General trends in the performance of Chinese Fintech entities and banks: 2013–19
Figure 1 reveals a significant decline in Fintech profitability over the 2013–2019 study frame. At the open of the period, Fintech entities’ ROA levels were stronger than those for traditional banks. By 2019 the situation had reversed. Through their engagement in riskier activities (P2P lending, for example), and the low regulatory compliance costs imposed on virtual banks during the early years of the sample period, Fintech entities initially outperformed traditional banks. However, by sample period end, Fintech lenders registered adverse ROA and ROE numbers (of -0.46 and -3.09), and major credit losses (with NPL and SML ratios of 4.93% and 5.61%).
Partially due to Fintech competition, ROA levels also declined for traditional banks over the period (see Fig. 1a). Within the traditional bank group, state-owned banks achieved the highest ROA. Nonetheless, all traditional bank sectors were subject to a declining ROA trend before stabilizing in 2017/8. ROE numbers also highlight a negative trend (see Fig. 1b). Due to Fintech entities’ strong equity issuance (Caixin 2020), traditional banks outperformed in ROE terms. Within the latter group, joint-stock commercial banks achieved the highest ROE levels.
Figure 1c reveals a steady upturn in NPLs in all traditional bank sectors during the course of the seven-year study period. Fintech entities experienced much greater variation in NPL rates, with a marked increase evident between 2015 and 2018. Figure 1d highlights SML trends. A steady initial increase is evident for traditional banks between 2013 and 2015 with indications of stabilization and improvement in the years thereafter. In contrast, Fintech entities experienced strong and volatile SML increases over the period. As today’s SMLs may become tomorrow’s NPLs, such trends sound a cautionary note on Fintech entities’ rising credit risks.
The marked difference in traditional and virtual banks’ prudential performance is indicative of segmentation in the respective entities’ client bases. Results are consistent with traditional banks’ retention of more conservative, risk-averse clients; and Fintech entities’ appeal to a more diverse, “techno-savvy” group of younger investors. Significant wealth differences are also likely between the client groups. Quantification of such segmentation effects awaits further inquiry [See earlier discussions and Endnote 3]. However, such investigation lies outside the scope of the present study.
Comparison of traditional bank and Fintech entities’ descriptive statistics
Building on background trends captured in Fig. 1, Table 2 reports descriptive statistics on the four performance measures (ROA, ROE, NPL and SML). Panel A presents descriptive statistics for traditional bank lenders. The remaining parts of Table 2 report performance statistics for lenders from state-owned (Panel B), joint stock commercial (Panel C), city commercial (Panel D), rural commercial (Panel E) and Fintech (Panel F) sectors.
For 2013–2019, profitability is greater among traditional banks than Fintech firms. Fintech entities register mean negative returns on both asset and equity (ROA and ROE). In contrast, traditional bank levels are 0.96% and 1.47% (see Table 2, Panel B). Additionally, Fintech companies’ non-performing loan (NPL) rates are more than three times those of traditional banks. Special mention loans (SML) follow a similar pattern, with rates twice as high for Fintech entities.
Within the group of traditional banks, state-owned banks (SBs) rank first in terms of ROA and second, behind joint-stock commercial banks (JSCBs), in terms of ROE. However, NPL rates are higher for these two groups compared to city commercial (CCBs) and rural commercial banks (RCBs). The higher NPL ratios suggest that SBs and JSCBs engage in riskier activities. One explanation is the “too big to fail effect” (Fan et al. 2012; Liang et al. 2014), inducing large banks to accept greater risk. However, Wang et al. (2018) demonstrate that the return “volatility connectedness” is lower among SBs than JSCBs.
SBs’ proximity to the state may nonetheless induce pressure to support strategically important but non-profit making industries. Adverse selection risk may also be higher if SOE borrowers’ executive compensation schemes incentivize non-disclosure of risk factors (Xu et al. 2014).
Table 3 presents summary statistics of explanatory variables. Panels A-E and F report respective statistics for traditional banks and Fintech lenders. Credit risk, as captured by LLP, is much higher for Fintech companies than for traditional banks (7.62% vs 3.06%). Indeed, regulations require Fintech companies to set aside more funds to cover credit risks (Caixin 2020). This requirement reflects Fintech entities’ strong growth orientation, as compared to traditional banks’ value focus. The gap in price-to-book (MB) valuations (19.76 versus 0.82) characterizes this difference. Moreover, many Chinese banks experienced MB ratios below 1.0 during the period (Cai 2014). Within the traditional bank group, larger banks, notably JSCBs and SBs display the lowest LLP ratios. Such banks’ greater access to state capital also mitigates the need for significant loan loss provisions. RCBs are the least risky traditional banks in terms of LLP ratios.
Liquidity ratios (= liquid assets/total assets) are notably higher for Fintech lenders compared to traditional banks (31.36% vs 14.20%). Such difference likely reflects Fintech entities’ preference for cash holdings to cover near-term technology and R&D expenditures. In contrast, traditional banks place greater reliance on longer-term investment. Within the traditional bank groups, JSCBs have the lowest liquidity ratio (12.54%) and SBs the highest (16.90%).
Relative to Fintech entities, traditional banks have a much lower equity to total assets ratio, EA (46.44% vs 6.99%). This difference reflects Fintech organizations’ strong reliance on external capital funding. Within the traditional bank group, uniform EA levels (of around 7%) are apparent. Similarly, Fintech entities’ operational risk (Ct/Ct-1) dwarfs that of traditional banks (2.33 vs 1.15). High recurrent costs in running digitalization and ecommerce platforms likely underlie this difference. Software/hardware costs, R&D expenditures and ecommerce access and distribution fees likely drive this difference. Among traditional banks, SBs (RCBs) display the lowest (highest) year-on-year increase in operating costs 1.07 (1.25). In sum, Ct/Ct-1 flags the importance of operating efficiency in relation to a lender’s profitability (Fang et al. 2019). Nonetheless, Wang et al (2021) report efficiency gains (in relation to factor productivity) from Chinese banks’ adoption of Fintech, and most notably in the area of Big Data.
Fintech entities also embrace greater market risk. Major differences in one-day value loss exposure exist between traditional bank and Fintech entities (77.96 vs 5.70). Among traditional banks, SBs (RCBs) exhibit by far the highest (lowest) VaR levels.
Table 3 reveals notable difference in TimeList values between traditional bank and Fintech entities. Traditional banks’ dividend pay-out ratios are materially higher (23.67% vs 8.00%). Such difference reflects Fintech entities’ reinvestment of profit into growth options. Within the traditional bank sector, SBs distribute the most generous dividends on average (30.53%). This outcome reflects SOEs’ greater inclination to distribute profit (McGuinness et al 2015). Welker et al. (2017) report regulatory incentives to distribute dividends for Chinese issuers planning equity issues. Forti and Schiozer (2015) identify the importance of dividend signaling post-GFC.
Table 4 reports descriptive statistics on macroeconomic and industry variables. The greatest variation is evident in market cap to GDP. A number of factors account for stock return volatility in China. These include limits on arbitrage (Gu et al. 2018), oil prices (Wen et al. 2019; Wang et al. 2019; and Xu et al. 2019), share reforms (Tan 2016) and short-selling restrictions (Xiong et al. 2017). Variation in inflation, interest rate spreads and reserve ratio requirements appear small in comparison. Such a picture reflects a period of relative stability in monetary policy (Chen et al. 2017). Table 4 also reveals major change in the number of Fintech entities, highlighting the evolving membership of China’s digital community (Leong et al. 2017).
Tables 15 and 16 report bivariate correlations for bank and Fintech groupings.
Fixed effects estimation results for Chinese banks
Table 5 presents empirical results from fixed effects regressions (without interaction terms) for the 40 Chinese listed banks. Hausman tests reveal a significant difference (at the 1% level) in coefficients between fixed and random effects models in all four regression specifications. Accordingly, we deem a fixed effects approach appropriate. Additionally, F-test statistics suggest the rejection of the null hypothesis of no joint significance (at the 1% level) in all four equations.
Results in Table 5 highlight a significant inverse association between LLP and bank profitability (ROA and ROE). Findings also suggest that rising credit risk quickly impinges on NPL positions. All of these outcomes offer support for Hypothesis 1. Consistent with the notion of liquidity as an expense, the coefficient on LQ is also negative and significant in ROA and ROE equations. Such findings support the contention in Hypothesis 2 that increased loan exposure (and illiquidity) induces higher bank profitability. According to Tan (2016), such an outcome is congruent with Chinese banks’ effective loan management. Results for NPL and SML equations in Table 5 suggest a significant positive relation between liquidity and prudential bank performance.
Results in the ROA and ROE equations of Table 5 indicate a significant positive link between bank profitability and the ratio of total equity to total assets (EA). As capital risk bears inverse relation to EA, findings support the contention in Hypothesis 3. In effect, banks’ robust capital position protects against unexpected losses (Athanasoglou et al. 2008). However, the significant positive coefficient on EA in NPL and SML regressions suggests banks raise capital to protect against an anticipated decline in prudential performance.
Table 5 results indicate that operational risk (Ct/Ct-1) has significant impact on SML and not NPL positions. Accordingly, there is only modest backing for Hypothesis 4 in relation to prudential bank performance. The coefficient on value-at-risk (VaR) is significant and negative in NPL and SML regressions. Such results offer support for the contention in Hypothesis 5 of a positive association between market risk and prudential performance. The estimated VaR coefficient is, nonetheless, small in all Table 5 regressions.
With regard to our sixth hypothesis, Table 5 reveals an inverse but insignificant association between dividend payout ratios (DPOR) and bank profitability. However, lower dividend payout, and thus greater reinvestment of profit, correlates with growing NPL concerns.
We also find support for the premise in Hypothesis 7 that performance deteriorates post-IPO (TimeList). The association is negative in ROA and ROE equations and positive in NPL equation. Results are consistent with banks listing when profit and asset quality are at near-term highs.
The dummy for foreign strategic equity holdings (ForStr) has a positive and significant relation with NPLs in Table 5. This outcome runs counter to Hypothesis 9. It reveals lower asset quality in bank entities with strategic foreign equity ownership. Luo et al. (2017) report that foreign banks’ establishment of branch networks in China instills efficiency gains in domestic banks. Results herein suggest counter effects when foreign parties establish equity stakes in domestic banks.
The impact of state ownership on profitability is also important. In terms of both ROA and ROE, state-owned commercial (SB) banks exhibit stronger profitability than JSCB, CCB and RCB entities. Results contrast with earlier evidence in Herrero et al. (2009) of enhanced JSCB performance. Findings on bank ownership in the Fintech era also contrast with results in Tan (2016), where city commercial banks yield a performance advantage. Ownership effects are thus very much time or study period dependent. The Fintech era of 2013–2019 offers important contrast with earlier studies. Our results for the Fintech era suggest greater profit resilience in large state-owned banks. Moreover, we surmise that SBs’ funding and political connections conferred major advantage during the recent period of Fintech disruption. The listing of SBs since 2005 has also helped in galvanizing banks’ governance and commercial lending policies (Jia 2009).
When examining NPL and SML performance indicators, we observe a negative relation with each of the three ownership dummies (SB, JSCB and CCB). Relative to rural banks, higher asset quality is evident in the three bank sectors. Consistent with He et al. (2017), state banks (SBs) demonstrate greater NP stability. Instructively, the SB dummy coefficient is larger than that for either JSCB or CCB dummies in Table 5 NPL regressions. Accordingly, results offer support for the contention in Hypothesis 8 of stronger financial and prudential performance in state-held banks.
For control variables, MB is strongly significant in all four equations in Table 5. The coefficient on MB is positive (negative) in ROA and ROE (NPL and SML) equations. Results suggest that growth options promote profitability and asset quality. Similarly, bank size (S) is significant in three out of four equations. Larger banks secure higher asset quality (through lower SML rates) and enhanced profitability. The positive impact of bank size on profitability supports an economies of scale narrative, and the attendant “too-big-to-fail” argument (Goddard et al. 2004b; Iannotta et al. 2007). Bank size also functions as a surrogate for political connections (Garcia-Herrero et al. 2009).
Variable Fintech exerts significant negative effect on bank profitability. Such an outcome underscores this study’s central premise: Fintech competition erodes traditional bank profitability. Evidence herein is consistent with Phan et al. (2019). Results in Column 4 of Table 5 indicate that increased Fintech competition correlates with rising (and thus weaker) SML positions. Results also point to a significant positive relation between capital market maturity (CapMkt) and prudential performance. This outcome extends the narrative, in Tan (2016, p. 92), by which equity market development supports bank financial performance. Borrower stock price reactions to loan announcements also inform on possible expropriation within the debtor entity (Huang et al. 2012). Such signals are increasingly important in light of China’s growing number of listed companies.
We also control for China’s reserve ratio requirement (RRR). Interestingly, RRR is increasing (decreasing) in bank profitability (asset quality) in Table 5. Additionally, and as in García-Herrero et al. (2009) and Tan (2016), a positive link exists between inflation (INF) and bank profitability. A narrowing of lending and funding margins (LIR-SIR) also weakens banks’ ROA and loan quality.
Table 6 reports fixed effects estimation results with interaction effects. Again, Hausman test results offer support for a fixed effects model approach in all four specifications. Significant interaction effects exist between risk variables and Ln (Fintech). The negative coefficient on Var* Ln (Fintech) suggests magnification of the positive effect of market risk on prudential performance (i.e., lower NPL and SML rates) as Fintech competition intensifies. In contrast, the positive effect of market risk on profitability (ROA and ROE) weakens with an increasing number of Fintech firms. Additionally, the adverse effect of credit risk (LLP) on profitability weakens with Fintech competition. The negative effect of liquidity (LQ) on ROE also weakens with Fintech competition, whereas the positive impact of liquidity on prudential performance intensifies with it. The opposite result is found for the capital variable (EA), where the positive association between capital and ROE weakens when Fintech competition intensifies. The positive and significant effect of EA*Fintech on SML rates suggests intensification of the negative association between excess capital and prudential performance during periods of rising Fintech numbers.
GMM estimation results for Chinese banks
Table 7 reports GMM estimation results within a dynamic model setting featuring lagged performance (but excluding interaction effects).Footnote 27 The magnitude of the coefficient on prior period performance is not only insignificant in two of the four regressions but also close to zero. Consistent with a competitive banking sector, performance effects are not persistent. Our results on this issue are broadly consistent with earlier findings in Tan (2016).
Overall, GMM and fixed effects results appear similar. Consistent with Hypothesis 1, higher credit risk inhibits profitability and asset quality. Table 7 regressions also signal some support for Hypothesis 2. Specifically, greater bank liquidity weakens profitability but supports prudential (loan quality) performance. Similar to Table 5, support for Hypothesis 3 is positive in relation to bank profitability but negative in terms of prudential effects. In particular, Table 7 results reveal that capital (EA) supports ROA and ROE performance but weakens loan quality (given rising to higher NPL and SML rates). Similar to fixed effect estimations, increased operational risk impairs asset quality (in relation to SML rates). As with Tables 5, 7 reveals that greater market risk improves NPL and SML rates. It also improves ROE bank performance. Such outcomes support Hypothesis 5. Again, the magnitude of estimated VaR coefficients are small.
Findings in Table 7 also support Hypothesis 6. Results suggest that lower dividend payout supports profitability. However, banks’ greater reinvestment of profit appears to weaken SML positions. GMM regressions also confirm the positive association between foreign strategic investor presence and performance (in terms of ROA and SML) In contrast, strategic foreign equity ownership correlates with weaker prudential performance (through higher NPL rates). Similar to fixed effects estimations, state-owned lenders (SBs) register stronger bank profitability. Based on the size of estimated coefficients, SBs are also more resilient in prudential terms (especially in relation to SMLs). As with Table 5, results in Table 7 reveal that financial and NPL performance weaken with time since IPO. Findings thus offer broad support for Hypothesis 8. Table 7 results suggest bank profitability and asset quality are increasing in MB and total assets (S).
Industry-specific and macroeconomic effects noted earlier remain intact. In particular, the rising number of Fintech firms has negative impact on banks’ ROE performance. Table 7 also confirms a positive and significant relationship between capital market maturity and prudential performance (as shown by the negative coefficient on CapMkt in NPL and SML regressions). As corroboration of Tan (2016, p. 92), capital market maturity also supports bank profitability. Table 7 results also show that narrower interest rate spreads (LIR-SIR) weaken ROE profitability.Footnote 28
Table 8 regressions incorporates risk factor interactions with number of Fintech firms.Footnote 29 First, the negative impact of credit risk on ROE performance becomes less pronounced with increased Fintech competition. Stronger Fintech competition also exacerbates traditional banks’ SML rates. Second, as Fintech numbers rise, the adverse effect of bank liquidity on profitability weakens and the negative effect on NPL rates intensifies. Accordingly, greater Fintech competition strengthens the positive impact of liquidity on asset quality (in view of lower NPL rates). Third, Fintech competition weakens the positive link between capital and profitability. In contrast, Fintech competition supports the role of equity capital in arresting SML rates. Fourth, Fintech interaction with operational risk results in lower NPLs. In GMM regressions, rising Fintech competition accentuates the positive impact of operational risk on ROA. Finally, as Fintech numbers rise, the positive relation between market risk and loan quality strengthens.
Fixed effect and GMM estimation results for Chinese Fintech entities
In a further stage of analysis, we investigate the determinants of financial and prudential performance for the group of 25 exchange-listed Chinese Fintech companies featured in this study. Tables 9 and 10 report relevant regression results. A number of broad and consistent themes emerge in relation to both fixed effect (Table 9) and GMM (Table 10) estimation approaches.
In relation to prudential performance, the important risk factors are LLP and VaR. Moreover, the direction of association is consistent with the underlying postulates in Hypotheses 1 and 5. For financial performance, the strongest associations with risk factors are evident in GMM results (Table 10). In particular, higher equity capital (EA) and market risk (VaR) levels strongly underpin financial performance. These outcomes provide important support for Hypotheses 3 and 5.
In contrast to the traditional bank group, performance in the Fintech sphere bears little to no association with dividends. Consistent with Hypothesis 6, this result reflects online lenders’ overarching need for cash for investment in growth options. Due, perhaps, to the very short period of exchange listing of all 25 Fintech entities, variable Timelist displays weak association with all performance measures. As a consequence, there is little to no support for Hypothesis 7 in respect to online lenders. This outcome contrasts with the strong support evident for banks.
As with traditional banks, asset size matters. Results in Tables 9, 10 reveal greater resilience in larger Fintech outfits (given a significant positive relation between S and each of ROA and ROE). However, asset quality bears inverse relation with online lender size (i.e., NPL and SML rates are increasing in S). Results also reveal that greater capital market development (CapMkt) is beneficial to Fintech firms in respect to financial and prudential outcomes. Results in this area appear stronger for Fintech than traditional banks. This finding most likely reflects Fintech entities’ greater dependence on external funding channels, most especially in respect to equity markets. Results in this area offer important development of Tan’s (2016) findings on the role of equity market development in supporting Chinese bank profitability during the pre-digitalization era.
A range of other important effects are also evident in Tables 9, 10. In particular, Fintech entities with higher MB valuations significantly underperform in ROE terms. Second, rising Fintech competition significantly erodes the ROA performance of online lenders. Third, online lenders’ financial and prudential performance significantly weakens as the differential between long- and short-term interest rates narrows. Such results highlight Fintech entrants’ particular vulnerability to macroeconomic and regulatory changes impacting term structure.
In a final area of analysis, we consider within-sample differences in traditional banks’ Fintech proficiencies. While virtually all of China’s major lenders have developed Fintech expertise, only six within the traditional bank sample had developed “wholly-owned fintech” entities (Lee 2019) within the 2013–19 study period. We consider this group of six separately via dummy HP6 (= 1 for banks with High Proficiency in Fintech). As with Lee (2019), we regard this cohort as Fintech leaders among traditional banks. Nine banks within the co-joined State Bank (SB) and Joint Stock Commercial Bank (JSCB) domain reside outside this group of six. Dummy LP9 (= 1 for banks with Lower Proficiency Fintech) characterizes such lenders. The third dummy (CCB) captures City Commercial banks. China Banking News (2019) reports that municipal banks, which include CCBs, as well as other regional lenders, generally suffer from one or more deficiencies in relation to core Fintech capabilities. Consequently, we categorize CCBs into a third group in which Fintech capabilities generally rank behind the LP9 group. The omitted dummy (= 3 banks) is for Rural Commercial Banks (RCBs), which we assume to have very low Fintech proficiencies.
If Fintech capability is an overarching weapon against disruption and loss of market share, one would expect stronger financial and prudential performance in the HP6 group. Descriptive statistics in Table 13 generally support this contention. The HP6 cohort has slightly higher mean ROA and ROE levels relative to the LP9 group. Similarly, NPL and SML rates appear slightly lower for the HP6 cohort. Moreover, and as shown in Fig. 2, the positive mean difference in financial returns between HP6 and LP9 groups, has grown in recent years. However, a significant performance gap is not apparent in multivariate analysis (see Table 14). Nonetheless, the preliminary evidence in Table 13 (Fig. 2) points to an emerging gap in the financial returns of HP6 and LP9 bank entities. Learning and implementation lags may mean that the performance benefits of Fintech adoption only become obvious over long-run horizons.