Financial markets trade tokens of rights and obligations, rather than goods or services but, in principle, the nominal world of money and credit mirrors the real world of houses and haircuts. Aristotle through Marx has told us why money ought to remain a means and not an end in itself, but no one can dispute the longstanding efficiencies and wealth that have been induced by the trading of financial instruments. In seventeenth-century Amsterdam, one could find markets for stocks, foreign currencies, bills of exchange, bank notes, annuities and futures. The last was known as windhandel or ‘trading in the wind’ (see De Marchi and Harrison Reference De Marchi, Harrison, De Marchi and Morgan1994). But this was only a faint shadow of what has come to pass over the last 50 years. The derivatives market that officially emerged in the 1970s and is now well over one hundred trillion dollars in size, suggests a world onto itself. The per capita value of derivatives for every person on our planet in 2006 was about $26,000, a value that clearly exceeds the average per capita income worldwide (p. 80). Many of the contracts offset the risks of other derivatives, so an unspecified portion and perhaps a majority are self-cancelling, but that plays all the more into the sense in which derivatives are about a virtual rather than a material world.

Donald MacKenzie, a leading sociologist of science, is already well known for his study of the efficient market hypothesis. One of his co-authors for two of the studies, Iain Hardie, was an investment banker before returning to academia, and is thus able to fill in some of the tacit knowledge that traders use. This is a short book, and some of the chapters are drawn from already-published articles, but sufficient work has been done to ensure that the book flows very well. Each chapter is chock full of observations and insights; even a selective reading will yield benefits. Chapters Four and Five, on derivatives and arbitrage respectively, are particularly recommended, both for their descriptive and implicit normative content. There are also interesting studies of hedge funds, corporate accounting and the market for carbon emissions, but I will only touch on these briefly.

There are three main assertions promoted throughout the book. One is that however virtual financial markets might seem, there are significant sociological, material and cognitive constraints that govern them. Another is that the concept of an economic agent needs to be rethought, particularly in light of the increased use of sophisticated algorithms and computing technology. And the third is that the science of economics ‘does things’ by, for example, selecting the parameters used to bundle risk in the derivatives market or assessing how to price carbon emissions. In this respect, it is part of the widely accepted view that scientific research has technological outcomes and thus that economists are not passive theorists of markets but agents in their evolution. These claims in turn have important implications for the core assumptions about rationality and about the efficiency of markets. The latter are only flicked out along the way rather than developed by the author, but because they emerge from substantial empirical research, are all the more intriguing.

Donald MacKenzie's text is based on a careful reading of the trade journals, the relevant scholarly literature in economics, on a substantial number of interviews with financiers, but above all, on extended embedding in the financial world, watching silently as accountants, traders and information processors go about their business. Kudos goes to him and his co-authors for gaining access to this world, one that is presumably more like a guild than not. In his conclusion, he admits that ‘getting permission to conduct ethnographic observation in financial markets can be extraordinarily hard’ (p. 179). In one study here, his co-authors were allowed to be participant-observers in a Lower Manhattan arbitrage trading room for 65 half-day visits over the course of almost three years. In another, to be related below, MacKenzie was able to sit in control central for the UK derivatives world, the unassuming office in the London Docklands where LIBOR (London Interbank Offered Rate), one of the ‘world's most consequential set of numbers’, is produced every business day.

If the reader expects that this book will read like a piece of journalism, let me emphasize that discretion governs every word. There is little that strays from the purely descriptive, other than to attend to the increased use of techniques and technology as financial markets grow and diversify. No doubt Professor MacKenzie is keen to honour whatever original agreements allowed him into the secret chambers in the first place, and to sustain the trust of the traders he studies so as to ensure access in the future. At times, he offers quotes from a particular (usually unnamed) broker, or provides the dialogue that he taped to show the range of judgements required. Sometimes he generalizes about the group as a whole, but he systematically dodges the ethical dimensions of his subject.

Let us take his case-study of the formation of LIBOR, the daily rate on which are indexed, as of 2006, over 84 trillion dollars of derivatives. He was the first to be an embedded observer. Those who undertake the calculations are paid employees, not traders, and they are merely responding to the judgement calls of a panel of 16 banks (most with a global reach) who report the 11 o'clock interest rate they have reason to believe other banks would charge for issuing them a specific loan (quantity, currency and temporal period are proscribed as part of the exercise). The highest and lowest rates are trimmed to diminish self-interested tugs on the rate, and an average is taken. Within the 30-minute window, the calculators follow up by email or telephone on any anomalies that catch their trained eyes, before doing the final computation and publishing the daily rate at about 11:45 every weekday morning. By 2006, some 400 000 people in the UK could receive the announcement of LIBOR on their mobile phones and buy or sell derivatives in a span of ten seconds. The largest derivatives market then was the ‘swaps market’, whereby one party paid a fixed rate for the contract while the other party paid the floating rate provided by LIBOR. Whoever bought low and sold high, albeit with a miniscule margin, made money.

While the calculators in the room that produced LIBOR may be unbiased, they are also under-informed since they cannot observe the trades and deliberations that lead up to the rates they receive. Those computing LIBOR ‘would lack the detailed knowledge needed to exercise informed judgement, so there is no practical alternative to reliance on those whose involvement means they may have ‘interests’ in the outcome’ (p. 81). As MacKenzie emphasizes, they are essentially beholden to the interbank loan market, which he describes as similar to a bank's trading room, but ‘more crowded, noisier, and more raucous’ (p. 80). Brokers assessing the spread of interest rates must also make informed judgements about the different credit ratings of banks; the sequencing of trades every morning, from the more aggressive to the more cautious, is also significant to the outcome. Although not stated explicitly, MacKenzie creates the space by which any reader could see the possibility for collusion. Some of the traders he interviewed speculated that LIBOR may be subject to manipulation by the leading banks, a phenomenon known as ‘herding’ such that each bank keeps its reported rate close to the others and thus injects a downward or upward bias in the rate. Large banks, after all, have substantial derivative funds indexed to LIBOR and thus have a vested interest. Even a shift by one basis point (one-one-hundredth of a per cent) can be viewed as suspicious. This is about as far out on a limb as MacKenzie is willing to go when it comes to questionable practices.

MacKenzie's account of present-day arbitrage seeks to emphasize the material and sociological features and the more obvious concentration of power. In a world of perfect information and instantaneous transmission of market data, there would be no arbitrage. Because of the imperfections, however, arbitrageurs induce market efficiencies and serve to anchor derivatives (the sale and purchase of the risk of assets) to the underlying assets. Arbitrage is as old as the hills, but the focus has shifted away from the purchase and sale of the same commodity geographically dispersed toward the near-simultaneous purchase and sale of similar rather than identical assets, such as bonds that expire in 2014 or 2015, or the prices of index futures and the price of the stocks making up the index. Even though the open-pit process of stock trading is virtually extinct, arbitrageurs and traders need to be close to the capital markets because it takes time for the information to flow and profits can hinge on a matter of microseconds. Electronic price transmission has changed the speed of arbitrage, which once relied on the telegraph or telephone or certain hand signals (‘the arb’) in the open pit. Much of it is now carried out by computers programmed to find the price discrepancies. But a glitch, or minor delay, even of a couple of seconds, can wreak financial havoc (see p. 93). The decline of face-to-face trading has dispersed the human bodies, but there is a recentralization governed by the technological systems. Centralization has also fostered mimicking whereby minor traders arbitrage around a single prominent arbitrageur, thus distorting the results. This has become more prevalent with the advent of hedge funds, essentially large off-shore investments that are less bound by the regulations of the key financial nations.

If there is one principle to which all economists assent, it is the principle of the gains from voluntary trade. But what is special about the options and derivative markets is that one party always stands to lose. When the practice of windhandel became prevalent, pious authorities raised objections, not just because it resembled gambling, but because it encouraged a vicious desire to bring ill upon another Christian (see De Marchi and Harrison Reference De Marchi, Harrison, De Marchi and Morgan1994). These contracts have not changed in essence; ‘a derivative can indeed seem to resemble a bet on the movement of the price of the underlying asset’ (p. 75). Increasingly, it is for entities that cannot be delivered in any straightforward sense, such as a stock index, and thus must be settled in cash. Nevertheless, one cannot have a legal market for pure speculation. Short selling requires an underlying asset, and that asset must have a minimum of non-hedging commercial activity of about 20–25 per cent of its value.

MacKenzie notes that the insight that one could separate hedging and speculation from genuine trading first gained momentum after Bretton Woods, specifically in the market for foreign currencies. The metrics used for derivatives, the choice of temporal and spatial parameters that render derivatives into a standardized product, for example the category cheapest-to-deliver bond-derivative, are critical to how the market proceeds. Academic economists played a significant role in creating these markets, but for MacKenzie, the jury is still out. ‘Successful choice of the specifications of derivatives contracts involves careful attention to sometimes conflicting interests . . . These interests are neither easy to determine [nor fixed] – extensive research often seems to be necessary to elicit them, giving contract design something of the flavour of economic experimentation’ (p. 70). He points to the recent advent of derivatives on housing indexes, pension indexes and even the weather. It occurred to me while writing this that there ought to be a derivatives market for visual art (paintings and sculptures) since the risk of the investment could be disassociated from the asset. A quick Google search (for what that is worth) elicited a working paper of January 2010 by two academics advocating the construction of a market with call options on an art index.

MacKenzie reviews the history of British legislation to show in what sense derivatives elide the accusation of pure gambling. In the eighteenth century, the law stipulated that an enforceable contract must pertain to something of material value or title to that value. By the nineteenth century, the law relaxed its material import and drew a distinction around the intent of the trading parties to deliver the asset, even if delivery did not take place. This became the acceptable practice for stock options and even futures on grain were often settled by cash rather than the delivery of the grain as long as the intent was manifest. It took substantial lobbying in the 1960s and 1970s to gain legal permission, in 1982, for treating a stock index as a comparable asset for such contracts. But in 1986, Britain removed all barriers to derivatives settled in cash, even if they failed the intent test and thus constituted de facto wagers. With a gesture to the concept of rational expectations, customers were to deposit the amount that covers the variance of the index, since their ‘spread bet’ would take a position similar to that taken by the aggregate of customers and thus minimize exposure to large market swings. Many bought derivatives under the outdated assumption that gambling debts were not enforceable, but a stock market crash of 1987 left them with liabilities exceeding their deposits.

The recognition that markets are global and dominated by access to information, including the speed of access, is not new. Spyglasses were critical to Venetian traders before Galileo spotted the moons of Jupiter. Telegraph cables, first laid in 1851, enabled the flow of information to redefine markets. Not surprisingly, the law of one price was formulated around this time, a law that is committed to the ubiquitous activities of arbitrageurs. One of the more obscure laws in the annals of economics was proposed by a Danish engineer, C.L. Madsen in 1876 and circulated widely in Europe for the next few years. Madsen's Law submits that there is a direct proportion between the number of telegraphs (T) and level of trade (U), such that T = c/d (U) where c is a constant and d the distance between the traders. It is a trivial law and did not endure, but it captures well the line of thought that motivates this book, namely that markets are materially constrained by the flow of information and that the distance between them is still a factor.

Traders need skills and rely on judgements formed over time, but they are much more the products of their material setting than is obvious. MacKenzie reviews some of the physical features of the now extinct open pit markets of the stock exchange or futures markets. Then it was important to have height (perhaps wearing platform shoes), and to be skilled at picking up on the fears expressed in the faces of others, and masking one's own. Now that most trades are done sitting in front of a computer, MacKenzie offers insights about the new material and physiological constraints. For one, traders rarely work alone, and usually face one another, sometimes in groups of six to eight at a large desk. These close quarters are important to the outcome, as traders learn to listen to the voices of their colleagues and digest facial cues while simultaneously interacting with clients from afar. MacKenzie notes how important it is that the voice technology allows for rapid silencing to prevent outside eavesdroppers. Apprentice traders must learn the habit of disengaging the outside world so that it becomes second nature, especially under moments of extreme stress when they are likely to forget. What is clear is that by working in shared spaces, traders detect collective trends from those sitting in the room as the day's traffic unfolds. MacKenzie relates typical dialogues that traders share, reacting to the day's events, the weather, political upheavals and, of course, key indicators such as rates of unemployment or inflation. The implication here, though not articulated, is that by working in groups, often in hot and crowded offices because of the high real estate costs, bandwagon effects are all the more likely. The very effort to become well-informed has led not to a diminution of the amplitudes of business cycles but to their augmentation.

One interesting philosophical question raised here pertains to human agency. In what sense are the movements of the interest rate truly about collective intentions and in what sense are those who set the daily rate (LIBOR) merely responding to forces far greater than their own sphere? To put it another way, where does one locate market forces given the complex socio-technical structure of financial trades? Goodhart's law suggests that the rate has become all the more detached from capital markets and is in effect an artificial construct of the group of proximate agents. MacKenzie is hard to pin down, but while there is room for human judgement (and error), the sense one has is that the collective results, aided and abetted by the new technology, decrease overall market efficiency.

MacKenzie also adopts the concept of agencement devised by sociologist Michel Callon. The point is to dilute the notion of individual actors and recognize the importance of the material and cognitive context, the use of machines and algorithms, the configuration of the workspace not to mention the social codes that forge client fidelity. Financial markets are made up of a chain of human and machine actions, what is known as ‘distributed tasks’ and as a result, there is ‘distributed cognition’. As a product of a group, the salient facts have a different standing from those forged individually.

A trader armed with the Black–Scholes model in an options market is a different agent, submits MacKenzie. However, this then tends to put undue weight on the featured metric of implied volatility, and as a result price movements are often distorted by this. MacKenzie considers the possibility that the use of such theoretical tools has changed the market, made it more rational and hence increased its liquidity. He suggests that there is a corresponding story to tell in the bond market, where yield measurements dominate. And when it comes to derivatives, the metric of base correlation counts most. But MacKenzie also voices scepticism about these tools. Most traders have no understanding of the theoretical foundations, and those in the credit-derivatives world, for example, have come to believe that the single-factor Gaussian copula model, while widely used, is inadequate. There are more complex models that add parameters, but unless both seller and buyer are using the same model, there is unlikely to be full communication. Old-fashioned Chartism is still prevalent (looking for trends using graphs of price movements) and because enough traders take it seriously (akin to astrology in the eyes of financial economists), other sceptics need to incorporate it insofar as it affects price movements.

Although most of the book was written before the recent financial crisis, the closing chapter reflects openly on the possibility that the new kinds of financial markets failed for some of the reasons MacKenzie detected in his sociological investigations. As the banks discovered at the time of the crisis, the LIBOR rates had become something of a fiction; banks held bonds and other liabilities that could only be sold at distressed prices. ‘The credit crisis – far from over at the time of writing – is, amongst other things, a crisis of the infrastructure of the financial world: not of its technological infrastructure, where only limited difficulties were manifest, but of its cognitive infrastructure, of its fact-generation mechanisms’ (p. 178). This quote helps capture the manner in which MacKenzie's studies highlight the inefficiencies and in-built biases of financial markets. This is an important study, one that probes profoundly into the nature of financial markets from the standpoint of the sociology of science and raises as many questions as it answers.