2.1 Literature review

Long-term focus on systemic risks and efficiency within the commercial banking sector has garnered extensive academic attention. Despite the established interest in the relationship between bank efficiency and risk, existing literature remains theoretically inconsistent and yields mixed findings regarding the impact of efficiency on risk. Variations in findings might arise from differences in variable measurement, analytical perspectives, and sample selection. The literature primarily bifurcates into two streams. One suggests that banking efficiency could mitigate risks. We collectively conclude this first set of theories as the “mitigating role of bank efficiency”.

Concerning the enhancement of efficiency to reduce risk, Dima et al. [11] identified a positive effect of bank efficiency on bank stability. Drawing on a sample of European commercial banks, Fiordelisi et al. [5]employed Granger causality tests to reveal that inefficient banks often exhibit higher risk levels, emphasizing that improving bank efficiency contributes to increased capital and enhances financial system stability. Saeed et al. [12]found that higher cost efficiency implies lower conventional banking risks. The improvement of bank efficiency also aids in reducing non-performing loan ratios [13]. Profitable banks have more diversified products and loan options to reduce risk [14]. Research around the crisis has also discovered the positive effects of efficiency in surviving crises. Based on a sample of U.S. banks, Assaf et al. [4] found that enhancing cost efficiency during normal periods has a proactive effect, effectively improving a bank’s profitability during crises and reducing the probability of bank default. Moreover, Beccalli et al. [15] and Mirzaei et al. [16] found that banks with high operational efficiency exhibit better stock performance in both normal and crisis periods.

Another strand of literature points out that higher efficiency is associated with higher risk. We collectively conclude this second set of theories as the “efficiency as potential hints of risk”. Higher efficiency levels could have been achieved at the cost of higher risk levels [16]. Bank’s appetite for taking higher risks increases [17]. Specifically, the process of enhancing efficiency may entail the aggregation of risk factors. Do Van Anh [18] observed that high-efficiency banks typically maintain lower capital levels. Moreover, a bank’s inefficiency often indicates lower asset quality, thereby increasing the likelihood of risk [19, 20]. The long-term relationship between a bank’s efficiency and non-performing loan ratios involves a delicate balance, contributing to an elevated risk of bank default [21]. Li et al. [22]even posit that bank efficiency can serve as an early warning indicator for the probability of a bank’s future bankruptcy. Beyond individual bank bankruptcy risk, Isik and Uygur [23] found that the level of efficiency in banks also acts as a forward-looking indicator of macroeconomic crises.

The existing literature has laid a solid foundation for research in this field but provides contradictory predictions. We posit that there remains ample room for further expansion and deepening of the investigation into this topic. Firstly, existing research on banking efficiency mostly tends to exhibit a single-dimension efficiency analysis, focusing either on cost efficiency or profit efficiency. Furthermore, there is a lack of uniformity in output settings while estimating efficiency, which results in conclusions that are deficient in comparability. Secondly, when analyzing the relationship between banking efficiency and risk, there is no consensus on the choice of the risk proxy. Most research focuses on the micro-level risks of individual banking entities, such as non-performing loan ratio, banks’ likelihood of default obtained from Moodys’KMV, and insolvency risk, and lacks attention to macro-level risks. Our paper makes several contributions to the literature on the relationship between bank efficiency and risk. First, unlike previous studies, we estimate both cost and profit efficiency estimators. We do this because cost and revenue efficiencies reflect two different managerial abilities. We posit that each of these efficiencies can have a distinct link with bank systemic risk. Moreover, this paper delves into research inconsistencies by setting output as liquidity creation when measuring efficiency. Liquidity creation serves as a comprehensive metric for measuring bank output and offers an integrated depiction of the overall operational situation of banks, which facilitates a more comprehensive and integrated comparison of bank efficiency, eliminating biases in conclusions that may arise from the selection of measurement scopes.

Second, this paper provides deeper insight into the relationship between bank liquidity creation efficiency and risk based on an anatomy of systemic risk. Previous studies have primarily analyzed the risk-bearing aspects of banks in pursuit of efficiency. Risk-bearing corresponds to the individual dimension of bank risk. There is a lack of existing literature explicitly addressing the relationship between efficiency and macro-level risk. The soundness of the banking sector is a vital ingredient of the financial system’s stability. This paper’s measurement of systemic risk breaks down risk into individual risk and systemic connectivity, aiding in a more precise and comprehensive delineation of the sources of risk.

Third, we investigate the underlying transmission mechanism through which risk is generated and prove the effect of digital transformation on balance efficiency enhancement and risk accumulation for banks. Prior studies showed that bank efficiency affects firms’ borrowing costs [8]and productivity [14], thus influencing economic growth [7]. The observed phenomenon in our study points out that enhanced profit efficiency is attributed to more intense risk-taking behaviour by banks, with loan allocations leaning towards the real estate sector. The ongoing digital transformation of commercial banks is conducive to reducing the connectivity between individual banks and the system, thus avoiding the impact of systemic risk.

2.2 Hypothesis development

Competition seems to be a significant driving force among the banks’ long-run stability determinants. The competitiveness of a bank is highly related to bank cost efficiency. Competitive banks are more likely to implement better monitoring mechanisms while liquidity creating and hence are less likely to suffer non-performing loans [24]. Also, these competitive banks can extract monopolistic rents from markets as reflected by higher interest rates charged on loans and lower interest rates offered for deposits [25, 26]. The process of transforming liquid deposits into illiquid liabilities is the core of bank liquidity creation, thus performing high liquidity creation profit efficiency. On the whole, efficient banks tend to be more competitive, which enhances bank stability and reduces the likelihood of exposure to systemic risk. As a result, Hypothesis 1 is formulated.

1. H1: Bank liquidity creation efficiency will reduce the bank’s overall systematic risk.

As early as 1997, Berger and De Young [26] point out that bank efficiency may be an important indicator of future problem loans. Low inefficient banks may tend to have loan performance problems due to poor senior management, which resulted in difficulties in monitoring both their costs and their loan customers, with the losses of capital generated by both these phenomena potentially even leading to failure. Larger banks can control their operating costs through better management and operational procedures and transfer the various categories of risks [11]. Low levels of efficiency could lead banks to try to boost returns by lowering their operating standards, such as less intensive credit monitoring.

1. H2(high management level hypothesis): High bank efficiency is an indicator of high management quality.

However, banks can also simply take on higher risks to compensate for lost returns and increase revenue, thus increasing profit efficiency [5]. High-profit efficiency during normal times may reflect excessive risk-taking that earns high returns in normal times but will create problems in subsequent crises [4]. High-risk investments generally have higher returns. To illustrate, banks’ investments in mortgage-backed securities (MBS) were remarkably profitable during the normal period before the subprime financial crisis. Still, they proved very risky and a significant contributing factor to bank failures and the crisis.

1. H3(high risk-taking appetize hypothesis): High bank efficiency is a reflection of high risk-taking appetite.

3.1 Variable estimation

3.1.1 Bank liquidity creation.

Liquidity transformation is considered the preeminent function of commercial banks. As the liquidity creator of society, the fundamental task is accepting short-term, liquid deposits and making long-term, illiquid loans, which directly facilitate production by affecting the successful channelling of savings to investment. An intuition for this function is that banks create liquidity by holding illiquid items in place of the non-bank public and giving the public liquid items. Bank liquidity creation measures how banks perform this function by focusing on liquidity classification and weight-assigning balance sheet items.

This study adopts the “Three-step methodology” as introduced by Berger and Bouwman [7] for the comprehensive measures of bank liquidity creation, categorizing balance sheet banking activities into three liquidity classes: liquid, semi-liquid, and illiquid. S1 Table provides the division standards details. A commercial bank’s liquidity creation is quantified through a weighted summation of these classifications, with the formula expressed as follows.

The measure helps us gain a deeper insight into banks’ role as liquidity creators and risk transformers. Considering the operational characteristics of Chinese banks, we draw on existing influential research [27] to adjust liquidity category and empowerment. Fig 1 shows the times series data of the liquidity creation scaled by assets(LC_TA) for different bank ownership structures since 2010. In the initial stages, state-owned and joint-stock banks exhibited a higher level of liquidity creation per unit of asset. However, in recent years, the capacity of rural commercial banks to serve the real economy has rapidly improved, increasingly approximating the levels observed in joint-stock banks. Building upon these shifts, a pivotal area for subsequent analysis will be the cost efficiency and profit efficiency of different types of banks in the liquidity creation process.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Year trend of LC_TA for different bank ownership structures.

This figure plots quarterly estimation results of LC_TA for different bank types within the sample period.

https://doi.org/10.1371/journal.pone.0313208.g001

3.1.2 Bank liquidity creation efficiency.

The concept of efficiency originates from measuring the achievement of bank economic goals by researchers of the last three decades. Banks often pursue the economic goals of cost minimization or profit maximization. The bank liquidity creation cost/profit efficiency tells us how close a bank’s cost/profit is to what a best-practice bank’s cost /profit, which operates in close proximity to the efficient production frontier, for producing the same current level of liquidity creation under the same conditions.

Commercial banks can fundamentally be conceptualized as input-output systems that produce financial products by incorporating essential factors such as fixed assets, human resources, and capital. In calculating bank efficiency, the most critical and preliminary step is the specification of input and output variables within the cost and profit frontier functions. We define bank total profit as pre-tax net income and total costs as the sum of interest expenses, fee and commission expenses, operating expenses, and other operating expenses. While there is a consensus within academic research regarding the inputs of commercial banks, there remains a disparity concerning the definition of outputs. Acknowledging the role of banks in the liquidity creation process, this paper innovatively posits liquidity creation as the bank’s output. Liquidity creation serves as a comprehensive measure of a bank’s overall operations. This setting has two-fold advantages. On the one hand, liquidity creation outperforms loans in reflecting how banks promote real economic growth [7, 28]. On the other hand, the calculation of liquidity creation is relatively standardized worldwide, making the efficiency measure more comparable across countries. We specify three input prices: w₁, the price of purchased funds (the ratio of total interest expenses to total liabilities); w₂, the price of labor (ratio of total personnel expenses to total assets); w₃, the price of physical capital (total operating and administrative expenses to total assets). We also include quantities of two fixed net puts (inputs or outputs): z₁, the total value of total bank off-balance sheet activities and z₂, bank financial equity capital. The details are listed in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Input-output variable definitions of bank efficiency.

https://doi.org/10.1371/journal.pone.0313208.t001

The literature suggests that bank efficiency can be calculated using either a non-parametric method (i.e., DEA) or a parametric approach (i.e., stochastic frontier analysis, SFA). We employ the most widely utilized methodologies, the stochastic frontier approach [29, 30] to gauge the efficiency when banks create the same liquidity under the same conditions, which not only posits specific functional forms for the efficiency frontier but also facilitates validity tests for estimated parameters. Building on the strengths of panel data, Greene [9] incorporates fixed effects within a frontier estimation framework. Drawing on his enhancements, the study specifies the stochastic frontier function for liquidity creation by banks in a panel dataset as follows: (1) where Q represents cost or profit incurred. f is the cost or profit function. Z and W denote inputs and their price. Y is output and a_i is the bank-fixed effect. u_i,t ∼ G is the inefficiency factor and is random error. We set output as the bank’s liquidity creation level. As previously mentioned, liquidity creation offers a more complete and accurate representation of how banks serve the economy. Banking liquidity creation activities, like loans [31] and off-balance sheet activities [32, 33], boost economic development. Based on this model, Li et al. [22] compared the liquidity creation efficiency of global banks in time and space dimensions. With the distribution of v and u given, the inefficiency factor for the single bank can be estimated by maximum likelihood. Drawing on Greene’s [9]estimation approach, the expectation of inefficiency term can be calculated as: (2) where ϕ and Φ are the probability density and cumulative distribution functions of normal distributions, respectively. The formula can be reduced as: (3)

In line with the intrinsic goals of cost minimization and profit maximization in a bank’s operational processes, we distinguish between liquidity creation cost efficiency and liquidity creation profit efficiency. The assessment involves measuring the proximity of a bank’s generated profits or costs, given the prices of input factors, the output of liquidity creation, and other variables, to the efficient frontier represented by the banks with maximum profits or minimum costs. The bank liquidity creation cost efficiency (LCCE) can be measured as the ratio of a bank’s actual cost for creating the same level of liquidity under the same conditions to the minimum cost (best practice) a bank would incur. This efficiency measurement reflects the degree of efficiency with which a bank operates relative to the best-performing institutions within the sector.

Similarly, we measure bank liquidity creation profit efficiency (LCPE) as the ratio of a bank’s actual profit to the profit a best-practice bank would attain. where and denote the estimated inefficiency factor regarding cost and profit.

Then we specification of Bank liquidity creation cost and profit functions. The traditional Cobb-Douglas production function, characterized by substitution elasticity of one and constant returns to scale, poses challenges in replicating the true frontier of bank production processes. On the contrary, the transcendental logarithmic (translog) production function is extensively employed in panel estimations of the effects of bank efficiency on bank performance. In addition, many researchers found that it fits banks’ data better than other functional forms. Drawing on existing research [29, 30, 34], we specify cost and profit functions as follows: where LC is liquidity creation defined in Section 3.1.1, denoting the bank system’s output. w_i and z_i are input prices and net inputs specified in Table 1, respectively. X includes control variables.

3.1.3 Systemic risk and decomposition.

We assess banks’ systemic risk by evaluating their sensitivity to shocks in the financial system, especially under distress events. The tail risk of individual financial institutions (IR) and the degree of connection with the financial system (SL) jointly determine the level of systemic risk. Therefore, we address such a distinction by decomposing the systemic risk estimator into two components to measure bank tail risk and systemic linkage, respectively.

This paper measures bank systemic through the method developed by van Oordt and Zhou [35]. The systemic risk of bank i is measured by the loss it experiences when an extremely adverse shock hits the financial system, namely, in the following tail model: (4) where R_i and R_s denote the stock return of bank i and the financial system s, is the value-at-risk of the system. notes a very small probability. With k worst ones out of n observations on the pair (R_i, R_s), p = k/n. The evaluation of is useful to assess the extreme loss on the stock in the event of a market crash. Van Oordt and Zhou [35] evaluated the potential drawbacks and applicable limitations of several kinds of measure methods for estimating and proposed an alternative estimator based on Extreme Value Theory(EVT).

Assuming R_s and R_i are both heavy-tailed, which can be expressed as where l_i(u), l_s(u) are slowly varying functions as u → ∞

Then, we can consider the tail dependence measure using the EVT approach. (5)

Eq 5 provides the basis of estimation of coefficient β^T. For statistic estimation, we mimic the limit of the extreme value of p → 0 by focusing on the lowest observations of the tail region, denoted as k. The probability p can be rewritten as p = k/n, where k/n → 0 as n → ∞. is the tail index estimator proposed by Hill [36], is the nonparametric estimator of τ_i ≔ lim_{p → 0}τ_i(p), and can be estimated nonparametrically as: where the losses , and I_{} denotes a indicator function.The biggest advantage of this measure is that systemic risk measure β^T can be expressed and decomposed logarithmically into two conceptual components: (6)

Component IR_i (individual tail risk) captures the level of bank tail risk. Since the denominator VaR_s(p) is homogeneous across all financial institutions, the cross-sectional variation in IR_i is solely due to the variation in the tail risks of individual banks, VaR_i(p). Component SL_i (systematic linkage) measures the strength of the link between the bank and the system in financial distress. Since τ_i(p) is unaffected by any strictly increasing monotonic transformation of R_i, cross-sectional differences in SL_i contain information only on the dependence between extreme shocks in the financial system and severe losses suffered by a particular bank, without being affected by the level of bank tail risk. As the sum of individual tail risk and systemic linkage, (beta) is systemic risk.

We use the return of the China Shenwan Banking Industry Index to proxy financial system return R_s, and set k/n ≈ 5%, similar to values in previous studies. Our baseline results fix a rolling estimation window of 8-quarter daily return for each bank.

The risk decomposition results for China’s bank sample are presented in Fig 2. The total systematic risk (beta) measure highlights several significant events in the Chinese market, such as the 2015 stock market crash, the 2008 global financial crisis, and the 2005 non-tradable shares reform. Beta rose sharply before these crashes and then plummeted. There was generally a negative correlation between IR and SL, particularly during crises where IR usually increases and SL decreases. This implies that banks suffered severe adverse tail shocks and became disconnected from the banking system. SL tends to be at a relatively high level before a crisis. Its decline is steep when a crisis occurs, although its volatility is relatively small during normal times.The individual tail risk of the bank and the risk of systematic linkage are in a state of mutual transformation. Before the crisis, banks were usually in a state of high linkage. However, during the crisis, the reasons for the decline in bank linkage are as follows. First, banks will fire sell assets to reduce indirect connection. Second, the shock has been transformed into an individual due to network contagion, which is reflected in an individual tail risk. This finding reflects the paradoxical robustness-while-fragile characteristic of the Chinese banking system. The distinct patterns of the two components also validate the advantages of decomposing the risk.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Decomposition of bank systemic risk in China.

This figure plots quarterly estimation results of total systematic risk(beta), individual tail risk(IR), and systematic linkage(SL) between sample periods. These three categories are defined in Section 3.1.3.

https://doi.org/10.1371/journal.pone.0313208.g002

3.1.4 Other bank characteristics and summary statistics.

To further mitigate the confounding effects of omitted variables on the results, this paperincorporates variables that reflect the latent risks in the banking operational process as control variables. The scale of a bank (Size) is closely correlated with its efficiency, whereby business scale may encompass additional risks, yet conversely, it could diminish overall risk due to risk diversification. Capital ratio (Capr) reflects the bank’s leverage level, and operational volatility in banking may be influenced by leverage amplification. Net return on equity (ROE) and non-performing loan ratio (NPL) directly depict the bank’s profitability and default risk in its primary lending operations. Meanwhile, the non-interest income ratio reflects the diversification trend in banking operations. Non-interest income (NII) may simultaneously enhance a bank’s profit potential and augment its sources of risk. Table 2 presents the summary statistics of key variables.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Summary statistics of key variables.

https://doi.org/10.1371/journal.pone.0313208.t002

In China, banks are typically categorized based on their equity structures into state-owned banks, joint-stock banks, city commercial banks, and rural commercial banks. These categories of banks exhibit significant differences in their business scope, funding costs, operational models, and efficiency objectives. Consequently, we conduct a deeply descriptive statistical analysis of the profit efficiency and cost efficiency for each type of bank. The results are presented in Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Descriptive statistical summary of liquidity creation efficiency measure.

https://doi.org/10.1371/journal.pone.0313208.t003

Profit efficiency among Chinese banks does not exhibit significant variation overall, whereas there are notable discrepancies in cost efficiency. Banks with the highest profit efficiency are predominantly state-owned, reflecting their market share dominance and relatively superior profitability. On the other hand, rural commercial banks and city commercial banks demonstrate higher cost efficiency, respectively. Despite lacking a cost advantage in terms of capital, these banks have proactively advanced unit cost output. Banks are becoming probability through methods such as project management to establish a competitive edge in cost control.

3.2 Data description

This paper collects panel data for a sample of China A-share listed banks over the period 2000–2021 from the China Stock Market & Accounting Research Database (CSMAR) and the Wind database. We have also manually supplemented some variable observations from bank annual reports. We have winsorized all continuous variables at the 1st and 99th percentiles to ensure consistency.

3.3 Empirical methodology

To explore the empirical relationship between systemic risk and bank liquidity creation efficiency, we estimate three regression models using our estimates of systemic risk(lnbeta), systemic linkage(lnSL), and bank tail risk(lnIR) as dependent variables, respectively. We estimate the coefficients in the following models from panel data. (7) Where θ_t is time-fixed effects, capturing variation in macroeconomic state variables and other sources of common variation in systemic risk over time and δ_i is bank fixed effects, capturing characteristics of a bank that do not change over time. The bank individual fixed effect and the time fixed effect are used to alleviate the interference of the important explanatory variable omission. Independent variable LCE_i,t−1 is lagged bank liquidity creation efficiency, including profit and cost efficiency defined previously. We follow the related literature [4, 37] and control a set of bank characteristics X_i,t. Finally, v_i,t, ϵ_i,t, e_i,t are the error terms respectively.

4.1 The double-edge effect of efficiency

Table 4 presents the empirical results of the Eq 7. Columns (1)-(3) show the double-edged effect of bank liquidity creation profit efficiency: Specifically, while there is a notable negative correlation between bank liquidity creation profit efficiency (LCPE) and individual risk (IR), the enhancement of LCPE is also linked to an increase in system linkage (SL), suggesting a possible trade-off between these two effects. The total impact of bank profit efficiency on systemic risk is negative, indicating the dominant role of individual risk (IR). The t-statistic of β₁ is -3.92, which means coefficient β₁ is statistically less than zero. The regression results support Hypothesis 1 which proposes that Bank liquidity creation efficiency will reduce the bank’s overall systematic risk. Despite the current lack of observable adverse impacts of liquidity creation profit efficiency on the overall systemic risk from the data, this may lead to an underappreciation of the inherent risks associated with increased systemic interconnectedness. Particularly in China, banks enhance their profit efficiency through non-traditional transactions among interbank or non-banking financial institutions, which increases their systematic linkage to the system. However, the potential risks accumulated related to systematic linkage are often overlooked, with attention solely focused on the changes in aggregate risk. It is noteworthy that efficiency improvement embodies a double-edged sword effect. Reducing individual risks while increasing the interconnectedness of the financial system may finally increase the risk of contagion.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. The relationship between bank liquidity creation efficiency and systemic risk.

https://doi.org/10.1371/journal.pone.0313208.t004

Columns (4)-(6) show that bank liquidity creation cost efficiency (LCCE) is not significantly related to risk. This supports the view that LCPE and LCCE reflect different abilities of banks, rendering the correlation between LCPE and LCCE indefinite [5]. Our findings for Chinese banks differ from U.S. banks, whose cost efficiency significantly correlates to risk rather than profit efficiency [4]. In the course of cost control, Chinese banks have not accumulated risk per se. Instead, it is in the pursuit of profit that they have heightened their interconnectedness with the financial system, thereby potentially precipitating a crisis of systemic contagion. This heightened linkage, while ostensibly fostering efficiency to profit maximization, inadvertently augments the financial system’s vulnerability to the propagation of risk. Consequently, a failure within one institution can cascade through the network, engendering a domino effect that threatens the entire system’s stability. This phenomenon underscores the need for vigilant regulatory oversight of bank behaviors engaging in profit chasing. Columns (7)-(9) incorporate both profit and cost efficiency into the regression model while controlling for potential mutual influences. The results remain unaffected. The relationship between cost efficiency and risk is insignificant, whereas the impact of profit efficiency on risk continues to exhibit a double-edged sword effect.

Theoretically, there are contradicting channels through which bank efficiency affects risk. High-cost efficiency may reflect superior managerial quality, resulting in favorable performance [26, 38], or indicate “skimping” on resources to screen and monitor loan applicants, leading to poor loan outcomes that become apparent only during subsequent financial crisis [26]. High-profit efficiency may be associated with high charter values that produce favorable performance [39, 40], or excessive risk-taking that earns high returns in normal times but creates problems in subsequent crises [4]. Thus, a natural question arises: does the relationship between bank liquidity creation efficiency and risk in China, diverging from the conclusions of existing literature, reflect a different story? We explore the answer by delving deeper into the channels through which bank efficiency influences risk.

4.2 Bank liquidity creation efficiency, managerial quality, and risk-taking

We employed the principal component analysis to construct a measure reflecting the managerial quality of banks. Based on relevant findings in corporate governance research [41], we selected indicators across six major categories: ownership structure indicators (including the shareholding ratio of the largest shareholder, the ratio of the shareholding by the second to fifth largest shareholders to that of the largest shareholder), nature of property indicator (a binary variable indicating whether the largest shareholder is a state-owned shareholder), controlling shareholder behavior indicators (the ratio of related transactions measured as the sum of sales to and purchases from related parties divided by total assets at the end of the period), management governance indicators (a binary variable indicating whether the company Chairman and CEO concurrently serve), board governance indicators (board size, proportion of independent directors), and external market competition indicators (market share measured as the company’s operating revenue divided by the total operating revenue of the industry). These indicators were subjected to principal component analysis, with the score of the first principal component serving as the indicator reflecting the managerial quality [42].

Concerning the measurement of risk-taking on the asset side of banks, the literature commonly employs indicators such as non-performing loan ratios and Z-scores. However, these variables essentially gauge banks’ post facto risk-taking behavior. To assess banks’ ex-ante risk-taking behavior, this study elected the proportion of risk-weighted assets to total assets as a measure.

We conduct panel regressions similar to the model 7, except that the dependent variables are replaced by the corresponding channel variables, namely managerial quality or risk-taking. Columns (1)-(3) in Table 5 show that the improvement in bank liquidity creation efficiency is not correlated with the improvement in managerial quality in China. This differs from the evidence that bank cost efficiency is associated with managerial quality in the U.S. [4], which implies that banks in China may improve liquidity creation cost efficiency by devoting fewer resources to credit screening and monitoring [26]. However, higher bank liquidity creation profit efficiency in China is accompanied by more risk-taking on loans, as Table 5 Columns (4)-(6) show. The empirical results support hypothesis 3 (high risk-taking appetize hypothesis) rather than hypothesis 2 (high management level hypothesis). Thus, an interesting question is: what loan structure do banks in China look like?

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Bank liquidity creation efficiency, bank managerial quality, and risk-taking.

https://doi.org/10.1371/journal.pone.0313208.t005

4.3 Bank liquidity creation profit efficiency and allocation of loan

China’s real estate industry experienced rapid and sustained development and acted as a significant contributor to the country’s economic growth. According to the China National Bureau of Statistics, real estate investment has accounted for more than 10% of GDP since 2009. The manufacturing industry is usually a proxy for the real sector and the key to industrial upgrading in China. We empirically explore the relationship between bank liquidity creation profit efficiency and loan allocation to these two industries. As is shown in Tables 6 and 7, the high liquidity creation profit efficiency of banks in China is closely associated with more investment in the real estate industry and less investment in the manufacturing industry, which implies that funds may be diverted away from the manufacturing sector. Banks played a crucial role in the real estate industry development and made steady and profitable returns, thereby enhancing their profit efficiency and helping alleviate individual tail risk, IR. However, as highlighted by Allen and Gale [43], the flow of a large amount of money from the banking industry into the real estate market creates a complex network between the banking and real estate industry, resulting in overlapping closed ties among banks and increasing the potential for financial contagion and systemic risk. This affects not only the link between banks and real estate firms but also the connection between banks themselves due to the concentration of credit resources in the real estate sector. The debt problems originated from Evergrande (a famous real estate enterprise in China) in 2021, which is a good example. Therefore, while increased liquidity creation profit efficiency may decrease individual tail risk IR, it also leads to a rise in systemic linkage, SL, which may result in the accumulation of bubbles and future risks. This echoes the double-edged effect of bank liquidity creation efficiency.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. The bank liquidity efficiency and bank loans allocation.

https://doi.org/10.1371/journal.pone.0313208.t006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 7. The bank liquidity efficiency and bank loans allocation.

https://doi.org/10.1371/journal.pone.0313208.t007