The Impact of International Financial Reporting Standards (IFRS) on Audit Quality Measurement: The Case of Selected Commercial Banks of Ethiopia (original) (raw)

The Impact of International Financial Reporting Standards (IFRS) on Audit Quality Measurement: The Case of Selected Commercial Banks of Ethiopia

Tilahun Minwuyelet

Department of Accounting, Addis Ababa University, Addis Ababa, Ethiopia

***Abstract***

***Prior researches have investigated the impact of mandatory International Financial Reporting Standards (IFRS) on various reporting, accounting and auditing areas. These studies recognized, in general, an increase in audit quality due to the adoption of IFRS. These studies mostly focus on Europe, while the effect of the Sarbanes-Oxley is often discussed in the United States. The effect in Ethiopia is less often a subject and this project will shed some light on the effects of mandatory adoption of IFRS in Ethiopia, while focusing on the effect on audit quality.***

***The research consists of companies registered on the Ethiopian Banking and Financial market between 1990 and 2022. This study applies two models to evaluate audit quality in terms of discretionary accruals and discretionary expenses. The first model is the “traditional” Modified Jones Model where the results are not showing a decreased level of discretionary accruals after the adoption of IFRS. The second model is the more recently developed model by Roychowdhury, the Real Earnings Management Model. These results also indicate that there is significant association between discretionary expenses and the adoption of IFRS. This finding is in line with findings of previous researches performed in the European region (EU) and other regions.***

***Additional analysis is required to elaborate further on the association of audit quality and the adoption of IFRS. This research has included various variables such as EBIT (earnings before interest and taxes) and ROA (return on assets), other variables (such as auditor size) were excluded due to the unavailability of data. Including these variables in future researches may add value to the results and analysis. Also, this research is conducted using the Modified Jones Model and Roychowdhury’s Real Earnings Management Model to calculate respectively discretionary accruals and discretionary expenses. Other models may shed additional light on the relation between audit quality and the adoption of IFRS.***

***Keywords: IFRS Modified Jones Model, Real Earnings Management & Discretionary Accruals***.

Introduction

The adoption of International Financial Reporting Standards (IFRS) is considered to be one of the most impacting regulatory changes in the history of financial reporting. Enhancement of the comparability of financial statements, improvement of corporate transparency and quality of financial reporting in general were the main goals of IFRS. This project examines if these goals, in particular strengthening audit quality, have been achieved.

The adoption of IFRS in Europe occurred in 2005. To this end, European Union (EU) was one of the initial adopters of IFRS and block grant donator of Ethiopia. As a consequence of the adoption of IFRS, audit frameworks have been affected which could have resulted in changes in discretionary accruals and therefore audit quality. In order to investigate the results of the adoption of IFRS in Europe, this study focuses on the development of discretionary accruals surrounding the years prior to the adoption of IFRS and after the adoption of IFRS. Furthermore, the study is conducted while focusing on Banking sectors registered in Ethiopia.

The effect of the adoption of IFRS has had a worldwide impact on the reporting framework and accounting standard. Investigating unexpected side-effects such as a change in audit quality may provide new insights regarding the total adoption costs and adoption effects. Also, increasing the audit quality has been one of the main objectives of the adoption of IFRS. This study contributes to the question if this objective has been achieved. In addition, sectors such as the banking and finance, insurances and others in Ethiopia have not overwhelmingly adopted IFRS and might consider adopting IFRS in the near future in spite of the good beginnings. This study gives an impression of the effects that occur after adopting IFRS from a domestic reporting standard. Therefore, this study adds value to future considerations regarding adopting IFRS and helps anticipating the effects of its adoption.

Literature Review

International Financial Reporting Standards (IFRS) set common rules so that financial statements can be consistent, transparent, and comparable around the world. IFRS are issued by the International Accounting Standards Board (IASB) formerly known as International Accounting Standards Committee (IASC). IFRS specify how companies must maintain and report their accounts, defining types of transactions, and other events with financial impact. IFRS were established to create a common accounting language so that businesses and their financial statements can be consistent and reliable from company to company and country to country. IFRS are sometimes confused with International Accounting Standards (IAS), which are the older standards that IFRS replaced (Barclay palmer 2021).

Obviously, there are major differences in financial reporting of companies in different countries. These differences result in complications for preparing, consolidating, auditing and interpreting published financial statements. There has been a greater need to bridge the gap between the differences in financial reporting standards among countries. To make this a reality, several organizations have been involved in trying to harmonize the financial reporting standards worldwide. The terms harmonization and standardization are used in most instances to describe the solution to solving the differences that pertain in national financial reporting standards. Harmonization is the process of increasing the compatibility of accounting practices by setting bounds to their degree of variation.

More than 144 countries around the world have adopted IFRS, which aims to establish a common global language for company accounting affairs. But all countries are not adopting IFRS and use the old one GAAP (general accepted accounting principle). A major difference between GAAP and IFRS is that GAAP is rule-based, whereas IFRS is principle-based. This disconnects manifests itself in specific details and interpretations. Basically, IFRS guidelines provide much less overall detail than GAAP. Consequently, the theoretical framework and principles of the IFRS leave more room for interpretation and may often require lengthy disclosures on financial statements. On the other hand, the consistent and intuitive principles of IFRS are more logically sound and may possibly better represent the economics of business transactions.

Methodology

This project will investigate whether earnings management (measured by audit quality) has increased or decreased due to the adoption of IFRS in Ethiopia especially in selected commercial banks. Earnings management will be measured in terms of audit quality where the discretionary accruals will be used as the dependent variable. As discussed in chapter 1 and in the literature review of this chapter, discretionary accruals have often been used to recognize earnings management in various empirical studies. If there is an increase in discretionary accruals, the statement can be made that the possibilities regarding earnings management have increased which means that the audit quality has decreased Discretionary accruals Possibilities to perform earnings management Audit quality, /Therefore, the main research question is described as: Research question 1: Is the adoption of IFRS in Ethiopia associated with an improve in audit quality for some commercial banks?

This result in the first hypothesis which can be stated as: H1: The audit quality, measured in discretionary accruals, has significant effect (positively) with the adoption of IFRS.

As discussed in chapter 1, the variable EBIT (Earnings before Interest and Taxes) has also been subjected to researches where findings were documented stating that EBIT has an influence on discretionary accruals. Therefore, in addition to the above research question, the following research question can be stated: research question 2: Is the level of EBIT affecting audit quality for selected commercial banks companies after the adoption of IFRS? Therefore, the second hypothesis is stated as: H2: A higher EBIT, compared to the year before, is associated to an increase of discretionary accruals in the years prior to the adoption of the IFRS compared to the period after the adoption IFRS?

Also discussed in chapter 1, the variable ROA (Return on Assets) has also been subjected to researches where findings were documented stating that ROA has an influence on discretionary accruals. Therefore, in addition to the above research question, the following research question can be stated, research question 3: Is the level of ROA affecting audit quality for selected commercial Ethiopian banks after the adoption of IFRS? In order to determine if ROA is outperforming the adoption of IFRS in terms of predicting audit quality, the second hypothesis is stated as:H3: A higher ROA has a higher impact on audit quality in the years prior to the adoption of the IFRS compared to the period during the IFRS

**3.1 Model Specification**

Looking at the different models discussed in chapter 1 that can measure discretional accruals, the following possibilities exist in terms of utilizing discretionary accruals:

The DeAngelo Model
The Healy Model
The Jones Model
The Modified Jones Model
Roychowdhury’s Real Earnings Management Model

3.1.1 The Modified Jones Model

As discussed in chapter 1, the modified Jones model is adjusted for the problems experienced in the regular Jones model such as possibilities to manipulate revenue. The modified Jones model minimizes this threat by using the net-revenue as a proxy for non-discretionary accruals. Also, as noted by Arezoo Aghaei Chadegani^{16} most studies related to discretionary accruals use this model for determining audit quality (and thus earnings management). This project will use the Modified Jones Model (Dechow, 1995) because most studies regarding discretionary accruals use this model to perform analysis. Additionally, two studies (3.4.3 and 3.4.4) discussed in the literature review with a similar focus have used different models (respectively Modified Jones Kothari model and properties of earnings to identify thresholds). It would be interesting to see how the results differ when using the modified Jones model.

Another reasons to choose Dechow over Kothari, is the ROA element which is included in the Kothari model. This project will use the performance measure ROA in a separate regression which makes in necessary to exclude ROA in the calculation to measure discretionary accruals. Paragraph 5.5 will elaborate further on this.

In the modified Jones model, the discretionary accruals are calculated by looking at the non-discretionary part of the total accruals:

= Change in revenue in year *t*

= Change in net receivables in year *t* minus year *t* – *1*

= Total value of property, plant and equipment as of December 31th of year *t*

*=* Total assets in year t

After calculating the non-discretionary part of the total accruals, the discretionary part of the total accruals in a certain year can easily be calculated by deducting the NDA (non-discretionary accruals) from the total accruals.

1^{6} Chadegani, A. (2011). Review of studies on audit quality.

Furthermore, the total accruals will be calculated by looking at the change in current assets, change in cash balance, and change in current liability and depreciation and amortization expenses:

= Total accruals for company in time period lagged for total assets

= Change in current assets for firm in time period to

= Change in cash balance for firm in time period to

= Change in current liabilities for firm in time period to

Change in short term debt for firm in time period to

= depreciation and amortization expenses for firm in time period to

Audit related variables such as audit fee and auditor size are not included in this research. In chapter one, an explanation is being given why the added value of adding the variable auditor size is negligible. Various studies, in particular Choi (2010)^{17}, have already empirically tested if audit fee is audit quality or discretionary accruals or earnings management.

Also, information regarding audit clients for Ethiopian Commercial Banks that are registered on the National Bank of Ethiopia is unavailable. However, this could be considered for future studies.

3.1.2 Roychowdhury’s Real Earnings Management Model

While the precious models use discretionary accruals to measure audit quality, the Real Earnings Management Model by Roychowdhury^{18} (2006) uses a different approach by looking at the discretionary expenses. This model tries to identify real activities manipulation that is being applied in order to avoid losses.

= The discretionary expenses in year *t*

*=* Total assets in year *t*

*St =* Total sales in year *t*

Analysis

This section describes the data, provides the descriptive results, regressions and correlation analyses results

**4.1** **Descriptive Statistics**

*Table.4.2: Summary Statistics of the Earnings Management Proxies*

| | count | mean | sd | median | min | max | skewness | kurtosis |
| Scaled TAcc | 3,717907291 | 0.476 | 0.427 | 2.572 | -9.756 | 14.902 | 3.109 | 20.679 |
| NDacc | (40,153,373,593) | 1.077 | 2.911 | -4.268 | -8.617 | 79.362 | 6.272 | 46.257 |
| Dacc | 43,871,280,884 | 1.554 | 0.929 | 2.573 | -9.573 | 14.901 | 0.069 | 10.068 |
| DiscrExpenses | 1,550,000,000 | 0.096 | 0.505 | 0.095 | 0.080 | 0.110 | 4.714 | 30.684 |
| abnDiscrExp | (1,093,000,000) | 0.068 | 0.209 | -0.130 | -0.110 | -0.150 | 2.369 | 18.100 |
| *N* | 2 | | | | | | | |
| | count | mean | sd | median | min | max | skewness | kurtosis |
| DeltaEBIT | 2,858,843,853 | 3.637 | 40.132 | 0.261 | 38.847 | 509.034 | 4.011 | 24.109 |
| ROA | 400,000,000 | 0.024 | 0.784 | 0.005 | 0.02 | 0.03 | 4.108 | 25.454 |
| IFRS | 2 | 0.042 | 0.412 | 0.075 | -0.352 | 0.1843 | 1.522 | 19.350 |
| Cash | 92,546,970,781000 | 1.063 | 10.980 | 12.939 | 3.366 | 19.146 | 27.401 | 13.660 |
| Current Assets | 362,280,866,038 | 4.161 | 52.767 | 23.313 | 8.217 | 30.409 | 17.821 | 48.083 |
| Current Liabilities | 60,941,604,000 | 3.385 | 35.761 | 0.281 | 8.217 | 30.409 | 19.211 | 50.323 |
| Debt in Current Liabilities | 203,682,000 | .003 | 60.379 | 0.040 | 0.000 | 2.036 | 25.480 | 91.821 |
| Depreciation Amortization | 650,427,000 | 0.123 | 65.340 | 0.215 | 0.104 | 0.325 | 18.609 | 41.890 |
| Total Assets | 621,038,400,227 | 120.540 | 262.334 | 123.606 | 28.892 | 621.038 | 110.232 | 160.767 |
| Revenue | 62,245,563,754 | 3.458 | 32.409 | 4.686 | 1.333 | 8.038 | 14.942 | 28.797 |
| Receivables | 98,121,583,000 | 5.451 | 46.980 | 26.804 | 4.547 | 49.060 | 15.344 | 36.066 |
| PPE | 23,056,943,029 | 0.512 | 11.210 | 2.256 | 0.221 | 4.256 | 20.086 | 25.208 |
| *N* | 2 | | | | | | | |

**Table 4-3: Summary Statistics of the Other Research Variables.**

The descriptive statistics on data obtained from financial statements prepared under IFRS and GAAP for the same firm year. The scaled total accrual is having a mean value of 0.476. This indicates that, on average, the company size (measured current assets) has increased by 47.6% over the years in the sample size. The non-discretionary accruals and discretionary accruals are respectively having mean values of 1.077 and 1.554 which indicates that discretionary accruals have increased relatively faster than non-discretionary accruals in Ethiopia. The mean value of discretionary expenses is equal to 0.096, this indicates that on average (over all the years in the sample size) the discretionary expenses (calculated by adding up research and development costs, selling costs and administrative expenses) equals 9 million and 600 thousand Eth.Birr. The mean value of abnormal discretionary expenses equals 0.068 which indicates that 6.8% of the total assets consist of abnormal discretionary expenses. Looking at the kurtosis values of all the variables, we notice large values (higher than 10) in almost all variables. Kurtosis represents the size of the tails of the distribution; a high kurtosis indicates a lot of extreme values. The distribution of this sample size consists of extreme values which is inherent to the characteristics of the companies in the sample size. The growth of capital is normally distributed looking at the two banks, Dashen and Nib. The kurtosis of the two most important variables (discretionary accruals and abnormal discretionary expenses, the proxies for the two applied models) are relatively high (higher that 10) which is a result of winsorizing

4.2 Multiple Regression Results Modified Jones Model


| | (1) |  
| VARIABLES | TACC/laggedTA |  
| ![img-0.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-0.jpeg) | 0.00007\*\*\* |  
| | (0.233) |  
| ![img-1.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-1.jpeg) | -0.233\*\*\* |  
| | 0.076 |  
| ![img-2.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-2.jpeg) | 0.076\*\*\* |  
| Observations | 2 |  
| R-squared | 0.0302 |  
| adj. R-squared | 0.0204 |  
| F-statistic | 18.0000 |  
| p(F) | 0.0001\* |  
| \*Standard errors in parentheses. Significance levels: \*\*\* p<0.01, \*\* p<0.05, \* p<0.1\* |

\*Table4-4: OLS estimation from the modified Jones Model\*

The R-squared indicates how the variation of the depending variable is being explained by the independent variable; in this model the R-squared equals 3.02%. The adjusted R-squared is the R-squared adjusted for the number of variables. The F-statistic combined with the p-value, tests the explanatory power of the model. The P value is smaller than 5% which indicates that the model is significant explanatory. Therefore, this model can be used to test the hypotheses discussed in the previous chapter.

\*Table4-5: OLS Regression Estimation\*

| | (1) | | (2) | | (3) | |  
| VARIABLES | DACC/laggedTA | VIF | DACC/laggedTA | VIF | DACC/laggedTA | VIF |  
| IFRS | 0.093\*\*\* | 10.76 | 0.071\*\*\* | | 0.082\*\*\* | 14.16 |  
| delta\_EBIT | | | 0.005 | 217.23 | | |  
| (IFRS X delta\_EBIT) | 0.001 | | 0.000 | 217.23 | | |  
| ROA | 0.001 | | 0.001 | | 0.001\*\* | |  
| (IFRS X ROA) | 0.000 | | 0.000 | | 0.000\*\*\* | 217.23 |  
| | | | | | (0.000) | |  
| Constant | 0.092\*\*\* | | 0.071\*\*\* | | -0.018\*\*\* | |  
| | (0.091) | | (0.091) | | (0.091) | |  
| Observations | 2 | | 2 | | 2 | |  
| R-squared | 0.032 | | 0.032 | | 0.016 | |  
| adj. R-squared | 0.0204 | | 0.032 | | 0.016 | |  
| F-statistic | 18.000 | | 12.916 | | 23.084 | |  
| p(F) | 0.001 | | 0.001 | | 0.000 | |  
| Dependent variable: ![img-3.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-3.jpeg)![img-4.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-4.jpeg) \= Discretionary accruals from the Modified Jones Model that proxies for audit quality;

Independent variables: ![img-5.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-5.jpeg)![img-6.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-6.jpeg) \= A dummy variable that equals 0(1) if year ![img-7.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-7.jpeg)![img-8.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-8.jpeg) falls in period before (after) IFRS adoption (testing hypothesis 1 in specification 1); ![img-9.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-9.jpeg)![img-10.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-10.jpeg) \= The change in EBIT for firm ![img-11.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-11.jpeg)![img-12.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-12.jpeg) at time period ![img-13.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-13.jpeg)![img-14.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-14.jpeg) to ![img-15.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-15.jpeg)![img-16.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-16.jpeg); ![img-17.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-17.jpeg)![img-18.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-18.jpeg) \= Return On Assets for firm ![img-19.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-19.jpeg)![img-20.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-20.jpeg) at time period ![img-21.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-21.jpeg)![img-22.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-22.jpeg);

Interaction variables: ![img-23.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-23.jpeg)![img-24.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-24.jpeg) \= The interaction between the dummy variable for the introduction of IFRS and the change in EBIT, that measures the differential effect of EBIT on the audit quality before and after the IFRS adoption (testing hypothesis 2 in specification 2); ![img-25.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-25.jpeg)![img-26.jpeg](https://attachments.academia-assets.com/markdown/94593530/img-26.jpeg) \= The interaction between the dummy variable for the introduction of IFRS and the ROA, that measures the differential effect of ROA on the audit quality before and after the IFRS adoption (testing hypothesis 3 in specification 3);

\*Standard errors in parentheses. Significance levels: \*\*\* p<0.01, \*\* p<0.05, \* p<0.1\* |

\*\*Discussion\*\*

The P values of all three regressions are significant (smaller than 5%) and this is an indication that this model is usable for conducting this research. Also, the VIF values are all lower that 5 which indicates no sign of multicollinearity.

The following hypotheses are defined and tested using the Modified Jones Model:

* H1: The audit quality, measured in discretionary accruals, has been affected positively by the adoption of IFRS.
* H2: A higher EBIT, compared to the year before, is associated to an increase of discretionary accruals in the years prior to the adoption of the IFRS compared to the period after the adoption IFRS.
* H3: A higher ROA has a higher impact on audit quality in the years prior to the adoption of the IFRS compared to the period during the IFRS.

In table 2, the hypotheses named above are respectively specified as (1), (2) and (3). In specification 1 of table 2, the coefficient of the dummy variable for the introduction of IFRS is significant at a 5% level for all three models. This results in the conclusion that the adoption of IFRS has had a positive effect on discretionary accruals. In other words, the discretionary accruals have irregularly increased after the adoption of IFRS. Therefore, H1 is rejected.

In specification 2 of table 2, the coefficient of the interaction of the difference in the EBIT with respect to the previous year with the dummy variable for the introduction of IFRS (IFRS x delta EBIT), is significant at a 1% level. This indicates that a higher EBIT (compared to prior year) has resulted in higher discretionary accruals in the period after the adoption of IFRS (compared to the period before the adoption of IFRS). Therefore, H2 is accepted.

In specification 3 of table 2, the coefficient of the interaction of ROA with the dummy variable for the introduction of IFRS is significant at a 5% level. This indicates that a higher ROA (compared to prior year) has resulted in higher discretionary accruals in the period after the adoption of IFRS (compared to the period before the adoption of IFRS). Therefore, H3 is accepted.

* ```  
1. **Multiple Regression Results Real Based Earnings Model**

*Table-4-6: OLS estimations for the real activities manipulation model (Roychuwdhury, 2008)*

| | (1) |
| VARIABLES | DiscrExpense |
| | 0.000*** |
| | (0.009) |
| | 0.025 |
| | (0.200) |
| Constant | 0.092*** |
| | (0.907) |
| Observations | 2 |
| R-squared | 0.3930 |
| adj. R-squared | 0.3710 |
| F-statistic | 39.4960 |
| p(F) | 0.0000 |
| *Dependent variable: scaled discretionary expenses.*

*Standard errors in parentheses. Significance levels: *** p<0.01, ** p<0.05, * p<0.1* |

**Discussion**

The R-squared indicates how the variation of the depending variable is being explained by the independent variable; in this model the R-squared equals 3, 93%. The adjusted R-squared is the R-squared adjusted for the number of variables. The F-statistic combined with the p-value, tests the explanatory power of the model. The P value is smaller than 5% which indicates that the model is significant explanatory. Therefore, this model can be used to test the hypotheses discussed in the previous chapter.

| Dependent variable: Abnormal discretionary expenses as defined in Roychowdhury (2006).

Independent variables: = A dummy variable that equals 0(1) if year falls in period before (after) IFRS adoption (testing hypothesis 1 in specification 1); = The change in EBIT for firm at time period to ; = Return On Assets for firm at time period ;

Interaction variables: = The interaction between the dummy variable for the introduction of IFRS and the change in EBIT, that measures the differential effect of EBIT on the audit quality before and after the IFRS adoption (testing hypothesis 2 in specification 2); = The interaction between the dummy variable for the introduction of IFRS and the ROA, that measures the differential effect of ROA on the audit quality before and after the IFRS adoption (testing hypothesis 3 in specification 3);

*Standard errors in parentheses. Significance levels: *** p<0.01, ** p<0.05, * p<0.1* |

**Discussion**

The P values of all three regressions are significant (smaller than 5%) and this is an indication that this model is usable for conducting this research. Also, the VIF values are all lower that 5 which indicates absence of multicollinearity.

Unlike the Modified Jones Model, the dummy variable Crisis is significant in this model. The abnormal discretionary expenses were lower during the financial crisis (started in 2013). A possible explanation why this variable is significant in this model and not in the Modified Jones model is the calculation of discretionary expenses. Discretionary expenses are calculated differently than discretionary accruals.

The following hypotheses are defined and tested using the Real Based Earnings Model:

H1: The audit quality, measured in discretionary accruals, has been affected positively by the adoption of IFRS.
H2: A higher EBIT, compared to the year before, is associated to an increase of discretionary accruals in the years prior to the adoption of the IFRS compared to the period after the adoption IFRS.
H3: A higher ROA has a higher impact on audit quality in the years prior to the adoption of the IFRS compared to the period during the IFRS.

In table 2, the hypotheses named above are respectively specified as (1), (2) and (3). In specification 1 of table 3, the coefficient of the dummy variable for the introduction of IFRS is significant at a (0.4) % level for all three models. This results in the conclusion that the adoption of IFRS has had significant effect on discretionary expenses. In other words, the discretionary expenses have increased after the adoption of IFRS. Therefore, H1 is accepted.

In specification 2 of table 3, the coefficient of the interaction of the difference in the EBIT with respect to the previous year with the dummy variable for the introduction of IFRS (IFRS x delta EBIT), is significant at a 5% level. This indicates that a higher EBIT (compared to prior year) has resulted in higher discretionary expenses in the period after the adoption of IFRS (compared to the period before the adoption of IFRS). Therefore, H2 is accepted.

In specification 3 of table 3, the coefficient of the interaction of ROA with the dummy variable for the introduction of IFRS is significant at a 5% level. This indicates that a higher ROA (compared to prior year) has resulted in lower discretionary expenses in the period after the adoption of IFRS (compared to the period before the adoption of IFRS). Therefore, H3 is accepted.

Correlations


In the table below, the correlations between the variables applied in the Modified Jones Model and Real Based Earnings Model are presented. \*\*Discussion\*\*

DACC and NDACC (-0.1791). An increase in discretionary accruals results in a decrease of the non-discretionary accruals. This is explainable due to the fact that the Modified Jones Model calculates discretionary accruals by deducting the total accruals with the non-discretionary accruals.

AbnDiscrExp and DiscrExp (-0.9230). A decrease in abnormal discretionary expenses results in an increase of the discretionary expenses. In the Real Based Earnings Model, the discretionary expenses consist of abnormal discretionary expenses and non-abnormal discretionary expenses. This negative relation is therefore not a surprise.

* ```  
1. **Limitations**

There are various limitations involved when conducting this research. These limitations will be discussed below.

Availability of data: Data prior to 2013 is not available or at least not sufficient in the database Compustat. Therefore, the sample size consists of observations after 2013. Furthermore, variables such as audit size and audit tenure for Ethiopia companies are unavailable in the database Compustat and are therefore excluded from this research.
Limitations of the Modified Jones Model: Despite that the Modified Jones Model is the most often used model regarding discretionary accruals, there are some limitations related to the modified Jones Model. The most important limitation is noted by M.D. Beneish^{[1]}, the Modified Jones Model is not recognizing accruals that result from the performance of the firm. In addition, Dechow, Sloan and Sweeney^{[2]} found that the Modified Jones Model is able to accurately handle large amounts of cash flows.
Not considered variables: different variables are discussed that may impact discretionary accruals. However, there is certain likelihood that variables that have not been subjected to prior research are impacting discretionary accruals. This research only restricts to the variables discussed in previous chapter.

1. **Conclusion**

In this chapter, the research has been conducted and the results have been used to determine the value of the different hypotheses. Almost all the hypotheses are accepted (except the first hypothesis which rejected by the modified Jones model) due to the finding that the P-value is lower than 5%. These results have higher significance which then bringing the following results:

*Table 4-10: conclusion tables of results*

Conclusion

The main research question is described as:

Did the adoption of IFRS in Ethiopia affected audit quality for commercial banks of Ethiopia?

The results in chapter 4 indicate that there is no effect on audit quality for commercial banks in Ethiopiaa after the adoption of IFRS when applying the Modified Jones Model or but does in Roychowdhury’s Earnings Management Model. Furthermore, in order to determine if variables other than discretionary accruals are affecting audit quality, the following sub research questions are defined:

Sub research question 1: Is the level of EBIT affecting audit quality for stock listed Ethiopian comercial banks after the adoption of IFRS?
Sub research question 2: Is the level of ROA affecting audit quality for stock listed Ethiopian companies after the adoption of IFRS?

The results in chapter 4 indicate that the level of EBIT is affecting audit quality for commercial banks in Ethiopia due to the adoption of IFRS when applying the Modified Jones Model or Roychowdhury’s Earnings Management Model. The effect of ROA is also showing the same results when addressed by the two used models. The Roychowdhury’s Earnings Management Model did find a significant association between ROA and audit quality due to the adoption of IFRS. In the mean time, the Modified Jones Model is also showing the same results. This model found that a high level of ROA is associated with a lower level of discretionary accruals after the adoption of IFRS. As discussed before, a high level of discretionary accruals is associated with a low level of earnings management and therefore a higher level of audit quality. It is interesting to see how these findings can be explained by economic theories and empirical results found in previous literature as discussed in chapter 2.

5.2.1. First Paper Contradicting the Findings

In the paper of Yi Lin Chua evidence is found that the adoption of IFRS has resulted in better accounting quality in Australia. However, the effect on earnings management is negative. These last finding is not in line with the findings of this project. We will take a brief look at the research design of Yi Lin Chua’s paper:

The period examined: 2013 to 2021;
Notable variables used: change in net income, change in variability of cash flows, Spearman correlation and regression of small positive earnings, Auditor size and bank effects;
Model used is based on Spearman correlation.

Looking at the key elements above, there are some differences comparing Yi Lin Chua’s paper with this project. Yi Lin Chua is not using discretionary accruals or discretionary expenses in any form in order to measure audit (or accounting quality), while this project is using discretionary information. In addition, Yi Lin Chua is using a model that is based on the Spearman correlation, while this project using the Modified Jones Model and Roychowdhury’s Earnings Management Model. Furthermore, this project controls for the financial crisis started in 2021, while Yi Lin Chua is only controlling for industry effects. Despite these differences, the findings of Yi Lin Chua regarding earnings management are different to this project.

5.2.2. Second Paper Contradicting the Findings

The paper of Thomas Jeanjean found evidence that the pervasiveness of earnings management did not declined in France after adopting IFRS. The research design is somewhat opposite compared to the research design of this project:

Discretionary accruals are used to measure earnings management;
The Modified Jones Model is being used;
Additionally, statistical thresholds are identified.

There are some differences between the paper of Thomas Jeanjean and this project such as focusing on France instead of Ethiopia and not using Roychowdhury’s Earnings Management Model. Also, the statistical threshold is an extensive part in the research performed by Thomas Jeanjean. Besides, despite these differences, the findings of Thomas Jeanjean are different as compared to the findings of this project.

5.2.3 First Paper Supporting the Findings

Panagiotis E. Dimitropoulos found evidence that the implementation of IFRS has contributed to les earnings management compared to local Greece GAAP. The research design is including the following notable aspects:

Used model is a combination of the Modified Jones Model and Basu’s model (a stock price model)
Control variables: firm leverage and growth opportunities;
The period examined: 2013 to 2021.

The results of Panaglotis same with this project most likely even if the use of a different model and different control variables. Another important difference between Panaglotis’s paper and this project is the researched region. Greece has been in a different economic situation than Ethiopia, both in terms of financial growth and the domestic GAAP (valid before the adoption of IFRS).

5.2.4 Third Paper Contradicting the Findings

Claudia Lopes found that the adoption of IFRS has resulted in a negative effect on accounting quality for firms within the Ethiopian banks. This paper used the discretionary accruals to evaluate the development of accounting quality throughout the years 2013 to 2021. The Modified Jones Model is used to calculate the discretionary accruals. Despite the focus on Ethiopia, the findings of Claudia Lopes are different to the findings of this project. This weakness the validation of our findings.

The final sub research question was described as:

Sub research question 3: Is the effect on audit quality due to the adoption of IFRS regarding Ethiopian commercial banks similar compared to companies in other regions?

The papers discussed were focusing on Dashen Bank and Nib Int’l bank for the purpose of this project. The paper that focused on these two banks show similar results and explanations are discussed in the section above.

The only difference in findings between the applied models (the Modified Jones Model or Roychowdhury’s Earnings Management Model) in this project is nothing rather is the same in H1.

Reference

Literature

Teferi Deyuu Alemi*, Dr. J.S.Pasricha(Professor). IFRS adoption progress in Ethiopia, Research Journal of Finance and Accounting, ISSN 2222-1697(paper) ISSN 2222-2847(online) Vol.7, No.1, 2016

Goshu Desalegn. Does IFRS adoption improve financial reporting quality: evidence from coomercial bank of Ethiopia, Journal of finance and accounting, ISSN 2222-697(paper)ISSN2222-2847(online) Vol.11, No.7, 2020.

Denberu Kenubeh Dagnew, implementation of IFRS standards, its practice’s challenges and prospects, the case of commercial banks in Ethiopia, Asia-Pacific Journal of Management research and innovation ,2021, DOI: 10.1177/2319510X21X211013597

Amanuel Tsegaye (2019). The effects of International Financial Reporting Standards (IFRS) adoption on audit fees in Ethiopia - The case of commercial banks

Alemgena Bekele (2016). The Benefits and Key Challenges Of Using International Financial Reporting Standard (Ifrs): Case Study Of Heineken Brewery Factory

Davidson, R. & Neu, D. (2010). A note on the association between audit firm size and audit quality. Contemporary Accounting Research, 86(1), 479-488.

Bartov, E (2000). Discretionary-Accruals Models and Audit Qualifications. Journal of Accounting and Economics, 31(3), 421-452. DOI: 10.1016/S0165-4101(01)00015-5.

Audit Quality and Accrual Persistence: Evidence from pre- and post SOX – Dennis Chambers and Jeff Payne (2010)

Lawrence, A., Minutti-Meza, M. & Zhang, P (2010). Can Big 4 Versus Non-Big 4 Differences in Audit-Quality Proxies be attributed to Client Characteristics? Accounting Review, 86(1), 259-288.

Knechel, W., Krishnan, G., Pevzner, M., Schefchik, Lori. & Velury, U. (2013). Audit Quality: Insights from the Acadamic Literature. Auditing: A Journal of Practice & Theory, 32(1).

Francis, J. & Krishnan, J. (1999). Accounting Accruals and Auditor Reporting Conservatism. Contemporary Accounting Research.

Armstrong, C. (2010). Market Reaction to the adoption of IFRS in Europe.

Nelson, S. & Shukeri, S. (2011). IFRS Adoption and Audit Timeliness: Evidence from Malaysia. DOI: 10.2139/ssrn.1967284.

Zeghal, D. & Chtourou, S. (2001). An analysis of the effect of mandatory adoption of IFRS on earnings management. Journal of International Accounting, Auditing and Taxation, 20(2), 61-72. DOI: 10.1016/2011.06.001.

JeanJean, T. & Xiaohong, L. (2012). Do accounting standards matter? An exploratory analysis of earnings management. Journal of Accounting and Public Policy, 27(6), 480-494. DOI: 10.1016/2008.09.2008.

Jeong-Bon, K., Xiaohong, L. & Zheng, L. (2012).The impact of mandatory IFRS Adoption on Audit Fees: Theory and Evidence. Accounting Review.

Lopes, C. & Cerqueira, A. (2010). Impact of IFRS adoption on accounting quality in European firms.

Philips, F., Libby, R. & Libby, P (2008). Fundamentals of Financial Accounting (4th edition). ISBN: 978-0077757854

Choi, J., Kim, J., Chansong, K., Zang, Y. (2010). Audit Office Size, Audit Quality and Audit Pricing.

Roychowdhury, S. (2006). Earnings Management through Real Activities Manipulation. Journal of Accounting and Economics, 42(3), 335-370. DOI: 10.1016/j.jacceco.2006.01.002

Ma, S. (2010). The puzzle of negative association between earnings quality and corporate performance. DOI: 10.2139/ssrn.2136693.

Fang, H., Chen, K. & Shiao, Y. (2006). The Study of Audit Independent System Effect on Manipulation of Discretionary Accruals of Companies.

Kothari, S., Leone, A. & Wasly, C. (2005). Performance matched discretionary accrual measures. Journal of Accounting and Economics, 39(1), 163-197. DOI: 10.1016/j.jacceco.2004.11.002.