This work provides a method of endogeneity corrected stochastic frontier for efficiency and productivity growth estimation in presence of product and labour market imperfections. The imperfections generate efficiency losses and hence affect the productivity estimates. A modified frontier function based on Cobb-Dauglus form that represents in terms of residue per unit of capital helps estimating the terms containing product and labour market imperfections. To correct the endogeneity issue for input selection, the estimation involves three-stages approaches. First, by applying Battese and Coelli (1992) error component model, the one-sided error is estimated. Second, the Levinsohn and Petrin (2003) approach is applied on the frontier values to estimate efficiency parameters that eliminates the input selection bias by using material cost as a proxy. Third, the endogeneity corrected technical efficiency is recovered from the modified one-sided error. The analysis of threedigit level of Indian industrial data across 17 major states for the period 2008-2016 portrays a strong presence of product and labour market imperfections. While the product market efficiency has deteriorated a bit at the aggregate level, it shows a marginal improvement in the allocative efficiency for the labour market. However, the efficiency level varies across industries. The productivity growth derived by adding the components of efficiency changes is similar to the Solow residual growth.