Volume 45, Issue 4 pp. 493-517
Full Access

An Analysis of Thai Financial Reporting Practices and the Impact of the 1997 Economic Crisis

SIRILUCK SUTTHACHAI

SIRILUCK SUTTHACHAI

Department of Accounting, Khon Kaen University

Search for more papers by this author
TERENCE E. COOKE

TERENCE E. COOKE

University of Exeter

Search for more papers by this author
First published: 25 November 2009
Citations: 12
Siriluck Sutthachai is a Lecturer in Accounting in the Department of Accounting, Khon Kaen University, and Terence E. Cooke ([email protected]) Emeritus Professor of Accounting, University of Exeter.

The authors would like to thank the two referees for their very helpful comments.

Abstract

We focus on listed Thai companies between 1993 and 2002 to ascertain whether the 1997 economic crisis, which we refer to as an economic disturbance, had an impact on financial reporting practices. Both changes in measurement and disclosure practices are considered and the period of study is divided into three sub-periods: the pre-economic crisis period (1993–96), the economic crisis period (1996–98) and the post-economic crisis period (1998–2002). The results show that there were significant increases in disclosure levels over the ten years but no substantial changes in measurement methods. This work takes on added relevance in the light of the recent (2007–08) economic bubble and subsequent financial crisis worldwide.

In the early 1990s many Asian countries, including Thailand, attracted considerable foreign capital investment as a result of high rates of interest. A speculative bubble based on high levels of gearing developed that eventually burst in 1997 to be replaced by a credit crisis and bankruptcies. With foreign money being withdrawn from banks, pressure mounted on currencies (Kaufman et al., 1999) and the Thai government was unable to protect the fixed exchange rate. The government reluctantly decided to float the baht with adverse consequences for growth. In 1996 GDP growth was 5.9 per cent but declined to −1.7 per cent in 1997 and −10.2 per cent in 1998 (Bank of Thailand, 2003). In addition, foreign currency denominated liabilities increased substantially in terms of the domestic currency, which had the adverse consequences of a more pronounced credit squeeze and increased bankruptcies. The number of non-performing loans (NPL) in the banking system increased remarkably and weakened the Thai financial system. The economic crisis also affected and weakened economic systems in other West Pacific Rim countries such as Indonesia, Malaysia, Singapore and Korea. The International Monetary Fund (IMF, 1999) was asked to provide financial assistance to the region and after its investigation suggested that countries affected by the financial crisis should strengthen their financial systems and governance, including financial reporting regulations and practices.

Thailand responded with new laws, accounting standards based on International Accounting Standards (IASs), and guidelines for good corporate governance. As a result, studies of Thailand's economic crisis have claimed that the financial structure and corporate governance mechanisms have changed substantially from dependency on banks to reliance on money markets (Jaikengkit, 2002). Such a change is thought to lead to greater disclosure (Jaikengkit, 2002), although we are not aware that this has been tested. Therefore, our research objective is to study financial reporting practices from 1993 to 2002 to provide evidence of an economic disturbance theory of financial reporting.

We suggest that an economic crisis, such as that occurring in the West Pacific Rim in 1997, represents a fundamental economic disturbance that changes the environment and the relationship between stakeholders. The result of an economic crisis may be to change perceptions of those in government and provide a response in terms of a revised regulatory framework. The idea of an economic disturbance theory was first advanced by Gort (1969). We argue that a major financial crisis, particularly one that extends beyond the confines of a single country, leads to a break with the past and temporary disequilibriums as stakeholders and government realign to the revised position that often includes a changed regulatory stance. Thus, we hypothesize that economic disturbance can lead to changes in financial reporting practices as a result of changes in uncertainty and in regulatory framework.

To examine the impact of the economic crisis, the ten-year period (1993–2002) is divided into three sub-periods: pre-economic crisis (1993–96), economic crisis (1996–98) and post-economic crisis (1998–2002). If changes in financial reporting practices occur most between 1996 and 1998 we can infer that this is a response to the economic disturbance.

Both measurement and disclosure practices are considered as part of our overall assessment of financial reporting. Based on previous studies (Hagerman and Zmijewski, 1979; Inoue and Thomas, 1996; Jaggi and Leung, 2003; Missonier-Piera, 2004), four accounting measurement issues were investigated: accounting for inventory, investments, borrowing costs and depreciation. For disclosure practices, we consider aggregate (mandatory and voluntary combined), mandatory and voluntary disclosure. Information is considered to be mandatory if it is required by law, or by regulations of the Stock Exchange of Thailand (SET) or financial reporting standards, whereas information that a company has no obligation to disclose is described as voluntary.

To study financial reporting practices and the impact of the 1997 economic crisis we investigate changes in the number of companies using a particular measurement method and changes in corporate disclosure indexes for the period 1993 to 2002. The results show statistically significant changes in Thai disclosure practices but not so for measurement practices. Levels of disclosure increased significantly from 1993 to 2002 and mandatory disclosure seems to be the main contributor. The rate of change in the extent of disclosure was greatest during the economic crisis period.

We further investigate the determinants of Thai financial reporting disclosure in order to assist our explanation of managerial choices. Multiple regression analysis is used to investigate relationships between firm-specific characteristics and disclosure. The results show a consistently significant relationship between firm size and disclosure over the periods studied for all types of disclosure. Other variables, including leverage ratio, ownership variables and auditor type, are significantly associated with disclosure. This suggests that changes in disclosure may have resulted from both regulation and other managerial incentives based on positive accounting theory. However, the regulations seem to play a greater part than other incentives in management selection of financial reporting policies. Also, we found that sample size had an impact on the regression results when testing the associations between levels of disclosure and explanatory variables.

THEORETICAL FRAMEWORK

An example of an economic disturbance that caused a major change in financial reporting occurred in the U.S. in the 1920s when an asset price explosion was followed by a collapse in markets that helped shape changes in the regulatory environment. The Securities Act 1933 and Securities and Exchange Act 1934 dramatically changed the regulatory environment and resulted in the formation of the Securities and Exchange Commission. In our proposed theoretical framework, economic disturbance is likely to change elements of the accounting environment. In turn, changes in the environment then influence managerial incentives in selecting accounting practices. The selected policies will indicate characteristics of financial reporting practices in a company. There are four main theories widely applied in financial reporting research, namely, environmental-determinism theory, positive accounting theory, the cost-benefit approach and the accounting-culture relationship. These four theories are complementary rather than competing explanations and our eclectic approach integrates them into an overall framework (see Figure 1).

Details are in the caption following the image


A THEORETICAL FRAMEWORK OF THE INFLUENCES ON FINANCIAL REPORTING

The Environment and Culture

The environment shapes financial reporting (Cooke and Wallace, 1990). We define the environment in terms of institutional factors, including the legal system, the economic system (financial system and economic growth), the educational system, the nature of enterprises (referring to ownership structure), and regulations. Many researchers also indicate that culture is also influential (see, e.g., the Hofstede–Gray explanation).

Management Incentives

In our theoretical framework, the environment and culture interact to determine management incentives. These incentives are derived from positive accounting theory and the cost-benefit approach. Positive accounting theory has three main strands: agency theory, capital market-association theory and political cost theory. Agency theory considers that management has contractual incentives in preparing financial reports as an outcome of a contract with shareholders and debtholders. If companies' shares are widely held and the amount of debt is high, management is required to prepare financial reports to discharge its accountability to shareholders and to satisfy any covenants in debt contracts (Watts and Zimmerman, 1986; Wong, 1988). Therefore, management generally selects optimistic measurement methods, in the sense of presenting a more laissez-faire approach (Gray, 1988), and a transparent approach to preparing financial reports. On the other hand, if companies are family-owned and have little debt, management has fewer contractual incentives (i.e., less pressure from shareholders and debtholders). Thus, management is likely to be more conservative with respect to measurement practices and more secretive in terms of disclosure (Gray, 1988; Nobes, 1998).

We recognize that managers have incentives to provide relevant information for user decision making. These incentives may affect managers' choice of accounting methods as well as other financial reporting decisions. In practice, it is often difficult to distinguish information and contracting efficiency perspectives. From an information perspective, managers select accounting policies to show how future cash flows will change, with consequential impact on the value of the firm and claims against the entity. From an efficient contracting perspective, managers select accounting policies that reflect reality in terms of underlying economic events, transactions and cash flows that have already taken place.

Capital market-association theory assumes that earnings management and information disclosure have an impact on capital markets. Therefore, management may have an asset pricing motivation (Fields et al., 2001) to select particular financial reporting practices. Management tends to select optimistic accounting methods and a transparent approach in order firstly to reduce information asymmetry between investors and management and therefore reduce the cost of capital, and secondly, to increase confidence of investors and creditors in a company's performance and consequently increase the amount of funds and debt that a company can raise.

Political cost theory suggests that government may intervene when substantial private profits are generated in order to transfer wealth (Watts and Zimmerman, 1986). Since government uses information in financial reports along with other material to decide whether to intervene in a particular industry, management may disclose both mandatory and voluntary information and select income-decreasing measurement methods in order to reduce the possibility of intervention and political costs (Watts and Zimmerman, 1986).

Management incentives may change if benefits from particular financial reporting policies outweigh costs. The current research chooses three costs (production, litigation and competitive disadvantages) and one benefit (public relations) as possible management incentives to selecting financial reporting practices (Gray and Roberts, 1989; Elliot and Jacobson, 1994).

Management cannot satisfy all the incentives mentioned above but will have dominant ones that influence choices between financial reporting policies. The choices of measurement practices can be categorized into two groups—conservative and optimistic accounting methods—whereas disclosure practices can be either transparent or secretive (Nobes, 1998). These choices will result in different financial reporting practices. We hypothesize that if economic disturbance changes the Thai environment considerably, then management may respond by changing measurement and disclosure practices for financial reporting.

CHANGES IN THE THAI ENVIRONMENT

Over the period 1993 to 2002 several changes occurred in the Thai environment as a result of the disturbance of an economic crisis. Before the 1997 crisis the relationship between banks and businesses in Thailand was very close, with loans being authorized mainly on the basis of private relationships. However, after the crisis, Thai banks had a problem with NPL and thus the relationship between bankers and businesses became more distant. Furthermore, many Thai banks were required to recapitalize and this allowed increased foreign shareholdings. Ownership structure changed and as a result more pressure from outside shareholders was put on banks to perform, with the consequence that stricter lending rules were imposed.

Changes in the structure of share ownership also occurred with many family-based entities becoming widely held groups, institution-owned and foreign-owned, particularly those with a listing (Wiwattanakantang, 1999; Suehiro, 2001).

With both changes in the financial system and the nature of business enterprises, management may have responded with contractual and asset-pricing motivations. Also, new laws and regulations were introduced to enhance financial reporting practices and respond to the economic disturbance. Two main changes to laws directly dealing with financial reporting were introduced: the Accounting Act B.E. 2543 (2000) and the Accounting Professions Act B.E. 2547 (2004). The first Act requires companies to comply with Thai Accounting Standards (TAS) and, in addition, penalties for non-compliance were introduced. The second Act strengthened the role of the profession and defined the regulatory role of the Ministry of Commerce. In addition, after 1997, the SET issued several new regulations requiring additional disclosure and specified that listed companies must comply with the TASs.

HYPOTHESES DEVELOPMENT

Watts and Zimmerman (1986) developed positive accounting theory suggesting that shareholders, debtholders, capital markets and the government will all influence management choice of accounting methods in a manner that maximizes managerial benefits.

Changes in the Thai economic system, the nature of enterprises, and raising finance from capital markets, may have caused an increase in the contractual and asset-pricing motivations in which management is expected to choose an optimistic accounting method in preparing financial reports. We assume that, compared to their respective alternatives, FIFO for inventory, asset recognition (interest capitalization), FIFO for investments and straight line depreciation are all income increasing. These assumptions are consistent with the work of Zmijewski and Hagerman (1981). Therefore, the following hypotheses are developed for each accounting method.

H1: The number of companies using FIFO for inventory valuation will increase, compared to companies using the weighted average and LIFO method, over the period 1993 to 2002, particularly in the period of economic disturbance.

H2: The number of companies using the asset-recognition method will increase, compared to companies using the expense-recognition method, over the period 1993 to 2002, particularly in the period of economic disturbance.

H3: The number of companies using FIFO to account for investments will increase over the period 1993 to 2002, compared to companies using identification and weighted average approaches, particularly in the period of economic disturbance.

H4: The number of companies using the straight-line method will increase over the period 1993 to 2002, compared to companies using the decreasing-balance method, particularly in the period of economic disturbance.

For financial reporting disclosure practices, Watts and Zimmerman (1986) proposed that if companies' shares are diversely held and the amount of debt is high, management will disclose information to discharge its accountability to those various groups and reassure debtholders about company performance. Furthermore, positive accounting theory suggests that management seeking outside resources will disclose more information in financial reports in order to reduce information asymmetry.

In Thailand, the economic crisis forced companies to raise capital from external resources. Furthermore, there were changes in ownership structure which led management and shareholders to form contractual relationships. Thus, for disclosure, contractual and asset-pricing motivations tend to influence management to choose a transparent approach to prepare financial reports. In addition, Thai laws and regulations require companies to disclose more information than in the past. As such, the following three hypotheses were developed:

The level of disclosure (aggregate H5; mandatory H6; voluntary H7) will increase over the period 1993 to 2002, particularly in the period of economic disturbance.

To assess the impact of the crisis period on financial reporting practices, we investigate whether there is a significant association between accounting practices and the years studied. We know that the accounting environment has changed over time and thus financial reporting practice may well follow those changes. However, in the year after the economic crisis (1998), the environment changed dramatically, more so than in other years. Based on environmental-determinism theory (Cooke and Wallace, 1990) and positive accounting theory (Watts and Zimmerman, 1986) we hypothesize that:

H8: There will be a significant association between measurement methods and the years studied and the degree of significance in the year 1998 will be more than those of other years.

H9: There will be a significant association between disclosure levels and the years studied and the degree of significance in the year 1998 will be more than those of other years.

Various firm characteristics have been identified to indicate managerial incentives in preparing financial reports. These characteristics include firm size (Craig and Diga, 1998; Camfferman and Cooke, 2002; Morris et al., 2004; Iatridis, 2006), leverage ratio (Craig and Diga, 1998; Wallace et al., 1999; Camfferman and Cooke, 2002; Iatridis, 2006), ownership structure (Chau and Gray, 2002; Morris et al., 2004; Ali et al., 2007), corporate governance (Haniffa and Cooke, 2002), industry classification (Cooke, 1992; Camfferman and Cooke, 2002; Haniffa and Cooke, 2002; Hope, 2003), Big 6 audit firms (Chalmers and Godfrey, 2004), listing status (Haniffa and Cooke, 2002; Hope, 2003), and profitability ratio (Camfferman and Cooke, 2002; Morris et al., 2004; Iatridis, 2006). These variables may help to explain variability in financial reporting practice but we were restricted by the availability of data. There is considerable discussion of relevant variables in the literature so we would refer to the above rather than provide the detail here. The available variables we include as having a potentially significant relationship with disclosure are: Firm size, H10; Leverage, H11; Profitability, H12; Industry type, H13; Auditor type, H14; Managerial ownership, H15; Ownership concentration, H16. We acknowledge that there may be an overlap between H15 and H16 to the extent that companies are run by inside shareholders such as family groups.

RESEARCH METHODOLOGY

Sample Selection and Responses

Sample companies had to be listed on the SET for the ten years from 1993 to 2002 so that changes in financial reporting practices could be observed. Survivorship bias is not thought to be a problem since we are not investigating company performance but instead looking at practice over time. In addition, companies were excluded, because of their potential distortionary impact, if they were a subsidiary of a foreign company, controlled by foreign companies, in the financial and insurance sectors, or in a rehabilitation plan (Cooke, 1989; Hossain et al., 1995; Singleton and Globerman, 2002). On this basis, 106 companies formed the sample, grouped into five industries: agribusiness (5 per cent), heavy industrial (29 per cent), consumer goods (29 per cent), services (30 per cent) and other sectors (7 per cent).

Overall, fifty-nine (55.7 per cent) provided accounts for each of the four years 1993, 1996, 1998 and 2002. A non-response bias test for disclosure practice was undertaken and the results show that there were no significant differences between the first and second responding groups in all four years (for voluntary disclosure), but there were significant differences in aggregate and mandatory disclosures in 2002. Therefore, non-response groups might have had an affect on the research results of aggregate and mandatory disclosures in 2002 if they had been included.

Research Instruments

The choice of measurement methods  Twenty-five annual reports were read to produce choices of measurement methods for four accounting issues previously identified. The following choices for each accounting issue were made: two alternative methods for each accounting issue along with ‘other methods’, ‘not disclosed’ and ‘not applicable’. The list of measurement methods is shown in the Appendix.

The development of a list of disclosure items  A list of disclosure items was developed using the following information: previous research, especially work on South-East Asian countries and developing countries; other related pronouncements and documents, such as the SET guidelines for listed companies' disclosure; and a companies' disclosure checklist used by one of the international auditing firms in Thailand. In addition, the Thai Civil and Commercial Code, the Ministerial Regulation No. 2, the Public Company Act, the SET announcements and TASs were reviewed in order to identify mandatory and voluntary disclosure items. The disclosure research instrument thereby established was then checked by four Thai accounting professionals in order to ensure relevance and comprehensiveness of the disclosure items list. Finally, twenty-five annual reports were read to check the applicability of the list for measuring the extent of disclosure. The final disclosure list consists of 266 disclosure items; however, it includes forty-nine items that changed from voluntary to mandatory during the period 1993 to 2002. In order to ensure that levels of disclosure in each year are comparable, these items are excluded. Therefore, the usable disclosure items list contains 217 items of which 170 are mandatory and 47 are voluntary throughout the period of assessment.

Data Collection

Data were collected from annual reports that contain both mandatory and voluntary information. Annual reports are one of the most important sources of information for investors' decision making (Chang et al., 1983; Abdelsalam, 1990; Ho and Wong; 2001) and are publicly available, accessible and used in similar research.

The number of companies using a particular measurement method was collected to assess measurement practices. For disclosure practices, an index was calculated as:

image

where nj is the number of relevant items for jth firm, and

xij is one if the ith item is disclosed or zero if the ith item is not disclosed so that 0 ≤Ij≤ 1.

Non-disclosure of financial reporting practices in annual reports exists. For measurement practices, Archer et al. (1995) and Parker and Morris (2001) have suggested that non-disclosure may mean either that management intends to conceal or that a company does not use that accounting method. For disclosure, non-disclosure may mean either that disclosure items are not relevant to a company or that management does not intend to disclose.

To deal with the non-disclosure problem, we read corporate annual reports in their entirety (Cooke, 1989; Chau and Gray, 2002; Camfferman and Cooke, 2002). This thorough reading helps to indicate whether non-disclosure is a result of non-applicability of the methods or management intention to conceal.

Research Methods

We used the chi-square test to assess measurement practices over time and also in the three sub-periods (pre-economic crisis, economic crisis and post-economic crisis). For disclosure, the Friedman test was used for testing practices over the period 1993 to 2002, along with the paired sample t-test and Wilcoxon sign-ranked test.

Ordinary least square (OLS) analysis was also undertaken to assess the relationship between firm characteristics and the disclosure index. The first regression model considers an association between firm characteristics and financial reporting practices to assess management behaviour in preparing financial reports. The second model includes dummy variables for the four years studied in order to investigate whether a point in time is an explanatory factor. The regression models are:

image
image
  • where Discjt= disclosure index,

  • β0 is the intercept,

  • Sizejt is firm size defined as total sales,

  • Levjt is leverage ratio defined as total debt to total assets,

  • Prosjt is profitability ratio defined as net income to total assets,

  • Ownj is (ownership) proportion of ordinary shares held by top ten shareholders,

  • Mgrj is (managerial ownership) proportion of ordinary shares held by management,

  • Auditorj is one if a company is audited by Big 4 and zero if otherwise,

  • Agrij is one if a company is in an agribusiness sector and zero if otherwise,

  • Indusj is one if a company is in a heavy industrial sector and zero if otherwise,

  • Consj is one if a company is in a consumer goods sector and zero if otherwise,

  • Servj is one if a company is in a service sector and zero if otherwise,

  • Othersj is one if a company is in other sectors and zero if otherwise,

  • Y1993 is year 1993,Y1996 is year 1996,Y1998 is year 1998,Y2002 is year 2002, and

  • εj is the residual term.

Before using both the t-test and the regression analysis, we tested the assumptions for a parametric test. On the basis of tests for kurtosis and skewness we concluded that the independent variables were non-normal. As such, we used the normal scores approach (van der Waerden method) to transform the data and then used the transformed variables in the regression equation, a method proposed by Cooke (1998).

RESULTS AND DISCUSSION

Descriptive Statistics

Measurement practices  Table 1 shows the frequency distribution of the number of companies for each accounting issue. The table shows that there were few changes over time in the number of companies using particular measurement methods. There are two main methods applied widely among the sample companies for inventory valuation: FIFO and weighted average methods. The choice between the two methods is almost equally divided in each year studied. The main measurement method in accounting for investments is the weighted average method whereas a few companies apply other methods (e.g. the identification method) over time. It is also noticeable that the number of companies disclosing their accounting method increases over time.

Table 1.
FREQUENCY DISTRIBUTION OF CHOICES OF ACCOUNTING METHODS AND TESTS OF DIFFERENCES
Year 1993 1996 1998 2002
Accounting for inventory
 FIFO 20 22 26 22
 Weighted average 25 27 26 31
 Other methods 6 4 3 2
 Not disclosed 3 2 2 2
 Not applicable 5 4 2 2
 Total 59 59 59 59
 Pearson χ2 (asym. Sig.) = 0.426 (1-sided)
Accounting for investment
 Weighted average 10 10 14 27
 Other methods 3 3 3 3
 Not disclosed 24 27 25 16
 Not applicable 22 19 17 13
 Total 59 59 59 59
Pearson χ2 (asym. Sig.) = 0.312 (1-sided)
Accounting for borrowing costs
 Recognized as expenses 2 1 1 0
 Recognized as assets 19 19 20 23
 Not disclosed 25 26 23 22
 Not applicable 13 13 15 14
 Total 59 59 59 59
 Pearson χ2 (asym. sig.) = 0.264 (1-sided)
Depreciation
 Straightline 57 57 59 59
 Diminishing 2 2 0 0
 Total 59 59 59 59
 Pearson χ2 (asym. sig.) = 0.127 (1-sided)
  • Note: The items ‘not disclosed’ and ‘not applicable’ are not included in the chi-square test.

With respect to accounting for borrowing costs, the majority of companies in the sample use the asset-recognition method and by 2002 no company used the expense-recognition method. For depreciation, the straight-line method is a dominant measurement method. Table 1 shows that only two companies applied other methods before the 1997 economic crisis but subsequently all companies used the straight-line method. The number of non-disclosers is high and the consequences are discussed later.

Disclosure practices  Table 2 shows mandatory, voluntary and aggregate disclosures and reveals that the average level of disclosure increased over the period 1993 to 2002. A considerable increase in the disclosure indexes occurred in the economic crisis period, as is evident from the maximum and minimum disclosure levels. The minimum level of aggregate disclosure increases from 39 per cent to 47 per cent and the mean rises from 50 per cent to 58 per cent. In addition, Table 2 shows that mandatory disclosure is the major contributor to the significantly increasing levels of aggregate disclosure, particularly during the economic crisis period, reflecting a response to the economic disturbance. Most changes in voluntary disclosure seem to occur in the post-economic crisis period, suggesting that the increasing level of voluntary disclosure may have contributed to aggregate disclosure in the post-economic crisis period more than in other periods.

Table 2.
LEVEL OF DISCLOSURE BY THAI LISTED COMPANIES
Year M Min. Max. SD Skewness Kurtosis K-S Lilliefors (Sig.)
Mandatory disclosure
1993 0.64 0.46 0.82 0.07 −0.01 0.05 0.200
1996 0.69 0.58 0.84 0.05 0.15 −0.02 0.200
1998 0.78 0.65 0.87 0.05 −0.43 −0.38 0.166
2002 0.82 0.71 0.90 0.05 −0.24 −0.57 0.200
Voluntary disclosure
1993 0.08 0.00 0.25 0.06 1.08 0.58 0.000
1996 0.07 0.00 0.20 0.05 1.08 0.23 0.000
1998 0.10 0.00 0.29 0.06 0.94 0.91 0.001
2002 0.18 0.05 0.40 0.08 0.49 −0.07 0.056
Aggregate disclosure
1993 0.46 0.33 0.58 0.06 0.09 −0.50 0.200
1996 0.50 0.39 0.61 0.05 0.06 −0.22 0.200
1998 0.58 0.47 0.67 0.05 −0.46 −0.38 0.002
2002 0.64 0.50 0.74 0.05 −0.29 0.30 0.200

Test of Differences in Financial Reporting Practices

Financial reporting measurement practice  Table 1 shows that some frequencies are less than five. To improve the power of the test we adopted the approach used by Parker and Morris (2001) who applied the Cochran's rule (1952 and 1954) by using the chi-square test with and without combining categories and then comparing these two results to find whether the two sets of data generate the same decision. If the data do not satisfy the Cochran's rule we use the Fisher's exact test. We calculate the chi-square tests across all years by using two sets of data.

Table 1 shows no significant values across all years for all accounting issues tested. When we combined the categories, the same results occur. Where the Cochran's rule was replaced by the Fisher's exact test the results are the same, suggesting no association between measurement practices and the period of time tested. Therefore, we can conclude that the proportion of companies using one of the measurement methods is not significantly different to the proportion of those using another method in each of the years studied. As such, we can reject H1–H4.

Financial reporting disclosure practice  Table 3 reports the results of the Friedman test, showing significant differences between the four years studied for aggregate, mandatory and voluntary disclosure indexes at the 0.01 level. Thus, over the ten years studied, there was a statistically significant increase in the levels of disclosure by Thai listed companies.

Table 3.
TEST OF DIFFERENCES IN LEVELS OF DISCLOSURE FROM 1993 TO 2002 USING THE FRIEDMAN TEST
Type of disclosure N χ2 df Asymp. sig.
Aggregate disclosure 59 175.82 3 0.000
Mandatory disclosure 59 175.543 3 0.000
Voluntary disclosure 59 82.58 3 0.000

Table 4 shows the problem of non-normal data for two-year testing and as a consequence the paired sample t-test is supported by the Wilcoxon sign-ranked non-parametric test. There are significant differences (1 per cent confidence level) in the levels of aggregate and mandatory disclosure in all three periods of time (1993–96, 1996–98, 1998–2002). Also, the results show that the period 1996 to 1998 has the biggest change in the disclosure index (highest t-values and highest Z-values from the Wilcoxon test). The results indicate that disclosure increased significantly over the economic crisis period. Furthermore, the similar patterns of increase in aggregate and mandatory disclosures support the view that mandatory disclosure is the main reason for increasing levels of aggregate disclosure.

Table 4.
PAIRED SAMPLE t TEST OF DIFFERENCES IN LEVELS OF THREE TYPES OF DISCLOSURE
Period Mean diff. SD t df Sig. (2-tailed) Wilcoxon test
Z Asymp. sig.
Aggregate disclosure
1993–96 0.03577 0.0391 7.0222 58 0.000 −5.3515 0.000
1996–98 0.08436 0.0418 15.4982 58 0.000 −6.6422 0.000
1998–2002 0.05826 0.0438 10.2243 58 0.000 −6.3478 0.000
Mandatory disclosure
1993–96 0.04381 0.0512 6.5757 58 0.000 −5.2232 0.000
1996–98 0.08761 0.0523 12.8738 58 0.000 −6.5818 0.000
1998–2002 0.04333 0.0513 6.4910 58 0.000 −5.0949 0.000
Voluntary disclosure
1993–96 −0.00831 0.0450 −1.418 58 0.162 −1.0744 0.283
1996–98 0.03610 0.0494 5.616 58 0.000 −4.8806 0.000
1998–2002 0.07271 0.0885 6.454 58 0.000 −5.1990 0.000

Statistically significant differences in the levels of voluntary disclosure were found in the economic crisis period and post-economic crisis period but not in the pre-economic crisis. We found that the post-economic crisis period has a higher t-value (6.454) than the economic crisis period. This implies that voluntary disclosure also contributed to overall increases in disclosure in the post-economic crisis period.

The results support H5 and H6. They also support the hypothesis that the level of voluntary disclosure changed over the period 1993 to 2002 but the biggest increase was not the period 1996 to 1998.

Multiple Regression Analysis

While measurement practices have not changed significantly disclosure has, particularly in the period of economic disturbance As such multiple regression analysis is undertaken to test various firm-specific characteristics to establish whether they help explain variability in disclosure scores. Because of the small sample size each year, we pooled the data and segregated it into two groups: before the crisis (pooling data for the years 1993 and 1996); and after the crisis (pooling data for the years 1998 and 2002). Unfortunately, before the economic crisis period, only seven of the sample companies disclose information about ownership structure. Due to this small sample size the ownership concentration and managerial ownership variables could not be included in the ‘before the crisis’ regression model but are included in the ‘after the crisis’ model. The regression results are presented in Table 5.

Table 5.
MULTIPLE REGRESSION ANALYSIS
Regression models—aggregate disclosure
Variables Before the crisis (Model A-1) After the crisis (Model B-1) After the crisis (Model C-1)
Data set 1 Data set 2 Data set 1 Data set 2 Data set 1 Data set 2
B (t-value) B (t-value) B (t-value) B (t-value) B (t-value) B (t-value)
Constant −0.007 −0.131 −0.331 −0.337 −0.199 −0.294
(−0.046) (−0.907) (−1.638) (−1.956) (−1.151) (−1.939)
Size 0.360 0.432 0.135 0.238 0.211 0.367
(3.668***) (4.442***) (1.061) (2.182**) (1.985*) (3.946***)
Leverage ratio 0.156 0.193 0.105 0.217 0.037 0.167
(−1.529) (1.918*) (0.765) (1.846*) (0.348) (1.807*)
Profitability ratio −0.229 −0.060 0.097 0.146 0.092 0.111
(−2.6678***) (−0.707) (0.819) (1.446) (0.967) (1.3333)
Own. concentration −0.277 −0.341
(−2.286**) (−3.297***)
Mgr. ownership −0.116 −0.312
(−0.825) (−2.605**)
Auditor type 0.086 0.165 0.454 0.323 0.431 0.410
(0.524) (1.013) (1.88*) (1.564) (2.301**) (2.495**)
Industry type
 Agribusiness −0.374 −0.004 −0.392 −0.133 −0.471 −0.006
(−1.925*) (−0.004) (−0.671) (−0.266) (−1.17) (−0.016)
 Cons. goods 0.289 0.155 0.128 0.262 −0.032 0.208
(1.476) (−0.020) (0.484) (1.156) (−0.147) (1.082)
 Services 0.400 0.280 −0.122 −0.081 −0.165 −0.108
(1.482) (0.802) (−0.438) (−0.342) (−0.767) (−0.57)
 Others −0.858 −0.001 1.018 0.815 0.696 0.645
(−2.404**) (1.050) (2.876***) (2.693***) (2.353**) (2.488**)
Model summary
 Adj. R2 0.312 0.323 0.221 0.444 0.148 0.344
F value 7.620 7.972 3.274 7.401 3.550 8.684
 Sig. F value 0.000 0.000 0.002 0.000 0.001 0.000
Sample size 118 118 81 81 118 118
Regression models—mandatory disclosure
Variables Before the crisis (Model A-2) After the crisis (Model B-2) After the crisis (Model C-2)
Data set 1 Data set 2 Data set 1 Data set 2 Data set 1 Data set 2
B (t-value) B (t-value) B (t-value) B (t-value) B (t-value) B (t-value)
Constant 0.073 −0.099 −0.097 −0.239 0.026 −0.217
(0.479) (−0.670) (−0.482) (−1.408) (0.144) (−1.412)
Size 0.088 0.398 0.036 0.226 0.019 0.347
(0.851) (3.995***) (0.28) (2.11**) (0.172) (3.694***)
Leverage ratio 0.080 0.191 0.048 0.263 −0.050 0.211
(0.747) (1.846*) (0.349) (2.273**) (−0.461) (2.263**)
Profitability ratio −0.291 −0.040 0.137 0.183 0.112 0.141
(−3.234***) (−0.459) (1.157) (1.846*) (1.139) (1.674*)
Own. concentration −0.202 −0.353
(−1.671*) (−3.462***)
Mgr. ownership 0.057 −0.311
(0.41) (−2.642**)
Auditor type 0.027 0.101 0.367 0.253 0.306 0.359
(0.155) (0.607) (1.522) (1.247) (1.581) (2.164**)
Industry type
 Agribusiness −1.215 −0.023 −0.715 −0.109 −1.056 −0.064
(−3.252***) (−0.065) (−1.225) (−0.222) (−2.541**) (−0.18)
 Cons. goods −0.628 −0.065 −0.260 0.133 −0.362 0.113
(−3.088***) (−0.327) (−0.98) (0.599) (−1.598) (0.584)
 Services 0.314 0.103 −0.097 −0.135 −0.238 −0.161
(1.531) (0.521) (−0.349) (−0.579) (−1.069) (−0.844)
 Others 0.638 0.520 0.639 0.661 0.536 0.583
(2.257**) (1.899*) (1.803*) (2.222**) (1.755*) (2.226**)
Model summary
 Adj. R2 0.246 0.290 0.090 0.445 0.094 0.330
F value 5.777 6.976 1.792 7.423 2.513 8.215
 Sig. F value 0.000 0.000 0.078 0.000 0.015 0.000
Sample size 118 118 81 81 118 118
Regression models—voluntary disclosure
Variables Before the crisis (Model A-3) After the crisis (Model B-3) After the crisis (Model C-3)
Data set 1 Data set 2 Data set 1 Data set 2 Data set 1 Data set 2
B (t-value) B (t-value) B (t-value) B (t-value) B (t-value) B (t-value)
Constant −0.194 −0.384 −0.558 −0.591 −0.360 −0.384
(−1.209) (−2.185) (−2.628) (−2.795) (−2.012) (−2.185)
Size 0.332 0.265 0.122 0.136 0.248 0.265
(3.077***) (2.46**) (0.908) (1.021) (2.265**) (2.46**)
Leverage ratio 0.050 −0.074 −0.023 −0.008 −0.077 −0.074
(0.449) (−0.696) (−0.161) (−0.054) (−0.703) (−0.696)
Profitability ratio −0.054 −0.061 −0.043 −0.029 −0.061 −0.061
(−0.575) (−0.632) (−0.349) (−0.235) (−0.623) (−0.632)
Own. concentration −0.090 −0.125
(−0.706) (−0.983)
Mgr. ownership −0.100 −0.100
(−0.678) (−0.681)
Auditor type 0.072 0.298 0.412 0.422 0.287 0.298
(0.397) (1.566) (1.621) (1.668) (1.481) (1.566)
Industry type
 Agribusiness −0.045 0.329 0.127 0.145 0.322 0.329
(−0.115) (0.804) (0.207) (0.237) (0.774) (0.804)
 Cons. goods 0.249 0.455 0.675 0.731 0.398 0.455
(1.164) (2.045**) (2.421**) (2.633**) (1.755*) (2.045**)
 Services 0.399 0.202 0.140 0.143 0.195 0.202
(1.853*) (0.924) (0.477) (0.492) (0.878) (0.924)
 Others −0.120 0.295 0.839 0.810 0.299 0.295
(−0.404) (0.983) (2.252**) (2.182**) (0.978) (0.983)
Model summary
 Adj. R2 0.117 0.101 0.110 0.135 0.077 0.101
F value 2.939 2.652 1.985 2.254 2.225 2.652
 Sig. F value 0.005 0.011 0.048 0.024 0.031 0.011
Sample size 118 118 81 81 118 118
  • ***  Significant at the 1% level,
  • **  ** Significant at the 5% level,
  • *  * Significant at the 10% level.
  • Data set 1 is a set of disclosure scores that does not include the non-applicable disclosure items in the calculation.
  • Data set 2 is a set of disclosure scores that include the non-applicable disclosure items in the calculation.

The adjusted R2 for all regression models is relatively low (compare Craig and Diga, 1998; Chau and Gray, 2002; Haniffa and Cooke, 2002). Although low values of the adjusted R2 are not uncommon in disclosure research (see Owusu-Ansah, 1998; Morris et al., 2004), the low values for the adjusted R2 and variability of the results over time suggest that the model's ability to generalize is somewhat limited (Field, 2004). Low values may be caused by inconsistent disclosure scoring for the non-applicable disclosure items. To test this we assume that every item in the disclosure checklist is applicable to every company to establish a further modified dataset. Thus, sample companies are penalized if they do not disclose information. We then ran regressions using these revised disclosure scores. The results are shown in Table 5, as Data set 2, compared with the results from the first set of data (Data set 1).

The values of adjusted R2 are higher than the first data set of disclosure scores, both before and after the economic crisis. The F values show the significance of the model in both time periods. The regression models also show that the size and leverage variables are consistently associated with aggregate disclosure over time. Furthermore, after the economic crisis, the regression model shows that ownership concentration and managerial ownership are negatively and significantly associated with aggregate disclosure, suggesting that the higher the percentage of ownership concentration and managerial ownership, the lower the extent of disclosure. The ownership structure seems to be a significant factor in explaining variability in disclosure levels. As can be seen in Table 5, the values of the adjusted R2 in each regression model are higher when ownership structure is included.

When we regressed firm-specific characteristics on mandatory disclosure using Data set 2 we also found similar results to aggregate disclosure. That is, mandatory disclosure level is significantly associated with firm size, leverage ratio, ownership concentration and managerial ownership.

For voluntary disclosure, before the economic crisis, the results show a statistically significant association between firm size and disclosure at the 1 per cent level. However, after the crisis there were no significant relationships between voluntary disclosure and the independent variables, except for industry type. This may be caused by either multicollinearity or the small sample size of eighty-one companies. However, the VIF and Tolerance values show no serious concerns for multicollinearity between the variables. Therefore, the sample size may have influenced the significance value of the size variable in the regression. We tested this assumption by omitting the ownership concentration and managerial ownership variables from the model to obtain a sample of 118 companies. The results show that voluntary disclosure is significantly related to both size and industry type at the 5 per cent level.

In addition, we pooled the data for all years together and ran regressions in order to find whether time has an association with disclosure levels. Table 6 shows the dramatic increase in the values of adjusted R2 after the year dummy variables are included. The adjusted R2 values increase to around 0.60 for aggregate and mandatory disclosure models and to about 0.40 for the voluntary disclosure model. This suggests that the year variables are influential regressors.

Table 6.
MULTIPLE REGRESSION ANALYSIS-WITH YEAR VARIABLE
Pooled data aggregate disclosure
Variables Data set 1 (all variables) Data set 2 (all variables) Data set 1 (no own.) Data set 2 (no own.)
B t B t B t B t
Constant −0.299 −1.370 −0.320 −1.403 −0.976 −10.491 −0.979 −10.248
Size −0.007 −0.093 0.097 1.246 0.121 2.683*** 0.235 5.054***
Leverage ratio 0.141 1.992** 0.235 3.180*** 0.122 2.668*** 0.206 4.379***
Profitability ratio 0.013 0.216 0.062 0.985 −0.016 −0.414 0.043 1.060
Own. concentration −0.208 −2.806*** −0.279 −3.595***
Mgr. ownership −0.171 −2.174** −0.290 −3.521***
Auditor type 0.134 0.975 0.066 0.461 0.186 2.421** 0.211 2.684***
Industry type
 Agribusiness −0.306 −0.908 −0.089 −0.253 −0.393 −2.366** 0.084 0.495
 Cons. goods 0.011 0.071 0.091 0.582 −0.094 −1.045 0.129 1.388
 Services 0.000 0.001 −0.023 −0.139 0.042 0.469 0.035 0.379
 Others 0.633 3.169*** 0.478 2.293** 0.361 2.905*** 0.325 2.552**
Years
 1996 0.399 4.014*** 0.382 3.743***
 1998 0.524 2.228** 0.505 2.053** 1.229 12.466*** 1.120 11.069***
 2002 1.267 5.358*** 1.186 4.801*** 1.960 19.141*** 1.724 16.398***
Model summary
 Adj. R2 0.505 0.581 0.713 0.697
F value 8.398 11.073 54.086 50.129
 Sig. F value 0.000 0.000 0.000 0.000
Sample size 88 88 236 236
Pooled data mandatory disclosure
Variables Data set 1 (all variables) Data set 2 (all variables) Data set 1 (no own. ) Data set 2 (no own.)
B t B t B t B t
Constant −0.149 −0.594 −0.266 −1.125 −0.896 −8.755 −0.969 −9.527
Size −0.067 −0.785 0.098 1.219 −0.020 −0.400 0.214 4.329***
Leverage ratio 0.101 1.244 0.246 3.2067*** 0.058 1.145 0.223 4.4578***
Profitability ratio 0.047 0.684 0.084 1.296 −0.014 −0.324 0.077 1.781*
Own. concentration −0.139 −1.637 −0.267 −3.322***
Mgr. ownership −0.033 −0.365 −0.274 −3.210***
Auditor type 0.122 0.776 0.037 0.249 0.119 1.409 0.178 2.124**
Industry type
 Agribusiness −0.484 −1.255 −0.078 −0.213 −0.739 −4.047*** 0.043 0.240
 Cons. goods −0.237 −1.384 −0.033 −0.206 −0.328 −3.295*** 0.054 0.545
 Services −0.008 −0.046 −0.096 −0.553 0.006 0.058 −0.034 −0.349
 Others 0.459 2.006** 0.357 1.653 0.355 2.600** 0.421 3.102***
Years
 1996 0.471 4.305*** 0.501 4.613***
 1998 0.585 2.171** 0.570 2.237** 1.354 12.493*** 1.186 11.008***
 2002 1.125 4.153*** 1.130 4.416*** 1.939 17.215*** 1.693 15.129***
Model summary
 Adj. R2 0.293 0.528 0.653 0.657
F value 4.010 9.109 41.185 41.905
 Sig. F value 0.000 0.000 0.000 0.000
Sample size 88 88 236 236
Pooled data voluntary disclosure
Variables Data set 1 (all variables) Data set 2 (all variables) Data set 1 (no own.) Data set 2 (no own.)
B t B t B t B t
Constant −0.472 −1.539 −0.467 −1.557 −0.566 −4.220 −0.585 −4.452
Size −0.029 −0.276 −0.024 −0.240 0.210 3.221*** 0.218 3.418***
Leverage ratio 0.127 1.276 0.142 1.456 0.063 0.954 0.065 0.999
Profitability ratio −0.060 −0.705 −0.056 −0.678 −0.075 −1.322 −0.078 −1.400
Own. Concentration −0.127 −1.217 −0.157 −1.541
Mgr. ownership −0.208 −1.881* −0.201 −1.853*
Auditor type 0.174 0.900 0.194 1.025 0.139 1.255 0.150 1.385
Industry type
 Agribusiness 0.136 0.287 0.141 0.304 0.158 0.662 0.193 0.824
 Cons. goods 0.582 2.771*** 0.623 3.033*** 0.300 2.306** 0.330 2.582**
 Services 0.224 0.996 0.214 0.969 0.293 2.270** 0.298 2.349**
 Others 0.598 2.133** 0.565 2.060** 0.069 0.385 0.055 0.311
Years
 1996 −0.175 −1.219 −0.166 −1.183
 1998 0.011 0.032 −0.017 −0.052 0.334 2.353** 0.345 2.480**
 2002 0.988 2.975*** 0.961 2.959*** 1.132 7.676*** 1.134 7.836***
Model summary
 Adj. R2 0.373 0.399 0.385 0.402
F value 5.317 5.809 14.346 15.367
 Sig. F value 0.000 0.000 0.000 0.000
Sample size 88 88 236 236
  • ***  Significant at the 1% level,
  • **  ** Significant at the 5% level,
  • *  * Significant at the 10% level.
  • Data set 1 is a set of disclosure scores that does not include the non-applicable disclosure items in the calculation.
  • Data set 2 is a set of disclosure scores that include the non-applicable disclosure items in the calculation.

For the all-variables models, the data for 1993 were omitted because of missing ownership data. Thus, only data for the years 1996, 1998 and 2002 were included in the regressions. The year 1996 variable is used as the base category and is the excluded dummy time variable to avoid perfect collinearity. The regression results show that 1998 and 2002 have higher disclosure levels than 1996. Apart from the year variable, the other variables significantly associated with disclosure are leverage, ownership concentration and managerial ownership, which is consistent with the results in Table 6. Also, the leverage ratio has a positive relationship, suggesting the higher the ratio the higher the disclosure level. On the other hand, the ownership concentration and managerial ownership have negative associations, indicating that the higher the concentration in ownership concentration and managerial ownership the lower the extent of disclosure.

It is noticeable that the size variable is not statistically significant, possibly because of the small sample size (eighty-eight companies). We obtained a larger sample size of 236 observations by excluding the ownership variables from the regression models and pooling the data for each of the four years (1993, 1996, 1998 and 2002). Here the 1993 year dummy is subsumed in the regression constant as it acts as the base category. The regression models show high values for the adjusted R2 and the size variable has a consistent association with the disclosure indexes for all types of disclosure in a similar manner to the results presented in Table 5. The results also reveal that leverage ratio and auditor type are significantly associated with aggregate and mandatory disclosure levels.

CONCLUSIONS

We have shown that Thai financial reporting disclosure practices have changed over the period 1993 to 2002 although measurement practices have not changed significantly. In addition to the substantial increase in disclosure over time, our statistical analysis indicates that changes in disclosure occurred mainly in the economic crisis period (1996–98), suggesting that the economic disturbance had an impact on disclosure practices. The research further investigated the determinants of financial reporting disclosure by dividing the periods studied into two: before and after the 1997 economic crisis. The regression analysis reveals that firm size is consistently associated with all types of disclosure. Other factors, including leverage ratio, ownership concentration, managerial ownership and auditor type, are significantly related to disclosure practice. In addition, the extent of disclosure varies over time.

The results imply that the economic crisis has had a rather limited impact on Thai measurement practices, suggesting that changes over time are somewhat ‘sticky’. In contrast, changes in disclosure practices suggest a substantial influence of the economic crisis, mainly as a result of changes in regulations. Apart from regulatory forces, contractual motivations and political costs incentives seem to have played their part on voluntary disclosures. An implication for other West Pacific Rim countries affected by the economic crisis is that if they want to develop their financial reporting practices to be accepted internationally then this must come from regulatory forces.

Appendix

accounting policy choices

Accounting issues Years
1993 1996 1998 2002
Inventory-cost measurement
 1 First-in first-out
 2 Weighted average
 3 Last-in first-out
 4 Identification method
 5 Other methods
 6 Not disclosed
 7 Not applicable
Accounting treatment for borrowing costs
 1 Borrowing costs are recognized as expenses
 2 Borrowing costs are capitalized as the cost of assets
 3 Other methods
 4 Not disclosed
 5 Not applicable
Depreciation methods
 1 Straight line method
 2 Diminishing balance
 3 Other methods
 4 Not disclosed
 5 Not applicable
Accounting for investment-cost of sale of securities
 1 First-in first-out
 2 Weighted average
 3 Other methods
 4 Not disclosed
 5 Not applicable

Footnotes

  • 1 Some argue against such developments. For example, Porter (1999) has identified a number of sources of authority that he believes are undermining democratic institutions. This includes supranational authority, the migration of policy making capacity to international institutions such as the International Accounting Standards Board. In addition Ball et al. (2003, p. 259) argue that ‘we are not implying that international institutional integration is desirable. Ball (1995) argues that it is undesirable. Our point is that there would be an enormous amount of economic, legal and political change required to make the demands placed on accounting information identical across nations, and to give managers and accountants identical reporting incentives worldwide. Such fundamental and widespread change is unlikely to occur, so homogenizing accounting standards alone will have a limited effect on the quality of accounting information actually reported.’
  • 2 See, for example, Mueller (1968), Radebaugh (1975), Meek and Saudagaran (1990), Iqbal et al. (1997), Radebaugh and Gray (1997), Nobes (1998), Choi et al. (1999), Hope (2003) and Nobes and Parker (2007).
  • 3 See, for example, Radebaugh (1975), Hofstede (1980, 1984), Gray (1988), Perera (1989), Baydoun and Willett (1995), Zarzeski (1996), Nobes (1998) and Haniffa and Cooke (2002).
  • 4 See, for example, Nobes (1998), Chanchani and MacGregor (1999), McSweeney (2002) and Baskerville (2003).
  • 5 Information about NPL and bank recapitalization can be found in Economic Monitors issued by the World Bank. The document is available at http://www.worldbank.or.th.
  • 6 See, for example, Priebjrivat (1992), Gray et al. (1995), Hossain et al. (1995), Craig and Diga (1998), Curuk (1999), Chau and Gray (2002) and Haniffa and Cooke (2002).
  • 7 The list of disclosure items is available upon request.
  • 8 See, for example, Cooke and Wallace (1990), Hossain et al. (1995), Curuk (1999) and Haniffa and Cooke (2002).
  • 9 Previous studies (e.g., Camfferman and Cooke, 2002; Haniffa and Cooke, 2002; Leventis and Weetman, 2004) also used this method in transforming the data to satisfy the normality assumption before running the regression analysis.
    • The full text of this article hosted at iucr.org is unavailable due to technical difficulties.