Up to this point the focus has been on the macroeconomics of exchange rate behaviour and we have argued, either directly or indirectly, that this focus is a valuable one. However, there are still important issues to address with respect to understanding the daily volume of foreign exchange traded globally and how the price of foreign exchange is actually set in the foreign exchange market. For example, we noted in Chapter 1 that on a day-to-day basis the current BIS estimate of the volume of gross trading in foreign exchange markets on a global basis is approximately $1.2 trillion,1 86% of which occurs between market makers alone. Since the total annual world trade ﬂow is around $4 trillion it is clearly difﬁcult to explain the massive foreign exchange trade in terms of standard macroeconomic fundamentals and therefore a number of researchers (see, for example, Frankel and Rose 1995a; Flood and Rose 1999; Lyons 2001) have proposed using a microeconomic-based modelling approach, namely, a market microstructure approach. This micro-based approach focuses on an array of institutional aspects of the foreign exchange market, such as price formation, the matching of buyers and sellers (i.e. market makers and brokers) and optimal dealer pricing policies. As we shall see later, one of the key explanations offered by the microstructural
approach for the huge daily volume in foreign exchange trade is the so-called ‘hot potato’ effect. The idea is that if an initial trade between a customer and a bank produces an unwanted position for the dealer, she will try to ofﬂoad this to another dealer and this process will continue until an equilibrium, where the initial foreign exchange position is willingly held, is reached. This may be seen as a form of risk management. The interpretation of high volume has important policy implications. For example, if, as many conjecture, high volume is a reﬂection of speculation some form of tax, such as the Tobin tax of throwing ‘sand in the wheels’ of international ﬁnance, may be the appropriate remedy. But if the volume reﬂects riskmanagement the tax would impede this andwould therefore be undesirable. Garman (1976) introduced the term market microstructure to deﬁne ‘moment-
to-moment trading activities in asset markets’. O’Hara (1995) gives a more general ﬁnance-based deﬁnition: ‘Market microstructure is the study of the process
ROMADO: “CHAP14” — 2007/1/11 — 12:47 — PAGE 340 — #2
and outcomes of exchanging assets under explicit trading rules. While much of Economics abstracts from the mechanics of trading, the microstructure literature analyses how speciﬁc trading mechanisms affect the price formation process.’ As we shall see in this chapter, such trading mechanisms may range from the role of a broker as an intermediary in the transaction process to the existence of (or lack of) a centralised trading location. In essence, market microstructure research exploits the structure provided by the existing trading patterns in a particular market and attempts to show how this affects price formation, returns and hence the (informational) efﬁciency of the market. In the microstructural literature, the focus on price formation is very different to the standard approach taken in the economics literature, where most of the discussion centres on market-clearing or equilibrium prices. In the context of the macroeconomics literature considered in this book this is perhaps at its clearest in the use of the rational expectations assumption, where behaviour out of equilibrium is not considered. Lyons (2001) ﬂeshes out the general deﬁnition of market microstructure, given
by O’Hara, and provides a deﬁnition which is much more speciﬁc to the foreign exchange market. He argues that what distinguishes the micro-approach from the asset market approach, considered extensively in the earlier chapters of this book, is that it relaxes three key assumptions of the asset approach. First, microstructural models recognise that some information relevant to exchange rates is not publicly available. For example, foreign exchange dealers regularly have access to information on trades – the concept of order ﬂow – which gives them inside information on how a currency might move in the future. A good example of this is central bank intervention which gives a trader, or traders, an indication of how the central bank perceives currency developments. Also traders engaged in exporting and importing take positions with foreign exchange dealers and thereby impart information on the evolution of the trade balance that is not available to the general public. This, of course, contrasts with the asset market model where all information is supposed to be publicly available. It is, however, consistent with Fama’s deﬁnition of strong-form efﬁciency (see Chapter 15), which encompasses both public and private information. Second, market microstructural models recognise that agents may differ in ways that can affect prices – that is, heterogeneity of expectations is signiﬁcant. One important way they differ is with respect to expectations formation. In nearly all macro-based models considered in this book, it is assumed that expectations are homogeneous and this is usually expressed using the conditional expectations operator, E(./It ). The essence of micro-based work is that the huge volumes of foreign exchange traded can only be explained by the existence of heterogeneous agents. In particular, why would agents trade with each other if they all held the same expectations? The kind of heterogeneity referred to in the microstructure literature may arise from a number of sources, such as differences in information, beliefs, preferences and wealth. Third, and as we noted in the previous paragraph, market microstructure models emphasise that trading mechanisms differ in ways that can affect prices. In sum, Lyons argues that the ‘hallmarks’ of any market microstructure model are order ﬂow and the bid-ask spread, which is the measure of price.