*4.1. Summary Statistics*

Table 2 reports the descriptive statistics of the matched sample firms for the variables used in our main regression and the corresponding correlation matrix. The first two rows contain the main dependent variables, i.e., LnPatents and LnInventions. The average number of patents (the raw value) was 16.02 before matching versus 29.39 after matching; the average number of invention patents was 7.30 before matching versus 14.19 after matching. The average number of invention patents (after matching) was smaller than the average number of utility patents (14.19 vs. 15.19), thus suggesting that applications for invention patents were harder to attain than utility patents [27]. As the correlation matrix shows, firms with higher ROA, lower leverage, and more cash holdings were more likely to invest in R&D; firms with a larger size, older age, and higher leverage had higher innovation outputs, that is, a higher number of patents.
