Misbehaving: The Making of Behavioral Economics

by Richard H. Thaler

  25

  The Battle of Closed-End Funds

  Shiller’s work wounded the price-is-right component of the efficient market hypothesis, but it was not considered a fatal attack. Disputes about methodology still lingered. And, although it was hard to justify what happened that week in October 1987, efficient market advocates were unwilling to rule out a rational explanation. In the spring of 1988, the University of Chicago held a conference about the crash, and one panel included Eugene Fama and me. Gene spoke first and said the market should be congratulated for how quickly it had reached its new equilibrium, meaning that something must have happened to cause people to revise down their estimates of the future returns on the stock market, and prices had adjusted immediately, just as they “should.”

  When it was my turn to speak, I asked the assembled experts if they thought that the present value of dividends had fallen 20% on Black Monday, as it was called. Only a few hands went up, and Gene’s was not among them. I raised my eyebrow as if to say, “Well?” Gene shot his hand up in the air, smiling. He was not ready to concede, but kept his good sense of humor.

  A smoking gun would be required to convince Fama and the rest of the efficient market clan. But as we saw earlier, intrinsic value cannot be determined with precision, which makes it hard to prove that stock prices deviate from intrinsic value. One possible approach to testing whether prices are “right” is to employ an important principle at the very heart of the EMH: the law of one price. The law asserts that in an efficient market, the same asset cannot simultaneously sell for two different prices. If that happened, there would be an immediate arbitrage opportunity, meaning a way to make a series of trades that are guaranteed to generate a profit at no risk. Imagine that gold sells for $1,000 an ounce in New York and $1,010 an ounce in London. Someone could buy gold contracts in New York and sell them in London, and if the transaction costs of making that trade are small, money can be made until the two prices converge. The existence of a multitude of smart traders who are constantly on the lookout for violations of the law of one price guarantees that it should hold, almost precisely and instantaneously. Finding a violation would strike at a core tenet of the EMH.

  And yet a violation was surprisingly easy to find. In fact, it had already been written about by, among others, Benjamin Graham. The law-breaking suspect was a type of mutual fund called a closed-end fund.

  With a more familiar open-end fund, investors can at any time put money into the fund or take money out, and all transactions are conducted at a price determined by the value of the fund’s underlying assets, the so-called Net Asset Value (NAV) of the fund. Imagine a fund that only buys shares in Apple Corporation, and one share of the Apple fund gets you one share of Apple stock. Suppose Apple is selling for $100 a share, and an investor wants to invest $1,000. The investor sends the fund $1,000 and gets back ten shares of the fund. If the investor later wants to withdraw from the fund, the amount returned will depend on the current price of Apple. If the price has doubled to $200 a share, when the investor cashes out she will get back $2,000 (less fees charged by the fund). The term “open-end” means that the assets managed by the fund can grow or shrink depending on the preferences of its investors.

  A closed-end fund works differently. Managers of the fund raise an initial amount of money, say $100 million, and that is it. No new money can be invested, and money cannot be withdrawn. (You can see the appeal of starting such a fund to the portfolio managers. Investors can’t withdraw their money!) Shares of the fund trade in the market, so if an investor wants to sell her shares, she has to do so at the fund’s market price. Go back to the hypothetical Apple fund example and suppose now the fund is organized as a closed-end fund, and as before, one share of the fund gets you one share of Apple stock. What is the market price for the Apple closed-end fund? One would assume net asset value, namely, the current price of Apple. If it were anything else, the law of one price would be violated, since it would be possible to buy Apple shares at two different prices, one determined by the market price of Apple shares, and the other by the price of the Apple fund.

  The EMH makes a clear prediction about the prices of closed-end fund shares: they will be equal to NAV. But a look at any table of the prices of closed-end fund shares reveals otherwise (see figure 15). These tables have three columns: one for the fund’s share price, one for the NAV, and another for the discount or premium measuring the percentage difference between the two prices. The very fact that there are three columns tells you that market prices are often different from NAV. While funds typically sell at discounts, often in the range of 10–20% below NAV, funds sometimes sell at a premium. This is a blatant violation of the law of one price. And an investor does not have to do any number-crunching to detect this anomaly, since it is displayed right there in the table. What is going on?

  FIGURE 15

  I did not know much about closed-end funds until I met Charles Lee. Charles was a doctoral student in accounting at Cornell, but his background hinted that he might have some interest in behavioral finance, so I managed to snag him as a research assistant during his first year in the program. When Charles took my doctoral class in behavioral economics, I suggested closed-end funds as a topic for a course project. He took the challenge.

  Around the time that Charles finished his paper for my class, Larry Summers had just written the first of a series of papers with three of his former students about what they called “noise traders.” The term “noise traders” comes from Fischer Black, who had made “noise” into a technical term in finance during his presidential address to the American Finance Association, using it as a contrast to the word “news.” The only thing that makes an Econ change his mind about an investment is genuine news, but Humans might react to something that does not qualify as news, such as seeing an ad for the company behind the investment that makes them laugh. In other words, SIFs are noise, and a noise trader, as Black and Summers use the term, makes decisions based on SIFs rather than actual news.

  Summers had earlier used more colorful language to capture the idea that noise might influence asset prices. He has an infamous but unpublished solo-authored paper on this theme that starts this way: “THERE ARE IDIOTS. Look around.”* Three graduate students who had met when they shared a suite during their first year as undergraduates—Brad De Long, Andrei Shleifer, and Robert Waldmann—joined Summers to produce a more rigorous, thorough, and polite version of the “idiots” paper. The model they proposed used closed-end funds as an example of the type of asset that their model might help understand, but they had not done any empirical testing. Charles and I thought we might be able to build on some of the work Charles had done for his term paper to fill in that gap, and we asked Andrei Shleifer, who had recently joined the faculty at the University of Chicago, to join us on this project. Charles, Andrei, and I then wrote a paper on closed-end funds, noting that there were four puzzles associated with these funds.

  When closed-end funds are started they are usually sold by brokers, who add a hefty commission of around 7% to the sale price. But within six months, the funds typically trade at a discount of more than 10%. So the first puzzle is: why does anyone buy an asset for $107 that will predictably be worth $90 in six months? This pattern had induced Benjamin Graham to refer to closed-end funds as “an expensive monument erected to the inertia and stupidity of stockholders.” This was a more polite way of saying “THERE ARE IDIOTS,” which remains the only satisfactory answer to this first puzzle.†

  The second puzzle is the existence of the discounts and premia mentioned earlier. Why does the fund trade at a price that is different from the value of its holdings?

  The third puzzle is that the discounts (and premia) vary quite a bit, across time and among funds. This is an important point because it rules out many simple explanations for the existence of discounts. One such explanation had held that the discount was necessary to compensate investors for the fact that the funds charge fees or mismanage the portfol
io. But if factors like this were the explanation, why do the discounts bounce around so much? Neither the fees nor the management vary much over time.

  The fourth puzzle is that when a closed-end fund selling at a discount decides to change its structure to an open-end fund, often under pressure from shareholders when it is selling at a large discount, its price converges to NAV. This fact rules out the possibility that net asset value was miscalculated. Collectively, the four puzzles created an efficient market conundrum.

  The primary goal of our paper was to draw a bit more attention to these puzzles. But our main research contribution was to understand a bit more about why discounts vary over time. We exploited an important fact about the U.S.-based closed-end funds we were studying: individual investors, as opposed to institutions, are the primary owners of these funds. We postulated that individual investors acted as the noise traders in this market; they would be more flighty than professional investors, such as pension funds and endowments, and thus they would be subject to shifting moods of optimism or pessimism, which we dubbed “investor sentiment.” We conjectured that when individual investors are feeling perky, discounts on closed-end funds shrink, but when they get depressed or scared, the discounts get bigger. This approach was very much in the spirit of Shiller’s take on social dynamics, and investor sentiment was clearly one example of “animal spirits.”

  The question was how to measure investor sentiment. To do so, we exploited the fact that individual investors are also more likely than institutional investors to own shares of small companies. Institutions shy away from the shares of small companies because these shares do not trade enough to provide the liquidity a big investor needs, and institutions such as mutual funds don’t buy shares of closed-end funds or other mutual funds because their customers don’t like the idea of paying two sets of fees. So, if the investor sentiment of individuals varies, we figured it would show up both in the discounts on closed-end funds and on the relative performance of small companies versus big companies. (Although shares in small companies do better on average, the difference varies, and in some periods big firms outperform small ones.)

  That is exactly what we found. The average discount on closed-end funds was correlated with the difference in returns between small and large company stocks; the greater the discount, the larger the difference in returns between those two types of stocks. This finding was the equivalent of finding footprints for Bigfoot or some other creature that is thought to be a myth.

  As I have said, we were by no means the first to write about closed-end funds. Economist Rex Thompson wrote his thesis on closed-end funds, and found that a strategy of buying the funds with the biggest discounts earned superior returns (a strategy also advocated by Benjamin Graham). The well-known efficient market guru Burton Malkiel, author of the perpetual best-seller A Random Walk Down Wall Street, has also advocated such a strategy. Nevertheless, our paper made some people upset, and it infuriated Merton Miller, the Nobel Prize–winning financial economist at the University of Chicago who was Shleifer’s senior colleague.

  To this day I do not know exactly what about our paper made Miller so upset, but I suspect that while others had written about these funds before, we were the first to do so since Graham without following the mannerly procedure of apologizing and making excuses for our anomalous findings. Instead, we seemed to be enjoying ourselves. On top of that, we were using one annoying anomaly, the small firm effect, to help explain another one, persistent discounts on closed-end funds. This is, for an Econ, the equivalent of taking the Lord’s name in vain while working on the Sabbath.

  Miller jumped into attack mode. We submitted our paper to the Journal of Finance, and the editor, René Stulz, sent it out to referees. Meanwhile, we learned that Miller was lobbying Professor Stulz to reject our paper. To his credit, Stulz accepted our paper and told Miller that if he disagreed with our findings, he should follow the usual procedure of writing a comment on the paper and submitting it to the Journal.

  Miller took Stulz up on this suggestion. He recruited Nai-fu Chen, a fellow Chicago professor, and Raymond Kan, a graduate student, to help with the research, and they submitted a comment on our paper. Miller was a sharp-witted guy, and the comment was written in his usual swashbuckling style. They started their paper this way: “Charles Lee, Andrei Shleifer, and Richard Thaler (1991) claim to solve not one, but two, long-standing puzzles—discounts on closed-end funds and the small firm effect. Both, according to Lee et al., are driven by the same waves of small investor sentiment. Killing two such elusive birds with one stone would be a neat trick indeed if Lee et al., could bring it off. But they can’t.”

  I won’t bore you with the substance of the debate, which was mostly about technical details. Following tradition, we wrote a “reply” to appear in the same issue of the journal, and introduced new data to buttress our claims, something Miller considered a violation of the usual protocol of such debates. He insisted on a reply to our reply, which meant that, following tradition, we as the original authors would get to throw the last set of stones.

  Naturally, in the final two comments both sides declared victory. I don’t know who won, but I do know that the unprecedented four-part pissing contest about our paper attracted a lot of attention. Thanks to Professor Miller, hundreds of financial economists were nudged to read our original paper, so by attacking us, Miller ended up doing us a big favor. Many readers of the Journal of Finance might otherwise not have noticed a paper about closed-end mutual funds. But nothing attracts attention more than a good fight.

  ________________

  * The only copy of this paper I have been able to find is one Fischer Black faxed to Summers with his handwritten comments on it. Next to the opening phrase about “IDIOTS” Black writes: “I call them ‘noise traders’. They trade on noise as if it were information.”

  † To be clear, it can be quite smart to invest in closed-end funds when they sell at a discount, but it is foolish to buy one when it is first issued and a commission is being charged.

  26

  Fruit Flies, Icebergs, and Negative Stock Prices

  The debate with Merton Miller obscured the most important point about closed-end funds: the blatant violation of the law of one price. It was as if we had discovered a unicorn and then had a long fight about what to call the color of the beast’s coat. Years later, after I had joined the University of Chicago, I revisited the law of one price with a Chicago colleague, Owen Lamont.

  At that time Owen was not really a behavioral economist. He was just an open-minded researcher who enjoyed stirring the pot and had a good eye for interesting problems. Owen is always a favored choice for the role of discussant at the behavioral finance seminars that Shiller and I organize at NBER. Zingers abound at these meetings, and Owen may hold the record for most points scored. Once he was asked to discuss a paper in which the authors had measured option traders’ anxiety levels during the trading day. The sensor technology employed was nifty, but a lot of us wondered what we should take away from this exercise. Owen opened his discussion with a summary: “The authors have definitively rejected the hypothesis that these traders are blocks of wood.”

  The interesting problem Owen had spotted was a blatant violation of the law of one price involving a company called 3Com. 3Com’s main business was in networking computers using Ethernet technology, but through a merger they had also acquired Palm, the maker of what at the time was considered a very spiffy handheld computer called the Palm Pilot. In the summer of 1999, when the stock of any respectable Silicon Valley technology company seemed to double every month or two, 3Com was being neglected, and its stock price was flat. 3Com management adopted a plan of action to increase its share price, and the plan involved divesting itself of its interest in Palm. On March 2, 2000, 3Com sold a fraction of its stake in Palm to the general public. In this transaction, called an equity carve-out, 3Com sold about 4% of its stake in Palm in the initial public offering, sold about 1% to a consortium of fi
rms, and retained ownership of 95% of the shares.

  This action in and of itself should worry efficient market advocates. What difference does it make whether Palm is located within 3Com or out on its own? If prices are “right,” then splitting a company up into two parts should not raise its value unless the parent company, 3Com in this case, was doing something dysfunctional in its management of Palm that was keeping that division from thriving. But of course, the 3Com management did not say they were getting a divorce from Palm to allow it to get out from under their mismanagement. Instead, they implied that Palm would somehow be magically worth more as a separate company than as a part of the parent company. Undoubtedly, they were hoping that as a separate company Palm would be valued more like the sexy technology companies around at that time, such as eBay, AOL, and Amazon. An efficient market advocate would be skeptical of this move. In a market consisting only of Econs, the value of 3Com is equal to the value of Palm plus the value of the rest of 3Com, and splitting them up would have no effect on the total value of the enterprise.

  But Econs were clearly not driving the stock prices of technology firms in the late 1990s. Puzzling as it might be, carving out Palm seemed to work. When the plan to separate Palm from the rest of the company was first announced on December 13, 1999, 3Com was selling for about $40 a share, and by the time the initial public offering for the Palm shares took place, on March 1, 2000, the 3Com price had risen to over $100 a share. That is quite a return for taking the costly step of turning Palm into a separate company! But the truly bizarre part was yet to come.