Hi B.T.,

I have read a few papers where lagged dependent variables are used in discrete and first-difference of operational returns. Although the measurements are based on accounting data, the principal may transfer over to financial data.

The general purpose of using this lagged dependent variables is a 'partial adjustment' for any mean reverting behaviour. The general theory behind the partial adjustment is that any profitable companies will be imitated by less profitable companies, therefore lose their competitive edge. The less profitable companies will be forced to adopt better capital investment decisions, therefore become more profitable. The overall effect is that operational returns are mean reverting in the long-term. This relates to Fama and French (2000)'s paper, forecasting operational performance.

Regarding stock data, if you lagged dependent variable is negative, it will also indicate a reversion towards an equilibrium value. Any price movement today partially reverts the following day. A positive lagged dependent variable would indicate drift, where the price movement today continues the following day.

I hope this helps.

Hi Daniel, thank you for your insight on the negative coefficient of the lagged dependent variable in a stock return model. I think your interpretation makes economic sense.  Any other possible interpretations are welcome.

Thank you for providing alternative interpretation of the negative coefficient of lagged dependent variable in stock return model. The reference you provided is appreciated too. More (if any) alternative interpretations are welcome.