Difference-in-differences test for micro effect of technological finance cooperation pilot in China

This paper takes 2011 and 2016 technological finance pilots in China as exogenous events and selects listed companies in the two pilot regions/cities as the experimental group. The listed companies in the non-pilot area are the control group. The changes in the corporate value of these two groups of companies before and after the pilot are analysed. The periods from 2008 to 2016 and from 2011 to 2019 are used as the sample periods for two technological finance pilots to ensure the reliability of the research conclusions. This paper excludes ST and PT companies. Continuous variables are processed with 1% upper and lower tailings, and finally, a total of 30,083 company annual observation values are obtained in two batches of technological finance pilots. The data in this paper come from CSMAR, WIND and RESSET databases.

The DID method is a commonly used method to evaluate the effect of policy implementation. The technological finance pilot can be regarded as a quasi-natural experiment. The value of the enterprises in the pilot and non-pilot areas may change before and after the pilot. By analysing the changes before and after the pilot, the impact of the technological finance pilot on corporate value can be derived. Considering that technological finance pilots may be non-random and there is a self-selection problem, this paper draws on the existing research methods of technological finance pilot effects, uses PSM-DID [5,6,7], and further conducts empirical tests to eliminate sample selection deviations and reverse causality to make sure the robustness of the results. The DID method was used to construct a model (1) to test the impact of technological finance pilots on corporate value. (1) Valueijt=βi0+βi1DIDit+βControlijt+λt+μj+εit {\rm{Valu}}{{\rm{e}}_{ijt}} = {\beta _{i0}} + {\beta _{i1}}{\rm{DI}}{{\rm{D}}_{it}} + \beta {\rm{Contro}}{{\rm{l}}_{ijt}} + {\lambda _t} + {\mu _j} + {\varepsilon _{it}}

Among them, the subscripts i, j and t represent the pilot batch, enterprise and time of technological finance, respectively. Value_ijt, the explained variable – corporate value, is expressed with TobinQ value, drawing on He Ying et al. [11]. DID_it is used to explain the implementation of technological finance pilot policies, which is the crossover term (DID_it = Treat_it × Imp_it) of the dummy variable (Treat_it) of the enterprise experimental group or control group and the policy implementation dummy variable (Imp_it). If the city where the sample company is located is the first (the second) batch of pilot cities of technological finance, the value of Treat_it is 1, otherwise it is 0. The value of Imp_it is 1, otherwise it is 0 after the first (second) pilot policy is implemented. Control_it is a set of control variables that affect the corporate value. Based on existing literature [12,13,14], this paper selects the following variables: RI (corporate risk-taking), Size (corporate Size), Roe (Return on Net Assets), Nprs (Net Sales Interest Rate), Tat (Turnover of total assets), Beta (market Beta) and Eps (Earnings per share); and Table 1 shows definition and measurement of these variables. λ_t is the time fixed effect, μ_j is the firm fixed effect, and ɛ_it is the random disturbance item. This paper focuses on the value of the coefficients β_i1 and their significance.

Table 2 reports the single factor test results of the main variables of two technological finance pilots. Panel A of Table 2 shows that after the first technological finance pilot, the value (mean) of the enterprises in the pilot areas has increased significantly, while the value (mean) of the enterprises in the non-pilot areas has not increased significantly. Panel B of Table 2 shows that after the second technological finance pilot, the value (mean, median) of the enterprises in both the non-pilot and pilot areas of technological finance has shown a significant downward trend. Obviously, the first (second) technological finance pilot had a significant positive (negative) impact on corporate value.

Table 3 reports the regression results of the model (1). The regression results in Panel A of Table 3 show the impact of the first technological finance pilot on corporate value. The year and corporate fixed effects are controlled in the regression results. It can be seen from Panel A of Table 3 that the models that use Value 1 and Value 2 to measure corporate value in columns (1) and (2), the DID regression coefficients are 0.129 and 0.140 at the 1% level, respectively, indicating that the first technological finance pilot has improved the value of the enterprises in the pilot areas. Panel B of Table 3 shows the impact of the second technological finance pilot on corporate value. The DID regression coefficients in columns (3) and (4) of Panel B in Table 3 are −0.101 and −0.130 at 10% level, respectively, which indicate that the second technological finance pilot has reduced the value of enterprises in the pilot area. From the perspective of control variables, the correlation between corporate value, risk-taking and Size, Roe, Tat, Beta, and Eps is basically consistent with the existing research. It can be seen that after controlling other factors, the first (second) technological finance pilot significantly increased (reduced) corporate value.

The premise of DID is that the common trend hypothesis is established, that is, the experimental group and the control group have the same changing trend before the policy is implemented. To test this hypothesis, this paper uses regression models and common trend graphs to verify. To reflect the dynamic effects of technological finance pilots on corporate value, the cross-product term (Treat × Year) of corporate dummy variables and time dummy variables in the technological finance pilot areas are used as explanatory variables, and control variables are added to construct a regression model. The regression results are shown in Table 4.

Panel A of Table 4 shows that the regression coefficients of the years before (pre_2, pre_1), current year (current) and 1 year after (post_1) the first technological finance pilot are not significant at 10% level, that is, before the implementation of the first pilot policy, there is no significant difference between the experimental group and the control group. The regression coefficients of every year after the first technological finance pilot (post_2, post_3, post_4, post_5) are significant, indicating that the impact of the first technological finance pilot on the value of the enterprise is lagging, and the lag period is 1 year. Panel B of Table 4 shows that the regression coefficients of the years (pre_4, pre_3, pre_2, pre_1) before the second technological finance pilot are not significant at 10% level, that is, before the second pilot, there is no significant difference between the experimental group and the control group. The regression coefficients of the current year (current) and the following 2 years (post_1, post_2) of the second technological finance pilot are significant at 5% level, indicating that the second technological finance pilot has a direct impact on the value of the enterprise, and there is no lag period, and the impact duration lasts for 2 years.

Figure 1 shows the common trend test chart of the enterprise value in the areas of two technological finance pilots. Panel A of Figure 1 shows the changing trend of the corporate value of the first pilot. Before the first pilot in 2011, the change curve of the corporate value of the control group and the experimental group was close to overlapping, the trend was roughly the same, and the difference was not obvious. After the first pilot, the two groups still had the same trend, but the growth rate of the corporate value of the experimental group was significantly greater than that of the control group with an obvious difference, and the parallel trend assumption was satisfied. Panel B of Figure 1 shows the changing trend of corporate value in the second technological finance pilot. Before the second pilot in 2016, the change curve of the corporate value of the control group and the experimental group was close to overlapping (without considering the occurrence of the stock market crash in 2015), and the changing trend was basically the same without an obvious difference. But after the implementation of the second pilot policy, the trends of the two groups were still the same, but the decline in corporate value of the experimental group was significantly greater than that of the control group, the difference was obvious, and the parallel trend hypothesis was satisfied. Above all, the common trend test of the experimental group and the control group of two technological finance pilots is established. The DID can be used to estimate the implementation effect of two technological finance pilots, which shows that the conclusions of this paper are robust.

Fig. 1

Based on the quasi-natural experiments of two technological finance pilots in China, this paper constructs DID and PSM-DID models based on the data of Shanghai and Shenzhen A-share listed companies from 2008 to 2019, to study the effect of the technological finance pilots on corporate value. This study found that the first technological finance pilot improved corporate value in the pilot area, while the second technological finance pilot led to a decline in corporate value in the pilot area. Following are the reasons for the different effects of two technological finance pilots.