Revisiting Tick Size: Implications from the SEC Tick Size Pilot

Abstract

This paper analyzes the effect of the SEC tick size pilot on market quality variables. We find the predicted trade-off from bigger tick sizes: greater cost of execution (spreads) and greater displayed liquidity in the form of increased depth. This suggests that with greater tick sizes investors can trade larger blocks of assets at the best prices (bid/ask) but at the same time the cost of small transactions is greater. Moreover, we find this effect to be significant only in assets for which the quoted spread is of a magnitude similar to the new tick size, thus the effect is significant for assets with a quoted spread which is proportionately large relative to the tick size. Increasing the tick size has a second effect, namely it reduces the minimum price variation. We find that this minimum price variation affects all assets in that we find a significant decrease in the number of changes in the best prices (bid or ask) in all assets. We also find that the increase in minimum price variation decreases the range of prices (the difference between the maximum and minimum price) though this effect is only weakly significant for assets which have quoted spreads that are significantly larger than the tick size. %This decrease in volatility could be the result of higher trading costs and lower volume and activity. The undesired effect comes from volume and activity. We also find a significant decrease in NASDAQ intraday trading volume as well as across the sum of all venues. This effect is contrary to the SEC's stated objective and that of the JOBS Act, namely to increase volume, and we find that the effect is most significant for the treatment group with the trade-at restriction (treatment group 3). Finally, we document a reduction in the level of market activity as measured by messages and trades that are posted and subsequently canceled in a short period of time (fleeting order). We find that the effect of the pilot is consistent with theoretical predictions and that this effect is statistically and economically significant for assets with quoted spreads that are of a similar magnitude as the tick size, and weakly or non-significant for assets with quoted spreads that are large relative to the tick size. These results suggest that the focus of evaluating the effect of tick sizes should be on the relative size of the asset's quoted spread, and not, as is often the case in the literature, the relative tick size as measured in terms of the asset's absolute price level.