To examine whether the Biopharmaceutical Act is effective only for biopharmaceutical firms rather than other industries, we choose the high-tech industry as the control industry because high-tech industry also has R&D intensity as high as that of biopharmaceutical firms in Taiwan.
4.4.1 Difference-in-differences Estimator (DID Estimator)
Table 7 presents the DID estimator of innovation for the inter-industry analysis. We also consider the possible continuous effect of the Biopharmaceutical Act and incorporate different time interval analyses in this table. Panel A and B present the DID estimator of the R&D investment and adjusted patent citations, respectively.
28 According to Yang et al. (2012), pharmaceutical firms usually have low R&D intensity because pharmaceutical firms generally produce generic drugs rather than patent drugs in Taiwan.
Table 7 DID Estimator: Inter-industry Analysis
Panel A: DID Estimator for R&D Investment: Inter-industry Analysis
t-1 t+1 Differences t-2 t+2 Differences t-3 t+3 Differences Treated 14.3224 14.7030 0.3806 13.5214 14.6223 1.1010 13.0075 13.4002 0.3927
(0.8686) (0.5928) (0.8183)
Control 1 17.9489 12.0816 -5.8673*** 17.1562 11.8311 -5.3251** 17.0038 11.5033 -5.5005**
(0.0067) (0.0253) (0.0305)
Control 2 18.5454 12.4521 -6.0932*** 18.0979 12.6801 -5.4178** 16.4575 11.3838 -5.0738**
(0.0027) (0.0108) (0.0141)
Control 3 16.6494 12.0499 -4.5995*** 15.8808 12.0349 -3.8458** 14.8170 11.0165 -3.8004**
(0.0018) (0.0108) (0.0109)
Control 4 16.6250 11.7890 -4.8361*** 16.4159 11.5156 -4.9003*** 15.0083 10.4198 -4.5885***
(0.0002) (0.0010) (0.0019)
Diff.1 -3.6265 2.6214 6.2479*** -3.6349 2.7912 6.4261*** -3.9962 1.8970 5.8932***
(0.0056) (0.0030) (0.0084)
Diff.2 -4.2230 2.2509 6.4739*** -4.5765 1.9422 6.5188*** -3.4500 2.0165 5.4665***
(0.0019) (0.0027) (0.0039)
Diff.3 -2.3271 2.6531 4.9802** -2.3594 2.5874 4.9468*** -1.8094 2.3837 4.1931***
(0.0178) (0.0084) (0.0066)
Diff.4 -2.3027 2.9140 5.2167** -2.8945 3.1067 6.0013*** -2.0007 2.9804 4.9812***
(0.0153) (0.0023) (0.0038)
Panel B: DID Estimator for Adjusted Patent Citations: Inter-industry Analysis
t-1 t+1 Differences t-2 t+2 Differences t-3 t+3 Differences
Treated 0.0253 0.0648 0.0395 0.0348 0.0477 0.0129 0.0689 0.0796 0.0107
(0.2171) (0.5468) (0.7261)
Control 1 0.3258 0.1355 -0.1902 0.0810 0.0927 0.0117 0.0675 0.0562 -0.0113
(0.2517) (0.8471) (0.8743)
Control 2 0.4472 0.2254 -0.2218* 0.2933 0.1812 -0.1122 0.0818 0.0805 -0.0014
(0.0606) (0.2705) (0.9762)
Control 3 0.4852 0.1749 -0.3103*** 0.2827 0.1790 -0.1037 0.1420 0.1772 0.0353
(0.0035) (0.1852) (0.4879)
Control 4 0.4164 0.1465 -0.2699*** 0.2266 0.1454 -0.0813 0.1279 0.1516 0.0237
(0.0081) (0.1657) (0.5809)
Diff.1 -0.3004 -0.0707 0.2297 -0.0462 -0.0450 0.0012 0.0013 0.0234 0.0220
(0.1731) (0.9851) (0.7702)
Diff.2 -0.4219 -0.1606 0.2613** -0.2585 -0.1334 0.1251 -0.0130 -0.0009 0.0121
(0.0308) (0.2313) (0.8358)
Panel B: DID Estimator for Adjusted Patent Citations: Inter-industry Analysis
t-1 t+1 Differences t-2 t+2 Differences t-3 t+3 Differences Diff.3 -0.4599 -0.1101 0.3498*** -0.2479 -0.1313 0.1166 -0.0731 -0.0976 -0.0245
(0.0014) (0.1526) (0.7009)
Diff.4 -0.3911 -0.0817 0.3094*** -0.1918 -0.0976 0.0942 -0.0590 -0.0720 -0.0130
(0.0034) (0.1337) (0.8247)
Note: This table presents the DID estimator of innovation for the inter-industry analysis. Panels A and B present the DID estimator of R&D investment and the DID estimator of adjusted patent citations, respectively. t is the event year, i.e. the year in which the Biopharmaceutical firm is approved by the Biopharmaceutical Act. Treated represents the treated firms, i.e.
approved biopharmaceutical firms. Control 1, Control 2, Control 3, and Control 4 respectively represent one, two, three, and four control firms to each treated firm. The control firms in the inter-industry analysis are high-tech firms. Diff.1, Diff.2, Diff.3, and Diff.4 represent the mean difference in the variables between Treated and Control 1, Control 2, Control 3, and Control 4 respectively. Numbers in the parentheses are p-values. ***,**, and * denote significance at the 1%, 5%, and 10% levels, respectively.
Panel A of Table 7 shows that the high-tech firms significantly decrease their R&D investment after the approval year although the approved biopharmaceutical firms do not change the proportions of R&D. The significantly positive DID estimators of R&D investment show that after the Biopharmaceutical Act, compared with high-tech firms, approved biopharmaceutical firms have a significantly higher proportion of R&D expenditure to total assets.
Panel B of Table 7 for the time interval (t-1, t+1) analysis demonstrates that high-tech firms significantly decrease their adjusted patent citations after the Biopharmaceutical Act. In this short time interval, the DID estimators of the adjusted patent citations are significantly positive, implying that the approved biopharmaceutical firms have significantly higher innovation output than high-tech firms after the Biopharmaceutical Act. However, the DID estimator results for time interval (t-2, t+2) and (t-3, t+3) are not significant. Therefore, the results show that the effect of the Biopharmaceutical Act on the innovation quality of the biopharmaceutical industry is less significant, and has only a short duration. This result of a short run effect is consistent with David et al. (2000).
4.4.2 Difference-in-differences Regression (DID Regression)
Table 8 shows the DID regression results for the inter-industry analysis. In Panel A, the significantly negative coefficients of Treatment show that the approved
Table 7 DID Estimator: Inter-industry Analysis (cont.)
biopharmaceutical firms have lower R&D investment than the high-tech firms. In addition, the significantly positive coefficients of the interaction term, After×Treatment, show that compared with high-tech firms, approved biopharmaceutical firms increase R&D investment significantly after the Biopharmaceutical Act. Thus, these findings demonstrate that compared with high-tech firms, approved biopharmaceutical firms are more encouraged to increase R&D investment by the act.
Table 8 DID Regression Results: Inter-industry Analysis
Panel A: DID Regression Results for R&D Investment: Inter-industry Analysis
One Matched Firm Two Matched Firms Three Matched Firms Four Matched Firms
(1) (2) (1) (2) (1) (2) (1) (2)
Aftert -1.4188 -1.3626 -2.1960*** -2.0770*** -0.6735 -0.4862 -0.0277 -0.0959 (0.1807) (0.1992) (0.0045) (0.0072) (0.2655) (0.4217) (0.9633) (0.8731) Treatmenti -3.1875*** -3.4854*** -3.3959*** -3.8954*** -1.9565*** -2.7109*** -1.8821*** -2.9477***
(0.0005) (0.0003) (0.0000) (0.0000) (0.0059) (0.0003) (0.0093) (0.0001) Aftert × Treatmenti 3.1546** 3.1323** 3.6625*** 3.5871*** 2.3407** 2.2497** 1.9755** 1.9896**
(0.0152) (0.0159) (0.0011) (0.0014) (0.0177) (0.0223) (0.0398) (0.0376) LN (TA)t -1.3558*** -1.3851*** -1.3597*** -1.3622*** -0.7284*** -0.7405*** -0.6099*** -0.6238***
(0.0002) (0.0001) (0.0000) (0.0000) (0.0007) (0.0006) (0.0012) (0.0009) RDt-1 0.4578*** 0.4547*** 0.5330*** 0.5273*** 0.5851*** 0.5758*** 0.6133*** 0.5999***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) ROAt -0.2356*** -0.2389*** -0.1988*** -0.2055*** -0.1721*** -0.1812*** -0.1725*** -0.1841***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Debt Ratiot -0.0132 -0.0257** -0.0361*** -0.0481***
(0.2954) (0.0148) (0.0000) (0.0000)
Observations 1,489 1,489 2,334 2,334 3,256 3,255 4,154 4,153
Adjusted R2 0.4900 0.4901 0.5243 0.5253 0.5226 0.5250 0.5496 0.5538
Panel B: DID Regression Results for Adjusted Patent Citations: Inter-industry Analysis One Matched Firm Two Matched Firms Three Matched Firms Four Matched Firms
(1) (2) (1) (2) (1) (2) (1) (2)
Aftert -0.0617** -0.0618** -0.0734*** -0.0724*** -0.0439* -0.0409 -0.0342* -0.0315 (0.0165) (0.0189) (0.0052) (0.0057) (0.0756) (0.1002) (0.0967) (0.0507)
Panel B: DID Regression Results for Adjusted Patent Citations: Inter-industry Analysis One Matched Firm Two Matched Firms Three Matched Firms Four Matched Firms
(1) (2) (1) (2) (1) (2) (1) (2)
Treatmenti -0.0602** -0.0624** -0.0846*** -0.0892*** -0.0943*** -0.0986*** -0.0797*** -0.0825***
(0.0113) (0.0140) (0.0039) (0.0041) (0.0026) (0.0034) (0.0056) (0.0076) Aftert × Treatmenti 0.0693** 0.0730** 0.0829** 0.0817** 0.0675* 0.0646 0.0591 0.0511 (0.0212) (0.0218) (0.0236) (0.0316) (0.0842) (0.1166) (0.1024) (0.1807) LN (1+NetSalest-1) 0.0041 0.0040 0.0050 0.0049 0.0092*** 0.0096*** 0.0084*** 0.0089***
(0.1087) (0.1484) (0.1215) (0.1498) (0.0057) (0.0067) (0.0043) (0.0042) RDt-1 0.0019*** 0.0019*** 0.0036*** 0.0035*** 0.0052*** 0.0052*** 0.0048*** 0.0047***
(0.0001) (0.0002) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Tobin’s Qt-1 -0.0008 0.0022 0.0034 0.0050*
(0.6933) (0.3806) (0.2308) (0.0567)
Observations 1,037 996 1,643 1,571 2,205 2,102 2,807 2,687
Adjusted R2 0.0552 0.0563 0.0641 0.0678 0.0853 0.0889 0.0807 0.0845
Note: This table presents the panel regression results of the inter-industry analysis, including the regression of R&D investment and adjusted citations with one, two, three and four matching control firms. The dependent variable of Panel B is the natural logarithm of 1+adjusted patent citation, i.e. LN (1+adjusted patent citation). The regression is shown in equation (1) of Section 3.3.4. Aftert = 1 if the firm is in the approval year or after approval year and 0 otherwise; Treatmenti = 1 if the firm is in the treated group and 0 otherwise. The treated firms are approved biopharmaceutical firms and control firms are high-tech firms. The definitions of the variables are presented in Appendix Table A1. Numbers in parentheses are p-values.***,**, and * denote significance at the 1%, 5%, and 10% levels, respectively.
Panel B of Table 8 shows significantly negative coefficients for Treatment, indicating that the approved biopharmaceutical firms have lower adjusted patent citations than the high-tech firms. In this panel, the coefficients of interaction term, After×Treatment, are positive significantly for one and two matched firms and are not significant for three and four matched firms. These results imply that approved biopharmaceutical firms are motivated more than high-tech firm to improve their innovation quality by the Biopharmaceutical Act, but the results are less significant.
In sum, both results of DID estimator and DID regression show that relative to high-Table 8 DID Regression Results: Inter-industry Analysis (cont.)
tech firms, the approved biopharmaceutical firms are more encouraged to invest in R&D activities and to improve their adjusted patent citations.29 These findings show the policy effectiveness of the Biopharmaceutical Act is only for biopharmaceutical firms (rather than firms in other high R&D intensity industries) on innovation improvement.
4.4.3 Subsample Analysis of Inter-industry
The subsection considers two inter-industry subsample analyses. First, we consider the possible effect of firm size and divide the sample into small and large firms for the subsample analysis of inter-industry because the firm sizes of biopharmaceutical firms are smaller than those of high-tech industries in Table 1. In addition, small firms usually lack collaterals and are hard to obtain external financing for R&D (David et al., 2000; Hall, 2002). Further, small firms also find it more difficult to appropriate the returns from R&D and thus have less motivation to invest in R&D (Chen et al., 2013). Therefore, small firms are more likely to have serious R&D underinvestment problems.
Table 9 shows the subsample DID regression result of inter-industry. Panels A.1 and A.2 of Table 9 exhibit significantly positive coefficients for the interaction term, After×Treatment, showing that for firms of similar sizes, approved biopharmaceutical firms have significantly higher R&D investment after the Biopharmaceutical Act than high-tech firms. However, the coefficients of the interaction term in the small firms are larger than those of large firms. These findings show that in the inter-industry analysis, the effect of the Biopharmaceutical Act on innovation investment may be stronger for small firms than for large ones. Small firms with more serious underinvestment problems may be stimulated to increase R&D investment after the Biopharmaceutical Act because the tax credits help to alleviate the financing constraint problem in small firms. These results are consistent with the concept of Baghana and Mohnen (2009) and Lokshin and Mohnen (2012), who argue that tax credit policy tends to be more effective in stimulating R&D input for small firms than for large firms.
29 The stimulation of innovation quality in the inter-industry comparison is less significant than the innovation investment.
Table 9 DID Regression Result of Inter-industry: Subsample Analysis for Different Firm Size
Panel A.1 DID Regression Results for R&D Investment in the Inter-industry Analysis: Small Firms One Matched Firm Two Matched Firms Three Matched Firms Four Matched Firms
(1) (2) (1) (2) (1) (2) (1) (2)
Aftert -2.6563 -2.6443 -3.5811*** -3.4917*** -3.1244*** -2.9833*** -2.5278*** -2.3399***
(0.6559) (0.1597) (0.0059) (0.0074) (0.0030) (0.0046) (0.0053) (0.0098) Treatmenti -2.5368 -2.6266 -3.0630** -3.3108** -2.5942* -3.0755** -1.1068 -1.7929 (0.1623) (0.1552) (0.0458) (0.0328) (0.0681) (0.0324) (0.4134) (0.1898) Aftert × Treatmenti 5.7595** 5.7399** 5.6477*** 5.4853*** 3.9550** 3.7895** 2.2866 2.1258 (0.0188) (0.0194) (0.0053) (0.0068) (0.0308) (0.0384) (0.1806) (0.2123) LN (TA)t -2.6272*** -2.6131*** -1.7853*** -1.7428*** -1.4603*** -1.4334*** -0.6776** -0.6314**
(0.0002) (0.0002) (0.0008) (0.0010) (0.0006) (0.0008) (0.0242) (0.0355) RDt-1 0.4280*** 0.4268*** 0.4836*** 0.4794*** 0.5712*** 0.5634*** 0.6058*** 0.5950***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) ROAt -0.2912*** -0.2936*** -0.2408*** -0.2471*** -0.1954*** -0.2055*** -0.1787*** -0.1923***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Debt Ratiot -0.0047 -0.0152 -0.0251** -0.0322***
(0.7906) (0.2800) (0.0299) (0.0016)
Observations 671 671 1,072 1,072 1,483 1,483 1,865 1,865
Adjusted R2 0.4839 0.4832 0.4832 0.4833 0.5288 0.5300 0.5355 0.5377
Panel A.2 DID Regression Results for R&D Investment in the Inter-industry Analysis: Large Firms One Matched Firm Two Matched Firms Three Matched Firms Four Matched Firms
(1) (2) (1) (2) (1) (2) (1) (2)
Aftert -0.4008 -0.6233 -0.3073 -0.3891 -0.6343 -0.7051 0.1333 0.0782
(0.6559) (0.4819) (0.6832) (0.6025) (0.2851) (0.2302) (0.7997) (0.8808) Treatmenti -2.6529*** -2.5552*** -3.0679*** -3.0077*** -2.5046*** -2.4400*** -2.3140*** -2.2546***
(0.0006) (0.0007) (0.0001) (0.0001) (0.0002) (0.0003) (0.0005) (0.0006) Aftert × Treatmenti 2.5006** 2.8499*** 2.6654** 2.8517*** 2.6098*** 2.7881*** 2.2241** 2.3671***
(0.0201) (0.0073) (0.0118) (0.0067) (0.0058) (0.0030) (0.0162) (0.0099) LN (TA)t -1.1017*** -1.0102*** -1.0538*** -0.9993*** -1.0786*** -1.0165*** -0.9345*** -0.8857***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Panel A.2 DID Regression Results for R&D Investment in the Inter-industry Analysis: Large Firms One Matched Firm Two Matched Firms Three Matched Firms Four Matched Firms
(1) (2) (1) (2) (1) (2) (1) (2)
RDt-1 0.4383*** 0.4387*** 0.5237*** 0.5250*** 0.5836*** 0.5852*** 0.6254*** 0.6267***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) ROAt -0.1489*** -0.1511*** -0.1579*** -0.1590*** -0.1052*** -0.1061*** -0.1117*** -0.1124***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Debt Ratiot 0.0012*** 0.0012*** 0.0012*** 0.0012***
(0.0000) (0.0000) (0.0000) (0.0000)
Observations 834 834 1,312 1,312 1,768 1,768 2,225 2,225
Adjusted R2 0.5347 0.5493 0.5435 0.5509 0.5522 0.5609 0.5746 0.5813
Panel B.1 DID Regression Results for Adjusted Patent Citations in the Inter-industry Analysis: Small Firms
One Matched Firm Two Matched Firms Three Matched Firms Four Matched Firms
(1) (2) (1) (2) (1) (2) (1) (2)
Aftert 0.0054 0.0122 -0.0291 -0.0263 -0.0109 -0.0090 0.0692** 0.0716**
(0.8621) (0.7011) (0.3005) (0.3623) (0.6943) (0.7514) (0.0237) (0.0507) Treatmenti 0.0264 0.0391 -0.0351 -0.0307 -0.0311 -0.0294 0.0647 0.0705 (0.3675) (0.2263) (0.2923) (0.4067) (0.4123) (0.4848) (0.1531) (0.1642) Aftert × Treatmenti -0.0404 -0.0490 0.0311 0.0189 0.0236 0.0150 -0.0437 -0.0488 (0.2573) (0.2090) (0.4247) (0.6613) (0.5912) (0.7573) (0.4096) (0.4067) LN (1+NetSalest-1) 0.0052* 0.0055* 0.0083** 0.0089** 0.0125*** 0.0126*** 0.0304*** 0.0316***
(0.0780) (0.0831) (0.0168) (0.0180) (0.0016) (0.0030) (0.0000) (0.0000) RDt-1 0.0016*** 0.0017*** 0.0026*** 0.0026*** 0.0024*** 0.0024*** 0.0021*** 0.0021***
(0.0008) (0.0010) (0.0000) (0.0000) (0.0000) (0.0000) (0.0001) (0.0002)
Tobin’s Qt-1 -0.0018 0.0050 0.0040 0.0018
(0.6086) (0.2172) (0.3623) (0.7291)
Observations 496 467 798 755 1,097 1,040 1,370 1,302
Adjusted R2 0.0424 0.0424 0.1229 0.1275 0.1087 0.1144 0.1062 0.1094
Table 9 DID Regression Result of Inter-industry: Subsample Analysis for Different Firm Size (cont.)
Panel B.2 DID Regression Results for Adjusted Patent Citations in the Inter-industry Analysis: Large Firms
One Matched Firm Two Matched Firms Three Matched Firms Four Matched Firms
(1) (2) (1) (2) (1) (2) (1) (2)
Aftert -0.0592 -0.0542 -0.1010 -0.1033 -0.0849 -0.0884 -0.1498* -0.1533*
(0.6612) (0.6919) (0.3273) (0.3221) (0.2875) (0.2743) (0.0530) (0.0507) Treatmenti -0.2198* -0.1860 -0.1631 -0.1454 -0.0863 -0.0653 -0.0774 -0.0557 (0.0604) (0.1275) (0.1182) (0.1840) (0.3429) (0.4957) (0.4300) (0.5890) Aftert × Treatmenti 0.2257 0.2121 0.2477* 0.2510* 0.2144* 0.2166* 0.2586* 0.2550*
(0.1509) (0.1921) (0.0794) (0.0873) (0.0850) (0.0948) (0.0539) (0.0685) LN (1+NetSalest-1) 0.0815*** 0.0842*** 0.0907*** 0.0916*** 0.0841*** 0.0859*** 0.0985*** 0.0999***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) RDt-1 0.0088** 0.0094** 0.0111*** 0.0114*** 0.0108*** 0.0111*** 0.0113*** 0.0113***
(0.0378) (0.0469) (0.0002) (0.0003) (0.0000) (0.0000) (0.0000) (0.0000)
Tobin’s Qt-1 -0.0028 -0.0051 -0.0049 -0.0036
(0.7623) (0.5588) (0.5331) (0.6696)
Observations 545 536 840 824 1,132 1,107 1,433 1,402
Adjusted R2 0.1652 0.1639 0.1470 0.1470 0.1307 0.1318 0.1370 0.1377
Note: This table presents the panel regression results of subsamples divided by different firm sizes, including small and large firms in the inter-industry analysis. Panels A.1 and A.2 show the regression results that explain the R&D investment for small and large firms, respectively.
Panels B.1 and B.2 show the regression results that explain the adjusted patent citations for these two subsamples. The dependent variable in Panels B.1 and B.2 is LN (1+adjusted patent citation). The regression is shown in equation (1) of Section 3.3.4. Aftert = 1 if the firm is in the approval year or after approval year and 0 otherwise; Treatmenti = 1 if the firm is in that treated group and 0 otherwise. The treated firms are approved biopharmaceutical firms and control firms are high-tech firms. The definitions of the variables are presented in Appendix Table A1. Numbers in parentheses are p-values. ***,**, and * denote significance at the 1%, 5%, and 10% levels, respectively.
Panels B.1 and B.2 of Table 9 present the inter-industry subsample analysis of adjusted patent citations. There are insignificant coefficients of the interaction term for the small firms but marginally significant coefficients of the interaction term for the large firms. These results show that only in the large firm group, approved biopharmaceutical Table 9 DID Regression Result of Inter-industry: Subsample Analysis for
Different Firm Size (cont.)
firms are more motivated to improve their innovation quality by the Biopharmaceutical Act than high-tech firms.
Next, we divide the sample into low and high R&D intensity firms because the previous section shows that the Biopharmaceutical Act encourages biopharmaceutical firms with low R&D intensity to invest more in innovation. This additional inter-industry subsample analysis explores whether the Biopharmaceutical Act also has a consistent effect in encouraging low R&D intensity firms in the biopharmaceutical industry rather than the high-tech industry.
The coefficients of the interaction term, After×Treatment, are significantly positive in Panel A.1 of Table 10 but not significant in Panel A.2 of Table 10. These results show that among low R&D intensity firms, the approved biopharmaceutical firms are more encouraged to increase innovation investments by the Biopharmaceutical Act than high-tech firms. In addition, for the high R&D intensity group, after the Biopharmaceutical Act, the approved biopharmaceutical firms do not have significantly different R&D intensity than the high-tech firms. This finding, which shows that the approved biopharmaceutical firms are motivated more to increase R&D investment by the Biopharmaceutical Act than high-tech firms, is driven primarily by the group of low R&D intensity firms. Further, in both Panel B.1 and B.2 of Table 10, the insignificant coefficients of interaction term, After×Treatment, show that the Biopharmaceutical Act does not result in any difference in the innovation quality between the approved biopharmaceutical firms and high-tech firms in either the low or the high R&D intensity firm groups. This finding shows that for both low and high R&D intensity firms, the Biopharmaceutical Act does not lead to any difference in the innovation quality between the approved biopharmaceutical firms and high-tech firms.
Table 10 DID Regression Result of Inter-industry: Subsample Analysis for Different R&D Intensity Level
Panel A.1 DID Regression Results for R&D Investment in the Inter-industry Analysis: Low R&D Intensity Firms
One Matched Firm Two Matched Firms Three Matched Firms Four Matched Firms
(1) (2) (1) (2) (1) (2) (1) (2)
Aftert 1.7445 2.2429* 0.5603 0.7918 0.6222 0.7712 1.1356** 1.2328***
(0.1440) (0.0557) (0.4719) (0.3054) (0.2847) (0.1834) (0.0165) (0.0094)
Panel A.1 DID Regression Results for R&D Investment in the Inter-industry Analysis: Low R&D Intensity Firms
One Matched Firm Two Matched Firms Three Matched Firms Four Matched Firms
(1) (2) (1) (2) (1) (2) (1) (2)
Treatmenti -0.689 1.9281* -0.3714 1.2276 0.4937 1.5582** 0.6711 1.2455*
(0.5063) (0.0839) (0.6438) (0.1542) (0.4709) (0.0326) (0.2802) (0.0589) Aftert × Treatmenti 3.0262** 2.3428* 3.6293*** 3.3018*** 3.0472*** 2.8048*** 2.3919*** 2.2496***
(0.0351) (0.0964) (0.0010) (0.0025) (0.0011) (0.0025) (0.0042) (0.0071) LN (TA)t -0.4707 -0.4486 -0.7691*** -0.8312*** -0.5744*** -0.6115*** -0.5526*** -0.5702***
(0.2082) (0.2203) (0.0036) (0.0015) (0.0043) (0.0023) (0.0008) (0.0006) RDt-1 0.2557*** 0.2704*** 0.3018*** 0.3137*** 0.3344*** 0.3439*** 0.3853*** 0.3913***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) ROAt -0.2801*** -0.2495*** -0.1872*** -0.1680*** -0.1345*** -0.1221*** -0.1077*** -0.1001***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Debt Ratiot 0.0817*** 0.0517*** 0.0344*** 0.0179***
(0.0000) (0.0000) (0.0001) (0.0099)
Observations 720 720 1,094 1,094 1,482 1,482 1,857 1,857
Adjusted R2 0.3521 0.3798 0.307 0.3211 0.2875 0.2952 0.2857 0.2879
Panel A.2 DID Regression Results for R&D Investment in the Inter-industry Analysis: High R&D Intensity Firms
One Matched Firm Two Matched Firms Three Matched Firms Four Matched Firms
(1) (2) (1) (2) (1) (2) (1) (2)
Aftert -0.8044 -0.1764 -2.6868** -2.0907 -1.7973 -1.4135 -0.7549 -0.2784 (0.6610) (0.0557) (0.0470) (0.1209) (0.1074) (0.2012) (0.4485) (0.7778) Treatmenti -6.1900*** -6.8924*** -7.0898*** -8.1367*** -4.4423*** -5.9052*** -4.3913*** -5.8374***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0005) (0.0000) (0.0005) (0.0000) Aftert × Treatmenti 2.2862 1.478 4.3563** 3.4420* 3.4653** 2.7279 2.5106 1.6675 (0.2885) (0.4975) (0.0209) (0.0673) (0.0474) (0.1154) (0.1454) (0.3288) LN (TA)t -1.2869** -1.2031* -1.3909*** -1.4913*** -1.1939*** -1.2449*** -0.6463** -0.7218**
(0.0367) (0.0505) (0.0020) (0.0009) (0.0006) (0.0003) (0.0451) (0.0237) RDt-1 0.3676*** 0.3654*** 0.4727*** 0.4645*** 0.5548*** 0.5391*** 0.5476*** 0.5320***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Table 10 DID Regression Result of Inter-industry: Subsample Analysis for Different R&D Intensity Level (cont.)
Panel A.2 DID Regression Results for R&D Investment in the Inter-industry Analysis: High R&D Intensity Firms
One Matched Firm Two Matched Firms Three Matched Firms Four Matched Firms
(1) (2) (1) (2) (1) (2) (1) (2)
ROAt -0.3316*** -0.3457*** -0.2965*** -0.3131*** -0.2274*** -0.2510*** -0.2445*** -0.2682***
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Debt Ratiot -0.0522** -0.0781*** -0.0850*** -0.0904***
(0.0288) (0.0000) (0.0000) (0.0000)
Observations 690 690 1,124 1,124 1,538 1,538 1,968 1,968
Adjusted R2 0.4856 0.4885 0.5706 0.5775 0.5543 0.5639 0.5345 0.5455
Panel B.1 DID Regression Results for Adjusted Patent Citations in the Inter-industry Analysis: Low R&D Intensity Firms
One Matched Firm Two Matched Firms Three Matched Firms Four Matched Firms
(1) (2) (1) (2) (1) (2) (1) (2)
Aftert -0.0156 -0.0149 0.0173 0.0175 0.0174 0.0173 0.0137 0.0136
(0.5316) (0.9097) (0.3891) (0.3987) (0.2954) (0.3129) (0.2983) (0.3144) Treatmenti -0.004 -0.0058 0.0001 -0.0001 -0.0009 -0.0013 -0.0030 -0.0016 (0.8549) (0.8106) (0.9966) (0.9972) (0.9660) (0.9537) (0.8691) (0.9364) Aftert × Treatmenti 0.0386 0.0384 -0.0056 -0.0043 -0.0030 -0.0010 -0.0042 -0.0024 (0.2095) (0.2447) (0.8509) (0.8918) (0.9131) (0.9726) (0.8652) (0.9269) LN (1+NetSalest-1) 0.0047* 0.0047 0.0040 0.0038 0.0034 0.0028 0.0003 -0.0004 (0.0813) (0.1193) (0.1419) (0.2163) (0.1944) (0.3300) (0.8787) (0.8551)
RDt-1 0.0005 0.0004 0.0015* 0.0014 0.0012 0.0011 0.0008 0.0008
(0.5052) (0.6346) (0.0732) (0.1043) (0.1440) (0.2048) (0.2234) (0.2922)
Tobin’s Qt-1 0.0012 -0.0007 -0.0013 -0.0023
(0.7205) (0.8421) (0.6979) (0.4156)
Observations 632 592 968 922 1,320 1,261 1,658 1,589
Adjusted R2 0.0245 0.0246 0.0155 0.0139 0.0213 0.0206 0.0166 0.0166
Table 10 DID Regression Result of Inter-industry: Subsample Analysis for Different R&D Intensity Level (cont.)
Panel B.2 DID Regression Results for Adjusted Patent Citations in the Inter-industry Analysis: High R&D Intensity Firms
One Matched Firm Two Matched Firms Three Matched Firms Four Matched Firms
(1) (2) (1) (2) (1) (2) (1) (2)
Aftert 0.0038 0.0054 -0.0954* -0.0915* -0.046 -0.0401 -0.1004** -0.0932**
(0.9344) (0.9097) (0.0711) (0.0811) (0.2955) (0.3613) (0.0114) (0.0183) Treatmenti -0.0734* -0.0779* -0.1178* -0.1163* 0.0115 0.0152 -0.0373 -0.0338 (0.0787) (0.0878) (0.0528) (0.0685) (0.8424) (0.8069) (0.5114) (0.5745) Aftert × Treatmenti 0.0187 0.0196 0.0979 0.0698 0.0350 -0.0030 0.0873 0.0378 (0.7497) (0.7517) (0.2290) (0.4027) (0.6501) (0.9706) (0.2593) (0.6360) LN (1+NetSalest-1) 0.0076* 0.0074 0.0195*** 0.0196*** 0.0358*** 0.0361*** 0.0322*** 0.0328***
(0.0876) (0.1341) (0.0021) (0.0038) (0.0000) (0.0000) (0.0000) (0.0000)
RDt-1 0.0015** 0.0013** 0.0006 0.0003 0.0008 0.0005 0.0016** 0.0013*
(0.0156) (0.0486) (0.4083) (0.6658) (0.2572) (0.5157) (0.0135) (0.0524)
Tobin’s Qt-1 0.0002 0.0081* 0.0102** 0.0142***
(0.9483) (0.0696) (0.0216) (0.0015)
Observations 598 558 986 929 1,368 1,290 1,754 1,663
Adjusted R2 0.0638 0.0696 0.0729 0.081 0.1151 0.1256 0.0884 0.0984
Note: This table presents the panel regression results of subsamples divided by different firm sizes, including small and large firms in the inter-industry analysis. Panels A.1 and A.2 show the regression results that explain the R&D investment for low and high R&D intensity firms, respectively. Panels B.1 and B.2 show the regression results that explain the adjusted patent citations for these two subsamples. The dependent variable in Panels B.1 and B.2 is LN (1+adjusted patent citation). The regression is shown in equation (1) of Section 3.3.4. Aftert = 1 if the firm is in the approval year or after approval year and 0 otherwise; Treatmenti = 1 if the firm is in the treated group and 0 otherwise. The treated firms are approved biopharmaceutical firms and control firms are high-tech firms. The definitions of the variables are presented in Appendix Table A1. Numbers in the parentheses are p-values. ***,**, and * denote significance at the 1%, 5%, and 10% levels, respectively.
5. Conclusion
This study investigates the impact of the Biopharmaceutical Act on firm innovation.
To overcome the endogeneity problem, we first use the PSM approach to identify suitable control firms and then adopt the DID approach to examine how the innovation activities of approved biopharmaceutical firms, relative to control firms, respond to the exogenous Table 10 DID Regression Result of Inter-industry: Subsample Analysis for
Different R&D Intensity Level (cont.)
shock of the Biopharmaceutical Act. To demonstrate the benefits and policy effectiveness of the Biopharmaceutical Act, we conduct both intra-industry and inter-industry analyses.
The results of the intra-industry analysis show that the Biopharmaceutical Act induces the approved biopharmaceutical firms to increase innovation investments. This finding is consistent with most previous studies which find a positive effect of tax credits on R&D.
The stimulation effect of the Biopharmaceutical Act on the innovation investments in the biopharmaceutical industry only exists among pharmaceutical and low R&D intensity firms. The subsample findings may have the similar economic implication since Yang et al. (2012) find that pharmaceutical firms usually have low R&D intensity. Pharmaceutical firms tend to underinvest more in R&D than non-pharmaceutical firms because of the high risk and fewer successful cases in new medicine, the long period required for innovation, and the substantial investment necessary. Low R&D intensity firms are more likely to underinvest in R&D. Therefore, these findings for pharmaceutical and low R&D intensity firms demonstrate the effectiveness of the Biopharmaceutical Act for firms with more serious underinvestment in R&D.
In addition, the inter-industry analysis supports the policy effectiveness of the Biopharmaceutical Act. The approved biopharmaceutical firms are motivated more to invest innovation and to improve innovation quality than high-tech firms. By investigating the SUI of Taiwan, Yang et al. (2012) find that the tax credits have more effect on R&D for industries with greater R&D intensity and suggest that the government should establish various tax credits. Therefore, our results support the argument of Yang et al. (2012) and confirm the effectiveness of the Biopharmaceutical Act, which grants the additional benefit of tax credits only for biopharmaceutical firms, while the SUI grants all industries the same preferential tax treatment.
Further, the results of the inter-industry analysis are dominated by low R&D intensity firms and small firms. These groups are more likely to suffer severe R&D underinvestment problems. Small firms often find it more difficult to appropriate the private returns of R&D and lack the physical assets to serve as collateral (David et al., 2000; Hall, 2002). These subsample results strengthen our finding that the policy effectiveness of Biopharmaceutical Act is greater for firms with more serious R&D underinvestment problems. This finding is also consistent with the contention of Baghana and Mohnen (2009) and Lokshin and Mohnen (2012) that tax credit policy tends to be more effective in stimulating the R&D investment for small firms than large firms.
Based on prior literature, if the government wants to have a stronger effect on R&D,
it should adopt tax credits rather than direct subsidies. In fact, the Biopharmaceutical Act primarily uses tax credits, which appears to be a good decision. After the empirical examination, we confirm the policy effectiveness of the Biopharmaceutical Act, especially for biopharmaceutical firms with more serious R&D underinvestment problems. In addition, studies show that the stimulating effect of tax credits is more rapid than that of direct subsidies (David et al., 2000). Our finding of a short run effect for the Biopharmaceutical Act confirms this result.
In conclusion, our empirical findings of the promoting effect of the Biopharmaceutical Act on innovation activities, support the theories regarding private R&D underinvestment.
This Act offers tax credits for R&D investment and holding shares of biopharmaceutical firms, and grants tax credits to the top executives and technology investors for new shares in biopharmaceutical firms. The tax credit regulations reduce the cost of R&D investment and increases equity financing opportunities. Thus, the increasing innovation activities resulting from tax credit regulations support the financial constraint theory in explaining the problem of R&D underinvestment.
In addition, the Biopharmaceutical Act offers non-tax credit preferential treatments to reduce the agency problem (i.e. it stimulates managers’ motivation) and to increase the incoming spillover effect by increasing cooperation opportunities. These non-tax credit treatments tend to support the agency theory and spillover theory in explaining R&D underinvestment. In this paper, discriminating between the tax credits and non-tax credits
In addition, the Biopharmaceutical Act offers non-tax credit preferential treatments to reduce the agency problem (i.e. it stimulates managers’ motivation) and to increase the incoming spillover effect by increasing cooperation opportunities. These non-tax credit treatments tend to support the agency theory and spillover theory in explaining R&D underinvestment. In this paper, discriminating between the tax credits and non-tax credits