Study of the Impact of Expected Credit Loss Model on the Quality of Accounting Information

. In 2017, the Ministry of Finance issued a new standard on financial instruments to achieve convergence with international accounting standards, in which the emergence of expected credit loss model has great significance and far-reaching impact on the development of enterprises in China. In this paper, the impact of expected credit loss model on the quality of corporate accounting information is studied by using Differences-in-Differences method. Using accounting conservatism as a proxy variable for accounting information quality, this paper finds that the expected credit loss model can improve the accounting conservatism of enterprises. The findings of this paper show that the expected credit loss model is effective in improving the quality of accounting information, thus justifying the feasibility and necessity of implementing the new financial instruments standard and providing a good empirical basis for deepening the reform of accounting standards in China in the future.


Introduction
Accounting standards are the standard for conducting accounting work and the basis for developing accounting system. The root of in-depth accounting practice lays a sound foundation for our economy to go global and plays an important role in maintaining the macroeconomic stability of our country. The world financial crisis in 2008 had a significant impact on the world economy and financial markets. After the crisis, many scholars believe that the complexity of the old financial instruments standard and the pro-cyclical effect of the incurred loss model are the culprits of this crisis, thus objectively accelerating the re-revision of the financial instruments standard. In July 2014, the International Accounting Standards Board (IASB) put the International Accounting Standard No.39-Financial Instruments (IAS 39) to replace International Financial Reporting Standard No. . In order to effectively address the practical issues faced by China The core changes of the new financial instruments standard include: First, the classification of financial assets is changed from "four classifications" to "three classifications": measured at amortized cost (AC), measured at fair value through other comprehensive income (FVOCI), measured at fair value through temporary profit or loss (FVTPL). This reclassification reduces the possibility of profit reconciliation and effectively curbs management's self-interest. Secondly, the impairment model of financial assets is changed from Incurred Loss Model (ILM) to Expected Credit Losses (ECL). The incurred loss model can only be impaired when there are objective conditions indicating that there are losses, which leads to the fact that enterprises do not take impairment in a good economic environment. But once the economic situation deteriorates, they will immediately collapse, because the impairment is not enough and there is a time lag in the accrual, and its "procyclicality" will promote the occurrence of economic crisis. The expected credit loss model takes into account the whole life cycle of financial assets to provide for impairment, which is more timely, broader in scope and more adequate than the incurred loss model, and has a certain degree of foresight,

FEMS 2022
Volume 19 (2022) 40 which is conducive to the effective avoidance of financial risks (Yuan Feng, 2021). Third, relevant provisions on hedge accounting are amended so as to better reflect the operation and management activities of companies. The emphasis of this paper is placed on the second aspect, which is the transformation of impairment models.
Some studies have found that expected credit loss models will have an impact on the quality of corporate accounting information. Some scholars argue that expected credit loss models can improve the timeliness, prudence (Huang Wenjie, 2021) and relevance of accounting information (Xiao Xu and Zeng Ting, 2015). Other scholars argue that expected credit loss models can reduce the reliability, comparability (Deng Yongqin et al., 2017), and truthfulness of accounting information (Huang Shizhong, 2015). In summary, this paper is motivated by the fact that the conclusions reached by the present studies are inconsistent, and there is a lack of systematic research and empirical tests on whether the expected credit loss model will ultimately have a positive or negative effect on the quality of accounting information. At present, there is a lack of systematic research and empirical tests.
This paper examines and tests whether the expected credit loss model can positively affect the quality of accounting information through the lens of accounting information quality based on the DID model by using the exogenous event of the implementation of the new financial instruments standard. The study finds that the expected credit loss model can improve the quality of accounting information of financial enterprise.
The contributions of this paper are mainly in three aspects: First, based on the institutional background and relevant data of China's new financial instruments standards, this paper provides an in-depth analysis of the significant impact of the change from the incurred loss model to the expected credit loss model, and investigates the impact of the expected credit loss model on the quality of accounting information in this context, which enriches the research literature related to the change of accounting standards. Second, this paper adopts the CScore model and uses accounting conservatism as a proxy variable for accounting information quality, and expands the research on the factors affecting accounting information quality of listed companies from the perspective of expected credit loss model. Third, this paper has certain practical value and inspirational significance. The new financial instruments standard is an important measure to prevent the financial crisis from coming again, and the change from the incurred loss model to the expected credit loss model is one of the most crucial aspects. A more adequate and timely provision for impairment is conducive to improving the quality of accounting information of financial enterprises and better maintaining the orderly and healthy development of the financial market.

Theoretical Analysis and Research Hypothesis
Accounting information is an important basis for enterprises to make decisions, and the quality of accounting information is a prerequisite for effective decision-making. High-quality accounting information can ensure the healthy development of the capital market, while low-quality accounting information can adversely affect the normal operation of the capital market. Therefore, the quality of accounting information has always been an important academic topic, which occupies a very critical position in both the practice and theory of accounting. Chinese scholars have also conducted many studies on accounting information quality and found that corporate governance (Liu Liguo  The incurred loss model was used under the old accounting standards, which required that impairment could only be recognized for assets with objective evidence of an actual loss. No impairment could be made even if there is an indication that a future loss is probable (Xu Aifei, 2018).

FEMS 2022
Volume 19 (2022) 41 Such an impairment model suffers from the disadvantages of not being timely, lacking foresight, and failing to alert investors and regulators to the approaching risk (Zhang Shanshan, 2015), which coupled with the accelerating effects of "pro-cyclicality" and "cliff effect", will likely financial crises and disruptions in financial markets. Instead, the expected credit loss model in the new financial instruments standard does not require the existence of objective evidence of impairment. Companies should recognize expected credit losses for the entire life of the financial instrument as long as there is a significant increase in the credit risk of the financial instrument from the date of initial recognition. If there is no significant increase in the credit risk of the financial instrument from the date of initial recognition, the expected credit losses should also be recognized for 12 months' loss. Compared with the incurred loss model, the expected credit loss model is an impairment model that changes from a fact-based to a forecast-based model (Wang Jingjing and Liu Guangzhong, 2014), which is less complex, more operational. It allows for faster and more impairment provisions, includes expected losses in profit or loss, reduces the impact of actual future losses, and effectively attenuates the "procyclicality" while also preventing the "cliff effect", thus increasing the relevance of accounting information (Xu Aifei, 2018). Although the expected credit loss model is forward-looking, historical accounting information is still a very important key point as a basis to adjust the current data so that the impairment charged under the expected credit loss model better reflects the current situation and future projections of the company (Jiang Lei and Liang Feiyuan, 2015). This shows that there is a very close relationship between the expected credit loss model and the quality of accounting information, which also proves the feasibility of this research. Expected credit loss model may improve the quality of accounting information. Firstly, Liu Xiaohua and Wang Hua (2015) found that international convergence of China's accounting standards can improve the quality of accounting information by using the Differences-in-Differences method through model building test analysis, and the beginning of China's implementation of the new financial instrument standards and the use of expected credit loss model is the key to achieve international convergence, so the quality of accounting information of Chinese companies is also improved. Secondly, the conversion of incurred loss model to expected credit loss model reduces the objectivity of accounting information to a certain extent, however, if companies can reasonably use the expected credit loss model and accurately predict the relevant indicators, it can reduce the " procyclicality " and thus enhance the truthfulness and usefulness of accounting information (Ke Jian, 2019). After the implementation of the new financial instruments standards, the use of the expected credit loss model will require more impairment provisions for financial assets, which will improve the prudence and timeliness of accounting information. For investors and other information users, the accounting information under the expected credit loss model can undoubtedly reflect the value changes of financial assets more accurately, so that they can obtain more reliable accounting information to make relevant decisions (Huang Wenjie, 2021). In addition, Mushkudiani Z and Shonia N (2016) find that the expected credit loss model achieves a reasonable compromise between providing relevant accounting information to information seekers and improving financial stability, effectively mitigating the amplifying effect of the incurred loss model on "pro-cyclicality" and enhances financial stability.
The expected credit loss model may also reduce the quality of accounting information. First, the provision for impairment does not require objective and sufficient evidence because the expected credit loss model removes the prerequisite of the actual occurrence of the loss event as the recognition of impairment, which increases the difficulty and cost of auditing and invariably leaves room for the management of the audited entity to adjust the surplus. Therefore, from the perspective of auditing, the quality of accounting statements under the expected credit loss model is reduced (Huang Shizhong, 2015).Secondly, the expected credit loss model is unclear in defining the "significant increase" in credit risk in the three stages of impairment, which requires a lot of professional subjective judgment and is difficult, leading to the lack of comparability of accounting information among different companies and further the decreasing in the quality of accounting information (Deng Yongqin et al. 2017). Xiao Xu and Zeng Ting (2015) pointed out that the expected credit loss model reduces reliability and increases relevance, so the relationship between reliability and relevance of accounting information must be reconciled.
In summary, since the new financial instruments standard has only been implemented for a short period, most of the academic research on the quality of corporate accounting information before and after the use of expected credit loss model is still at the stage of theoretical research. The views are not uniform, and there is a lack of empirical research on this aspect. The impacts the expected credit loss model on the quality of accounting information of listed companies needs further empirical research and testing. Therefore, this paper proposes the following two opposite hypotheses: H1a: The expected credit loss model can improve the quality of accounting information of a company when other conditions are not changed.
H1b: The expected credit loss model will reduce the quality of accounting information of a company when other conditions are not changed.

Study Sample
In view of the fact that the new financial instruments standard is effective from 2018 for A+Hshare companies and from 2019 for A-share listed companies, and that the financial industry is the most affected industry among all industries, this paper selects all A-share and A+H-share companies in the financial industry from 2016 to 2018 as the sample, and preprocesses the initial sample as follows: (1) The samples of A+H-share companies in the financial sector that take the lead in implementing the new standards for financial instruments in 2018 shall be the treatment group, and A-share companies in the financial sector that will implement the new standards for financial instruments after 2019 shall be the control group. (2) In order to make the sample size corresponding to each variable in the descriptive statistics consistent and thus the results more objective, 34 samples are excluded from this paper. (3) On this basis, the samples with missing data are eliminated to obtain a total of 197 "company-year" samples. The financial research data in this paper are obtained from CSMAR database, and the data processing software is Stata16.0.

Research Model
This paper constructs a differences-in-differences (DID) based on this quasi-natural experiment implemented by the new financial instruments standard, citing Liu Yaoyao et al. (2021), which is used to test the impact of the expected credit loss model on the quality of accounting information.
where: the explanatory variable is the quality of accounting information, measured by accounting conservatism (CScore). Treat is an event variable used to define A+H-share companies in the financial sector that are the first to use the expected credit loss model in 2018 (Treat=1) and A-share companies that have not yet used the expected credit loss model (Treat=0). Post is a time variable used to distinguish between after the expected credit loss model is used (Post=1) and before it is used (Post=0). DID is the cross product of Treat and Post, that is, Treat * Post. The fixed effect of year is controlled in model (1). Since Post is highly correlated with Year, only Treat and DID are included in the model in this paper, while Post is not included. In the conservatism test below, this paper controls for firm fixed effects and the test results do not change significantly. The regression coefficient of the cross product term DID is the focus of this paper. If it is significantly positive, it indicates that the expected credit loss model can significantly improve the quality of accounting information of financial firms and hypothesis H1a is verified, while if it is significantly negative, hypothesis H1b holds.

3.3.1Expected credit loss model
(1) The expected credit loss model is determined for the experimental and control groups, generating the event variable (Treat). Financial industry A+H-share companies that used the expected credit loss model during the study period are the experimental group (Treat=1), otherwise they are the control group (Treat=0). The specific distribution of the event variable (Treat) is shown in Table  1 The cross product term of event variable (Treat) and the time variable (Post) is the core explanatory variable that needs to be focused on in this paper.

Quality of accounting information
Accounting conservatism is an important principle in financial accounting, and accounting conservatism is closely related to the quality of accounting information, which directly affects the profit of enterprises. Therefore, this paper uses accounting conservatism as a measure of accounting information quality, and the higher the accounting conservatism of financial enterprises, the better the quality of accounting information. According to Article 18 of China's Company Accounting Standards-Basic Standards, "Conservatism means that an enterprise should exercise due care in accounting recognition, measurement and reporting of transactions or events, and should not overestimate assets or earnings or underestimate liabilities or expenses." According to Basu (1997), accounting conservatism means that the recognition of "bad news" in accounting surplus should be timelier and more adequate than the recognition of "good news", i.e., liabilities should not be underestimated and assets should not be overestimated, so as to improve the quality of accounting information.
In this paper, by citing the CScore model of Khan and Watts (2009), the measure of accounting conservatism is as follows. (3) and (4) into model (2) respectively, model (5) has been generated.
In the above equation, EPS is earnings per share. P is the stock price at the beginning of year t (using the price of the stock on the last trading day in April of year t). R is the stock return (12-month buy-and-hold return from May of year t to April of the following year, calculated based on the monthly individual stock return considering reinvestment of cash dividends). D is a dummy variable BCP Business & Management

FEMS 2022
Volume 19 (2022) 44 that takes the value of 1 if R < 0 and 0 vice versa. SIZE is the total assets and MB is the market valueto-book ratio, i.e., market value/total assets. LEV is debt to asset ratio, i.e., the value of total liabilities/total assets. Gscore (β0) is the timeliness of reflecting "good news" at the firm level. Cscore (β1) is the timeliness of reflecting "bad news" at the firm level, which is an indicator of accounting conservatism. β0+β1 is the incremental timeliness of reflecting "bad news" over "good news" in accounting surplus.
In this paper, we first bring model (3) and model (4) into model (2) to obtain model (5), and then regress model (5) by year and industry, and bring back the obtained regression coefficients γ1~γ4 with model (4) to calculate the accounting conservatism (CScore) of company i in year t+1. The larger the value of CScore, the higher the accounting conservatism of the company and the higher the quality of accounting information.
Control variables. With reference to the previous studies related to accounting conservatism, the following control variables are selected in this paper, including company size (Size), debt to asset ratio (LEV), return on net assets (ROA), internal control (Icindex), and number of people tracked by analysts (Analyst). The main variables in this paper are defined in Table 2.

Descriptive Statistics
The descriptive statistics of the main variables in this paper are presented in Table 3. The mean value of the proxy variable for the quality of accounting information (CScore) is -0.754 and the standard deviation is 3.212, indicating that the accounting conservatism varies widely across financial firms. The mean value of Treat is 0.122, which indicates that 12.

Empirical Results
The regression results of the effect of expected credit loss model on accounting information are presented in Table 4. Column (1) shows the regression results before the inclusion of control variables. The coefficient of DID is 2.163 (t=3.00) and is significantly positive at the 1% level. Column (2) shows the regression results after the inclusion of control variables. The coefficient of DID is 1.641 (t=2.57) and is significantly positive at the 5% level. From the above results, it is shown that the expected credit loss model is significantly and positively related to the proxy variable accounting conservatism (CScore) of accounting information quality, that is the expected credit loss model can improve the accounting information quality of financial firms, and the hypothesis H1a is thus verified.

Conservatism Test
In the above research design section of this paper, model (1) controls for annual fixed effects. Therefore, the conservatism is tested by controlling for firm fixed effects in model (1) by regressing again, and the results are shown in Table 5 below. From the regression results in Table 5, the coefficients of DID without and with the inclusion of control variables are still significantly positive at the 1% and 5% levels respectively, indicating that the expected credit loss model's effect on the improvement of accounting information quality still exists, thus providing further conservatism evidence for the previous empirical analysis.

Research Conclusions
This paper constructs a DID model using a sample of listed companies in the financial industry from 2016-2018 to test the impact of expected credit loss model on accounting information quality. It finds that the accounting conservatism of A+H-share listed companies that are the first to use the expected credit loss model significantly improves compared to A-share listed companies that do not use the expected credit loss mode. This paper uses accounting conservatism as a proxy variable for accounting information quality, thus demonstrating the significant effect of using the expected credit loss model on improving accounting information quality. The findings of this study confirm the effectiveness of the expected credit loss model on the improvement of accounting information quality, and also affirm the feasibility and necessity of the implementation of the new financial instruments' standard, which provides a good empirical basis for deepening the reform of accounting standards in China in the future.

Research Implications
The expected credit loss model can improve the quality of accounting information and promote the healthy development of China's financial market. In the future, the relevant financial sector in China should further improve the expected credit loss model under the premise of continuous convergence with international accounting standards, so as to seek common ground while reserving BCP Business & Management

FEMS 2022
Volume 19 (2022) 47 differences and make it more adaptable to the development of China's economic market and society, thus further improving the quality of accounting information of enterprises. In the context of the new financial instruments standards, enterprises should also focus on improving the quality of their own accounting information, enhancing their own business level, business ability and professional competence to have better internal control and risk management, and avoid major mistakes caused by subjective judgment in forecasting.

Research Gaps and Outlook
This study has certain limitations that need to be improved and refined in subsequent in-depth studies. Although this paper confirms the positive effect of expected credit loss model on accounting information quality and makes a reasonable theoretical elaboration, it does not provide a more direct test of accounting information quality itself. In the future, we need to further improve the research design and conduct further systematic research on its impact.