A clear-cut empirical implication of the inventory model is that in setting their prices dealers take their inventory levels into account. More precisely, when their holding of a risky asset increases, they offer liquidity at better prices so as to induce their customers to buy. The opposite applies when their inventory decreases. There is a vast empirical literature on this issue. Hasbrouck and Sofianos (1993) and Madhavan and Smidt (1993) show that inventory changes have an effect on prices, so that quote revisions are negatively related to the specialists’ trades. They also show that quoted prices induce mean reversion in inventory towards the target portfolio. What is ambiguous, however, is the time horizon of inventory mean reversion. Both Hasbrouck and Sofianos (1993) and Subrahmanyam (2008) find evidence that specialists react slowly to inventory shocks, so that inventories can be persistent for up to two months. This slow adjustment is not due to hedging with positions in other stocks since, as Naik and Yadav (2003) more recently showed using data from the London Stock Exchange, dealer firms’ quote changes are significantly related to changes in their ordinary inventories, not to changes in equivalent inventories in other stocks. Thus it remains a puzzle
why it takes so long for inventory effects to be manifested in financial market data.