Friday, November 15, 2013

Empirical Evidence for the Endogeneity of the Money Supply in the United States from 1971-2008

Dear Readers,

Let me start off by sincerely apologizing for my abrupt and now lengthy absence from the blogosphere. Although I have been absent from blogging, my interest in the endogenous money hypothesis and Modern Monetary Theory (MMT) continues to grow.

This semester I have been taking a directed readings course on those topics with another GMU PhD student, Paul Mueller. As part of the course, we are co-writing two papers that will hopefully be published in an academic journal. Our first paper, “Empirical Evidence for the Endogeneity of the Money Supply in the United States from 1971-2008,” is now sufficiently complete to make publicly available.

The unique aspects of our paper are the incorporation of Divisia monetary aggregates and a focus on broader measures of the money supply (i.e. M3 and M4). While the paper remains in draft status (so please do not cite this version), we would greatly appreciate comments and suggestions for improving the paper before a final draft is submitted for publication. The paper can be downloaded at SSRN (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2355178). Here is the abstract:

“This paper demonstrates that contrary to orthodox monetary theory, fluctuations in total commercial bank loans affect the quantity of various money aggregates, including the monetary base, but not vice versa. The Granger causality tests that we run on lagged quarterly data strongly suggest that changes in the money supply depend on private demand for commercial loans, not “exogenous” changes in the monetary base. Our findings strongly contradict the notion of a fixed “money multiplier.” The theory behind endogenous money is that banks issue new loans (credit) on demand and look for reserves later. The Federal Reserve must ultimately accommodate increases in demand for reserves from the banking sector to maintain an interest rate target and, more importantly, financial stability. These findings suggest that most economists need to revise their theories about monetary policy and credit expansion.”


Thank you in advance for taking the time to help and for continuing to follow my blog. 

(Note: In the past couple days it has come to my attention that using Granger-causality tests on the first differences of variables may lead to invalid results. Apparently a separate technique, created by Toda and Yamamato, generates more valid results using extra lags of the variables in levels as exogenous variables in the regression. The preliminary results using this method don’t materially alter the paper’s conclusion. Since each technique has been used in recently published articles, we would appreciate insight from other economists on which method (or both) to include our final version.) 

17 comments:

  1. Fantastic. Seems to me this area of economics could use much more empirical and quantitative support.

    ReplyDelete
  2. Very interesting stuff Josh and just what we need right now.

    http://diaryofarepublicanhater.blogspot.com/2013/11/cullen-roche-on-qe-joshua-wojinlower.html

    ReplyDelete
  3. Don't worry, Mark Sadowski will come up with some carefully cherry-picked data which he will transform in just the right way using some obscure technique to demonstrate that in fact everything you say is wrong and everything he says is right.

    ReplyDelete
    Replies
    1. You know it Anon but it's ok I'm expecting him.

      Delete
    2. Wojinlower's paper on endogenous money and the money supply a must read http://diaryofarepublicanhater.blogspot.com/2013/11/right-away-i-love-joshua-wojinlowers.html

      Delete
    3. Wasn't it Mark Sadowski who brought it to Joshua's attention that using Granger-causality tests on the first differences of variables may lead to invalid results? And from whom did he learn about the Toda and Yamamato technique?

      Delete
    4. ATR and Mike - Thanks for the support!

      Anon - You are certainly free to your opinion but so far Mark has provided thoughtful commentary and been willing to engage in productive dialogue.

      Mark - Yes you were on both accounts. If we end up going that direction I will definitely make your help publicly known. We left the T-Y technique out of the current paper to see whether or not people make the critique you note. We're also seeking guidance on that topic from GMU professors that are more familiar with econometrics.

      Delete
  4. "Our findings strongly contradict the notion of a fixed “money multiplier.”"

    That's good, because so does empirical reality:

    http://research.stlouisfed.org/fred2/graph/?graph_id=123223&category_id=0

    Every Econ 102 textbook I've ever seen teaches the money multiplier is a function of three variables. The currency ratio is always portrayed as the depositors' choice, the reserve ratio (above required, if any) is always the lenders' choice, and the total amount of currency and reserves (the monetary base) is the central bank's choice (even if supplied through the discount window), and all are dependent on the conduct of monetary policy by the central bank.

    "Yet, according to orthodox monetary theory, the massive expansion of the monetary base should have done far more than stabilize the price level. It should have led to a significant increase in the price level (Meltzer 2013)."

    Alan Meltzer's recent statements concerning contemporary US monetary policy severely contradict much of what he has written about the Great Depression. And I seriously doubt that the period in question (1971-2008) is relevant to our current circumstances.

    Also, from my brief skimming of your paper I see almost no discussion of the work monetary economists have done (e.g. Sims, Bernanke/Blinder, Christiano/Eichenbaum/Evans etc.) since Friedman and Schwartz published their monetary history half a century ago. In my opinion you are unfairly characterizing what most modern monetary economists actually believe.

    Nevertheless, good luck with getting the paper published. I'm certain that the Journal of Post Keynesian Economics will love it.

    ReplyDelete
    Replies
    1. Mark,

      Thank you for taking the time to skim the paper and provide feedback.

      I can't speculate on Econ 102 textbooks, but my personal experience at the graduate level (both from classes and conferences) suggests that the "money multiplier" is not only still taught but often presumed to hold within broader macro models. Given that, it is worth repeating the finding until a larger consensus is reached.

      Perhaps our paper is not clear enough on the distinction between exogeneity in the sense of direct control or statistics (or I misread your comment). We don't deny the central bank can directly control the quantity of base money, however we do disagree that base money is exogenous in the statistical sense. I completely agree that this finding is dependent on the central bank's conduct and does not accurately depict the current environment.

      You may be correct that we are unfairly characterizing modern monetary economists and I will consider expanding that discussion. I recognize that New Keynesians, for example, basically exclude money from their models and rely on a central bank that targets short-term interest rates. Our paper is not especially intended to undermine these recent views, but rather deal with the fact (mentioned earlier) that many economists still learn that the Fed normally affects NGDP/inflation/unemployment by exogenously adjusting the broad money supply through the monetary base. If we agree that is has not done so for most of its history, that's wonderful. In that case, maybe the used of Divisia monetary aggregates and broader money supply measures will provide further evidence.

      Thanks again for the comments and I appreciate any further comments you have now or in the future.

      Delete
    2. Joshua,
      I've had a chance to go through your results and compare them with my notes and my own results and so I have a question and some further comments which I hope you find helpful.

      1) When you say "commercial bank loans" are you referring to "commercial bank loans and leases" or some other measure? (You may want to make this more specific.)

      2) The only published paper with Granger causality results on the US monetary base, commercial bank loans and money supply that I am aware of is Palley (1994). Palley uses monthly data from January 1973 through June 1990 and looks at two measures of commercial bank loans: 1) commercial bank loans and leases and 2) commercial bank loans and securities. He finds that the monetary base Granger causes commercial bank loans (both measures). He then uses this fact to argue that the evidence supports Structural Endogeneity over Accomodationist Endogeneity. This finding also contradicts one of your results.

      When I tried to reproduce Palley's results, using monthly data over the same time period as Palley, like you I found that commercial bank loans (both measures) Granger cause the monetary base. There are a few possibilities why Palley's results are different from mine. First, he uses differenced data, whereas I did my tests in levels. Second, his data comes from CITIBASE, whereas mine comes from FRED. I've checked and there are in fact slight differences in the data. Finally, there is the remote possibility that I am mischaracterizing Palley's results because I'm going by an earlier draft of his paper, as I do not have a copy of the final published version.

      3) I find that there is bidirectional Granger causality between the monetary base and NGDP at the 1% significance level over the period 1959Q1 through 2013Q2. This is the only one of my personal results which contradicts yours. Two possible explanations for our different results are: 1) I did my tests in levels and not differences, and 2) We are using different time periods.

      And one minor editing point. In the conclusion you say:

      "On this count, empirical evidence is presently limited to a handful of studies covering select developed countries."

      One of the studies you list looks at Malaysia, which is usually characterized as a developing country.

      Thanks,
      Mark

      Delete
    3. Mark,
      Thank you for continuing to offer helpful comments.

      1) We are referring to commercial bank loans and leases (from FRED). I had questioned whether or not to make that more explicit, so I appreciate the input.

      2) One of the inspirations for the paper was simply the sparse amount of empirical evidence in the academic literature and lack of updates in nearly 20 years.

      3) I'll have to review my notes on the tests in levels I ran the other night, but I thought that result was unchanged. Even if the difference is merely due to time, it's an interesting point for future consideration.

      Lastly, good catch on the incorrect designation of Malaysia.

      Thanks again.

      Delete
  5. Joshua,

    This is a great initiative. I agree with your point 2 above that it's a good time for an update. I just want to add to Mark's comment that you may wish to check out Eichenbaum and Singleton (1986). There is an interesting result on page 133. The money supply appears to Granger cause real output during the Fed's "monetarist" experiment. The contention is that money may have been in some sense more exogenous during the period when the Fed switch to a nonborrowed reserves operating policy instrument. For this reason, it may be worth considering the results in light of changes in the Fed's policy regime. As hinted above by Mark, the December 2008 changes also alter system dynamics.

    http://www.nber.org/chapters/c4246.pdf

    Check out also the comment on p.144 by G. Mankiw (who I find has a very good grasp of methodological aspects, notwithstanding his politics and other policy preferences).

    ReplyDelete
    Replies
    1. Circuit,

      My apologies for not replying sooner. Thank you very much for the comments and suggested reading. I'm presently working on revising the paper and hopefully will have a much improved version available within the next few weeks.

      Thanks again.

      Delete
  6. Have to look at the paper.

    However IMO, the sharp dividing point 1971 is unnecessary. I will leave you will all sorts of contradictions.

    ReplyDelete
    Replies
    1. Sorry second line should read:

      "It will leave you will all sorts of contradictions."

      Delete
    2. Ramanan,

      Did you already download the revised version? Due to a potential conflict with the publishing journal we had to remove the paper from SSRN, at least temporarily.

      I appreciate your comment regarding the date choice. To clarify, data from the Center for Financial Stability only begins in 1967, so for the purposes of this study we could not have gone back much further. I recognize that the central bank was acting in an accommodating manner prior to the official close of the gold window, but thought using a more defined institutional setting might eliminate some questions.

      Anyways, your comments (and blog) are always appreciated.

      Delete
    3. I am a bit confused. Get the error message for the two links.

      Delete