Believe those who are seeking the truth. Doubt those who find it. Andre Gide

Monday, April 15, 2013

This is not rocket science

I'd like to offer a few thoughts on this piece by Brad DeLong.

He seems to think that some people in the profession are confused about things like the natural rate of interest and its relation to the market rate. What's ailing the economy is so painfully obvious. Why are these dopes trying to make things harder than they are? This is not rocket science after all.

Well, I freely admit to being a bit confused. Let me explain why.

First, the "natural rate of interest" is a term that seems to mean different things to different people. According to DeLong, the natural rate of interest is the (real) interest rate consistent with "full employment." Well, that's all fine and good, except that he does not define what he means by "full employment." (I certainly hope he does not define it as the level of employment consistent with the natural rate of interest!)

In my view, the natural rate of interest and the full employment level of employment are theoretical objects. They may or may not exist in reality. Economists use these terms to help them interpret the world. Nobody knows for sure just by looking at the data where the economy sits in relation to the natural rate of interest or full employment (however these terms are defined theoretically).

One might of course try to estimate their values by sample averages in the data. But this approach is not without problems. The employment-population ratio in the U.S. shows secular variation. Moreover, it varies across different demographic groups. Take a look at this, for example. While there are more sophisticated ways of estimating these things, the whole exercise is predicated on the assumption that these objects actually exist. (They most certainly do, if only in the minds of some economists).

In any case, with that little bit out of the way, I'd like to see whether I can make sense of the various claims that DeLong makes in his column.

1. The current natural interest rate is much lower than it is normally--the natural rate is too low--and that is a problem.

One way to understand this statement is with the classic "loanable funds" market diagram. There is a supply for loanable funds S(r,Y), increasing in the real interest rate r, and increasing in the level of real income Y. The the demand for loanable funds, or investment demand, I(r) is a decreasing function of r. Let Y* denote the full-employment level of GDP. Then, in a closed economy, the natural rate of interest r* satisfies S(r*,Y*) = I(r*).

Now, imagine that investment demand collapses for some reason. Moreover, assume that the zero lower bound (ZLB) is not binding. What happens to the full-employment level of Y? I am not sure what DeLong is assuming here. I think he assumes that it remains unchanged at Y*, but that actual Y may move along some "short run" (sticky price) AS curve. So, are we currently in the short-run or long-run? He does not say. I say that after four years, the sticky price frictions are probably no longer relevant, especially since U.S. CPI is close to its long-run trend. If this is the case, then he must be assuming that Y* is currently at its pre-crisis level (or would be, if the ZLB was not a problem).

But what is the problem here? If the interest rate is free to move, then the economy achieves full employment and GDP remains at "potential" Y*. (This, despite the collapse in investment, which lowers the future capital stock, future real GDP, etc.? How can Y* remain unchanged years after lower-than-normal investment?) I am led to infer from this discussion that he views the real problem as stemming from the ZLB, which he tackles in his next statement.

2. The current market interest rate is higher than the natural rate--the market rate is too high--and that is a problem.

Alright, suppose that r0 cannot be attained because of the ZLB on nominal interest rates (despite the fact that yields on longer maturities are not at the ZLB, but anyway, ignore this too.) Then the market interest rate r1 is too high; i.e. r1  > r0. If this situation  persists, then according to this simple model, the level of output must decline to some Q satisfying S(r1,Q) = I(r1); where Q < Y*. This is also a problem; see also Krugman.

If this latter problem is indeed the problem, then there are simple fixes. First, the Fed could raise its inflation target. Inflation serves as a tax on saving because it lowers the real rate of interest (given the ZLB). At the same time, it stimulates spending. The fiscal authority could achieve the same result by just taxing saving directly. I wonder, however, how many people really believe this to be *the* problem.

*The* problem seems more related to whatever caused the collapse in investment demand in the first place. I'm sure that most economists, including DeLong, would agree with this. But the key question, in my view, is precisely what factors (institutional arrangements and exogenous shocks) caused the collapse and whether the desired policy response should be made contingent on specific causes. My experience in working with economic models is that the optimal policy response generally depends a great deal what one assumes about the underlying shock. I believe that this lesson likely holds for real economies too and I wish that economists would spend more time talking about this.

3. Increasing G--printing more Treasury bonds, selling them, and buying goods and services--(a) increases the supply of safe assets, (b) lowers the proper value of safe assets via supply and demand, thus (c ) raises the "natural" rate of interest, and (d) could fix our problems if the policy raises the natural rate of interest so much that it is no longer lower than the market rate of interest.

Listen folks, I am not against increasing government spending (see for example, here). But we have to be clear about what economists mean by "increasing G." What they mean is an increase in spending in any form. This is the proverbial "digging up holes and filling them up again" prescription. Or Krugman's famous "let's build a bunch of stuff and export it to Mars" prescription. Of course, they do not mean that this is the way resources should be used. The statement is simply that even if the resources were to be used in a wasteful manner, it would work. Well, forgive me for not being so comfortable with that proposition.

Now, suppose we instead take Steve Williamson's view (more correctly, my interpretation of his view) that the collapse in investment spending is related to the evaporation of private label liquidity products (like the MBS products that used to circulate in the repo market). In fact, Steve has an interesting paper where he lays out the details of a specific model here: Liquidity, Monetary Policy, and the Financial Crisis. Brad may want to read this himself, once he works his way through Wicksell.

Unfortunately, there is no simple way to summarize Williamson's model in a short blog post. As in reality, what happens depends on a specific set of circumstances. In one case he considers, the shock that afflicts the banking sector drives up the demand for relatively safe securities, which drives the real interest rate down. Hence, the statement that "the interest rate is too low" (relative to the rate that would prevail if the economy worked perfectly). There is no talk of "natural" rate of interest or "full employment" in Willamson's paper. In Descartes-like fashion, we could say that Willamson had no use for that language.

In his model (as in reality), U.S. Treasuries are substitutes for the now missing private collateral objects. In his "scarce interest-bearing asset" case, a one time open market operation by the Fed (purchase of UST) has a perverse effect: it diminishes the supply of good collateral and therefore raises its price; i.e., it lowers the real interest rate--but in a way that is bad for the economy because it further depresses investment (that is, the lower interest rate is a reflection of a downward shift in the investment demand schedule.) Williamson calls this the "illiquidity effect."

[Note: I am not arguing that this is in fact what is happening. It may or may not constitute one of many forces that are currently in operation. It is a property that emerges from a standard economic model with realistic frictions. It is standard practice to use theory to help guide us what to look for in the data. For example, it was theory that motivated modern day NIPA design, not the other way around!]

The policy prescription in Williamson's model is for the Treasury to expand its supply of debt, both to meet the hightened demand and to cover for the shortfall in (private) supply. What is the Treasury to do with its proceeds? In principle, the Treasury could buy up private securities (this is like a banking operation--an open market operation, but with Treasuries instead of money). Or taxes could be cut. Or, yes, an increase in G too (but whether and how much to do this should be based on standard cost-benefit calculations.)

Unlike the "this is not rocket science model" that DeLong likes to work with, there are many interesting cases that emerge in the Williamson model. Macroeconomists should go read it. His paper is by no means the last word on the subject. Indeed, it may turn out one day to be all wrong. But that is the nature of research. There are still a lot of things we do not understand about the way the macroeconomy works.  Well, for those of us who remain confused, in any case.


  1. "The fiscal authority could achieve the same result by just taxing saving directly."

    Eh? What does that mean?

    I don't think monetary policy works by tracking a 'natural rate', except in the long run. In the short run, the Fed is signaling what direction it wants to go. Most people figure that the Fed Funds rate is below its long term equilibrium, but so what? It's an overnight rate, not a long term rate. It can be below its long run equilibrium without the economy getting a 'jolt'.

    1. Max, What I meant to say is that, given enough instruments, there exists a fiscal policy that replicates what the monetary authority can achieve (in theory, at least).

    2. Ok, if you say so!

      Paradoxically, the Fed's interest rate floor may be artificially depressing interest rates. If the Fed got rid of the zero bound, a plausible outcome would be higher long term interest rates immediately, and a higher Fed Funds rate sooner. The function of a negative Fed Funds rate would not be to adjust to a negative 'natural rate', but rather to reassert the Fed's credibility.

  2. David,

    Williamson's point about current rates being too low is not really that different than a New Keynesian saying there is a short-run natural interest rate and a long-run natural interest rate. I think Williamson is making a distinction without a difference here. Brad and him are not that far apart, especially so on the policy implications. Willamson's contribution is to make this point in the context of a shortage of transaction assets.

    Where I disagree with Williamson is on this point:

    Treasuries are substitutes for the now missing private collateral objects. In his "scarce interest-bearing asset" case, a one time open market operation by the Fed (purchase of UST) has a perverse effect: it diminishes the supply of good collateral and therefore raises its price; i.e., it lowers the real interest rate--but in a way that is bad for the economy because it further depresses investment

    This assumes (1) the OMO is temporary and (2) there is no market segmentation. If the OMO is permanent and thus expected to raise future nominal spending, it could also raise expected real growth and in turn increase current demand for financial intermediation. This would lead to more privately produced safe assets. Josh and I have a paper we are finishing up that shows this theoretically and empirically. For now, you can see some of the preliminary empirical results in this post (look for VAR IRF links):

    On (2), I think Williamson is ignoring the glaring signs of market segmentation in spreads between risky assets and safe assets. This is where the ZLB becomes important, it leads to more market segmentation. See here for more:

    1. Thanks David.

      In the experiment I cited, I believe that the OMO is permanent. There is, however, no market segmentation (he has a newer paper out that has a "preferred habitat" flavor to it).

      I'm looking forward to reading your paper with Josh! Can you please send it to me?

  3. I agree with Brad on this one, and I think possibly your response proves his point.

    When all is said and done, I think both you and Brad agree that there are not enough jobs, and that higher levels of economic activity would be consistent with a higher number of jobs.

    I think the problem (and actually the solution) can easily and most usefully be described in very elementary vernacular that any non-economist (non-rocket scientist) can understand.

    The profession (all professions) can do well applying Occam's razor to verbiage and the various guff and jargon specialists always gravitate to; the simplest word is the best word.
    Jobs: more
    Economy: bigger

    just my two cents from a non-rocket scientist who would prefer not to have unemployed friends.

  4. The implications of a steeply negative natural rate for l(r) are interesting. I think it implies, "on a risk-adjusted basis, the marginal investment project has a negative NPV". If this this is the case, I wonder whether it fits with the current observed peak in corporate margins. What would DeLong respond?

  5. It seems that basic arithmetic and fact checking are both rocket science to economists.

    Today, I am telling Obama to issue an executive order imposing a moratorium on economists making public or private statements until we have a proper licensing and regulatory scheme in place.

    Talk about the hockey sticks of all hockey sticks in data.

    Moreover, my point that 97% of economists are incompetent is now beyond dispute. Did anyone before now bother to read the paper, check the math, etc.?

    How Much Unemployment Was Caused by Reinhart and Rogoff's Arithmetic Mistake?

    1. According to Yglesias, Reinhardt & Rogoff's mistake was discovered within 24 hours of releasing their Excel data. So economists as a whole look relatively diligent in this incident.

    2. how long before anyone bothered to look?

    3. Unfortunately there is no licensing requirements that would keep stupidity like John D's from the internet. It seems however that common sense has prevented anyone from paying any attention to him.

  6. David

    I appreciate your blog, even though I disagree with it most of the time (forces me to sharpen up my own arguments). I do think that if you were an engineer you would never, you know, actually build anything. You would always be stuck at the start posing deep questions. Does the natural rate of oscillation of this system really exist? How can we be sure...? In the meantime we better not commit to anything, we'll just keep saying on the one hand A, on the other B, and shooting down anyone who dares to take a position.

  7. I read this comment on my as saying:

    (a) Brad is right, but
    (b) I don't want to admit it, so I will get all snarky, and
    (c) no, there actually isn't any significant value added in Williamson vis-a-vis Wicksell, but
    (d) LOOK! Halley's Comet!!!
    (e) You can raise the (safe) natural interest rate via printing bonds and cutting taxes or via asset swaps (or, indeed, loan guarantees) as well as via printing bonds and buying goods and services or raising expectations of inflation.

    Am I wrong?


    Brad DeLong

  8. Brad:

    On your point (e), I think that it is a logically valid statement that emerges as a proposition in a class of economic theories. So you are correct in making the statement, but you are incorrect as portraying it as "obvious" or as a "self-evident truth." The statement relies on a set of assumptions, that may or may not constitute reasonably good approximations for reality. Moreover, I believe it is very difficult to ascertain the empirical content of the statement because objects like the "natural rate of interest" are not directly observable.

    As to your point (c), I think that Wicksell was a great economist, but I wouldn't go so far as to suggest that he answered all outstanding macroeconomic questions to the point that no further research is needed. Williamson's paper should be viewed as just one piece in a process of ongoing research. Whether this program ultimately adds "significant" value remains to be seen.

    So, my main objection with most of what you say (in your blog, not your body of research) is not that I think you are necessarily "wrong;" rather, it is the way you portray your knowledge as "unquestioned God-given truth." I think you might take this approach because it works in promoting a particular political view, which is fine. But that doesn't mean I have to like it when academics do this. And if I feel the need to call someone out on it, I will do it.

    Thanks for your comment.

  9. make $19,659.60 per month
    don't miss this opportunity
    join now and get your free mastercard

    For more information, contact me :

    Website :
    E-Mail :
    Skype : makemoneyleaders