Bid/ask Spreads And The Costs Of Making Markets

The purpose of this thesis is to test the received explanations of the determinants of bid/ask spreads for common equity securities. This thesis employs trade-to-trade data for twenty-four securities listed on the Toronto Stock Exchange (TSE) over the period from August, 1988 through December, 1992. The literature is extended by simultaneously modelling the cost components (inventory, asymmetric information, fixed, and search costs) and volatility of realized bid/ask spreads using an exponential autoregressive conditional heteroscedasticity (EARCH) framework. Because certain small trades on the TSE are auto-filled, registered traders do not always make a decision on trade involvement--some trades are non-discretionary. The theoretical model of spreads used in this thesis extends Glosten and Harris\u27 (52) (GH) model by distinguishing between non-discretionary (auto-filled) and discretionary trades to allow testing for differences in cost structures. Determination of the relative contribution of the various cost components is also based on GH.;The contributions of this thesis are threefold: (1) the costs of trading are dependent on the trade type--auto-filled versus discretionary; (2) the volatility of realized spreads is modelled as an exponential autoregressive conditionally heteroscedastic (EARCH) process which is a function of information flow; and (3) richer, more exact intra-daily data is used to eliminate problems currently recognized in the literature. The primary conclusions are: (1) for non-discretionary trades, the average cost per share per trade is {dollar}0.0683 of which 96\% is due to fixed costs and 4\% is an asymmetric information cost; (2)~for discretionary trades, the average cost is \{dollar}0.0273 of which 76% is fixed and 24% is information: and (3) search (liquidity) and inventory costs are not statistically different from 0. These results are consistent over three different sub-periods