I am taking "Decision Analysis" this semester through Stevens Institute of Technology, which is part of the reason I have been epically failing at posting anything new. However, I had to sit down and write a bit tonight for an assignment, so I am posting here. The book is “Against the Gods: The Remarkable Story of Risk” by Peter L. Bernstein, and I HIGHLY recommend it to just about everyone – not just engineers or economists.
Bernstein presents the evolution of the understanding of risk and its application and influence on society in general and decision making in particular. Bernstein takes us from the Greeks, who “presumably … admitted no possibility of regularity in earthly events” and the perspective that “precision was the monopoly of the gods (44)” through the centuries to an environment where the mathematical concepts of probability and statistics are well understood. However, the Greeks were not entirely incorrect – regularity in decision making is still not, nor is it likely to ever be, easily achievable. Rational decision making is facilitated by the understanding that risk is based on the somewhat objective parameters of probability and severity. However, the weight the decision maker places on the outcome is also crucial, and is very subjective. Better decisions can be made when the situation is well understood; the more information, the better. However, no matter how much data mining is done, all decisions are made on the basis of limited data. Trending is one way to understand the past and is often the basis for making decisions about the future. When paired with the concept of a normal distribution and the law of law numbers, the principle of regression to the mean can assist decision makers. Unfortunately, even when a decision situation is well understood, facts are subjective as is utility. The utility of additional gain of a quantity also tends to decrease and is dependent on the quantity previously possessed. The key point of the book is that despite all the variability possible in the decision making process, decisions made using disciplined procedures will, over time, outperform guesses, fate, destiny, or other ad hoc methods.
The outcome of a decision has uncertainty. This uncertainty typically involves a desired and undesired outcome. Risk is tied to the occurrence of the undesired outcome, or the non-occurrence of the desired outcome. Risk is based on probability, severity, and also the importance that the desired outcome will occur (p. 104), therefore “both gravity and probability should influence a decision (p. 71).” Bernstein writes that “a decision should involve the strength of our desire for a particular outcome as well as the degree of our belief about the probability of that outcome.” The discussion in the book that the weight or importance placed on an outcome should factor into the decision making process is one that seems very logical but is a step that is also neglected or misused. I often don’t consciously consider the importance of an outcome when making a decision, or only consider the short-term implications. For example, I often decide to stay up after finishing my work to watch a show or movie. The consequence of the decision to stay up later is that I will be more tired than if I were to go to bed at that time, and the probability that I will be more tire tired is nearly unity. In making that decision, I am weighing the pleasure of relaxing in front of the TV to be more important than my mental acuity at work the next day. However, in such a case, I often ignore or fail to consider the weight of the consequence one step down from the immediate outcome. Considering how I weight the outcomes of my decisions will be one takeaway from this book.
The idea of weight also made me think that some people will put more importance on the severity of the outcome than on the probability (the example from the book being the one elephant in Russia that was killed by a bomb – severity was high, probability was very low). For example, for some people the fear of flying is overwhelming; they see the practice as incredibly risky and dangerous. Of course, most people place a high importance on their own safety. But, people who prefer not to fly for this reason are considering the severity of the undesired occurrence (a crash) much more than the very low likelihood of that event.
A second key concept from “Against the Gods” is that all decisions are made on the basis of limited data (p. 73). We know the key components of risk (probability, severity, and weight), but how are these parameters quantified? The likelihood of a future event is often (and subjectively) extrapolated from history (e.g. stock market) or based on physics and statistics (e.g. spin of a roulette wheel). Whether in deciding to invest in a particular index fund or to bet on number 23, “we all have to make decisions on the basis of limited data (p. 73).” Bernoulli explained we must assume that “under similar conditions, the occurrence (or non-occurrence) of an event in the future will follow the same pattern as was observed in the past.” Furthermore, “an estimate of probabilities after the fact also is impossible unless we can assume that the past is a reliable guide to the future. The difficulty of that assignment requires no elaboration. The past, or whatever data we choose to analyze, is only a fragment of reality.” So, Bernoulli summarizes that we have to try to obtain as much data as possible, but be aware that this data is not comprehensive. This fact is very useful to recognize, if maddening. The question remains – how much data is enough to make a sound decision? And is the amount of data dependent on the importance of the decision? To these questions, I answer, “it depends,” and “yes,” respectively. It becomes apparent that setting up a framework for a decision is not a linear process, but rather an iterative one in which it can be beneficial to use feedback from your own thought process in order to optimize the decision framework.
Regression to the mean is the third important topic from “Against the Gods.” Francis Galton was a pioneer in this area, finding that the offspring of a couple tended to be closer to the average height than the height of the parents. Essentially, a shorter couple would tend to have children taller than the “Height of Mid-Parents” (average height of parents), and taller couples would have shorter children. Quetelet contributed to the idea of normal distribution, which is based on the population mean, and found that the measurements of people tend to fit the “bell curve.” While regression to the mean can be useful in making decisions about the future, it also ties back to the concept of limited data. The mean at any point is only a mean of a snapshot population in history, thus is again “a fragment.” For example, when considering height, there would be many other parameters that contribute to the value (e.g. nutrition), such that the mean height of the population of London in the 1800s would not be a good indicator for predicting height of the population of Taiwan in the 21st century. In terms of the economic environment, it is “perilous in the extreme to assume that prosperity is just around the corner simply because it has always been just around the corner (173).”
Better understanding of regression to the mean will be useful in such areas of my life as investing. My age and circumstances have positioned me favorable to invest from the beginning of my career in a bear market and to have confidence that the market will improve for years to come. Knowing that what goes up must regress to the mean will make me more likely to adjust my portfolio in the future to a more conservative distribution to protect myself against future recessions.
We have discussed methods to better determine the probability that a particular outcome will occur, which can then be used to aid in the decision-making process. However, determining the gravity or severity of a particular outcome is not cut and dry. The fourth key concept is that facts are subjective. Early in the book when discussing existence of God, Bernstein points out that whether or not to believe in God is “a choice in which the value of the outcome and the likelihood that it may occur will differ because the consequences of the two outcomes are different (p. 70).” Bernstein state that the “role of facts is to provide a single answer to expected value (the facts are the same for everyone), [but] the subjective process will produce as many answers as there are human beings involved (105).” An example might be a water treatment company that is going to ensure that the microbe level in treated water must be less than x%, which will cost $y. At that level, the chance of an illness related to the water quality is z%. Consumers would want z=0, where the company will want to minimize y. A certain cost $y corresponds to a particular microbe level x%. It is clear that the two stakeholders here are aware of the same facts, but the facts mean different things to each group.
A good example of this that I deal with is risk aversion. Every person has a different level of risk tolerance corresponding to each scenario. The facts may be the same for two people but could easily result in opposite decisions. I typically like to arrive at the airport such that I am in the security line no later than 45 minutes before my flight. Since I typically print my boarding pass at home and rarely check a bag, this means that I walk into the airport about 45 minutes before my departure time. I do not think it very likely that I will miss a flight, and also feel that I am able to control the situation even if it is not ideal (e.g. ask a TSA representative if a shorter line is available if I am going to be late). My husband, on the other hand, may have the same facts as me (flight status, weather conditions, parking availability) and the same conditions (no luggage, boarding pass in-hand), but will come to a different conclusion. He always decides to arrive in the airport at 90 minutes before the flight. The facts are the same, but the decision is different.
We have learned that incomplete data and subjective interpretation of the facts make decision-making challenging. Compounding this is the fifth concept from “Against the Gods,” which is that utility is subjective and is not a constant. Bentham stated that utility is “that property in any object, whereby it tends to produce benefit, advantage, pleasure, good, or happiness (p. 189).” Bernoulli further described that “the “utility resulting from any small increase in wealth will be inversely proportionate to the quantity of goods previously possessed (p. 105).” Utility is tied to the risk parameter of “weight” or “importance.” If a decision could result in an increase in wealth, that increase would correspond to a certain utility. The utility, however, does not remain proportional to that initial increase. Bernstein says that “rational decision-makers will try to maximize expected utility – usefulness or satisfaction – rather than expected value. Expected utility is calculated by the same method as that used to calculate expected value but with utility serving as the weighting factor.” The converse of gain is perhaps more severe: “the disutility caused by a loss will always exceed the positive utility provided by a gain of equal size (p. 112).”
Kahneman and Tversky later deviated slightly from this description of utility, finding that the “valuation of a risky opportunity appears to depend far more on the reference point from which the possible gain or loss will occur than on the final value of the assets which would result (p. 274).” This explanation resonated with me. As a relatively inexperienced full-time engineer, I have recent memories of having almost no discretionary income while attending college. Back then, gambling $25 on a single hand of Blackjack seemed ridiculous. While I am not the type to gamble more than once every few years, my current income makes the possibility of losing $25 on Blackjack more palatable. This simple example can be extrapolated to the way I handle my personal investments.
Finally, and in summary, decisions made using disciplined procedures outperform “seat of pants” methods (p. 336). The possibility of losing is an integral part of risk management (p. 234), and the probability, severity, and weight of that consequence must be understood. There are many methods to gather data, make predictions, and assign probability, but the decision makers must understand that no data set is ever complete, nor can it ever be fully accurate. Facts are also subjective and “uncertainty lies in the intentions of others (p. 232)”. The unknown intentions of others can lead to speculation about strategies to, for example, hedge against loss or to capitalize on a perceived weakness in the stock market. A volatile stock market may be frightening to some, but “volatility represents opportunity rather than risk, at least to the extent that volatile securities tend to provide higher returns (p. 261). However, volatility doesn’t mean anything in terms of risk “until coupled with a consequence” (261). Again it is clear that a decision framework is essential but creating one is not a linear process.
Today’s technology gives us the advantage of being able to analyze vast amounts of data, to run different predictive models, and to assist in taking subjectivity out of evaluation criteria. However, a computer “only answers questions; it does not ask them (p. 336).” We must understand the parameters that make up risk along with the shortfalls of any decision-making process. It is also critical to know that we are always making decisions with imperfect or incomplete data, and that there is really no such thing as a “sure thing.” Creating a decision framework in which the decision maker is identified, the possible outcomes are understood, the importance of the outcome to the decision maker is clear and the consequence of failure is defined is crucial to making a rational, sound decision.