Absent a crystal ball, diversification is one of the best tools available to the prudent investor. The insight that drives diversification is pretty simple: while there's a chance that the price of any one stock may plummet, it's unlikely that the same will happen to multiple stocks, and some are even likely to rise. But if most investors are in a risk-neutral position because of diversification, should the legal system have a rule that compensates investors who suffer a loss due to securities fraud? In other words, isn't loss due to fraud therefore just another form of risk that savvy investors ought to be able to minimize through diversification?
In her paper Are Investors' Gains and Losses from Securities Fraud Equal over Time? Theory and Evidence, presented at last Thursday's WIP talk, Professor Alicia Davis evaluates the claim that, for any single diversified investor, gains and losses from fraud will net out to zero over time. Using a combination of probability theory and data generated by computer-simulated trading, Professor Davis identifies several significant flaws in the notion that fraud-related losses and gains perfectly cancel out one another over time, suggesting that the legal rule of compensation might still have a useful role to play in securities regulation.
Critics of compensation, Professor Davis notes, commonly analogize the risks of trading fraud-tainted stocks to participating in a coin flipping game where the player wins $1 for heads and loses $1 for tails. The more the coin is flipped, the more likely the player will break even. While this is mathematically true in a limited sense, Professor Davis demonstrates that, thanks to probability theory, it is not the whole truth. As the number of flips goes up, the range of possible outcomes also increases. That is, after two flips, the player can only be up $2, down $2, or neutral. But after, say, a million flips, the player could be up $1,000, down $5,000, or any number of other possibilities. Therefore, while the expectation of having a roughly equal number of heads and tails increases with the number of flips, the chance that the player actually, exactly breaks even is, in Professor Davis's words, "vanishingly small."
Compensation critics might respond by saying they aren't concerned with investors breaking exactly even, but instead believe that being near break-even—say, within a few percentage points—is good enough. Anticipating this response, Davis notes that, in order to be 95% certain that the outcome will fall within +/-1% of break-even, an investor would need to make 10,000 trades—a number far greater than the 700 trades an average mutual fund makes over a ten-year period. Most investors, both individuals and institutions alike, are therefore likely to fall far short of reaching any degree of certainty.
Furthermore, Professor Davis notes that, unlike the coin flip game, gains and losses from fraud-tainted stocks are unlikely to be symmetrical. Because of this, an investor could break-even in the sense that the numbers of winning and losing trades are equal, but nonetheless experience a net loss. This point leads into the second half of Professor Davis's paper, a computer-simulated study of the risks of trading fraud-tainted stocks. By varying characteristics such as trading strategy, initial capital invested, percentage of initial invested capital held in cash, number of stocks in portfolio, and annual turnover rate, Professor Davis constructed 14 simulated investor types representing both institutions and individuals. She then placed these investor types in a simulated trading universe, lasting ten years and comprised of the nearly 15,000 stocks that were publicly traded between 1996 and 2006. Using data drawn from securities fraud class action settlements from the same period, Professor Davis populated this universe with 653 fraudulent stocks.
At the conclusion of the simulation, Professor Davis noticed some striking results. First, the absolute number of fraud-tainted stocks an investor trades is rather low—indeed, the most active trader, with more than 3,100 trades, only traded on average 29.4 fraud-tainted stocks. Despite this low number, most investors are nevertheless almost certain to encounter a fraud-tainted stock at some point, and often with the potential for significant losses. Generally speaking, institutional investors are better protected against fraud than individuals, but both groups are at risk of substantial fraud-related losses.
This insight becomes particularly worrisome when one considers the "risk of ruin," or the possibility that an investor will exit the market after suffering a catastrophic loss, even if there is a theoretical possibility of eventually recouping that loss and breaking even. Investor types in Professor Davis's study were programmed to remain invested in the market for the entire ten years. Actual investors have different operating parameters, and the "risk of ruin" suggests that actual losses could possibly far exceed those predicted by Professor Davis's model.
Professor Davis concludes her study by noting that its results do not "necessarily lead to clear policy prescriptions." However, the study does raise some provocative points, many of which were elaborated upon during the Q&A session following Professor Davis's presentation. For instance, if securities regulation ought only be focused on safeguarding overall investor profits, then it may be worth reconsidering fraud victim compensation. The study indicates, however, that securities fraud raises troublesome distributive concerns, and the current system of compensation may accurately reflect society's desire to protect against asymmetrical harms—that is, investors are generally more concerned with avoiding losses than they are with experiencing gains.
Building on an insight first observed by Frank Easterbrook and Daniel Fischel, Professor Davis further suggests that eliminating compensation may have a negative effect on allocative efficiency, as without the confidence that they can recover some of their fraud-related losses, investors might expend resources that could be put to more productive use than protecting against fraud, or may even leave the market altogether. Furthermore, the study's conclusions point to future inquiry that could contribute to the scholarly debate about compensation, such as comparing the investor types most likely to experience losses from fraud with the plaintiffs most likely to win compensation in securities litigation.
While disclaiming any policy prescriptions, Professor Davis ably shows that there is more thinking to be done on the subject of securities fraud victim compensation.