Tuesday, December 29, 2015

Why a Bottle of Beaujolais is not the same as a Collateralized Debt Obligation (Updated May 2016)

I have updated this blogpost with a link to the new version of my paper. The new revised paper has the title of "Pricing Assets in an Economy with Two Types of People". 


Brad DeLong kindly tweeted a link to a working paper (updated to new version May 21st 2016) I wrote last year. Matt Yglesias asks Brad to explain the paper. Let me take a stab at that.

Every graduate student of economics learns, early in her career, that markets work well. The idea that ‘markets work well’ has a well defined meaning: allocating resources by buying and selling goods in free markets does at least as well as any other way of allocating them. Let me be more precise.

A society, to an economist, is a bunch of people and a bunch of goods. A good is something that people want. For example, a ticket to see the latest Star Wars movie is a good. A bottle of Beaujolais is a good: and so is a banana. I could go on. But the basic idea here is that everyone in society has preferences over different bundles of goods. I personally would prefer a bottle of Beaujolais and a banana to a trip to the movies: but you may rank things differently.

We need a way of thinking about abstract ways of allocating goods amongst all of the people in the society that is general enough to include the activity of buying and selling goods in markets as one possible allocation mechanism. Economists think about that idea by introducing an abstract individual that we call the social planner. The social planner is a non-existent benevolent dictator who knows the preferences of every person in society.

Imagine that we dump all of the goods that exist in a big pile in the middle of a very large imaginary room. Now let the social planner allocate them to people. For example, she might give everyone equal amounts of every good. That might sound like a good idea, but some people might not drink wine. They would prefer an extra loaf of bread to a bottle of Beaujolais. That idea suggests that some ways of allocating goods are better than others. If the social planner finds a way of allocating goods among people that can’t be improved on, without making someone in society worse off, we say that that allocation is Pareto Optimal.

There is not just one Pareto Optimal way of allocating goods. There are many. And some of them are very bad from a moral perspective. For example, if the social planner gives everything to one selfish person: that allocation is Pareto Optimal. Why? Because, in order to give food to starving children we need to take it away from the selfish person. And that, by assumption, makes him worse off. Pareto Optimality is a very weak concept.

But although Pareto Optimality a very weak concept it is an interesting concept because, if an allocation of goods is not Pareto Optimal, it is very bad indeed. Everybody in society, from the very richest to the very poorest person, could agree upon an intervention that would change things for the better.

Graduate students of economics learn, early in their careers, that markets allocations are Pareto Optimal. Markets may not produce outcomes that you or I judge to be morally acceptable. But they do not leave room for any obvious improvements that we could all agree upon. That idea, with a little reflection, seems to me to be obviously wrong. The ‘Global sunspots…’ paper provides one reason why.

Ok. That's the background. To understand my ‘Global sunspots paper…’ I need to go a little further by elaborating on the idea of a ‘good’.

Writing in 1959, the Nobel laureate, Gerard Debreu, suggested that the theory of markets that I just explained, is much more general than it at first seems. Debreu asked us to think of a good in a new way. A loaf of bread is not just a loaf of bread. It is indexed by location, date and state of nature. A loaf of bread in Paris on March 9th 2016 in the rain, is a different good from a loaf of bread in London England on July 20th if the sun is shining.

Financial economists took that idea and they ran with it. They argued that the financial markets provide people with ways of trading goods across space, time and states of nature. And because market allocations are Pareto Optimal, there is no possible intervention in the financial markets that can make us all better off. Government should get out of the way and let the magic of the market do what it does best.

According to financial economists, the financial instruments that are traded on Wall Street are unequivocally good. Equities, bonds, and exotic derivatives like collateralized debt obligations (CDOs)  all provide opportunities that connect people across time and space. To justify their confidence in the magic of the market, financial economists appeal to the mantra that they learn as graduate students: market allocations are Pareto Optimal. What could possibly be wrong with that?

The problem is a simple and obvious one. In order for markets to work well: we must all be able to take part in them. One of the most important functions of the financial markets is the ability to make bets on the future. If I think that the S&P 500 will crash next month, but you don’t, we have an opportunity to trade. And if I face different outcomes if the market crashes, or if it booms, I will try to insure myself by selling the market short in boom states and paying for that trade by buying the market in the crash state. It is my ability to make those trades, which ensures that market movements are not capricious. If a crash occurs, it occurs for a reason. And that reason is connected with the fundamentals of the economy. That, at last, is the theory. That theory is wrong.

That’s where sunspots come in. Writing in 1983, David Cass and Karl Shell picked up on a phrase that had been used much earlier by Stanley Jevons. Jevons thought that sunspots influence the business cycle through their effect on the weather. Cass and Shell meant something very different. They used the term as a spoof to mean capricious movements in market prices, and in the goods that we all consume, that are unrelated to fundamentals.

That leads me to the conclusion of my global sunspots paper. Because we cannot trade in financial markets that open before we are born, most of the movements in financial markets are Pareto inefficient. Financial markets go up. Financial markets go down. If you are lucky enough to enter the labor market in a boom; you will have a happy and prosperous future. If your first job occurs in the wake of a financial crisis: tough luck.

A bottle of Beaujolais is not the same as a Collateralized Debt Obligation. Why? Because trades in CDOs affect the welfare of the unborn: and the unborn are not around to trade in the CDO market. Stay tuned. I have much more to say about this idea in my forthcoming book ‘Prosperity for All’, to be published by Oxford University Press in 2016.


  1. Roger: I still don't get it. OK, the unborn cannot insure themselves against the risk of being born in a bad year. But why does that create sunspots? If we banned left-handed people from buying home insurance, we would not have a Pareto Optimal allocation of resources, but I don't see how it would create multiple equilibria.

  2. Nick
    The existence of sunspots and the existence of multiple equilibria are separate questions. If there were perfect insurance, there would still be multiple equilibria. These equilibria, in the case I study in the paper, would all converge to the same steady state. The effects of different initial conditions would soon work themselves out and, however the world started, we would observe the same steady state.

    In the absence of complete insurance, it is as if there is a new initial condition every period. Sunspots are randomizations across different initial conditions and each new shock is associated with a persistent path that converges back to the same steady state.

  3. Well done as explanation! De Vroey take note.

  4. Did you ever bother to look at Ole Peters' work? I brought him to your attention.

  5. There's something unsatisfying about this analysis as it chases the analysis down the rabbit hole to detect that there's a MISSING UNSTATED ASSUMPTION: SHOCK!

    Meanwhile the whole edifice is so dripping in assumptions - most obviously the idea of zero transactions costs - that one wonders why it's been taken as far as it has.

    This isn't to suggest that Debreu wasn't engaging in worthwhile theoretical work, but the idea that that then founds a whole dominant way of thinking to the exclusion of alternatives is scientism rampant. That's the sad state of our discipline.

    1. Don Ross, Economics and allegations of scientism
      (forthcoming in M. Pigliucci & M. Boudry, eds., Science Unlimited? University of Chicago Press)

    2. Interestingly, Debreu, along with most contemporary GE theorists, were (and still are) strongly opposed to the idea of continuous market clearing. Ken Arrow and Frank Hahn both take (took) the view that markets are almost always in disequilibrium. It was Lucas, and later Ed Prescott, who developed the modern macro concept that markets are in equilibrium at every point in time.

      The equilibrium viewpoint is not just a part of the new-classical macro that was promoted by economists at Minnesota, Chicago and Rochester, it also permeates all of the New-Keynesian models constructed before the recent crisis. In those models, temporary monopoly power allows prices to be away from their long-run steady state values, but the demand is equal to the supply of every commodity, including labor.

  6. Roger,
    You have a worthwhile point to make. But why do you have to make it by tossing out phrases like this:
    "Graduate students of economics learn, early in their careers, that markets allocations are Pareto Optimal."
    You and I (and any professional economist) know what you mean by this. We learn that competitive equilibria may possess desirable properties under some stringent conditions. These conditions, we were taught, almost certainly do not hold in reality.
    The way you write makes it seem like we've all been brainwashed to believe in the unqualified virtue of Minkowski's separating hyperplanes. Good marketing technique maybe. Is that what you're after? If so, then I think it stinks!

    1. The generalization sounds plausible to me. Big name economists and economists in general did not put up a very big fight against the financialization of the economy and the global system these past decades. Quite the contrary they pushed it while the lead economic policy-maker and regulator Alan Greenspan denied bubbles could exist.

      Then came the epic housing bubble and mother of all market failures as the financial system would have collapsed had the taxpayer not bailed it out.

      As Bill Clinton had said, "the era of Big Government is over." The era of markets was here and we see the results.

    2. What are you saying? That economists favor these policies solely because of the first welfare theorem?

      You see what you are doing here, Roger? It is less than useful.

    3. David
      I am glad that you find my ideas worthwhile.

      It is naive to think that the average graduate student, or indeed, the average practicing economist, views the first welfare theorem as a benchmark that almost never holds in reality. Graduate students are taught to think of the efficiency of markets as the normal case. The Coase theorem is just one extreme of that point of view.

      Are economists brainwashed? Acclimatized into a culture is perhaps a better way of phrasing it.

      Do I see what I'm doing? Well, if you mean my attempt to explain some fairly complex ideas in ways that the average person can understand then, yes, I do.

      And as an aside: I firmly believe that, most of the time, markets are the right way to allocate commodities. I also believe that many macroeconomists have failed to pay enough attention to the exceptions. In my view, there are two. One is the Pareto inefficiencies of asset markets that I drew attention to in this post. The other, is the Pareto inefficient state of labor markets that I have drawn attention to in other work. In my view, Macroeconomics is, or should be, about little else.

    4. Peter
      While I agree with much of what you say, there is also much to be said in favor of the era of markets. I believe that Greenspan was aware that there was a market bubble but, in his view, it was easier to clean up after the bubble burst than to 'take away the bunch bowl in the middle of the party'. That approach worked in the 1990s: it failed in 2008 because the interest rate had hit the zero bound.

      The moral of these episodes is not to give up on markets. It is to design institutions in which markets can function more effectively. Like Keynes, I believe that we must finds ways to rescue capitalism from itself. The alternative, as tried in Mao' China or Kruschev's Soviet Union, is not, to me, an attractive alternative.

  7. «This isn't to suggest that Debreu wasn't engaging in worthwhile theoretical work»

    Then let me "suggest" that Arrow-Debreu(-Lucas) models are completely worthless for the study of economics at the very least for lacking microfoundations (that is, they are wholly inconsistent both mathematically and semantically with the neoclassical microfoundations they claim to have), thus according the Lucas critique itself. :-)

    For example, and without even considering R Farmer's clever attempt to show that intertemporal "tatonnement" is a laughable idiocy, because market participants have finite lives, the Cambridges Capital controversy showed that any neoclassical model with more than *one* good in it (which must work as both consumption and investment good) cannot exclude multiple equilibria and path dependency and a number of other issues.

    «scientism rampant. That's the sad state of our discipline.»

    As to "scientism rampant", intertemporal "tatonnement" that clears infinite markets each with infinite sellers and buyers who are all however impersonated by the same immortal agent trading that same one type of good across all of them is "the state of the art", and R Farmer is not the first to try to show how comically stupid (or clever...) some details of that "state of the art" are, but as the Cambridges Capital Controversy shows, that model is necessary to "support" JB Clark's "three parables" (of which one is the central truthiness of Economics, that the "distribution of income" of that one immortal agent is solely determined by the agent's productivity) which are just too important for the purpose of propaganda.

    "Economics" is the Church Of JB Clark, Scientist (hopefully no offence to the members of that other similarly named church).

  8. I am a bit disappointed by this "reveal" of the intent behind the sunspots paper, because it boils down as per my previous comment to a weak criticism of the assumption of immortality of the single agent that impersonates infinite sellers and buyers across infinite markets across eternity trading a single good with itself.

    It is a weak criticism also because much better ones can be done, and also because it can be easily countered with a "narrative": a person's parent can trade on her behalf before she is born, and in fact until her age of majority.

    I had guessed wrong in a comment in B deLong's blog that it was a demonstration of the potential usefulness of instrumental "variables" in empirical models (as Jevons suggested), also for example P Sylos-Labini's (and later A Alesina's) "political variable".

    BTW in my previous comment I mentioned JB Clark's "three parables", and a nice summary involving the Cambridges Capital Controversy and them is here:


  9. Both Debreu and this, leave out primate politics that values certain items for status alone, (e.g. Gucci handbags for $1250, 1947 Cheval-Blanc at $300,000, etc.) and the reflexivity of the market that bases its buy and sell decisions to a great degree on the last decisions. Combine that with the reality of poor real information that means many trades are mostly blind.

  10. I like the basic point, that bargaining between generations is pretty difficult, not only because future generations are not around to express their preferences, but because they have no significant way of bribing us to change our behavior. I would say they have no way at all, if I believed that we are all self-interested: but since we base at least some of our behavior on their future welfare, one thing they could do is relieve us of some of that burden, or at least convince us to redistribute it from one effort to another. But they can’t ship the product of their future factories back through time, and they can’t ship the money produced by their future Fed back through time, to pay us to behave some different way---to invest more in infrastructure that they would use, for example. They depend on our good will for that.

    And they can’t pay us the value of the marginal product of our technical progress. What would we owe Newton or Liebniz, for example, for providing us with mathematical tools that form at least one of the foundations our modern prosperity? How can we pay them? We can’t. The world is governed by “the small and arrogant oligarchy of those who merely happen to be walking about” at any given point in history (Chesterton), and the unborn will make do with whatever we choose to leave to them, whether it’s a thriving economy filled with brilliant technology and abundant infrastructure, or a world in ruins from global warming. They can rail against us if they choose, but in fact they have no alternative but to simply accept the result of the choices we make.

    I guess the only other point I want to make is pretty obvious, but should be stated anyway. It is that the explanation above attributes far too high a value to mere Pareto optimality, at least in one phrase: “if an allocation of goods is not Pareto Optimal, it is very bad indeed.” Really? I don’t think so. I don’t think anyone thinks so. An equal allocation of each good to each person is unlikely to be Pareto optimal, of course, and it could be vastly improved by market trading. But it’s not necessarily bad: it’s just suboptimal. It’s the start of trade, and allows movement, through trade, toward a better allocation. But that quote closely follows the usual description of an allocation of everything to one “selfish person” as Pareto optimal, and describes that as “bad from a moral perspective”---it’s also bad from a market perspective, since no trading toward a better allocation is possible from that beginning. Between the two, I’d much rather start with the non-Pareto optimal equal allocation than with the Pareto optimal allocation of all to a single person. Pareto optimality, obviously, is not everything.

    1. I too stopped short at the statement, "if an allocation of goods is not Pareto Optimal, it is very bad indeed." It does not follow. Pareto "optimality" derives only from the sign, not the magnitude of change: 'no one is made worse off' by an intervention == the change in value is 0 or +1 for every individual. And a state is Pareto optimal or it is not. So there is no introducing comparative statements like "very bad indeed".
      Of course, I take this to be an argument against the usefulness of Pareto "whatever you want to call it other than optimality, which it most definitely is not". It isn't 'efficiency' either; efficiency is about output quantity/unit input.

    2. «I like the basic point, that bargaining between generations is pretty difficult, not only because future generations are not around to express their preferences, but because they have no significant way of bribing us to change our behavior»

      That's a nice summary of the "intergeneration compensation" issue, but my reading of R Farmer's post is that this is not the basic point he makes, but the *example* he makes to illustrate the point he makes. The basic point is another still.

      The point R Farmer makes above is that it is *essentially* not realistic to postulate a single agent that always existed and always will and trades with itself across eternity to achieve optimal intertemporal allocation of resources.

      Now this seems absolutely obvious, but that "*essentially*" I used above was used advisedly, and related to the real "basic point".

      It is about the argument of that swindler M Friedman that an Economics model does not need to be realistic, it only needs to produce realistic results (consider delusionary concepts like "the quantity of money" to see an application).

      Now if a model at the center of which there is an immortal "representative" agent trading with itself across all time *could* produce "realistic" results then according to Friedman's argument it would not be an *essentially* wrong model and the agent would be in some sense "representative".

      But R Former's argument is that it cannot produce realistic results, because it cannot clear the markets with respect to the trading of unborn people, because they cannot trade being unborn. He also assumes implicitly that in the model one can only make trades about the future, not the past.

      Which sort of brings the discussion onto multigenerational models, where there is a "represenative" agent per generation. But these models all suffer the fatal flaw of not "supporting" the central truthiness of Economics, that absent government interference unregulated (and infinite in number trading infinite goods across all time with infinite sellers and infinite buyers, but this special casing is rarely mentioned) markets deliver Pareto optimal productivity determined distribution of income (of a single immortal agent that knows without error the probability of all future events, another special case that rarely gets mentioned).

    3. Oops, another special-casing that is rarely mentioned and that indeed I forgot to mention myself so far, is that it is assumed in Arrow-Debreu(-Lucas) models that there is an initial allocation of resources that cannot be questioned, the so called "endowments".

      The "initial endowments" are what I think R Farmer alludes to in «Now let the social planner allocate them to people».

      Given initial endowments, whose distribution cannot be questioned, and intertemporal trading with usual special-case assumptions, Pareto-optimality of which R Farmer talks above a bit sideways has a very important role: in practice it means that nobody can be made worse off than their "initial endowments", because "democratically" everybody has a right of veto on any change in the "initial endowments" (and their fruits).

      That is Economics then becomes as intended a direct continuation of the theology of "All things bright and beautiful":

      The rich man in his castle
      The poor man at his gate
      God made them high and lowly
      And ordered their estate.

      Also sprache Arrow-Debreu(-Lucas). :-)

  11. Roger, I am not an economist. I have a question. Scott Sumner says there is a shortage of money, and certainly we can see that work out on main street where there is no turnover of money and business is slow. But I have seen that there is a shortage of bonds, and have written about it on Talkmarkets. So, if bonds are used as collateral for printing money, which the Fed says occurs, we are in a dilemma right? The Fed takes a lot of collateral that we may not think is so sound. Perhaps that is their way out to print more money. Could you share your opinion about all this hoarding, of bonds and money?

    1. Gary
      To say that bonds are 'collateral' for money is not quite right. A central bank is like King Midas. Everything Midas touched turned to gold. Everything the Fed buys turns to money.

      Private banks can become insolvent. Central banks cannot; at least, in the usual sense of that word.

      The liabilities of the Fed consist of accounts of private banks held at the Fed. These are not liabilities in the sense that you or I have liabilities. If I borrow money from you, the loan is a liability that I will, eventually, have to repay. If the Fed 'borrows' from the public by buying commercial paper, the liability it creates is a dollar bill that will never be repaid.

      The main way the Fed has altered the investment landscape, in my view, is by changing the duration and risk composition of assets held by the public. In that respect, money and treasuries were perfect substitutes when the interest rate on treasuries was effectively zero.

      If the assets purchased by the Fed turn out to be 'unsound', there will be a reduction in the profits of the Fed. These are ultimately returned to the taxpayer so, in that sense, there will be a fiscal consequence.

    2. I am sorry to take your time. I forgot to add that Bonds are collateral in swaps clearing houses. It is big demand. That seems to be ongoing and permanent and will keep rates low, which would be a problem. And as you say, if the Fed buys less pristine assets than long bonds, it could hurt the Fed and ultimately the national debt.


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