if you do succeed in getting the money into the REAL economy — i.e. actual consumers and businesses (as opposed to banks), then the effect is immediate and explosive. Bernanke may, in some ways, be like these monetarists — and so he is just so confused why an interest rate on reserves could be so bad — “well it’s just a tiny interest rate — how could that matter?”, he might say. Furthermore, keeping alive weak banks and feeding them the liquidity is not so effective either. Can you imagine what would be happening if most of the weak banks were sold off instead of bailed out? I suspect the monetary base would be much smaller and M2 growth (and NGDP growth) much more vigorous now.

    On a side note, I suspect Friedman, if alive, would have felt the Fed caused this. The amazing thing is that during this crisis M2 DROPPED in the middle of 2008!!! See this chart:

    2008-present M2 growth

    That would have caused Friedman to jump out of his chair and note that the Fed was MUCH too tight.

    I think Friedman’s biggest intellectual mistake was that he, later in life, started to focus on ‘freezing’ the monetary base. That was a mistake, or at best confusing. You can’t freeze the monetary base and expect results, except maybe in a totally free banking situation where private banks can print money and block deposit convertability at will. And even then I’m not sure. Maybe Friedman understood this and the monetary base freezing idea was meant to be accompanied by free banking, but he wasn’t clear (at least in what I have read).

  9. Gravatar of Devin Finbarr Devin Finbarr
    20. April 2009 at 21:27


    My view is that the monetarists do under-estimate how hard it is to get money into people’s hands. Buying short-term Treasuries is a crappy way to do it, in some cases. And on this the Keynesians are right.

    It baffles me why people think it is so difficult to get money into people’s hands. Add a few zeroes to the treasury’s account at the Fed, and mail out checks to all Americans. Or declare a tax holiday and fund all government operations by printing money. Or create accounts at the Fed for every American, and mail them all debit cards. The dollar is a fiat currency, getting money into people’s hands is trivial.

    Buying short-term Treasuries is a crappy way to do it, in some cases.

    Buying short term treasuries is inflationary neither in theory nor in practice. A treasury is simply a note backed by the full faith and credit of the USG, that says, “this will be a dollar on date X”. When the Fed buys treasuries at market value, it simply replaces one note backed by the USG with another note backed by the USG of the same market value. It has done exactly nothing. If the Fed pays more than market value, or buys so much that it drives up the price way beyond even face value, then the extent of that overpaying will be inflationary. But the inflation will be non-neutral – it will be a massive giveaway to the bond holders.

    I simply do not understand why monetarists focus on inflating via buying treasuries or manipulating interest rates. These methods either do not work, or if they do work, they work via massive wealth transfers. A reflation should be as neutral as possible and as quick as possible. The best way to do this is a tax holiday or a tax refund.

  10. Gravatar of Devin Finbarr Devin Finbarr
    20. April 2009 at 21:38


    You can’t freeze the monetary base and expect results,

    The transition issues would be a nightmare, but once you got through that, I think a frozen monetary base would be quite nice.

    Recessions and depressions happen because of sudden shocks to the supply or demand of money. With a frozen money supply, you could not have supply shocks. I suspect you would also not have demand shocks either. With a fixed money supply you also could not have maturity transformation, and maturity transformation is the primary cause of demand shocks, recessions, and depressions.

  11. Gravatar of Jeremy Goodridge Jeremy Goodridge
    20. April 2009 at 23:50


    You say:

    “Buying short term treasuries is inflationary neither in theory nor in practice. A treasury is simply a note backed by the full faith and credit of the USG, that says, “this will be a dollar on date X”. When the Fed buys treasuries at market value, it simply replaces one note backed by the USG with another note backed by the USG of the same market value. It has done exactly nothing. If the Fed pays more than market value, or buys so much that it drives up the price way beyond even face value, then the extent of that overpaying will be inflationary. But the inflation will be non-neutral – it will be a massive giveaway to the bond holders.”

    I agree with you on this. Everyone focuses on buying short-term securities because this has been the traditional approach. But it’s really just the modern version of the gold standard. It’s the box that prevents us from seeing that get money into people’s hands really can be quite easy.

    As you suggest, “Add a few zeroes to the treasury’s account at the Fed, and mail out checks to all Americans. Or declare a tax holiday and fund all government operations by printing money. Or create accounts at the Fed for every American, and mail them all debit cards. The dollar is a fiat currency, getting money into people’s hands is trivial.”

    It really isn’t hard to get money out there — once your mind is open to the possibilities. But as long as the Fed feels it needs to BUY AAA stuff to get money out, things will always be difficult — monetarists underestimate that the mechanism does matter. If instead the Fed had an NGDP target and THAT’s what constrained it, it would be much more open to doing things that look more ‘fiscal’ in nature — like your proposal above. Right now, there’s this attitude that the above proposal is ‘irresponsible’, but it’s only irresponsible when it’s not constrained by an NGDP target.

    And I think your most important point is this:

    “A reflation should be as neutral as possible and as quick as possible. The best way to do this is a tax holiday or a tax refund.”

    It’s just like the ideal for what our tax structure should be — as neutral as possible. If we thought this way, we would be printing money and sending people checks or providing across-the-board tax relief to reduce withholding.


  12. Gravatar of Bill Woolsey Bill Woolsey
    21. April 2009 at 02:49

    Your key point is that the Fed should have used quantitative easing to prevent the drop in nominal income over the last six months.
    Describing this as the Fed causing the drop in nominal income causes confusion. While you are correct that F&S argument that the Fed turned a recession in 29 into a great depression by failing to act as lender of last resort in a way that offset the decrease in the money multiplier, your citation of Krugman claiming that F&S are wrong to call this the Fed “causing” the problem points to the same confusion.

  13. Gravatar of Nick Rowe Nick Rowe
    21. April 2009 at 02:56

    Jeremy Goodridge said: “But maybe it’s the case that the fixation on short-term rates is equivalent to the gold standard fixation of the 1930’s “” an idea that is so powerful, you can’t get out of it.”

    YES! And the closer you get to the real practitioners in monetary policy, in both central and commercial banks, the tighter the hold of that idea. Any other proposal for monetary policy just gets translated straight back into “But what does that mean for short-term interest rates?” It’s hard to think of it as a rabbit when you’ve spent your life hunting duck. It’s easier for historians (and historians of thought), who have been forced to think about rabbit hunting as well.

    Keynes, on the hold of ideas, modes of thinking, etc.

  14. Gravatar of Bill Woolsey Bill Woolsey
    21. April 2009 at 03:05


    T-bills and the monetary base are not the same.

    The monetary base serves as medium of account. There is no variable money price for base money, because the dollar is defined as a unit of base money. T-bills do not serve as the medium of account. T-bills have a variable market price.

    The monetary base (along with various other financial instruments) serve as the medium of exchange. T-bills do not.

    The medium of exchange can be spent into existence. People accept it without intending to hold it. To issue more T-bills, the Treasury must sell them to someone who wants to hold them.

    Further, an increase in the demand for the medium of exchange can be actualized by an individual by pulling money out of the stream of income and output. Since no one is routinely paid in T-bills, people who want to hold more T-bills need to buy them.

    There is a market for T-bills, with specialzed dealers who quote variable market prices for them.

    This doesn’t mean that open market operations in T-bills always have an impact. I think that we are currently in a special situation where there isn’t much impact. But to state that they are both claims against the government simply failes to understand what is important about the monetary base.

  15. Gravatar of ssumner ssumner
    21. April 2009 at 04:34

    Septizoniom2, I will try, but check out a post from late February, “Why did monetary policy fail”

    Patrick, Yes. Luck rewards the well prepared. Friedman was a superb debater, and a superb critic of Keynesian monetary theory. His own monetary theories? No so good.

    Dilip, I shouldn’t say Mankiw necessarily opposes my idea, but I think he worries about private cash hoarding, and thus seeks a more complete solution. I prefer half a loaf to none.

    Larry, In my last paragraph to “Reply to Krugman” I talked about what is at stake, no has a moral right to let ego influence their policy views.

    BTW, I guess I gave the impression I was quitting blogging. Not so, I just meant I have finished off the backlog of things I wanted to say when I started out. So I will be more reactive. Even today’s comments gives me more ideas.

    Nick, In a sense you may be right about motivation, although even that doesn’t really explain current policy. A few remarks (more aimed at the Fed than your interpretation of the Fed.)

    1. The rapidly changing events were actually of the Fed’s own creation. The markets crashed in early October when not much was going on, not during events like Lehman in mid-September. They gave the Fed several weeks to turn the steering wheel, and when they saw it had no intention of doing so, and would offer no credible policy to keep NGDP from falling, they collapsed. That’s why I don’t buy the road condition excuse, the driver is only exonerated if he had no time. The Fed didn’t cut rates once between late April and October 6. And they didn’t cut rates on October 6 either–instead they instigated one of the most contractionary policies in their 95 year history—a policy of paying banks above free market rates to hoard reserves.

    I’m sure you are right about new ideas–indeed your comment suggests something I should work on. I just felt I was starting to repeat myself, but I suppose that is inevitable at some point.

    I am only familiar with the U.S. interest rates in the 1930s. Keep in mind we were only off gold for nine months, and only a few days of those nine months saw a Fisher effect (in November 1933) You are exactly right about expansionary policy reducing risk spreads, and we have unusually strong evidence as I regressed the (AAA-Baa) spread against daily movements in the dollar during 1933. Since the dollar’s depreciation was exogenous (caused by a series of FDR comments) the results seem air-tight to me.
    I think real risk-free rates stayed pretty low because the economy was so far depressed in 1933, that their wasn’t much expansion of plant and equipment even after the economy started growing. So nominal T-bill rates were low and stable as I recall.

  16. Gravatar of septizoniom2 septizoniom2
    21. April 2009 at 04:40

    thank you. would look forward to your efforts. i have read that post in the past. i wonder if the fed is laboring under the burden of being still in the miasma of a gold standard without realizing it.

  17. Gravatar of ssumner ssumner
    21. April 2009 at 05:23

    Devin, Krugman fears to much money would expose the central banks to dangerous capital losses if that had to pull the money back later, and thus sell bonds, to prevent an inflation overshoot. I think he worries much too much about this issue.

    I agree that M2 is not useful today.

    As I said, there is no “baseline” stable monetary policy, so debates about causation only make sense as policy counterfactuals. Bernanke does think the Fed caused the 1929-33 collapse, so his defenders can hardly argue that I am applying an unfair standard, as rates were cut sharply and the MB boosted sharply during those years as well.

    Josh, Thanks, I’ll look at that piece.

    Krugman does understand that the multiplier fell. His argument is that if the Fed had put even more (base) money out into circulation, the only (or main) thing that would have happened is the multiplier would have fallen even more.

    I agree with you on Japan, and had a brief critique of his view in this post.

    Jeremy, I hope I am wrong about how we will look back on this short term (where Bernanke is a hero.) I hope to see my 50 year prediction come true, but I doubt I’ll live that long. Maybe when I am dead someone will say “Sumner was right” (and others like Thompson who also think the Fed caused the crash.)

    BTW, F&S were wrong in assuming that there would have been a severe recession anyway. As I show in my Depression manuscript the initial 1929-30 downturn was caused by the simultaneous hoarding of gold by key central banks (U.S., France, UK, etc.) It made the gold reserve ratio soar after October 1929. Tight money caused the first year of the Depression. (Totally missed by IS-LM types.) After late 1930 it was a mixture of tight money in the gold bloc, and widespread bank failures. There were no “bubbles” in 1929 that could explain the onset of Depression. The 1987 stock crash was just as bad, and didn’t lead to even a tiny, tiny slowdown. The economy stayed strong for almost 3 more years.

    Krugman is not inept in his use of language, he is careful, and clever, and I’d say misleading at times. Here’s just one example I recall from his NYT column:

    He said neoliberal policies had not always been helpful in Latin America. He noted that Mexican growth had continued to stagnate despite the election of President Fox on a neoliberal platform.

    The preceding statement (not quote, but close enough) addresses two audiences at once. To his leftist admirers he seems to support their anti-capitalist prejudices. But he also winks to development economists, hinting he knows that Fox was not able to get any of his market reforms through Congress. That’s why he didn’t simply say neoliberal reforms failed in Mexico. The Brits call this “dog-whistle politics” and Krugman is the supreme master of that art.

    I agree with your criticism of the strange way he looks at the politics of liquidity traps. He has all sorts of moral outrage for fiscal policy failures, but gives even the most conservative central banks like the BOJ a sort of pass. I’m sure you can find quotes of him suggesting they create inflation expectations, but his usual attitude is a sort of “well, there’s nothing to be done there, they have too conservative a reputation.” Well then change the reputation!!!

    Devin, Tax holidays might work, but it would balloon the deficit. OMOs are much better, and cash and T-securities are not that close a substitute in the long run. Try shopping at your local WalMart with a fist full of T-bills.

    Devin#2; Gold supplies were fairly stable in the period after the Civil War, but increases in the demand for gold would still create occasional depressions.

    Bill, QE makes it sound exotic, how about “ordinary open market purchases”. And how about no interest on reserves, rather than interest on reserves—there it is the Fed getting more exotic.

    Also keep in mind that Bernanke told us (and the markets by implication) that the Fed could handle a zero interest rate environment. That they had other ways to inflate. So let’s see them.

    I think the heart of our disagreement is about the term “cause”. I think it can only be defined in terms of plausible policy counterfactuals. And I think expected inflation or expected NGDP targeting were plausible counterfactuals. Bernanke had argued in the past that they tried to offset shocks and keep AD growing at a satisfactory rate, so they should have continued doing that. As I mentioned above, most people seem to have a baseline scenario in their minds for when the Fed is “doing nothing” That is, most people have a preferred policy indicator. Unfortunately, there seem to be dozens of such indicators, which makes each person’s useless. Some might think doing nothing is hold the base constant, some might think doing nothing is hold the fed funds rate constant, or M2, or the inflation spread between indexed and conventional bonds (Hetzel’s preferred indicator.) Mine is expected NGDP growth. For 25 years that was pretty stable–when it fell sharply after September I viewed it as a policy change. The forward-looking Taylor rule is also now being grossly violated, if one prefers that indicator.

    I am happy if people interpret my claim for causation as being no stronger or weaker than F&S regarding 1929-33, as I think most economists find their condemnation of the Fed to be pretty convincing. Remember that I did not claim the Fed caused the onset of the recession in December 2007, but only the worsening 9 months later, so there’s another similarity to F&S.

    Nick, That’s a good point. I think the whole subject of the psychology of how economists think about money and monetary policy is worth a book. I’ve noticed that everyone I talk to is fixed on some way of looking at monetary economics–and if you disturb their view with a thought experiment they often get confused for a moment–until they can work it into their IS-LM framework. People like Bill and I (and I presume you as well) are always having to battle people who can’t imagine monetary economics in any way other than interest rates (which I regard as a relatively unimportant aspect of policy.)

    Bill, Excellent point, my thoughts exactly.

  18. Gravatar of Alex Alex
    21. April 2009 at 05:43


    Some comments about rules:

    * There is no rule that dominates all other rules (including no rule at all) in every possible scenario. At some point you will face a shock that under your rule will give you at a worse outcome than some other rule. So judging the performance of the rule under that outcome alone can be misleading.

    * In relation to the previous point, rules have to be evaluated in view of their overall performance over many years and different situations.

    * The effectiveness of rules depends on the commitment capabilities of the policy authority. Rules should not be changed just because now it is not working out as we wished. If you go changing your rules all the time then you are screwed.

    * Rules are not forever. At some point you will face a shock so big that it will either force you or that you will find it optimal to choose to abandon your rule. But still you should try to hold to it as long as you can (it shows your commitment to it and increases your reputation), let the market see that you really had no choice.

    So for example your view is that tight monetary policy caused the 2001 crisis in Argentina. But it was that same monetary policy rule (fixed exchange rate) that stopped the hyperinflation of the early 90’s and helped the economy grow. Economic conditions changed (the dollar appreciated, and deficits accumulated) and eventually the rule was no longer good for the economy. Still the Central Bank resisted to change it. Actually the Central Bank could not change the exchange rate, only Congress could. By 2001 the economy was in a bad situation which forced the abandoning of the rule and eventually lead to a recession. So did the rule work or not? Well it is hard to say. But I think the 90’s showed some things (which we actually already knew but hoped they had changed):

    * There is no independence of the Central Bank in Argentina.
    * The government has no ability or will to close budget deficits.
    * Money in Banks is not safe in Argentina.
    * There is no respect for private contracts.

    All of this in the end guarantee that no rule will work in Argentina because we know that as soon as the rule starts to under perform it will be abandoned.

    So should the Fed have abandoned the gold standard sooner in the 30’s? Maybe, but what would have been the reputation cost of doing it?

    Should the Fed now be worried about preventing 10% unemployment or a complete meltdown of the banking system?

    I don’t know. What I know is that the US is not Argentina because in part it didn’t use to go out printing its way out of every crisis. The US is not Argentina because it went through it worst recession in 50 years just to bring inflation down. The US is not Argentina because after the civil war it paid back the debt that was issued by a government which no longer existed. Sure, printing can help in this case but do we really want to do it?


  19. Gravatar of JimP JimP
    21. April 2009 at 06:03


    You have made reference various times to articles you have written, and to a book manuscript you have. Are any of these articles available on the web? And – I don’t suppose you would be willing to post that book. I wouldn’t either. But I sure do wish it would get published. Is there anything people could do to get it published?

  20. Gravatar of Josh Josh
    21. April 2009 at 06:19


    Re: Krugman and the money multiplier:

    I think that we are talking about two different things. You are stating that Krugman understands the money multiplier (something I would dare not refute) and that his perspective is that increasing the monetary base could not have gotten us out of the Depression. I don’t buy that theory, but that is not what I am talking about.

    What I am referring to is not whether monetary policy was impotent, but rather Krugman’s claim that the Depression was not the fault of the Federal Reserve. Of course, this depends on one’s definition of “fault” and whether we are willing to concede that the Depression need not have had one cause (which unfortunately seems to be how the debate is always framed). In any event, Krugman backs up the claim that the Federal Reserve cannot be blamed because the monetary base was constant, while M2 was falling. He writes (in his JME response), “a nearly flat monetary base is more or less what we would have expected if there had been no Federal Reserve, and the United States had been on a pure commodity money standard. The Fed should have done better””but there’s no evidence that it caused the Depression.”

    Of course, this analysis is not complete. If we look at monetary policy a bit earlier, James Hamilton has detailed the fact that the the Fed sold $393 million worth of securities from 1927 through 1928. By August 1928, there were only $80 million worth of securities left on the Fed’s balance sheet. I think that this is clear evidence that, in fact, the Federal Reserve was too tight and the failure increase the monetary base can be seen as a further example that the Fed was too tight.

    Further, as Nelson and Schwartz note:

    “Krugman will only consider the aggregate of the monetary base in judging the stance of 1930s monetary policy in the United States. As we have already discussed, the aggregate of the monetary base matters, but analysis of its components is also important. That analysis confirms a contractionary policy in the 1930s, contrary to Krugman’s position. The decline in the M2 money multiplier also reflects monetary policy. Alternative monetary arrangements (i.e., better Federal Reserve policy in the 1930s, or the counterfactual case of the pre-Federal Reserve clearinghouse arrangements continuing in the 1930s) would have lessened the collapse of the M2 multiplier. Thus, looking at the aggregate base alone to judge the impact of monetary policy’s influence on M2 is inadequate.”

    This is the point that I was trying to make. Krugman uses the monetary base to say that Federal Reserve policy was not too tight when such an analysis leaves out very important details.

  21. Gravatar of Jeff Jeff
    21. April 2009 at 06:45


    Given the ridiculous level of housing prices relative to income and rents, real housing prices were bound to crash. Add in insane leverage ratios and a financial debt crisis was bound to happen. Rogoff and Reinhart tell us that always leads to a bad recession.

    An NGDP target implies that when the forecast predicts a recession, you deliberately create inflation. Consider that if real GDP is expected to fall at a 5 percent rate for a couple of quarters, your 5 percent NGDP target means the Fed aims to create 10 percent inflation. Are you really surprised that not many people think that’s a great idea? Centuries of experience show that governments are far more likely to inflate than deflate, so it’s not surprising that prudent people do not want to see double digit inflation get started.

    An alternative is to have the Fed target the expected price level. If the forecast predicts substantial deflation, the Fed is too tight. Using this metric, you can still argue that money has been too tight. But it seems to me that criticizing the Fed for not trying to create double-digit inflation is going too far.

    BTW, I really like your emphasis on the gold standard. I found Temin’s “Lessons from the Great Depresssion” very convincing on that score many years ago. The strongest point of his argument, as I recall, is that while a country that is losing gold reserves is forced to tighten, a country (like the U.S.) that is gaining reserves is not thereby pushed to ease it’s monetary policy. The asymmetry means that a severe shock anywhere leads to gold hoarding and the attendant tight money everywhere. You couldn’t come up with a worse system if you tried.

  22. Gravatar of Jeremy Goodridge Jeremy Goodridge
    21. April 2009 at 07:57


    I would add two more Krugman misunderstandings:

    1. A flat monetary base within the US is not what would have happened without a Federal Reserve. In an international gold standard, gold moved to where it was most demanded. And gold was the primary component of the monetary base. So, while worldwide gold might have been flat, within a country, monetary base could move up and down. It was highly imperfect, and particularly susceptible to international crises, but it was not characterized by a stable monetary base over short periods WITHIN individual countries. In fact in the early part of the great depression, monetary base would have RISEN somewhat if the Fed did nothing, but instead if dropped. That is hardly nothing. And as we know with all these things, it’s a lot easier to prevent the panic than to cure it once it happens. So, it’s not unreasonable to think that if the Fed had in fact not neutralized gold inflows in 1929-1930, world history might really have been different.

    2. And finally, non-convertibility of deposits was a crucial feature of the pre-Federal reserve days. It was the primary way that increased demand for money didn’t result in a collapse of the true money supply ( that includes deposits). But we never allowed that mechanism to operate post federal reserve because the federal reserve’s ‘lender of last resort’ capacity was supposed to solve that. As Friedman and Schwartz point out, it’s the ultimate irony that the longest bank holiday in history came AFTER the federal reserve was created even though the federal reserve was created in large part to prevent such bank holidays from happening. So, here again, monetary base measures miss a big part of federal reserve function which was in fact to prevent bank runs from causing the banking system to collapse (and thus deposits and thus aggregate demand).


  23. Gravatar of Jeremy Goodridge Jeremy Goodridge
    21. April 2009 at 08:12


    Another problem with the whole monetary base discussion, less applicable to the 1930’s, but VERY applicable to today is the fact that monetary base INCLUDES bank deposits with the federal reserve. And today, that’s where most of the monetary base is going. And there’s a good reason for it — we are paying interest on reserves. So, here, again, monetary base is a mistaken measure of Fed policy.


  24. Gravatar of David Stinson David Stinson
    21. April 2009 at 08:28

    Hi Scott.

    I take it from your post above that you may be posting less or at least differently. I wanted to let you know how much I had appreciated your blog. As I mentioned previously, I have been out of school for awhile and am sort of feeling my way back to a state of reduced-out-of-datedness in the macro department. I have learned a lot and found it stimulating. Your blog is also refreshingly free of the barely concealed political activism and “crisis-opportunism” that seems to characterize many economics blogs these days.

    Regarding your main point, and setting aside the question of the Fed’s role in causing the bubble, when you argue that the Fed caused the Great Unravelling (that’s what I am calling it), do you mean “caused” in the sense of a catalyst for an otherwise inevitable event or do you mean “caused” in the sense that had the Fed acted differently last fall, we could have continued as we were indefinitely?
    If it’s the latter, I am assuming that you would also believe that the bubble was not really a bubble in that it was sustainable and not characterized by a cluster of private sector error or fatal incompatibilities in the real economy. Am I right in assuming that?

    Nick & Scott,

    I think this was discussed previously, but can’t the central banks preference for intervening, in the first instance, at the short end be explained by a desire to preserve maximum reversibility? Intervening at the longer end increases the potential for significant capital losses on assets it purchases (if it is successful in raising inflationary expectations). Besides, I would have thought that the whole idea that the interest rate is part of the monetary transmission mechanism is based (at least in part?) on the notion that central banks’ intervention in one part of the term/risk structure of interest rates can affect other elements (due to private sector arbitrage) of the structure (on at least a ” loose” basis, if not a “tight” one). I am thinking as well that government intervention at other selective points in the risk/term structure raises the possibility that the lender of “last” resort may become the lender of “only” resort.


    BTW, I enjoyed your post of the other day about interest rate targeting although I must admit that am still struggling to get a precise grip on what the natural rate of interest really represents in conceptual terms.

    You mentioned above that:

    “Not knowing the history, my guess would be that short safe rates increased, but that spreads between safe and risky rates decreased, and that risky rates may even have decreased. Do you know if that’s what happened? Because I think that’s what would happen now with QE (or something similar).”

    I understand why you would expect QE to cause risky/long rates to fall and safe/risky spreads to fall, but why would you necessarily expect QE to result in higher short rates?

    Scott & others re Krugman,

    One might make the following conjectures:

    a) Economists who favour old-fashioned fiscal stimulus may have a tendency, unconscious or otherwise, in their enthusiasm for advocating their preferred policy tools to try and undermine the effectiveness of competing policy devices.

    b) Correct me if I am wrong, but the argument for fiscal stimulus seems to be conceptually more dependent on what at its base is some sort of “market failure story” than does the effectiveness of monetary policy. I can see that interventionists of the fiscal variety may find the liquidity trap narrative appealing in that it involves people doing nasty-sounding things like “hoarding” money which may, so the story would go, be individually “rational” in an impossibly narrow, misguided and lemming-like way but which would collectively be irrational.

  25. Gravatar of Jeremy Goodridge Jeremy Goodridge
    21. April 2009 at 08:47


    You say:

    “Correct me if I am wrong, but the argument for fiscal stimulus seems to be conceptually more dependent on what at its base is some sort of “market failure story” than does the effectiveness of monetary policy. I can see that interventionists of the fiscal variety may find the liquidity trap narrative appealing in that it involves people doing nasty-sounding things like “hoarding” money which may, so the story would go, be individually “rational” in an impossibly narrow, misguided and lemming-like way but which would collectively be irrational.”

    Yes, I think you are right. The extreme anti-monetarist would say, “It doesn’t matter how much money you print, it will just get hoarded”. Krugman himself has some of this — for example when he emphasizes the monetary base. He’s basically saying, “see they are printing lots of money, but it’s just getting hoarded by banks…”.

    But then, Krugman isn’t totally like this when he proposes inflation targeting. If there is an infinite desire to hoard, of course, even inflation targeting wouldn’t work. So, he must believe there is SOME point at which people would stop hoarding.

    So the view Krugman has now is, “sure let’s have the Fed print like crazy, do QE, buy lots of long-term, riskier stuff, but then let’s not count on it totally and still have the treasury spend money on real stuff to boost GDP”.


  26. Gravatar of Jeremy Goodridge Jeremy Goodridge
    21. April 2009 at 08:54

    By the way, my quotes aren’t real quotes by Krugman but just things he might say, or someone with his views might say.


  27. Gravatar of david glasner david glasner
    21. April 2009 at 09:59

    Scott, Your ruminations about the Monetary History prompt me to ask you what do you find so compelling about the Monetary History, and why do you feel any need to defend it against Krugman’s criticisms? I would have thought that you would have used Krugman as a point of departure from which to draw attention to the shortcomings of the explanation of the Great Depression offered by the Monetary History. For example, you curtly dismiss Krugman’s point about the behavior of the monetary base between 1929-33 with the observation that under the gold standard the monetary base is endogenous. Ahem, well doesn’t that also apply to M1 and M2 about which Friedman was obsessively preoccupied? Is that not precisely the point of concentrating on the gold market as you do so skillfully in your own empirical/historical work on the Great Depression. Why defend a bad theory just because it has the adjective “monetary” in front of it? I am tempted to launch into another one of my diatribes about the baneful influence Friedman exercised on the development of monetary theory, an influence largely propagated by way of the Monetary History, even though that work is itself devoid of any explicit monetary theorizing. But I don’t want to use your blog to further belabor that pet peeve.

  28. Gravatar of Nick Rowe Nick Rowe
    21. April 2009 at 10:16

    David Stinson @ 8.28:

    Thanks David!

    You are not the only one struggling with the natural rate concept!

    In reply to your question, a successful QE policy would raise short safe nominal interest rates because:
    1. Expected inflation would increase.
    2. Consumption and investment demand would increase because expected future real income and real demand would increase.
    3. Consumption and investment demand would increase because the banking and financial system would be stronger, because of higher expected inflation and future real income.

    In short, the pathological fears which cause people to want to hold short, safe, liquid assets only would diminish.

  29. Gravatar of Bill Woolsey Bill Woolsey
    21. April 2009 at 13:21


    How can you possibly believe that the size of the recession (the drop in real output) is independent of the drop in nominal expenditure? How is it that the end of a housing bubble or “insane” leverage levels lead to a drop in real output? Why exactly?

    To the degree that these things require a reallocation of resources, so that the productive capacity of the economy grows more slowly, stagnates, or even shrinks, then maintaining the growth path of nominal income will result in somewhat higher inflation and somewhat lower real wages and other real incomes even if nominal wages and other incomes continue on their previous growth path (of course.)

    But there cannot be a problem of a generalized inability to sell. There cannot be a “problem” of no one spending because they are too much in debt. There cannot be a problem of people saving too much because of worry. Or saving too much to rebuild lost wealth. With nominal income growing on target, people can save all they want. They can pay off their debts if they want. Of course, what they want to do might be different.

    The nominal income target automatically fixes any problem that involves some kind of general–people just can’t afford to buy things or won’t buy things.

    You end up with a situation where people are buying plenty of things, just maybe not the things we are set up to produce. So there are sectors of the economy with strong sales, rising prices (the inflation,) profits, and growing employment. Yes, there are shrinking sectors of the economy. Perhaps shrinking sectors shrink more rapidly that growing sectors grow. That is why aggregate output might be depressed. Perhaps the shinking sectors lose employment faster than growing sectors hire. That is why unemployment is high. But there is plenty of prosperity and a clear path to recovery.

    No… the story about a burst bubble and insane leverage leading to falling output is due to falling sales of product. And that _means_ falling nominal income in the aggregate.

  30. Gravatar of Larry Larry
    21. April 2009 at 16:44

    @Alex – There is no rule that dominates all other rules (including no rule at all) in every possible scenario.

    I think later rules typically handle more cases than earlier rules. I.e, rule changes do not have to be a crisis-driven, random walk. Greenspan’s rules were better than Strong’s and would have worked better than Strong’s at that time, both for the Depression and for the cycles that preceded it. Bernanke’s rules could be a further improvement. IOW, policy rules, like laws of physics, can incorporate new experiences without sacrificing the abiliyt to handle the old ones.

  31. Gravatar of ssumner ssumner
    21. April 2009 at 18:42

    Septizoniom2; Yes, there are parallels to the gold standard mindset. They couldn’t envision stable fiat regimes, our central bankers can’t envision non-interest rate regimes. BTW, Nick Rowe now has a good post on trying to imagine monetary policy without reference to interest rates. I hope people will check it out as it complements my own ideas.

    Alex, The perfect rule mimics the social welfare function for losses from macro instability. To complicated to discuss here, but Robert Hall did some articles on this in the mid-1980s.

    I’ll try to do a post soon on Argentina, so I’ll refrain from detailed comments here.

    I want to prevent both 10% unemployment and banking collapse. Monetary expansion can do both.

    JimP, At some point I will go online if I can’t find a publisher. Then I could constantly improve the book through comments. It would start as “the Depression by Scott Sumner” and end up “the Wikipedia Depression” If I mention something and you want to see it I can email it to you.

    Josh, I understand your point better now, and it makes sense. I have a very different view from both Krugman and F&S. Some earlier posts like “1929, 1937” “Certain dates in 1932” and “George Warren” give you some sense of my take on the Depression.

    I disagree with Hamilton’s view that money was too tight, and the Austrian view that it was too loose. I think it was exactly right during 1927-29. F&S point out that 1928-29 was the only modern expansion to feature deflation–music to the ears of NGDP types like me, Bill and George Selgin. (Well, maybe a bit too easy for George.) But policy was way too tight after August 1929.

    Jeff, Maybe the bad recession causes the house price collapse. I don’t think that happened this time, but I think the recession worsened the problem more than necessary. On the growth rate, if we have 5% nominal growth we certainly won’t get 10% inflation, we’d get 1 or 2%. Also see Bill’s response.

    The asymmetry point is interesting, but it also applies to interest rates (and money.) The gold ratio cannot go below zero, nor can nominal interest rates, nor can the money supply. The first two might be serious policy constraints in certain situations, but not the money supply. This is why credible money supply targeting is never completely ineffective.

    Jeremy, I agree there is no Great Depression w/o the Fed. Probably the initial post-war depression would have been a bit worse. I need to do a post on the 1921 depression sometime.

    David, Thanks, the blog will continue. The bubble gradually unwound for two years with no major damage to non-housing sectors of the economy. That process should have been allowed to continue. Instead the Fed let NGDP expectations plunge, severely damaging non-housing sectors of the economy.

    The commenter Jon says the short run only idea is a myth. He says the Fed has always bought lots of medium term debt.

    You may be right about Krugman–but I’ve been accused of trying to read his mind so I should let others draw their own conclusions.

    Jeremy, I think Krugman knows that there is some QE that would work (although last time I said that he jumped on me–so let me be more specific and say I assume he knows that if the Fed bought up all of planet Earth it would “work”), but he is worried about the risks associated with unconventional QE.

    David, You are right that there are more problems with F&S’s book than I implied in this piece. I will atone with a future post sometime showing how Hawtrey’s account is more accurate than F&S. (For those who don’t know, both David and I developed neo-Hawtreyan views of the Depression separately in the 1980s. He is less fond of Friedman than I am. But we both favor Hawtrey’s view.)

    Nick, I agree with those mechanisms.

    Bill, I run into this sort of misunderstanding all the time. Economics courses need to do a better job explaining the relationship between NGDP and RGDP. They are unrelated in the long run, and closely related in the short run.

    Larry, I agree that we get better over time, but there are setbacks. After Strong died things got worse before they got better, and it’s pretty hard to argue with Strong’s record. But long term I agree with you.

  32. Gravatar of Devin Finbarr Devin Finbarr
    21. April 2009 at 19:58

    Bill and Scott,

    I strongly encourage you both to check out Warren Mosler’s article The Seven Deadly Innocent Frauds of Economic Policy. Also check out some of Winterspeak’s thoughts, here and here and here. My arguments may sound very odd, but it’s not because I am not familiar with standard macroeconomic theory. My views about QE and the nature of treasuries were once much closer to yours. But I read a bunch of stuff that made my head hurt, gave it a lot of thought, debated really smart people, and ended up convinced that the standard view is quite wrong.

    The best way to think of a T-bill is as a dollar with a not valid date. Thinking as T-bills as “debt” is a relic of the gold standard era when the bills were denominated in a currency the government could not print. But it simply does not make sense to think of T-bills as debt when USG has an infinite supply of dollars. And moreover, because everyone knows T-bills are backed by the full faith and credit of the USG, the bulk of the inflation happens when the government runs the deficit, and a much smaller amount happens as the T-bills mature. In other words, we already paid for the Iraq war, it was paid via seinorage at the time the government ran the deficit.

    Because the treasury has a not valid date, its current value is slightly below the value at maturity. A one year T-bill paying out $100, will have a net present value of ~$95-$99, depending on inflation expectations and time preference.

    If the Fed buys up those treasuries at the $99 price, the balance sheet of the bank or company has not changed. And if its balance sheet has not changed, it can not suddenly start spending that dollar. For whatever reason the bank was holding that T-Bill, presumably it still needs to hold that dollar. Banks, individuals, sovereign wealth funds, hold treasuries because they desire or are required to hoard a good that is backed by the full faith and credit of the U.S. Exchanging a dollar for that treasury does nothing to change that.

    Or think of it another way. Imagine there was a one day treasury bill. If the Fed engaged in open market operations to buy these bills, would that be inflationary? What about ten day treasury bills? Thirty days? One year? What would the inflationary impact be if the Fed monetized all T-bills with a maturity of 30-days? All the bills maturing in less than a year?

    And also, in a world where most private savings exists in money market funds that hold T-Bills as assets, the similarity between T-Bills and dollars is even more direct.

    I’d also note, again, that bank lending is not constrained by reserve ratios or base lending. It’s constrained by capital requirements and the lack of borrowers. ( Canada has no reserve ratio, but the same credit freeze) OMO changes neither, and thus does not stimulate lending.


    Further, an increase in the demand for the medium of exchange can be actualized by an individual by pulling money out of the stream of income and output.

    I do not understand this sentence. Money is never in a stream. All dollars, at all times, are in someones possession. An increase in the demand for dollars simply means that people are less willing to trade their dollars for other goods. ( this accordingly results in lower prices, and thus lower GNP. This gives the appearance that less money is flowing around and thus money “fell out of a stream”, but really no such thing happened. It’s a lot clearer to look at monetary economics through the lens of supply and demand.

    The increase in demand happens because people want to increase the ratio of dollars to expenditures, because they feel the ratio is dangerously low ( this happens after a bank run or asset bubble crash has wiped out paper wealth). But in refusing to trade dollars for other goods, aggregate demand falls, triggering the depression. Thus the only way for people to restore their desired dollar paper wealth to dollar expenditures ratio, without lowering dollar aggregate demand, is for the government to provide dollars and future dollars that people can hoard. The government can do this by issuing dollars or T-bills, the point is to repair the balance sheets of the private sector, and either will do just fine for that purpose. ( The only other thing the government could do is try and reflate the bubble, thus repairing the balance sheet by making non-dollar assets more valuable. I do not think trying to reflate bubbles is a good idea, for reasons that should be obvious.)

    I think that we are currently in a special situation where there isn’t much impact.

    So what is this special situation? Why is it happening?


    Devin, Tax holidays might work, but it would balloon the deficit.

    I said the government should print the money and keep printing until it hits it target, then stop. If it overshoots ( which is very unlikely, since as soon as it stops printing, demand for money will increase, undershooting is much more of a concern) it can always tax away the excess.

    Do you agree that a monetary injection should be as neutral as possible? As in, it should not be a wealth transfer? Let’s say the Fed decides it needs to increase the money supply by 20%. Should this all be done via open market operations? And say I have $50,000 in cash, short term cd’s and short term bonds. If the monetary injection is neutral, I should end up with $60,000 in cash, CD’s and bonds. How does this happen via open market operations?

    OMOs are much better, and cash and T-securities are not that close a substitute in the long run.

    Long term T-Bills and dollars are not close substitutes. Short term T-bills and dollars are very close substitutes. The difference between a 30-day T-bill and a dollar is 30 days.

    Try shopping at your local WalMart with a fist full of T-bills.

    Many big transactions occur via direct transfers between money market funds. What do you think that money market fund is holding? It’s not fist full of green bills…

    Krugman fears to much money would expose the central banks to dangerous capital losses if that had to pull the money back later, and thus sell bonds, to prevent an inflation overshoot. I think he worries much too much about this issue.

    The USG is a currency issuer, it never, ever has to worry about dollar losses (although the Fed does have to worry about a Congress that has no understanding of monetary economics). Or, if inflation overshoots it can simply can sterilize via raising taxes a bit or mucking with capital requirements. Really though, the only way to overshoot with inflation is if Wall St’s Infinite Loan Machine turns back on, and since that was really a form of legalized counterfeiting, it should be kept off permanently.

    Gold supplies were fairly stable in the period after the Civil War, but increases in the demand for gold would still create occasional depressions.

    The key is not to let the supply of notes saying “this is redeemable on demand for an ounce of gold” to outpace the actual supply of redeemable gold. Do that, and you will never have a bank run or depression. And if by some chance you do allow inflation under a gold standard, and then people start redeeming their notes, you must immediately devalue, the quicker the better. Obey these rules and a gold standard will be very stable.

  33. Gravatar of Scott Lawton Scott Lawton
    21. April 2009 at 20:26

    Bill typed:
    “How is it that the end of a housing bubble or “insane” leverage levels lead to a drop in real output? Why exactly?”

    Here’s the narrative: the apparent prosperity (GDP growth) was an illusion. Spending was funded by debt. So: the bubble bursts, no more home equity withdrawals, the artificial level of spending goes away, GDP goes negative.

    If true, then it’s quite easy to believe that an NGDP target will just make up the difference with inflation … i.e more artificial spending.

    Where does the argument go wrong?

    “With nominal income growing on target…”

    But the argument is that spending was fueled by debt not income. People were living beyond their means; borrowing from the future.

    (No need to answer in the comments, but I’d love to see a post on this.)

  34. Gravatar of Jon Jon
    21. April 2009 at 21:01

    “The commenter Jon says the short run only idea is a myth. He says the Fed has always bought lots of medium term debt.”

    Yes, I’ve clearly repeated myself enough on that point. I’ve been meaning to pull-together the data to show this graphically, but alas the relevant data is buried in unstructured plain text in the H4.1 reports.

    But I just realized that I had something “good enough” laying around. It doesn’t show the total balance sheet (although historically that’s irrelevant). You’ll have to trust me that the time period isn’t selective (2000-2007).


    I still hope to get around to calculating the mean maturity over 1980-2008, but alas, that may simply involve too much effort for mere rhetorical purposes.

  35. Gravatar of Bill Woosley Bill Woosley
    22. April 2009 at 02:31

    Scott Lawton:

    Do you think this is what Jeff had in mind?

    Anyway the story you tell is that consumption spending in the economy was funded by borrowing that used the equity in homes as collateral. With the end of the bubble, there is less equity in homes and so less borrowing and so less consumption.

    Leaving aside talk about “illusion” or “artificial,” when this consumption goes away, there is less spending, less sales, and so, less production.

    You have, like most people, come up with a “demand-side” account for the nature of recession. Again, less borrowing, less consumption spending, fewer consumer goods sold, and so, fewer consumer goods produced, and so less output.

    If nominal income targeting can be accomplished, there is no drop in spending and no drop in production. Now, you may believe that whatever spending is generated is artificial, but it is a contradiction to argue that low spending causes the drop in production and so the increase in spending must cause inflation.

    For what it is worth, there are deeper problems with this account. For every borrower there is a lender. During the boom, the consumption spending was funded by lending. What would the lenders have done with the money if they hadn’t lent it for consumer loans collateralized by home equity? The fundamental error in this approach is to assume they would have held money. Or, perhaps, that all of the consumer loans were funded by newly created money that people did not want to hold.

    Anyway, now for the burst bubble phase. The people that were funding consumption by borrowing against their home equity are no longer borrowing. What do the people that would have lent to them do with the money? The assumption here is that it is held as an increase in cash balances. (Or, perhaps, that less new money is created. Or that the quantity of money created by the bankign system shrinks.)

    Suppose that these over-indebted borrowers don’t just stop borrowing more, but begin to pay down debts. What do those receiving debt repayments do with the money? The assumption you are making is that they hold the money. They might. Or, that debt repayments involve a decrease in the quantity of money. Maybe.

    Looking at the problem in a different way, the end of the bubble reduced net worth. Assuming that people wanted that net worth for some reason, they must now increase saving to rebuild that net worth. An increase in saving reduces the natural interest rate. This has two kinds of effects. It reduces the quantity of saving supplied, which is more consumption and it increases the quantity of investment demanded, which is more spending on capital goods. There “should” be a reallocation of expenditure from consumption to investment.

    If this effect on interest rates was so large that “the” natural real interest rate becomes negative, and so negative that it is less than the expected inflation rate, then the zero bound on nominal interest rates becomes relevant. And so, you see the neo-wicksellian argument as to why the drop in housing prices is relevant. Because the financial system is based on zero interest currency, a sufficiently large increase in saving perhaps cannot be coordianted by sufficiently low nominal interest rate. The stories about holding money are answered “yes” not because of a failure to think through the issue, but because people will hold unlimited quantities of zero interest currency before lending it out at more than slightly negative nominal interest rate.

    Finally, most spending is funded out of current income. There were only two quarters in the last 10 years where personal saving was negative. All the quarters with postive personal saving far outstrip those two negative quarters. On the whole, U.S. households have been spending less than their disposable income all along. In aggregate, the flow of U.S. income has been been funding the flow of U.S. consumption. Obviously, some households have been borrowing heavily to fund consumption. There are other households that have been saving. The debt-funded consumption binge was a shift of consumption between households. So, the households that were in debt can consume less by failing to go even more in debt and even pay down their debts while the households that were saving can consume more of their income instead of saving it and lending it or increase consumption out of repaid debts.

    This isn’t to say that they should do this. It is rather that they could. Consumption spending could rise from the level of 2008 without any increase in debt and with debt decreasing. Failure to see this comes from overaggregation of analysis.

    And, of course, increased aggregate spending is consistent with increased aggregate saving by all households. It just requires higher investment–more spending on capital goods.

    Above, I told you a story about the zero negative bound. I have explained many times that I think that it does not apply to the current situation. Some interest rates are near zero, but others are well above zero.

    I think it is likely that the end of the housing bubble will increase saving and lower the natural interest rate. But, the stampede to safety–T-bills, FDIC insured deposits, and reserve balances at the Fed–is a separate issue. It is just some interest rates that are near the zero bound. Thinking about “the” interest rate is close to worthless when thinking about changes in the constellation of interest rates.

    Anyway, the “debt” story is based upon a tissue of fallacies regarding macroeconomics. It is a common error. Nominal income is the sum of spending on consumption, investment etc. Borrowing is used to fund some of this. Current debt impacts borrowing. Too much debt, less borrowing, less spending, and so, lower nominal income.

    Sounds plausible. But the monetary disequilibrium approach asks–what would have been done with the money? Where did the money come from? What happens to money? If the result is a shortage or surplus of money, then we have the explanation of why nominal income (aggregate expenditures) is impacted.

    The notion that too much debt depresses spending and lower spending reduces output, and increased spending causes inflation is a different sort of contradiction.

    As I explained before, if ending the bubble requires a reallocation of resources, then production and employment will be depressed. Nominal income targettting will result in higher inflation than otherwise would have occurred.

    But the notion that we had too much debt, and so, too much spending, and so we must now suffer less spending and less production–that is wrong.

    Oh.. this is a closed economy analysis. If we are looking at the U.S. and thinking about borrowing from other countries, then, we can, in aggregate, pay down debts. But only if we have a trade surplus. In other words, we stop buing chinese toys and underwear and we start sending them refriderators. Nominal income is maintained, along with production, but we don’t get the use of that production here in the U.S. We send more of it to foreigners.

  36. Gravatar of TheTradingReport » Blog Archive » links for 2009-04-22 TheTradingReport » Blog Archive » links for 2009-04-22
    22. April 2009 at 03:06

    […] Scott Sumner: How did Friedman and Schwartz persuade us? […]

  37. Gravatar of ssumner ssumner
    22. April 2009 at 03:41

    BTW, I am aware of Mankiw’s new piece on negative interest on reserves, and will have a long response (hopefully today.) Because of classes today it may have to wait until tonight. Please be patient if some comments don’t get addressed until tomorrow.

  38. Gravatar of Bill Woolsey Bill Woolsey
    22. April 2009 at 07:39


    I aways understand monetary economics thinking about supply and demand.

    There is a flow of expenditures and production and income. The reason for focusing on that is because that is what is important. Income and output. People use the income they earn to purchase goods and services. That is mostly what is happening in the real world. Well, I guess I would say that is what is most important.

    Do financial transactions enhance this (say by shifting consumption between households or allowing resources to be allocated away from consumption to investment) or do financial transactions disrupt the process. (People saving by allowing the money they earn as income to accumulate as growing incomes by not spending that money on anything else.)

    Your arguments that T-bills are the same as money are false. Your focus on the possibility that T-bills could be paid off by the creating of money are pointless. That is, they could be paid off that way, but you cannot draw conclusions about the real world if you assume that they are always paid off with newly created money. Otherwise, a shortage of T-bills results in higher prices and lower yields for them. A surplus of T-bills results in lower prices and higher yields for them.

    It is possible that T-bills will by higher taxes, paid off by reduced other spending, or simply not paid off. The special case where they are always paid off with newly issued base money can’t be the basis of monetary theory.

    I personally own no T-bills. I have a small amount of currency in my wallet now. The notion that if somehow, my portfoilio of T-bills when up to $1000, I would get rid of my currency (in my wallet) to bring my holdings of goverment guaranteed securities as close to the current $20 is false. I know it is false.

    I am not sure exactly where you are going wrong, but I suspect is that you are assuming that the reason people hold base money is for the same reason they would hold a share of Apple stock. And that T-bills and base money are related in the same way as Apple common and preferred stock. And so, in equilibrium, they must perform in similar ways.

    As far as I can tell, this sort of reasoning comes from the fact that we have better formal models of investment in equity (well…) And this medium of exchange business is difficult to formalize, especially if it all has to be based upon intertermporal utility maximization of representative agents. And so, by identifying “science” with formalism, we ignore everything we know about the medium of exchange.

    But, perhaps I am wrong and your headaches come from some other source.

    I know why I have money in my checking account. It is passing through between income and expenditure. When my ballance is low, I reduce by expenditures. When it is high, I spend more. It is not an investment in Bank of America. Why do I accept money for payment? It is because I believe that people will in turn accept it. It is not because I consider BoA a good investment prospect.

    My currency holdings are trivial, but I certainly am not holding the $20 of curreny in my wallet today because I think it is a good investment in the U.S. goverment. I hold it because I may want to buy something with it and I believe it will be accepted. That is what the medium of exchange means.

    Anyway, the special condition today is that there is a very high demand for T-bills and the shortest one have very low interest rates. If the interest rate turns negative to the point where it approximates the storage cost of currency, then the lower bound will have hit. The demand for currency is really a spillover from the demand for T-bills. If that is the situation, then issuing currency and buying T-bills will have no impact. The zero (or really slightly below zero) bound on nominal interest rates results in a shortage of T-bills that cannot be cleared by having T-bill prices rise and their yields fall to a level that is sufficiently negative. That shortage spills over into a shortage of base money. Fixing the shortage of base money by exacerbating the shortage of T-bills will not be effective.

    If the shortage of T-bills at the zero (or slightly less than zero) lower bound on yields caused the frustrated T-bill buyers to go out to eat at restaurants (couldn’t get any T-bills, might as well go out to eat,) then this would not cause a decrease in aggregate expenditure.

    They could print out currency to pay off T-bills, (but maybe they won’t) so T-bills and currency are really the same thing, so open market operations in T-bills cannot matter, well it breaks down after the first step. What might happen doesn’t determine what will happen.

  39. Gravatar of Bill Woolsey Bill Woolsey
    22. April 2009 at 07:46

    the disruption is people saving by allowing the income they earn to accumulate as growing money balances rather than spending.

    I think the economy is about people achieving their goals, yes, but the key element of this is people supplying labor to earn income to purchase consumer goods and services. Investment is about shifting resources from the production of consumer goods to capital goods so that more consumer goods can be produced in the future. Labor, consumer goods now and in the future, capital goods to enhance the production of consumer goods, etc.

    I sometimes wonder if people who have a special interest in finance have a different perspective because they don’t see this as what is “really” going on.

    To me, finance is about shifting funds between households and firms to enhance this process. Somehouseholds consumer more, others less. Resources shift from consumer goods to capital goods, or in reverse. Monetary disequilibrium is a problem because it disrupts this process–the process, the flow of production, income, and consumption.

    I am sure that this basic vision colors my approach.

  40. Gravatar of Scott Lawton Scott Lawton
    22. April 2009 at 17:03

    Bill: thanks much for the detailed reply. I think this issue is important, so here’s more.

    Bill typed:
    “Do you think this is what Jeff had in mind?”

    Not sure, but in any case the view became more common as housing prices kept moving farther above historical norms. Here’s an Aug 2008 look back at a Nouriel Roubini “warning” in Sep 2006:

    Another source: finance guy Barry Ritholtz at his blog, The Big Picture: … though of course it would take time to wade through the archives.

    Yet another: … though he also predicted a huge collapse in the dollar, which IMHO ignores the “flight to safety” issue.

    (I don’t claim that my quick narrative does justice to their arguments, only that they were visible skeptics on the way up.)

    “For what it is worth, there are deeper problems with this account. For every borrower there is a lender.”

    Agreed. I think it’s one of the classic “seen vs. unseen” issues, like Bastiat’s broken window. (Yet another bit of econ that’s not intuitive.)

    But events seem to have “proven” the narrative, so it takes extra effort to explain.

    “What would the lenders have done with the money if they hadn’t lent it for consumer loans collateralized by home equity?”

    My guess: invested somewhere else (possibly in a different unsustainable manner!) with the “help” of financial companies who paid huge bonuses based on short-term results.

    “An increase in saving…”

    Including (I think) recyled dollars from China and the oil countries.

    “An increase in saving reduces the natural interest rate … which … increases the quantity of investment demanded, which is more spending on capital goods”

    I think that’s actually one of the issues that got us into this mess.

    The argument: money (especially from overseas) was chasing both safety and return. Sub-prime created return. Banks sliced things up, “insured” them, and got the rating agencies to declare them “prime”. Everyone in the middle got a fee, and the investor got a “safe” product with a somewhat higher return.

    And the homeowner got cash to spend.

    Politics helped things along: Clinton, Bush and Congress all pushed to increase home ownership. Fannie/Freddie seem to have “done their part” by chasing return (and compensation) rather than stability.

    The Taylor Rule argues that Greenspan (and the Open Market committed) had a role, by holding interest rates down too long. (Higher rates could have had 2 beneficial effects: fewer sub-prime mortgages and more incentive to put money in plain-old money market funds and such.)

  41. Gravatar of ssumner ssumner
    22. April 2009 at 17:49

    It’s late so I’ll just make a few brief comments:

    Devin, When you said MMMFs don’t hold a fistful of dollars, that’s exactly my point. Federal Reserve notes are different from other assets, and they are used for different types of transactions. My new post today (Response to Mankiw) has more to say about this, if you are interested.

    I don’t know why you’d want stimulus to be completely neutral. Isn’t the hope that it will raise real GDP? Now if you mean normal monetary policy then I might agree. If you go for 5% NGDP growth or 2% inflation, then that can be built into contracts–and should be neutral if expected.

    I don’t see your point about the gold standard. If the gold standard only works when the supply of paper money is kept fixed, why not just skip the gold standard and go with a fixed paper money system?

    We know that OMOs have powerful real effects, and yet according to your theory they should never have any effects, it’s just swapping cash for T-bills. But these operations strongly effect real stock prices and real GDP.

    Scott, I agree with some of Bill’s response, but will just add this if it helps. The supply-side must be modeled. Thus if you are going to argue that less consumption causes output to fall you need something like sticky wages as an explanation. But then if you have sticky wages you can no longer say an increase in NGDP will merely raise prices, it will also raise real GDP.

    Jon and Bill, Thanks. I’m out of time for now.

  42. Gravatar of Devin Finbarr Devin Finbarr
    22. April 2009 at 18:05


    Do you agree that there is no default risk on treasuries? That the government will never, ever bounce a dollar denominated check? Certainly the government and the market both believe this. Every government regulation treats treasuries as zero default risk. Capital requirements treat treasuries as zero default risk. Virtually every trader, bank, sovereign wealth fund, etc treats treasuries as zero default risk.

    Note that the actual supply of currency (M0) is $2 trillion and yet outstanding Treasury obligations are $11 trillion. Suppose the treasuries were obligated in a non-fiat currency. Suppose there were only 2,000 tons of gold in the world, and the government owed 11,000 trillion tons. What would the default risk of the government be? Answer: pretty darn close to 100%. Certainly not anything close to 0%. In this situation, a Treasury note would be toilet paper.

    Yet in real life the default risk of the government is 0. Just the facts that a) the government can print money and b) the government has promised to never default on treasuries mean that treasuries have zero default risk, even though the government probably never will actually print the money.

    In reality, the treasuries just roll over when they mature. It’s actually much more convenient for the private sector to hold short term treasuries than actual dollars. Short term treasuries can be held as assets backing zero-term bank accounts and money market accounts. The owner of the account has zero risk, can draw a check against the account, plus he gets a bit of interest. A very short term treasury bill has all the advantages of a dollar, plus it earns a bit of interest – it’s like a dollar++.

    I know why I have money in my checking account. It is passing through between income and expenditure.

    Of course, you don’t have money in a checking account. At least not currency. A checking account is just another example of how a government guarantee to print a dollar to back an asset, means that asset is worth at least a dollar, even if it never actually prints one note.

    Without the FDIC guaranteeing your checking account, your account would vaporize. Our banking system is basically 0% reserve, if government guarantees were withdrawn, there would be the mother of all bank runs.

    What M1 and M2 represent are “contingent dollars”. And as long as the government’s promises are good and the printing press unbroken, a “contingent dollar” is as good as an actual dollar. A T-Bill is a “contingent dollar” with a not-valid until date.

    The only reason this bizzarro world of contingent dollars exists, is because there is still populist pressure not to actually print “real dollars”. So the government inflates by guaranteeing various assets ( bank accounts, treasuries, money market funds), thus converting them into contingent dollars.

    Anyway, the special condition today is that there is a very high demand for T-bills and the shortest one have very low interest rates. If the interest rate turns negative to the point where it approximates the storage cost of currency, then the lower bound will have hit. The demand for currency is really a spillover from the demand for T-bills.

    What people demand right now are assets that will not fall in dollar price. Dollars will not fall in dollar price by definition. And if you are willing to hold a treasury until maturity, neither will a treasury. Since a treasury pays a tiny bit of interest, it is preferable to holding an actual dollar.

    We may be really be close to being on the same page. It’s just what you consider a “special case”, I consider current reality. The average maturity of treasury bills has grown shorter and shorter over the years, which means they are becoming close and closer to dollars.

    From your other post –

    for every borrower there is a lender.

    In a sane world, this would be true. Unfortunately, we do not live in a sane world. In fractional reserve reality, banks literally create money out of thin air. A bank makes a loan, and if that loan causes its reserves to fall below the required level, it borrows the reserves from the Fed. All that matters are capital requirements, the availability of borrowers, and the federal funds rate.

    You may also notice, that in a world where every borrower was matched with a lender, there would be no way for M2 to double in a decade. But it did.

  43. Gravatar of Devin Finbarr Devin Finbarr
    22. April 2009 at 18:28

    I don’t know why you’d want stimulus to be completely neutral. Isn’t the hope that it will raise real GDP?

    By neutral, I mean a reflation should not redistribute wealth. For example, let’s imagine a case where every dollar bill with an odd number serial number disappeared. USG would obviously want to reflate. Krugman wants to reflate by spending money on construction. This is will result in a huge transfer of real resources to the government, to solve a purely nominal problem. You seem to want to reflate by putting money in the hands of banks ( presumably so they will lend it out?). But this will result in a transfer of real economic resources to activities funded by borrowing. I want the government to simply replace the vanished bills. If there is no good accounting of who lost the bills, the government should still aim to be as neutral as possible.

    I don’t see your point about the gold standard. If the gold standard only works when the supply of paper money is kept fixed, why not just skip the gold standard and go with a fixed paper money system?

    Sure, that would work too – if you trust the government. The reason people like gold standards is that it keeps the government honest. If paper is redeemable in gold, the government is limited in the amount it can inflate, as people will start redeeming their notes if they feel the notes are getting devalued.

  44. Gravatar of Greg Ransom Greg Ransom
    22. April 2009 at 19:16

    I’m all for this:

    “I need to do a post on the 1921 depression sometime.”

  45. Gravatar of Devin Finbarr Devin Finbarr
    22. April 2009 at 19:16

    Scott –

    We know that OMOs have powerful real effects, and yet according to your theory they should never have any effects, it’s just swapping cash for T-bills.

    When the Fed uses OMOs to change the federal funds rate, that has substantial effects. Using OMO to change the federal funds is something of an anachronism, a leftover from fixed exchanged rate world. The Fed could just set the rate directly, it would be the same thing. But right now, the Fed Funds rate is already at zero. If you use some sort of negative interest rate scheme, then yeah, you can get an arbitrary amount of inflation.

    It’s straight out debt monetization that my theory predicts won’t work, and I don’t see any evidence that it ever has.

  46. Gravatar of ssumner ssumner
    23. April 2009 at 06:00

    Devin, I know the first comment was addressed to Bill, but a couple observations. Cash is nothing like T-bills for three reasons:

    1. Small denominations (useful in transactions)
    2. Fixed nominal value (useful in transactions)
    3. Anonymous (useful in tax evasion.)

    As a result the demand for cash is almost totally unrelated to the demand for T-debt. The first two points explain the transactions demand for cash, which is actually the smaller demand. For the sorts of things that people use cash for (small transactions) T-bonds are not even a close substitute. Nobody would even think about substituting T-bills for cash in their wallets. On the other hand cash is a poor investment compared to FDIC-insured deposits or T-bills. The other demand for cash is related to tax evasion, and again, T-bills are a very poor substitute, because they are not as anonymous. Thus people often hold large cash balances (in the millions of $s) even when T-bill yields are 5% or 10%. Why? To evade taxes.

    Sharply falling AD (like right now) hits the rich and the poor unemployed hardest. But the middle class also suffers. I am simply calling for reversing that process. It would help the rich and the poor the most, but also help the middle class. I don’t see redistribution as an import long run effect of M-policy, as long as it is aimed at something like a stable inflation or NGDP rate

    So we need a gold standard because we can’t trust the government? And you say a gold standard only works if the government handles it the right way? I don’t get that. The government created the worst economic disaster in American history under the gold standard of 1929-33—how can we trust a government to run a gold standard any better than a fiat regime?

    Greg, My hunch is that we’d agree on that one. The 1921 depression was handled much better than 1929-30. I think libertarians and Austrians are basically right about 1921. BTW, I also think Harding is underrated on not just economic grounds, but many others as well.

    Devin, No, the Fed cannot directly set the FFrate. It can only target the rate by changing the base. The base is the key to monetary policy, the FFrate is a trivial side effect.

    By the way “debt monetization” and open market purchases are exactly the same thing. So every time they monetize debt it effects the FFrate (except at zero rates.) So if you say debt monetization doesn’t matter, then neither does changes in the ffrate.

  47. Gravatar of Bill Woolsey Bill Woolsey
    24. April 2009 at 06:49

    I don’t agree that there is zero default risk for government bonds. As a voter, I have a small political influence on the government. I can imagine that others could think like me, (not that I expect that to happen in this regard.) And I would support a default–partial payment of govenrment bonds, to creating money to pay off bonds.

    So, not only is it possible that government would default on debt. I know one person who prefers that to printing money to pay off bonds. I would advocate a nominal income growth rule and that would limit money creation. Using money creating to pay off debts–violates the rule. My prefereds solution would be to cut current expenditures. Perhaps some kinds of tax increaces would be acceptable. But printing up money, never.

    Anyway, basing monetary theory on the assumption that debt is always paid off with newly printed money is false. It could happen. That doesn’t mean it well happen.

    The notion that FDIC insurance is necessary for the existence of checkable deposits is false. There were checkable deposits before FDIC insurance. There have been checkable deposits under fiat money regimes without deposit insurance.

  48. Gravatar of Devin Finbarr Devin Finbarr
    24. April 2009 at 07:59


    If the national debt was denominated in gold (say it was 10,000 tons of gold) and there was only 2,000 tons of gold in the world, what do you think the default risk would be?

    And what do you think the default risk now, in our world of fiat money?

    You don’t have to be exact for either, just give me an order of magnitude. Tell me what you personally would do.

    The notion that FDIC insurance is necessary for the existence of checkable deposits is false. There were checkable deposits before FDIC insurance.

    Uninsured checking deposits barely worked when reserve ratios were much higher than currently. There were continually bank runs and suspensions of redemption. Under current ratios of obligations to base money, checking accounts would vaporize if people honestly believed that the government would not bail them out. I sure as heck no I would be running to the bank.

    Were you paying attention to the run on the money market funds last September? Do you think the Fed was right in stepping in to guarantee them? What do you think would have happened if the Fed had allowed the run to continue?

  49. Gravatar of Jeff Jeff
    24. April 2009 at 10:27


    Sorry about taking so long to respond, I’ve been pretty busy with my day job. I’m telling a supply story, not a demand one. The housing crash and the subsequent financial crisis has severely damaged financial intermediaries and now hampers their ability to perform their usual function. This is a supply shock, akin to throwing sand in the gears. Responding to an adverse supply shock with expansionary monetary policy is a big mistake. See the 1970’s.

    Scott, in your response to me you said that pegging NGDP at 5 percent would only involve 1 or 2 percent inflation. In other words, you just don’t believe the housing crash can cause a drop in real output. Well, Reinhart and Rogoff have looked at dozens of financial crises in many different countries and found that debt crises usually do end up in recession. Now it may be that the central banks in each and every one of those crises was really the one at fault, but still, doesn’t it give you the slightest pause?

  50. Gravatar of ssumner ssumner
    24. April 2009 at 17:01

    Devin, I don’t agree with your view that the banking system was fragile before FDIC. Take the Great Depression. By October 1930 we had suffered a worse depression than anything since WWII, our modern system would be devastated by that type of depression. And guess what, we had not yet experienced even one banking panic yet. The first panic, in November and December 1930, was fairly small. Our pre-1933 system was very strong. Now it is true that the Great Depression got so severe than many banks did eventually fail, but most survived. Few modern banks could survive such a fall in NGDP. If the Fed had not caused the Great Depression, our pre-1933 banking system would have done fine–it was far superior to the modern system, much more conservatively managed.

    Jeff, During the worst of our financial crisis (second half of 2008), the dollar appreciated strongly against the euro. Please go out and check how many of those crises examined by Rogoff saw their currencies experience a powerful rally in the teeth of the crisis, and then we can discuss their relevance to the American supply shock/demand shock debate.

    Here is the bottom line. In a supply shock output falls and inflation rises. In a demand shock both output and inflation fall. So what does this recession look like?

    According to your view FDR’s dollar devaluation program should have produced lots of inflation and not much real growth. After all, the banking system was by then even more flat on its back than today. In the first 4 months the WPI rose 14% and industrial production rose 57%. Now I will grant you this, after July 1933 FDR raised wages 20% (a supply shock) and output slipped into stagflation. We are not doing anything that bad, but there are some disturbing moves underway in Washington. Even so I’ll stick to my 1-2% trend growth as the likely worst case from the statists in Washington.

    People that know a lot more about banking than me insist that there are still 1000s of sound banks in the U.S., and that firms unable to get funds from one bank can go to another. Lending is still occurring. I do know that standards have tightened, so it is harder to get loans. But this partly reflects the higher default risk from the fall in NGDP. If NGDP were rising at 5%, defaults would drop off sharply, and the banking system that now looks dysfunctional would suddenly spring back to life, just as in 1933.

  51. Gravatar of Alex Alex
    25. April 2009 at 08:01


    “By October 1930 we had suffered a worse depression than anything since WWII, our modern system would be devastated by that type of depression.”

    I don’t understand the first part I think you got some dates mixed up. About how our modern system would have had behaved in a situation in the 30’s I think you are right. If we had a drop in NGDP between 11 and 24% like in the 29-32 period then the banks would be toasted the thing is can we have one now? You seem to see the problem of the banks as a separate issue, while I think it is all related, NGPD falls banks fail, bsnks fail M falls, M falls NGDP falls even further (by M falls I mean money supply relative to demand). In 2008.9 the Fed did not stop the initial fall in NGDP but it sure did stop the cycle by not letting the banks fail and by injecting liquidity into the system. This brings me to thinking that actually Bernanke does not have any intention to inflate the economy. The injection of liquidity was never intended to spill out of the banking system to the rest of the economy. The Fed just wants to show that it is ready to fullfill its roll of lender of last resort and it did it by putting the cash in advance but to prevent that the cash leaves the system it is paying interests on reserves.


  52. Gravatar of ssumner ssumner
    25. April 2009 at 12:39

    Alex, No my dates are exactly right. October 1930 was 14 months into the Depression, and it was already worse than anything we have had since WW2. And there were as yet no banking panics. That was exactly my point. By early 1933 NGDP had roughly fallen in half.

    I agree that causality goes both ways, but in the Depression it was definitely falling NGDP that caused the initial banking problems. Once bank panics occurred, that clearly made the Depression even worse. Without a Great Depression, however, the U.S. banking system would have been fine during the 1930s.

    I don’t agree that Fed policy helped in late 2008. The money that was injected did no good, as the Fed paid banks to hoard it. I also don’t understand your comment about trying to prevent the cash from leaving the system. The cash does absolutely no good unless it leaves the system. If the banks hoard the cash as ERs, then the multiplier simply falls (or velocity falls if you prefer.)_

  53. Gravatar of Clayton Clayton
    26. April 2009 at 15:03

    I wanted to address Scotts supply vs. demand argument, but I think it’s more effective to drill right down to the NGDP argument and back into my supply vs. demand issue (if it ends up relevant).


    I certainly agree with Scott that the failure to achieve inflation expectations (or even try to target inflation) has had negative real effects. In this regard, the Fed has certainly and categorically failed in at least one of its mandates by exacerbating the recession. I’m not even sure a die hard rational expectations economist could conclude otherwise since, at some point, inflation was less than expectations (which should be contractionary).

    However, I have trouble getting beyond this point. Assuming the Fed always succeeds in inflation targeting (or that they use trend targeting to make up for unsuccessful years)… what benefit comes from taking the next step to NGDP targeting?


    The only answer I can come up with is that Scott believes either:

    (1) that money is not truly neutral if we successfully target trend inflation to 2%


    (2) if real growth does not hit a specific target, (2a) it must be due to a non-rational real factor and this irrationality (2b) can and (2c) should be “corrected” through monetary policy

    … or perhaps some combination of both (1) and (2)


    I started following the arguments on both sides, but neither were particularly concise so I’ll start by asking Scott if he believes either of these to be the case (and the basis of his NGDP policy) or if I’ve missed some other line of logic justifying NGDP.

  54. Gravatar of Clayton Clayton
    26. April 2009 at 15:05

    Oh… and since it may help you understand why I came to these two conclusions, let me follow by saying…

    If (2a) is not true and the real growth has slowed for a rational reason (i.e. a fall in productivity), then a NGDP target would call for increased inflation and either:
    – Rational expectations would completely eliminate the impulse (which makes inflation targeting equally effective) or
    – The inflation would drive real factors above “classical” equilibrium. This, being unsustainable, could only contribute to economic distortions (capital misallocations, a bubble, etc.).

    In either case, this seems to be undesirable. Indeed, it seems to me that the optimal real growth trend *must* be exactly the same throughout the years for this to work *as well* as inflation targeting as any unexpected change in real growth trends could only cause inflation expectations to deviate from inflation (NGDP trend net real growth).

  55. Gravatar of ssumner ssumner
    27. April 2009 at 15:32

    Clayton, I agree with Earl Thompson that a nominal wage target might be the best way to stabilize the economy, and prevent business cycles. I don’t think that is politically feasible. NGDP is the next best thing. Suppose there is a boom, but inflation is stable (as in 2004-07). In that case NGDP growth will accelerate and so will nominal wage growth. My plan would have a bit tighter money in years like 2004-07, to slow the boom a bit, and more expansionary policy during recesion years to reduce unemployment. Under my plan the average rate of inflation would be no different, but output would be more stable

    In my view business cycles are not caused by price level instability, but rather nominal wage instability. Because nominal wages are sticky, when the equilibrium nominal wage rate fluctuates, it causes employment fluctuations. So you need a monetary policy where workers will want to voluntarily ask for the same pay increase every year (in aggregate.)

  56. Gravatar of Bill Woolsey Bill Woolsey
    28. April 2009 at 02:40


    Please explain exactly how the sand in the gears caused by problems in real intermediation cause a reduction in output.

    Sand in the gears, less production, more inflation…. I want more than that.

    In my view, the job of credit markets is to move funds from less to more valuable uses. If the funds remain in their less valued uses, how much real loss would be generated?

    I think it comes down to reallocation of resources again. That is, structural unemployment and difficulties in expanding areas of the economy where there is demand relative to the ease of shutting down production where there isn’t demand.

    However, with nominal income targeting there will be sectors of the economy with growing demand, rising prices (the inflation we will observe,) rising profits, production, and employment.

    Firms already finance a good bit of investment with retained earnings. Households already fund almost all of consumption out of current income. Stable nominal income with no credit transactions basically means those things become 100%. Remember, for every borrower who uses credt to spend, there is a lender who is lending rather than spend.

    Just about every account of this process that suggests credit problems sneaks in monetary disequilibrium and a drop in aggregate expenditure. Generally, the second best usage is assumed to be an accumulation of money balances, and so the failure of credit markets involves a disruption of the flow of output, income, and expedenditure–less spending, fewer sales, and less production.

    As for other crises, Scott asked how many involved an appreciation of the exchange rate. More to the point, I would say that central banks efforts to defend the exchange rate is the key reason for a nominal contraction. Avoiding (or dampening, more likely) and increase in the “cost of living” coming through higher prices for imported consumer goods, amounts to the same thing. Further, having a large foreign debt denominated in foreign currency may well create an extra incentive to maintain the external value of the currency.

  57. Gravatar of Jeff Jeff
    28. April 2009 at 09:20

    Scott (and Bill):

    I don’t disagree with you as much as you think. I do agree that a big drop in expected inflation and a rise in the value of the dollar indicates that money was too tight. My only disagreement is with the choice of target. I favor targeting a path for the expected price level rather than a path for nominal GDP, as the former is less likely to ease money in response to an adverse supply shock.

    So far as we know, there has never been a successful large barter economy. The double coincidence of wants is just too high a hurdle. But money is more than a medium of exchange, it is also the numeraire. As such, the only way it’s value can change without distorting other relative prices is for every price in the entire economy to change in exactly the same proportion. This has never happened, and probably never will.

    The implication is that inflation and deflation have never been neutral and most likely never will be. Changes in the value of money change relative prices, and to the extent that these changes are not easily predicted, they increase uncertainty and reduce welfare. Almost everyone agrees that unexpected changes in the price level are bad, but I’d go further and say that even expected changes are bad. You can see my argument for this here.

  58. Gravatar of ssumner ssumner
    28. April 2009 at 17:15

    Jeff, That’s a nice article, I have a flood of comments in the newer posts, so I’ll just make a couple brief points, and try to do an NGDP vs CPI post this summer when I have more time to respond. My two brief points for NGDP are:

    1. Does better if wage stickiness is a bigger problem than price adjustment (as I believe to be the case.)

    2. During oil shocks it’s best if non-oil prices don’t have to fall. Actually this relates to my wage argument, because oil shocks require lower real wages. If nominal wages are sticky and you target the CPI, unemployment will rise

    But I will say this, either a NGDP or price level target would be much better than the current mess. So I agree we are not that far apart.

  59. Gravatar of Clayton Clayton
    29. April 2009 at 05:29

    Incidentally, there’s a pretty interesting (though massively counteintuitive) argument for (1) being true… which we may want to return to. However, let’s follow your logic:

    “In my view business cycles are not caused by price level instability, but rather nominal wage instability. Because nominal wages are sticky, when the equilibrium nominal wage rate fluctuates, it causes employment fluctuations. So you need a monetary policy where workers will want to voluntarily ask for the same pay increase every year (in aggregate.)”

    So you’re argument is (2b), specifically nominal wage rigidity. Incidentally, I think a lot of people will agree on (2b) but for a number of vastly different reasons (with different policy implications). But let’s roll with it…

    Are you arguing that upside wage rigidity is also a problem? I don’t hear a lot of people complainging that they cannot switch jobs or get adequate raises in a boom. Further, I don’t think the historical record is necessarily clear. Inflation during the 2004-2007 period averaged closer to 3% per year which would imply a somewhat more stimulative monetary policy. 10-yr TIPS spreads (I couldn’t quickly find shorter period data) during the period averaged closer to 2.5% – Rational expectations says that trend inflation targetting would have been less stimulative during those years as well.

    Additionally, I think the real question is the the relation between aggregate income and prices (which should drive real decisions). If we temporarily assume that wages are below equilibrium (monetary policy is racing ahead of wages), this need not automatically shift real production out of equilibrium. Indeed, high corporate profits during the period suggest that, if labor was underpaid, much of it was being paid to capital holders. We may want to discourage this wealth transfer on moral grounds, but it shouldn’t necessarily interfere with equilibrium production or aggregate income.

    [As an aside and sorta like the “aha” in (1), this raises questions about what you really want to trend. If you target trend labor-income and income share has been temporarily shifted to capital, a labor target could easily result in distorted real purchasing decisions. Incidentally, I’m not sure if NGDP has the same flaw.]

    Let me stop there… is upward wage rigidity really a serious problem or are we mostly fixing downward wage rigidity?

  60. Gravatar of ssumner ssumner
    29. April 2009 at 15:06

    Clayton, I think they are both flaws. A lot of people believe that easy money contributed to the housing bubble. I am doubtful that is was the key factor, but it may have played a role. However I do believe monetary policy was too expansionary during 2004-07. An expansionary policy would merely raise inflation, leaving real GDP unchanged, if there was no wage stickiness. The problem is that when NGDP rises, wages are slow to adjust, and thus real output also rises. So yes, that is a problem.

  61. Gravatar of Clayton Clayton
    5. May 2009 at 04:39


    I believe that the fundamental problem in your argument is not the basic NGDP logic but the argumentation.

    You’re basically saying… believe my interpretation because this evidence agrees with it… when the evidence also exculpates the popular, current interpretation (or at least exculpates trend inflation targeting). You can’t prove that NGDP is right/better (in their minds) by using “too stimulatory” evidence that would also be “too stimulatory” under a trend inflation target.

    People don’t leave models because you present them evidence that — at least on the surface — also support their existing model. Even a deep mathematical “proof” that showed that NGDP was closer to predicting real output would likely be within the margin of error of things like calibrations of (and assumptions in) the model — never mind that *you* don’t trust macro-models).

    I actually thing NGDP has some merit to consider… but only because I looked at the model in a way that DOES NOT permit a partial equivalence of the evidence. Consider presenting a question more like this:


    Independent of the source of the problem, (I believe) we all accept that the savings rate needs to rise to restore equilibrium to the economy.

    All things being equal, an increased savings rate — a fall in consumption and a rise in savings/investment — would (at least temporarily) drive down the price of consumer goods and up the price of investment goods.

    As a result, NGDP neutrality would be CPI disinflationary (or deflationary). CPI trending should be NGDP accelerating. As we have to differing monetary prescriptions to analyze (and thus one must present evidence either for or against a model):

    – What transmission mechanisms exist that would drive this difference into the macro-economy?
    – Which is actually “neutral” and why?
    – Can you really achieve neutrality if you’re only looking at half the puzzle?
    – Does PPI (no) or some other index (labor maybe, others maybe) address this question?
    – Does NGDP address this question?


    Notice that this looks more like (1) in my original logical structure, that CPI is not a “neutral” policy.

    I think you can start to build a case for NGDP that is diametrically opposite to the case for CPI.

    For example, under NGDP policies… downward wage rigidity would combine with CPI disinflation/deflation to create unemployment on the consumer goods side. This is bad from a full-employment perspective. However, most economists could be convinced that it’s actually *necessary* to create this structural unemployment so that the labor can be reallocated to investment (capital goods) industries — both on simple macro-allocation grounds and due to the more micro short term disequilibrium in the price of these goods.

    You also get to play some rhetorical games. There are plenty of people who think that “core CPI” is wrong because it excludes food and energy. You can now leverage that “objection” by saying… why exclude all of these other things as well?

    Unless you see an obvious flaw, I would float the very small story among your colleagues and see what kind of response you get. The simple thought experiment was enough to convince me to question my implicit assumption that CPI is “neutral”. Unless someone provides a compelling counter-argument (in which case I’d love to see it), you could consider posting the same thought experiment on here… see if you can undermine more people’s confidence in CPI as a “neutral” variable.

  62. Gravatar of ssumner ssumner
    6. May 2009 at 11:15

    Clayton, I don’t follow your first comment. The evidence also favors what currently popular theory? I mentioned booms because many critics of inflation targeting say it led to overly expansionary policies in 2005-06 and 1999-2000. If so, NGDP would have been better. That was my only point.

    I simply sketched out the reasons I like NGDP targeting. It would take a 50 page research paper to prove it’s superiority over something like inflation targeting, and even the 50 paper paper won’t convince many people who weren’t already inclined to accept the hypothesis. My response was not intended to convince anyone.

    You have some interesting ideas, but my plate is full right now so I probably won’t address this issue for a while. My current focus is on monetary stimulus.

  63. Gravatar of Clayton Clayton
    7. May 2009 at 17:50

    You keep going back to this 2005-06 policy thing…

    I repeat… Inflation during the 2004-2007 period averaged closer to 3% per year [and 3.3% from 2005-2006]… 10-yr TIPS spreads during the period averaged closer to 2.5% (I couldn’t find shorter data)… Rational expectations [RE] says that successful inflation targetting would have been less stimulative during those years as well.

    Here’s a way better chart for short(er) terms:

    Unless the short term TIPS spreads are way higher (not a natural conclusion given the term structure implied by the chart), this period is “overly expansionary” in *both* RE inflation targetting and NGDP frameworks.

    **That was my first point… you can’t convert inflation targetters to NGDP if your *primary* evidence comes to the same “overly expansionary” conclusion in an RE inflation targetting framework. If you’re trying to persuade to NGDP targetting, you have to pick data where NGDP is right and RE inflation targetting is *wrong*. The same point would be made since the fall… inflation is below expectations… so you don’t disprove inflation targetting… so you’ll never convince someone to give it up in favor of NGDP on that evidence alone.

    If you just want the Fed to manage and meet inflation expectations, you’d be better off dropping the whole NGDP framework — it distracts from your key point. You could get all the inflation targetters on board by arguing that the Fed failed to hit RE targets. You could have your debate about whether or not QE can reverse any disinflation.

    Instead, you use the NGDP argument or conclusion so liberally that it sound like one has to agree with NGDP to agree with your policy prescriptions (or at least it takes a nobel prize to separate the two).

  64. Gravatar of ssumner ssumner
    21. May 2009 at 04:24

    Clayton, These are all good points. I was probably wrong about 2005-06 as you say. I was partly responding to Austrian types who thought NGDP targeting might not have been more contractionary than actual policy during those years.

    I suppose for some examples (1998-2000) I rely on widely publicized arguments that inflation targeting threw us off. I don’t know if those widespread perceptions were accurate, but if they were then NGDP would have done better.

    You are right about the last half of 2008. But what about December 2007 to August 2008? Didn’t inflation send a very different signal from NGDP? And didn’t NGDP turn out to be right in that case?

    Having said all that, your point about 2005-06, and the last few months does shift me a bit toward the view that inflation targeting would be almost as good as NGDP. I should emphasize that point more often. My argument for NGDP futures targeting is really an argument for SOME form of futures targeting.

  65. Gravatar of IS-LMが嫌いな理由 by Scott Sumner – 道草 IS-LMが嫌いな理由 by Scott Sumner – 道草
    9. October 2011 at 03:52

    […] PS. 種種の問題を部分均衡的な見方(金利のモデル、NGDPのモデル、NGDPを物価と実質産出に分割するモデルなど)で観察するミルトン・フリードマンの嗜好を私も共有している。彼がもし今生きていればもっと優れた反LS-LM的な仕事をしただろう。ところで私は今、フリードマンとシュウォーツの書物、 Monetary History についての重要なエントリを書く準備をしている(訳者注: この未訳エントリ)のだが、そのエントリが今回の補足になるだろう。クルーグマンのF&S批判を批評することになるだろう。彼の批評には複雑な気持ちだが、クルーグマンが提起した問題を論じることは、私がどこから来たかを読者が理解するための大きな助けになるだろう。 […]

  66. Gravatar of Ray Lopez Ray Lopez
    18. November 2015 at 19:43

    For a devastating critique of Friedman and Schwartz’ Monetary History, by a sympathetic reader, see Christina Romer’s paper, found online:

    Title: NBER Macroeconomics Annual 1989, Author/Editor: Olivier Jean Blanchard and Stanley Fischer, editors, Publisher: MIT Press, ISBN: 0-262-02296-6 Conference Date: March 10-11, 1989 Publication Date: 1989 Chapter Title: Does Monetary Policy Matter? A New Test in the Spirit of Friedman and Schwartz Chapter Author: Christina D. Romer, David H. Romer

    Though Romer concludes the Fed does matter, it shows Friedman’s narrative is a house of sand.

Leave a Reply