28.02.2015 Views

Exchange Rate Economics: Theories and Evidence

Exchange Rate Economics: Theories and Evidence

Exchange Rate Economics: Theories and Evidence

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Market microstructure approach 363<br />

of trade). In addition to the advantage of the longer time period compared to<br />

Lyons,Yao’s dealer also has a substantial volume of trade with non-dealer customers<br />

(25% relative to 1% in the case of the Lyons data set) making his mix of<br />

trade more representative of the market average.<br />

Evans (1997) also uses Reuters D 2000–1 data,sourced from a customised feed at<br />

the Bank of Engl<strong>and</strong>,on interbank trading for the period 1 May–31 August ,1996.<br />

In particular,he has access to time-stamped tick-by-tick data on all transactions<br />

for nine USD bilateral currencies (the price,a bought-<strong>and</strong>-sold indicator <strong>and</strong><br />

cumulative trading volume). This data has the advantage over that of Lyons <strong>and</strong><br />

Yao since it contains 24 hour transaction data on the whole of the interbank<br />

market,rather than one dealer,<strong>and</strong> makes it possible to analyse order flow’s role<br />

in price determination for the overall market. As Lyons (2001) notes,the fact that<br />

this data set spans such a long time span,<strong>and</strong> for so many currencies,is important<br />

since it allows analysing exchange rate determination from more of an asset price<br />

perspective than is possible with the other data sets <strong>and</strong>,second,it permits analysis<br />

of intra-day patterns with more precision than prior studies (noted earlier) which<br />

used indicative quotes. However,to set against these advantages,the data set only<br />

consists of transaction prices,rather than the spread,<strong>and</strong> also does not include<br />

the inventory position of dealers. Although the data is date-stamped to the second<br />

decimal place,it is not necessarily continuous <strong>and</strong> for this reason Evans analyses<br />

the data on a 5 minute by 5 minute basis. He shows that there is pronounced<br />

heterogeneity across the markets in the sense that,in terms of transactions,the<br />

DM is a much more active market than either the guilder or krona.<br />

The nature of the data allows Evans to construct measures of market-wide<br />

dem<strong>and</strong> for currencies of the form:<br />

D t =<br />

(Number of dollar purchases) − (Number of dollar sales)<br />

,(14.28)<br />

Number of dollar purchases + sales<br />

<strong>and</strong> he finds that there is a strong statistically positive correlation between this<br />

excess dem<strong>and</strong> measure <strong>and</strong> the change in the exchange rate. Evans argues that<br />

the source of this correlation can be traced to an informational asymmetry between<br />

traders during bilateral conversations. In particular,the nature of the foreign<br />

exchange market – decentralised trading,lack of transparency <strong>and</strong> heterogenous<br />

information – means that there is a tendency of under-adjustment of quotes with<br />

respect to excess dem<strong>and</strong> <strong>and</strong> the lack of transparency stops traders from observing<br />

the extent to which the arrival of private information is correlated across the market.<br />

This,in turn,creates the informational asymmetry referred to earlier.<br />

Using a VAR analysis,Evans also examines the interactions between the transaction<br />

prices,quantities <strong>and</strong> quotes. This analysis suggests that the posting of quotes<br />

is a distinct activity rather than an adjunct of trading,although the data also indicate<br />

that quote <strong>and</strong> trading activity are not made independently from each other.<br />

Furthermore,changes in trading activity within the interbank market significantly<br />

affect quotes,while innovations in quote activity affect transactions. These complex<br />

interactions are not consistent with any extant microstructure model.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!