Introduction

Currently, the increasing complexity of the auditing environment, coupled with various uncertainties in the operational and financial management of enterprises, poses significant challenges for auditing organizations in mitigating detection risks. In 2024, financial bureaus of China’s Ministry of Finance imposed administrative penalties on 231 accounting firms and 514 certified public accountants (\(CPAs\)). Additionally, 181 accounting firms faced confiscation of unlawful income and fines totaling RMB 18,281,200, which reflects a year-on-year increase of 15.36%. The rise in penalties imposed on audit firms has been more substantial compared to previous years, making the resolution of these challenges a pressing concern for audit firms.

Nowadays, all kinds of audit institutions have realized that they must actively promote the digital transformation of auditing and enhance the proficiency of audit talents in the use of audit techniques and tools, to improve the quality of auditing and control the level of detection risk. Given the increasing importance of digital audit talent, it is crucial to explore whether improving the adequacy of digital audit talent can effectively reduce the level of detection risk. From the perspective of auditees, a higher degree of digitization correlates with improved corporate governance, internal control, and risk management. Consequently, the quality of data concerning business operations, production management, and capital finance is enhanced, leading to increased accuracy and transparency in financial information disclosure. This, in turn, reduces both audit costs and the detection risk encountered by auditing institutions. Conversely, from the perspective of audit institutions and their auditors, digital technology serves as a double-edged sword: while it improves audit efficiency and quality, it also necessitates that auditors acquire advanced digital skills and adapt to new auditing scenarios. This includes navigating the complexities of computer program code implementation, model construction, and parameter configuration, which may introduce new risks related to data security, network security, and other types of technical detection risks. Therefore, understanding whether the adequacy of digital audit talent reduces the level of detection risk in audit institutions holds significant theoretical and practical implications for the sustainable and healthy development of these institutions.

To further verify the above research objectives, the research framework of this paper is designed as follows: First, it systematically reviews the current research status and gaps in the fields of digital audit talent, audit digitization, and detection risk, laying the theoretical foundation for this study; Secondly, based on the digital data of the accounting firm industry disclosed by the Chinese Institute of Certified Public Accountants (\(CICPA\)) from 2020 to 2022, the impact of the adequacy of digital audit talent on the level of detection risk and its intermediary mechanism are discussed, and the corresponding research hypotheses are proposed; Third, empirical testing is conducted using this model, which includes robustness checks and heterogeneity tests. Finally, the research findings are synthesized to provide a summary of theoretical contributions and practical implications, accompanied by a discussion of the study’s limitations and suggestions for future research directions.

The potential contribution of this paper is mainly reflected in two aspects: Firstly, it verifies the logical assumption that audit institutions can improve their degree of audit digitization and thus reduce the level of detection risk by increasing the adequacy of digital audit talent, which extends the theoretical research perspective of detection risk and provides new empirical support; Secondly, it provides strategic suggestions and references for audit institutions on how to push forward their audit digitization transformation and improve their digital audit capability, which has some value in terms of how to effectively reduce detection risk in practice. In addition, the findings and conclusions of this paper will help the audit industry and audit institutions to enhance their understanding of the value and significance of audit digitization.

Theoretical basis and concept definition

Definition and characteristics of digital audit talent

Digital audit talent encompasses individuals who possess both traditional auditing expertise and proficiency in digital technologies, including big data analytics, artificial intelligence, and blockchain1. This paper utilizes the talent framework collaboratively developed by the Digital Manufacturing and Design Innovation Institute (\(DMDII\)) in the United States and Manpower Group2, the world’s third-largest human resources service company, to categorize digital audit talent into two primary types: knowledge-oriented audit talent and technology-oriented audit talent. Knowledge-oriented audit talent includes individuals with project experience related to digitization or the capability to employ digital methods and tools to address challenges encountered in audit work. In contrast, technology-oriented audit talent refers to individuals who contribute to the development of digital infrastructure, formulate technical roadmaps, provide essential technical support, and offer technical assurance.

Considering the digital auditing experience of auditors, audit expertise theory3 posits that the specialized knowledge and experience auditors accumulate in specific domains can significantly enhance audit quality4. This aligns with the Resource-Based View, which emphasizes heterogeneous resources as sources of competitive advantage. Academia generally agrees that, in the context of digital transformation, auditors can develop digital audit expertise by engaging in digital audit projects and acquiring relevant experience and skills. Furthermore, the widespread adoption of digital credentials and transactions by clients renders traditional paper-based auditing techniques inadequate, compelling auditors to cultivate a digital audit mindset5. For instance, Manita et al.6 argue that auditors with a significant proportion of digital clients possess digital audit talent, as they gain extensive experience through participation in digital audit practices, thereby forming unique digital audit expertise. Feliciano et al.7 assert that digital audit talent encompasses auditors who have developed data analysis capabilities and knowledge through the auditing of clients undergoing digital transformation.

From the perspective of utilizing digital methods and tools, the academic community generally holds that digital audit professionals must master technologies such as data mining and machine learning to enhance the efficiency and quality of audit work8. Zhang9 pointed out that auditors should learn to employ Robotic Process Automation (\(RPA\)) ‘assistants’ to automate certain aspects of the audit process, effectively ‘replacing humans’ in performing repetitive tasks. Consequently, audit personnel would primarily need to review results, manage exceptions, and coordinate activities. Atadoga et al.10 indicated that auditors should leverage technologies like cloud computing to integrate resources from various stakeholders, thereby improving the efficiency of personnel in collecting audit evidence. Xu et al.11 noted that when auditors conduct tests on subjects such as revenue, they can utilize data regression technology systems to perform large-scale sampling and data processing, overcoming the limitations of traditional testing methods, which often involve complex sample selection, challenges in matching, and weak representativeness.

From the perspective of possessing professional digital theoretical knowledge, Amy et al.12 proposed that digital audit talent must exhibit interdisciplinary learning abilities, risk identification skills, and the capacity for continuous learning and innovation. Qasim et al.13 pointed out that intelligent accounting professionals should possess a combination of skills that encompass multiple disciplines, data literacy, an understanding of digital intelligence technologies, and digital intelligence management to meet the demands of enterprise digital transformation. Lee et al.14 emphasized that auditors should evolve into ‘high-quality hybrid professionals,’ with a continuous learning ability recognized as one of their essential qualities. Anderson-Gough et al.15 explored the concepts of ‘performance’ and ‘meritocracy’ in performance evaluation and career development, asserting that digital audit talent must possess both technical capabilities and social compatibility.

From the perspective of practical application development capabilities, this type of talent not only assists in the construction of information equipment and other infrastructure during the early stages of digital transformation but also significantly enhances the application efficiency and update frequency of digital technologies in subsequent operations, thereby promoting the deep integration of digital technologies with audit business16. Technical audit talent is responsible for establishing essential digital functions, and the scope of this role is relatively broad, encompassing numerous initial tasks in the digital transformation of audit institutions, such as digital design engineers, cybersecurity engineers in manufacturing, and data architects17. Qualified technical audit professionals should possess skills in technology and computer science, digital competencies, and robotics or automation programming18. They must be capable of understanding and applying existing audit software and technologies, comprehending business models and needs, and exhibiting independent research, development, and innovation capabilities to meet the evolving digital demands of audit institutions19. Additionally, they should possess strong communication skills and a spirit of teamwork to ensure the smooth progression of digital transformation efforts20.

In summary, digital audit talent encompasses professionals who have accumulated extensive knowledge and experience in digital audit projects, thereby developing expertise in this field. These individuals are adept at utilizing advanced tools and technologies, which significantly enhance audit efficiency. Furthermore, they possess a strong foundation in digital theoretical knowledge and application development capabilities, which powerfully drive the digital transformation of audit institutions.

Examine the connotation and influencing factors of the level of detection risk

The traditional audit risk model is defined as the product of inherent risk, control risk, and detection risk. Detection risk is the risk that the auditor’s procedures performed to reduce audit risk to an acceptably low level will not detect a material misstatement that exists and could be material, either individually or when aggregated with other misstatements, to the financial statements21. In a digital environment, Earley22 suggests that the connotation of detection risk has expanded to include not only the risk of misstatement in traditional financial data, but also new dimensions such as information system control risk and data analysis model risk. From the perspective of factors influencing detection risk, research by Kokina et al.23 has found that the application of digital technology can, on the one hand, reduce traditional detection risk. For example, Bhaskar24 argues that the detection risk is related to the audit procedures selected by the \(CPA\) s; that is, appropriate audit procedures can reduce detection risk. On the other hand, it also introduces new risk factors, such as algorithm bias and data quality issues. For example, Stefaniak et al.25 believe that the professional competence of \(CPAs\), such as professional knowledge and professional judgment, is the primary factor affecting detection risk, and the quality control risk of accounting firms also affects detection risk. Glover et al.26 point out that improper supervision and audit focus may threaten the appropriate application of auditor skepticism and further affect detection risk. In terms of measuring detection risk, scholars such as Fearnley et al.27 have used a combination of network analysis and fuzzy comprehensive evaluation method to quantitatively assess detection risk and construct an audit detection risk evaluation index system.

Definition and measurement of audit digitization

The audit digitization refers to the extent to which digital technologies and tools are applied during the audit implementation process28. It is a systematic process that introduces modern digital intelligence technologies, such as big data analytics and artificial intelligence, to assist audit institutions and auditors in accurately identifying and assessing potential risks, thereby improving audit efficiency29. Krahelet et al.30 pointed out that audit digitization not only signifies the updating of technical tools but also represents a fundamental transformation in audit methodologies and thinking patterns. Enhancing the audit digitization level can, to a certain extent, influence the efficiency, accuracy, and quality of audit work by reducing human errors and increasing the automation of audit tasks.

Existing research primarily assesses this phenomenon across three dimensions: the breadth of technology application, which encompasses the scope of technologies such as big data analytics, artificial intelligence, and blockchain29; the depth of process integration, reflecting the extent to which digital technologies are integrated with traditional audit processes1; and the strength of decision support, indicating the degree to which digital tools facilitate audit judgment31. The most recent measurement dimension emphasizes the breadth of generative AI application. For example, Kokina et al.23 proposed a new measurement indicator for the audit digitization level – AI-Penetration, including technology coverage (the proportion of audit processes using generative AI) and the depth of human–machine collaboration (the frequency of interaction between AI and auditors).

Research hypotheses

Analysis of the relationship between the adequacy of digital audit talent and the level of detection risk

The rapid development of digital audit has imposed new requirements on auditors’ competency structures, while traditional audit model face challenges in controlling detection risk. Against this backdrop, the adequacy of digital audit talent has become a critical factor influencing the level of detection risk.

Earley22 states that auditors’ digital expertise can significantly improve the internal control levels of clients undergoing digital transformation. Zhang9 states that \(RPA\) implementation in audits significantly reduces human error-related risks. Babina32 believes that digital audit talents can empower audit development through technology by building system platforms, using new digital technology tools, and data analysis software, thereby enhancing the ability to see through corporate financial and non-financial information, enriching audit risk prevention methods, and effectively improving audit quality. Otia et al.33 emphasize that auditors need to master new skills in digital transformation and improve their capabilities through training programs. Auditors with digital skills can better handle complex data and identify risk patterns.

Based on the aforementioned research, this study argues that digital auditors can utilize advanced data analysis tools and technologies to develop and optimize automated audit processes, thereby reducing manual operations in audit procedures, such as \(RPA\). This approach enables auditors to collect and process large volumes of data more effectively, allowing for more accurate identification of potential risk points and further enhancing the internal control levels of enterprises34. Consequently, we propose the following hypothesis:

Hypothesis 1 (H1)

The adequacy of digital audit talent is negatively associated with the level of detection risk.

Analysis of the relationship between the adequacy of digital audit talent and the degree of audit digitization

Existing research demonstrates that the adequacy of digital audit talent not only directly affects technology implementation outcomes but also exhibits a dynamic interactive relationship with the degree of audit digitization. Specifically, on the one hand, digital audit professionals can propel comprehensive digital transformation in organizational audit practices. Drawing insights from Haislip et al.35, they proposed a theoretical framework for ‘talent-technology adoption’ that indicates sufficient talent reserves can reduce organizational resistance to technology adoption and accelerate digitization processes. This provides direct theoretical support for the positive correlation between talent adequacy and advancements in digitization. Furthermore, Manita et al.6 complement this perspective by highlighting that audit partners with a digital vision significantly influence technology investment decisions within accounting firms, thereby reinforcing the critical role of talent in driving audit digitization. On the other hand, advancements in audit digitization reciprocally reshape talent demand structures. Huang et al.34, believe that the deployment of \(RPA\) requires audit teams to be equipped with “technical translators,” namely talent who understand both auditing standards and automation tools. Enterprises lacking such interdisciplinary talent often find their \(RPA\) projects stalled in the pilot phase, unable to scale. Based on the theoretical foundations and empirical evidence provided by the aforementioned studies, this paper proposes the following hypothesis:

Hypothesis 2 (H2)

The adequacy of digital audit talent is positively correlated with the degree of audit digitization.

Analysis of the relationship between the adequacy of digital audit talent, the degree of audit digitization, and the level of detection risk

Talent is the core element driving the development of auditing. Kokina et al.23 pointed out that excellent talents can bring better product quality and higher work efficiency to organizations, laying a foundational argument for the critical role of talent in auditing advancement. Digital auditors can empower audit businesses by building system platforms and data analysis software, providing auditors with new technologies, tools, and methods. This empowerment enhances the ability to interpret corporate financial and non-financial information, enriches risk prevention means, and effectively improves audit quality. Krahel and Titera30. further supported this mechanism by arguing that audit digitization serves as a bridge between technical talent and audit efficiency through methodological and tool transformation, clarifying how the application of talent-driven technology enhances practical outcomes. Janvrin et al.31 confirmed that digital tools improve risk identification capabilities through process integration, thereby validating the effectiveness of technical empowerment in risk management. Ultimately, this creates a logical chain where digital talent fosters digitization, which in turn influences detection risk. This mediating pathway is consistent with the “technology-capability-risk” framework proposed by Lugli and Bertacchini28, providing a theoretical foundation for understanding the dynamic relationship between talent, digitization, and audit risk. Based on this analysis, we propose the following hypothesis:

Hypothesis 3 (H3)

The more adequate the digital audit talent, the higher the level of detection risk can be reduced by increasing the degree of audit digitization.

Research design

Based on the above research, this paper constructs the research hypotheses, systematically demonstrates the relationship between the three types of variables, namely, the adequacy of digital audit talent (explanatory variable), the degree of audit digitization (mediating variable) and the level of detection risk (explained variable), and conducts robustness and heterogeneity tests on the regression results. The research framework is shown in Fig. 1.

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

Research framework.

Data sources

Based on the research hypothesis, this paper takes domestic accounting firms as the primary research subject. It compiles and refines a panel dataset comprising 300 observations from 2020 to 2022 as the empirical sample (In view of the missing data prior to 2020, the scope of this study is limited to these three years to ensure the accuracy of the empirical analysis and the reliability of the data). The relevant data encompasses three dimensions: First, the adequacy of digital audit talent resources, measured by the number of personnel recruited in information technology, digital construction, and other related technical fields. The data is derived from information disclosed on the official websites of accounting firms via the “Drops Credit” platform (a publicly accessible enterprise credit intelligence platform for small and medium-sized enterprises, accessed at: https://shuidi.cn). Second, the degree of audit digitization, is measured by the number of personnel engaged in digital construction within the firms in the industry. The data is derived from information disclosed on the official website of \(CICPA\); And third, the level of detection risk is measured by the natural logarithm of the penalty amount imposed on audit institutions and their auditors. The data on penalty amounts is derived from information disclosed on the official website of the China Securities Regulatory Commission (\(CSRC\)).

Variable design and construction

Explanatory variable—the adequacy of digital audit talent (IT)

As indicated by the literature review, digital audit talent can be categorized into knowledge-oriented audit talent and technology-oriented audit talent. This paper employs the number of technology-oriented audit talent hired—the proxy for the adequacy of digital audit talent—as the explanatory variable. Drawing on the research by Haislip et al.35, this study selects keywords related to information technology and digital construction, including job roles such as network security engineer, cloud engineer, data developer, \(RPA\) developer, web front-end developer, C#/NET/Java software engineer, C++ developer, and software developer, to filter recruitment information from accounting firms. Initially, this study employs web scraping technology to collect recruitment data from the official websites of accounting firms via the “Drops Credit” platform. Subsequently, a meticulous manual screening process is conducted to identify and categorize recruitment information related to information technology and digital construction talent released by each accounting firm over various years. Finally, through data aggregation, this paper quantifies the number of digital audit talent hires by each accounting firm across different years. Table 1 below presents some of the recruitment announcements obtained.

Table 1 This table is a partial list of digital audit talent recruitment announcements (2020–2022).
Full size table

Explained variable—the level of detection risk (DR)

This study utilizes the natural logarithm of the penalty amounts levied on audit firms and their auditors as the dependent variable, referred to as the level of detection risk. According to the theory proposed by Chen et al.36, regulatory penalties effectively quantify the severity of audit deficiencies. Therefore, this paper employs regulatory penalties as a proxy variable for the level of detection risk. Based on publicly available data from the \(CSRC\) and the Ministry of Finance from 2020 to 2022, approximately 70% to 80% of all penalties directly stem from financial statement audit engagements (e.g., deficiencies in audit procedures, failure to detect financial fraud), while the remaining 20% to 30% involved violations related to non-audit services such as taxation and consulting (e.g., improper tax planning, internal control consulting flaws). The penalty amounts considered in this paper encompass the sum of the following three types of penalties imposed on accounting firms and signing \(CPAs\) within the respective year due to violations in financial statement audit engagements: criminal penalties, administrative penalties, and professional sanctions. To ensure accuracy, the reasons for the penalties were individually verified through the penalty decisions of the \(CSRC\) and announcements from the \(CICPA\).

Mediating variable—the degree of audit digitization (AD)

This paper uses the degree of audit digitization as a mediating variable. Previous studies have primarily employed theoretical analysis, case studies, interviews, and questionnaires to examine the indicators of digitization in accounting firms. These indicators include the proportion of information technology that significantly impacts accounting firms and the utilization of information systems within these firms. Due to the lack of specific data directly reflecting the degree of audit digitization, this study selects the data related to “industry informatization construction” (Number of people engaged in digital work in the professional area of this accounting firm) disclosed on the official website of the \(CICPA\) and uses it to quantitatively evaluate the degree of audit digitization.

Control variables

Referring to the professional contribution index of accounting firms on the official website of the \(CICPA\), as well as the analyses of auditors’ characteristics37, this paper chooses the intensity of auditor training intensity (\(TI\)), auditor professional competence (\(PC\)), and the year (\(YR\)) as the control variables. Among them, auditor professional competence includes variables such as number of Certified Public Accountants (\(CPAs\)), number of professional standard-setters in the industry (\(PS\)), number of people who compile and mark Certified Public Accountant exams (\(PE)\), number of continuing education lectures in the industry (\(ET\)) and number of practice quality inspections conducted in the industry (\(QI)\). The variable definitions and descriptions are shown in Table 2.

Table 2 Definition and description of all variables.
Full size table

Empirical econometric model design

Based on variable definitions and research hypotheses, this paper constructs the following econometric model:

Modeling the effect of the adequacy of digital audit talent on the level of detection risk

$$DR={i}_{1}+c\times IT+{\beta }_{1}\times V+{\varepsilon }_{1}$$
(1)

Equation (1) represents the extent to which the adequacy of digital audit talent affects the level of detection risk, i.e., the magnitude of the change in the level of detection risk triggered by a one-unit change in the adequacy of digital audit talent when control variables are included. Where \(V\) is the control variable and \({\upvarepsilon }_{1}\) is the error term in the model.

Modeling of the effect of the adequacy of digital audit talent on the degree of audit digitization

$$AD={i}_{2}+a\times IT+{\beta }_{2}\times V+{\varepsilon }_{2}$$
(2)

Equation (2) represents the magnitude of the change in degree of audit digitization triggered by a one-unit change in the adequacy of digital audit talent. Where \(V\) is the control variable and \({\varepsilon }_{2}\) is the error term in the model.

Modeling of the effect of the adequacy of digital audit talent on the level of detection risk through the degree of audit digitization

$$DR={i}_{3}+c{\prime}\times IT+b\times AD+{\beta }_{3}\times V+{\varepsilon }_{3}$$
(3)

Equation (3) represents the pathway where the adequacy of digital audit talent affects the level of detection risk through its impact on the degree of audit digitization, i.e., a one-unit change in the adequacy of digital audit talent causes a change of \(b\) units in degree of audit digitization, which ultimately leads to a change in the level of detection risk. Where \(V\) is the control variable, and \({\varepsilon }_{3}\) is the error term in the model.

Model empirical evidence and testing

In testing whether the adequacy of digital audit talent can reduce the level of detection risk by influencing the degree of audit digitization, this paper substitutes all indicator data into the regression equations to calculate the coefficients, thereby obtaining the three regression equations for the mediating effect. The significance of the results is then used to determine whether the model holds.

Descriptive statistics

In this paper, variable data were subjected to descriptive statistics using \(Stata\) 17 software, and the results are shown in Table 3.

Table 3 Descriptive statistics.
Full size table

In Table 3, the explained variable has a mean (\(Mean\)) of 2.20, a standard deviation (\(SD\)) of 5.40, a median (\(Median\)) of 0, and a maximum (\(Max\)) of 16.91; the explanatory variable has a mean of 1.46, a median of 0, and a maximum value of 70. Both variables exhibit a higher mean than median, indicating the presence of extreme large values in the sample. This aligns with the current status of audit digitization implementation and transformation among Chinese audit institutions, where the degree of digitization varies significantly.

Regression results

The regression results for research hypotheses H1–H3 are shown in Table 4 below.

Table 4 Regression results for model variables.
Full size table

Assumptions H1 validation scenarios

Column (1) presents the relationship between the explanatory variable “adequacy of digital audit talent” and the explained variable “level of detection risk”. The results show that: \(DR=-0.131\times IT+{\varepsilon }_{1}\), i.e., for each unit increase in the adequacy of digital audit talent, the level of detection risk decreases by 0.131 units. Moreover, it is significant at the level of 10%, indicating that the adequacy of digital audit talent has a marginally significant negative impact on the level of detection risk.

Thus, research hypothesis H1 is supported: the greater the adequacy of digital audit talent, the lower the level of detection risk.

Assumptions H2 validation scenarios

Column (2) presents the effect of the explanatory variables on the mediating variables. The results show that:\(AD=0.017\times IT+{\varepsilon }_{2}\), i.e., for each unit increase in the adequacy of digital audit talent, the degree of audit digitization increases by 0.017 units, and the effect is significant at the 10% level. This indicates that the adequacy of digital audit talent has a marginally significant positive effect on the degree of audit digitization.

It follows that H2 is validated: the higher the adequacy of digital audit talent, the higher the degree of audit digitization.

Assumptions H3 validation scenarios

The results of Column (3) show that:\(DR=-0.148\times IT+1.008\times AD+{\varepsilon }_{3}\). This indicates that, while controlling for the degree of auditing digitization, the direct negative impact of the sufficiency of digital auditing talents on the level of detection risk is -0.148 and is significant at the 5% level (p < 0.05). This suggests that an increase in the adequacy of digital audit talent reduces the level of detection risk. Thus, the H3 is verified. A detailed breakdown of the mediating and direct effects is provided in Table 5.

Table 5 Effect types, coefficients, and interpretations.
Full size table

The coefficient of \(AD\)→\(DR\) in the model is 1.008** (significantly positive). This finding can be attributed to several factors: Firstly, during the process of digital transformation, audit institutions may encounter challenges related to technology implementation, staff training, and process adjustments. Secondly, they may face new technological risks, including data security vulnerabilities and system failure risks, which can result in a temporary increase in the level of detection risk. Additionally, successful digital transformation necessitates extensive internal support and coordination within audit institutions. If there is resistance to change within the audit institution, the full potential of digital technologies may not be realized, thereby introducing new risks.

Robustness tests—replacement of explanatory variables

In this paper, we adopt the method of replacing the explanatory variables to conduct a robustness test on the regression results38, replacing the original “the adequacy of digital audit talent” with “the number of IT personnel in audit institution”. These data come directly from the ranking document of accounting firms released by the \(CICPA\), using raw, unprocessed data. As shown in Table 6, the regression test results are basically consistent with the previous findings, indicating that the constructed model has strong robustness.

Table 6 Robustness tests—replacing explanatory variables.
Full size table

From the results in Table 6, after replacing the explanatory variables, the core explanatory variable’s regression coefficient remains significant at the 1% level. This indicates that the results are robust to different measures of the explanatory variable, i.e., the larger the number of digital audit talent in the audit organization, the more the level of detection risk is reduced.

Heterogeneity tests—effect of audit firm size

Existing studies have shown that the size of the audit institution affects the identification of detection risk levels39. Specifically, the larger the size of the audit institution, the more capable it is of adapting to the changes brought about by the digital transformation of auditing—equipped with more professional resources and technical capabilities—and thus can accurately assess detection risk levels and improve audit quality40. Based on the above analysis, this paper argues that the larger the audit institution, the more it can strengthen the inhibiting effect of audit digital transformation on the level of detection risk.

In this study, we incorporate the policy issued by the Ministry of Finance and the State Administration for Industry and Commerce, which encourages large and medium-sized audit institutions to adopt the organizational form of a special general partnership. Firms are categorized into two groups based on their annual revenues: large firms (those with revenue greater than or equal to the median) and small firms (those with revenue less than the median). We use the SORT function in Excel to classify the firms into these two categories for comparative analysis, utilizing revenue data from 2020 to 2022. Specifically, category C1 comprises firms with annual revenues exceeding the median, while category C2 includes firms with annual revenues below the median. The aim of this study is to investigate the heterogeneity in the relationship between the adequacy of digital audit talent and the level of detection risk across firms of varying revenue sizes. The results of this study are presented in Table 7.

Table 7 Analysis of model heterogeneity.
Full size table

From the results of Table 7, it is found that model validation for accounting firms with annual revenues below the median (hereinafter collectively referred to as ‘small firms’) shows that a greater adequacy of digital audit talent reduces the level of detection risk. In contrast, the results for accounting firms with annual revenues above the median (hereinafter collectively referred to as ‘large firms’) show that more digital audit talent increases the level of detection risk.

The likely reason for this is that small firms usually have limited resources: the adequacy of digital audit talent significantly enhances their technical competence, enabling more effective use of digital tools to identify and reduce the level of detection risk. The new technologies and ideas brought by these talents can generate greater marginal benefits in small firms, helping firms to better cope with the complex and changing audit environment, improve audit quality, and reduce the level of detection risk.

In contrast, large firms are typically rich in resources, possess advanced technical equipment, and have established audit processes and quality control systems. However, the adequacy of digital audit talent can introduce new challenges. Firstly, large firms often have complex interpersonal relationships and organizational structures, which may require time for newly recruited digital audit professionals to adapt and integrate. This adjustment period can temporarily affect the efficiency and coordination of audit activities. Secondly, the adequacy of digital audit talent may lead to an over-complexity in technology application or inconsistencies across different departments, thereby increasing communication costs and coordination difficulties during the audit process. Additionally, if large firms fail to effectively integrate the new technologies introduced by digital audit professionals with existing audit processes, it may result in resource wastage and an elevated level of detection risk. Large firms may also be more inclined to undertake large, complex audit projects that carry inherent risks. Consequently, digital audit talent may face heightened pressure and challenges in managing these complex projects, potentially leading to an increase in the level of detection risk if project management and risk controls are inadequate. In contrast, smaller firms, due to the relatively limited scope of their projects, can leverage their professional advantages more effectively, resulting in more significant reductions in the level of detection risk.

In summary, the relationship between the size of audit institutions and the adequacy of digital audit talent as well as the adequacy of digital audit talent and the level of detection risk warrants further research and attention. Future studies could investigate the specific practices and effects of audit institutions of varying sizes in managing the level of detection risk, as well as strategies for promoting the healthy development of the industry through regulation and policy guidance. Additionally, audit firms should develop a reasonable strategy for the allocation of audit resources based on their specific circumstances to achieve an optimal alignment between the level of detection risk management and the enhancement of audit quality.

Discussion and strategy proposal

This study empirically examines the impact of the adequacy of digital audit talent on the level of detection risk, with a focus on the mediating role of the degree of audit digitization, using panel data disclosed on the official website of the \(\text{CICPA}\) from 2020 to 2022. The findings reveal three interrelated insights: (1) the adequacy of digital audit talent significantly reduces the level of detection risk through direct effects; (2) the degree of audit digitization acts as a complex mediator, with a suppression effect that partially offsets the risk-reducing benefits of talent; and (3) the relationship exhibits striking heterogeneity across firms of different sizes. In the following sections, we will elaborate on these findings, including their alignment with existing literature, theoretical contributions, and practical implications.

Analysis of the empirical results

Direct effect of digital audit talent: validation of strategic talent value

The core finding that the adequacy of digital audit talent negatively correlates with the level of detection risk (H1 supported, direct effect = -0.148, p < 0.05) aligns with the Resource-Based View (RBV) of the firm, which emphasizes heterogeneous resources as sources of competitive advantage (Bedard, 1989)3 Digital audit talent, defined as professionals with dual competencies in traditional auditing and digital technologies(Zhang et al. 2021)3, constitute such a strategic resource due to their scarcity and inimitability.

Specifically, these talents may mitigate risk through three interrelated mechanisms: First, by leveraging tools like RPA (Zhang, 2019)9 and data regression systems (Xu et al. 2019)11, they transform “sampling risk”—a fundamental limitation of traditional auditing—into “algorithm optimization risk,” which can be continuously reduced through iterative model improvement. For instance, auditors proficient in data regression techniques can effectively overcome sample selection biases, thereby enabling more accurate risk assessments. Second, their ability to synthesize auditing expertise with digital skills (Qasim & Kharbat, 2020)3 enables more accurate identification of hidden risks in unstructured data, such as anomalies in client transaction logs or system access records. Third, They design closed-loop workflows that reduce human intervention, thereby minimizing error-related risks (Babina, 2020)3.

Mediating role of audit digitization: a double-edged sword

While digital audit talent promotes audit digitization (H3 supported, coefficient = 0.017, p < 0.1), the mediating effect of digitization itself is unexpectedly positive (\(AD\)→\(DR\) coefficient = 1.008, p < 0.05), resulting in a suppression effect. This means that while talent directly reduces risk, the digitization process partially offsets this benefit—an insight that enriches our understanding of technology’s role in auditing.

This finding extends Krahel and Titera’s30 proposition that digitization acts as a “bridge” between talent and efficiency by revealing its dark side. The short-term risks brought about by digitization may stem from: (1) technical challenges, such as data security vulnerabilities (Atadoga et al. 2024)10 and algorithmic biases (Kokina and Davenport 2017)23; (2) organizational frictions, including resistance to change (Haislip et al. 2016)35 and misalignment between new technologies and existing processes; and (3) skill gaps, where rapid digitization outpaces staff adaptability, leading to improper tool usage. For example, firms adopting cloud computing (Atadoga et al. 2024)10 may face integration issues that compromise data integrity, temporarily increasing the level of detection risk.

Notably, this suppression effect does not diminish the long-term value of digitization; rather, it underscores the transitional costs associated with it. As observed by Lugli and Bertacchini28 in the Italian context, the benefits of digitization—such as improved risk coverage—emerge gradually. And with the further improvement of audit digitization—such as the enhancement of digital audit talents’ skills, the matching of technologies with processes, and the optimization of algorithms—this inhibitory effect may diminish. However, it is essential to manage the short-term disruptions that accompany this process, a nuance that prior studies have often overlooked by focusing solely on its positive impacts.

Therefore, we recommend that auditing institutions should continue to increase the recruitment and cultivation of digital auditing talents. In the long run, the benefits will outweigh the short-term impact.

Heterogeneity by firm size: resource constraints and organizational complexity

The size-based heterogeneity presented in Table 7 challenges the conventional assumption that larger firms are better positioned to leverage digital transformation (Vasarhelyi et al. 2012)40. In contrast, our findings indicate that small firms significantly benefit from digital talent (coefficient = − 0.149, p < 0.1). This may be due to their limited resources, which amplify the marginal value of talent in enhancing technical capabilities. Furthermore, these firms often lack established digital infrastructure, making talent-driven digitization a critical solution for addressing gaps in risk detection. Large firms demonstrate a counterintuitive positive correlation (coefficient = 0.535, p < 0.001), whereby an increase in digital talent temporarily elevates the level of detection risk. This phenomenon likely arises from organizational complexity; rigid hierarchies, siloed departments, and coordination costs (Anderson-Gough et al. 2024)15 impede the effective integration of new talent and technologies. For instance, large firms may encounter conflicts between legacy audit processes and emerging digital tools, resulting in inefficiencies or errors. This divergence highlights the ineffectiveness of “one-size-fits-all” digital strategies. Small firms should prioritize talent acquisition to establish fundamental capabilities. Large firms should concentrate on organizational alignment—such as streamlining processes and fostering cross-departmental collaboration—to fully leverage the potential of their talent.

Theoretical contributions

This study advances academic discourse in three ways:

  1. (1)

    Extending the Resource-Based View: By quantifying how digital talent mitigates risk through both direct and indirect pathways (via digitization), we substantiate the role of talent as a strategic asset in risk governance, extending beyond its traditional function of enhancing efficiency.

  2. (2)

    Unpacking Digitization’s Complexity: The identification of a suppression effect within the mediation model challenges linear perspectives on technology adoption, emphasizing that digitization entails trade-offs between short-term disruptions and long-term benefits.

  3. (3)

    Recontextualizing Firm Size Effects: Our findings enhance the understanding of size-related dynamics by demonstrating that organizational complexity, rather than mere resource abundance, significantly influences the outcomes of digital transformation.

Policy recommendations

The digital audit talent cultivation system should be further strengthened

This study recommends advancing the construction of the cultivation system from three dimensions. First, universities should establish an interdisciplinary curriculum system. Taking the University of Pennsylvania’s “Audit Analytics and Technology” program as a model, courses such as Python programming and blockchain auditing should be designated as compulsory modules, ensuring that technical courses comprise no less than 40% of the total instructional hours. Furthermore, the Ministry of Education could incorporate “digital audit” into discipline evaluation criteria to promote the joint development of certification courses by universities and accounting firms. Second, the auditing industry could establish a mandatory digital competency certification system, requiring \(CPAs\) to complete training in machine learning, natural language processing, and other relevant areas. This training could be linked to \(CPAs\)’ continuing education credits, thereby intensifying the cultivation of digital audit talent. Third, the capability requirements for audit instructors should be further enhanced by implementing a ‘dual-competency’ faculty reform. This reform would mandate that instructors obtain tool certifications, such as SAS and Tableau, and incorporate digital competencies into the Class A accreditation standards for the audit majors.

The audit institutions should optimize the allocation mechanism of digital audit talents to enhance the efficiency of talent utilization

A “tiered strategy” should be adopted for firms of different sizes.

Given the limited resources of small firms, the focus should be on recruiting 1–2 multi-skilled technical talents, such as professionals with both Certified Public Accountant qualifications and Python programming skills. These individuals can prioritize addressing the firm’s data cleaning and basic automation needs, leveraging minimal investment for maximum efficiency gains. For small firms, SaaS-based auditing tools are a more ideal choice. Taking Alibaba Cloud’s auditing suite as an example, it has the characteristics of convenient deployment and low cost, which can effectively reduce infrastructure investment, while avoiding the complex technical integration problems caused by excessive pursuit of digitization, thereby avoiding the inhibitory effects that may occur in the initial stage of digital transformation. The successful case of “lightweight transformation” in the Singapore SME Digitization Programme provides a referential experience model for the digital practice of small firms.

For large firms, establishing a cross-departmental digital committee is a crucial step. Drawing on Deloitte’s “Digital Transformation Office” model, this committee, composed of professionals from various departments such as audit, IT, and risk management, can break down departmental barriers and organically embed digital talent into the entire audit project process. This approach effectively reduces communication and coordination costs between departments, enhancing the application efficiency and collaborative effect of digital technology in audit work. Furthermore, given the high technical integration risks faced by large firms during digital transformation, establishing a digital transformation risk reserve is necessary. It is recommended to allocate 3–5% of the annual budget as a risk reserve, specifically to address potential data security incidents, technical failures, and other risk events. This mechanism can provide the firm with the necessary financial support when risks occur, ensuring the continuous advancement of digital transformation efforts and reducing the impact of risks on business operations.

The government should increase investment support for digital construction

Regulatory bodies should adopt a “dynamic guidance framework,” implementing different policies for different periods and stages. Within 1–2 years, regulators should focus on establishing a “digital maturity rating system.” This system should quantitatively assess the level of digital development of firms, especially for firms with an Adoption of Digitization < 30%, with a focus on monitoring the compliance of their technology applications, including whether data desensitization processing meets standards and whether algorithm transparency meets requirements. Through strict supervision, firms will be urged to standardize digital operations and reduce potential risks.

Within 3–5 years, regulatory bodies should introduce a “suppression effect mitigation plan.” For large firms, detailed process integration guidelines should be provided, such as referencing the ISO/IEC 17025 audit digitization standard, to help them optimize internal processes and improve the success rate of digital transformation. For small firms, talent training subsidies should be provided to encourage them to strengthen the cultivation and introduction of digital audit talent, and improve overall digital capabilities.

After 5 years, regulatory bodies can incorporate suppression effect indicators, such as the matching degree of technology and processes, and the adaptation degree of talent skills and digital needs, into the audit quality evaluation system. Abandoning the previous single “digital investment” assessment method, a more comprehensive and scientific evaluation mechanism should be established to guide firms to continuously focus on the quality and effect of digital transformation, and promote the healthy digital development of the entire industry.

Limitations of the present study

It should be noted that this study also has some limitations. This study is based on audit industry data for the period of 2020–2022, and although it verifies the significant impact of the adequacy of digital audit talent on the level of detection risk through the degree of audit digitization, data limitations (scope and period) may have a certain impact on the comprehensiveness and representativeness of the study’s conclusions.

Future research should aim to further expand in two key areas. First, it is essential to obtain richer, multidimensional data samples, particularly long-term panel data across various regions and industries. This approach will help reveal the dynamic evolution of the relationship between the adequacy of digital audit talent and the level of detection risk. Second, it is vital to explore other potential mediating or moderating variables, such as the maturity of the technological environment and the level of internal corporate governance. This exploration will provide a more comprehensive understanding of the complex mechanisms through which digital audit transformation influences audit quality and risk control, as well as the impact of the adequacy of digital audit talent on these aspects. Third, it is necessary to compare compare cross-country patterns, particularly between civil law (e.g., China) and common law (e.g., U.S.) jurisdictions.

Conclusions and outlooks

This study empirically examines the impact of the adequacy of digital audit talent on the level of detection risk, with a focus on the mediating role of audit digitization, using panel data from \(CICPA\) (2020–2022). The core findings are summarized as follows:

  1. (1)

    Direct risk-mitigating effect of digital audit talent: The adequacy of digital audit talent significantly reduces detection risk (direct effect = -0.148**, p < 0.05). This effect is independent of the degree of audit digitization, emphasizing that talent serves as a foundational strategic resource for risk control, operating through mechanisms such as RPA-driven process automation and data regression-based full-sample analysis.

  2. (2)

    Dual role of audit digitization as a “double-edged sword”: While digital talent promotes audit digitization (coefficient = 0.017*, p < 0.1), digitization itself exhibits a short-term suppressing effect (AD → DR) (coefficient = 1.008**, p < 0.05). The critical point for its risk transformation lies in the synergy between technology-process matching and the upgrading of talent skills.

  3. (3)

    Size-dependent heterogeneity: Small firms benefit significantly from digital talent (coefficient = − 0.149*, p < 0.1) due to their flat organizational structures, while large firms face amplified risk in the short term (coefficient = 0.535***, p < 0.001) due to coordination costs in complex hierarchies. This necessitates large firms to unlock the value of talent through organizational restructuring.

In theoretical terms, this study extends the Resource-Based View by positioning digital audit talent as a strategic asset in risk management and unpacks the complex mechanisms of audit digitization. Practically, this study provides actionable insights: small firms should prioritize the acquisition of digital talent to establish foundational capabilities; large firms need to streamline their organizational processes to facilitate the integration of technology; and policymakers should implement differentiated support measures, such as talent subsidies for small firms and integration consulting for large firms, to promote balanced development within the industry.

In an era of escalating regulatory penalties and digital transformation pressures, this study underscores digital audit talent as the linchpin balancing risk control and technological innovation. While short-term digitization frictions are inevitable, sustained investment in talent—paired with context-appropriate strategies—will drive long-term improvements in audit quality. These findings provide a roadmap for emerging markets navigating the complexities of audit digitization.