Estimate firm-year accounting conservatism

Tri Tri Nguyen
May 12, 2017


I am a PhD student. My second paper focuses on accounting conservatism. I want to estimate firm-year conservatism following Khan and Watts (2009, JAE). In their model: GSCORE and CSCORE are regressed on LEV, MTB, SIZE (year t). They argue that those variables are determinants of accounting conservatism. My question is that: can I use LEVt-1, MTBt-1, SIZEt-1 (year t-1) to estitmate GSCORE and CSCORE? If not possible, please advise my what isssues associated with the use of lagged LEV, SIZE, MTB to estimate accounting conservatism.

Thank you very much! Tri Tri

Ties de Kok
May 17, 2017

The Khan and Watts approach is a two step procedure. The first step is a cross-sectional regression that is run per year. The second step is not a regression, it just an equation that you fill in with the coefficients from the first stage corresponding to that firm year and the values for Size, MTB, and Lev for that specific firm-year observation. 

I don't know whether this answers your question? Why would you want to use the lagged values? You can use the lagged values if you also take the "lagged" coefficient estimates but this will just give you the "lagged" GSCORE / CSCORE. 

Tri Tri Nguyen
May 27, 2017

Thank you very much! It does not answer my question, but it is helpful in fact. Best regards, Tri