Gold, not money, drove the interwar macroeconomy. (1932, pt. 2 of 5)

Austin Frakt has an interesting new post that praises a recent paper by Joshua Angrist and Jorn Steffen-Pischke.  They argue that the econometrics used in recent applied micro research (but not macro) has dramatically improved.  Frakt discusses their paper:

They argue that the “credibility revolution” experienced in empirical microeconomics since Leamer’s critique is due principally to a greater focus on research design not on sensitivity analysis.

A “research design” is a characterization of the logic that connects the data to the causal inferences the researcher asserts they support. It is essentially an argument as to why someone ought to believe the results.

I like this quotation.  As you know I have been very critical of traditional macroeconomic research, which I believe has not adequately addressed the identification problem.  In my work on the Great Depression I tend to focus on market responses to macro policy shocks.  I don’t claim this overcomes all identification problems, but I do believe it is much easier to make persuasive arguments about causality if you can observe policy announcements having a dramatic and immediate impact on asset prices.  Today’s excerpt from chapter 6 tries to use this approach, but I don’t have a smoking gun linking gold market shocks and the financial markets.  Nevertheless I thought there was a lot in interesting circumstantial evidence.

6.b  The Spring Open Market Purchases and the Run on the Dollar

There are several plausible alternatives to the hypothesis that Congressional turmoil sparked the dollar crisis.  Temin argued that the spring 1932 OMPs led to the gold outflow, noting that the outflow ended in July when the pace of open market purchases slowed dramatically. In contrast, Friedman and Schwartz argued that the gold bloc nations converted their dollar assets into gold for domestic political reasons unrelated to Fed policy.

Unfortunately, it is difficult to know exactly when the financial markets received information regarding shifts in Fed policy.  We do know that the day after the key April 12 meeting of the Open Market Policy Conference the NYT accurately predicted a sharp acceleration in the Fed’s weekly purchases of securities.  And neither that meeting, nor any of the other key meetings in early 1932, was accompanied by unusual stock price movements.  Yet T-bondprices did increase during this period, and the April 14th NYT suggested that the bond market rally over the previous five days was triggered by expectations an acceleration in the open market purchases of securities.  This picture is still further clouded by the fact that the NYT also suggested that sizable stock (and bond) market rallies on April 14 and 21 were attributable to greater than expected open market purchases.[1]

[1]  See the NYT (April 15, p. 27, and April 22, p. 28).  The Dow rose 3.4 percent on April 14th and 4.3 percent on April 21st, 1932.

Both the British devaluation of 1931 and the American devaluation of 1933 provide overwhelming evidence that, during the Great Depression, currency devaluations increased real (domestic) stock values.[1]  Short of actual devaluation, however, it is difficult to ascertain the impact of expansionary monetary policies that might increase the probability of devaluation, and even more difficult to decipher the market response to uncertainty over legislative proposals to force a more expansionary monetary policy.  The financial markets seemed to welcome expansionary policies, but only so long as they did not trigger a crisis of confidence which neutralized the impact:

           “The financial community is in general strongly opposed to the [Bonus bill] scheme, but it is a curiously sympathetic opposition in many quarters.  The professed object of proponents of the plan, which is to bring about an advance in the price level, is viewed widely as commendable, but the method by which it is sought to achieve this end is felt to be unsound.  It would result, in the opinion of most bankers, in great disturbance to confidence here and abroad and would, despite this sacrifice, fail of achieving its purpose.” (NYT, 4/12/32, p. 27, emphasis added.)

In addition to the Bonus bill, Wall Street also faced uncertainty over the Goldsborough bill, which would have instructed the Fed to raise prices back to the average level of the mid-1920s.  The May 7th CFC argued that this bill would force a devaluation of the dollar and suggested that the bill was contributing to the U.S. gold outflow.  The April 27th NYT(p.25) noted that a few Wall Street analysts supported the Goldsborough bill because they were afraid that without Congressional pressure the Fed would abandon its policy of purchasing government securities, but later suggested that the majority view was more complex:

           “Wall Street as a whole is strongly in sympathy with the idea behind the Goldsborough bill, which imposes a mandate upon the Treasury and the Federal Reserve System to lower the purchasing power of the dollar, but is opposed ever more strongly to the bill on practical grounds.”  (NYT, 5/3/32, p. 29.)

The financial press attributed a sharp decline in the dollar and T-bonds on May 3rd and 4th to a positive House vote on the Goldsborough bill (and European misunderstanding of the U.S. political system.)[2]

[1]  This evidence is not restricted to the time of the initial devaluation.  Real U.S. stock prices were highly correlated with the price of gold throughout the April 1933 – February 1934 dollar devaluation.

[2]  Foreign currencies rose sharply on May 3rd, and on May 4th several European currencies reached their highest levels since October 1931.  T-bonds fell 19/32 on May 4th and 1 13/32 on May 14th.  Although the bill passed the House by a (unexpectedly) overwhelming margin, the NYT (5/4/32, p. 29) gave the bill little chance of being enacted by the Senate.

On May 5th, the NYT(p. 29) referred to a “flight of funds from this market, set in motion by the action of the House of Representatives on Monday in passing the Goldsborough bill”.  Wall Street’s response, however, was ambiguous.  Although the May 4th NYT(p. 29) noted that the Goldsborough bill “may have depressed stocks” (the Dow declined 2.2 percent on May 3rd), the stock market rose on May 4th, despite an even sharper break in government bond prices, and an even weaker dollar.  The ambivalent reaction to the Goldsborough bill may have reflected investor support for a more expansionary monetary policy, but apprehension over the possibility of a dollar crisis.

The political climate in Washington D.C. changed abruptly on the evening of May 5th: “Hoover Sends Congress Sharp Message, Saying Its Inaction Disturbs The Nation; Demands Quick Balancing Of Budget”.[1]  The next day the Dow soared by 9.1 percent,[2] and the following day the NYT headlines reported “Tax Bill Completed” and “Bonus Bill Buried In House’s Red Tape”.  Although the budget fight would drag on for several more months, it was becoming apparent that even as the probability that Congress would approve the more radical legislation diminished, the stock market remained mired far below its mid-March levels.  When the dollar again declined on May 9th, the NYT suggested that:

          “There was the usual disposition in Wall Street to ascribe this movement solely to the vagaries of Congress.  That explanation possibly lost some force from the fact, which other markets than foreign exchange have appeared to reflect, that the Congressional situation has in the last few days taken a distinct turn for the better.  What apparently has been happening is that the foreign central banks, notably that of Holland, have been pursuing somewhat aggressively the policy of turning their foreign exchange reserves into gold.” (5/10/32, p. 29)

The NYTblamed inflation concerns partly on the fear that the Goldsborough bill would force the U.S. off the gold standard, but also on the fact that (due to gold outflows) the Fed had now been forced to invoke the provisions of Glass-Steagall which allowed eligible paper to be augmented with government securities.  They also noted that:

          “The contrast presented by these fears of inflation and the unremitting fall in the prices of stocks and commodities is one of the curious aspects of the current situation.  Obviously if there was any genuine belief in the likelihood of inflation, the attempt of capital to escape depreciation by conversion into some tangible form of wealth would be reflected at once in a demand for commodities and common stocks” (5/14/32, p. 23)

[1] NYT, 5/6/32, p. 1.

[2]  The 15 percent wage cut announced by U.S. Steel may have also contributed to the strong rally in stock prices.

Although the preceding observation has some validity if dollar depreciation is thought to be imminent, the opposite result can occur if an immediate devaluation is considered unlikely.  Under a fiat money system, fears of future inflation will reduce the demand for money, thereby causing an immediate increase in prices.  Under a gold standard, however, fears of future devaluation can be deflationary.  After a currency is devalued, nominal prices continue to be quoted in terms of the medium of exchange (money), and the nominal price of the medium of account (gold) will increase.  Thus, under a gold standard fears of devaluation will increase the demand for the medium of account, whereas under a fiat money system fears of future inflation will reduce the demand for the medium of account (i.e. currency.)

In principle, the problem of gold hoarding could have been neutralized by the expansionary effects of the Fed’s open market purchases.  But this effect was already being more than offset by contractionary policies in Europe.  The problem was not so much the U.S. gold outflow, (a fall in the U.S. gold ratio is the expected consequence of an expansionary policy), but rather the fact that the gold bloc central banks sterilized their gold inflows by sharply increasing their gold ratios. They weren’t playing by the rules of the game.

As it became apparent that the budget situation wasn’t the only factor affecting the dollar, investors began to view the gold outflows as an independent disturbance to the foreign exchange market.  Commenting on European views that the outflow was depressing the dollar, the NYT noted ironically that:

“It resembles the reasoning which attained much popularity in this country a year or more ago; which began by declaring that the stock market decline was the result of unfavorable trade [i.e. business] conditions, and ended by insisting still more vigorously that the trade situation was the direct result of the decline in stocks.”  (5/22/32, p. N7.)

If the gold outflow had been due solely to the actions of private agents, then it would be natural to consider the flows as simply an endogenous response to economic shocks.  But given the wide discretion practiced by the gold bloc central banks, it is not unreasonable to view the flows as being at least partially exogenous, and as being an independent factor contributing to the weakness of the dollar.

During the latter part of May the U.S. gold outflow accelerated, the dollar became very weak, and stock prices continued to decline.  The situation was worsened by political developments in Germany and France that increased tensions and reduced expectations for the upcoming Lausanne Conference.  On May 31st Hoover made another surprise appearance before the Senate to press for quick passage of the budget.  His dire warnings that the situation had deteriorated sharply in the previous few days, however, did not inspire the same confidence among investors as did his May 5th speech.[1]

Later we will see that Hoover’s penchant for over-dramatizing economic crises actually hurt the markets, and perhaps the economy, during the fall of 1932.  Nevertheless, Congress did begin to move more aggressively, and the Dow rose by 13.7 percent during the first four days of June, an increase that the June 5 NYTattributed to Congressional progress on the budget.  A June 15th NYTheadline reported that “France Withdraws Her Last Gold Here; Dollar Value Rises”, “Our Bankers Are Elated” and “Dollar Is Dominant Again”.  Yet the crisis was not quite over.  The markets faced one more banking panic, an additional month of Congressional budget battles, and completion of the Lausanne Conference, before a sustained recovery could begin in mid-July.

From June 15 to July 8 the Dow declined steadily and the June 26th NYT(p. F1) argued that the end of the gold outflow had put critics of the Fed OMPs in an “awkward position”.  The NYTdid concede that the slowdown in the rate at which the Fed was buying securities may have helped to stem the outflow, but argued that the Fed could now reach its excess reserve target with smaller weekly purchases.  Unfortunately, because the Fed’s excess reserve targeting procedure automatically synchronized these events, it is not possible to know whether cessation of the gold outflow slowed the open market purchases, or vice versa.  As less gold flowed out, the Fed didn’t need to purchase as many securities to hit their reserve target.  Fortunately, the recovery that began in the summer of 1932 provides independent evidence that gold hoarding was a problem for the markets, and for the broader economy.

Given the overwhelming importance of German developments for the U.S. stock market during 1931, it is surprising how little market reaction there was to the Lausanne Conference.  The NYT did report that stock prices

[1]  Stock prices declined immediately after the speech.

rallied after 2:00 PM on July 6th on rumors of a Lausanne agreement.  And the market opened strongly on July 8th, following the announcement of an agreement which essentially ended German reparations.  But despite the agreement, the Dow closed at its lowest level of the Depression on July 8th, and the July 9th NYT (p. 19) noted “Stocks moved up and down with a strange indifference to the overshadowing reparations agreement”.  The next day the NYT argued that Wall Street pessimism was due to the perception that a broader agreement for (allied) war debt rescheduling was also needed, but unlikely to occur because of opposition in Congress, and in the Democratic Party platform.

On July 11th the NYTreported that the Lausanne Agreement contained a secret “Gentlemen’s Agreement” that made the reparations cancellation conditional on the U.S. forgiving the allied war debts.  The U.S. immediately disavowed any involvement in this “understanding,” which dissipated some of the goodwill that might have resulted from the treaty.  Nevertheless, it is unlikely that anticipation of these problems could fully explain the stock market’s weakness during the first week of July, particularly in light of the fact that the price of YPBs soared by 45.6 percent between June 29 and July 9.

An alternative explanation for the weak stock market was the renewed banking difficulties in the Midwest.  During late June and early July, Midwestern banking problems centered in Chicago led to an upsurge in currency hoarding, and as early as June 25th the NYTwas suggesting that the Chicago bank failures were depressing stock prices.  Lastrapes and Selgin (1995) provided evidence that a “check tax” which took effect on June 21 may have substantially increased the currency/deposit ratio.  If so, then this tax could have contributed to the ongoing banking difficulties in the Midwest.  The banking problems may have also helped to delay the rally in the dollar until mid-July.

As late as July 11th, NYTheadlines continued to show little evidence that a dramatic reversal in the dollar’s position was imminent:  “Possibility of [Gold] Shipments to America Discussed, but Doubted”.  Although radical legislation was now considered unlikely, Congressional deliberations continued to overhang the market.  The July 12th NYT noted that a late rally in the stock market was due to rumors of Congressional adjournment on the following day, and argued that adjournment was necessary before there could be a return to confidence in the dollar.

During the final days of the 72nd Congress, deliberations continued on one last issue with important monetary implications.  Carter Glass had proposed legislation which would allow banks to substantially increase their issuance of national bank notes, as a way of forestalling even more radical legislation.  And in early June the Glass bill did displace the Goldsborough bill.  Although the Glass bill theoretically allowed the issue of over $1 billion in new currency, there were doubts that it would have much impact.  The actual increase in national bank notes over the next 12 months was slightly over $200 million (roughly 4 percent of the total currency stock.)

The July 17th issue of the NYT (p. F1) attributed a decline in the T-bond market to the (apparent) rejection of the Glass bill on the eve of Congressional adjournment.  The bill actually was adopted by Congress late in the evening on the 16th, and therefore the subsequent market reaction provides a useful confirmation of the NYT’s analytical skills.  On Monday, July 18th, the price of T-bonds rose 1 5/32, the largest daily movement during the entire month of July.  Four weeks later T-bonds prices fell 1 8/32 on Attorney General Mitchell’s ruling that national bank notes could only be issued for a three year period (ending in 1935.)

 6.c   Signs of Recovery

The period from July 8 to September 7 saw one of the largest stock market rallies during any two month period in American history.  The Dow nearly doubled, and other indices showed considerably larger gains on very heavy volume.  As late as July 19, however, the market had risen only slightly from its Depression lows.  And since by that date Congress had already adjourned, the Midwest banking panic had ended, and the press was already predicting the imminent termination of the Fed’s open market purchases, it is difficult to account for the subsequent boom in stock prices.

The analysis in the preceding section suggests that the dollar crisis contributed to the spring 1932 stock market crash.  If so, then a turnaround in the dollar’s position would be expected to boost stock prices.  A July 29th NYT headline reported that “Stocks Rise Again In Year’s Heaviest Trading”, and in addition to rising commodity prices, attributed the gain to a “spectacular rise of the dollar in terms of foreign currencies which reflected the further reinforcement of the gold position of the United States.”  Another story noted the “discovery that the United States is securely anchored to the gold standard” and then asserted that:

“Certainly the financial community does not relish the return of the bulk of the gold shipped from New York to Europe a few months ago, in view of the fact that many other nations need the metal far more than the United States, but at least there is a distinct feeling of satisfaction that there is a movement in this direction and that the ill-starred raids on the dollar are over.”  (7/29/32, p. 23)

The NYTmay have actually understated the gains from the gold inflow.  Most of the previous outflows had gone to nations with less need for the gold than the U.S., such as Britain, France, Holland, and Switzerland, rather than to nations in a precarious financial position, such as Germany.  And news that private gold hoarding ceased in July was even more unambiguously positive; as dishoarding pushed the world monetary gold stock up by $143 million in August, and another $131 million in September.

France was easily the largest gold hoarder during the Great Depression.  Her monetary gold stock rose almost continually from $711 million in late 1926, to over $3.2 billion at the end of June 1932 (nearly 30 percent of the world total.)  After June, French gold imports slowed to a trickle, and on August 5th the NYT (p. 21) cited the “new [French] situation” as one of the factors which contributed to the return flow of gold.  They also suggested these flows were leading to inflation expectations which were being reflected in soaring stock and commodity prices.  This supports the view of Johnson, and others, that French gold hoarding played a significant role in the Depression.  On August 11 the NYT reported that, in a significant policy shift, foreign central banks were now buying dollar assets with gold.  By this time, the Dow had already risen 68.3 percent from its July low.

During August, confidence in the international monetary system returned.  The August 19th NYT(p. 23) argued that Britain was moving inexorably toward resumption of gold payments “without asking the permission of the Keyneses and the Cassels.” The Dow reached a peak on September 7th, and the next day the NYT (p. 29) reported that the Dutch guilder and Swiss franc declined as confidence returned and investors pulled funds out of “neutral” countries.

Along with the rise in stock and commodity prices, industrial production rose 2.9 percent in August, 6.5 percent in September, and 3.5 percent in October.  It looked like the Depression was ending.  But then industrial production leveled off in November, and declined 2.6 percent in December.  By March 1933 output had almost fallen back to the lows of July.  Although it is unclear exactly what caused the recovery to abort, there is strong evidence that the looming presidential election had a dramatic impact on the stock market from mid-September onwards.



32 Responses to “Gold, not money, drove the interwar macroeconomy. (1932, pt. 2 of 5)”

  1. Gravatar of JimP JimP
    15. March 2010 at 07:59

    Another supporter – not that Obama gives a damn.

    He is obsessed with healthcare the same way other people are obsessed with Martians.

  2. Gravatar of Mike Sandifer Mike Sandifer
    15. March 2010 at 09:07

    There’s been a revolution in behavioral economics in recent years as well, especially with respect to neurophysiology. I recently found and blogged about this paper, which focuses on a neural model of utility:

    That economics is a biological science seems obvious to me, but a surprisingly large number of economists and physicians I speak with seem skeptical about this seeminly self-evident claim.

  3. Gravatar of Mike Sandifer Mike Sandifer
    15. March 2010 at 09:11


    Just out of curiosity, how many people should die and/or go bankrupt before health care reform becomes a central priority?

    I’m not a fan of the Obama approach, but I’m less of a fan of the human suffering that takes place sans a less than otimal one.

  4. Gravatar of Philo Philo
    15. March 2010 at 09:49

    Mike Sandifer,

    Your implicit claim, that anything the Congress is likely to do about health care is going to save a lot of people from death or bankruptcy, strikes me as a bad joke.

    And why whine about conditions in such a rich country as the U.S., when so much of the world is really poor?

  5. Gravatar of JimP JimP
    15. March 2010 at 09:55


    Health care is not a positive right – it is an economic good which costs. To me the issue is – how much will the Obama plan cost and what will it do and is there a better way.

    I think if the plan passes, which it looks like it might, it will be a disaster for us all.

    I think all those “cost savings” in the bill are fakes and a fraud. They have put off taxing Cadillac plans till the end of the universe – and the same will happen with all the other ones.

    Health care is no more of a central priority than food or rent – but this plan will force people – by straight government force – to choose health care over rent or food. What is just or fair about that?

    It forces young people – by straight force – to pay for the health care costs of old people. What is fair about that? Nothing.

  6. Gravatar of ssumner ssumner
    15. March 2010 at 10:35

    JimP, Thanks, I like to see talk about nominal targeting.

    Mike, The issue has always been how to measure utility. As far as I know, that problem has never really been solved. Indeed it isn’t even quite clear exactly what utility is.

    As far as health care is concerned, I also wish to see reform. But more in the direction of HSAs, not (public or private) health insurance.

  7. Gravatar of Mike Sandifer Mike Sandifer
    15. March 2010 at 11:21


    I suspect that only economists think utility is still a mystery. The mystery was laregly solved within psychology years ago.

    I think utility is calculated by multiplying the net required rate of intake of reinforcement, which would code for all resources needed to optimize inclusive fitness(reproductive success), by the ratio of a given reinforcer’s expected objective gain and the sum of all available net reinforcing options, minus the same for losses, temporally discounted. You could also call this “demand.”

    This is a reinterpretation of Hernnstein’s matching law(can google it), which has experimental support going back to at least the early 1960s.

    The aspects of expectations are accounted for with learning models, of which I prefer the stochastic variety.

    I realize you may have no time to read into this sort of thing, but the paper I link to above considers such an approach in detail, as does my new blog.

  8. Gravatar of Mike Sandifer Mike Sandifer
    15. March 2010 at 20:09

    Philo, if we decide to tighten our belts and donate unprecedented amouns of money to the developing world, in effect becoming a major engine of economic development in the world, fine. But, we aren’t going to do that. Even if we were, the least we can do for our people is give them the best health care we can afford.

    And yes, we can easily afford this. $100 billion-per-year in a $14 trillion+ economy is less than trivial. Even if the program will cost ten times that amount, I’m all for it. We can certainly cut defense spending, for example.

    JimP, show me a fair government and I’ll show you world peace.

  9. Gravatar of StatsGuy StatsGuy
    15. March 2010 at 21:23

    “The financial markets seemed to welcome expansionary policies, but only so long as they did not trigger a crisis of confidence which neutralized the impact:”

    “During the latter part of May the U.S. gold outflow accelerated, the dollar became very weak, and stock prices continued to decline.”

    These two points are linked. Weakening the dollar by permitting gold outflow (aka, by increasing net indebtedness of the country, or alternatively by decreasing the net creditor position – assuming we treat gold as real money and dollars as promise notes) is not stimulative. This is comparable to an event which triggers a run on the dollar by saying “everyone expects us to devalue, but not today”. The community arbitrages against this occurrence, yielding a run on the dollar. Stimulative action requires action that favorably modifies the net balance of wealth in favor debtors. In 1932, that meant cutting back on the value of promises (dollars) to pay in “real” wealth (aka, gold). Attempts to supply dollars while weakening the international balance sheet forced up interest rates due to an expectation of “default”, or future monetization.

    The WORST CASE SCENARIO, is creating an expectation of future default (or inflation/monetization) without actually monetizing anything. Everyone pays a rate premium, but without the stimulus. It’s a massive transfer of wealth in the wrong direction at the wrong time.

  10. Gravatar of Doc Merlin Doc Merlin
    15. March 2010 at 22:57

    “Temin argued that the spring 1932 OMPs led to the gold outflow, noting that the outflow ended in July when the pace of open market purchases slowed dramatically. In contrast, Friedman and Schwartz argued that the gold bloc nations converted their dollar assets into gold for domestic political reasons unrelated to Fed policy.”

    Hrm, are we seeing this exact same thing over again now, with the BRIC countries lowering their dollar portfolios and increasing their exposure to commodities and precious metals?
    In terms of data, the start of this recession seemed a lot more like the start of the 1920 recession to me than 1929 (with lots of commodity price spikes) , but the talk of dollar troubles with foreign central banks has me worried. What is your take on this, Scott.

  11. Gravatar of scott sumner scott sumner
    16. March 2010 at 06:00

    Mike, Just tell me in very simple terms what you think utility is maximizing. Is it happiness? Genetic fitness? Reproductive success? And then tell me why I should agree with whatever definition you choose. Suppose I say “utility is measured by how big someone’s stamp collection is.” Who’s to say I am wrong? I am sure I missed something in your answer. You seem to suggest utility is some objective scientific concept, not merely a vague and fuzzy definition.

    You are right, I don’t have time to read new books now, perhaps you could briefly tell me what you think people are maximizing. And is it what you think people should be maximizing? Or should they maximize something different from what they are maximizing, as many people believe?

    Mike, Singapore’s government spends 1.2% of GDP on health care, and has 100% health care coverage and a higher life expectancy that any European country. Our government spends more than 7% of GDP on health care, and we have 45 million uninsured. If you suggest we should junk our system adopt Singapore’s, I’ll support you 100%. If you suggest pouring billions more into our flawed private insurance model, I am not going to support you.

    Statsguy, You make some good points. BTW, one key here is that the expectation of devaluation was never high enough to force the US to abandon the gold peg. Under floating rates an expectation of future devaluation causes the current spot rate to depreciate. That did not happen in 1932, which is one reason why, as you say, we had the worst of both worlds.

    The only difference in emphasis is that I think the key was getting prices higher so that production again became profitable (assuming sticky wages.) You seem to focus more on getting prices higher to improve the net wealth position of debtors. Both factors may play a role.

    Doc Merlin, I see your point but you always have to be careful with these analogies. In the early 1930s gold was money, so an increase in the real value of that particular commodity was deflationary. Now cash is money, so if people swap cash for commodities we now get inflation. I am not saying they are no lessons for today from this period, I think there are, and will end chapter 6 with a long discussion of where we are making the same mistakes. But the gold is money factor complicates drawing parallels.

    I agree that the start of this recession was more like 1920. The difference is that the problem in 1921 was big a decline in the monetary base, whereas this around it was increased demand for base money. But otherwise they are similar. Also, wages were more flexible back then, so the recovery was quicker.

  12. Gravatar of Mike Sandifer Mike Sandifer
    16. March 2010 at 15:33

    Scott, utility is simply about optimizing the passing of genes, directly or indirectly.

    Now, unless you want to skip the rest:

    The definition I offered above is based on decades of experimental research in both animals and people.

    This link has a nice chart in figure 2 that diagrams the process by which stimuli are encoded for utility and then behavioral responses:

    The equation toward the end in the chart is the matching law equation, and is obviously very simple, but it has held up well since it was formulated in the 60s.

    Basically, it says that behavioral options exist in a sort of portfolio, with each option divided by the values of all other options at a given moment. However, unlike the Markowitz formulation, diversification is defined differently. I

    formulate it in terms of expected gains – expected losses, multiplied by the required rate of intake of all reinforcement (resources necessary to pass genes), and you have utility, with a familiar negatively accelerated curve.

  13. Gravatar of jsalvatier jsalvatier
    16. March 2010 at 16:00

    please change your FAQ link on the sidebar to, it still tries to link to your old blog, which makes it difficult to link people who have frequently asked questions to your frequently given answers.

  14. Gravatar of ssumner ssumner
    17. March 2010 at 07:44

    Mike, I don’t agree with your definition, and no amount of experimental results can change my mind if I don’t agree with your basic definition. If you want to argue that utility is something like happiness, and that people maximize happiness by doing things that improve their genetic fitness, then I would look at the evidence. I probably wouldn’t agree, but I’d consider it a reasonable hypothesis.

    But if you define utility in terms of gentic fitness then I am just not interested. I have no interest defining utility that way, it doesn’t seem to be a very useful definition. And no amount of evidence can show a defintion is “correct” if most people don’t use that definition.

    Do you believe people collect stamps because it improves their genetic fitness?

    jsalvatier, That’s odd, on my computer when I hit FAQs it leads me to the new blog. I’m not sure why yours is different, but I’ll look into it.

  15. Gravatar of Mike Sandifer Mike Sandifer
    17. March 2010 at 09:02


    Okay, I can see now where some objections from some economists might come from. Unfortunately, the issue of genetic fitness just isn’t that simple, and one can’t ignore the way the brain functions when thinking about things like stamp collecting.

    My guess is that stamp collecting may have orginally reflected sentimentality on the part of collectors. That is simply a byproduct of a metacognitive capacity for achieving some satisfaction by having uniquely specific stimuli(the stamps), which will elicit specific happy memories, or even tragedies that one may feel guilty forgetting(guilt is just the opportunity cost of actions taken or forgone, again with a relationship to gentic fitness that is not so straightforward).

    Since stamp collecting became popular, through coincident interests, then through deliberate sharing, then motivated by monetary gain, you perhaps end up with the market you get today.

    Psychology is a hard science, mathematical field. this is not reflected in the work of practioners, many of whom are the equivalent of palm readers.

    Even emotions can be defined concretely. For example, anger is a response to an outcome that is worse than expected. Since expectations(learned or unlearned) determine behavioral investment, violations thereof necessarily indicate a loss, and the expectations were necessarily unrealistic.

    There is a phenomenon known as “hangovers,” which have, in the case of anger responses, people attacking soda machines, yelling at people in traffic with their windows up, etc. This is due to brains wired to treat everyone as if they’ll be encountered on an ongoing basis, as mobiiity was vastly less for most of the 200,000 years of human prehistory. Strangers were much more rarely encountered. Also, it was far less common for most of that time for the inanimate to cause violations of expectations. Much more commonly, it was conspecifics(other humans).

    Anger obviously motivates punishment, and the degree to which we can refrain from responding and make proper discriminations with respect to the targets of anger, is inversely related to the magnitude of the anger.

    Anyway, that’s just an example of this perspective. This model of anger is based on many series of experiments that began with this paper by Azrin, et al:

    Azrin N H, Hutchinson R R & Hake D F. Extinction-induced aggression. J. Exp. Anal. Behav. 9:191-204, 1966.

    This is a very highly developed field and is what I’ve referred to in a couple of previous comments about the behavioral economic development of a new microeconomics that might be applied to macro at some point. Unfortunately, I have no idea how that might be done, especially given that I don’t even understand macro.

    This is not psychology as most people think they know it.

  16. Gravatar of Mike Sandifer Mike Sandifer
    17. March 2010 at 09:23

    I should’ve mentioned that stamp collecting is in a category of incidental behaviors that the unique metacognitive capacity of humans allows.

    You probably won’t even read what I’ve written so far, so I’ll stop here, but I can go into the metacognitive capacity more, and symmetry, transitivity, and learning in general to expand on the points above.

    I’ll end by saying that while many write about the mysteries of consciousness, there aren’t any mysteries I can see at the abstact level. Even many of the physiological mysteries are giving way.

  17. Gravatar of ssumner ssumner
    18. March 2010 at 07:11

    Mike, I still consider consciousness to be deeply mysterious.

    You have lots of interesting ideas, but I don’t see how any relate to the question of how to define utlity. Recall that a definition is one thing, and a theory trying to explain how that variable is determiend is another very different thing. You seem to be trying to explain things like why people get angry. But how does this realte to utility? Some people I know like to get angry, others don’t.

  18. Gravatar of Num-Three-Ers « Entitled to an Opinion Num-Three-Ers « Entitled to an Opinion
    18. March 2010 at 19:06

    […] 18, 2010 Num-Three-Ers Posted by teageegeepea under Uncategorized Leave a Comment  Via Scott Sumner comes the question “What took the ‘con’ out of econometrics?“. I bet many […]

  19. Gravatar of Mike Sandifer Mike Sandifer
    18. March 2010 at 23:26


    I was just tryng to offer a tidbit from teh science of psychology that most people are wholly unaware of, with the anger thing.

    That paper I link to above defines utility well, in my opinion. And that matching law equation holds exactly when the reward centers in the brain are directly electrically stimulated. This models how behaviors are distributed.

    But, we have to go a bit deeper for utility. I hate to do this, and it’s my fault. My attempts at explanations haven’t been clear enough. But, here is a little aid(greatly simplified, for easier illustration:

    From a post of mine,

    U = N [ g / ( h + g ) – L / ( h – L )],


    U = utility or net motivation (subjective net expected gain or loss, or you can call this simply demand),

    g = objective gain

    L = objective loss or response cost,

    h = baseline rate of intake of reinforcement, or “have” (mood),

    N = required rate of reinforcement (Need).

    Ultimately, this is a model relating the required rates of intake of unconditioned reinforcers to the resources perceived available, with the ultimate goal being to optimize inclusive fitness, or reproductive success. This is a reinterpretation of the various matching law models.

    So, that’s how I define utility.

  20. Gravatar of Mike Sandifer Mike Sandifer
    18. March 2010 at 23:37

    I should add, that when you look at the derivatives, and other wise play with this, you even get familar results(in no particular order), including

    1. Increased risk aversion with lower moods.

    2. A shifting of behavior toward more proofitable gains, with lower moods. For instance, higher calorie foods, more immediate sex (when search costs are relatively low), etc.

    3. A roughly inverted U-shaped consumption curve, indicating the over-reward effect. I see this in microeconomic texts as well, with respect to wages.

    4.) A definition for depression as a motivational trap. As mood decreases, the subjective values of both gains and losses increase, with more rapidly for losses. Hence, the formerly paradoxical finding in research that there was a higher need for reward in the depressed, but less net motivation to seek it.

    5.) Increases in the magnitude of the objective gains and losses, with a constant difference between them relative to mood, leads to more risk aversion. Or, the bigger the decision, the more risk averse.

    6. The negatively accelerated curve familiar to economists

    7. A distribution of behaviors consistent with portfolio theory.

    Maybe I’m leaving something out, but you get the basic idea.

  21. Gravatar of ssumner ssumner
    19. March 2010 at 05:56

    Mike, I have no idea what terms like “objective gain” mean. or how they differ from “subjective gain.” I presume subjective gain would be something like how better off you think you are, but what is objective gain?

    I also don’t know what you mean by “mood.” You put this into an equation. Does that mean “mood” can be measured? If so, how?

  22. Gravatar of Mike Sandifer Mike Sandifer
    19. March 2010 at 20:07

    “Objective gain” refers to a measurable input, such as rate of calories or volume of water consumed, etc.

    “Mood” is simply the sum of the all net reinforcing options at a given moment. All inputs and outputs are ultimately translated into mood, as a common currency.

    Ways to measure mood include:

    1. Comparing differential rates of substitution, baseline versus manipulation. Changes versus baseline will obviously reveal the derivative of the utility curve over the relevant range. This can be done globally, when relevant variables cannot be controlled or accounted for with regard to specific behaviors. However, there are some excellent ways of controlling for relevant variables.

    2. Compare differential response latencies, or the time between stimulus detection and the conditioned or unconditioned response.

    3. Or, in a more direct absolute sense, measure energy expenditure as revealed in autonomic response indicators such as heart rate, respiration rate, skin conductivity, pupil dilation, etc.

    4.) Measure ability to delay gratification. Lower moods lead to more temporal discounting.

    5.) Measure the risks and costs willingly taken to faciliate gains.

    6.) Measure responses to anxiolytic drugs, or other relevant psychotropics.

    7.) Measure the degree of tolerable unpredictability of gains and losses(similar to 5).

    I’m sure I’ve left some things out, but I am about to collpase, so I better get some sleep.

  23. Gravatar of Mike Sandifer Mike Sandifer
    24. March 2010 at 01:30

    I shouldn’t post so early in the morning. I really flubbed the post before the last one. I meant subjective gains and losses increase with lower mood. Anyway, I think you know I’m usually not so sloppy, though I understand it you don’t particularly remember.

  24. Gravatar of ssumner ssumner
    24. March 2010 at 06:29

    Mike, I just don’t see where any of this is going. I can’t imagine what caloric intake has to do with utility. I still don’t think you have even begun to define the term ‘utility.’ Some people want more calories, some want less. But we need to figure out what utility is, not calories.

    Why do you assume that the way people respond to stimuli measures utility?

  25. Gravatar of Mike Sandifer Mike Sandifer
    26. March 2010 at 14:17


    “In reply to “Mike, I just don’t see where any of this is going. I can’t imagine what caloric intake has to do with utility. I still don’t think you have even begun to define the term ‘utility.’ Some people want more calories, some want less. But we need to figure out what utility is, not calories.”

    Caloric intake is just one requirment for living organisms to behave. It is one example of the many resources required to live and pass genes. Others include water, rates and quality of social interactions, frequency and adaptive value of sex, etc.

    As previously stated, the sum of all of the rates of net intake of such resources, past, current, and expected, temporally discounted, determines “mood,” thusly operationalized.

    So, there are two types of utility. There is total utility, which is the long run average rate of intake of biological resources. There is also moment utility, which involves the temporally sensitive reactions to current and expected events relative to long run utility.

    So, if we break it down, the utility of engaging in any behavior is reflected in the ratio of its benefit over all other currently favorable options. (you also multiply this by the rate of intake required to sustain the organism).

    Now there is just one more complication. Break the ratio up into expected objective(easily measured) gain over all other options and subtract the expected objective loss(costs), also over all other options, and you have the utility of a given behavioral option.

    You also replied:

    “Why do you assume that the way people respond to stimuli measures utility?”

    The way people respond to stimuli is an external reflection of the brain’s internal computation of utility.

    I refer to a specific chart there that illustrates this nicely:

    This comes from the short html paper linked to in prior replies at

    This discussion has apparently been more of a distraction than anticipated, so if you want to cut to the chase in the future, refer to that paper.

  26. Gravatar of Mike Sandifer Mike Sandifer
    27. March 2010 at 08:16

    If you lack time, I consider utility net motivation, or you can call it demand.

    This is not just an abstract concept, but has even been measured via direct brain stimulation.

    When and if you ever have the time and interest, you might want to look at this paper:

    There is a direct and essentially exact causal relationship between brain stimulation and exhibited behavior. The allocation of behavior exactly follows the matching law.

    If you ever want a copy of the paper, which is ostensibly unlikely, I’ll scan it and send it to you.

    While you may still not agree with this definition of utility, the experimental research is moving ahead, no matter what any economist thinks. Neuroeconomics is the real study of utility within individuals.

  27. Gravatar of ssumner ssumner
    27. March 2010 at 10:28

    Mike, Your paper says:

    “Following modern usage in economics, utility is used here to represent the basis for choice. If an animal consistently chooses option x in a given set of circumstances, then that option is said to have a higher utility than the competing alternatives at the time of decision.”

    Doesn’t that make it tautological that every person maximizes utility? Obviously there are many people who define utility differently, as there are many people don’t think that humans maximize utility.

  28. Gravatar of Mike Sandifer Mike Sandifer
    1. April 2010 at 15:43


    Sorry about the lag in my response, but I’m so busy lately that I hardly have time to consider anything unrelated to work.

    I think that paper makes more than just a semantical claim. The suggestion is that the way behavior is allocated over reinforcing options, or in response to direct brain stimulation reward, is optimal. or nearly so given the environments in which brains evolved.

    In the example I used above about anger, clearly getting angry at fellow drivers on the road makes little sense in large cities today, in which strangers usually remain strangers. But, over almost all of 200,000 years, human beings had significantly less mobility and hence others encountered would likely be around for sometime. The punishment anger motivates only makes sense in that evoluationary context(this is known as evolutionary overhang).

    Likewise, behavior that is ostensibly sub-optimal are almost always optimal in the evolutionary context. Exceptions include new, random genetic mutations, most of which are maladaptive.

    To me, utility refers to the usefulness of something, be it a stimulus, a behavior, etc., and the approach in that above paper allows one to determine how useful something is, both subjectively and objectively.

    I should point out, that it’s really saying that behavior, and the internal calculations of utility, are allocated according to the weighted average of the net expaected gain. This was the central approach addressed by Markowitz in his portfolio theory, as I recall(though the treatment of risk is different, etc).

  29. Gravatar of ssumner ssumner
    2. April 2010 at 08:21

    Mike, You talk about “usefulness” and “net expected gain” without defining these terms. Useful for what? Gain in what terms? What is the criteria? Happiness? Money? Genetic fitness? Fame and glory? I have no idea what people should be maximizing, and I don’t see anyone else with good answers.

  30. Gravatar of Mike Sandifer Mike Sandifer
    2. April 2010 at 16:27


    I think we’re retreading some points here. Utility, or “usefulness” is in terms of passing genes.

    Evidence includes the willingness of the males of most mammalian species tested to forego food and water in exchange for time with novel mating opportunities, even unto death. Also, in most species, mothers are more willing to defend siblings than offspring when faced with an inescapable dilemma. This is due to the fact that siblings have a larger number of genes in common than parents and offspring.

    There are many other indirect ways to facilitate the efficient passing of genes as well, including the less fit in a family taking supporting roles vis-a-vis the better fit, or even the former committing suicide to reserve resources for the latter(this suggests a neural program that kicks in during severe depressino, and is triggered by some psychotropics).

    Everything we sense and do is related to maximizing fitness, even in many cases in which it is not immediately obvious. This is even true when our evolved nature is incompatible with modern life, and this indeed tells us more about the environments in which evolution took place than the obvious cases. Of course, again, there are random genetic mutations, but these are very rare.

    This opens the door to many “just so” stories, but clever experimental tests can at least determine whether available evidence is consistent with a given hypothesis.

    To give just one “just so” story, take the case of the relative social sensitivity of children versus adults. Children care much more about social approval in general, especially among their peers. They also face more social pressures, including ridicule and bullying. The story goes that this is due to the aforementioned lack of mobility that was a human reality for 200,000 years. Hence, children evolved to treat each other as if they would always be within the same social group. So, these early interactions help to begin to establish a more or less permanent social hierarchy, with severe consequences for those deemed less fit.

    More generally, we tend to like people we see as fit. This includes people who are smart or physically fit, or attractive, as well as those who are similar to us. In the latter case, our learned mindsets presumably reflect our best perception of our adaptive needs, and hence we see those similar as also being adaptive and having confluent interests. This means greater incentives for cooperation, or “friendship.” Indeed, we value people we consider more fit, as cooperation with them will likely yield more benefits with regard to resources.

    So, I’ve taken a different approach here, trying to repaint the world in terms of adaptive(inclusive) fitness. This is a very different mindset for most, and hence it is very easy to miss some of the subtleties that are necessary for understaning the economics of passing genes.

  31. Gravatar of ssumner ssumner
    2. April 2010 at 17:12

    Mike, When I decide what movie to go see, it is based on which one I think I will enjoy more. I couldn’t care less about how it affects my genetic fitness. If that was my goal I’d probably watch soap operas to learn strategies for manipulating people. Nor do I like people better who are attractive.

    It seems to me that your comment is a defense of evolutionary psych. I have nothing against E.P., I just don’t see it as helping us understand what utility is. When economists use the term ‘utility,’ they are not talking about “usefulness.” If evolutionary psychologists define utility as “usefulness” that is fine. But be aware that is isn’t the concept economists are referring to.

  32. Gravatar of Mike Sandifer Mike Sandifer
    3. April 2010 at 03:34


    The idea that anything you do can be absent utility as I see it ignores the fact that the brain is wired in specific ways to allow you to enjoy the movies you like. Also, the very ways we learn follow specific evolutionary determined rules. Preferences are never without the influence of genetic “usefulness,” again except for most genetic mutations.

    If we take this fundamentally, it is not merely by chance that we generally enjoy novel experiences, when available resources do not rule out a range of them due to dangerous volatility. Novelty serves the purpose of motivating diversification of sources of resources, including the less obvious, indirect examples.

    Then, it is not accidental that we enjoy novel movies about social interactions, natural landscapes, etc. The importance of these things involve the passing of genes, via mating opportunities, cooperation, and competition in the former case, and human-friendly environments in the latter. Preference fron this perspective is never sans evolutionary significance.

    Take also the fact that we value foods at all, that we tend to value high calorie foods and distribute our choices with regard to food composition according to the matching law. Take also cognitive biases, some of which naturally arise as a function of small world networks(neural networks); the fact that we enjoy or find aversive certain more direct means of neurotransmitter manipulation(drug use); or just the economics of energy expenditure in general.

    Just because these links to evolutionary fitness are not always obvious, doesn’t mean they don’t exist. Frankly, most of these links have already been discovered.

    With regard to the definition of utility being “usefulness,” you are much more qualified to know whether it is related to microeconomic definitions. However, taking the matching law definition results in a negatively accelerated curve, which can certainly only coincidentally be similar to that for economic utility, unlikely though it may be.

    Anyway, unless I misunderstand economic utility, any contrary formulation should be seen as a useful approximation at best, and nonsense at worst.

    I think the same holds true for your views of EMH, except that the difference in predictive power between it and ecological approaches may often be slight. Hence, you holding on to them may have no significant consequences. Ecological approaches to investment may often be merely scientific curiosities.

Leave a Reply