Tuesday, January 28, 2014

What's a Central Bank Good For?

The New York Times this morning comes with an article on Narayana Kocherlakota. As journalism goes, this isn't bad, though I think it leaves a false impression of why Kocherlakota was appointed, or what the Minneapolis Fed is about as a research institution. The article makes a big deal of how Kocherlakota's views on monetary policy have supposedly changed since his appointment:
Mr. Kocherlakota’s outspoken advocacy for stronger action is particularly striking because he spent his first three years at the Minneapolis Fed, after his appointment in 2009, loudly arguing that the Fed should do less. It also represents a sharp break from the Minneapolis Fed’s longstanding association with economists who contend that monetary policy lacks the power to reduce unemployment.
That's correct, but it might lead you to think that everyone who works or has worked for the Minneapolis Fed thinks that monetary policy is irrelevant. If you analyzed all of the research coming out of the Minneapolis Fed for the last 40 years, I don't think you would find it any more representative of that view than what you see in the economics profession at large. Of course there were prominent people in Minneapolis - Prescott being the standout - who pushed that view. But anyone who has been around the Minneapolis Fed knows that the people in that place take nothing for granted. Certainly Prescott's view of the world is not taken as religion on the premises.

I disagree with this statement:
Mr. Kocherlakota’s shift has surprised and dismayed some people at the Minneapolis Fed and economists who supported his appointment because they expected him to be a principled opponent of the Fed’s stimulus campaign.
I don't think the potential policy positions of Kocherlakota were a concern of anyone who cared about who the replacement for Gary Stern would be. Within the institution, as in any research institution, people I think were concerned about having a leader in place who understood research and the role it could play in policy. Much work had gone into building a top notch research institution in Minneapolis, beginning in the early 1970s, and everyone who contributed to that effort seemed interested in furthering the tradition. Kocherlakota was chosen by the Board of Directors of the Minneapolis Fed, but I think the views of researchers in the institution played an important role in the choice. And I don't think there were litmus tests about policy views that went into that determination.

Here's another point of contention:
In October, Mr. Kocherlakota fired Patrick J. Kehoe, an economics professor at the University of Minnesota, from his position as a monetary policy adviser at the Minneapolis Fed. Another Fed adviser, Ellen R. McGrattan, was given a new position as a consultant. Both Professor Kehoe and Ms. McGrattan are proponents of the view that monetary policy has little power to lift an economy from recession.
That last sentence is incorrect. Factually, it's hard to nail Kehoe as a proponent of monetary policy neutrality, as I argued here. This also gives the impression that the disagreements with Kehoe and McGrattan were based on policy. As far as I can tell, that's wrong.

Here's something interesting. Prescott is quoted in the article as follows:
It is an established scientific fact that monetary policy has had virtually no effect on output and employment in the U.S. since the formation of the Fed.
So, my guess is that, when you read that, you're wondering what kind of lunatic this Prescott is. I've known Ed for long enough to know that he's a very deep thinker, and a serious scientist. He also loves to be provocative. So, what could Prescott be thinking when he says the Fed is irrelevant?

Let me play devil's advocate, and argue Prescott's point. The U.S. has had a central bank only since 1914, but somehow managed to surpass the U.K. - the home of the first central bank, established in the 1694 - in terms of per capita income sometime during the 19th century. But maybe you think that the Fed was important as an institution for eliminating or at least mitigating the effects of financial panics, which occurred repeatedly in the U.S. after the Civil War and before the Federal Reserve Act was passed? On this, Canada is an important counterexample. There was no Canadian central bank until 1935, but Canada managed to avoid the financial panics that occurred in the U.S. in the 19th century. Thus, it appears that central banking is not necessary for development, nor does it appear to be necessary to prevent panics.

What about the Great Depression? Friedman and Schwartz thought that the Fed's behavior during that period was important, at least in propagating the Great Depression. But in Kehoe and Prescott's book, Hal Cole and Lee Ohanian make the case that we can make sense of the Great Depression without thinking much about monetary factors, or banking, for that matter. Again, if we were to view central banking as the key to preventing banking crises, obviously the presence of the central bank didn't prevent 1/3 of U.S. banks from failing in 1933, and Canada sailed through the early part of the Great Depression with no central bank, and no bank failures.

In terms of aggregate fluctuations, it was once thought that a cursory look at the time series showed the benefits of stabilization policy - clearly post-WWII volatility in real GDP was much smaller than pre-Great Depression real GDP. But Christina Romer's work called that into question. If you do the measurement in a different way, it's not so clear. In more recent times, we all know about the Great Moderation, which was supposed to have been a great victory for monetary policy, but was followed by the immoderate recession of 2008-09.

If we look at what explains the differences in levels of per capita income across countries, I don't think anyone argues that central banking is an important factor. But there is plenty of work ascribing the dispersion in standards of living in the world to misallocation in factors of production, human capital, productivity at the firm level, political factors, and property rights. Ross Levine makes the case that financial factors are important for economic growth (poor financial arrangements could be part of what the factor misallocation comes from), but it's not clear what link might exist between central banking and general financial health.

So, I don't want to put words in Prescott's mouth, but that's the case he might make. On the other side of the argument, we have Friedman and Schwartz's monetary history, and a wealth of other empirical evidence aimed at showing that monetary policy matters. I think there are few macroeconomists who would argue that short-run nonneutralities of money are non-existent, though many of them would give you an argument about the wisdom of exploiting those nonneutralities.

Further, in spite of the fact that some countries functioned well in the past without central banking, in the complicated world of modern finance, it's possible these institutions are critical. Friedman was of course a big advocate of laissez-faire, but not in monetary arrangements - he thought the government should be the monopoly supplier of currency, and that the central bank had an important role to play. We may now consider some of his monetary policy prescriptions wrongheaded, but if Friedman were alive and were Prescott's colleague, he would be giving him a hard time over lunch.

In any case, we should not be dismissive of what Prescott is saying. We need to think carefully about what our central banks are good for. Perhaps central bankers, and the public, have been taking the role of monetary policy too seriously. One can certainly make a case, as I have, that the Fed has recently been inclined to confidently leap into unknown territory, with too little thought for the consequences.

55 comments:

  1. I wonder whether Canada is a good comparison point in this case, since the historical banking system there relied on provincial monopolies, e.g. Toronto-Dominion Bank in Ontario, Bank of Montreal in Quebec, ScotiaBank in Nova Scotia, etc.

    Is there a less regionalized example that we could use to make the same point that you make here?

    ReplyDelete
    Replies
    1. They're not provincial monopolies. All of the ones you mentioned branch nationally. In the small town I grew up in (in Ontario), there were 10,000 people, and there were branches of all the banks you mention.

      Delete
    2. I meant that they were monopolies during the 1st half of the 20th century. But it does look like I was wrong about that. Oops! I should have paid more attention when I took that tour of the Canadian Currency Museum!

      Delete
    3. Yes, national branching was also a feature of the pre-1935 era when the chartered banks could issue notes.

      Delete
    4. Of course some of us grew up in Alberta. The "fact" that the eastern banks just mentioned, were unwilling to finance the early oil industry was deeply ingrained. It was said to reflect their market power at the time. We were taught in school that the oil industry had to get US based funding at significant extra costs and difficulty. (I wonder if there is any academic documentation of this history.)

      Delete
    5. They taught you this in school? I guess the academic prior would be that the banks would have a hard time foregoing a profit opportunity. My brother, who lives in Calgary and looks for oil and gas, now believes this stuff too. He also denies that global warming is related to the activity of human beings.

      Delete
    6. It is not so crazy to imagine the bankers not funding the projects. They did not have much oil industry knowledge, so it might have seemed extra risky. They did have market power so local investments were lucrative. And banking was not a free entry industry in Canada in the 1900s as far as I can tell. Furthermore, there was a lot of failed drilling over an extended period, and the drilling location are far from Montreal and Toronto making monitoring a hassle.
      So it is logically possible.

      Delete
    7. 1. Monopoly power presumably affects all regions in the same way. Hard to see why that implies that the Bank of Montreal discriminates against Albertans.
      2. The whole idea behind branch banking is that you have people in the local branch who are experts on local industries. The Calgary branches of the Bank of Montreal should hire people who know about the oil industry, presumably. And the job of people in the industry is to convince the banks that they have good projects to fund. Maybe they weren't very good at that, or had to learn.

      Delete
  2. "The U.S. has had a central bank only since 1914, but somehow managed to surpass the U.K. - the home of the first central bank, established in the 1694"

    Nope, that title goes to the Swedish Riksbank which was established in 1668.

    ReplyDelete
    Replies
    1. I just googled "first central bank," and got this:

      https://www.firstcentral.net/

      That's the First Central Bank in Warrenburg Missouri. Apparently in Warrenburg they are confused about what "central bank" means.

      I stand corrected, though of course we could quarrel about whether the Bank of Amsterdam (early 17th century) counts.

      Delete
  3. Central banks keep inflation on low and predictable levels, can sometimes stabilize output in recessions and prevent speculative attacks upon a currency (we all know what happened in Europe).
    I fail to see how an institutional setting without central banks (I know to little about history so I have no idea how this is supposed to work, are their multiple currencies floating around, do individual banks issue their own IOUs ?) could improve upon any of these three purposes of central banks.

    ReplyDelete
    Replies
    1. I'm inclined to agree with you. On the other hand, there were these historical monetary systems that seemed to work pretty well. A successful one that I know about is Canada, pre-1935. There were government issued circulating notes - ones and twos - and private notes issued by the chartered banks in Canada, in denominations of five and above. The banks each redeemed notes of other banks, and there was a clearing system. The notes traded at par. There was some consolidation in this system from the late 19th century on, but no banking panics, and one bank failure between 1900 and 1935. So the question would be: why can't that work today, in the United States, for example?

      Delete
    2. Another primary goal of a Central Bank -- and indeed the original purpose that motivated their creation -- was financial stability. In particular, acting as a lender of last resort and supplying liquidity in times of crisis.

      Delete
  4. In any case, we should not be dismissive of what Prescott is saying.

    I think, actually, we kind of should.

    ReplyDelete
    Replies
    1. That's a dumb thing to say. Why exactly should we be dismissive of Prescott? Because he is very thoughtful? Because he is a first rate scientist? Because he is very well read? Or because of the general impact he has had on the profession?

      Delete
    2. Noah, I've been reading what you write, and you haven't given any sound reasons to back up "kind of should." For me, it's all useless blather. Write about what you know.

      Delete
    3. If Noah writes only about what he knows, then he'll have to completely stop writing. Which would be good for everyone.

      Delete
    4. Because he is very thoughtful? Because he is a first rate scientist? Because he is very well read?

      I think you forgot to capitalize "thoughtful" and "scientist"...

      Noah, I've been reading what you write, and you haven't given any sound reasons to back up "kind of should."

      True, but you did the same in 2010 when you wrote: "I doubt that there were any people in the room yesterday who took Ed seriously."

      http://newmonetarism.blogspot.com/2010/07/sed-report.html

      For me, it's all useless blather.

      I am sorry that my writings have not provided you with value in this instance. However, you must admit that the price is right.

      If Noah writes only about what he knows, then he'll have to completely stop writing. Which would be good for everyone.

      ZING!!

      Delete
    5. "I am sorry that my writings have not provided you with value in this instance."

      You seem to think they provide value in other instances. It is an established scientific fact that your writings provide zero value.

      Delete
    6. 'True, but you did the same in 2010 when you wrote: "I doubt that there were any people in the room yesterday who took Ed seriously."'

      Yes, I wrote that a few hours after I heard him talk. There's an important lesson there, which I explained to you elsewhere. Don't treat the man lightly. There's a lot going on in that mind that it's useful to know about. He's not some armchair guy dishing out opinions off the top of his head.

      Delete
    7. Not to stir up trouble, but here is a different hypothesis...

      http://orderstatistic.wordpress.com/2014/01/30/the-wisdom-of-laureates/comment-page-1/#comment-127

      Delete
    8. No thoughts on that Chris House quote?

      Steve?

      Anonymous butthurt troll?

      ...Bueller?

      Delete
    9. If anyone has a God complex it is Noah. He seems to think that his thoughts are valuable.

      Delete
    10. Aww, there you are, Anonymous butthurt troll. I missed your cute, vulnerable earnest-ness.

      Delete
    11. Had to look up "butthurt troll." Not always up on the language of younguns.

      I left some comments on the Chris House post. I agree with Chris that the possession of a Nobel Prize is no guarantee that the prizewinner's utterances on policy will be any good. I have known many extremely bright people, some of whom have Nobel prizes. Others don't even have a high school diploma. Some are very much cognizant of their limitations. Some of them are oblivious to the fact that their superior intellectual capacity applies to only a few specialized tasks, and that on many dimensions they are dumbasses. Sometimes conferring a Nobel prize on such a person just encourages more bad behavior.

      This doesn't describe Ed Prescott. As I said, sometimes he just likes making people angry. If you were to ask him why he said what he did to the guy from the NYT, he could give you a fully articulated, coherent economic argument for it. If he says it, it's going to be hard to understand. But get him to write it down. Then, you'll know exactly what he's saying. You can rebut it alright, but you'll have to work at it.

      Delete
    12. You wanna seriously go on pretending that Prescott's lie about (short-run) monetary neutrality "an established scientific fact" is just meant to be a provocation?

      This is equivalent to excusing another type of right-winger claiming that the Earth is flat and just a few thousand years old. If somebody defended such a liar he would be rightly called what he is: a fellow anti-scientific ideologue.

      Delete
    13. Instead of resorting to name-calling, tell us what the evidence is. Then we can discuss it.

      Delete
    14. "Aww, there you are, Anonymous butthurt troll. I missed your cute, vulnerable earnest-ness."

      Noah, let me know when you publish something. I'm not holding my breath.

      Delete
  5. How do you think Prescott could explain Paul Volker's big tightening around 1980 and the severe recession that followed almost immediately after? If monetary policy didn't cause that, then what did? What 'real' factor occurred right at that exact time to technology or whatever, that almost instantly caused the worst recession since the great Depression?

    ReplyDelete
    Replies
    1. In Chapter 13 of my macro book, I go through a version of an RBC model for undergraduates, and discuss the issues, and certainly that comes up.

      One thing you have to address is the first chart in the chapter. What you see there are deviations from trend in the Solow residual and real GDP, and you'll see that the first closely tracks the second. Further, there is a large negative Solow residual where the "Volcker" recession occurs. So, you have to address that. Prescott is confronting you with something. He says: look, I have a very simple theory, and it seems consistent with some features of observed times series. He doesn't say it's the last word. This is supposed to move science along. Your challenge is to come up with something better.

      Like you, I think monetary policy is important, but I'm not sure quite how important it is, or what the important mechanism(s) is (are) that make it work. So, for example, if I think that New Keynesian models are the way to go, I could show how a NK model can replicate the Volcker recession. For example, take an estimated NK model, plug in the time path for the fed funds rate over that period, and see what happens. I've never seen that done actually, and it's not obvious what you would get.

      So, nothing is obvious here. Prescott presents you with a challenge. He says: Be serious. Show me your model. Show me how it fits features of the data. Demonstrate to me what its quantitative properties are so that we can all evaluate it. That's why we take him seriously.

      Delete
    2. And see anonymous 5:49 am below. That's another way of thinking about this. Take Smets/Wouters, who argue that they have fit the data well by including a number of shocks, and ask how important the monetary shocks were over the sample. The answer is that they don't matter much.

      Delete
    3. Richard H. Serlin asks a very insightful question and I think it is important to examine it. [[Note: As I was looking for a quick Cut 'n Paste job, I ran across the important quote on http://socialdemocracy21stcentury.blogspot.com/2011/02/reaganomics-analysis.html . I have ABSOLUTELY nothing to do with this site. It merely supplies the quote! Good read, however...]]

      P.C Roberts stated in Supply Side Revolution:
      "“Secretary Regan, Undersecretary Sprinkel... requested from Federal Reserve Chairman Paul Volcker was a gradual 50 percent reduction in the growth rate of the money supply spread over six years...
      "Instead of evenly spreading the reduction in money growth over a six-year period, the Federal Reserve delivered 75 percent of the reduction in the first year. In November 1981 the Treasury staff reported that the main monetary measure, called M1-B, which consists of currency plus all kinds of checking accounts, had fallen over the previous six months, for only the second time since 1959–60. There was actually less money in the economy in October than there had been the previous April. The growth in M1-B over the year was meagre, roughly equivalent to figures for the recession years of 1970 to 1975. This was ferocious indeed and, with the tax cut delayed, recession inevitable (Roberts 1984: 116).

      To the question at hand, there's your explanation of Volcker, acknowledged by Roberts, one of the architects of the original Supply Side Theory. It is a straightforward acknowledgement the Monetarism has a Valid Domain. Where PCR falls down is in his belief that "Tax Cuts" provided the necessary shelter from the whims of the State.

      We are getting $1 T in Tax Cuts a year right now. These will become Tax Increases when the Bill comes due. If they are repaid (at all...) they will be repaid in inflated dollars and the State, entwined in every aspect of the economy, now has a vested interest in creating inflation, making it appear that it is not responsible for the fractured and non-productive economy it has caused.

      Wanniski - the Anti- Demand Supply Sider - has the Better of this argument. What are the impediments to people meeting in the Market Place to exchange goods? On this view, the EPA, for example, is a Tax on Goods and Services placed by the government and as such, is intimately bound to the Monetarism describing the economy. The Recession of the 1980s flushed out great swaths of non-productive economic (/sarc) activity.

      This lesson has been lost!

      CW

      Delete
    4. "If we look at what explains the differences in levels of per capita income across countries, I don't think anyone argues that central banking is an important factor."

      And this also may be relevant to anonymous 5:49:

      In any edition of "Stocks for the Long Run", by Wharton financial economist Jeremy Siegel, early in the book you'll see a graph of total real returns of U.S. stocks since 1802. In the current edition it's on page 82, and we see that the annualized return from 1802 to 2012 is 6.6%, so $1 grows to $707,997. The eighth wonder of the world at a very high long run return.

      One striking thing is that even the 1929 crash and the Great Depression doesn't look like much deviation from the long-term trend when you see this very long time series graph – In the long run, it didn't mean much. It was pretty much technology with good governance. But in the long run,... And, if you had to live through that, it was devastating for tens of millions of families.

      And, it's not just GDP, sadly the only thing considered in our measure of recessions and depressions. GDP can be fine and growing well, while there's high unemployment, including high long-term unemployment.

      In addition, technological advance and good governance are the big factors, and GDP and stocks grew well before central banking, but if good central banking could have added even another ½% to the long-term growth rate, that's a gigantic amount over decades and centuries. And it's hard to tell with long-term time series econometrics how much impact monetary policy had, and would have had, given that there's so much regime change, and other factor change, over time that's hard to quantify, changes in technology, changes in governance, in public education, etc.

      But case studies do appear strong that monetary policy can have powerful real effects, especially if applied powerfully. Obviously for the worst with hyperinflations; I don't think even Prescott would argue there's no real effects from that. But it seems for the better substantially too – as it looks like we saw very quickly right after Volker stopped the punishment.

      Finally, you may not be willing to really answer a question loke this in public, but you really don't think Prescott constantly says things like this to make look better his libertarian ideology, to constantly discourage among the profession, and the public, government action, for libertarian reasons, not public welfare, or maximizing total societal utility reasons?

      Delete
    5. You don't discount useful science because you're worried about the political views of the scientist.

      Delete
    6. M1 is endogenous, so it can't just "fall". Good God some of you people are seriously dim.

      Delete
    7. Professor Williamson says: "Prescott is confronting you with something. He says: look, I have a very simple theory, and it seems consistent with some features of observed times series. He doesn't say it's the last word."

      Prescott says: "It is an established scientific fact that monetary policy has had virtually no effect on output and employment in the U.S. since the formation of the Fed." How is that not saying that the issue is settled?

      Delete
  6. There is a very clear and uncontrovesial interpretation of what Prescott was saying: Milton Friedman was wrong in that, if you do an analysis of variance of which shocks explain most of U.S. GDP and inflation fluctuations, monetary shocks will explain a tiny fraction of either. This is not a radical, out of mainstream view, it is in fact what comes out of SVAR studies and of DSGE estimation. If you want to see for yourself, open Smets and Wouter's AER and look for the table with the analysis of variance.

    What this doesn't mean (and this is where Prescott was too vague and open to attack), is that monetary policy is impotent if a Central Banker decides to use it in an unexpected way, or that monetary policy rules don't matter, since they may affect how macroeconomic aggregates react to real shocks. But if you want to know the source of fluctuations, you are better off looking elsewhere.

    ReplyDelete
    Replies
    1. That's very important. Prescott chose his words carefully. He said "has had," not "can have."

      Delete
    2. Since when is impact of monetary policy reduced to effects of unpredictable monetary shocks? If central bank followed e.g. the Taylor rule perfectly, there would be no such shocks - but that (by itself) surely doesn't imply that monetary policy is irrelevant.

      The most straightforward interpretation of Prescott's remark is that time paths of output and employment would have been "virtually" unchanged had the Fed chosen a different path for policy. Maybe Prescott believes this - but in no way is it "an established scientific fact".

      Delete
    3. That's a big problem in extracting information from the data on how important monetary policy is. For example, in some NK models, if the policy rule is optimal, the equilibrium allocation will be the same as in the underlying RBC model. So, in the resulting time series you won't see any evidence that monetary policy matters.

      Delete
    4. I'm new in this literature...So, I'm wondering what's the latest research trying to tackle that problem?

      Delete
    5. You could start by reading Chapter 7 in the latest Handbook of Monetary Economics:

      http://www.sciencedirect.com/science/handbooks/15734498

      That's the state of the art in contemporary NK models, and what they say about policy. I certainly don't endorse all of that - it has plenty of problems - but that's a place to start.

      Delete
  7. The fact that a country can do fine without a central bank does NOT show that monetary policy is neutral. Central banks can cause a lot of problems! Just ask one of the economists at freebanking.org

    ReplyDelete
    Replies
    1. Some more thoughts on monetary policy. If it's neutral, who do asset markets jump in response to central bank announcements? If one wants to just throw out the EMH and say all these traders are reacting irrationally to meaningless noise, can you just bet against whatever movement occurs after an announcement and rake in the money? John Cochrane at least limited his rejection of the EMH to novel situations which markets may need learn about first to respond properly. Prescott is saying something far more radical.

      Delete
    2. "Central banks can cause a lot of problems!"

      Agreed. But how much central banking do you need? How many central bankers and economists does it take to get close enough to optimality? What's the value of all that agonizing that goes to thinking about QE, forward guidance, etc.?

      Delete
    3. "How much central banking" is an unusual way to think about it. The amount of effort that goes into running a CB seems rather trivial relative to its effect on the economy, although Prescott claims it's an established scientific fact that the latter is zero.

      Going back to the original question, Milton Friedman at one time claimed he'd like to have zero human beings running the CB, as a computer should just increase the money supply at a known rate. Scott Sumner wants instead of a computer to rely on a market of speculators on NGDP futures contracts. Evaluating those proposals based on how much "agonizing" goes into setting policy would be strange.

      Delete
    4. By "how much central banking," I mean how much resources should we use up on this problem. There are maybe 400 economists with PhDs at the Board, and another 400 at the regional Feds, with additional support staff and research assistants. What's the opportunity cost? Would we have been better off if Bernanke had spent 8 years at Princeton, and the Fed had hired some schmuck to do the job? Suppose we put Scott Sumner in charge and let him do NGDP targeting. Likely that would be worse than what we have now. But would the welfare cost be that large?

      Delete
    5. Comparing the QE-block to EMU-block, yes I'd say the welfare consequences of bad monetary policy are quite large. Going back to the Great Depression, the U.S seems to have been worse off than Canada as a result of its central bank, and I'm sure you've seen the graph of how different countries recovered after dropping the interwar gold standard. Zimbabwe monetary problems were undoubtedly entangled with its fiscal problems, but most agree it's better off after dollarization.

      Delete
    6. 1. Hard to say it's obvious that QE does anything at all.
      2. "... the U.S seems to have been worse off than Canada as a result of its central bank..." You could check this, but I think the decline in real GDP in the two countries was very similar in the Great Depression.
      3. As you say, hyperinflation is clearly bad, but you can lay the blame with fiscal policy. See Sargent's "four big inflations" paper.

      Delete
  8. Maybe what really matters is banking regulation. If banks' funding is sufficiently liquid and capital cushions deep enough, maybe the banking system would work just as well without a central bank?

    KP

    ReplyDelete
    Replies
    1. Yes. Note that the successful monetary systems that worked without central banking were indeed regulated. For example, entry was restricted for the Canadian banks in the system I mentioned. To get a bank charter, you had to go to Parliament.

      Delete
  9. For what it is worth, in my opinion the following paper goes further in explicitly estimating the proportion of variation in real output attributable to monetary policy than any other:

    What are the effects of monetary policy on output? Results from an agnostic identification procedure
    By Harald Uhlig
    (2005)

    Abstract:
    "This paper proposes to estimate the effects of monetary policy shocks by a new agnostic method, imposing sign restrictions on the impulse responses of prices, nonborrowed reserves and the federal funds rate in response to a monetary policy shock. No restrictions are imposed on the response of real GDP to answer the key question in the title. I find that ‘‘contractionary’’ monetary policy shocks have no clear effect on real GDP, even though prices move only gradually in response to a monetary policy shock. Neutrality of monetary policy shocks is not inconsistent with the data."

    http://www.benoitmojon.com/pdf/Uhlig%20JME%20sign%20agnostic%20identification.pdf

    Uhlig notes in a footnote on the first page his gratitude to Chris Sims and a long list of other economists (including Bernanke) for their helpful discussions and comments.

    On page 384 Uhlig notes:

    "‘‘Contractionary’’ monetary policy shocks have an ambiguous effect on real GDP. With 2/3 probability, a typical shock will move real GDP by up to +/- 0.2 percent, consistent with the conventional view, but also consistent with e.g. monetary neutrality. Indeed, the usual label ‘‘contractionary’’ may thus be misleading, if output is moved up. Monetary policy shocks account for probably less than 25% of the variance for the 1-year or more ahead forecast revision of real output, and may easily account for less than 2% at any given horizon."

    On page 392 he states:

    "I have followed the empirical approach in Bernanke and Mihov (1998a, b), who have used real GDP, the GDP deflator, a commodity price index, total reserves, nonborrowed reserves and the federal funds rate for the U.S. at monthly frequencies from January 1965 to December 1996.

    He explains his results in greater detail on page 398-399:

    "According to the median estimates, shown as the middle lines in this figure, monetary policy shocks account for 5 10% of the variations in real GDP at all horizons, for up to 20% of the long-horizon variations in prices and 15% of the variation in interest rates at the short horizon, falling off after that. Explaining just two or so percent of the real GDP variations at any horizon is within the 64% error band: it thus seems fairly likely, that monetary policy has practically no effect on real GDP. This may either be due to monetary policy shocks having little real effect, or due to a Federal Reserve Bank keeping a steady hand on the wheel, as argued by Cochrane (1994), Woodford (1994) or Bernanke (1996)."

    Examination of the graph on page 400 reveals that the peak proportion of variation in real GDP attributable to monetary policy shocks occurs at the 2 month horizon. It also reveals that 16% of the time the proportion of variation is greater than 40% at the 2 month horizon. In a 32 year period that would mean that would have occurred in approximately 61 months.

    ReplyDelete
    Replies
    1. I remember this paper. Harald is more careful than anyone when interpreting VAR results. Some caveats still apply of course. In work with VARs, you always have to ask whether you believe the identifying restrictions - even if they're "agnostic."

      Delete
    2. Especially the Bayesian VAR with sign restriction, which Uhlig used in his paper. To some extent, it seems to me that you can always manipulate your gibbs sampling to get your desirable impulse response.

      Delete
  10. Ed Prescott used to say that 70% of business cycle volatility was due to productivity shocks. Mankiw made two good points at a Fed conference on this claim in the 1990s:

    1. If monetary policy in the post war period was efficient, most remaining business cycle volatility must be due to non-monetary factors because monetary policy is not an independent source of instability.
    2. The real question is can the central bank cause a recession if it wanted too?

    ReplyDelete