ABSTRACT
Using a loan guarantee contract for credit financing is the primary method for firms to obtain external financial resources. We examine whether obtaining loan guarantees from third parties can promote innovation behaviour, and investigate the underlying mechanism. Using the 2007–2016 data of Chinese listed firms, we find that (i) firms with loan guarantees have lower innovation input but much higher patents output in that very year, (ii) higher the guarantee coverage cannot promote R&D expenditure in that very year and the following year. The results indicate that loan guarantees can promote innovation behaviour because it can ease the financial constraints the debtor facing, rather than the risk sharing mechanism.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 We use KZ index (Kaplan and Zingales Citation1997) to measure financial constraints. KZ index is calculated as followed. We use financial data of A-share listed companies (removing firms in financial industry) from 2007 to 2016. After deleting missing data and that has IPO in the given year, we obtain the calculation sample including 19,314 firm years. We use five variables to measure degree of financial constraints:operating net cash flows to total assets (CF/Assett-1), cash holding to total assets (DIV/Assett-1), cash dividend to total assets (CASH/Assett-1), leverage, and Tobin’s Q, and calculate the median of these variables in each year. For each observation, if CF/Assett-1 (DIV/Assett-1, orCASH/Assett-1)is less than its median, we set kz1 (kz2, orkz3) = 1, otherwise kz1(kz2, orkz3) = 0; if Leverage (Tobin’s Q) is larger than its median, we set kz4 (or kz5) = 1, otherwise kz4 (or kz5) = 0. Next, kz = kz1+ kz2+ kz3+ kz4+ kz5. kz<2 means a firm less financially constrainted, while kz>3 implies a firm suffering more serious financial constraints. Given kz as dependent variable, we conduct ordered logistic regression to obtain the estimated coefficients of above five variables. These coefficients can be used to calculate KZ index of each observation in our sample.