CHINA JOURNAL OF ACCOUNTING STUDIES 2019, VOL. 7, NO. 3, 345–363 https://doi.org/10.1080/21697213.2019.1701252 ARTICLE Is the national talent project eﬀective? Evidence from the Chinese academic accounting leading talents project a b a Yin Xingqiang , Hu Ning and Zhang Limin a b School of Accountancy, Shanghai University of Finance and Economics, Shanghai, China; School of Accountancy, Southwestern University of Finance and Economics, Chengdu, China ABSTRACT KEYWORDS Talent project evaluation; Based on the list of re-examination and selection of the two sessions Chinese academic of China’s Academic Accounting Leading Talent Project of the accounting leading talents; Ministry of Finance, we directly evaluate the policy eﬀect of the talent Diﬀerence in diﬀerence project with a Diﬀerence-in-Diﬀerence approach. We ﬁnd that, com- pared with the talents who are not selected in the project, the chosen talents publish more papers after being selected (18.2% higher), and more articles signed as the ﬁrst author. The incremental downloads and citation rates of these papers after the selection are also more signiﬁcant than those of the unselected ones. A series of sensitivity test also supports our main ﬁndings. Additional research ﬁnds that the results of the talent project are more signiﬁcant for scholars who published fewer papers in the past and for scholars who got docto- rates degree from ‘non-985’ universities. The above ﬁndings docu- ment that Chinese government projects aimed at the selection and cultivation of talent have a signiﬁcantly positive eﬀect. 1. Introduction In the face of increasingly ﬁerce competition from international technology and talents, the Central Committee of the Communist Party of China (CPC) and the State Council of the People’s Republic of China (PRC) have actively been breaking down barriers to personnel training, updating the concept of talent training, innovating talent training models, and promoting comprehensive reform of the personnel training system. Experts are selected and cultivated in various disciplines through a series of policies and methods such as Father of China’s Hydrogen Bomb, and the Two Academies (the Chinese Academy of Sciences, and the Chinese Academy of Engineering). Such experts have great signiﬁ- cance for and far-reaching inﬂuence on the invigoration of China through science and education and the policy on developing a quality workforce. To a certain extent, it has boosted the rapid development of China in the past 40 years. General Secretary Xi Jinping CONTACT Hu Ning email@example.com School of Accountancy, Southwestern University of Finance and Economics, Chengdu, China Paper accepted by Kangtao Ye. For example, in May 1985, Deng Xiaoping emphasised at the National Education Work Conference that “national strength and the economic development are increasingly dependent on the quality of workers and on the quantity and quality of intellectuals. “Subsequently, the CPC Central Committee and the State Council promulgated the document ‘Decision of the Central Committee of the Communist Party of China on the Reform of the Education System’ aimed at guiding ‘more talents and better talents’. © 2019 Accounting Society of China 346 Y. XINGQIANG ET AL. clearly stated in the report of the 19th National Congress that talents are strategic resources for realising national rejuvenation and winning international competitive sta- tus. Provinces actively compete to attract talent away from other provinces, further highlighting the importance of ‘prosperity for talented people’ (Chen, 2011). The new era brings new missions and challenges for the selection and cultivation of high-tech talents, who will directly aﬀect the continued growth of China’s economy and compre- hensive national strength. Reasonable and eﬀective human capital development is a key driver of economic growth (Benhabib & Spiegel, 1994; Islam, 1995; Yang, Gong, & Zhang, 2006), and is also the focus of policymakers’ continued attention. According to endogenous growth theory (also known as the ‘new growth theory’), knowledge is the only source of economic growth that can continue to receive incremental returns (Becker & Barro, 1988; King & Robson, 1993; Romer, 1987). Talent selection projects are important all over the world for promoting political oﬃcials or selecting social elites; this is especially true for China, which is still in a period of transition. The government has led the cultivation of a large number of professional talents, as highlighted by various Chinese talents search programmes, such as the Two Academies, the Changjiang Scholars Programme, the Thousand Talents programme, and the New Century Talent Project. Since the Ministry of Education and the Li Ka Shing Foundation of Hong Kong jointly funded the ‘Changjiang Scholars Award Scheme’ in 1998 to raise the academic status of China’s higher education institutions and revitalise China’s higher education, most provinces and cities in China have followed suit to attract talents. According to our statistics as shown in Table 1, until now there are more than 50 kinds of talents and scholars projects, often named after famous mountains, rivers, or regions. As various talent projects are being carried out in full swing, the performance of these projects has attracted the attention of the society and the government. On the one hand, the government selects relevant candidates by setting selection criteria (including mate- rial scoring, professional written tests, and interviews, etc.), by inviting experts and scholars with various professional backgrounds or industry expertise to carry out intensive training for selected talents, by matching funding to generate incentives (Adams & Griliches, 1998; Jacob & Lefgren, 2011; Payne & Siow, 2003), and by oﬀering congenial Table 1. Various academic titles in diﬀerent places in China. Named after rivers Named after mountains Named after regions Yellow River Scholars, Qianjiang Scholar Taishan Scholar, Huangshan Scholar Oasis Scholar, Yan Zhao Scholar Zhujiang Scholar, Wanjiang Scholar Huashan Scholar, Hengshan Scholar Chutian Scholar, Tianfu Scholar Minjiang scholars, Three Gorges Hengshan Scholar, Lushan Scholar San Qin Scholars, San Jin Scholars Scholars Zhijiang Scholar, Longjiang Scholar Tianshan Scholar, Everest Scholar Qianling Scholar, Ba Gui Scholar Xiangjiang Scholar, Songjiang Scholar Kunlun Scholar, Jinggangshan Beiyang Scholar, Qilu Scholar Scholar Liangjiang Scholar, Ganjiang Scholars Central Plains Scholar, Oriental Scholar Qiongzhou Scholar Statistics do not include all kinds of scholars in China. Following the ﬁrst talent policies introduced by Guangdong province, other places (such as Beijing, Shanghai, Zhejiang, Jiangsu, Xi’an, Chengdu, Chongqing, and Guizhou) issued preferential policies (involving settlement, child education, and remuneration packages) to attract talents. CHINA JOURNAL OF ACCOUNTING STUDIES 347 academic environments (Burt, 2001; Granovetter, 2005;Jaﬀe, Trajtenberg, & Henderson, 1993). It will ultimately contribute to the accumulation of human capital, and promote selected scientiﬁc research or academic exploration. That is to say, the leading talent project is human capital investment by the government, and the cultivated human capital will bring about the increase and improvement of output (Arora & Gambardella, 2005; Ouyang, Ye, Cui, Zhang, & Qiu, 2015; Tian, Sun, & Lu, 2015). On the other hand, because of information asymmetry, long-term continuous investment of the human capital, relatively slow output process (Azoulay, Zivin, & Manso, 2011), and vague indicators of measuring output, those entrusted by the government with the task of selection may easy to succumb to moral hazard or indulge in favouritism (Durante, Labartino, & Perotti, 2011; Fisman, Shi, Wang, & Xu, 2017; Zinovyeva & Bagues, 2015). What is more, some talent projects lack supervision and governance in later stages. Even if these projects can select excellent talents, some of them may stop focusing on academic work after they obtain resources, a high salary, or administrative positions. Talent projects supported by the government may have signiﬁcant increase in output, measured by publications and patents, due to the accumulation of human capital; however, information asymmetry and agency cost make it challenging to evaluate the real role of talent projects. Our central question is the following: Have various government-led talent selection projects successfully completed their expected goals, and ultimately promoted the increase in the output of selected talents? Although the relevant issues need to be urgently explored, there is little empirical literature on it; two exceptions are Jaﬀe(2002) and Jacob and Lefgren (2011). The Ministry of Finance launched the National Accounting Leadership (Backup) Talent Development Project in September 2005 to improve and cultivate a group of high-level accounting talents with broad research horizons, a knowledge base that is optimised for their jobs, practical experience, outstanding innovation ability, and high professional ethics. This is in line with the national strategy of strengthening the country through talents, by actively adapting to economic and social developmental needs. The Ministry of Finance has so far cumulatively recruited 1,658 scholars in 41 groups, in four categories – enterprises, administrative institutions, certiﬁed public accountants, and academics. Of these, 716 students have graduated after completion of the 6-year training programmes. Most of the relevant literature on the evaluation of accounting talent projects has conﬁrmed that innovation output (such as the publication of academic papers) increases after the selection of scholars; this is taken as evidence of successful implementation (Ouyang et al., 2015; Tian et al., 2015). Although this kind of research demonstrates eﬀectiveness of talent projects to a certain extent, it ignores candidates from the shortlist who were not selected; consequently, this literature might be measuring the learning eﬀect of participating in the programmes and the macro time trend, rather than the appropriateness of the initial selection. It is diﬃcult to directly and accurately assess the causal eﬀect of the implementation of the talent selection projects without considering scholars who were shortlisted but not selected – this is what we do. The talent projects disclose the list of candidates shortlisted and the ones ﬁnally selected. This selection and publication process has actually formed a quasi-natural experiment that provides a possibility to evaluate the eﬀect of the talent projects. Through the distribution of the re-test and the school ﬁnally selected, we ﬁnd that the geographical distribution of the re-examined and ﬁnally selected scholars is relatively 348 Y. XINGQIANG ET AL. uniform over 19 provinces, municipalities, and autonomous regions in China – the eastern coast (Shanghai, Guangdong, and Xiamen), the northeast (Dalian), the central (Wuhan and Hunan), the southwest (Sichuan, Chongqing, and Yunnan), and the northwest (Shanxi and Xinjiang). See Table 2 for details. This paper adopts the Diﬀerence-in-Diﬀerence (DID) model and attempts to directly evaluate the eﬀectiveness of the talent project. Using data from the re-examination and selection in two sessions of the academic accounting leading talent programmes under the Ministry of Finance, we compare the ﬁnalists and the unselected scholars in terms of their output changes after the selection process. Compared with scholars who were shortlisted but not selected, the ﬁnalists published more papers, and the probability of producing high-quality authoritative journals was 18.2% higher. In particular, selected scholars published 12 more papers on average in CSSCI journals compared to the unselected ones. Articles published by the selected scholars also had signiﬁcantly larger incremental downloads and citation rates. We also ﬁnd that, among those selected, the output increment is more signiﬁcant for scholars who had fewer publications in leading journals prior to their selection and got their doctoral degrees from ‘non-985’ universities. What is more, the number of papers co-authored by the selected talents after the selection also increased. This paper has three main contributions. First, talent selection projects may have apparent output eﬀects due to government support and investment; however, it may be challenging to play the role of talent cultivation due to information asymmetry or agency problems. Based on the theory of human capital, this paper enriches the research on the accumulation of human capital and the improvement of output by means of the research scene of the academic accounting of the Ministry of Finance. In particular, unlike the literature that has examined the general human capital accumulation and economic growth (e.g. Benhabib & Spiegel, 1994; Islam, 1995; Yang et al., 2006), this paper supple- ments and enriches the research about the outcomes of the accumulation of senior human capital from the perspective of professional talent cultivation. Secondly, because of lack of primary data and the ambiguity of measures of ability, most literature aﬃrms the validity of the talent projects (e.g. Ouyang et al., 2015; Tian et al., 2015), based on the change in academic achievements before and after selection, but this research result is not only vulnerable to individual learning eﬀects and selection errors but also ignores the prevailing inﬂuence of external environments, such as the diﬃculty of publishing in periodicals and time trends. So, it is diﬃcult to assess the causal eﬀect of talent project implementation comprehensively and accurately. Based on the data (the re-examination list) of two sessions of academic accounting leading talent projects of the Ministry of Finance, this paper uses Diﬀerent-in-Diﬀerence (DID) to reveal the causal eﬀect between the talent project and its innovation output, and directly evaluate the policy eﬀect of the talent project. Thirdly, since the establishment of the ‘Changjiang Reward Programme’, talent projects have been extensively implemented at both national and provincial levels. Has the government-driven talent project achieved the expected goal or not? Prior research has not given direct empirical evidence. This paper eﬀectively evaluates the policy eﬀects of talent projects through the implementation of the Ministry of Finance’s academic accounting leading talent project. It also provides precise policy evaluation and empirical reference for the selection and implementation of relevant government talent projects. CHINA JOURNAL OF ACCOUNTING STUDIES 349 Table 2. Overview of the talents entering the re-test and ﬁnal selection in 2011 and 2013. Panel A: The number of the talents entering the re-test and ﬁnal selection in 2011 University Number of the Number of the Proportion retested selected Central University of Finance and Economics 4 2 0.5 Renmin University of China 3 2 0.67 Sun Yat-sen University 3 2 0.67 Jinan University 3 1 0.33 Southwest University of Finance and Economics 3 2 0.67 Shanghai Lixin College of Accounting 2 1 0.5 Xiamen University 2 2 1 Shandong Economic University 2 1 0.5 Central South University of Finance, Economics and Law 1 0 0 Yunnan University of Finance and Economics 1 1 1 Inner Mongolia University 1 0 0 Inner Mongolia University of Technology 1 0 0 Beijing Jiaotong University 1 0 0 Beijing Business University 1 1 1 Beijing University of Technology 1 1 1 Beijing University of Aeronautics and Astronautics 1 0 0 South China University of Technology 1 1 1 Nanjing University of Finance and Economics 1 1 1 Sichuan University 1 1 1 Fudan University 1 1 1 Anhui University of Technology 1 0 0 Anhui University of Finance and Economics 1 1 1 Guangdong Institute of Finance 1 0 0 Xinjiang University Of Finance and Economics 1 1 1 Hangzhou University of Electronic Science and 10 0 Technology Wuhan University 1 1 1 Tsinghua University 1 1 1 Suzhou University 1 0 0 Zhengzhou Institute of Aeronautical Industry 11 1 Management Capital University of Economics and Business 1 1 1 Total 44 26 0.591 Panel B: The number of the talents entering the re-test and ﬁnal selection in 2013 University Number of the Number of the Proportion retested selected Central South University of Finance, Economics and Law 4 2 0.5 Central University of Finance and Economics 4 2 0.5 Dongbei University of Finance and Economics 3 2 0.67 Sun Yat-sen University 2 1 0.5 Beijing Business University 2 1 0.5 Xiamen University 2 2 1 Zhejiang Gongshang University 2 1 0.5 Southwest University of Finance and Economics 2 1 0.5 Xi’an Jiaotong University 2 1 0.5 Shanghai Jiaotong University 1 1 1 Shanghai Maritime University 1 0 0 Shanghai University of Finance and Economics 1 1 1 Renmin University of China 1 1 1 China University of Mining and Technology 1 0 0 Yunnan University of Finance and Economics 1 1 1 Inner Mongolia University 1 1 1 Beijing Jiaotong University 1 1 1 Beijing Foreign Studies University 1 1 1 Beijing University of Technology 1 0 0 Beijing University of Aeronautics and Astronautics 1 1 1 Beijing University of Posts and Telecommunications 1 1 1 East China University of Science and Technology 1 0 0 (Continued) 350 Y. XINGQIANG ET AL. Table 2. (Continued). Nanjing Audit University 1 1 1 Nanjing Institute of Finance and Economics 1 0 0 Sichuan University 1 0 0 Anhui University of Finance and Economics 1 0 0 University of Foreign Trade and Economics 1 1 1 Guangdong Institute of Business 1 1 1 Jinan University 1 0 0 Jiangxi University of Finance and Economics 1 1 1 Tsinghua University 1 1 1 Hunan University 1 1 1 Shihezi University 1 1 1 Fujian Agricultural and Forestry University 1 1 1 Suzhou University 1 1 1 Northwest University of Technology 1 0 0 Southwest University 1 1 1 Chongqing University 1 0 0 Capital University of Economics and Business 1 1 1 Total 53 33 0.623 The structure of the rest is as follows: The second section introduces the background of the leading talent project, summarises a possible channel through which the talent project aﬀects the innovation output based on the theory of human capital, and presents the research questions of this paper. The third section presents the design of the empirical research and oﬀers basic descriptive statistics. The fourth section oﬀers the test result, regression analysis, and tests of robustness. The ﬁfth section contains further analysis and discussion. The sixth section concludes and oﬀers policy implications. 2. Research background, theoretical analysis and research questions 2.1. Research background As a basic and applied discipline, accounting has for a long time played an irreplaceable role in economic transformation. The rapid and high-quality develop- ment of the economy requires high-quality accounting support. China’scontinuing education and personnel training in accounting has achieved remarkable results since the economic reform and opening up. It initially formed a network of con- tinuing education and training for junior, middle, and senior accounting personnel and accounting staﬀ nationwide, which has improved the professional skill and the knowledge structure of accountants. By the end of 2004, the number of accoun- tantswithjuniororintermediateaccounting professional technical qualiﬁcations was 5.168 million, constituting 52.48% of the total accounting staﬀ in China, while this proportion was 33.88% in 1993. The number of accounting personnel with senior accountant qualiﬁcations was 68 thousand, and the corresponding ratio rose from 0.4% in 1993 to 0.7% in 2004. However, with deepening economic globalisa- tion and rapid advancement in science and technology, the structure and overall quality of accountants cannot meet the needs of the economy and the society, especially when it comes to high-skilled and comprehensive senior accounting talents urgently needed for modernisation. In response, the Ministry of Finance issued the ‘Notice of Senior Accounting Talents Training from the Ministry of CHINA JOURNAL OF ACCOUNTING STUDIES 351 Finance’ and ‘Notice of the Programme for the approval of the Accounting Society of China (ASC) on the selection and training of senior accounting talents and accounting academic leaders reserve talents from the Ministry of Finance. These two notices initiated the selection and training of practical and academic account- ing leaders. The accounting talent project is an essential part of the construction of a talent team, and a necessary force and means to maintain the market economic order, promote scientiﬁc development, and promote social harmony. As a state-led professional talent training programme, the national accounting leading talent training programme is regarded as the most eﬀective strategic investment project in terms of input and social impact, bringing the latest training methods to China’s accounting professionals (Ouyang et al., 2015). Establishment of this programme is a milestone in the cultivation of academic scholarship in Chinese accounting. It not only provides a broad platform for academic accounting scholars but also further broadens the horizons of academic accounting research. It will provide a reliable way of bridging the gap between China and advanced countries in accounting research (Tian et al., 2015). Similar to other talent selection, the selection of leading academic accounting talents has three steps. First, the applicants submit relevant materials, such as education experience and previous research output, which the selection agency evaluates and grades in order to determine the list of participants for the initial test. Second, an initial written assessment is carried out. Most questions are sub- jective and divergent, which is used to test the participants’ ability to use basic theory while remaining open to speculation. On the basis of the original ratings in the ﬁrst step and the average scores in the ﬁrst test, the selection committee determines the list of participants for a re-test. This re-test is the third and ﬁnal step. It takes the form of an interview, which mainly examines the participants’ communication skills and on-the-spot resilience. The selection agency then deter- mines the ﬁnal list of the national accounting leaders based on the total score of the three steps. For example, the 2011 governmental academic accounting leading talent training and selection work was launched in April 2011. The initial selection had 133 outstanding talents from 83 universities in 26 regions (provinces, autonomous regions or municipalities). Of these 44 candidates were interviewed, and 26 of them were selected as the 2011 national academic accounting leader talents. The interview enrolment rate and interview pass rate were 33.08% and 59.09% respec- tively. Since the project discloses the list of candidates who enter the re-test and the ﬁnal selection, the process forms a quasi-natural experiment and provides an oppor- tunity to evaluate the talent project. In addition, unlike other comprehensive talent projects, the Ministry of Finance’s academic accounting leading talent project is a professional talent project and a state-led professional talent project with Chinese characteristics. Therefore, this scenario not only provides some empirical insights for other related talent selection project but also provides more reliable support for the evaluation of talent project policies. We thank two anonymous reviewers for encouraging us to introduce more institutional background and details of the selection process. 352 Y. XINGQIANG ET AL. 2.2. Theoretical analysis and research questions Due to the information asymmetry in the talent selection process and the uncertainty of innovation output, the success of the talent selection project is mainly determined by the fairness and reasonability in the selection process, and by whether it provides useful guidance and support for the scholars chosen. Actually, the eﬀectiveness of implementing the talent selection project mainly have the following two aspects. We discuss details of these two aspects below. On the positive side, the leading talent project implemented by the Ministry of Finance can promote research output by enhancing human capital of those selected. As men- tioned before, candidates are required to submit detailed professional information and go through a step-by-step screening mechanism that confers an informational advantage on the government when it comes to selecting talented persons. Second, the ASC will invite experts and professors from diﬀerent backgrounds to conduct intensive training in various academic and practical ﬁelds for the selected talents, which not only helps to broaden their horizons, optimise their knowledge structure, and enrich their practical experience, but also helps develop their ability to discern and strengthens their academic innovation (Azoulay et al., 2011). In addition, various resources or platforms provided by the government eﬀectively enhance the publication and patent output of schools and scholars (Adams & Griliches, 1998; Jacob & Lefgren, 2011; Payne & Siow, 2003). Many of the selected talents have similar academic backgrounds, basic knowledge, and academic interest, which undoubtedly greatly reduces the communication cost and learning cost among those selected (Iaria, Schwarz, & Waldinger, 2018). Frequent learning, commu- nication, and discussion are also conducive to the formation of certain social network relationships among the selected talents, and this social connection due to the social network is conducive to the formation of mutual competition and incentives (Coleman, 1988; Granovetter, 2005) and learning eﬀects (Burt, 2001;Jaﬀe, 1989;Jaﬀe et al., 1993), which in turn may lead to an increase in outcomes and output. Based on this the academic accounting leading talent selection project of the Ministry of Finance will help improve the output of the selected talents through human capital accumulation (Arora & Gambardella, 2005; Ouyang et al., 2015; Tian et al., 2015). On the negative side, given the information asymmetry, does the government have enough capacity to select relatively better talents from the pool of candidates, especially when departments or persons entrusted by the government may be subject to moral hazard and shirking (Harrison & Harrell, 1993)? If not, the project can hardly select the truly talented candidates. Bielby and Bielby (1999) shows that the organisational form of selection plays a key role improving the allocation of talent in the labour market. Studies have shown that the long-term continuous investment and relatively slow output process in the high human capital industry (for example, education, scientiﬁc research, music), lead to relatively vague evaluation indicators of the talents. Combined with the diﬀerences in the age and level of the candidates, this makes it diﬃcult to ﬁnd a uniﬁed evaluation system (Seifert & Hadida, 2006), thereby hindering the process of selecting outstanding candidates. Relational transactions based on closeness of social ties and geographical proximity is common in China. Therefore, insider bias in the selection process could render various talent projects ineﬀective (Fisman et al., 2017). Fisman et al. (2017) further show that the CHINA JOURNAL OF ACCOUNTING STUDIES 353 rent-seeking behaviour is typical in the selection process of Chinese academicians, which distorts the allocation of human capital, in turn reducing the eﬃciency of resource allocation. In addition, even when the agencies select outstanding talents following a rigorous selection plan, the eﬀectiveness of the selected projects is closely related to the external environment and personal characteristics of the talents (Linder & Peters, 1987). For example, when being selected as a young scholar can help one obtain scientiﬁc research resources, high salaries, or administrative positions, some of the chosen talents may stop at an opportune moment instead of focusing on academic research. That is to say, the selected skills may lack incentives to achieve the expected training objectives of the talent project due to weak post-supervisory governance and changes in personal preferences, which impairs the talent cultivation role of the project. Based on the above analysis, this paper attempts to explore the following two unre- solved and important questions: Does the accounting academic leading talent project improve the innovation output of selected talents? What is the mechanism through which a successful talent search project works, and what are the reasons behind a failed one? 3. Research design and descriptive statistics 3.1. Model design and variable deﬁnition In order to evaluate the eﬀect of the academic accounting leading talent project, we construct the following regression model (1): DOutput ¼ β þ β Nominated þ Control þ ε (1) i i 0 1 In model (1) the subscript i represents the person. Following the related literature (e.g. Guo & Li, 2017; Liu & Zhao, 2017), the dependent variable (DOutput) is the output of the talents. Because publication of academic papers is time-consuming, we adopt several diﬀerent measures of output – the change in the total number of papers before and after the selection year (Dnum), the change in the probability of publishing China’s author- itative journals (Dtopdum), the change of total ﬁrst-author papers (Dsumﬁrst), the number of times of the papers were downloaded (Dload), and the change in the number of citations (Dcite). We calculate the total (Sum) and the maximum (Max) values of the downloads and citations. The independent variable (Nominated) equals 1 if a scholar is nominated as a leading talent, and 0 otherwise. In order to avoid the selection bias caused by personal ability, we use the pre-selection output of the talents as a control variable; this includes the number of papers and the number of projects that were awarded a research grant. In particular, we check the following variables before the scholar was selected: whether at least one paper was published in a top Chinese journal (Beftop), the maximum number of the papers down- loaded (Befmaxload), and the number of grant/funding applications (Befproject). Further, we also control for other individual characteristics of the candidates, such as having administrative duties (Duty), gender (Sex), years after PhD (Time), whether they are returnees from abroad or not (Foreign), and whether or not they applied for the project The top journals mainly consist of Chinese Social Science, Economic Research, Economics (Quarterly), World Economy, Management World, Financial Research, Accounting Research, and Audit Research. The regression results of only China Social Science, Economic Research, and Management World as top journals are consistent with the main conclusions. 354 Y. XINGQIANG ET AL. the ﬁrst opportunity (Rep). We also control the school-level variables such as whether they come from 985 universities (Empsch), whether the university belongs to economics and ﬁnance (Ecosch), whether their incumbent mater is located in Beijing (Beijing), and the total number of people selected along with them (Anum). Table 3 shows the concrete deﬁnitions of the variables. We use the ordinary least square (OLS) regression with clustering by the individual level and adjusting the robust standard error and controlling the ﬁxed eﬀects of time and province. The data used in this paper comes primarily from the oﬃcial website of the Ministry of Finance of the PRC, the oﬃcial site of China’s Accounting Association, the oﬃcial website of Xiamen National Accounting Institute, Baidu Encyclopaedia, and the schools the scholars come from. We obtained 97 initial samples from the retests and ﬁnalists of the fourth (2011) and ﬁfth (2013) sessions of the national academic accounting leader of the Ministry of Finance. At the end we are left with 81 samples after excluding some samples with missing values. 3.2. Descriptive statistics and inter-group diﬀerence test The main descriptive statistics of variables are shown in Table 4.The diﬀerence CSSCI journals published before and after the selection of leading talents ranged from Table 3. Main variable deﬁnitions. Variable Variable type code Variable description Dependent variable Dnum The diﬀerence between the total number of CSSCI journals published before and after the selection. Dtopdum The diﬀerence in the probability of publishing a paper in China’s top journals before and after the selection. Dtopsum The diﬀerence between the total number of the papers published in China’s top journals before and after the selection. Dsumﬁrst The diﬀerence between the total number of ﬁrst-author papers published before and after the selection. Dload The diﬀerence between the total number of downloads before and after the selection. Dcite The diﬀerence between the number of citations before and after the selection. Independent variable Selected If a scholar is selected, the value of Selected is 1; 0 otherwise. Control variable Beftop If there is at least one paper that published in China’s top journals before the selection, Beftop equals to 1; 0 otherwise. Befproject The logarithm of 1 plus the number of national foundations before nomination. Befmaxload The logarithm of the sum of 1 and maximum downloads for the published papers published before nominated. Duty If a talent holds an administrative position above the branch level, Duty equals to 1; 0 otherwise. Sex If the gender of a talent is male, Sex equals to 1; 0 otherwise. Time The gap between the year when a talent obtained a doctoral degree and the year when he or she was nominated. Foreign If a talent has a experience of studying abroad, Foreign equals to 1; 0 otherwise. Rep If candidate participate in twice. Rep equals to 1, 0 otherwise. Ecosch If a talent comes from a ﬁnancial and economics university, Ecosch equals to 1; 0 otherwise. Empsch If a talent comes from a university that belongings to project 985, Empsch equals to 1; 0 otherwise. Beijing If a talent comes from a university that is located in Beijing, Beijing equals to 1; 0 otherwise. Anum The total number of people who entered the retest at the same class in the same session. This comprises the National Natural Science Foundation of China and the National Social Science Foundation. CHINA JOURNAL OF ACCOUNTING STUDIES 355 Table 4. Descriptive statistics of the main variables. Variable Mean Sd Min p25 p50 p75 Max Dnum 0.381 0.957 −2.079 −0.223 0.288 0.799 2.708 Dtopdum 0.111 0.570 −1.000 0.000 0.000 0.000 1.000 Dtopsum 0.199 0.756 −1.609 0.000 0.000 0.693 2.708 Dsumﬁrst 0.347 0.942 −2.079 −0.223 0.251 0.847 2.398 DMaxload 0.353 3.049 −8.008 −0.990 −0.080 0.748 9.353 DSumload 0.521 3.404 −8.617 −0.965 −0.034 0.985 10.940 DMaxcite −0.496 2.093 −5.666 −1.766 −0.740 0.172 5.889 DSumcite −0.446 2.437 −6.541 −1.819 −0.850 0.454 6.979 Selected 0.580 0.497 0.000 0.000 1.000 1.000 1.000 Beftop 0.556 0.500 0.000 0.000 1.000 1.000 1.000 Befpronum 0.618 0.478 0.000 0.000 0.693 1.099 1.792 Befmaxload 6.961 2.683 0.000 6.798 7.847 8.507 9.613 Duty 0.494 0.503 0.000 0.000 0.000 1.000 1.000 Sex 0.716 0.454 0.000 0.000 1.000 1.000 1.000 Time 4.049 1.809 1.000 3.000 4.000 5.000 10.000 Foreign 0.111 0.316 0.000 0.000 0.000 0.000 1.000 Rep 0.099 0.300 0.000 0.000 0.000 0.000 1.000 Ecosch 0.333 0.474 0.000 0.000 0.000 1.000 1.000 Empsch 0.309 0.465 0.000 0.000 0.000 1.000 1.000 Beijing 0.296 0.459 0.000 0.000 0.000 1.000 1.000 Anum 1.914 1.120 1.000 1.000 1.000 3.000 4.000 −2.079 to 2.708, with a mean value of 0.381 greater than 0, which indicates that the signiﬁcant changes CSSCI journal articles published by leading talents before and after the selection. And on average, the number of publications has increased since the selection. From the maximum, minimum, mean and standard deviation, whether it is the diﬀerence between the probability of publishing an authoritative journal (Dtopdum) and the number (Dtopsum), or the diﬀerence between the total ﬁrst- author number of published articles (Dﬁrstsum) and the downloads (Dload) and the citation (Dcite) of the published papers are quite diﬀerent in the sample interval. Overall, there was an absolute increase in results after the selection. The nominated proportion of the two leading talents was 58%, that is, 52% of the skills only entered the re-examination and were not nominated, which also provided the possibility for the problems to be studied in this paper. In addition, from the descriptive statistics, it can also get that 9.9% of the participants in the sample have two experiences, 11.1% are overseas, nearly 30% incumbent school are from Beijing, and 30.9% are from 985 universities. A third of the students are from ﬁnancial institutions. This paper mainly uses the idea of Diﬀerence-in-Diﬀerences (DID) model to evaluate the policy eﬀect of talent projects, and an important assumption of DID is that the parallel trend hypothesis is met. For this reason, we deﬁne Treat that if it is nominated as 1, conversely, it is 0, and respectively test the diﬀerence between treatment group (Treat = 1) and control group (Treat = 0) before and after their selection. As shown in Table 5, before the selection, the only variable to diﬀer signiﬁcantly between the selected and unselected candidates is the number of top-journal articles published before the selection date. This variation is desirable as it satisﬁes the premise of the DID regression model. After selection these talents increased their output by signiﬁcantly more than those who were not selected. 356 Y. XINGQIANG ET AL. Table 5. Diﬀerence test between the nominated or not. Variable Before being selected (Bef) After being selected (Aft) Output Types Treat = 0 Treat = 1 Diﬀ Chi2 Treat = 0 Treat = 1 Diﬀ Chi2 Num Mean 1.849 1.935 −0.086 0.058 1.746 2.023 −0.276* 2.547 Median 1.869 1.946 1.792 2.079 Topdum Mean 0.618 0.809 −0.191* 0.000 0.529 0.809 −0.279*** 0.000 Median 1.000 1.000 1.000 1.000 Topsum Mean 0.697 1.062 −0.365** 3.562* 0.705 1.038 −0.334** 0.852 Median 0.693 1.099 0.693 1.099 Sumﬁrst Mean 13.353 15.085 −1.732 0.479 12.412 17.936 −5.524 2.913* Median 7.000 8.000 6.000 12.000 Maxload Mean 7.061 7.347 −0.286 4.653** 6.477 6.826 −0.350** 2.913* Median 7.010 7.469 6.597 6.934 Sumload Mean 7.825 8.109 −0.284 1.579 7.170 7.761 −0.592*** 9.350*** Median 7.838 8.283 7.207 7.856 Maxcite Mean 4.355 4.544 −0.189 0.127 2.872 3.452 −0.580** 1.772 Median 4.494 4.635 3.332 3.555 Sumcite Mean 5.083 5.372 −0.289 1.579 3.396 4.093 −0.697** 1.579 Median 5.257 5.565 3.670 4.205 (1) *, **, *** respectively indicate the signiﬁcance level of 10%, 5%, 1% (the same below); (2) mean median test using the ttable3 command, Diﬀ value is the mean diﬀerence test result, The Chi2 value is the result of the median diﬀerence test. 4. Empirical results and regression analysis 4.1. Basic regression results Table 6 reports on the academic achievements of selected talents before and after the selection. The results show that compared with those who weren’t selected, the selected talents signiﬁcantly increase their output of papers after the selection: 12 more papers on average, more than 0.492 extra CSSCI journal publications on average, an 18.2% higher probability of publications in top journals, and 0.3 extra publications in top journals. We use Max or Sum, both downloads (Dload) and citations (Dcite), and the conclusion stands. Compared with those who were not selected, the number and quality of academic achievement of the nominated leading talents are signiﬁcantly higher. This partly vali- dates the output incentive eﬀect of the leading talent projects. 4.2. Robustness test In this section we test the robustness of the regression results and conclusions. 4.2.1. Redeﬁning the time interval We redeﬁne the dependent variable in the main test to allow for shorter windows around the time of selection – three years in Panel A, and ﬁve years in Panel B. The regression results are shown in Table 7. The variables Dload and Dcite are still signiﬁcant but less so, possibly because publication takes a long time to disseminate. The above results still support the main research conclusions broadly. 4.2.2. Alternative measures of scholarly achievement In the theoretical analysis to adapt to the current demands of economic and social development, an essential goal of the accounting talent project mentioned is to train CHINA JOURNAL OF ACCOUNTING STUDIES 357 Table 6. Successful selection and academic achievement. (1) (2) (3) (4) (5) (6) (7) (8) Dload Dcite Output Dnum Dtopdum Dtopsum DSumﬁrst Max Sum Max Sum Selected 0.492** 0.182* 0.260* 11.727** 0.384* 0.617** 0.619* 0.656* (2.62) (1.76) (1.71) (2.03) (1.86) (2.16) (1.90) (1.81) Beﬀtop −0.363 −0.985*** −1.026*** 0.815 −0.069 −0.385 0.027 0.159 (−1.35) (−5.84) (−5.12) (0.11) (−0.23) (−0.89) (0.05) (0.31) Befmaxload −0.064 0.138*** 0.136* −2.041 −0.835*** −0.765*** −1.026*** −1.067*** (−0.46) (2.94) (1.80) (−0.75) (−7.61) (−3.83) (−5.94) (−4.96) Befproject −0.079 −0.022 −0.277 −8.911 0.373 0.243 0.366 0.402 (−0.32) (−0.21) (−1.60) (−1.06) (1.53) (0.72) (1.31) (1.11) Rep 0.244 0.069 0.401 6.144 0.113 0.516 0.344 0.491 (0.85) (0.53) (1.44) (0.82) (0.39) (1.17) (0.79) (0.90) Anum 0.103 0.087** 0.175** 2.027 0.145 0.203 0.081 0.080 (1.10) (2.02) (2.10) (0.74) (1.36) (1.43) (0.56) (0.47) Beijing −0.191 0.078 −0.116 −10.12 0.184 −0.023 −0.058 −0.205 (−0.87) (0.80) (−0.60) (−1.58) (0.84) (−0.07) (−0.21) (−0.54) Sex 0.095 0.339*** 0.347* −0.222 0.698*** 0.662* 0.672* 0.628 (0.39) (3.18) (1.85) (−0.04) (3.09) (1.84) (1.71) (1.34) Duty −0.401** −0.241** −0.347** −8.584* −0.160 −0.273 −0.036 −0.132 (−2.41) (−2.38) (−2.61) (−1.93) (−0.75) (−1.04) (−0.13) (−0.41) Time −0.177*** 0.014 −0.029 −5.507*** −0.058 −0.161** −0.145* −0.196** (−3.35) (0.48) (−0.73) (−3.65) (−0.99) (−2.14) (−1.95) (−2.18) Foreign 0.046 0.277 0.430** 5.119 0.117 0.089 −0.441 −0.109 (0.21) (1.50) (2.07) (0.82) (0.42) (0.25) (−0.96) (−0.23) Ecosch 0.046 0.150 0.053 6.292 −0.090 0.019 0.049 0.116 (0.15) (1.08) (0.21) (0.87) (−0.26) (0.04) (0.10) (0.20) Empsch −0.004 0.129 0.029 8.422 0.000 0.199 −0.016 0.170 (−0.02) (1.10) (0.15) (1.44) (0.00) (0.71) (−0.05) (0.49) Cons 1.457 −0.977*** −0.515 36.062* 5.674*** 5.643*** 6.797*** 7.141*** (1.36) (−2.81) (−0.84) (1.75) (6.61) (3.71) (5.33) (4.47) N 818181 81 818181 81 Year/province yes yes yes yes yes yes yes yes R 0.358 0.592 0.489 0.314 0.609 0.523 0.519 0.490 The values in parentheses are t values, and *, **, and *** indicate the signiﬁcance level of 10%, 5%, and 1%, respectively. Table 7. The result of resetting time interval. Panel A: Output three years changes before and after the selection. (1) (2) (3) (4) (5) (6) (7) (8) Dload3 Dcite3 Output Dnum3 Dtopdum3 Dtopsum3 DSumﬁrst3 Max Sum Max Sum Selected 0.400* 0.309** 0.235 0.518** 0.897 1.010 0.471 0.567 (1.92) (2.12) (1.22) (2.25) (1.38) (1.44) (1.25) (1.29) Control yes yes yes yes yes yes yes yes N 81 81 81 81 818181 81 Year/province yes yes yes yes yes yes yes yes R 0.490 0.479 0.306 0.386 0.647 0.658 0.688 0.679 Panel B: Output ﬁve years changes before and after the selection Selected 0.351* 0.284** 0.298 0.508** 0.558 0.667 0.341 0.411 (1.69) (2.17) (1.52) (2.19) (1.11) (1.20) (0.84) (0.87) Control yes yes yes yes yes yes yes yes N 81 81 81 81 818181 81 Year/province yes yes yes yes yes yes yes yes R 0.528 0.470 0.326 0.392 0.728 0.731 0.619 0.626 358 Y. XINGQIANG ET AL. Table 8. Diﬀerent measures of scholarly achievement. (1) (2) (3) (4) (5) (6) Probit Tobit Probit Tobit Tobit Probit Dawadum Dawasum Dengdum Dengsum Project Didum Selected 7.727*** 2.309*** 1.378** 1.234*** 0.449 4.176*** (3.01) (4.82) (2.14) (2.89) (1.38) (3.32) Control yes yes yes yes yes yes N69 79 70 81 81 81 Year/province yes yes yes yes yes yes Pseudo R 0.581 0.464 0.74 P值 0.000 0.000 0.000 Partial regression shows some diﬀerences from the full sample due to the loss of sample size when using probability estimation. a group of high-level accounting talents with international vision, rich practical experi- ence, outstanding ability to innovate, high professional ethics. In view of this broad goal of the project, we explore additional measures of scholars’ achievements, not just the number of papers. By reviewing one by one the resumes of scholars who entered the re- examination list, we count the change in the following variables before and after the talent selection process – the probability of obtaining either the Relevant Paper Award of the Ministry of Finance or the Provincial Philosophy Social Science Award (Dawadum), the total number of honours obtained (Dawasum), the probability of publishing in a foreign SSCI journal (Dengdum), the number of publications (Dengsum), the number of projects received from the Ministry of Finance or a company (DProject), and the probability of becoming an independent director of a listed company (Didum). The basic regression results are shown in Table 8. Relative to those not selected ones, those selected have a higher probability of winning the honorary title of high-level account- ing talents – column 1 (Selected) has a coeﬃcient of 7.727 and is signiﬁcant at 1% level; the number of honours also increases signiﬁcantly as seen from the coeﬃcient of 2.309 in column 2 (Selected), which is signiﬁcant at 1% level. It lends support to the theory that the output of selected talents receives recognition from peer experts. After the selection, the probability of publishing in foreign journals also increases (column 3 (Selected) has acoeﬃcient of 1.378, signiﬁcant at 5% level), and the number of published foreign journals also rises signiﬁcantly (in column 4 (Selected) the coeﬃcient 1.234 is signiﬁcant at 1% level). The projects handed to the scholars, either by the Ministry of Finance or a company, also have increased after the selection of leading talents (column 5 has coeﬃcient is 0.449, which is not signiﬁcant), and a higher possibility of becoming an independent director of listed companies (column 6 has coeﬃcient 4.176, signiﬁcant at the 1% level). The above results support the main conclusions through diﬀerent measures. 4.2.3. Deleting consecutive samples Eight participants in the sample participated twice. In order to avoid these twice- repeated participations from interfering in the study, we report regression results after excluding these repeated participations as shown in Table 9.We ﬁnd support for our main ﬁndings. CHINA JOURNAL OF ACCOUNTING STUDIES 359 Table 9. Eliminate repeated participations. (1) (2) (3) (4) (5) (6) (7) (8) Dload Dcite Output Dnum Dtopdum Dtopsum DSumﬁrst Max Sum Max Sum Nominated 0.512** 0.189* 0.256 11.925** 0.263 0.536* 0.664* 0.679* (2.48) (1.74) (1.51) (2.02) (1.11) (1.69) (1.92) (1.73) Control yes yes yes yes yes yes yes yes N 73 73 73 73 73 737373 Year/province yes yes yes yes yes yes yes yes R 0.193 0.469 0.328 0.145 0.520 0.419 0.403 0.374 5. Additional tests The main conclusion is that the state-led accounting talent project promotes output and cultivates talent. But are the results driven by the government selecting candidates with better potential (ability hypothesis) or by the training and platform received by those selected (cultivation hypothesis)? To answer this, the paper carries out further analysis based on the characteristics of the candidates’ previous research, the graduate schools that awarded these scholars their PhD degrees, and the platform construction of the project. 5.1. Heterogeneous inﬂuence of candidates’ characteristics In order to test the impact channels, we characterise the comprehensive ability and research potential of scholars by using two measures – the number of top journals published before the scholars were nominated (Beﬀtopsum), and whether their doctoral degrees were awarded by ‘985 universities’ (Docsch). If the ability hypothesis holds, it is expected that the selected talents will do better if they have more early publications in top journals or a doctoral degree from a ‘985 university’. On the contrary, if the cultivation hypothesis holds, we do not expect to see this. The regression results are shown in Table 10 and Table 11.The results show that while selection is followed by higher output (the coeﬃcient of Selected is signiﬁcantly positive), the eﬀect is weaker for scholars who published more top-journal articles before selection (the coeﬃcient of Selected×Beﬀtopsum in Table 10 is signiﬁcantly negative). Table 10. Research capability before the selection, selection, and output change. (1) (2) (3) (4) (5) (6) (7) (8) Dload Dcite Output Dnum Dtopdum Dtopsum DSumﬁrst Max Sum Max Sum Selected 0.721*** 0.182 0.464*** 17.261** 0.518** 0.971*** 0.945** 1.116*** (3.32) (1.46) (2.74) (2.58) (2.21) (3.05) (2.63) (2.81) Beﬀtopsum −0.344 −0.985*** −1.009*** 1.291 −0.058 −0.354 0.055 0.199 (−1.22) (−5.80) (−4.79) (0.16) (−0.19) (−0.80) (0.11) (0.38) Selected×Beﬀtopsum −0.123*** 0.000 −0.109*** −2.964* −0.072 −0.190*** −0.175** −0.246** (−2.84) (0.01) (−2.74) (−1.69) (−1.36) (−2.83) (−2.32) (−2.63) Control yes yes yes yes yes yes yes yes N 81 81 8181 8181 81 81 Year/province yes yes yes yes yes yes yes yes R 0.267 0.498 0.423 0.201 0.528 0.459 0.444 0.428 360 Y. XINGQIANG ET AL. Table 11. Phd graduate school, selection, and output change. (1) (2) (3) (4) (5) (6) (7) (8) Dload Dcite Output Dnum Dtopdum Dtopsum DSumﬁrst Max Sum Max Sum Selected 0.913*** 0.358*** 0.567** 15.403** 0.680* 1.216** 0.739 1.154* (3.32) (2.70) (2.64) (2.33) (1.80) (2.56) (1.39) (1.99) Docsch −0.012 0.126 0.023 8.352 −0.006 0.188 −0.018 0.160 (−0.06) (1.10) (0.12) (1.41) (−0.03) (0.70) (−0.06) (0.47) Selected×Docsch −0.619** −0.259* −0.452** −5.414 −0.436 −0.882* −0.177 −0.733 (−2.13) (−1.86) (−2.06) (−0.79) (−1.06) (−1.72) (−0.30) (−1.13) Control yes yes yes yes yes yes yes yes N 81 81 81 818181 81 81 Year/province yes yes yes yes yes yes yes yes R 0.259 0.520 0.402 0.161 0.531 0.447 0.410 0.390 In addition, if the scholar graduated from a ‘985 university’, we expect that the coeﬃ- cient of Selected × Beﬀtopsum in Table 11 to be signiﬁcantly negative, as is indeed the case. The possible reason is that these scholars have had better training, opportunities, and platforms; so the incremental inﬂuence of the leading talents’ promotion eﬀects is rela- tively weakened. There is a stronger impact on ‘non-985 university’ doctorates, who lack some resources and access to certain platforms. The above results reveal that compared with those scholars who have strong scientiﬁc foundation and high comprehensive ability, the incremental output promotion eﬀect of leading talent projects has a greater impact on scholars with relatively weaker foundation and lower comprehensive ability. This veriﬁes that talent projects have an important role in human capital cultivation and accumulation. 5.2. Policy support and platform construction Domestic talent projects are often accompanied by provision of resources and construc- tion of platforms. There are two possible mechanisms in this paper. On the one hand the Ministry of Finance will invite senior experts from diﬀerent disciplines (both academics and practitioners) to deliver seminars. These special lectures facilitating continuous innovation by not only broadening the horizons of selected talents, but also encouraging scholars to focus on classical theoretical controversies, current research frontiers, and realistic problems. For example, the project explicitly says, ‘The forms of the training project include special lectures, seminars, forums, etc. It can help the selected talents to consolidate basic theories, upgrade knowledge structure, and expand management horizon by learning and communicating in the National Accounting Institute. It can also help the selected talents to determine the direction of the scholars’ learning and the speciﬁc tasks involved in scientiﬁc research and practice by this training and comprehen- sive test. In addition, after the leading talents are selected, the Ministry of Finance will also give special supporting funds to the talents, such as the National Accounting Leadership (Backup) Talents Academic (ﬁfth) Training Project.’ For example, the 12th joint training of the national accounting leading talent training project (at Xiamen University in October 2017) invited Ba Shu-Song (the chief economist of China Banking Association), Professor Lu Hong-De (Chung Yuan Christian University of Taiwan), Mr. Tian Feng (Alibaba), Ma Bin (Tencent Company), Professor Zhang Wei-Guo (Shanghai University of Finance and Economics), Professor Huang Shi-Zhong (Xiamen National Accounting Institute), Shi Yao-Bin (Ministry of Finance), Wu Xiao-Qing (Central Committee of the Democratic National Construction Association). CHINA JOURNAL OF ACCOUNTING STUDIES 361 Table 12. Changes in cooperation papers after being nominated among the candidates. (1) (2) (3) (4) (5) (6) Logit Probit Nbreg Poisson Tobit Cooperation Dcondum Dcondum Dcosum Dcosum Dcosum LnDcosum Selected 5.070*** 2.955*** 2.436*** 2.436*** 3.461*** 1.715*** (2.74) (3.24) (2.80) (2.80) (3.66) (3.55) Control yes yes yes yes yes yes N 60 60 818181 81 Year/province yes yes yes yes yes yes PseudoR 0.366 0.367 0.325 0.300 0.361 P 0.000 There is a loss of sample when using Logit and Probi, which is a certain diﬀerence from other sample. On the other hand, a more important mechanism may come from the opportunity and platform that talents have for communication and cooperation, such as the series of special trainings organised by the Ministry of Finance, or irregular academic forums hosted by scholars themselves. The two variables Dconsum and Dcondum measure the change in, respectively, the collaboration probability and the number of collaborative papers (Dconsum) among all participants in the re-examination list. Regression results reported in Table 12 shows that both variables increased more compared to scholars who were not allowed to participate in the re-examination. The above results consistently show that the leading talent project provides an excellent communication platform, and has contributed to collaborative output 6. Conclusions and policy implications Nowadays, with the rapid development of science and technology, the emergence of a knowledge economy and increasingly ﬁerce competition for talents, the retention of talents has become the key to sustainable competitiveness of countries and regions. Eﬃcient allocation of accounting talents, a key high-level human capital, plays an important role in sustaining stable growth and development of the ﬁeld of accounting and econom- ics. In addition, the selection and training process of talent projects leads to the production and inheritance of knowledge, which is the only source increasing returns to scale in the economy. Therefore, the eﬃciency of knowledge production and human capital allocation is not only related to the professional development of talents themselves, but also directly aﬀects the strength and the development of a nation. Therefore, exploring the eﬀectiveness of talent selection projects is of great theoretical value and practical signiﬁcance. Based on the diﬀerence between the two periodicals of the Ministry of Finance’s re- examination list and the ﬁnal selection list, we use the basic idea of Diﬀerence-in- Diﬀerence (DID) model to test this. We ﬁnd that Chinese government selection projects have signiﬁcantly positive eﬀect on promoting output and cultivating talent. Speciﬁcally, compared with the talents who are not selected in the project, (1) the selected talents publish more papers after being selected, (2) among these papers published in Chinese core journals, the selected talents publish more papers as the ﬁrst author, (3) the prob- ability that these papers are published in Chinese top journals by the selected scholars is For example, take the National Accounting Leading Talents Lecture Series at Shihezi University http://sem.shzu.edu.cn/ 2016/0918/c975a83481/page.htm. 362 Y. XINGQIANG ET AL. 18.2% higher, and (4) these articles have larger incremental downloads and citation rates. Furthermore, we also ﬁnd that the talent project is more pronounced for scholars who published fewer papers in top Chinese journals in the past and for scholars who were awarded doctoral degrees by ‘non-985’ universities. In addition, the number of papers on which the talents collaborated also increased. The possible policy prescription of this paper is that, although the national selection project has a speciﬁc role in talent cultivation as a whole, it needs to pay special attention to the openness, fairness, and reasonableness of the selection process. By improving the selection process, problems arising from diverse sources such as information asymmetry, moral hazard, and personal characteristics can help select the very best. On this basis, we can devise further policies to support the selected scholars, especially those scholars who start with a disadvantage and consequently oﬀer high returns to human capital investment. Our paper tries to ﬁgure out the reason for the increasing output of the leadership talent project –‘the project screened better persons’ versus ‘the project enriched talents’. However, in the absence of detailed data on total scores of re-examinees, we compare the diﬀerences before selection between selected and unselected scholars following previous literature. This leaves room for the possibility that the project selected those with stronger scientiﬁc research potential. It means that further research should consider more detailed re-test scores or ranking data to open the black box about the mechanism of talents projects, and separate the two possible explanations. Acknowledgments We are grateful for numerous helpful comments from two anonymous reviewers. We acknowledge ﬁnancial support from the Postgraduate Innovation Project of Shanghai University of Finance and Economics (projects CXJJ-2018-305 and CXJJ-2018-306). Disclosure statement No potential conﬂict of interest was reported by the authors. ORCID Hu Ning http://orcid.org/0000-0001-9252-1070 References Adams, J., & Griliches, Z. (1998). Research productivity in a system of universities. Annals of INSEE, 49, 127–162. Arora, A., & Gambardella, A. (2005). The impact of NSF support for basic research in economics. Annales d’ Economie et de Statistique, 79–80,91–117. (in Chinese). Azoulay, P., Zivin, J.S.G., & Manso, G. (2011). Incentives and creativity: Evidence from the academic life sciences. The RAND Journal of Economics, 42(3), 527–554. Becker, G.S., & Barro, R.J. (1988). A reformulation of the economic theory of fertility. Quarterly Journal of Economics, 103(1), 1–25. Benhabib, J., & Spiegel, M.M. (1994). The role of human capital in economic development evidence from aggregate cross-country data. Journal of Monetary Economics, 34(2), 143–173. CHINA JOURNAL OF ACCOUNTING STUDIES 363 Bielby, W.T., & Bielby, D.D. (1999). Organizational mediation of project-based labor markets: Talent agencies and the careers of screenwriters. American Sociological Review, 64(1), 64–85. Burt, R. (2001). Structural holes versus network closure as social capital, social capital: Theory and research. New York, NY: Aldine de Gruyter. Chen, X. (2011). Tolerance and economic performance in American metropolitan sreas: An empirical investigation. Sociological Forum, 26(1), 71–97. Coleman, J.S. (1988). Social capital in the creation of human capital. American Journal of Sociology, 94,95–120. Durante, R., Labartino, G., & Perotti, R. (2011). Academic dynasties: Decentralization and familism in Italian academia. Working Paper. Retreived from https://www.nber.org/papers/w17672 Fisman, R., Shi, J., Wang, Y., & Xu, R. (2017). Social ties and favoritism in Chinese science. Journal of Political Economy, 126(3), 1134–1171. Granovetter, M. (2005). The impact of social structure on economic outcomes. Journal of Economic Perspectives, 19(1), 33–50. Guo,F., & Li,X.(2017). Editorial preference, relationship draft, and citation rate discounts: Evidence from China’s authoritative journals in economics. China Economic Quarterly, 4,1237–1260. (in Chinese). Harrison, P.D., & Harrell, A. (1993). Impact of “adverse selection” on managers’ project evaluation decisions. Academy of Management Journal, 36, 635–643. Iaria, A., Schwarz, C., & Waldinger, F. (2018). Frontier knowledge and scientiﬁc production: Evidence from the collapse of international science. Quarterly Journal of Economics, 133(2), 927–991. Islam, N. (1995). Growth empirics: A panel data approach. Quarterly Journal of Economics, 110(4), 1127–1170. Jacob, B.A., & Lefgren, L. (2011). The impact of research grant funding on scientiﬁc productivity. Journal of Public Economics, 95(9–10), 1168–1177. Jaﬀe, A.B. (1989). Real eﬀects of academic research. American Economic Review, 79, 957–970. Jaﬀe, A.B. (2002). Building a program evaluation into the design of public research support programs. Oxford Review of Economic Policy, 18(1), 22–34. Jaﬀe, A.B., Trajtenberg, M., & Henderson, R. (1993). Geographic localization of knowledge spillovers as evidenced by patent citations. Quarterly Journal of Economics, 108(3), 577–598. King, M.A., & Robson, M.H. (1993). A dynamic model of investment and endogenous growth. The Scandinavian Journal of Economics, 95(4), 445–466. Linder, S.H., & Peters, B.G. (1987). A design perspective on policy implementation: The fallacies of misplaced prescription. Review of Policy Research, 6(3), 459–475. Liu, R., & Zhao, R. (2017). Does the anonymous review system promote China’s economic progress?: Based on the method of diﬀerence-in-diﬀerence. China Economic Quarterly, 1,173–204. (in Chinese). Ouyang, Z., Ye, X., Cui, H., Zhang, G., & Qiu, T. (2015). The eﬀectiveness and optimization research on the national accounting leadership or backup talent training: Based on the Analysis of the tenth-anniversary questionnaire. Accounting Research, 12,3–16. (in Chinese). Payne, A., & Siow, A. (2003). Does federal research funding increase university research output? Advances in Economics and Policy, 3, 1538–1637. Romer, P.M. (1987). Growth based on increasing returns due to specialization. American Economic Review, 77,56–62. Seifert, M., & Hadida, A.L. (2006). Facilitating talent selection decisions in the music industry. Management Decision, 44(6), 790–808. Tian, Z., Sun, J., & Lu, W. (2015). Review and prospect about the frontiers of accounting research: Analysis of the research results based on academic accounting leading talents. Accounting Research,(11), 11–19. (in Chinese). Yang, J., Gong, L., & Zhang, Q. (2006). Human capital formation and its impact on economic growth: An endogenous growth model with education and health inputs and its testing. Management World, 5,10–18. (in Chinese). Zinovyeva, N., & Bagues, M. (2015). The role of connections in academic promotions. American Economic Journal: Applied Economics, 7, 264–292.
China Journal of Accounting Studies
– Taylor & Francis
Published: Jul 3, 2019
Keywords: Talent project evaluation; Chinese academic accounting leading talents; Difference in difference