مقاله سال ۲۰۱۷

Do heterogeneously informed currency traders differ in their use of information? If so, how does private information impact their trade size? What is the relationship of trade size to foreign exchange (FX) rate volatility? This paper seeks answers to these questions by linking the information process to order size patterns while relying on a trading strategy designed in an electronic FX market. Extending (Easley et al., 1997b; Easley and O’Hara, 1987), our high-frequency trading setup allows informed and uninformed traders to place orders sequentially in continuous time.1 To test the predictions of the strategies, we derive tractable likelihood functions that identify the variation in trade size associated with the orders of informed and uninformed FX traders.

Based on a retail electronic trading platform dataset, our empirical analysis reveals several notable findings. First, we empirically show that large orders are likely to be executed by informed traders rather than uninformed traders. This evidence is particularly pronounced for buy orders and remains strong regardless of the choice of size thresholds. These results highlight the importance of the information content of trade size (i.e., informativeness) in characterizing currency transaction data. More broadly, a direct implication of this analysis is that order flow size could be informative by itself even in the absence of information shocks.2 Second, an estimated logit model suggests that large trade size appears to be an endogenous factor that depends on price volatility. This finding supports the intraday trading invariance principle proposed by Andersen et al. (2015). Finally, we assess the distributional characteristics of price increments and show that excess kurtosis in exchange rate data, corresponding to large price-contingent trades, is significantly lower than that in small trades. Our motivation for this assessment directly builds on the argument of Osler and Savaser (2011) and Osler (2005), who provide evidence that pricecontingent trading could solely explain the extreme price cascades in the transactions of an FX dealer. We emphasize that the source of extreme events could be attributed to the informativeness of trade size: uninformed traders tend to place small orders as a range of extreme stop-loss and take-profit trades. This may result in jump cascades or excess kurtosis observed in transaction data. In addition to quotation bursts in equities (Gençay et al., 2016), the size and informational content of trades could thus be additional drivers of currency jumps.

The remainder of the paper is organized as follows. Section 2 introduces the model and outlines the trading environment. Section 3 describes our data. In Section 4, we present and discuss the empirical results. Section 5 concludes the paper.