Tuesday, November 27, 2012

Explaining Stephen Williamson to the world (and himself)

Stephen Williamson catches a lot of flack on the net. Some is undeserved, some is deserved, but a big chunk is probably due to the fact that he and his fellow New Monetarists have a communications problem. People don't understand what they're up to. So here's my attempt to bring Steve down to earth and explain to the world the importance of the research being done by him and his colleagues. I'll go about this by adding a bit of historical context. After a quick tour of the history of monetary thought, readers will be able to see where in the greater scheme of things the New Monetarists fit. Now Steve doesn't know much about the history of economic thought - he thinks it's unimportant. So in a way, I'm explaining not just Steve to the world, but Steve to Steve.

One of the big problems in economics is how to deal with two significant but divergent streams of economic thought - monetary theory and real theory (ie. microeconomics). Put differently, there's "money" and then there's everything else - consumption, preferences, exchange, etc.

One strategy has been to let the two streams stay apart. The so-called "classical dichotomy" is one way to do this. Use the quantity theory to explain money. Use utility functions for the microeconomics. The two need never meet. Under the dichotomy, money is just a veil that has no effect on the underlying real economy.

Similiar to this is the idea of assuming a frictionless world with complete markets and perfect knowledge. The Walrasian, or Arrow Debreu model, represents this. In a Walrasian environment, an incredibly canny auctioneer sets the prices for all conceivable trades, dictates who is to deal with who, and clears these trades instantaneously at a centralized clearing house. It does this all before the market even opens. Given that this   initial setup makes barter hassle-free, there is no role left for mediums of exchange. Thus the entire discipline of monetary economics can be safely assumed away.

Now obviously this is unsatisfactory since we know that in the real world, yen and dollars and pounds do exist, and banking crisis can throw economies into a tailspin. Thus arises the second approach to dealing with the divergent streams - the effort to integrate monetary theory and microeconomic theory. This is what Steve and his merry band of new monetarists are working on.

Now Steve isn't the first to try and integrate real economics and monetary economics. The problem has a very long lineage. While the classical thinkers like Adam Smith and David Ricardo are typically accused of advocating the classical dichotomy, this is incorrect. By 1752 David Hume, for instance, was writing that in the short run, at least, changes in the money supply might affect the real economy. In debasing the coinage, a temporary stimulus might be given to real economic activity. This effect was due to the delay in full adjustment to the debasement, a delay in turn arising from people's inability to detect the change. For his part, Adam Smith created a sophisticated theory of real economic disturbances based on the inappropriate expansion in the supply of of bills of exchange, a popular form of credit used as a medium of exchange by England's merchant class. That's a bill of exchange below.

The problem with classical thinking wasn't that it maintained a so-called dichotomy, but that it linked money to the real economy through a stylized collection of macroeconomic stories. Furthermore, the macroeconomics, insofar as it was linked to an underlying set of principles, was using the wrong real theory of the microeconomy. To classical thinkers, value was determined by the amount of objective labour imbued in that object. It was only with the ideas of Carl Menger, Leon Walras, and William Stanley Jevons in the 1870s and 1880s that the neoclassical way of thinking was adopted in which value was determined subjectively and at the margin.

With the re-orientation of the idea of value, the classical methods for tethering the real and monetary economies had to be reconfigured. Luckily for the world, Carl Menger had not only established the basis for modern microeconomics, but he also described the conditions that would lead people to choose certain items to serve as liquid mediums of exchange. According to Menger, while all commodities are valued according to the marginal "bit" of utility provided by owning them, different commodities provide some "extra" utility if they are also more marketable than others. The peculiar marketability of a good is determined by that good's superior characteristics in the eye of the market - durability, transportability, fungibility, etc - as well as the features of the world that might encourage that good's marketability, including legal tender laws and custom.

Menger therefore single-handedly integrated monatery theory with microeconomics. That's why, when you read papers written by Steve, Randall Wright, Nobuhiro Kiyotaki, and other "New Monetarists", they always include a Menger quote or two.

Despite his contribution, it was perceived at the time that Menger had not entirely solved the problem. The money-like items in Menger's book are all commodities; gold, cattle, and sea shells. But the economy had long since become one driven by the exchange of credit, not commodities. Furthermore, state money, some of it seemingly fiat, had begun to emerge. Menger never touched on any of these issues.

The problem of integrating credit and fiat money into microeconomics fell to one of Menger's successors, Ludwig von Mises, in his 1913 book the Theory of Money and Credit. According to Mises, a fiat medium of exchange could have no marginal utility without an already existing purchasing power, but since a fiat token was by definition intrinsically worthless, it lacked the utility necessary to have any pre-existing purchasing power. This was a chicken and egg problem. Mises's solution was his famous (on the internet, at least) regression theorem. If people looked to the purchasing power of money yesterday to derive its marginal utility today, and people yesterday looked to the price from the day before, and so on to infinity, then it was possible to escape from the circularity of the problem. The "birth point" of fiat money occurred the day it was released from its commodity underpinnings. On that day, people relied on the purchasing power from the prior day, when money was still a useful commodity, to calculate the marginal utility of money today, when it was now a fiat token.

This technique of integrating fiat money into microeconomics was not popular with economists of the day. John Hicks derisively referred to Mises's conclusion as "money is the ghost-of-gold." Hicks went on to attempt his own integration of fiat money into microeconomics by seizing on John Maynard Keynes's liquidity preference. Keynes had introduced the idea that people might value money not solely for its saleability, but also for its speculative properties, namely by comparing it to the value of bonds. Hicks, and James Tobin who followed him, constructed a portfolio-choice theory of money whereby fiat tokens were just one of many assets that individuals held in their portfolios. The inevitable question was this: why hold any 0% yielding tokens at all when a high-yielding and riskless bond dominates the token's return? As long as people are risk averse, said Hicks and Tobin, then they will seek to build a collection of uncorrelated assets. Money, therefore, was valuable because it served the purposes of people in diversifying their retirement portfolios.

There was a major problem with the portfolio approach to money - while it managed to integrate money as a store-of-value into economics, it was incapable of doing the same with money's role as a medium-of-exchange. We know from experience that people don't just hold dollars because they prefer them to stocks and bonds on the margin, they hold them because dollars can buy stuff, stocks and bonds can't.

Don Patinkin found a simple way to integrate money as-medium-of-exchange into microeconomics: just include some item called M into the utility function. A utility function lists people's consumption preferences. By including money (real balances) into the utility function, individuals would be forced to rank the relative benefits of not only financial assets to money, but also stuff to money. Patinkin motivated money's inclusion into the utility function by noting that people who were caught without money to pay for something might be "financially embarrassed". Thus, the possession of money balances provided a form of insurance, and therefore utility.

Like Hicks, Patinkin didn't think that Mises's regression theorem was necessary to get fiat money to circulate in an economy. By virtue of having M in their utility function, people had a real demand for fiat money. Each individual's entire money demand schedule at all price-levels was to be submitted to the auctioneer prior to the market opening. The auctioneer would crunch the entire mass of demand schedules and determine the fiat money's market-clearing price level, then let the market open. Voila, a valued fiat money emerges. 

Don Patinkin
Patinkin incorporated the medium-of-exchange function of money into a Walrasian general equilibrium setting which, as I've already pointed out, had no place for money. Later critics pointed out the problem with this approach. Frank Hahn showed that it was possible for an economy to be in equilibrium even though the value of money was zero. This implied that the world could function fine without money, thank you very much. This conclusion was alarming since our basic intuition tells us that money makes things easier by salvaging us from a world of uncomfortable barter. Secondly, Robert Clower pointed out that any one of the many goods in Patinkin's utility function might be used as a medium of exchange. What, apart from arbitrary force, makes it necessary for the M in the utility function to be the medium of exchange? Now it might be possible to have the model dictate the use of M as the medium of exchange, but this constriction of the range of exchange choices is welfare-reducing. This again runs contrary to our intuition about money and barter, since we tend to assume that the use of money is welfare enhancing.

It is from this critique of Patinkin's money-in-the-utility-function approach that Steve's work descends. According to Steve, putting money into the utility function is a sloppy way of integrating monetary theory into microeconomics. We shouldn't be using models that force people to use some item called M given that M is unnecessary in a Walrasian environment.

Instead, guys like Steve are drilling down into the deeper reasons for why people hold money. They try to isolate these reasons by slowly throwing away the idealized assumptions of the Walrasian market. One of the Walrasian assumptions that has disappeared in New Monetarist models is centralized clearing. Now, people meet bilaterally and at random in order to trade. Nor does the Walrasian auctioneer call out prices. Rather, people negotiate prices among themselves. By deconstructing the Walrasian model (Steve calls this adding frictions), New Monetarists are trying to show why it is essential for people hold mediums-of-exchange. The Kiyotaki/Wright (1993) model, which took its inspiration from search frictions in labour economics, seems to be the wellspring from which all subsequent literature in the field is derived.

This approach is somewhat unique in modern economics. Those using the dominant New Keynesian framework are usually content to integrate monetary theory into microeconomic theory on a shallower level. New Keynesians proceed directly from Patinkin and either plunk money right into the utility function, are set some sort of constraint that forces individuals to always hold cash in advance of all trades. People use money at gunpoint (if at all).

Is the New Keynesian technique good enough for the job? Not if the centuries old effort to finally integrate monetary theory into microeconomics yields something totally unique. It seems to me the New Monetarist effort at integration is similar to Carl Menger's. In his book Principles of Economics (1871), Menger laboriously worked through all the frictions that might lead people to settle on certain items as mediums-of-exchange. But having done this, Menger never proceeded to inquire into the macroeconomic implications of his own unique way of integrating the two bodies of though. Nor did Austrian economists follow through on Menger, preferring instead to build their macro on a Wicksellian platform.  New Monetarist macroeconomics could be what Mengerian macroeconomics would have been. That's assuming, of course, that guys like Steve actually succeed in properly integrating money into micro, a subject I'm not entirely comfortable on commenting on since the literature is vast and complicated. Nevertheless, it'll be interesting to watch.

Sources: a mash of David Laidler, Perry Mehrling, George Selgin, Robert Lucas, Robert Dimand, and more


  1. JP: very good post!

    Minor quibble. Whether you like von Mises answer or not, it is an answer to a question that other models duck. Even if you build a demand for the medium of exchange into the model, there is always a second equilibrium in which the price of fiat M is zero, so it can't be used as a medium of exchange, so nobody demands it.

    Could we all just spontaneously start using some worthless object as medium of exchange? How do we pick one equilibrium vs the other? History is a very big Schelling focal point.

    1. Thanks, Nick.

      We've debated this before so you know I'm not a huge regression theory fan. But I do agree with you on your point that unlike others, Mises didn't try to duck the problem.

      In theory it seems like we could all spontaneously start using worthless objects as mediums of exchange, but given the fact that we rarely see this occur (bitcoin is the only example I know) I lean to the idea that we need some object to already have some fundamental value before we start to trade it. A liquidity premium gets built on top of this fundamental value... it's this portion of an item's value that might be influenced by Mises's backward expectations. A liquidity premium seems to me like a Schelling focal point.

    2. JP:
      But a liquidity premium creates a profit opportunity for short sellers. If silver gets a premium, short sellers just start selling IOU's promising the delivery of 1 oz of silver or an equivalent value of copper. Those IOU's start to be used as money and silver loses some of its premium. Then the short sellers deliver the oz (or equal value of copper) and take their profit.

    3. Hi Mike,

      In an ideal world I'm sure there would be no liquidity premium. But if we assume limitations on short-selling then the effect would arise. Have you ever tried sell a teapot short? There's not much of a market for that. You should give the Kreps-Harrison paper a lookover, I mentioned it here. They forbid short sales to help create a liquidity premium effect.

      You're also assuming that a silver IOU will necessarily compete with underlying silver as a medium of exchange. It seems to me that liquidity is sticky... it doesn't easily migrate to other similar objects. It takes a lot of work, or a lot of luck, to make your IOU liquid.

    4. Well, yes, in certain crazy markets it's hard to sell short. For example you mentioned that bitcoin has interest rates ranging from 25%-200%, so short selling is hard. But the market for silver is not crazy, and people have been issuing silver IOU's for probably as long as they've been trading silver. Historically, more trades have been effected using wooden tallies, clay tablets, etc., than using actual silver.

      While I'm at it, a word about bitcoin: It's value comes either its monetary usefulness or from its curiosity value (sort of like baseball cards, rare stamps, etc.). If it's monetary usefulness, then that's vulnerable to short selling. If it's curiosity value, then that's not vulnerable to short selling. So remember how you asked me if I thought that long term bitcoin=0 and I said yes? Well I take it back. Because of that short-selling argument I now think that bitcoin has value for the same reason as baseball cards, so I think that long term bitcoins will act like long term baseball cards.

      My grandparents never expected that a piece of paper with Honus Wagner's picture on it might one day be worth a fortune, just like I never expected that a certain arrangement of computer bits might someday be worth something.

    5. I don't think it's either/or.

      Bitcoin's value can be decomposed into that portion that arises from its value as a thing (say a curio) and a second portion that arises from value in monetary trade. Sounds like I'll have to have another bitcoin post soon. Let's see what I can cook up.

    6. I shouldn't have said either/or. I meant its value comes from some combination of curiosity/liquidity. But a liquidity premium would be vulnerable to short selling, so I think the liquidity premium must be very low, not enough to make short selling profitable.