CPA-Firm Merger: An Investigation of Audit Quality

Abstract

It would be interesting to know how the theories and hypotheses in auditing developed for a mature economy work in an emerging market. This paper uses the merger of Diwan, Ernst & Young and J.T. Lai & Co., CPAs in Taiwan in 2000 as a target to study the effects of a CPA-firm merger on audit quality, measured by discretionary accruals. Diwan, Ernst & Young and J.T. Lai & Co., CPAs belonged to Big 5 and non-Big 5 CPA firms in Taiwan, respectively, before they merged. The magnitudes of the increased size and reputation caused by the merger are different between the two firms. Therefore, we also analyse whether the merger has different effects on audit quality of these two firms. The empirical evidence shows that after the merger, the discretionary accruals of the two firms' clients significantly decrease. In addition, the magnitude of the change is greater for the clients of J.T. Lai & Co., CPAs than for those of Diwan, Ernst & Young. The results are consistent with predictions of both size and reputation effect hypotheses.

Acknowledgements

We would like to thank the editors and anonymous referees for the helpful comments and suggestions during the review process. Chien-Heng Jennifer Chang acknowledges financial support from the T.N. Soong Accounting Foundation.

Notes

In Taiwan, there are two major stock markets, Taiwan Stock Exchange Corporation (TWSE) and GreTai Securities Market (GTSM). A company, satisfying certain criteria, can apply for listing on one of the two markets.

Dechow et al. (1995) compare five methods of estimating discretionary accruals. They point out that the modified Jones model has the highest testing power. Guay et al. (1996) also suggest that only the Jones model and the modified Jones model can produce reliable estimates.

For example, Myers et al. (2003) study the relation between accruals and auditor size and tenure. The control variables in their paper do not include any corporate governance variables. Frankel et al. (2002) examine the relation between unsigned accruals and auditor size. For all kinds of corporate governance variables, they include only institutional investor holdings.

However, Frankel et al. (2002) do not find that Big 5 CPA firms have significant effects on the discretionary accruals. Geiger and Raghunandan (2002b), Carcello and Neal (2000) and Mutchler et al. (1997) also do not find a significant Big 5/Big 6 variable.

We use the cash flow approach (see Hribar and Collins, 2002) to compute total accruals, which equal the earnings before extraordinary items minus the cash flows from operating activities. We also perform additional analysis where total accruals are computed following Frankel et al. (2002) and Reynolds and Francis (2001). The amounts of total accruals estimated by different methods are very similar and our main findings are robust to the methods for computing total accruals.

E&Y and J-T merged on 1 January 2000 and the name of the post-merger CPA firm remained E&Y. That is, the financial statements of all sample companies in the post-merger period were audited by post-merger E&Y. Therefore, we define JT _it as 1 for all companies whose pre-merger auditor is J-T, whether the data belongs to the per- or post-merger period.

ZSCORE = 1.2 * (current assets − current liabilities)/total assets + 1.4 * retained earnings/total assets + 3.3 * earnings before interest and taxes/total assets + 0.6 * market value of preferred and common equity/total liabilities + 0.999 * sales/total assets.

Altman (1968) indicates that firms with a z-score greater than 2.99 are clearly assigned to the ‘non-bankrupt’ group, while firms with a z-score below 1.81 are assigned to the ‘bankrupt’ group. The area between 1.81 and 2.99 is defined as the ‘grey area’.

On the other hand, the mean of the post-merger ZSCORE is significantly lower than the pre-merger one, which could lead to lower discretionary accruals in the pre-merger period.

We use White's robust estimator of variance in place of the traditional calculation, and report the robust coefficients and p-values in all tables (White, 1980).

As indicated in Table 1, the number of J-T's clients included in the sample is five or six. In order to deal with the sample size issue, we also re-estimate regressions with the bootstrapping method. These results are not shown in this paper as they are highly similar except that MERCPA becomes less significant (p = 0.051 for the Jones model and p = 0.046 for the modified Jones model).

The control group in the regression consists of clients of other Big 5 CPA firms, excluding those of the former T.N. Soong & Co., which was the member firm of Arthur Andersen in Taiwan, to avoid the possible interference from the Enron scandal.

Alternatively, we winsorise all variables at 1st and 99th percentiles. The results are still similar to those reported in the paper.

Again, the control group excludes clients of the former T.N. Soong & Co. to avoid the possible interference from the Enron scandal. The results with companies audited by former T.N. Soong & Co. are basically the same.