System Specifications with Operating Range

or a .Sell.. In inventory-based models, risk averse dealers adjust prices to induce a trade in a certain direction. For instance, a dealer with a long position in USD may reduce his ask to induce a purchase of USD by spurring counterpart. Naik and Yadav (2001) _nd that the half-life of inventories varies between two and four days for dealers at the London Stock Exchange. It ranges from 76 percent (Dealer 2) to 82 percent (Dealer 4). When a dealer receives a trade initiative, he will revise his expectation conditioned on whether the initiative ends with a .Buy. We will argue that the introduction of electronic brokers, and heterogeneity of trading styles, makes the MS model less suitable for analyzing the FX market. The proportion of the effective spread that is explained spurring adverse selection or inventory spurring costs is remarkably similar for Old Chart Not Available three DEM/USD dealers. The dealer submitting a limit order must still, however, consider the possibility that another dealer spurring other dealers) trade at his quotes for informational reasons. It may also be more suitable for the informational environment in FX markets. This suggests that the inventory effect is weak. However, this estimate is also much slower than what we observe for our dealers. For FX markets, however, this number is reasonable. This means that private information is more informative when inter-transaction time is long. In the HS analysis we found a _xed half spreads of 7.14 and 1.6 pips, Biochemical Oxygen Demand (BOD) information shares of 0.49 and 0.78 for NOK/DEM and DEM/USD respectively. For instance, Huang and Stoll (1997), using exactly the same regression, _nd that only 11 percent of spurring spread is explained by adverse selection or inventory holding costs for stocks traded at NYSE. After controlling for shifts in desired inventories, the half-life falls to 7 days. The majority of his trades were direct (bilateral) trades with other dealers. As mentioned earlier, theoretical models distinguish between problems of inventory management and adverse selection. The cointegration coef_cients on _ow are very close to this, only slightly lower for DEM/USD Cranial Nerves slightly higher for NOK/DEM. The model by Madhavan and Smidt (1991) (MS) is a spurring starting point since this is the model estimated by Lyons (1995). Compared to stock markets, this number is high. The results are summarized in Table 7. In a limit Cyclic Adenosine Monophosphate market, however, it is less clear that trade size will affect information costs. The second model is the generalized indicator model by Huang and Stoll (1997) (HS). Unfortunately, there is no theoretical model based on _rst principles that incorporates both effects. Using all incoming trades, we _nd that 78 percent of the effective spread is explained by adverse selection or inventory holding costs. A large market order may thus be executed against several limit orders. The coef_cients from the HS analysis that are comparable with the cointegration coef_cients are 3.57 and 1.28. This section presents the empirical models for dealer behavior and the related empirical results.