A STUDY OF BANKING COMPANIES IN BANGLADESH

Category: Accounting

Time lag in financial report publication and audit delay are intertwined and used interchangeably in financial reporting literature. As a result, in most cases timeliness have actually dealt with audit delays. The long audit delay normally leads to an even longer publication delay as companies in Bangladesh are reluctant in calling the annual general meetings (AGM) of shareholders in years with poor financial performance and/or low or no dividend announcement prospects. Although the Companies Act, 1994 requires all companies, listed and unlisted, to furnish their annual accounts before the AGM within nine months of expiry of their respective financial years, a significant portion of these companies do not comply with this requirement. Many companies do not submit their annual accounts with the Registrar of Joint Stock Companies for several years. Some companies are found to take up to seven years to present audited financial statements before the AGM of shareholders (Karim, Ahmed & Islam, 2006). The usefulness of the information disclosed in company annual reports (CARs) will decline as the time lag increases, and it has been argued by Abdulla (1996) that the longer the period between year end and publication of the annual report, the higher the chances that the information will be leaked to some interested investors.

In addition to the above, the steady increases in foreign investment into the country and greater degree of financial liberalization are expected to improve the timeliness of financial reporting in the country over time. This study focuses on one aspect of corporate financial reporting in banking companies - timeliness. The study aims at exploring whether the developments in the financial reporting regulatory environment have been successful in significantly reducing the time lag in publishing financial statements by banking sector in Bangladesh.

2. Scoring the Index

There are various approaches available to develop a scoring scheme to determine the disclosure level of corporate annual reports from works of other researchers. Wallace (1988), Cooke (1991 and 1992), Robbin and Austin (1986), Hossain et al. (1994) and Ahmed and Nicholls (1994) adopted a dichotomous procedure in which an item scores one if disclosed and zero if not disclosed. The approach used by Courtis (1979), Barrett (1976 and 1977) and Marston (1986) was for a weighted disclosure index to be employed. In some cases the researchers predetermined the weights subjectively.

In the present study, the annual reports of the Banking organizations have been examined against unweighted index. In case of unweighted disclosure index, while measuring the level and extent of disclosure, the disclosure will be heated as a dichotomous variable. Here, the only consideration is that if a company discloses an item of information in its corporate annual report it has been awarded '1' and if not, it has been awarded '0'. In the disclosure model which we have used and followed in this study, the total disclosure score for a banking company taken to be additive. In our study, unweighted disclosure index have been considered. Among the total of 144 items in the disclosure index we had 44 items, which consisted of sub-items. For scoring the sub-items the total score'1' has been distributed equally because we have thought that all the sub-items are equally important.

3. Measuring Timeliness: Audit Lag, AGM Lag, and Total Lag

Three measures of timeliness are defined: (1) audit lag, (2) AGM lag, and (3) total lag:

Audit lag: Audit lag means the interval of days between balance date and the date of the auditor's report. [The interval of the number of days from the year-end to the date recorded as the opinion signature date on the auditor's report. Or the open interval of the number of days from the year-end to the date recorded as the opinion signature date in the auditor's report.] Audit delay represents the number of days elapsed between the balance sheet date and the date auditor(s) sign(s) the financial statements.

AGM lag: AGM lag means the interval of days between the date of the auditor's report and the date of the annual general meeting. The open interval of the number of days from the opinion signature date on the auditor's report to the date of settlement of annual general meeting (AGM). AGM lag represents the number of days elapsed between the date of signing auditor's report and the day on which the AGM is actually held.

Total lag: Total lag refers the interval of days between balance date and the date of the annual general meeting. The open interval of the number of days from the year end to the date of settlement of annual general meeting. Total lag represents the total interval time after balance date before the directors formally present financial results to the owner of the entity.

Audit lag and AGM lag subdivide total lag into two components: the interval of days it takes before audited accounting information becomes available for release (through the press), and the time management takes to organize all necessary activities to bring on the company's annual general meeting.

4. Establish the Impact of Disclosure Score on Timeliness

To test whether disclosure level of sample banks is affected by the timeliness of the bank's financial reporting, we have taken different measures of timeliness such as audit lag, AGM lag and total lag to regress them with disclosure score. Accordingly, we have formulated different null hypotheses. The following specific hypotheses have been tested regarding timeliness.

H1: Timeliness as measured by audit lag does not affect the disclosure score of the sample banks.

H2: Timeliness as measured by AGM lag does not affect the disclosure score of the sample banks.

H3: Timeliness as measured by total lag does not affect the disclosure score of the sample banks.

H4: There is no significant association between three measures of timeliness [viz., audit lag {AuL}, AGM lag {AgmL} and total lag {TotL} and the extent of disclosure.

5. Methodology

This section considers the research methodology of the present study. The aims of this chapter are to set up the foundation of the statistical analysis and to highlight the statistical tools, techniques and applications of performance measurement regarding disclosure of financial reporting. The present research decides to adopt the disclosure index method of measurement. Using the disclosure index method the level of disclosure is measured by selecting items that might be disclosed in annual reports and giving banks a score based on the number of disclosures actually made. Despite its inherent problems (e.g., subjective judgments), disclosure index approach has become an important vehicle for the measurement of the company information disclosure. Some of the researchers used weighted disclosure indices while other researchers used unweighted disclosure indices.

An unweighted index is the ratio of the value of the number of items a company discloses divided by total value that it could disclose (Hossain, 1998). Under an unweighted disclosure index, all items of information in the index are considered equally important to the average user. The choice of an unweighted index over a weighted one does not produce substantially different results (e.g. Chow and Wong-Boren, 1987; p.537) and there are researchers who favored the use of unweighted indices (e.g. Spero, 1979; p.57 and Rubbins and Austin, 1986). The present researcher has decided to follow the unweighted disclosure index approach in his study because there are merits in using both the methods and the results obtained have been the same for the use of either methods.

6. Type of Data

The present study is based on the secondary information. The secondary information has been collected from the corporate annual reports of the selected banking organizations. In the present study the comprehensive corporate annual reports of the selected banking companies of the selected years will be used to compute the disclosure scores. Banks' annual reports encompass most of the secondary information of this research. We have also conducted intensive library work to collect other secondary data.

7. Selection of Sample Bank

The banking sector of Bangladesh comprises of four categories of scheduled banks. As of June 2008, 49 scheduled banks are operating in Bangladesh with a network of 6236 branches. The scheduled banks have 3699 branches in the suburban or rural areas, which is 59.32 percent of the total number. (Source: Bangladesh Bank and Securities and Exchange Commission).

Before the study, it was decided that the sample would cover the annual reports of the banking companies in Bangladesh for the year 2002 - 2006. The planned size of the sample represented 12 banks. The sample banks are listed on Dhaka Stock Exchange in Bangladesh. The sample banks were selected by using of judgmental sampling approach, which is based on advance commencement of business as banking companies in Bangladesh, (viz., the banks are old according to the year of incorporation) and the availability of Annual reports.

8. Tools of Analysis

Statistical tools like average, standard deviation, simple and multiple regression, correlation, t tests have been used to analyze the data. In the unweighted disclosure index disclosure of individual items has been treated as a dichotomous variable. Here, the only consideration is whether or not a bank discloses an item of information in its corporate annual report. If a bank discloses an item of information in its annual report it will be awarded '1' and if not it will be awarded '0'. The disclosure model for the unweighted disclosure thus measures the total disclosure (TD) score for a bank as additive as follows-

TD =

Where,

d = 1 if the item di is disclosed

0 if the item di is not disclosed

n = number if items

9.(a) Regression Studies between Disclosure Score vs. Audit Lag (AuL)

Table showing the Regression Studies between Disclosure Score vs. Audit Lag (AuL)

From the regression result we observe that values of R2 are small in case of year 2002, 2004, 2005 and 2006 and the significance levels of the regression co-efficients are above .05 for the said years. Further, we observe that values of R2 are high in case of year 2003 and the significance level of the regression co-efficient is below .05 for this year. So, our null hypothesis is rejected which means that timeliness as measured by AuL affects the disclosure score.

9.(b) Regression Studies between Disclosure Score vs. AGM Lag (AgmL)

Table showing the Regressional Studies between Disclosure Score vs. AGM Lag (AgmL)

From the regression result we observe that values of R2 are small and the significance levels of the regression co-efficient are above .05. So, our null hypothesis is accepted at significance level .05, which means that timeliness as measured by AgmL does not affect the disclosure score. Further, we observe that values of R2 are relatively high in case of year 2002 and 2006 and the significance levels of the regression co-efficient are below .12 of this year.

9.(c) Regression Studies between Disclosure Score vs. Total Lag (TotL)

Table showing the Regressional Studies between Disclosure Score vs. Total Lag (TotL)

From the regression result we observe that values of R2 are small and the significance levels of the regression co-efficient are above .05. So, our null hypothesis is accepted at significance level .05, which means that timeliness as measured by AgmL does not affect the disclosure score. Further, we observe that values of R2 are relatively high in case of year 2002, 2003 and 2006 and the significance levels of the regression co-efficient is below .11 in case of year 2003 and regression co-efficient are below .16 in case of year 2002 and 2006.

10. Independent Variables with their Labels and Relationships in the Regression

The description of the three independent variables, their labels and expected signs and relationships are presented in Table 4.

Table showing the list of independent variables, their labels and expected signs and relationships in the regression

11. Multiple Regression Models

Multiple linear regression techniques are used to test two alternative versions of each hypothesis. The model is created using UDI as the dependent variable.

UDI =  + 1 AuL + 2 AgmL + 3 TotL + 

Where UDI = total score received each sample bank under unweighted disclosure index;

 = the constant, and

 = the error term.

12. Regression Studies between Disclosure Score vs. Multiple Variables

The goal of this chapter is to examine the association between the extent of information disclosure in published annual reports and timeliness. Accordingly we have formulated null hypotheses (H4). The anticipated association is examined by that hypothesis. The multiple linear regression technique is used to test the two alternative hypotheses. To test whether disclosure level of sample banks is affected by multiple variables, we have taken the measures such as audit lag (AuL), AGM lag (AgmL) and total (TotL) to regress them with disclosure score.

Table showing the Regressional Studies between Disclosure Score vs. Multiple Variables

From the multiple regression result we observe that values of R2 are small in case of year 2002, 2004, 2005 and 2006 and the significance levels of the regression co-efficient are above .05 for the said years. Further, we observe that values of R2 are high in case of year 2003 and the significance levels of the regression co-efficient are below .05 of this year. So, our null hypothesis is rejected which means that timeliness as measured by Multiple variables affects the disclosure score only of the context of the year 2003.

13. Discussion and Conclusion

This study reports the results of an empirical examination of the association between financial reporting timeliness and disclosure score. Three measures of timeliness are used. One, in terms of the number of days it takes a company to have the audit completed, the second the number of days it takes a bank to hold its AGM since the date of its fiscal year audit completed, and finally, the number of days it takes a bank to hold its total since the date of its fiscal year-end. Two levels of analyses are carried out. First, an analysis of the trend of extent in the three lags and second, statistical analysis on the impact of disclosure score on timeliness.

Results show that AGM, and total delays are not associated with disclosure score, i.e., there has been no significant positive relationship in corporate timeliness in reporting with the disclosure scores in reporting during the period under study. Though audit lag has the significant positive relationship.

During the post-regulatory period i.e., post- 'BRPD circular- 3' period, timeliness has deteriorated significantly which suggests that regulatory changes have failed to bring about improvement in the quality of financial reporting of banking sector in Bangladesh with respect to timeliness. The findings of this study can be used in the debate on the efficacy of regulatory pressure on financial reporting of banking sector in Bangladesh. The regulatory changes through BRPD circular brought about the banking sector in Bangladesh throughout the study period have been substantial. It was expected that the change would improve the age-old problem of chronic publication delay in corporate financial reporting of banking sector in the country. Mean publication delayhas shown mixed trend over the years.

The regression models of disclosure score formulated here are identical to identify the relationship between disclosure score and timeliness attributes in the sample banks. The inclusion of the timeliness attributes used in the regression models has been discussed. The results suggest that the explanatory variables used in the studies of timeliness proved to be significantly negatively associated with disclosure score. AGM lag and total lag found not to be significantly associated with disclosure score. Other variable 'audit lag' failed to establish significant positive relationship with the disclosure score. From the multiple regressions result shows that only for the year 2003 timeliness attributes affects the disclosure score and in case of year 2002, 2004, 2005 and 2006 the timeliness attributes do not affect the disclosure more. From the results of this study the following conclusions can be drawn. Firstly, there appears to be an unusually audit delay made by the Bangladeshi listed banking companies soon after the balance sheet date. The average interval of time between balance sheet date and the date of auditor's report is 3.4 months (the mean audit lag over the 5 years period ranges from 95 days in 2005 to 121 days in 2003 with a mean delay of 102 days for the entire population). Although the minimum audit delay is low (43 days), the average audit lag is 102 days as against approximately 40 days after their clients' balance sheet dates in the USA, and approximately 80 days in the case of the listed companies in New Zealand and Australia. So, banking companies in Bangladesh are taking relatively more time to complete audit of their accounts. As a result the appeal of the information provided by the bank annual reports cannot help the users to take their decision in time if it takes another 102 days to arrange annual general meeting. With regard to timeliness as a qualitative objective of financial statements, this evidence can be regarded as unsatisfactory. However, the findings of this study may be generalized after taking into consideration its limitations.

Directions for future research:

The present research considers only the annual reports for listed banking companies. This study does not consider non-listed or non-financial or manufacturing companies. Further research can be undertaken to measure audit delay longitudinally in company characteristics (multi groups of companies) to determine whether the trend of audit delay has improved over time. Such a study would provide additional insights into the underlying causes for the audit delay in developing countries in general and in Bangladesh in particular. However, if anyone includes listed non-financial companies in the sample, he can attempt to examine the relationship between audit delay and industry type i.e., non-financial as '1' and financial as '0'. The results may be different if the number of company characteristics was increased or another set of variables were examined. Although the sample consisting 12 listed banks (25% of the population) in Bangladesh is reasonable, further research can be undertaken with a larger sample. This might be useful with respect to the stability of the regression equation.