29.03.2013 Views

SEC Follow Up Exhibits Part C SEC_OEA_FCIC_001760-2501

SEC Follow Up Exhibits Part C SEC_OEA_FCIC_001760-2501

SEC Follow Up Exhibits Part C SEC_OEA_FCIC_001760-2501

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

Reg SHO Pilot Report 2/12/2007<br />

V Methods<br />

A. Comparing Pilot and Control Stocks<br />

We derive the results in this report using several empirical approaches. The first<br />

approach compares the pilot sample to the control sample both before and after the start of the<br />

pilot. The approach allows us to apply a standard t-test to the difference of the differences. As<br />

long as the pilot and control samples are generally well constructed, this approach will control<br />

for changes in market conditions that are unrelated to the pilot. However, this approach may not<br />

fully control for differences within the samples. Further, it is cumbersome when examining the<br />

differential effect of the pilot on sub-samples. Therefore, we will also employ a regression<br />

approach.<br />

The regression approach includes multiple test variables at the same time and also<br />

controls for unforeseen differences between the pilot and control samples. Because the number<br />

of stocks changes over the sample period, we employ daily cross-sectional regressions for each<br />

sample day after the start of the pilot and then we average the coefficients. We employ<br />

variations on the following regression:<br />

Market Quality = α + β Pilot + β Pre Pilot Market Quality<br />

(1)<br />

i 1 i 2<br />

i<br />

where Market Quality is one of several dependent variables (e.g., effective spread) measured for<br />

each stock on one trading day, Pilot is equal to one for pilot stocks and zero for other stocks, and<br />

Pre-Pilot Market Quality is the average daily level of the dependent variable measured over the<br />

Pre-Pilot Period. If the dependent variable is generally larger for pilot stocks than control stocks,<br />

then β1 will greater than zero. If this relation is persistent across time, then the average β1 will be<br />

greater than zero. The Pre-Pilot Market Quality variable helps ensure that the coefficients on the<br />

Prepared by the Office of Economic Analysis 28

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

Saved successfully!

Ooh no, something went wrong!