| This dissertation examines voluntary market-making behavior, namely scalping, in futures markets. The data used are time-stamped electronic transactions in soybeans on the Dalian Futures Exchange in China marked with individual traders' identities. Previous studies have followed single scalpers or broader aggregates, but never all participants in a market. The transactions over seven months in 1999 and 2000 involved 125 brokerage houses representing 6,213 distinct customers. Among these many traders, about 7 or 8 during any month followed the short-term strategies normally classified as scalping, accounting just themselves for 10% of volume. This finding is itself interesting, for scalping is usually imagined to be a feature of traditional pits rather than electronic markets.; Using these transaction data, a bivariate autoregressive conditional intensity (ACI) model is estimated for two most actively traded futures contracts in the sample. The ACI model, a new econometric tool for analyzing times-series count data, is appropriate because the number of scalpers present in a five-minute interval, the dependent variable, is a small integer. The bivariate ACI also allows for scalpers' participation in simultaneous contracts, which helps account for the endogeneity of scalpers' trading. The results suggest that volatility and transaction levels of non-scalpers increase the likelihood of scalper presence, however, scalper spread does not appear to explain scalper presence. Furthermore, Non-scalpers' trading appears to be the source of price volatility.; An analysis of the relationship between scalpers' realized spread and trading frequency shows that most scalpers are trading below the optimum level. Two vector autoregression models are estimated to determine the immediate and longer term interactions among the profits and the trading frequencies of different groups of traders. The results suggest that large-volume traders are likely to be “informed” traders; scalpers and small-volume traders, who account for 95 percent of all traders, are likely to be “uninformed”. Evidently, the customers of large brokers are likely to be more informed than those of small brokers. It is also found that scalpers' realized spread as a group increases when others trade more often, consistent with the finding that scalpers derive more income from a higher demand for liquidity. |
