Dr. Stephen Owusu-Ansah Assistant Professor of Accounting Department of Accounting & MIS King Fahd University of Petroleum & Minerals P. O. Box 809 Dhahran 31261 Saudi Arabia Tel: + 966 0(3) 860-4285 Fax: + 966 0(3) 860-3489 E-mail: stephen@kfupm.edu.sa

(Forthcoming in Accounting & Business Research, Vol. 30, No. 3, Summer 2000)

__________ I acknowledge the hospitality and the continued support of the officers of the Zimbabwe Stock Exchange, the publisher, and the librarian of The Financial Gazette in Harare, Zimbabwe. I am grateful for the travel grant awarded by the Wincott Foundation, England to enable me to pay a study visit to the Zimbabwe Stock Exchange, and for the use of the facilities of King Fahd University of Petroleum & Minerals, Saudi Arabia. I am indebted to R. S. Olusegun Wallace, Donal McKillop, Ajay Adhikari, Joseph Abeka and Yakubu Umar for their helpful comments and suggestions. I also thank the two anonymous reviewers of the previous versions of this article and the editor, Ken Peasnell, for their constructive criticisms and suggestions. Comments from the participants of the staff seminar of the Commerce Divisions of the University of Canterbury and Lincoln University, New Zealand are acknowledged. All remaining errors are, however, mine.

ABSTRACT This article reports on the results of an empirical investigation of the timeliness of annual reporting by 47 non-financial companies listed on the Zimbabwe Stock Exchange. It also reports on the factors that affect timely reporting by these companies. The results of a descriptive analysis indicate that 98 percent of the companies in the sample reported promptly to the public (i.e., submitted their audited annual reports to the Zimbabwe Stock Exchange by the regulatory deadline). A two-stage least squares regression identified company size, profitability and company age as statistically significant explanators of the differences in the timeliness of annual reports issued by the sample companies. No evidence was found to support the monitoring costs theory argument which suggests that highly-geared companies are timely reporters. Furthermore, the empirical data indicate that audit reporting lead time is significantly associated with the timeliness with which sample companies release their preliminary annual earnings announcement, but not with the timeliness of their audited annual reports. Plausible explanations for these findings along with the limitations of the underlying research are provided.

1. Introduction Timely reporting contributes to the prompt and efficient performance of stock markets in their pricing and evaluation functions. Timely reporting helps to mitigate (or reduce the level of ) insider trading, leaks and rumours in the market. As a result, most stock exchanges in the world including the Zimbabwe Stock Exchange (ZSE), demand from their listed companies prompt release of audited financial reports to the markets. While several studies have investigated the informational

‘efficiency’ of emerging stock markets, there is virtually no systematic analysis of both audit-related and corporate-specific factors likely to affect the timeliness of corporate financial reporting in these markets. Thus, researchers have been pre-occupied with the effect of released information on prices of securities traded on emerging markets, and not with the timeliness of the information and the factors that affect timely reporting. Research on the latter may improve our understanding of whether information released in emerging markets is timely, and the factors that account for this phenomenon. This article presents such an analysis by focusing on the experience of non-financial companies listed on the ZSE. For several reasons, the ZSE serves as a useful context for investigating the characteristics of financial reporting in emerging capital markets. First, because some of the securities traded on the ZSE are also constituent stocks of the International Finance Corporation’s Global and Investible Indices (i.e., the ZSE is a typical emerging stock market); international investors may be interested in the timely reporting profile of its listed companies. Second, the ZSE is second oldest stock market in Africa (second to the Johannesburg Stock Exchange in South Africa), but little is known about it. Third, Zimbabwe is an important country in Africa in that its professional accountancy

body presently serves (in conjunction with South Africa) on the board of the International Accounting Standards Committee. a general requirement of timely reporting is imposed by two regulatory sources in Zimbabwe: (i) the 1952 Companies Act (Chapter 190 of Zimbabwe Laws). The construct. section seven presents the conclusions and the limitations of the study. The regulatory framework for timely reporting in Zimbabwe To secure immediate release of information which might materially affect market activities and prices of listed securities. Both prior analytical and empirical evidence provide strong support for this contention. Stale information is of little use to market participants in their investment decision-making processes. Empirical studies have also shown that timely information affects the prices of securities on the market (Givoly & Palmon. 1984. 2. Section five describes the sample selection criteria and the statistical tests employed. 1982. and all those that are entitled to attend a company’s annual general meeting (AGM) at . The Companies Act requires that audited annual accounts be sent to shareholders. and its operationalisation in this study are also discussed in this section. Timeliness in financial reporting is a significant characteristic of accounting information. and (ii) the ZSE Listing Agreement. timeliness of corporate reporting. Kross & Schroeder. Chambers & Penman. Feltham (1972) demonstrated analytically that timely information affects a decision-maker’s expected pay-off. Section six reports and discusses the results of the statistical tests. Finally. The remainder of this article is arranged as follows: Section two describes the statutory and the institutional requirements relating to the timeliness of corporate financial reporting in Zimbabwe. Section three provides a brief overview of the relevant literature. 1984). Section four focuses on the development of testable hypotheses.

The Act obliges companies registered under it to hold their AGM within 26 weeks (182 days) immediately following the end of their financial year..g. send to shareholders) audited annual accounts within 24 weeks (168 days) after the close of their financial year. The second type is mainly concerned with the pattern. companies undertake to send their audited annual accounts to their shareholders at least 21 days prior to their AGM. reporting lag. As a condition for listing. imprisonment or both. Dyer & McHugh (1975) found that company size and . and total reporting lags over a six-year period to 1971. Chambers & Penman. In effect. the timely reporting requirement is somewhat more stringent. within 160 days after financial year-end). Further. 1975). Only the literature on the latter is reviewed here because the present study belongs to that category. listed companies are required to incorporate a provision to this effect in their articles of association. 1984).e.. and the factors influencing timely reporting behaviour (e.e.least 14 days before the date of the AGM. year-end closing date.. and profitability.. They studied preliminary. They examined the association between the reporting behaviour of 120 companies listed on the Sydney Stock Exchange in June 1971 and company size. 3. The directors of a company that fails to comply with these requirements are punished either by a fine. The first type is concerned with the impact of timely reporting on variability of stock returns (e. Dyer & McHugh. With regard to companies listed on the ZSE. and to submit three copies of their accounts to the chief executive of the ZSE at least 22 days before the dates of their AGM (i. auditors’ signature.g. Dyer & McHugh (1975) pioneered an empirical investigation into the timeliness of annual financial reporting of Australian companies. companies in Zimbabwe are supposed to publish (i. Review of the relevant literature The literature on the timeliness of corporate annual financial reporting is of two main types.

they tested three . however. on average. those in the mining and exploration and in the service industries were 1 slow reporters. number of pages of annual report. Courtis (1976) extended this line of research by examining the association between four additional corporate attributes and reporting delay of New Zealand listed companies. it would have been more appropriate for Courtis to have examined the activities and attributes of auditors instead. leading audit firms (representatives of the former Big-eight) take 53 days to sign audit reports as against 90 days for small audit firms in New Zealand. however. While supporting Gilling’s (1977) view that reporting lag is an outcome of the interaction between a company’s management and its auditors.year-end closing date were significantly associated with total lag. Commenting on Courtis’(1976) study. He found that whereas companies in the fuel and energy and in the finance industries were fast reporters. Accordingly. The attributes examined were company age. They. that neither approach sufficiently explains the reporting behaviour of companies. Courtis (1976) observed that the companies in his sample took almost 4 2 months beyond their financial year-end to report to shareholders. taken by New Zealand auditors to certify a client company’s accounts. and industry. and found that. Davies & Whittred (1980) argued. Gilling (1977) investigated the association between attributes of auditors and auditing lag. He attributed this lack of punctuality to the three months. 1980: 55). and the auditors who express opinion on them. Gilling suggested that since Courtis (1976) was concerned primarily with auditing lag. none of the attributes examined was significantly associated with auditors’ signature lag. on average. Gilling (1977) argued that timely reporting depends on both the management of the reporting company who prepares the accounts. Except for industry. They suggested that attributes of audit firms and those of client companies ‘which jointly determine the duration of yearend audit …’ should be examined (Davies & Whittred. number of shareholders. reported no statistically significant relationship between timely reporting and profitability.

They reported that structured audit approach was significantly associated with longer audit report lag. Using univariate analysis. For instance. (1993) investigated the multivariate relationships between structured audit technology. changes in auditing firm and the presence or otherwise of extraordinary items.. Garsombke (1981) found that there was a statistically significant difference in timely reporting between companies listed on either the New York Stock Exchange or the Over-the-Counter market. The other subsequent studies on North American countries while using sophisticated research designs. auditor-type. the type of audit firm. 1982.. Bamber et al. Their study also provides evidence on the association between company size. . listing status. namely.empirical indicants that capture year-end interaction between management and auditors. corporate-specific factors. management attitude and favourableness of reported earnings. Ashton et al. and those listed on the American Stock Exchange. and US exchange listing and the sample companies’ reporting lag within each country. Bamber et al. 1989 for Canadian companies). and timely reporting behaviour of 120 US companies. and Ashton et al. audit-related factors and timeliness of audit reports in the US. 1987. Garsombke (1981) extended this research paradigm to North America by examining the relationship between company size. In a comparative study by Frost & Pownall (1994). they found that both domestic and foreign companies listed in the US are more timely reporters than their counterparts listed in the UK. focused only on audit delay (see Givoly & Palmon.. month of financial year-end. The results of their statistical analyses indicated that the type of audit firm and the presence of extraordinary items are related to the reporting behaviour of the sampled 100 companies listed on Australian Associated Stock Exchange as of 31 December 1972 and 31 December 1977. 1993 for US companies.

This study investigates some of these factors that are relevant to the socio-economic conditions in Zimbabwe. While Ng & Tai’s (1994) study focused only on factors that affect audit delay in Hong Kong. company-specific factors are those that either enable management to produce a more timely annual report or reduce costs associated with undue delay in reporting. The present study contributes to the literature by providing such evidence. 4. . on timely reporting behaviour of companies in emerging economies. 1995: 3). and to use a simultaneous-equation estimation approach. gearing and company age. auditor-type was not included in this study because all the companies in the sample were audited by representatives of the international Big-five audit firms operating in Zimbabwe. as suggested by Davies & Whittred (1980). the month of financial year-end and the complexity of a company’s operations. 1998).1 To date. Company size (SIZE) 1 2 3 These studies are those conducted by Ng & Tai (1994) and Abdulla (1996). The audit service market in Zimbabwe is concentrated and dominated by the representatives of the Big-five (Owusu-Ansah. 4. it is the first study to examine timeliness of corporate reporting in an African country.. Audit-related factors are those that are likely to impede (or help) the auditor in carrying out the audit assignment and issuing the audit report promptly. no study has systematically examined the relative effects of both company-specific and auditrelated factors. Certain company-specific factors such as company size. For the audit-related factors. The present study is more comprehensive than the two studies because it investigates both corporate-specific and audit-related factors. and company-specific factors3 including company size. In reality. In contrast. the demarcation between company-specific factors and audit-related factors is not clear cut as suggested here. profitability and gearing can also be classified as auditrelated factors (see Simnett et al.1.Prior studies have focused primarily on developed markets with only two studies investigating timely reporting behaviour of companies listed on emerging stock markets. The impact of each factor on timeliness is theorised below. The audit-related factors2 include the presence of extra-ordinary and/or contingent items. Factors affecting timely reporting behaviour Timely reporting is a function of audit-related and company-specific factors. and for which data were available. Abdulla’s (1996) was concerned with corporate-specific determinants of timeliness in Bahrain. Also. profitability.

Frost & Pownall. Dyer & McHugh.. Audit delays are. they have more resources. hypothesised in the alternative form that: H1: Ceteris paribus. minimised and this enables the companies to report promptly to the public. large companies tend to have strong internal control systems with the consequence that auditors spend less time in conducting compliance and substantive tests. therefore. more accounting staff and sophisticated accounting information systems that result in more timely annual reports..g. large companies tend to be followed by a relatively large number of financial analysts who usually rely on timely release of annual reports to confirm and revise their expectations of companies’ present and future economic prospects. large companies are timely reporters for several reasons. the available empirical evidence suggests that large companies report on a more timely basis than their smaller counterparts (e. 1989. 1980. 1994). Ashton et al. large companies are able to install and operate modern computerassisted devices that are fast in processing and monitoring inventories and production operations. Profitability (PROFIT) . First. 1995). there is an incentive for such companies to report timely to avoid speculative trading in their shares.The timeliness of an annual report release is a function of the reporting company’s size. Finally. Company size is measured by the year-end total assets of each company as in prior studies (e. Simnett et al. Anecdotal evidence suggests that financial analysts interpret undue delay in reporting as an attempt to conceal information that may adversely affect the value of a company. Third. 1975.2. 4. It is. Usually. 1977.. reporting lead time is a decreasing function of a company’s size.. Davies & Whittred.g. Furthermore. Since large companies are generally followed by a relatively large number of analysts. Second. Gilling. therefore. The use of these devices results in a faster preparation of accounts and more timely annual reporting.

The opposite is true of a company with bad news (negative performance). auditors take much time to audit failing (high risk) companies as a defence against any potential future litigation. For instance. however.Profitability is expected to influence a company’s timely reporting behaviour. a company with good news (positive performance) will experience a rise in the market values of both its outstanding equity shares and management. Dye & Sridhar (1995) state that companies with successful results (good news) will report more promptly than those with failing operations or that have sustained losses (bad news). He observed that about 47 percent of the companies in his sample took four or more months following their financial year-end to report publicly. 1980. 1965. and for corporate managerial skills (Manne.3. The market for corporate managerial skills uses the performance of a company to set management’s outside opportunity wage. mixed. Fama. 4. For instance. in the alternative form. therefore. Gearing (GEAR) . and more than a fifth filed for bankruptcy before the release of their annual reports. It is. Watts & Zimmerman. This is because profitability measures a company’s efficiency of operations. Profitability is measured by return on capital employed (ROCE). 1986). The performance of a company has a signalling effect on both the markets for corporate securities. it is hypothesised. In contrast. reporting lead time is a decreasing function of profitability. Lawrence (1983) provided evidence that financially distressed companies in the US unduly delayed the release of their annual reports. reasonable to expect the management of a successful company to report its good news to the public on a timely basis. that: H2: Ceteris paribus. In contrast. Empirical evidence is. In spite of the inconsistency of the empirical results. Dyer & McHugh (1975) reported no association between profitability and total reporting lag in Australia.

Because such audit firms are known to expend more resources (Palmrose. 1986). 1993) in their audit engagements. This theory argues that different levels of audit quality would be demanded by different companies depending on their ownership structure. usually one of the Big-five audit firms (Firth & Smith.. however. two opposing views in the literature on the relationship between gearing and reporting lead time. and use high quality staff (Chan et al. The credibility of accounts is enhanced by association with a reputable auditor. management is likely to desire a higher quality audit to add credibility to the accounts as a means to reduce monitoring costs. One common clause is to require management to report promptly and at a certain frequency to enable debtholders to re-assess the long-term financial performance and position of companies. There are. In contrast.The relationship between timely reporting and the extent to which a company’s total assets is financed by fixed return capital is unclear. increases the probability of the external auditor being sued. it is likely that audit delay will be minimal and this results in speedy reporting by client companies. Another explanation for the negative relationship between gearing and timely reporting is provided by the monitoring costs theory. 1976). One view holds that highgeared companies report more promptly than low-geared companies. . 1991) on two grounds. to improve his defence against any possible legal suit. 1995). a high gearing ratio increases the probability of a company’s failure which. The auditor may have to perform more audit work. the other view holds that gearing usually associates positively with timely reporting (Carslaw & Kaplan. in turn. When agency costs are high. 1992). debtholders typically include clauses in debt contracts to constrain the activities of management (Jensen & Meckling. This school of thought contends that because high-geared companies have the incentive to invest sub-optimally. First. which may lengthen the time of the audit assignment (Simnett et al..

Similarly. the financial effect of which is to be determined by future events which may or may not occur’ (International Accounting Standards Committee. and thus are expected to require additional audit time. a company is coded one if it reported an extra-ordinary and/or contingent item. . and zero if otherwise. and the estimation of the amount involved. The third hypothesis is stated in the alternative form as: H3: Ceteris paribus. In the regression model. companies that report extra-ordinary and/or contingent items are more likely to release their annual reports late compared to those that do not report extra-ordinary and/or contingent items. Extra-ordinary and/or contingent items (EXTRA) The presence of extra-ordinary items and/or contingent items in a company’s accounts is expected to affect the timeliness of reporting to the public. they are likely to engage the auditor in a lengthy discussion and negotiation if he/she disputes the nature. a directional expectation is not formulated for gearing which is proxied by the ratio of long-term debts to total assets. 1991). The available empirical evidence indicates that companies reporting either extraordinary and/or contingent items are late reporters as they require careful audit investigation (see Ng & Tai. 4. Therefore. their existence. reporting lead time of a company is a function of its gearing ratio. Because of the uncertainty involved in the estimation of the outcomes of extra-ordinary and contingent items. By definition. audit of debt is relatively more time consuming than audit of equity especially if the number of the debtholders is large (Carslaw & Kaplan. the fourth alternative hypothesis is stated as: H4: Ceteris paribus. On the basis of the preceding discussion. extra-ordinary items reflect material non-recurring events that arise from business/economic activities that are not part of the company’s normal operations. a priori speculation of the direction of the relationship between gearing and timely reporting cannot be provided. Hence.4.Second. 1995: 182). contingent items relate to ‘conditions or situations at balance sheet date. 1994).

Because it is costly to complete an audit on schedule during the ‘busy audit season’. 1995). Although the direction of the effect of the month of the financial year-end on reporting lead time is unclear. Simnett et al. To complete an audit on schedule during the ‘busy audit season’ may require over-stretching of resources including the payment of over-time allowances. 4. Month of financial year-end (MONTH) It is expected that the month of the year in which a company’s financial year ends would influence its reporting lead time. 1991.5. All other months were treated as ‘non-busy audit season. This may result in possible audit delay as the workload of auditors increases tremendously. mixed (see Davies & Whittred. affects the reporting lead time of companies whose financial year-ends fall within the busy audit season. 1980. auditors have the incentive to delay the final audit. The most common financial year-ends for all the companies listed on the ZSE are 31 March and 31 December. the following directional hypothesis. there will be much demand for the services of auditors operating in that country.. The empirical evidence on the relevance of ‘busy audit season’ reported in the literature is. 1989. Newton & Ashton. in turn. reporting lead time of a company is an increasing function of the proximity of its financial year-end to the busy audit season in Zimbabwe. and because they are required by law to have their final accounts audited.6.’ A dummy variable. This. Carslaw & Kaplan. is coded one for ‘busy audit season’. however. MONTH. Complexity of operations (OPERA) . stated in the alternative form is tested: H5: Ceteris paribus. so these months were considered as ‘busy audit season’ in Zimbabwe.4. and zero if otherwise. If most companies in a country have their financial year-ends within a particular period of time.

reporting lead time of a company is an increasing function of the complexity of its operations. Thus. This hypothesised relationship is supported by Ashton et al. As a company continues and its accountants learn more. Thus. This proposition is based on the learning curve theory. the complexity of a company’s operations is captured by the number of different lines of business in which a sample company operates. . the theory suggests that a reduction in reporting time would occur as the number of annual reports produced is increased. it is hypothesised in the alternative form that: H6: Ceteris paribus. As a result. Company age (AGE) It is proposed in this study that promptness in financial reporting by a company is influenced by its age (i. well-established company is likely to be more proficient in gathering. On the basis of an analysis in Bamber et al. (1993).. a positive relationship between operational complexity and audit delay is expected. the time by which the company will eventually release its financial report to the public. (1987) who found a significant positive relationship between operational complexity of a company and audit delay. its development and growth).7. 4. In the context of this study. processing and releasing information when needed because of learning experience. reporting lead time of a company is a decreasing function of its age. and hence. an older. Thus. the ‘teething problems’ which would cause unusual delays are minimised.e. The degree of complexity of a company’s operations which depends on the number and locations of its operating units (branches) and diversification of its product lines and market is more likely to affect the time that an auditor takes to complete his/her audit assignment. it is hypothesised in the alternative form that: H7: Ceteris paribus.It is expected that the degree of complexity of a company’s operations will influence how timely the company reports to the public.

their auditors and audit reporting practices. and the timely reporting . eleven did not meet the requirement of data availability. The data for the analyses were derived from the annual reports of the sample companies. First. This step resulted in 58 companies surviving. measured here from the date of its flotation on the ZSE. a company’s age should be measured from the date of its incorporation. but it is not required by law or regulation in Zimbabwe to publicly report annually when a company is private. data on a company’s preliminary earnings and final reporting dates were available. Of the remaining 58 non-financial companies. but it is a good representation of the population of non-financial companies whose equities were traded on the ZSE in 1994. therefore.4 A listed company was included in the sample if it satisfied two conditions.Ideally.1. not included in the sample. its findings are still relevant today as there have not been any major changes in the number of companies listed on the ZSE. 5. it was a non-financial company. The Zimbabwean Companies Act exempts financial companies from complying with certain reporting requirements which may unduly influence either favourably or unfavourably their reporting behaviour. the final sample is not considered as being random. Research design and methodology 5. Sample selection and data collection A sample of 47 non-financial companies was selected from the 64 companies that were listed on the ZSE as of 31 December 1994. except those on OPERA which were taken from Remo Investment Brokers (Private) Limited (1994). and were therefore. A sample company’s age is. Because of the above selection procedure. 4 Although this study uses 1994 data. Second.

(ii) preliminary earnings announcement date. The audit report dates were taken from the annual reports of each sample company and were used to compute the AUDRLT for each company. three reporting events for each sample company were specified as: (i) audit report date. An audit reporting lead time (AUDRLT) is the number of calendar days between a sample company’s financial year-end and the date of its external auditor’s report. Gregory & Van Horn (1963) defined delay as ‘the length of time between the cut-off point (the time no transactions are accepted for inclusion in the particular report) and the distribution of reports to users. The Financial Gazette. Conceptual definition of timeliness Conceptually. A reporting lead time is the number of calendar days between a sample company’s financial year-end and the day on which each of the three reporting events occurred.’ Delay. The Chronicle. Determination of reporting lead time Following Dyer & McHugh (1975). A preliminary earnings reporting lead time (PRERLT) is the number of days between each sample company’s financial year-end and the earliest date on which its earnings announcement (abridged annual accounts) was published in one of the following daily and weekly newspapers in Zimbabwe: Business Herald. The Daily requirements in Zimbabwe between 1994 and 1999.5. and (iii) final report date (the date on which an annual report is filed with the ZSE). is described as reporting lead time.3. timeliness denotes a quality of: (i) ‘being available at a suitable time’ or (ii) ‘being well-timed’ (Gregory & Van Horn. 5. 1963: 576). Reporting lead time is defined here as the length of time a company takes after its financial year-end to release its financial information to the public.2. The Sunday News. The key variable in timeliness is the delay in release of annual reports. in this study. The data were collected in May 1996 when the writer paid a .

which by construction. all the equations in the system are estimated simultaneously. should not correlate with the error term of the equation in which it appears as an explanatory variable (Equation [2]). except that it is published weekly. Oppong (1993) used The Financial Gazette as his data source for a price-earnings study on the ZSE. The dates on which the annual reports are received by the ZSE are presumed to be the dates on which they have been released to the public in this study.Gazette. Equation [2]. The Financial Gazette is an equivalent of the US Wall Street Journal and the UK Financial Times in Zimbabwe. 5 6 study visit to the ZSE in connection with his doctoral research project.. and The Herald. As its name implies. weekly readership was highest for the Business Herald. 1991). however. The two cross-sectional reporting lead time models were estimated with a two-stage least squares (2SLS) technique. either PRERLT or FINRLT). 5. The PRERLT and FINRLT are post-audit reporting lead times (Post-AUDRLT). a simultaneous-equation regression system of two cross-sectional models was used. The ZSE has a custom of writing the dates on which annual reports of listed companies are received on their front covers. the same explanatory variables were used but with different dependent variables depending on which reporting lead time is being estimated (i.6 For the equation of interest. In the second stage of the estimation process. followed by The Financial Gazette. The test to confirm whether or not an endogenous regressor correlates with an error term of a structural model is called a simultaneity test (Pindyck & Rubinfeld. and The Chronicle.500 adults aged between 16 and 60. the RIZ found that The Herald is the most widely read daily newspaper followed by The Daily Gazette. .e. A final reporting lead time (FINRLT) is the number of calendar days between a sample company’s financial year-end and the date on which its audited annual report is received by the ZSE. These dates were taken from the inhouse library of the ZSE. The FINRLT for each company was computed from these dates.5 Back copies of these newspapers were thoroughly examined in the library of the publisher of The Financial Gazette in Harare to obtain the first dates of publication of the abridged annual accounts of each sample company. This regression yields estimated endogenous regressor.4. the estimated endogenous regressor is incorporated in Equation [2] as an explanatory variable. Model specification To analyse the influence of each factor on the timely reporting behaviour of the companies in the sample. the 2SLS is a two-stage process. the endogenous regressor is regressed on some exogenous variables. In the first stage. The use of the 2SLS becomes necessary in simultaneous equation models where an endogenous regressor (AUDRLT in this study) becomes stochastic. In reality. and usually correlates with the error term of the equation in which it appears as an explanatory variable (the Equation [2] in this study). As far as financial newspapers were concerned. Using a stratified random sample of 2. A readership survey conducted by Research International Zimbabwe (RIZ) in 1994 showed that these newspapers enjoy the highest readership rate in Zimbabwe.

the following simultaneous regression models which are assumed to hold for each sample company were estimated: AUDRLT = f(SIZE. AGE) [1] [2] 7 The null hypothesis that there is no problem of simultaneity is rejected. This suggests that the time a company takes to release either its preliminary earnings or final accounts is influenced by how long it takes its auditors to certify the accounts. AGE. the events of preliminary annual earnings announcement and final release of audited accounts will chronologically follow the date on which the external auditor certifies the accounts.0000. Hence. First. it was statistically assessed whether Equations [1] and [2] are simultaneously determined by using Hausman’s (1978) simultaneity specification test as suggested in Pindyck & Rubinfeld (1991: 303-305) and Gujarati (1995: 670-671). Rather. the empirical evidence in Panel B of Table 1 indicates that auditors in Zimbabwe. Hence. Second. The results of the Hausman test suggest the presence of a simultaneity problem. on a priori reasoning. MONTH. while that for the SqrtFINRLT model is 29. OPERA) Post-AUDRLT = f(AUDRLT.53 with p-value = 0. .0000.The use of the 2SLS technique is justified as follows. EXTRA. it is reasonable to assume that both the PRERLT and FINRLT are a function of the AUDRLT. if there is a problem of simultaneity because OLS estimates will be inconsistent regardless of sample size. SIZE. The F-statistic for the PRERLT regression model is 62. based on an argument that a chronological ordering of events does not conclusively establish causality. Finally. while it takes the client company between seven and 23 days after the auditor has certified the accounts to release its preliminary and final accounts to the public.26 with p-value = 0. GEAR. PROFIT.7 Pindyck & Rubinfeld (1991) and Gujarati (1995) have cautioned against the use of ordinary least squares (OLS) regression technique. on average. GEAR. take 62 days after a client company’s financial year-end to certify the latter’s accounts. they recommended the use of a 2SLS regression technique whose estimates will be both consistent and efficient. PROFIT.

a normal probability plot of the data on total assets suggests that its distribution is not normal. but was within the provisions of the Companies Act. PRERLT and FINRLT) are called endogenous variables. A high and significant correlation exists among the three reporting lead times. suggesting that each of them can be a surrogate measure of timeliness. An examination of the actual reporting dates indicated that these companies reported well in advance of the 160-day requirement of the ZSE.As stated earlier. Oppong (1993) also noted a similar annual reporting behaviour of the companies 8 In the context of simultaneous equation models. [TABLE 1 ABOUT HERE] Table 2 indicates that about 98 percent of the sample companies submitted their audited annual reports to the ZSE by the required date in each case. Also. the jointly dependent variables (AUDRLT.1. As a result. 6. . Summary statistics Table 1 presents summary statistics of the variables used in this study. Post-AUDRLT is represented by either PRERLT or FINRLT depending on which reporting lead time is being estimated. The results of a Shapiro-Wilk’s test indicated that the distribution for each of the reporting lead times is fairly normal except that of FINRLT. The mean timeliness measured by earnings announcement date is only eight days longer than that measured by auditor’s report date. analysis and discussion 6. while those that are truly non-stochastic or can be so regarded (such as AUDRLT. For this reason a square root of FINRLT (SqrtFINRLT) is used in all the regression models.8 Equation [1] is a function of audit-related factors and some corporate-specific factors that may affect audit reporting lead time. Only one company in the sample reported late. Panel A of that table presents the results of correlation and normality tests performed on the reporting lead times. The audit-related factors indirectly affect PRERLT and FINRLT. the measure of SIZE used in all the regression models is natural log transformed (LnSIZE). Empirical results.

No pairwise correlation coefficient is higher than 0. He observed that preliminary annual earnings information.80. Panel A of Table 3 presents the correlation matrix of the empirical indicants of the factors and AUDLRT.01 level.2. . AUDRLT also influences significantly (at the 0. were released by the end of the third month following their financial yearend.9 Like the results of the simple correlation procedure. and as such. The coefficients of LnSIZE. PROFIT and OPERA) are called exogenous variables. a post-estimation variance-covariance matrix of the estimators (VCEs) of the PRERLT model is provided in Panel B of Table 3. Though not significant. hypothesis H3.listed on the ZSE. Also. As expected. 1968: 366). and in several cases. SIZE. is not supported by the data. it is called exogenous regressor. [TABLE 2 ABOUT HERE] 6. The F-statistic indicates an overall satisfactory fit which is significant at the 0. because many simultaneous estimation techniques break down in the face of multicollinearity (see Farley & Leavitt. PROFIT and AGE are all statistically significant at the conventional levels10. Thus. as suggested by Gujarati (1995) this implies that multicollinearity is not a serious problem. the nature and the degree of multicollinearity among the explanatory factors and AUDRLT were assessed. because AUDRLT is both endogenous and exogenous to the system. the positive sign on GEAR supports the school of thought that audit of debt is more time consuming than that of equity. the actual annual reports. Regression analysis Before the models were estimated.01 level) when a sample company announced its annual preliminary earnings. [TABLE 3 ABOUT HERE] Panel A of Table 4 reports the results of the PRERLT regression model. However. H2 and H7 respectively. high-geared companies are likely to be late reporters. based on the monitoring costs theory. the VCEs suggest that the degree of multicollinearity is not severe. and have the expected signs suggested by hypotheses H1.

First. 1999a: 186-190). a Pregibon’s goodness-of-link test was performed on both the PRERLT and SqrtFINRLT models. Second. no evidence was found to support the monitoring costs theory as speculated in hypothesis H3. but are not statistically significant in explaining how timely a sample company submitted its audited annual accounts to the ZSE. if the prediction squared has an explanatory power (Pregibon. see Pregibon (1980). normality and independence of the residuals of the models were examined. those of AGE and MONTH are significant at the 0. but the MONTH variable has a ‘wrong’ sign. The graphs for both models showed no discernible patterns 9 The VCEs of the SqrtFINRLT model (not reported) were the same as those of the PRELRT model. 11 The link test examines the fit of a hypothesised model. It involves a refitting of the model. Again. On the other hand. A hypothesised model is mis-specified.11 The null hypothesis that these models were mis-specified. 10 .10 level for the coefficients of PROFIT and AGE is too high for significance. and that some other significant explanatory variables can be found was not supported by the link test. Some researchers may argue that the 0. For different estimation procedures of the link test . a residual-versusfitted plot was graphed for each model. PROFIT and AGE have the expected signs.05 level. one should not find any additional significant independent variables except by chance. The coefficients of LnSIZE and OPERA are statistically significant at the 0.10 level.[TABLE 4 ABOUT HERE] The results of the SqrtFINRLT regression model are presented in Panel B of Table 4. The test is based on the significance of the prediction squared. Of the hypothesised relationships between AUDRLT and the seven exogenous variables in Equation [1]. The reported results for Equation [2] in Table 4 were subject to several diagnostic tests. Stata Corporation. but with the prediction and the prediction squared as explanatory variables. To check for the constant variance assumption. The coefficients of AUDRLT.05 level. AGE and MONTH are supported by the data (Table 4). 1979. the coefficient of LnSIZE is significant at the 0. The test is based on the idea that if a model is properly specified. While the overall fit of the model is not statistically significant. the assumptions of constant variance. only those between AUDRLT and LnSIZE. OPERA.

in the data points suggesting that the assumption of constant variance can be maintained for both models. However. Because standard errors are biased when estimated residuals are serially correlated. a 2SLS regression with robust standard errors was run. the coefficient of PROFIT became a significant explanator of the timely reporting behaviour of the sample companies. 1999b: 189). 1999c: 257). The negative coefficient of LnSIZE signifies that larger companies are more prompt reporters than smaller companies.14 (0. 1970 for relative efficiency of the runs test).73 (0.3. the assumption that the residuals of the models are independent cannot be maintained. The results of the robust regression (not reported here) were not significantly different from those reported in Table 4. there should be no pattern to the estimated residuals plotted against predicted values (Stata Corporation. And for the SqrtFINRLT model. The results of a non-parametric Onesample runs test of randomness suggest that the residuals are negatively serially correlated (see Geary. 6. The results of the robust regression (not reported here) for the AUDRLT model were the same as those reported in Table 4. These results could be that because large companies in Zimbabwe are affiliated with multinational corporations. Davies & Whittred (1980). and as result auditors spend less time in conducting compliance and substantive tests. These results corroborate the conclusions in Dyer & McHugh (1975). The hypotheses that the residuals of the PRERLT and the SqrtFINRLT models are not normally distributed are rejected at the 0. while the coefficient of GEAR became statistically significant at the 0. for the PRERLT model.4600) and -0. and so they tend to have access to modern technology and are able to produce their accounts on a timely basis. The robust 2SLS analysis relaxes the independence assumption (Stata Corporation. Frost & Pownall (1994). and Abdulla (1996).13 The Z-statistics (p-values) for the models with PRERLT and SqrtFINRLT as dependent variables are -0. Another explanation is that large companies tend to have strong internal control systems and efficient audit committees. However. 12 13 In a well-fitted model. . Givoly & Palmon (1982). that of the AGE variable lost its statistical significance.890) respectively.12 A Shapiro-Wilk’s test for normality of the estimated residuals indicates that the normality assumption is not violated. Discussion of regression results The results show a significant inverse relationship between LnSIZE and both PRERLT and SqrtFINRLT.10 level.05 level.

prolongs a client company’s reporting lead time. it is not statistically significant. In contrast. Rather. (1995). in turn. Lawrence (1983). auditors tend to do more substantive tests to avoid any potential future legal suit. A complementary view is that because companies reporting losses are viewed relatively as risky.’ This finding is also consistent with that of Simnett et al. Conclusions. This. This result. This result is not surprising given the tendency of stock markets to reward profitable companies more than they reward unprofitable ones. While the negative relationship between profitability and timely reporting found in this study is consistent with the conclusions in Courtis (1976). however.The negative sign of the PROFIT variable in both models suggests that successful companies (those with good news) report more promptly relative to their counterparts with poor operating results (those with bad news). therefore. 7. Whittred & Zimmer (1984) and Abdulla (1996). it contradicts those reported in Dyer & McHugh (1975) and Garsombke (1981). The significance of the nature of the relationship between AUDRLT and PRERLT confirms Givoly & Palmon’s (1992: 491) assertion that the ‘single most important determinant of the timeliness of the earnings announcement is the length of the audit. it suggests that there are other factors such as delay by printing houses in Zimbabwe that might be responsible for the delay in releasing the final annual report to the public. have the incentive to inform the public of their superior performance by releasing their annual reports quickly. Profitable companies. The study also investigated the factors . limitations and implications for future research This article reports the results of a study which examined the pattern and timeliness of financial reporting practices of 47 listed companies in Zimbabwe. while the relationship between AUDRLT and SqrtFINRLT is positive. does not contradict nor diminish the strength of Givoly & Palmon’s (1992) assertion.

age and audit reporting lead time. such as the use of audit committees by sample companies. profitability. Future research studies may consider non-linear models. depending on how the dependent variable (Post-AUDRLT) of the specified model is measured. quality of their internal control systems and the effects of crosslisting. Second. It is not likely that the relationship will always be linear. company age and audit reporting lead time affected how fast a sample company announced its preliminary earnings (PRERLT). The results of a 2SLS regression with robust standard errors indicate that both company size and profitability are significant predictors of timely reporting in Zimbabwe. The results of a 2SLS regression show that timeliness in reporting by ZSE listed companies is influenced by their size (measured by total assets).. Third. The descriptive analysis indicates that about 98 percent of the sample companies reported within the regulatory stipulated time period (i. While company size.e.that influenced timely reporting of these companies. but correlate significantly with each other. this study did not consider all relevant factors that might affect timeliness in reporting. due to either lack of data on these variables in Zimbabwe or the difficulty in measuring them. the functional form of models examining the relationship between timeliness and its determinants warrants further investigation. profitability (measured by ROCE). The study also found that preliminary earnings announcement date. 160days after a company’ financial year-end). the results of this study should not be generalised to financial companies listed on the ZSE. The above findings and conclusions are subject to a number of limitations. regardless of how timeliness is measured. First. Fourth. only size influenced the timeliness by which a sample company released its final audited annual report (FINRLT). audit reporting date and the date on which final accounts are released are not identically the same. These factors merit exploration in future research work on timeliness. the size of the sample used .

in this study is small. Future research may examine the same sample of companies over a period of time to ascertain the trend in their timely reporting behaviour. this study investigates the timely reporting behaviour of ZSE listed companies at a particular point in time. Finally. . Future research may extend the statistical design employed in this study to other emerging economies or developed economies where a large sample size is likely to be obtained.

E.. U. N. (1991). In Doupnik. R. T. Journal of Political Economy. (1984). E. and Gwilliam. J. Autumn: 275-292. Journal of Accounting Research. 13(3). R. (1975). and McHugh. November: 362-368. Journal of Business Finance and Accounting. Winter: 21-32. Accounting and Business Research. Sarasota. and Kaplan. J. Auditing: A Journal of Practice and Theory. A. ‘Determinants of audit fees for quoted UK companies’. S. (1993). and Whittred. 16(1). April: 228-307. A. J. F. R. Bamber. S.. Farley. H. H. J. Y. (1972).). M. and Sridhar. E. P. D. B. A. S. H. ‘An empirical analysis of audit delay’. Willingham. (1995). ‘The association between selected corporate attributes and timeliness in corporate reporting: Further analysis’. (1968). Courtis. ‘An examination of audit delay: Further evidence from New Zealand’. M. 9: 73-88. and Leavitt. J. E. 25(2). Ashton. R. and Schoderbek.. Chambers. M.. (1993). K. ‘Industry-wide disclosure dynamics’. J. S. and Elliott. B.. J. J. (1996). (eds. (1980). Ashton. P. A. ‘Relationships between timeliness in corporate reporting and corporate attributes’. Journal of Accounting Research. (1976). G. Journal of Marketing Research. Accounting and Business Research. Spring: 21-47. S. Winter: 45-56. Spring: 1-23. Journal of Accounting Research. November: 765-786. Information Evaluation. and Penman. Journal of Accounting Research. P. Spring: 157-174. P. J. ‘Agency problems and the theory of the firm’. S. Davies. 5. 88(2). L. June: 48-60. Contemporary Accounting Research. ‘Audit delay and the timeliness of corporate reporting’. Feltham. (1989). 22(1). Spring: 657-673. ‘The timeliness of Bahraini annual reports’. 6(25). Graul. ‘A model of the distribution of branded personal products in Jamaica’. D. 5(2). H. (1987). C. 20(6). A. ‘Timeliness of reporting and the stock price reaction to earnings announcements’. ‘Audit structure and other determinants of audit report lag: An empirical analysis’. and Salter. Carslaw. and Newton. . Autumn: 204-219. C. A. 22(85). K. Advances in International Accounting. Bamber. Dye.. G. Florida: American Accounting Association. 33(1). Ezzamel. Chan. Studies in Accounting Research No. Fama. Abacus. (1980). R. P.References Abdulla. S. ‘The timeliness of the Australian annual report’. Dyer. 12(1). 5.

Spring: 153-176. L. 21(2). R. D. H. (1963). Spring: 75-102. 8(29). Jensen. Auditing: A Journal of Practice and Theory. Journal of Accounting Research. R. D.Firth. International Accounting Standards Committee (1995). 57(3). 73(2). (1978). and Van Horn. Hausman. Belmont. ‘An empirical investigation of the effect of quarterly earnings announcement on stock returns’. Communication via Annual Reports. and Pownall. H. Lawrence. London: International Accounting Standards Committee: 181-188. Gujarati. ‘International Accounting Standard IAS 10 (reformatted 1994)’. Applied Economics. H. Journal of Accounting Research. J. and Ashton. 8 Supplement: 22-37. ‘Reporting delays for failed firms’. (1982). A. G. R. November: 1251-1271. 3: 305-360. and Meckling. Gilling. Manne. A. (1989). ‘Relative efficiency of count of sign changes for assessing residual autoregression in least squares regression’. (1970). (1992). J. Third Edition. In Courtis. California: Wadsworth Publishing Company. Accounting Review. Autumn: 606-610. AFM Exploratory Series No. (1977). W. Journal of Political Economics. M. April: 110-120. Gregory. Journal of Accounting Research. 22(1). C. and Palmon. (1965). J. H. W. ‘Timeliness of annual earnings announcements: Some empirical evidence’. 32(1). P. ‘The association between audit technology and audit delay’. (1994). Geary. ‘Selection of auditor firms by companies in the new issues market’. D. ‘Mergers and the market for corporate control’. Armidale. C. and Schroeder. Accounting and Business Research. (ed. D. ‘Theory of the firm: Managerial behaviour. M. K. 57(1): 123-127. G. Kross. 11. Automatic Data-Processing Systems: Principles and Procedures. (1981). . ‘Accounting disclosure practices in the United States and the United Kingdom’. 46(6). ‘Specification tests in econometrics’. Journal of Financial Economics. Winter: 34-36. A. International Accounting Standards 1995. and Smith. D. (1995). N. 24: 247-255. Givoly.508. Newton. M. C. agency costs and ownership structure’. R. (1983). Inc. Econometrica. Second Edition. D. (1976). Basic Econometrics. C. Garsombke. Frost. New South Wales: University of New England: 204-218.). Biometrika. E. London: McGraw-Hill. ‘Timeliness in corporate reporting: Some further comment’. H. (1984). A. July: 486. ‘Timeliness of corporate financial disclosure’.

Part I Developing prediction equations’. In Reckers. Accounting Review. 29(1): 15-24. G. (ed. Third Edition. Texas: Stata Press. University of Toronto. Journal of Accounting Research. Aitken. R. Statistics: Graphics and Data Management: Stata Reference H-O . Florida: Harcourt Brace and Company. Y. (1995). and Tai. Spring: 97-110. Texas: Stata Press. 28(1): 71-77. Pindyck. M. College Station. Release 6. V. H. R. Applied Statistics. D. ‘Goodness of link tests for generalised linear models’. Ramanathan. ‘Data analytic methods for generalised linear models’. D. 26(1). Middlesex University. Pregibon. L. S. (1994). (1995). Snee. Pregibon. and Rubinfeld. R. 59(2). (1998). J.D. April: 67-79. Advances in Accounting. Journal of Quality Technology. J. March: 43-59. British Accounting Review. K. D. Z. Watts. ‘Price-earnings research and the emerging capital markets: The case of Zimbabwe’. ‘Audit fees and auditor size: Further evidence’. Texas: Stata Press. Release 6. England. thesis. 24(4). (1973). B. A. P. L.. Stata Corporation (1999b). and Zimmerman. R. Canada. Third Edition. College Station. Inc. Englewood Cliffs.Ng. S. and Zimmer. ‘The adequacy of corporate mandatory disclosure practices on emerging markets: A case study of the Zimbabwe Stock Exchange. (1993). M. April: 287-295.D. Oppong. (1986). Harare: Remo Investment Brokers [Private] Limited. ‘An empirical examination of the determinations of audit delay in Hong Kong’. New York: McGraw-Hill. Release 6. ‘Timeliness of financial reporting and financial distress’. P. L. (1986). ‘The determinant of audit delay’. and Firth. (1984).). M. Remo Investment Brokers [Private] Limited (1994). D. (1980). Unpublished Ph. Introductory Econometrics with Applications. Investor’s Guide: Introduction to the Zimbabwe Stock Exchange. Palmrose. Whittred. Simnett. F. Positive Accounting Theory. Statistics: Graphics and Data Management: Stata User’s Guide. New Jersey: Prentice-Hall. Owusu-Ansah. Statistics: Graphics and Data Management: Stata Reference P-St. (1979). (1991). 13: 1-20. Stata Corporation (1999c). . I. R. 5(2). Orlando. thesis. Choo.’ Unpublished Ph. Stata Corporation (1999a). International Journal of Accounting. P. Department of Statistics.. College Station. ‘Some aspects of nonorthogonal data analysis. Econometric Models and Economic Forecasts.

of days since financial year-end to the date of external auditor’s report No.45 0 0 0 1 1 Max. 17 25 34 9.0000 0. as suggested in the econometrics literature.11 ? + + + – ________________________________________________________________________ † Because the explanatory variables measure different effects on reporting lead time. 115 147 176 14.Table 1: Summary statistics of the reporting lead times and the factors investigated (n = 47) ________________________________________________________________________ Panel A: Pearson correlation matrix and normality test of reporting lead time Reporting lead time AUDRLT PRERLT FINRLT SqrtFINRLT Shapiro-Wilk’s test for normality _____________________________ Z-statistic Significance AUDRLT 1.4848** 0.6441** 0.98 1. dev.04 0.08 26.34 0.71216 PRERLT 0.7262** 0.01 (two-tailed).0000 0.8045** 0.560 0. To be able to compare the numerical value of one regression coefficient with that of another.81 1 1 9 49 Mean Std. predicted directions and summary statistics of variables Variable† AUDRLT PRERLT FINRLT SIZE PROFIT GEAR EXTRA MONTH OPERA AGE Proxy of variable No.70 69.0000 0.0000 0. * p < 0.8217** 1.07957* SqrtFINRLT 0.49 85.728 0.09 23. and (iv) collinearity problems among the explanatory variables are reduced.27 9. According to Snee (1973) .56 32. estimated with maximum precision.408 0.64 -3. of years since listing on the ZSE Direction + ? ? – – Min. (ii) computer round-off errors are minimised.50 1. ** p < 0.6406** 0.40 8. of days since financial year-end to the date of preliminary earnings announcement No. x i and s i denote the mean and standard deviation of the i th variable. therefore.67 22. the estimated coefficients will not be comparable in numerical terms (Ramanathan.71 22. of days since financial year-end to the date of receipt of final accounts by the ZSE Natural log of year-end total assets Return on capital employed (%) Percentage of total assets financed by longterm (interest-bearing) debts Extra-ordinary and/or contingent items Busy and non-busy audit seasons in Zimbabwe Different lines of business No.23319 FINRLT 0.53 3.178 0.6406** 1.02 5. the variables in the regression models were standardised. (iii) regression coefficients can be ranked in order of absolute values to determine which variables are most important.15 12.8217** 1.06 85.7262** 0.82 14.4848** 0.8045** 1.6441** -0.42931 ______________________________________________________________________________ Panel B: Proxies.87 0.10 (two-tailed). 61. by subtracting the sample mean from the raw score ( X ) and dividing the difference by the sample standard deviation as shown in the formula below: i (x − x ) s i i i where standardisation of variables has the following advantages: (i) regression coefficients are estimated within the range of the data.79 0.62 16. 1995: 165-166). and are.

4 0.4 27.1 10.7 0.5 100.4 27.4 6.7 0.1 25. % Companies actually reporting by the required date No.4 2.0 2.0 6.4 6. 0 3 13 0 1 5 3 3 3 3 1 12 47 % Companies expected to report by the end of † No.4 6.0 3 3 3 3 1 11 0 3 13 0 1 5 46 6.1 10.7 0.4 6.4 6.0 6.0 6.0 3 3 3 3 1 12 0 3 13 0 1 5 47 6.0 2.4 6.6 97.9 ________________________________________________________________________________ † This is based on the 160-day reporting requirement of the Zimbabwe Stock Exchange.5 0.4 6. .6 100.6 6.1 25.4 27. % ________________________________________________________________________________ January February March April May June July August September October November December Total 0.4 6.4 2.Table 2: The pattern and timeliness of annual financial reporting of the sample companies ________________________________________________________________________________ Calendar month Companies with financial year ending in No.4 2.4 6.1 23.1 10.0 2.

132 0.069 -0.0000 Constant | 0.000 Gear | -0.026 0.0008 0.0006 1.000 Month | -0.0000 _______________________________________________________________________________________________ * p < 0.000 Profit | -0.000 Opera | 0. .4096 -0.166 -0.195 -0.000 Extra | -0.0547 1.036 -0.066 1.0020 0.1159 0.4691 1.1457 1.030 0.173 0.017 -0.098 1.000 Age | -0.3042 -0.0219 0.110 1. † The variance-covariance of the estimators of the SqrtFINRLT model (not shown here) were the same as those of the PRELRT model.288 1.004 0.0715 1.167 -0.252* -0.05 (two-tailed).017 -0. ** p < 0.2387 0.063 0.074 -0.290* 0.0009 0.217 -0.0001 0.053 0.089 -0.3356 0.0000 Gear | 0.1 (two-tailed).129 1.152 1.Table 3: Results of collinearity tests on empirical indicants of the factors investigated (n = 47) ________________________________________________________________________________________________ Panel A: Pearson product-moment correlation matrix | LnSize Profit Gear Extra Month Opera Age Audrlt ---------------+----------------------------------------------------------------------LnSize | 1.344** 1.000 Audrlt | -0.000 ________________________________________________________________________________________________________________________________________________________________ Panel B: Variance-covariance matrix of the estimators of the PRERLT model† | Audrlt LnSize Gear Profit Age Constant ---------------+-----------------------------------------------------------------------Audrlt | 1.0000 Profit | 0.117 0.264* -0.0000 Age | 0.0000 LnSize | 0.

2811213 ______________________________________________________________________________________________________________________ Panel B: Final reporting lead time (SqrtFINRLT) as dependent variable Equation Obs.5609456 .964 0.5471575 0.2517417 Age(-) | -.344 -.2482651 Age(-) | -.0000992 .0410714 Gear(?) | .2358306 .1462686 -1.0447577 .197123 Constant | 8.95928 0.1734 -------------------------------------------------------------------------------| Coef.3716338 .050 -.332319 LnSize(-) | -.2945728 .275744 6. Root MSE R-squared Adj. R-squared F-statistic Prob > F -------------------------------------------------------------------------------SqrtFinrlt(2) 47 1.7584619 .2805792 .014 0.2673903 -1.672 0.160108 2.547 0. Interval] -------------+-----------------------------------------------------------------SqrtFinrlt(2)| Audrlt(+) | .121 -.6900824 Gear(?) | 2.285 0.6259 5.17859 -.684 0.684 0.33 0. Err.95928 0.737 0.800178 1.845 0.036 -1.337 0.2805792 .53538 5.7234029 .267637 -------------------------------------------------------------------------------Audrlt(1) | LnSize(-) | -.1456831 0.0179729 .014 -11.47976 2.297586 -1.155319 0.2164706 40.1955024 .2806695 .4123 0.007 2.672 0.081 0.2715 Audrlt(1) 47 .57 0.49346 .1456831 0.484 0.1955024 .001 0.2548859 .479962 Constant | 69.429766 Profit(-) | -3.413923 Age(-) | -3.1656 1.2811213 _______________________________________________________________________________________________ † The p-values are for one-tailed tests except that for the GEAR variable which is for a two-tailed test because its hypothesis is specified in a non-directional manner.2723621 .1502676 0. t P>|t|† [95% Conf.999 -.934335 2.936 -.764 0.1399297 -0.0009 Audrlt(1) 47 . .2365847 -1.082 -8.0000992 .16 0.2273118 .1567812 1.4302345 .001 0.2198 0.4913586 . Root MSE R-squared Adj.0511737 Constant | -.045 -.340 -.1532362 -0.403 0.337 0.2029107 .1003539 Opera(+) | .1532362 -0.2947711 Profit(-) | -.23418 73.999 -.26663 LnSize(-) | -5.1010342 .033466 . Std.0447577 .0616844 .2723621 .1399297 -0.088 -7.989 -.1513131 -1.313707 2. Std.344615 -1.0019133 .Table 4: Two-stage least squares regression estimates ________________________________________________________________________________ Panel A: Preliminary earnings reporting lead time (PRERLT) as dependent variable -------------------------------------------------------------------------------Equation Obs.252 -.1590211 -1.085 -.2945728 .0798 1.2548859 .4049791 †† Month(+) | -.1513131 -1.094 -.764 0.2029107 .57 0.415 0.000 8.989 -.4661803 Profit(-) | -.5609456 .2563 0.596755 -2. t P>|t|† [95% Conf.050 0.375 0.1462686 -1.102249 33.0410714 Gear(?) | .6022298 .804135 24.0019133 .167 -.5894819 Extra(+) | .1590211 -1.2414273 -0.344 -.72533 -------------------------------------------------------------------------------Audrlt(1) | LnSize(-) | -.793 0. Interval] -------------+-----------------------------------------------------------------Prerlt(2) | Audrlt(+) | 13.6022298 .2947711 Profit(-) | -.1734 -------------------------------------------------------------------------------| Coef.1003539 Opera(+) | . Err.1010342 .050 -.977 0.416 0.186 0.393295 9.077011 2.6666 0.321124 2.4913586 .014 0.094 -.341 -2.0616844 .05617 1.0465458 Gear(?) | .2849479 .830466 .403 0.042 -.252 -.1567812 1.3716338 .2219353 0.045 -.1502676 0. R-squared F-statistic Prob > F -------------------------------------------------------------------------------Prerlt(2) 47 14.2849479 .2482651 Age(-) | -.8776953 1.0511737 Constant | -.000 65.737 0.0798 1.5894819 Extra(+) | .4049791 †† Month(+) | -.2198 0.

309.†† To isolate the partial effect of each of the busy audit seasons (DEC and MAR) and that of the non-busy audit season (OTHERS). dropped by the statistical software used because too many parameters were included in the models vis-à-vis the number of observations.097). The MAR dummy variable was. however. . The results (not reported here) indicated that only the coefficient of OTHERS was significant (t-statistic = 1. the models were re-estimated with these variables included as an additional set of explanatory variables. p-value = 0.

Sign up to vote on this title
UsefulNot useful