Weather and Time Series Determinants of Liquidity in a Limit Order Market

Abstract

When liquidity is measured by the bid-ask spread or price impact, markets with more trading activity are typically more liquid than markets with less trading activity. But showing a causal connection from trading activity to spreads is difficult because these variables are endogenous. In the case of Finland's fully electronic limit order market, we use deseasonalized sunshine as an instrument for trading activity, and find that indeed higher trading activity causes lower spreads in the time series. We introduce another instrument for spreads and show that causality runs the other way as well: lower bid-ask spreads invite more trading activity. By using the lagged CBOE Volatility Index as an instrument, we also find that an exogenous increase in intra-day volatility causes larger spreads.

Keywords: Market microstructure, price impact, instrumental variables

JEL Classification: G1