Research Question? Can We Relate Post Entry Performance of Individual establishments to establishment specific characterstics?
Prior Work Audretsch(1991) showed that survival rates of industries happen at industry level and not at business level. hij – baseline rate of hazard Xij – Vector of Covariates containing 3 factors Long run average cost function - degree to which new business is burdened by inherent size disadvantage Technological regime change( Industry growth (Price/Cost Margin)
Model Estimation To estimate the β’s the Partial Likelihood function Coefficients are estimated by maximizing the PL.(Log Likelihood function) In case of Ties a modified PL is used.
Model Estimation S(t) – vector of the sum of covariates for episodes at time t, m – number of exits at time t
Data Specification Longitudanal data does not have start and end date [5 year interval used] SIC predicted estimation. – data from D&B file provides virtual census 12251 new establishments in 1976 studied for 10 years. Firm specific characterstics – innovation rate=number of innovations/employment Establishmnet size = MES Industry employment growth , price – cost margin Macroeconomic employment rate , real interest rate (Economic report of president)
Critique While Survival Analysis is right, handling of censoring (right/left and interval) is not handled in his model Not clear whether the coefficients are significant (Wald test of coefficients) Is the model adequate to explain the phenomenon?