Does Government Subsidy Affect Firm Survival? Evidence from Chinese Manufacturing Firms

ABSTRACT

This article applies a matching approach to deal with the selection bias and use the complementary log-log model to analyze the impacts of subsidy on Chinese manufacturing firms’ survival from 1998 to 2007. Our empirical results show that government subsidies significantly decrease the likelihood of firm exit. However, the effect decreases as the level of subsidies increases for private and foreign firms, but displays a nonlinear relationship across subsidy levels for SOEs. We also show the effects vary across the levels of institutional quality measured by the prevalence of rent seeking and government intervention. Further results suggest that the potential channels include increased investment in intangible and fixed assets as well as enhanced profitability.

KEY WORDS:

• firm survival
• government connection
• subsidy
• China

Notes

1. At the end of our observation period, some of the firms could still be in operation. However, we cannot observe enough information.

2. In robustness tests, we will obtain estimates based on continuous-time hazard models with a parametric baseline hazard function that can control for unobserved firm heterogeneity and estimates based on the Cox model correcting the partial likelihood function for ties using the method of Breslow (1974) for comparability with previous studies.

3. In China, because of the planned economy before the 1980s, numerous older firms are mainly state-owned. Since the 1990s, China has started to promote reforms in the SOEs, and this is an essential part of China’s economic transition. During this process, a large number of SOEs have been privatized, merged or reorganized, resulting in massive exits of older firms. Moreover, as older firms or larger firms may suffer from x-inefficiency, squared terms of size and age are included in our estimations (Liu and Li, 2015).

4. Appendix 1 defines all the variables used in this aticle. Our data have been deflated by the deflators taken from the China Statistical Yearbook (various issues) published by the National Bureau of Statistics of China. We use the provincial capital goods deflator to deflate the capital variables and the gross domestic product (GDP) deflator to deflate other variables.

5. Detailed information of the NBS database and its cleaning procedures are well summarized by Brandt et al. (2014).

6. All the correlation coefficients are less than 0.44, and most of them are very small, alleviating the concern of the multicollinearity problem when they are used simultaneously in the same regression.

7. The data come from World bank (2006).

8. We group the levels of affiliation into four groups with higher values indicating affiliation with lower level: (1) central and provincial level; (2) prefectural and county level; (3) street, town and village and below; (4) no affiliation.

Additional information

Funding

This work was supported by the Beijing Social Science Foundation [17LJC008]; Youth project in Humanities and Social Science of Ministry of Education [18YJC790212].