Sunday, December 20, 2015

Scott Sumner and Musical Chairs

Since 2009, Scott Sumner has been a big proponent of nominal GDP targeting. He sees nominal wages as slow to adjust and he has sketched a simple model, the musical chairs model, to explain why his policy should be adopted.

I am a new convert to these arguments. That is my loss. I had assumed, incorrectly, that Scott was proposing that central banks should simply adjust the coefficients on their interest rate policies, so called Taylor Rules, to raise the nominal interest rate when nominal GDP growth is above target and to lower it when nominal GDP growth is below target.I will refer to that variant of NGDP targeting, as growth rate targeting. An alternative, NGDP level targeting, would make these interest rate adjustments in response to deviations of nominal GDP from a target growth path. For an elaboration of that view, see, for example, the article by Evan Koenig, Vice President of the Dallas Fed.

Viewed in this light; NGDP targeting, of either variety,  is not a particularly new idea. Nor does it represent a departure from the body of New Keynesian economics that grew up in the decades since 1983, when Ed Prescott sought to banish money from macroeconomic models. Scott is saying much much more than that.

The  novel aspect of Scott’s proposal, and one that I endorse wholeheartedly, is the means that he advocates to achieve his goal. Scott proposes that central banks and/or national treasuries should set up markets for nominal GDP futures. Robert Shiller has made a similar suggestion. He proposes that national treasuries finance their borrowing by issuing securities that pay off a dividend that is proportional to nominal GDP. He calls these ‘Trills’; where a Trill is a claim to one trillionth of GDP in perpetuity.

In my own work, I have drawn attention to the remarkable stable connection between the real value of the stock market, and the unemployment rate. I interpret that connection through the lens of a causal theory in which expectations drive asset values, and asset values drive aggregate demand. I have suggested that central banks trade an exchange traded fund to stabilize real economic activity. Hold that thought as the word ‘real’ represents a significant point where Scott and I differ.

Trading Trills, trading GDP futures and trading an ETF, are all methods of targeting nominal wealth. I do not want to quibble over the exact method: and I readily concede that Trills or GDP futures have advantages over ETFs. The important insight here, is that wealth, or permanent income, drives aggregate demand and that expectations cause inefficient fluctuations in aggregate demand that can be stabilized through relatively straightforward interventions.

In the simplest macroeconomic models, GDP measured in wage units, is proportional to employment:

PY/W = bL

where P is ‘the’ price level, Y is real output, W is the money wage, L is employment and b is the inverse of labor’s share of income. Scott points out that money wages move slowly and that, as a  consequence, stabilizing PY will stabilize employment, and eventually, wages and prices. Scott bases his ideas on Samuelson’s neoclassical synthesis (see Pearce and Hoover). According to this idea, the economy is Keynesian in the short run, when prices and wages are sticky, and classical in the long-run.  

In Scott’s world, the economy homes in on the natural rate of unemployment just as surely as a heat-seeking missile converges to its target. Scott’s intellectual heritage is firmly monetarist. If Milton Friedman were alive today, one might imagine that Scott would find a supporter for his ideas.

My own heritage is different. I have sought to wed post-Keynesian insights with new-classical ideas by resuscitating the idea that the economy is not self-stabilizing. There are many equilibrium unemployment rates and any one of them may be an equilibrium. 

I wholeheartedly endorse Scott’s proposal for open market trades in GDP futures. And, like Robert Shiller, I would like to see the creation of a market for Trills. Unlike Scott, I do not endorse the proposal to stabilize either the level or the growth rate of nominal GDP. Trades in GDP futures should aim to stabilize the unemployment rate. And here is my biggest difference from Scott: trades in GDP futures should be seen as a complement to inflation targeting: not as a substitute.

As I explained in my 2013 book,  How the Economy Works, the economy is not a rocking horse, always returning to the same rest point, a metaphor that originates with Wicksell. It is a boat on the ocean with a broken rudder that requires active political interventions to steer it to a safe harbor.

Policy interventions have two dimensions: not one. Central banks should continue to set the overnight interest rate in an effort to target the inflation rate. They should adopt a second instrument, the purchase and sale of GDP futures, to target real economic activity. For more on this idea, stay tuned. I have forthcoming book in 2016  with Oxford Univesity Press with the working title: Prosperity for All. It will be available in mid 2016.


  1. Yep. I think the big difference between you and Scott is that Scott sees the LRAS curve as a thin vertical line and you see it as a very thick vertical line (correct?).

  2. I am surprised that you endorse nominal GDP targeting while simultaneousy asserting that asset values drive aggregate demand.

    The problem with nominal GDP targeting is that it ignores financial assets, the same error that the Taylor rule commits. A policy that keeps an eye on nominal GDP alone naturally advocates a loose monetary policy when nominal GDP growth is below trend. In so doing it sets the stage for asset price inflation. Later, when nominal GDP growth gets close to the trend value, the central bank tightens monetary policy, leading to a crash in one or more asset markets. This makes consumers cut consumption, leading to a recession.

    You cannot conduct monetary policy while ignoring money. Right now, as the graph in the link below shows, monetary growth (my measure) is getting close to zero. Next year, it will fall further and result in a financial crash. In 2017 a recession may be expected unless the Fed responds with another QE.

  3. Philip
    A Trill, as defined by Shiller, IS ( or would be if it existed ) a financial asset. Pegging the price of Trills through open market trades of Trills for treasuries should exert a powerful influence on the financial markets.

  4. Prof Farmer,
    Theories must be judged by the predictions they make. I predict a major financial crash next year. I hope you will read my book "Macroeconomics Redefined" after that.

  5. Thanks Roger, I don't think I've ever seen a blogger explain my ideas as accurately as you did here. A few questions:

    1. Using Nick LRAS metaphor, does the LRAS get thinner if the target growth rate (NGDP or inflation) is greater? Here I'm thinking of the criticism that the natural rate hypothesis does not hold at very low inflation rates, due to downward wage stickiness. Krugman discusses this issue.

    2. What about trying to shift the LRAS (if it is too low) with countercyclical employer side payroll taxes, or wage subsidies for low wage workers.

    Or do you think the problem I am missing (with NGDPLT) can be addressed with a different sort of monetary policy?

    BTW, I just came out with a book on the Great Depression (actually written about 10 years ago) which focuses heavily on the response of stock prices (the Dow) to policy oriented economic shocks during the 1930s. Not surprisingly, stocks were highly correlated with industrial production. In addition, stocks seem to have responded as expected to news related to the policy shocks that led to the Great Depression. So I have a lot of sympathy for your attempt to bring the stock market into macroeconomics. I get annoyed when people say policy X doesn't have any effect, when the stock market indicates it in fact does have an effect.

  6. Thanks for the endorsement Scott. Your originality is a breath of fresh air on the blogosphere.

    In response to your questions. Nick likes the thick AS curve metaphor. It's not my favorite way of characterizing my work. I don't believe that the existence of multiple steady state equilibria is contingent on whether or not we are at the ZLB.

    I have made two changes to standard Keynesian theory. First, I have provided a reconciliation of the 45 degree line as an AS curve in the Keynesian cross model, with rational behavior by households and firms. That reconciliation is based on relatively standard labor search theory but without Nash Bargaining to close the model. Second, I have dropped the Phillips curve. I close my models instead with the assumption that beliefs are fundamental. Because there are multiple steady state equilibria, beliefs can be both fundamental and rational in the sense of rational expectations.

    Your suggestion of shifting the AS curve with taxes and subsidies may help. I regard that as an empirical question. Policy changes 'push' against beliefs. The belief function is a mapping from observable variables to beliefs about future prices. I regard it as the product of learning behavior by many individuals with much the same status as preferences. Just as we think that preferences are learned as children, but slowly changing as adults, so are beliefs.

    I'm looking forward to reading your book on the Great Depression. By the way: I agree entirely with the interpretation of the General Theory you outline on your blog; at least of the first four or five chapters. I suspect we differ in our interpretation of the causes of the Depression. I'm not saying that policy in the 1920s and 1930s was perfect. But I ascribe a good deal of blame for the crash to the inability of asset markets to allocate capital efficiently over time. That, in my view, is due to a second market failure that is linked to incomplete participation. We cannot trade assets contingent on the state of the world we are born into.

    I have a book coming out next year that explains these ideas in more depth.

  7. So, BAC has said markets are manipulated by the Fed, by mispricing risk. That was clearly true in the housing bubble. So, how does this manipulation match up with your idea of a stable economy? I am not an economist. I would like to know why you guys think that everything happens as if there was a free market when bankers admit that is not the case!

  8. O/T: Roger, I'm curious what you think of this ranking of "allocation conditions" from strong to weak. (you and search/matching theory are mentioned there in the 1st sentence)

  9. Prof. Farmer,

    In your equation above is the labor share 1/b the exponent that would be constant in the traditional Solow production function (i.e. the real labor share)?

    And which FRED series does W = money wage correspond to? This one?


Note: Only a member of this blog may post a comment.