A challenge to the conventional wisdom surrounding financial risk, providing insight into why easy solutions to control the financial system are doomed to fail
Finance plays a key role in the prosperity of the modern world, but it also brings grave dangers. We seek to manage those threats with a vast array of sophisticated mathematical tools and techniques of financial risk management. Too often, though, we fail to address the greatest risk—the peril posed by our own behavior.
Jón Daníelsson argues that critical risk is generated from within, through the interactions of individuals and perpetuated by their beliefs, objectives, abilities, and prejudices. He asserts that the widespread belief that risk originates outside the financial system frustrates our ability to measure and manage it, and the likely consequences of new regulations will help alleviate small-scale risks but, perversely, encourage excessive risk taking. Daníelsson uses lessons from past and recent crises to show that diversity is the best way to safeguard our financial system.
The Illusion of Control covers all the standard concepts linked to risk management. While the author props up to shoot down many a strawman to create the feeling that he is making some radical new points, there is almost nothing radically original. Readers new to the topic might find the book a good introductory guide, but it is unlikely to provide new insights to those familiar.
Awareness of risk management tools or their importance does not make anyone less vulnerable in the financial world. Adherence requires enormous discipline and continuous training. Many professionals could use the book as a refresher despite familiarity with the topics covered. The book's discussions might just prompt some to re-check something long forgotten, making a quick read worthwhile.
The author's arguments are straightforward. Criticisms are largely reserved for nameless policymakers - keeping the content even less controversial. The book's excessive focus on risks measured in the form of liquid securities' price movements makes it miss many other measures of risks, though. The book barely mentions the importance of balance sheet analysis, rating agency models, evaluation of qualitative factors like management quality, liquidity and opaqueness of structures involved, the keyman risk, and a number of other tools employed by loan officers, investors in private markets, or even those in public markets.
The author's arguments against artificial intelligence (for whatever it means) expose the book's underbelly - ie, its inability to discuss anything recent and forward-looking. The reality of risk sciences, more than anything else in practical life, is that what is measured and managed is a function of the available tools. As tools evolve, so do methods and practices. Discussing risk without discussing the latest in data sciences or criticizing "artificial intelligence" with vague generalities is equivalent to discussing modern physics with Einstein's discoveries and suggesting sciences stop there.
The author's recommendations are often towards models that are too static (with the changing inputs) and increasingly archaic. From calculators through spreadsheets, applications providing ever more complex statistical results, and programmed databases through to machine learning algorithms now, Riskometers are constantly changing. The roles of data and processing power continue to rise, while the input of human decision-making - still critical - continues to evolve (and need one daresay, diminish). The new-age risk models - which too will surely fail - even at the most unsophisticated organizations use analysis far more complex than the ARCH, value-at-risk, etc. invented decades ago. The book's understanding of what artificial intelligence means and how it is changing is plain wrong and self-serving (to the criticisms offered).
The end sections have other fears, which are also made up. For example, the author worries needlessly about everyone using the same risk models in a couple of contexts without realizing the real-life participants' needs to do better risk management than their peers as a competitive tool. While risk models are never perfect - as the author shows, this is theoretically given with reality in its infinite potential always shapeshifting to fill the corners left vacant by limited risk models - few prudent actors stop at implementing the bare minimum suggested by policy directives. Of course, it neither means policymakers have no roles to play nor implies an absence of nefarious players ignoring the most elementary guidelines for ill-gotten gains. The main point the book misses is that risk management is more in the intent and details, as it needs to use as much data/processing power as possible while also having the best human oversight.
A thoughtful work on the limits of risk modeling and how regulators ought to pursue macroprudential policy. Heavily inspired by Keynesian/Knighian conceptions of risk. Some poorly-informed comments on AI driven less by fundamental computational/mathematical understandings of the topic and more by pop presentations of the topic.
Repeats the same 3 points over and over. The author is clearly an expert in (macro) regulations and the politics around it, so for someone new to the field the first third of the book is actually a very decent introduction., but that doesn't excuse the unnecessary repetitiveness. Also, some of the points on algorithmic trading, risk management for Quant funds and the 2007 quant bust are just incorrect, or at least dumbed down to the point of being incorrect.
Very opinionated with strong sourcing and theory behind it. Even after working on finance regulation for quite some time I honestly had to pause and research up some of the concepts and mainly abbreviations used and it is not the most approachable book.
However, it did help me in a major way! It provided a good founding for some of my feelings (nice), it provided very good sources to understanding crisis and most importantly it did change some of my opinions on financial regulation. If you are interested in the topic, highly recommended. It is of course very much centered on the high strategic level of risk and crisis, not geared towards for example some more specific parts (alternative investment funds) and some risks closer to security/ resilience. That actually is a great recommendation - and anybody who quotes Schneier has my heart.
The only part which was a bit wild was the AI regulation part. Yes, many problems with that, probably not a way forward, but the level of critique was not really up to the challenge.