On the Paradox of Thrift and Liquidity Traps

The paradox of thrift is generating a considerable amount of attention this week. I’ll give a brief summary of the debate and then my thoughts.

This was started by our friend David Beckworth (inspired by Martin Wolf):

He is alluding here to the Paradox of Thrift, the idea that if everyone tries to save–which makes sense individually–during a recession, then aggregate spending will fall.

David then goes on to note that:

Let me explain why the Paradox of Thrift is really just an excess demand for money problem. First, individual households can save three ways: (1) by cutting back on consumer spending and hoarding money, (2) by spending income on stocks, bonds, or real estate and (3) by paying down debt. In the first case, all households attempt to increase their holdings of money by cutting back on expenditures. However, if there is a fixed amount of money this will create an excess demand for it and a painful adjustment process will occur. If , on the other hand, the Fed adjusts the money supply to match the increased money demand then the painful adjustment is avoided and monetary equilibrium is maintained.

Nick Rowe chimes in with this in the comments:

Yes! There is no paradox of thrift. There is a paradox of hoarding the medium of exchange. That’s because there are two ways to buy more money: sell more other things; buy less other things. One of those two options is always open to the individual, but not to everyone.

The *only* case where Say’s Law is wrong is when there is an excess demand (or supply) of money, the medium of exchange.

Brad DeLong has taken umbrage with these statements:

The hole in David’s argument is, I think, where he says “the Fed adjusts the money supply” without saying how. Suppose that we have a situation–like we have today–where people are trying to cut back on their expenditure on currently-produced goods and services in order to build up their stocks of safe assets: places where they can park their wealth and be confident it will not melt away when their back is turned. They switch spending away from currently-produced goods and services and try to build up their stocks of safe assets–extremely senior and well-collateralized private bonds, government securities, and liquid cash money. Now suppose that the Federal Reserve increases the money supply by buying government securities for cash. It has altered the supply of money, yes. But it interest rates are already very low on short-term government paper–if the value of money comes not from its liquidity but from its safety–then households and businesses will still feel themselves short of safe assets and still cut back on their spending on currently-produced goods and services and the expansion of the money supply will have no effect on anything. The rise in the money stock will be offset by a fall in velocity. The transactions-fueling balances of the economy will not change because the extra money created by the Federal Reserve will be sopped up by an additional precautionary demand for money induced by the fall in the stock of the other safe assets that households and businesses wanted to hold.

In addition, we also get perspective on the framework that DeLong is operating within:

I am still frustrated that all of this seems so clear to me and is to opaque to so many other smart people. Personally, I blame Olivier Blanchard for making us spend three weeks on Lloyd Metzler’s “Wealth, Saving, and the Rate of Interest” in my first year of graduate school…

Regardless of what anyone has said to this point in the debate, the key to understanding the differences between folks like David Beckworth, Nick Rowe, and Scott Sumner and those like Brad DeLong and Paul Krugman is encapsulated in a key assumption made by Lloyd Metzler (that persists in the New Keynesian framework).

In Metzler’s analysis, bonds and capital are assumed to be perfect substitutes. As a result, we can focus all of our attention on the interest rate. Monetarists and quasi-monetarists, however, do not accept the view that the sole effect of monetary policy is given by its effect on “the” interest rate. This was the primary source of criticism of Monetarists of Keynesian models. Brunner and Meltzer spent a great deal of time and effort emphasizing that exclusive focus on the interest rate could be misleading. In fact, if you read the papers in Jerome Stein’s Monetarism, it is clear that this is the central focus of the debate from the monetarist perspective — although not from the Keynesian perspective. What’s more, the paper in that volume by Tobin and Buiter even notes that a useful extension to their model would be to drop Metzler’s assumption.

Why does this matter? It matters for a variety of reasons, but most importantly, it matters for the concept of a liquidity trap as DeLong describes above. Brunner and Meltzer (1968) showed that when the demand for money and other assets are a function of an index of interest rates rather than just the short-term interest rate, one can rule out the existence of a liquidity trap. (So long as money and the other assets are not perfect substitutes, an index of interest rate is necessary for analysis.)

Thus, it is not that David and Nick haven’t read Metzler, it is that their framework implicitly, if not explicitly, rejects Metzler’s assumption. Relative prices matter.

P.S. Much of the New Keynesian paradigm, at least that espoused by Woodford, highlights the importance not only of actual central bank policy, but expectations about central bank policy. For a discussion about expectations, see this post by Scott Sumner.

7 responses to “On the Paradox of Thrift and Liquidity Traps

  1. Josh: This was a helpful post. I read Meltzer years ago, but don’t remember all the details. And I always get him muddled with Metzler 😉

    Damn! I just got them muddled again!

  2. David Beckworth

    Ditto what Nick said. You are my go to guy for the history of monetary economic thought.

  3. The Brunner-Meltzer result, as you describe it, sounds pretty implausible to me. Why can’t all the interest rates in the index go to zero? I don’t see how you could possibly rule out the possibility that the Wicksellian natural rate for every bond in the index could be lower than the negative of the expected inflation rate.

    Also, there are statutory constraints on central bank activities which may prevent a central bank from buying what it needs to buy to increase the supply of money sufficiently without at the same time increasing the demand. Doesn’t the assumption that “bonds are the only lever through which monetary policy can affect capital markets” get you the same result as the assumption that “bonds and capital are perfect substitutes”?

    We can all agree that, if the Fed were able to purchase capital directly and willing to purchase a sufficient amount, it could absolutely negate the paradox of thrift. But that isn’t the world in which we actually live.

  4. Andy,

    There seems to be both areas of agreement and disagreement here.

    First, regarding the index of interest rates you wrote, “Why can’t all the interest rates in the index go to zero?”

    Well, I suppose that they could depending on how narrowly you define the index, but I don’t find that scenario very plausible. Short term rates are near zero, but longer term rates are not — and that is just looking at bonds. If we expand our spectrum of rates beyond bonds, this becomes even less plausible.

    This brings me to my larger point. A liquidity trap requires that money becomes a perfect substitute for all other financial assets. I think that you agree with me on this point (at least as a theoretical proposition) since you wrote, “if the Fed were able to purchase capital directly and willing to purchase a sufficient amount, it could absolutely negate the paradox of thrift.”

    Nonetheless, you argue that the Fed is constrained and thus cannot buy other assets. A liquidity trap is therefore possible because the Fed cannot take the action that is necessary. I am not convinced that this is correct. The Fed has purchased a large amount of non-traditional assets. What would preclude them from buying stocks, for example? How would this type of behavior differ from what they are already doing?

    Finally, although it deviates from the current discussion, I would be remiss if I didn’t say something about expectations. I don’t like to think about monetary policy in terms of interest rates — or at least “the” interest rate. I don’t think that interest rates are a useful signal for monetary policy. I similarly don’t think that monetary aggregates are, in and of themselves, useful indicators. The only truly useful indicator of policy is how much an economic variable deviates from target. In other words, it is very easy to assess whether policy is too loose (or too tight) in an economy that has, say, a nominal income target. It is not as easy when the central bank has no stated objective. (A nominal income target — if credible — would also signal that the central bank stands ready to increase the monetary base in the event of an increase in the demand for base money.)

  5. DeLong things that the crisis was caused by the flight to safety, however the price behaviour of very safe but less liquid assets tells us that the cause of the crisis was the flight to liquidity that was not fully accommodated by the Fed:


    However at this point liquidity gap between long term government bonds and reserves is very low, and traditional QE will have very low bang for the buck. I believe credit easing is the answer.

  6. Pingback: Welcome NRO Readers | The Everyday Economist

  7. Pingback: “Ben Volcker” and the monetary transmission mechanism « The Market Monetarist

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s