Customer Reviews

An Engine, Not a Camera: How Financial Models Shape Markets (Inside Technology)

5 star:		(5)
4 star:		(1)
3 star:		(0)
2 star:		(0)
1 star:		(0)

 

 

Many financial analysts and financial journalists have pointed to quantitative trading and the subprime mortgage markets as being the major cause behind the extreme volatility in the financial markets in the summer of 2007. This book therefore seems fitting for this particular time in financial history, if only at a bare minimum to educate the reader about the use of mathematical modeling in financial analysis and financial engineering. As the subtitle of the book indicates, the author's main thesis is that the use of mathematical models can actually change the dynamics of the markets themselves, moving them possibly to territories even more uncertain that they were invented to describe. Quantitative trading, now done by most of the major players in the financial markets, is dependent of course on mathematical modeling, some of which uses highly sophisticated reasoning patterns and artificial intelligence. Most of these models are proprietary, and therefore one cannot ascertain their efficacy in the acquisition of wealth for the organizations that deploy them. However, with a little pertinacity one can acquire a good understanding of their workings by studying the academic literature.

Some of the predominant models in the public domain are discussed in this book, mostly from an historical perspective but the author inserts some of the relevant mathematics in its appendices for the more mathematically sophisticated reader. In general the author makes a plausible case for his main thesis, but at times his conclusions are based on mere anecdotes, and he makes the typical mistake of imputing power and influence to individuals that is unsubstantiated. It is very tempting, especially among those individuals or institutions that are involved in trading, or even responsible for innovations in the same, to believe that they are the cause for some of volatility in the financial markets. But such claims, even if they seem reasonable or intuitively clear, must be substantiated with careful statistical analysis, which can be time-consuming and difficult, and few individuals it seems are willing to devote themselves to such a project. The author though is aware of this, for he states very early on in the book that historical sources may not be sufficient to allow one to decide if the influences are real. In addition, he cautions the reader to "look not just at what participants say and write but also at whether the processes in question involve procedures and material devices that incorporate economics."

The author labels the idea that economics as an academic project is actually part of economic processes the `performativity of economics', which he further breaks down into subclasses that serve to clarify the distinctions he wishes to make. One of these is more of a passive notion, called "generic" performativity, which is used to describe the participant's use of economic theories or data without emphasizing their effects on economic processes. If such effects take place, this is called "effective" performativity, which is then specialized to "Barnesian" performativity. The latter is used to describe the situations where the practical use of economic theory makes economic processes resemble what they are described to be by economic theory. Barnesian performativity is to be contrasted with `counterperformativity' where the actual use of economic models makes economic processes not resemble their description by these models. The author discusses how to detect Barnesian performativity, but warns of the difficulty in proving that movements in prices are following certain model predictions.

But aside from the qualitative/historical emphasis that the author makes in this book and the small number of unsubstantiated claims of model-market influence, the reader will take away a better understanding of such topics as the capital asset pricing model, the Black-Scholes-Merton model of option pricing, the Modigliani-Miller theory of capital structure, a description of Levy processes and their role in econometrics, and most interestingly, a different explanation for the demise of Long Term Capital Management. All of these topics, coupled with the intellectual honesty and literary skill of the author, make this book a highly interesting contribution to the financial literature.

Both the science and the art and practice of finance have experienced phenomenal growth since the 1970s.

As a science, finance has evolved from a descriptive outpost on the economic frontiers to become of that discipline's central topics. During the same period, the financial markets changed from what often seems today like sleepy outposts of liquidity into dynamic centers for financial engineering. In the 1970s, the world was being introduced to commodity hedging and options trading. By the early part of the 21st century, derivatives contracts totaling more than $273 trillion were outstanding worldwide.

Donald MacKenzie, a sociology professor at the University of Edinburgh, argues in An Engine, Not a Camera, the trends are connected. Paraphrasing Milton Friedman, he argues the emergence of economic models were an engine of inquiry rather than a camera to reproduce empirical facts. As the science of finance became authoritative, the markets were altered. These new, Nobel Prize-winning theories, elegant mathematical markets models, were more than external analyses. They evolved into intrinsic parts of the financial process.

Beginning with a discussion of the work of Franco Modigliani and Merton Miller, the Capital Asset Pricing Model and Random Walk, MacKenzie takes the reader on a journey through the development of the Black-Scholes-Merton model, The Crash of 1987, Long-Term Capital Management and the Russian government's default in 1998 to bind the threads of his thesis.

Detailed, astute, well-written, and with much of the technical detail relegated to the appendices, this book weaves economics, financial theory, economic sociology and science and technology studies into an essential read for anyone with a serious interest in the financial markets.

By far the least interesting part of this fascinating book, it seems to me, is its ostensible purpose. The title more or less says it all: financial models don't merely describe the world around them; they play an active part in shaping that world. The shorthand for this concept is "performativity": MacKenzie wants to argue that financial models are "performative." (Ungainly as that word is, MacKenzie steps to the next level when he gives us "counterperformativity.")

In the economic world outside of finance, this is straightforward enough. If you spend enough time imagining that humans are rational actors who only look out for their own self-interest, for instance, and the theorems furthermore say that you should be a rational actor who only looks out for his own self-interest, you will -- suprise of surprises -- probably find yourself acting purely self-interested.

This is easy enough as an intuitive proposition. To really argue it, you'd need some data. The data would need to show that the model somehow creates the greed, or that people become greedy faster than they otherwise would have. You'd then be required to show that the causality doesn't run in the other direction: it's not that greedy people are just drawn to this model, but that they become greedy. Something like this.

Perhaps my tone suggests what I think about this, namely that it doesn't interest me very much. And so it goes for the bits -- and thankfully they are just bits -- about performativity in An Engine, Not A Camera. The finance markets are a good place to look for data: there's a lot of extremely high-quality stock-price data, at one-day resolution or lower, for the last few decades. And there happen to be a couple models -- the Capital Asset Pricing Model and the Black-Scholes option-pricing formula -- that are in wide use. Conditions couldn't be more ripe for testing some hypotheses.

One fundamental concept sitting underneath Black-Scholes is that opportunities for riskless profit don't persist; that is, there are no arbitrage opportunities. In order to determine the rational price for a stock option, we construct a riskless portfolio on the basis of that option. Since it's riskless, and since there are no arbitrage opportunities, our riskless portfolio must fetch the same price as any other riskless asset. (Treasury bonds with the same duration as the option are often taken to be riskless assets.)

If we want to show that financial models have a hand in shaping the financial world, we could show that Black-Scholes actually helped to destroy arbitrage opportunities. This is a straightforward enough exercise; it's an exercise that plays a central role in the fall of Long-Term Capital Management. There the story was pretty clear-cut: a hedge fund (and its imitators) sought and destroyed arbitrage opportunities so successfully that there were none left to find. In a perfectly efficient market, Long-Term Capital is broke. And so it was.

In that sense, financial models are performative. But there's also (wait for it) "counterperformativity," in which the model is self-negating. Index funds are a good example here. Index funds are an idea that basically sprang out of the Capital Asset Pricing Model, in which the market's own volatility is the baseline against which stock volatilities are measured (this measure of volatility is known as "beta"). Whenever a stock gets added to the S&P 500, that stock's price rises more than it rationally should. The reason is that index funds are required to buy that stock automatically now that it's part of the index. So index funds are a built-in source of arbitrage that wouldn't have existed without CAPM. (One might wonder why arbitrageurs, knowing this ahead of time, wouldn't swoop in and bid away the irrationality. I don't know the answer offhand, but I suspect that index funds' sheer size is part of the answer: stock purchases that large need larger short sales than arbitrageurs are prepared to conduct. Plus there are more restraints on short sales than on purchases.)

Like I said, I don't think much of this is very interesting. To me, MacKenzie's book functions much better as a history and semi-technical explanation of modern finance. Read it as history, rather than as sociology or philosophy.

Among other fun technical details in MacKenzie's book, I'd single out the Cox-Rubinstein model. It's a simplified model of option pricing, which leads to the Black-Scholes model in the limit as prices are allowed to change continuously. And it's an elegant piece of work.

The natural next steps after An Engine, Not A Camera are more formal works on option pricing and stochastic processes. We'd want to learn about martingales, Brownian motion, stochastic differential equations; and generalizations of Black-Scholes that allow for credit constraints, legal limitations on short-selling, discontinuous price changes, corporate dividends, and heavier-tailed price distributions. I'm going to be diving into Grimmett and Stirzaker in the near future.

MacKenzie has written a wonderful and brilliant book on a very complex set of issues; the development of mathematical finance theory. He has managed to take some very complex ideas and place them in highly readable and understandable ways and in addition places you in the midst of the process and the people as it has evolved over the past fifty years.

He starts with a discussion of the commodity exchanges. The point he makes is that the exchanges could exist only with the railroad and standards, inspected wheat and corn, so that one no longer worried about a specific bushel of wheat or corn from a specific farmer, but only about the property rights to a bushel, no matter where it came from. Once that existed the whole process was off and running.

He then takes the reader through the evolution of the CAPM model with Modigliani and Miller. He adds a bit of the Samuelson saga and the growing difference between MIT and Chicago schools. Having been on the edge of that set of discussions he does a splendid job.

These observations of the schools and personalities should be exceptionally germane to the current environment which is dominated by the Harvard types who filtered through MIT. After all, Larry Summers was MIT undergrad and Harvard grad. Bernake is MIT grad. The author lends great credibility and human feeling for these characters while ensuring the understanding of the new principles developed.

He then proceeds through stock options and the Black-Scholes work, and then takes you into the pits of traders and then through the various falls that occurred over the years.

This is not a book for understanding finance. It was not intended for that. This is, however, a must book for understanding the mind-set which got us to where we are now.

There were tons of good intentions but a lack of real world reality. There is a clear understanding that the models be accepted not that they predict the actual. Like economists generally, these finance type economists develop their models in a manner which is then antithesis of engineers.

Engineers are all too often introduced to their trade by being shown examples of failure. Most engineers remember the film of the collapse of the Tacoma Narrows Bridge due to inherent instabilities in its design. Then they are told that they should never allow that to happen again. Thus, engineers over design and always look for failure modes.

These financial economists, as demonstrated by the author, are driven by concepts and theories and are devoid of any reality. Thus, the never ending collapses are inevitable with such mind sets. This book is a great lesson for the future as well as the present.

MacKenzie has an interesting take on the development of financial models, technology and organized exchanges, focusing on the latter half of the 20th century. He sees the three components as interacting as markets evolve. For example, the index futures exchanges did not really take off until options pricing theory made it possible to create spreadsheets to assist traders in options pricing. The incorporation of computers greatly increased market efficiency.

MacKenzie analyzes the development of modern finance theory and its interplay with market evolution from a social-scientific, anthropological point of view. The people involved in the development of theory and markets take center stage; the author conducted dozens of interviews with academics and practitioners, and even reviewed the private papers of some. The amount of research incorporated into the analysis is impressive.

One word of caution: although the book does not contain much math, and what is included is relegated to appendices, a strong familiarity with the development of financial models, at the level, say, of Bernstein's "Capital Ideas", would be greatly beneficial to one's enjoyment of this book.

Summary: If you are interested in this topic -- read this book! You won't be disappointed.

I am going to concentrate my review on those chapters of the book that deal with the comparison -contrast done by the author between the financial models/theories created at the University of Chicago's Department of Economics and Booth School of Business- the Capital Asset Pricing Model,Black-Scholes Equation ,Efficient Market Hypothesis and the Rational Expectationist hypothesis,all based on the Normal distribution or log normal distribution-and Benoit Mandelbrot's work ,that has now been replicated by thousands of Econophysicists worldwide in all financial markets,based on discontinuities,jumps in the data,nonsmoothness,power laws and the Cauchy Distribution's fat tails and extreme kurtosis.
The author comes to the conclusion that Mandelbrot's approach was rejected by economists because his technical analysis was difficult to handle and use ,did not give clear cut,precise,exact answers and was hard to interpret. Mandelbrot's scientific work was rejected because it was not analytically tractable to the economics profession.

The problem with this answer is that all goodness of fit tests and exploratory data analysis shows overwhelmingly that Mandelbrot is correct and the University of Chicago's Fama ,Stigler,Friedman,Cochrane ,Lucas,Markowitz,Merton,Miller,Scholes,etc. ,are wrong.Furthermore,they have been proven wrong since at least 1963,which is the year that Mandelbrot demonstrated empirical evidence that overwhelmingly demonstrated that the normal or log normal distributions fell a long way short of close to being able to repesent the time series data on price changes in specific commodity and financial markets.

The author does not take the final step that he needed to take given the evidence .Economics is not a science or art because massive empirical and statistical evidence counts for nothing in this field.Economics can be viewed as a Ptolomaic and/or Procrustian anti -scientific approach based on a priori claims and ideological presuppositions.Tournament chess palyers are more scientific.Economics is a field very similar to Ptolomaic astronomy.Ptolomaic astronomers engaged in the creation of artificially constructed models based on ideological and a priori claims to knowledge about the real world that have been falsified by empirical work many times over.

I highly recommend this book even though the author does not take the last step he needed to, given the evidence discussed .