An Investigation of Market Invariance for the Tokyo Stock Exchange

We test the invariance-of-bet hypothesis from Kyle and Obizhaeva (2016) for one of the largest order-driven equity markets: the Tokyo Stock Exchange (TSE). The pooled regression coefficients of the log of the number of trades on the log of trading activities range from 0.586 to 0.679, close to the theoretical value of two-thirds predicted by the invariance-of-bet hypothesis. We further calibrate the order-splitting multiplier by combining the TSE’s unique feature and the equation derived from the market invariance hypothesis. The calibrated order-splitting multipliers decrease as the minimum tick size increases from 1 to 50 JPY. Market depth is negatively correlated with the order-splitting multiplier. These findings suggest that increased order-splitting activities may give rise to the estimated coefficients deviating from the theoretical value implied by the market invariance hypothesis