Plight of the fortune tellers: Why we need to manage financial risk differently

Plight of the Fortune Tellers: Why We Need to Manage Financial Risk Differently

Volumn , Issue , 2010, Pages

  Rebonato, Riccardo ^a  

^a NONE

Author keywords

[No Author keywords available]

Indexed keywords

EID: 84883937393     PISSN: None     EISSN: None     Source Type: Book    
DOI: None     Document Type: Book

References (61)

1. That is, for readers familiar with Markowitz's theory, unless the mix of businesses of a bank already lies on a point of the efficient frontier.
2. J. H. Cochrane and C. L. Culp (2003), Equilibrium asset pricing and discount factors: overview and implications for derivatives valuation and risk management, in Modern Risk Management-A History (ed. P. Field), chapter 5 (London: Risk Books).
3. For an intelligent and fascinating discussion of the shortcomings of pundits and experts at predicting political events.
4. P. Tetlock (2005), Expert Political Judgement (Princeton University Press).
5. To make the story simpler, I have changed the number of ships and ducats from Bernoulli's story. This example (with yet another combination of ships and ducats) is quoted in C. Gollier (2004), The Economics of Risk and Time (Princeton University Press). The treatment in that book gave me the idea to use it for this discussion.
6. An important and closely related field of study goes under the name of "bounded rationality."
7. G. Gigerenzer and R. Selten (2002), Bounded Rationality-The Adaptive Toolbox (Cambridge, MA: MIT Press).
8. J. P. Forgas, K. D. Williams, and W. Von Hippel (eds) (2003), Social Judgements (Cambridge University Press), and references therein. The terms System I and System II are commonly, but not universally, used.
9. For instance, the terms "intuitive mode" and "controlled mode" are used by Kahneman
10. D. Kahneman (2002), "Maps of bounded rationality: a perspective on intuitive judgement and choice," Nobel Prize Lecture (December 2002)).
11. J. LeDoux (1996), The Emotional Brain (New York: Simon & Schuster).
12. See also on a related topic work by G. Lowenstein, B. Knutson, and D. Prelec, recently published in Neuron, about the activity of different parts of the cortex involved in assessing trade-offs between pleasurable and adverse outcomes when evaluating a purchase (quoted in The Economist, January 13, 2007, p. 73).
13. For a simple discussion of the evolutionary advantage conferred by decisional heuristics.
14. P. M. Todd (2002), Fast and frugal heuristics for environmentally bounded minds, in Bounded Rationality-The Adaptive Toolbox (ed. G. Gigerenzer and R. Selten), chapter 4 (Cambridge, MA: MIT Press).
15. For a modern discussion of the importance of salience,
16. M. Bacharach (2006), Beyond Individual Choice: Teams and Frames in Game Theory (Princeton University Press).
17. In some (rare) cases we may know virtually nothing about a phenomenon before we begin collecting the evidence. In these situations our prior belief is said to be "diffuse." Diffuse priors are not without conceptual problems
18. P. Gregory (2005), Bayesian Logical Data Analysis for the Physical Sciences (Cambridge University Press), for a good discussion), but I cannot deal with this topic here. In practical applications I tend to find that, if too much is made of these logically valid objections, the troublemaker is, statistically speaking, often up to no good.
19. This does not necessarily mean that this is the "right" choice. What I have presented is a variation on the theme of an old paper by Paul Samuelson (P. Samuelson (1963), Risk and uncertainty: a fallacy of large numbers, Scientia 98:108-13). Samuelson, an economics professor, offered a colleague a fifty-fifty bet to win $200 or lose $100. The colleague (who became known in the large subsequent literature as "SC,"for "Samuelson's colleague"), turned down the bet, but said that he would have accepted the same bet if repeated 100 times. Samuelson went on to prove that SC acted "irrationally" within the normative framework of utility theory. The point here is that turning down the one-shot bet and accepting the series of identical repeated bets seems natural and commonsensical (and, for what it is worth, it would be my choice as well), but it is not uncontroversial. True diversification, on the other hand, is uncontroversial.
20. See also the discussion of Rabin's critique of utility theory in chapter 4 on this topic.
21. More precisely, in frequentist statistics, parameters, such as the tail probability of the coin, are fixed and not random. So either the coin is fair or it is not.
22. As I said, most, but not all, of the newspapers behaved this way. Indeed, the Financial Times and the Economist were two of the few sober exceptions in the British press and provided a rare example of selfrestraint and responsibility in their reporting at the time. Five to ten years later, when their more balanced assessments proved much closer to the mark than those of their competitors, they unfortunately failed to reap any direct advantage from their self-restraint. Presumably, they had simply sold fewer copies at the time of the hype than they could have. The tools at the disposal of the more sober newspapers to reap an advantage by calling the scaremongers' bluff were too blunt.
23. Operational risk is the risk of losses arising from factors such as fraud, human error, terrorist acts, disruptions to the business, etc. An inelegant definition of operational risk used to be "all risk that is left after trading and credit risk."
24. I will show in chapter 10 that, when it comes to decision making, a "division of labor" of sorts takes place in many institutions.
25. For the moment let us neglect this more realistic but more complex dynamics, and let us just call the decision maker "the risk manager."
26. The astute reader will have noticed that the distribution associated with trade B crosses the A distribution in figure 4.2 (the "no-brainer") only once, but twice in figure 4.3 (the "ambiguous" case). This observation is actually general: if a distribution with a higher mean only crosses another distribution of profits and losses once, then it "stochastically dominates" the latter. This means that any utility-maximizing investor would prefer the distribution to the right.
27. W. T. Ziemba (2003), The Stochastic Programming Approach to Asset Liability Management (Charlottesville, VA: Association for Investment Management and Research).
28. Economists explain that this is the reason why we pay a lot for assets that covary negatively with our consumption.
29. J. Cochrane (2001), Asset Pricing, chapter 1 (Princeton University Press).
30. For a discussion of this separation between "brawn and brains,"
31. A. Shleifer (2000), Inefficient Markets, chapters 1 and 4 (in particular), Clarendon Lectures in Economics (Oxford University Press).
32. It is difficult, but not impossible. Very smart traders, especially using options, can sometimes put together strategies that display these positive features. As usual, they are likely to destroy the opportunities just by exploiting them. Bad for the traders, but good for the market, which has become a bit more efficient.
33. We are neglecting in this discussion the positive expected drift from investing in equities. Over a short horizon this matters little.
34. Some large fund managers who offer their services to the general public currently only charge (when all expenses are included) as little as one-third of a percentage point per annum for the service. If you happen to be an institutional investor you will be able to negotiate even better terms.
35. Even if you are a hardcore efficient marketeer, market efficiency does not come about by divine intervention. Prices can only incorporate all the publicly available information about the various stocks because someone has done all the necessary homework. The process by means of which new information becomes translated into prices may be fast very, very fast, even-but it cannot be infinitely fast. It is almost certainly far too rapid for a day trader sitting at home with his laptop computer to exploit. But an army of professional, sophisticated managers equipped with the latest state-of-the-art software and hardware can be just a bit ahead of the pack. Indeed, it is just by their actions that market can remain efficient.
36. Explanations along these lines belong to the "limit-to-arbitrage" school of thought.
37. Traders become very averse to losses because they can be "stopped out." Different traders whose losses are reckoned from the beginning of the financial year or on a twelve-month rolling basis are observed to display different risk-taking behavior.
38. Perhaps this is because financial firms are managed not by the ultimate shareholders (who may well care about their consumption pattern), but by their agents (the board, the CEO, the CFO, etc.) and it is much easier to incentivize these agents in terms of their profits and losses. The ultimate owners of a firm may well therefore be motivated in their choices by the absolute level of their wealth, but in designing the incentives for their agents, they end up making choices on the basis of profits or losses.
39. We should distinguish here between diversifiable and undiversifiable risk.
40. For a discussion of this point
41. C. L. Culp (2003), The Modigliani-Miller propositions, in Modern Risk Management: A History, chapter 6 (London: Risk Books). My statement and the ensuing discussion refer to undiversifiable risk.
42. VaR, for instance, lacks sub-additivity (i.e., I can increase the total VaR of two portfolios by putting them together, contrary to intuitive notions of diversification and risk). This problem can be fixed by using a germane measure (conditional expected shortfall): roughly speaking, this is an average of the losses past the desired percentile.
43. Since the historical-simulation method appears to make use of no free parameters, it is often referred to as a nonparametric approach. Yet an extremely important parameter is used in the historical-simulation method, i.e., the length of the statistical record (that is, the length of the so-called "statistical window") to be used in the analysis. This implicitly tells us what constitutes the relevant past and does so in a binary manner (fully relevant-and included in the data set-or totally irrelevant- outside the boundaries of our data set).
44. "Financial crashes are 'outliers' " is indeed the title of chapter 3 of Sornette's bookWhyStock Markets Crash, devoted to the study of extreme events in a variety of financial markets. An "outlier" in this context is an occurrence that does not belong to the population of normal events. It is one of the "Errors of Nature, Sports and Monsters" referred to in the quote that opens this section.
45. I am making the assumption here that the nature of the phenomenon is timescale invariant. So, having monthly data about corporate defaults is better, from a frequentist point of view, than having yearly data. This is because the underlying phenomenon does not change as a function of our sampling interval. The same may not be true, however, for tic-by-tic price moves and yearly price moves.
46. See the discussion in chapter 7.
47. Establishing this correspondence between market price and fundamental value is essential if one wants to argue that securities held for the long term (say, in an investment portfolio) should also be markedto- market daily. This observation may seem of rather narrow interest, but it can have a major impact on the risk regulation imposed on pension fund managers. Since this is a multitrillion-dollar industry, even mundane accountancy niceties can have seismic effects.
48. D. Poirier (1995), Intermediate Statistics and Econometrics (Cambridge, MA: MIT Press), reaches strikingly similar conclusions as he discusses the use of past data in assessing the probability of catastrophic failure before the fateful launch of the Challenger space shuttle in 1986. In his words, "we note that because there had been previously no catastrophic failures, it is difficult to think of an empirically based relative frequency interpretation of this concept. Rather, a 'degree of belief' interpretation seems necessary."
49. Before we get started, we must keep in mind that the economiccapital view of the world presupposes that it makes sense for the firm (i.e., in our context, the bank) to engage in diversification. This means that the bank managers should not consider investments in isolation, but should look at how they fit into the overall portfolio. Diversification is not left wholly to the investor. If this seems too obvious even to state, let us remind ourselves that this is not how finance theory views diversification. If you have never lost any sleep over this, you can read on with abandon, innocent of any doubts that the idea of engaging in diversification may be conceptually flawed-personally, I do not think it is, for the reasons discussed in chapter 5.
50. For the sake of honesty, though, I thought I should place the disclaimer right at the beginning of the discussion.
51. Raising and holding equity capital is more costly due to a combination of taxation rules and agency and information costs.
52. K. Froot and J. Stein (1996), Risk management, capital budgeting and capital structure, National Bureau of Economic ResearchWorking Paper 5403 (Cambridge, MA), or R. C. Merton andA. F. Perold (1993), The theory of risk capital in financial firms, Journal of Applied Corporate Finance 5:16-32.
53. In a similar vein, readers familiar with asset allocation will know very well that Markowitz portfolios (which assume perfect knowledge of probabilities and correlations) can be radically, not marginally, different from Black-Litterman portfolios (which assume more diffuse beliefs about the same quantities).
54. A. Greenspan (1998), The Role of Capital in Optimal Banking Supervision and Regulation, FRBNY Economic Policy Review (October 1998).
55. Noone has stated thismore clearly than Culp and Miller,whoentitled the introductory article of a book they edited: "Why a firm hedges affects how a firm hedges" (their italics).
56. C. L. Culp and M. H. Miller (1999), Introduction: why a firm hedges affects how a firm hedges, in Corporate Hedging in Theory and Practice: Lessons from Metallgeschaft (ed. C. L. Culp and M. H. Miller), pp. xix-xxiii (London: Risk Books).
57. Indeed, we have seen in chapter 5 that, according to academia, the reason why private-firm risk managers should continue to draw a salary is not self-evident at all. By enriching the Finance 101 view about risk management within firms ("there should be none"), we did find a role for judicious control of risk in a private bank. Ultimately,we argued, risk management plays an important role in allowing the shareholders (and the bondholders) to discern more clearly, through the fog of financial noise, information about the bank itself.
58. We will consider profits and losses, despite the theoretical problems already discussed in chapter 4, simply because we would not even know how to begin to think in terms of "total consumption" (or its proxies) when it comes to the practical decisions the agents who run a financial institution have to make.
59. B. Cornell (1999), The Equity Risk Premium (Chichester: John Wiley & Sons), or E. Dimson, P. Marsh, and M. Staunton (2002), Triumph of the Optimists (Princeton University Press).
60. It may seem that this is where risk management and asset allocation differ most. Actually, this is not the case. There is a strong similarity between the "grafting" procedure just described and the Black- Litterman approach, which explicitly solicits from the investor his personal (nondata) views about returns, and directly incorporates them into the investment decision.
61. I do not want to give the impression that experimental financial psychology is virgin territory. Far from it. What is mainly lacking is cross-disciplinary fertilization and a link between the useful research that does exist and risk-management practitioners (and regulators).