Illiquidity in the Japanese Day-Ahead Electricity Market

Historical data of system prices and traded quantities of electricity over the 48 half-hour intra-daily intervals in the Japan Electric Power Exchange are analyzed. Viewed as a panel dataset of the 48 di erent commodities in 7 di erent markets (days of a week) or 336 di erent contracts over 288 weeks, the data allow me to compute two representative measures of illiquidity, namely, Amihud's price-impact measure and Roll's implied spread cost measure from November 2006 to April 2012. These measures are based on the absolute weekly returns of each of 336 contracts divided by the corresponding volume of traded electricity, and on the rst-order serial covariance of weekly returns, respectively. Two measures closely comove but they contribute to the returns in di erent magnitudes, suggesting that each of them captures both common and distinctive aspects of illiquidity in the JEPX market. Once the lagged returns are controlled for in a dynamic panel framework, the in uence of price-impact measures on returns dominate that of spread measures. The price-impact measure and traded volume contribute the return variations in opposite signs, i.e., positively and negatively. It suggests that the assessment of illiquidity requires a careful treatment of these confounding factors.