Does high debt ratio influence Chinese firms’ performance? A semiparametric stochastic frontier approach with zero inefficiency

A flexible stochastic production frontier model with panel data

“…All four models yield the same ranking for the top 10 firms. A closer look at the ownerships of firms shows that six of them are non‐state owned enterprises (Non‐SOEs), consistent with recent findings that non‐SOEs are more efficient in Chinese economy (Wang et al, 2021). The ranking for the bottom firms differs across models.…”

“…We source our data from Malikov et al (2020), which uses a firm-level panel dataset from Chinese Industrial Enterprises Database during 2000-2006, collected through annual surveys by the National Bureau of Statistics of China (NBSP). The dataset provides reliable measures on variables because NBSP endeavors to ensure consistency in data collection across regions and industries and audits selected companies to ensure data accuracy (Wang et al, 2021). The dataset is also suitable in our model because it offers a large number of firms with shorter time span.…”

“…$$$ {\alpha}_i $$$ is the fixed effect, capturing the latent heterogeneity, which can be freely correlated with $$$ X $$$. Technologies in our model may be different for different firms and different time periods, since environmental variables from $$$ {X}_{it} $$$ vary over $$$ i $$$ and $$$ t $$$ and are likely to be the determinants of technologies (Wang et al, 2021). In case of no environmental variables exist, technologies may change across firm due to intercept shifts through $$$ {\alpha}_i $$$.…”

“…On the one hand, firms with exports are more efficient because of learning by exporting, efficient reallocation of resources, and better managerial performance due to competitive pressure (Clerides et al, 1998). On the other hand, the imperfect market in China allows exporting firms of certain ownership types, particularly state‐owned enterprises, to have soft budge constraint, which may induce moral hazard problem that increases inefficiency (Wang et al, 2021). We reexamine this puzzling issue in the light of our new model.…”

“…There are many other contexts where our model fits. For instance, the nonparametric part can be either production or cost frontier and the parametric part be a nonlinear probability function for firms being fully efficient (Wang et al, 2021). Alternatively, the model can capture effects of determinants, such as oil price, that affect exchange rate (Xiao et al, 2014).…”