Does International Accounting Standard No. 27 Improve Investment Efficiency?

Abstract

We examine how the adoption of International Accounting Standard No. 27 (hereafter, IAS 27) consolidation rules affects firm-level investment efficiency. IAS 27, effective in Taiwan for fiscal years beginning after 2005, defines the “control” criteria for consolidated entities as majority control rights rather than majority financial ownership. IAS 27 discourages firms’ ability to manage earnings through the use of unconsolidated entities and reduces information asymmetry between managers and shareholders. Consistent with the standard’s intended objectives, we document that firms experience a significant increase in investment efficiency after adopting IAS 27. Firms subject to overinvestment (underinvestment) are more likely to reduce (increase) investment toward a more optimal level after IAS 27 adoption. We also find that foreign investors increase their shareholdings in Taiwanese firms after the adoption of IAS 27.

Keywords

international accounting standards International Accounting Standard No. 27 IAS 27 consolidated financial statements investment efficiency

Introduction

On March 11, 2010, the Financial Accounting Standards Board (FASB) and the International Accounting Standards Board (IASB) jointly issued the Exposure Draft, Conceptual Framework for Financial Reporting: The Reporting Entity. Both boards agree that when one entity has “control” over investments in another entity, the entity with control should present consolidated financial statements (Paragraph RE8). However, the two boards used different criteria to define control. Accounting Research Bulletin (hereafter, ARB) 51 states that the consolidation should be based on financial control (AICPA, 1959, para. 20). In the absence of a definite and comprehensive pronouncement on “control,” most firms use majority ownership as a general policy for consolidation and manage the 50% threshold to omit some subsidiaries from consolidation (e.g., Blacconiere & Hopkins, 2002; Hartgraves & Benston, 2002; Hoyle, Schaefer, & Doupnik, 2011).

In contrast, IAS 27 (2003) defines control for consolidation using the principle of whether an investing entity has effective control over the financial and operating decisions of the voting entities. For instance, according to Paragraph 13 of IAS 27, control can exist even when the parent has less than 50% ownership. Control can exist when the parent has the power to appoint or remove the majority of the members of the board of directors or the equivalent governing body and control of the entity is by that board or body or when it has the power to cast the majority of votes at meetings of the board of directors or an equivalent governing body and control of the entity is by that board or body. Therefore, compared with ARB 51, IAS 27 creates barriers for firms which use a simple ownership rule to circumvent consolidation (Hsu & Pourjalali, 2014).

In this study, we examine whether the consolidation criteria defined in IAS 27, which reflect a broader and more comprehensive definition of control, increase the ability of external capital providers to monitor managerial investment activities. We expect that adopting effective control under IAS 27 will improve investment efficiency to a larger extent than adopting financial control under ARB 51 for the following two reasons. First, prior studies (e.g., Hsu & Pourjalali, 2014) find that adopting IAS 27 helps improve earnings quality and reduce abnormal intercompany transactions. By imposing the broader definition of control under IAS 27, more affiliated companies should be consolidated and the corresponding profits through affiliated transactions be eliminated in full. This requirement should improve the precision of consolidated earnings information by reducing managers’ ability to manage earnings through affiliated transactions. Consistent with the findings of Biddle, Hilary, and Verdi (2009), we expect a higher quality of consolidated statements under IAS 27 to increase investment efficiency.

The second reason is that a broader definition of control may curb managerial incentives to engage in value destroying activities such as empire building in investees with ample capital. In scandals such as at Enron, firms keep investees’ assets and liabilities off the balance sheet, which allows them to hide the risks associated with investees’ debts (Ketz, 2003). In addition, the parent firm can use one investee’s assets as collateral to raise funds and then divert the funds to other investees (Lamont, 1997; Rajan, Servaes, & Zingales, 2000; Scharfstein, 1998). Concurrently, if rational investors anticipate this tendency, they will increase the cost of capital, or ration capital, which may lead financially constrained firms to underinvest. Only when these investees are properly included in the consolidation can accounting information be more aligned with the true financial condition. Thus, we hypothesize that adoption of the control concept under IAS 27 leads to more efficient firm-level investment activities by reducing both over- and underinvestment.

To test whether the adoption of the broader and more comprehensive control concept in IAS 27 improves firms’ investment efficiency, we use a change in Taiwanese accounting standards as a natural experiment with an exogenous shock. All Taiwanese public firms are required to prepare consolidated financial statements in accordance with Taiwan’s Statement of Financial Accounting Standards No. 7, Consolidated Financial Statements (TSFAS 7). This statement is similar to ARB No. 51 and its revision, FAS 94, before 2005 and to IAS 27 after 2005 (Hsu & Pourjalali, 2014). Taiwan’s adoption of IAS 27 provides a unique setting in that it facilitates the contrast of consolidation methods between U.S. generally accepted accounting principles (GAAP) and International Financial Reporting Standards (IFRS). While a large body of research (e.g., Florou & Pope, 2012; Li, 2010) examines the economic consequences of a switch from non-U.S. domestic accounting standards to IFRS, few studies directly test whether firms using U.S. GAAP experience similar benefits from adopting IFRS. For example, Barth, Landsman, Lang, and Williams (2012) compare the value relevance of IFRS with U.S. GAAP between 1995 and 2006 by matching firms domiciled in 27 countries that adopted IFRS with U.S. firms using U.S. GAAP. However, the two sets of firms may not be comparable because IFRS-adopting countries and U.S. firms operate in different socioeconomic and legal environments. Our setting allows us to avoid the issue by examining the economic consequence of a switch from U.S. GAAP to IFRS using firms themselves as a control. Consequently, the change in the consolidation standard in Taiwan provides a more powerful setting to test whether the effective control definition under IAS 27 improves firm-level investment efficiency.

We compute the amount of investment as the sum of capital expenditure and R&D expenditure. Following the literature (e.g., Biddle et al., 2009), we perform both conditional and unconditional tests to examine whether investment is more efficient in the post-IAS 27 adoption period than in the pre-IAS 27 adoption period. We define the post-IAS 27 (pre-IAS 27) adoption period as firm-year observations occurring between 2006 and 2008 (between 2001 and 2003).¹ We remove years 2004 and 2005, which we consider transitional years, to reduce measurement errors.

For the conditional tests, we expect that firms subject to overinvestment (underinvestment) are more likely to reduce (increase) investment toward the optimal level after the adoption of IAS 27, compared with the pre-IAS 27 period. Our empirical results are consistent with our prediction, suggesting that after adoption of the control concept in IAS 27, investment efficiency improved by reducing both over- and underinvestment. For the unconditional tests, we first estimate investment efficiency by estimating a model of investment as a function of growth opportunities (i.e., sales growth) and use the residuals from the estimation. A positive (negative) residual is considered overinvestment (underinvestment). Higher absolute values of residuals (positive or negative) indicate more deviation from the optimal level. Then, we examine the extent to which overinvestment (underinvestment) is reduced after the adoption of IAS 27. Again, consistent with our prediction, we find that the adoption of IAS 27 leads to more efficient investment decisions by managers of Taiwanese firms. Our results are robust to the inclusion of various control variables such as firm size; market-to-book ratio; standard deviations of cash flows from operation, sales, and investment; Altman’s Z score (Altman, 1968); asset tangibility; industry-level leverage; cash flows from operation scaled by sales; dividend-paying status; firm age; operating cycle; loss occurrence; analyst coverage; and foreign investors’ shareholdings.

To test whether an increase in investment efficiency after the adoption of IAS 27 is concentrated in firms directly affected by IAS 27 adoption, we first separate our sample firms into two groups: the change group and the no-effect group. We assign a firm to the change group if it reclassifies a less-than-50%-owned investee from an associate during the pre-IAS 27 period to a subsidiary in the post-IAS 27 period. We assign a firm to the no-effect group if it does not have such a reclassification. If our findings based on the entire sample are concentrated in the change group, this result strengthens our claim that the positive relation between the adoption of IAS 27 and investment efficiency is less likely driven by macroeconomic or other regulatory factors. However, if we report similar positive relations between the adoption of IAS 27 and investment efficiency in both groups, our results may be due to common macroeconomic shocks. Consistent with our hypothesis, we find that managers’ investment decisions are more efficient, that is, both under- and overinvestment behaviors decrease, after IAS 27 became effective only in the change group where the impact of the adoption of IAS 27 is significant. In contrast, firms in the no-effect group do not experience any improvement in investment after the adoption of IAS 27, further supporting our hypothesis and ruling out the possibility that our results are driven by other concurrent economic or regulatory events.

We also examine whether the adoption of IAS 27 leads to an increase in foreign investors’ demand for equities of Taiwanese firms as one potential benefit from adopting the broader and more comprehensive control concept. We hypothesize that foreign investors increase their stock ownership after the adoption of IAS 27 for two reasons. First, IAS 27 provides more transparent and useful information (Hsu & Pourjalali, 2014) and thus potentially reduces home bias significantly, resulting in an increase in foreign investors’ shareholdings (Aggarwal, Klapper, & Wysocki, 2005; Covrig, DeFond, & Hung, 2007). Second, as more efficient investment usually increases accounting performance and firm value (Jung, Lee, & Weber, 2014; Titman, Wei, & Xie, 2004) and foreign investors generally prefer firms with good accounting performance (e.g., Kang & Stulz, 1997), IAS 27 may also increase foreign investors’ shareholdings through increased investment efficiency. Consistent with our prediction, we find that stock ownership by foreign investors increases significantly after IAS 27 was adopted in Taiwan. Similar to our results on investment efficiency, the increase in foreign investors’ ownership is more pronounced for firms in the change group. In contrast, we find no significant increase in foreign investors’ shareholdings for firms in the no-effect group, the firms that are not significantly affected by the adoption of IAS 27.

This article contributes to the existing literature in several ways. First, by using the exogenous shock from the adoption of IAS 27, our study shows that the more comprehensive and broader control concept has a potential to improve managers’ investment decisions. These findings have significant implications for the consolidation project between the FASB and IASB.² Second, while contemporary studies similarly investigate whether IFRS improves firm-level investment efficiency (e.g., Biddle, Callahan, Hong, & Knowles, 2013; Chen, Young, & Zhuang, 2013), the nature of their research design leads them to analyze the aggregate effect of accounting standards on investment. In contrast, we focus on a specific standard, IAS 27, to understand the implications of a specific policy choice—the control concept for consolidated financial statements—for investment efficiency. Third, prior studies examining the impact of accounting quality on investment efficiency (e.g., Biddle & Hilary, 2006; Biddle et al., 2009) suffer from the endogeneity problem to some extent; that is, it is plausible that firms with higher investment efficiency are more likely to have better financial reporting quality as better operating performance may make earnings management unnecessary. In contrast, by utilizing the exogenous shock from the adoption of IAS 27 and subsamples where the impact of IAS 27 adoption is distinct between groups, we minimize this endogeneity concern.

The remainder of this article is organized as follows: The “Prior Literature and Hypothesis Development” section reviews prior literature on the definition of control for, and the value relevance of, consolidated statements. This section also provides the institutional background on accounting standards for consolidation requirements in Taiwan, followed by the development of the research hypotheses. The research design and sampling procedures are given in the “Data and Descriptive Statistics” section, “Empirical Results” section reports the empirical results, “Sensitivity Analysis” section discusses the sensitivity tests, and “Summary and Conclusion” section provides the summary and concluding remarks.

Prior Literature and Hypothesis Development

Consolidation and the Equity Method

Many firms have strong incentives to maintain technical eligibility to use the equity method for some investees rather than full consolidation (e.g., Hoyle et al., 2011). One main reason is that relative to consolidation, the equity method masks inefficient cross-subsidization within the firm, as has often been documented (e.g., Rajan et al., 2000; Shin & Stulz, 1998). Companies can also keep investees’ assets and liabilities off the balance sheet, which allows them to hide the risks associated with investees’ debts (Ketz, 2003).³ Under the equity method (used for significant investments—usually ranging from 20% to 50% of the entity’s equity), the unrealized income from intercompany transactions is fractionally eliminated, only to the extent of the investor’s ownership (American Institute of Certified Public Accountants, 1971, para. 22, IAS 28). Thus, if firms can manage the control criteria such that firms can claim that they do not have control over investees and do not need to consolidate them, they have more latitude to hide economic performance in consolidated statements. For instance, the Coca-Cola Company, a U.S.-based firm, structured many of its investments in companies at just below a 50% ownership level, thereby avoiding full consolidation (Hartgraves & Benston, 2002). We review two control criteria under U.S. GAAP and IFRS in the following section.

The Control Concept for Consolidation Under ARB 51 and IAS 27

ARB 51 concludes that consolidation should be based on financial control (FASB, 1987, para. 20), which requires U.S. companies to fully consolidate all majority-owned voting entities into their financial statements. However, in the absence of a definite and comprehensive pronouncement on control, most firms use majority ownership as a general policy for consolidation and manage the 50% threshold to omit some subsidiaries from consolidation (e.g., Blacconiere & Hopkins, 2002; Hartgraves & Benston, 2002; Hoyle et al., 2011).

The bright-line tests under ARB 51 have invited opportunistic interpretation by corporate executives. In the case of the Coca-Cola Company, for example, Coca-Cola Enterprises should be an integral part of the company because Coca-Cola can determine all the terms for the affiliated transactions with enterprises. Ketz (2003) and Atanasov, Boone, and Haushalter (2010) show that by avoiding consolidation, Coca-Cola is able to inflate its accounting performance without fully removing abnormal intercompany transactions.

However, International Accounting Standard No. 27 (IAS 27) defines consolidation criteria using the principle of whether an investing entity has effective control over the financial and operating decisions of the voting entities. In Paragraph 4, “Control is the power to govern the financial and operating policies of voting entities and is presumed to exist when the parent owns more than half of the voting power directly or indirectly.” However, control can also exist when the parent has less than 50% ownership (Paragraph 13 of IAS 27): It exists when the parent has (a) power over more than half of the voting rights by virtue of an agreement with other investors, (b) power to govern the financial and operating policies of the entity under a statute or an agreement, (c) power to appoint or remove the majority of the members of the board of directors or equivalent governing body and control of the entity is by that board or body, or (d) power to cast the majority of votes at meetings of the board of directors or an equivalent governing body and control of the entity is by that board or body.

As evidenced above, compared with ARB 51/FAS 94, IAS 27 broadens the definition of a subsidiary, requiring more controlled investees to be accounted for in consolidated statements. Hsu, Duh, and Cheng (2012) find that consolidated statements are more value relevant under IAS 27 than under ARB 51. Hsu and Pourjalali (2014) further find that adopting IAS 27 decreases managers’ ability to manage earnings through affiliated transactions and reduces the likelihood that unrealized losses/incomes of affiliated transactions are included in financial statements. They also find that exclusion/reduction of abnormal affiliated transactions in consolidated financial statements enhances investors’ ability to predict future earnings, and they find that future earnings response coefficients (FERCs) improve significantly as a result of the adoption of IAS 27. However, neither study explores whether consolidated financial statements based on effective control reduce a firm’s agency problems, such as adverse selection and moral hazard. Our study attempts to fill the gap by examining the real impact of the control concept for consolidations on managers’ investment decisions.

The Adoption of IAS 27 in Taiwan

All Taiwanese public firms are required to prepare consolidated financial statements in accordance with TSFAS 7. Taiwanese financial accounting standards followed U.S. GAAP as their primary reference from their inception in 1984 until the IFRS convergence projects in the late 1990s. To converge with international accounting standards, TSFAS No. 7 was revised to shift from ARB 51/FAS 94 to IAS 27 in 2004 with an implementation date of January 1, 2005. Consequently, TSFAS 7 is similar to U.S. GAAP’s ARB 51 and its revision, FAS 94, before 2005 and is similar to IAS 27 after 2005.

Under ARB 51, consolidation was required only when a firm had a controlling financial interest over investees, which was usually interpreted as direct or indirect majority ownership. Under IAS 27, however, consolidation for voting entities is required when the firm has decision-making power over both financial benefits and operating decisions. We find no other basis for consolidation in Taiwan before 2005 than the use of 50% of the outstanding voting shares as a threshold. After the adoption of IAS 27, the new basis for consolidation (the effective control concept) more than doubled the number of reported subsidiaries in consolidated financial statements for a substantial majority of Taiwanese public companies (Hsu & Pourjalali, 2014).

Hypothesis Development

Prior literature (e.g., Stein, 2003) argues that information asymmetry between managers and outside suppliers of capital can result in agency problems such as adverse selection and moral hazard, both of which affect investment efficiency. We examine whether adopting IAS 27 mitigates these agency problems and thus increases investment efficiency.

First, we expect that IAS 27 mitigates adverse selection because consolidated financial statements under IAS 27 increase the precision of accounting information more than those under ARB 51. Ideally, consolidation imposes reporting constraints on firms by reducing their ability to hide costs and risks associated with the investee or to report fictitious gains through affiliated transactions (Bauman, 2003; Ketz, 2003). Compared with ARB 51/FAS 94, IAS 27 broadens the definition of a subsidiary, requiring additional controlled investees to be accounted for. Thus, adopting IAS 27 may reduce managers’ ability to manage earnings through affiliated transactions, as it increases the likelihood that unrealized losses/incomes from affiliated transactions are eliminated.

Consistent with this view, Hsu and Pourjalali (2014) find that the inclusion of more investees under IAS 27 decreases abnormal affiliated transactions and increases earnings quality, which is reflected in the FERC (forward earnings response coefficient). Thus, with the improved quality of accounting information, the extent of information asymmetry and hence adverse selection is expected to decrease. A reduction in adverse selection lowers the likelihood that managers will issue capital when their firm is overvalued and as a result will discourage managers from overinvesting (Biddle et al., 2009). In addition, when the precision of accounting information is improved, rational investors perceive the firm to be less risky and require a lower rate of return, which in turn decreases the cost of capital for the firm (Francis, LaFond, Olsson, & Schipper, 2004; Lambert, Leuz, & Verrecchia, 2007). This lower cost of capital reduces the likelihood of underinvestment because managers have better access to funding when positive net present value investment opportunities are available (Myers & Majluf, 1984).

Second, we expect that adopting IAS 27 also directly reduces moral hazard problems. A parent company may structure transactions between itself and its investees in a way that allows profits to be shifted from the investees to the parent company or allows costs to be shifted from the parent company to its affiliate (e.g., Thomas, Herrmann, & Inoue, 2004). Prior studies also document evidence of inefficient cross-subsidization within the firm (e.g., Rajan et al., 2000; Shin & Stulz, 1998). Firms can also keep investees’ assets and liabilities off the balance sheet and hide the risks associated with investees’ debts (Ketz, 2003), potentially leading to overinvestment. Not only can inappropriate financial reporting practices exist between a parent company and investees but also improper real activities. For example, the parent firm can easily raise funds by using one investee’s assets as collateral (Lamont, 1997; Rajan et al., 2000; Scharfstein, 1998), again resulting in a potential overinvestment problem. If the broader definition of control under IAS 27 produces higher quality information which assists investors in more careful monitoring of managers (Bushman & Smith, 2001), it may curb managerial incentives to engage in overinvestment problems such as empire building. At the same time, if rational investors anticipate this tendency, they will increase the cost of capital, or ration capital to price-protect themselves, leading to underinvestment. If adopting IAS 27 reduces the above behaviors, outside suppliers of capital, anticipating the decrease in affiliated transactions, will not ration capital ex ante and the underinvestment problem will decrease. Based on the above arguments, we hypothesize that investment efficiency improves after IAS 27 adoption, on average for Taiwanese firms as follows:

Hypothesis 1a: After the adoption of IAS 27, the extent of overinvestment is reduced.

Hypothesis 1b: After the adoption of IAS 27, the extent of underinvestment is reduced.

However, the impact of IAS 27 on investment efficiency is not necessarily uniform across all Taiwanese firms. We believe that IAS 27 adoption is particularly relevant to firms with at least one less-than-50%-owned investee which was reported to be an associate (noncontrolled investee) before IAS 27, but was reported as a subsidiary (controlled investee) after IAS 27 without an increase in ownership. Thus, we expect that investment efficiency improvement exists only for such firms, after the adoption of IAS 27. This leads to our second hypothesis:

Hypothesis 2a: The extent of overinvestment is reduced only in firms affected by IAS 27 adoption.

Hypothesis 2b: The extent of underinvestment is reduced only in firms affected by IAS 27 adoption.

For the third hypothesis, we examine whether the economic benefits from adopting IAS 27 are reflected in the attraction of foreign capital. Specifically, following the recent literature in this area, we focus on the effect of IAS 27 adoption on a change in foreign investors’ shareholdings and predict that the adoption of IAS 27 increases foreign investors’ shareholdings for two reasons.

First, better financial reporting quality, including better disclosure quality, and thus lower information asymmetry between managers and shareholders after adopting IAS 27 may significantly increase foreign investors’ shareholdings. Florou and Pope (2012) examine the relation between mandatory adoption of IFRS and institutional investment decisions and show that based on an international sample of 10,852 unique firms from 45 countries from 2003 to 2006, institutional investor demand for equities increased for IFRS adopters.⁴ More related, Covrig et al. (2007) show that one of the results of voluntary IAS adoption is more information which is more familiar to foreign investors and thus reduces home bias among them, resulting in an increase in foreign mutual fund ownership. Baik, Kang, Kim, and Lee (2013) also show that foreign institutional investors prefer low information asymmetry stocks more than domestic institutional investors do because foreign investors suffer from home bias. If the adoption of IAS 27 significantly reduces home bias by reducing information asymmetry and agency costs, foreign investors will increase their shareholdings in Taiwanese firms.

Second, as predicted in the first hypothesis, if the adoption of IAS 27 improves investment efficiency, then the firm will attract more foreign capital as more efficient investment enhances future operating performance and firm value (Jung et al., 2014; Titman et al., 2004) and foreign investors prefer a strong accounting performance (Kang & Stulz, 1997).

In sum, we expect that an improvement in financial reporting quality and more efficient investment decisions by managers after the adoption of IAS 27 lead to more stock ownership by foreign investors and, consistent with our second hypothesis, the effect of IAS 27 adoption on foreign investor shareholdings is more pronounced for firms directly affected by IAS 27. This leads to the following hypothesis:

Hypothesis 3a: After the adoption of IAS 27, stock ownership by foreign investors increases.

Hypothesis 3b: After the adoption of IAS 27, stock ownership by foreign investors increases only in firms affected by IAS 27 adoption.

Data and Descriptive Statistics

Sample Composition and Data

Our study is based on all the nonfinancial firms listed on the Taiwan Stock Exchange (TSE) during the period 2001 to 2008. All accounting and market data are collected from the Taiwan Economic Journal (TEJ) database. Our sample selection begins with all listed firms having “associate” and “subsidiary” data available in the TEJ database for any year from 2001 to 2008. We define 2001-2003 as the pre-IAS 27 period and 2006-2008 as the post-IAS 27 period. We exclude 2004 and 2005 from our sample to clearly separate before adoption from after adoption of IAS 27; these 2 years are considered a transitional period. We use multiple years in the pre- and postperiods to increase the power of the tests and to minimize the effect that any 1 year’s (possibly unrepresentative) data might have on our results.

Our sample selection process begins with 4,228 firm-year observations and 751 firms for the period 2001 to 2008. We eliminate 916 firm-year observations for which the data for the ownership of subsidiaries and associates are either missing or incomplete, 245 firm-year observations for firms that made new acquisitions after IAS 27, 459 firm-year observations for which the required data for investment efficiency are not available, 94 firm-year observations for which the foreign investor information is not available, and 125 firm-year observations of firms that do not have at least one observation in both the pre- and post-IAS 27 periods. To mitigate the effect of outliers on the results, we winsorize all the continuous variables at their 1% and 99% levels. The final sample contains 2,389 firm-year observations for 420 unique firms.

Descriptive Statistics

Table 1, Panel A, provides information on sample distribution before IAS 27 adoption (from 2001 to 2003) and after IAS 27 adoption (from 2006 to 2008). From our sample of 2,389 firm-year observations, 1,165 (1,224) firm-year observations are assigned to the pre-IAS 27 (post-IAS 27) period. To minimize the impact of potentially confounding macroeconomic factors, we separate our sample into the change group and the no-effect group. A firm is put into the change sample if it is significantly affected by IAS 27 and into the no-effect group if it is not affected by IAS 27. The number of observations in the change group is 1,769, whereas the number of observations in the no-effect group is 620. Panel B shows the difference in sample distribution between the change group and the no-effect group by industry composition. In both groups, about 50% of observations are concentrated in the electronics industry, and the difference between the two groups is not economically significant. Other than the electronics industry, no industry accounts for more than 10% of the observations, suggesting that beyond electronics, our observations are not clustered in certain industries. More importantly, industry composition between the change and no-effect groups is not significantly different, implying that the differences between the two groups are unlikely to be due to industry characteristics.

Table 1.

Sample^a Distribution Before and After IAS 27.

Panel A: Sample Distribution by Year.
Sample	Pre-IAS 27 period				Post-IAS 27 period				whole period observations
Sample	2001	2002	2003	Total	2006	2007	2008	Total
Full	357	398	410	1,165	406	418	400	1,224	2,389
Change^b	270	300	298	868	307	304	290	901	1,769
No-effect	87	98	112	297	99	114	110	323	620
Panel B: Sample Distribution by Industry Code.
				Change sample		No-effect sample		Difference test between two groups (%)
Industry				Observations	%	Observations	%
Cement (11)				28	1.58	2	0.32	1.26
Food (12)				57	3.22	14	2.26	0.9
Plastics (13)				70	3.96	15	2.42	1.54
Textile (14)				98	5.54	34	5.48	0.06
Electronic and machinery (15)				86	4.86	35	5.65	−0.79
Appliance and cable (16)				26	1.47	14	2.26	−0.79
Chemicals (17)				49	2.77	51	8.23	−5.46
Glass (18)				22	1.24	1	0.16	1.08
Paper (19)				22	1.24	3	0.48	0.76
Steel and iron (20)				64	3.62	24	3.87	−0.25
Rubber (21)				27	1.53	7	1.13	0.40
Automobile (22)				22	1.24	1	0.16	1.08
Electronics (23)				871	49.24	301	48.55	0.69
Construction (25)				81	4.58	40	6.45	−1.87
Transportation (26)				56	3.17	12	1.94	1.23
Tourism (27)				8	0.45	6	0.97	−0.52
Wholesale (29)				34	1.92	8	1.29	0.63
Utilities (97)				10	0.57	12	1.94	−1.37
Others (99)				138	7.80	40	6.45	1.35
Total				1,769	100	620	100	χ² = 19.39 (p = .15)

Note. IAS 27 = International Accounting Standard No. 27.

All firms were listed on the Taiwan Stock Exchange from 2001 to 2008 (excluding 2004-2005) and had available data for all analyses and did not acquire new equity stakes during the sample period.

We split our firms into a change sample (1,817 observations) if they reclassify a less-than-50%-owned investee from an associate during the pre-IAS 27 period to a subsidiary in the post-IAS 27 period, and a no-effect sample (651 observations) if they do not have such reclassifications.

Table 2 presents descriptive statistics for our variables. One of the key variables is the amount of investment, INV. For the change sample, INV has a mean of 8.54% and a median of 6.06% in the pre-IAS 27 period. In the post-IAS 27 period, the mean (median) of INV is 8% (6.17%). The difference in INV between pre-IAS 27 and post-IAS 27 periods is significant at the 5% level. However, this statistic is not very meaningful, as we need to consider the likelihood of over- or underinvestment for our analysis. The mean (median) of foreign investor ownership pre-IAS 27 is 8.16% (3.08%), whereas the mean (median) of foreign investor ownership increases dramatically, to 15.83% (10.83%) post-IAS 27. The difference between the two periods is significant (p = .001), confirming our hypothesis that foreign investors hold more equity after the adoption of IAS 27. In contrast, for the no-effect sample, we observe no change in the mean of INV between the two periods. While we find that the mean of foreign investor ownership increases from 10.99% to 11.18% after the adoption of IAS 27, the difference is not significant.

Table 2.

Descriptive Statistics.

			Change sample (n = 1,769)				Difference in mean	No-effect sample (n = 620)				Difference in mean
	Total sample		2001-2003		2006-2008			2001-2003		2006-2008
	M	Median	M	Median	M	Median		M	Median	M	Median
INV	8.06	4.99	8.54	6.06	8.00	6.17	−0.54**	7.45	4.81	7.48	4.68	0.03
OverInvest	0.48	0.51	0.40	0.39	0.49	0.50	0.09	0.44	0.44	0.57	0.56	0.13
Analyst	3.69	3.21	3.76	4.00	3.66	3.00	−0.10	3.56	3.00	3.62	3.00	0.06
FINV	11.81	9.54	8.16	3.08	15.83	10.83	7.67***	10.99	8.57	11.18	9.27	0.19
SIZE	15.68	15.15	15.84	15.74	16.01	15.92	0.17*	14.91	14.81	14.89	14.94	−0.02
MB	1.64	0.98	1.51	0.80	1.45	0.98	−0.06*	1.60	0.80	1.71	1.05	0.11*
ZSCORE	2.31	2.13	2.20	1.81	2.35	2.13	0.15	2.35	1.97	2.36	2.38	0.01
CYCLE	4.98	4.97	5.00	4.99	4.94	4.95	−0.06	5.04	5.03	4.93	4.96	−0.11
LOSS	0.23	0.00	0.21	0.00	0.20	0.00	−0.01	0.23	0.00	0.16	0.00	−0.07
TAN	0.31	0.29	0.32	0.30	0.29	0.27	−0.03	0.34	0.33	0.29	0.26	−0.05
MEAN_LEV	0.14	0.13	0.17	0.15	0.12	0.10	−0.05*	0.17	0.15	0.11	0.10	−0.06*
CFO_SALE	0.09	0.07	0.11	0.07	0.09	0.07	−0.02	0.09	0.07	0.10	0.06	0.01
VOL_OCF	0.07	0.06	0.07	0.05	0.07	0.06	−0.00	0.07	0.05	0.08	0.07	0.01
VOL_SALE	0.20	0.15	0.18	0.13	0.23	0.17	0.05*	0.16	0.11	0.22	0.18	0.06*
VOL_INV	0.07	0.05	0.08	0.06	0.08	0.06	−0.00	0.07	0.05	0.06	0.04	−0.01
AGE	8.28	6.00	6.93	4.00	10.79	8.00	3.86***	4.81	3.00	7.97	5.00	3.16***
DIV	0.58	1.00	0.47	0.00	0.71	1.00	0.24**	0.48	0.00	0.61	1.00	0.13**

Note. For variable definition, see the appendix.

, **, and *** indicate significance at 10%, 5%, and 1% levels, respectively, in a two-tailed test.

Turning to other variables, we do not find any pronounced differences between pre-IAS 27 and post-IAS 27 in the change sample except in DIV, an indicator variable for dividend payer.⁵ The mean of DIV is 47% pre-IAS 27, but increases to 71% post-IAS 27. The difference is significant at the 5% level. We find similar observations for the no-effect sample. The mean of DIV is 48% pre-IAS 27, but increases to 61% post-IAS 27. Although the difference is significant, the magnitude of increase (= 13%) is lower than that (= 24%) for the change sample.

Empirical Results

Investment Efficiency After IAS 27 Adoption for the Change and No-Effect Subsamples

Conditional tests

Our first hypothesis suggests that the adoption of IAS 27 improves investment efficiency. We test this hypothesis using the following regression (1):

\begin{matrix} {INV}_{i t} = α + β_{1} O v e r I n v e s t_{i t - 1} + β_{2} P O S T_{i t - 1} + β_{3} P O S T_{i t - 1} \times O v e r I n v e s t_{i t - 1} + β_{4} {Analyst}_{i t - 1} + β_{5} {Analyst}_{i t - 1} \times O v e r I n v e s t_{i t - 1} \\ + β_{6} F I N V_{i t - 1} + β_{7} F I N V_{i t - 1} \times O v e r I n v e s t_{i t - 1} + \sum_{j = 1}^{13} ϒ_{j} C o n t r o l_{i j t - 1} + ε_{i t}, \end{matrix}

where INV is the sum of research and development, capital, and acquisition expenditures less the sale of property, plant, and equipment multiplied by 100 and scaled by the lagged total assets. OverInvest is the an ex-ante measure indicating the likelihood that a firm will under- or overinvest. This measure is based on two liquidity measures identified in the literature: cash holdings and leverage. The literature suggests that firms with higher cash reserves or lower leverage tend to overinvest. Following Biddle, Hilary, and Verdi (2009), we first rank each firm-year observation into deciles based on cash holdings and leverage, respectively. As leverage is opposite to cash holdings in terms of liquidity implications and thus the extent of overinvestment, we multiply leverage by −1 before ranking. Then, we rescale the rankings to range from 0 to 1 and average them, so that our composite measure still ranges from 0 to 1. In other words, firms in both the highest cash reserves decile and the lowest leverage decile have the value of 1 for OverInvest, suggesting that such firms are most likely to overinvest. POST is an indicator variable which is equal to 1 if a firm-year observation occurs between 2006 and 2008 and 0 if a firm-year observation occurs between 2001 and 2003. Analyst is the number of stock analysts following each firm. FINV is the foreign investors’ shareholdings.

Our interest is in the indicator variable POST and its interaction with OverInvest. If OverInvest equals 0, then firms are financially constrained and are more likely to underinvest. If adopting IAS 27 reduces underinvestment, then the coefficient on POST is expected to be positive. However, if OverInvest equals 1, then firms are unconstrained and are more likely to overinvest. In this case, if adopting IAS 27 reduces overinvestment, then the sum of the coefficients on POST and POST×OverInvest (i.e., $β_{2}$ + $β_{3}$ ) is expected to be negative.

Because of other factors also affecting managers’ investment decisions, a number of control variables are included in the multivariate analysis. Similar to Biddle et al. (2009), we control for the determinants of investment such as firm size; market-to-book ratio; standard deviations of CFO, sales, and investment; Altman’s Z score; tangibility; industry level-leverage; CFO/Sale; DIV; AGE; operating cycle; and loss occurrence. The control variables are defined as in Biddle et al. (2009) and also detailed in the appendix. Finally, we include industry dummies and year dummies to control for variations in investment across industries and over time.

The results of estimating Model (1) are presented in Table 3. We first estimate Model (1) with the entire sample, which enables us to examine the overall impact of IAS 27 adoption on firm investment behavior. As expected, the coefficient on POST is 2.243 and significantly positive (p = .001), supporting our first hypothesis that IAS 27 adoption increases investment when firms are more likely to underinvest, mitigating the underinvestment problem. The reduction in the likelihood of underinvestment is 2.243%, which represents about 25% of the average investment level of the sample. The coefficient on POST×OverInvest is significantly negative (p = .001). Furthermore, the sum of POST and POST×OverInvest is −1.215 and significantly negative as well (p = .01). Specifically, adopting IAS 27 reduces overinvestment by 1.215% of total assets in the total sample, representing about 15% of the average investment level. The results support our hypothesis that IAS 27 adoption reduces investment when firms are more likely to overinvest, alleviating the overinvestment problem.

Table 3.

Change in Investment Efficiency for the Change and No-Effect Samples—Conditional Tests.

	Predicted sign	(1) Full sample		(2) Change sample		(3) No-effect sample
Intercept		−17.517	(−4.78)***	−16.806	(−3.68)***	−13.258	(−2.02)*
OverInvest		2.618	(1.98)*	4.001	(1.94)*	−0.051	(−0.03)
POST		2.243	(4.15)***	2.624	(3.14)**	2.040	(2.33)*
OverInvest×POST		−3.458	(−3.31)***	−4.870	(−2.95)**	−1.822	(−1.07)
Analyst	+	0.462	(3.58)***	0.590	(2.62)**	0.396	(2.32)*
OverInvest×Analyst		0.060	(0.25)	0.007	(0.01)	0.347	(1.23)
FINV	+	0.063	(2.56)*	0.079	(2.26)**	0.098	(2.29)*
OverInvest×FINV	−	−0.088	(−2.12)**	−0.068	(−1.10)	−0.086	(−1.22)
SIZE	+	0.553	(3.86)***	0.582	(2.86)**	0.403	(1.43)
MB	+	0.106	(0.62)	0.609	(2.17)*	−0.162	(−0.85)
ZSCORE	−	−0.096	(−0.46)	−0.136	(−0.41)	−0.018	(−0.06)
CYCLE		0.750	(3.00)**	1.129	(3.03)**	0.498	(1.31)
LOSS	−	−1.483	(−4.56)***	−1.735	(−3.50)***	−1.475	(−2.68)**
TAN	+	15.746	(14.62)***	18.140	(11.36)***	14.902	(9.46)***
MEAN_LEV		−3.328	(−1.06)	−10.533	(−3.48)***	−11.010	(−1.99)*
CFO_SALE		1.445	(2.10)*	3.722	(2.60)**	−0.309	(−0.43)
VOL_OCF		2.351	(0.88)	−4.121	(−1.05)	−0.493	(−0.12)
VOL_SALE		2.227	(2.24)*	4.419	(3.04)**	1.998	(1.32)
VOL_INV		3.260	(1.15)	2.606	(0.64)	11.895	(2.21)*
AGE	−	−0.148	(−8.58)***	−0.157	(−8.07)***	−0.170	(−5.37)***
DIV	−	0.618	(2.39)**	0.908	(2.28)*	0.206	(0.52)
Test POST+OverInvest×POST	−	−1.215	(2.21)**	−2.246	(2.55)**	0.218	(0.54)
Industry and year dummies		Included		Included		Included
n		2,389		1,769		620
Adjusted R²		.342		.361		.338

Note. POST_it is an indicator variable equal to 1 (0) when the observation lies in the period 2006-2008 (2001-2003). Other variables are as defined in the appendix. We run the following regression:

Investment_i,t = α+β₁OverInvest_i,t−1+β₁POST_i,t−1+β₃POST_i,t−1×OverInvest_i,t−1+β₄Analyst_i,t−1+β₅AnalystCoverage_i,t−1×OverInvest_i,t−1+β₆FINV_i,t−1+β₇FINV_i,t−1×OverInvest_i,t−1+γ Control variables +ϵ_i,t,

where all variables including control variables (reported in the appendix) are as defined in Biddle, Hilary, and Verdi (2009). Control variables are firm size; market-to-book ratio; standard deviations of CFO, sales, and investment; Z score; tangibility; industry-level leverage; CFO/sales; dividend; age; operating cycle; and loss occurrence.

, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively, in a two-tailed test.

Turning to the effect of analyst coverage on investment efficiency, the coefficient on Analyst is significantly positive, suggesting that firms with more analyst coverage are less likely to underinvest. However, the coefficient on Analyst×OverInvest is insignificant, suggesting that analyst coverage does not mitigate the overinvestment problem in our sample firms. We also control for the effect of foreign investor shareholding on investment efficiency. The coefficient on FINV is again significantly positive, documenting the role of foreign investors in reducing underinvestment. The coefficient on FINV×OverInvest is significantly negative at the 5% level, implying that firms owned more by foreign investors are less likely to have an overinvestment problem. As a multicollinearity problem may exist in the model, we calculate variance inflation factors (VIFs) for all the variables in the model and find that all individual VIFs are less than 9, reducing this concern.⁶

To test the second hypothesis, we estimate Model (1) for two subsamples of firms: the change sample and the no-effect sample. Following Hsu and Pourjalali (2014), we categorize firms in the change sample if they have at least one less-than-50%-owned investee, which was reported in 2004 to be an associate (noncontrolled investee) but was reported in 2005 a subsidiary (controlled investee) without an increase in ownership. The change sample is our treatment sample. Firms in the no-effect sample classify an investee as an associate in 2004 and continue to classify the investee as an associate in 2005 with no change in ownership. Thus, IAS 27 should have no impact on these firms. The no-effect sample helps to control for the effect on investment efficiency of any macroeconomic and regulatory changes other than IAS 27 that occurred during our sample period. Thus, if the results based on the entire sample are driven by firms in the change sample and not the no-effect sample, it increases our confidence that the documented increase in investment efficiency from pre-IAS 27 to post-IAS 27 is mainly due to the impact of IAS 27.

We first estimate Model (1) only with observations in the change sample. Consistent with the results based on the entire sample, the coefficient on POST is significantly positive, whereas the coefficient on POST×OverInvest and the sum of the coefficient on POST and the coefficient on POST×OverInvest are significantly negative. In contrast, when estimating Model (1) with the no-effect sample, the coefficient on POST×OverInvest is insignificant, although the coefficient on POST is still significantly positive. These results suggest that the increased investment efficiency through reducing overinvestment is mainly due to the adoption of IAS 27. However, based on Model (1) conditional tests, it is not clear that the increase in investment efficiency through the reduction in underinvestment is due to the adoption of IAS 27.

The coefficients on other control variables are generally consistent with prior studies. Consistent with the literature on investment (Hoshi, Kashyap, & Scharfstein, 1991; Kang & Stulz, 2000; Lang, Ofek, & Stulz, 1996), the extent of investment is negatively (positively) related to loss occurrence, leverage, and firm age (size and dividend payment status). Similar to firms operating in the United States, in Taiwan, larger and more profitable firms are more likely to increase investment, while more leveraged and older firms tend to have a lower level of investment.

Unconditional tests

We also employ unconditional tests for the relation between IAS 27 and investment efficiency. We first measure investment efficiency by estimating the following model of investment as a function of sales growth⁷ and use the residuals from the estimation.

IN V_{it + 1} = α + β_{1} SGROWT H_{it - 1} + ε_{it + 1},

where SGROWTH_it−1 is sales growth from year t− 1 to year t− 2.

Equation 2 is estimated for each industry-year based on the TEJ industry classification for all industries with at least 20 available observations in a year. The residual of firm-year observations is quartile ranked. We form a variable (INVEFF) that takes the value of 1 if the residual is in the bottom quartile of the distributions, the value of 0 if it is in the middle two quartiles, and the value of 2 if it is in the top quartile. The bottom quartile of unexplained investment is classified as underinvestment (UNDER_INV) because these firms have the most negative residuals, whereas the top quartile is classified as overinvestment (OVER_INV), as the firms have the most positive residuals. We use observations in the middle two quartiles as the normal (or more optimal) investment group and estimate the multinomial logistic regression that tests the likelihood that a firm is less likely to be in the top quartile (overinvest) or in the bottom quartile of unexplained investment (underinvest) after adopting IAS 27. In the multinomial logistic model with three outcome categories (1, 0, 2) and a constant term (β) denoted by the vector x, two logit functions are described as follows (Hosmer & Lemeshow, 2000). The two logit regressions are considered simultaneously but separately for the likelihood of over- and underinvestment as dependent variables. We expect the coefficient on POST in Equation 3 to be negative.

\ln [\frac{P (Y = 1 | X)}{P (Y = 0 | X)}] = α + β_{11} POS T_{i} + β_{12} Analys t_{it - 1} + β_{13} FIN V_{it - 1} + \sum_{j = 1}^{13} ϒ_{1 j} Contro l_{ijt - 1} + ε_{it},

\ln [\frac{P (Y = 2 | X)}{P (Y = 0 | X)}] = α + β_{21} POS T_{i} + β_{22} Analys t_{it - 1} + β_{23} FIN V_{it - 1} + \sum_{j = 1}^{13} ϒ_{2 j} Contro l_{ijt - 1} + ε_{it},

where all explanatory and control variables are the same as in Equation 1.

Results are reported in Table 4 separately, depending on whether the dependent variable is underinvestment (UNDER_INV) or overinvestment (OVER_INV). Panel A presents the results of estimating model (3a) when the dependent variable is UNDER_INV. We first estimate Model (3a) with the entire sample. Similar to Panel A of Table 3, the coefficient on POST is significantly negative (coefficient = −0.441; p≤ .01), consistent with our hypothesis that the adoption of IAS 27 decreases underinvestment in Taiwanese firms. Similar to Table 3, to further strengthen the argument that our results are mainly due to the adoption of IAS 27, we reestimate Model (3a) with both subsamples: the change group and the no-effect group. We find that the coefficient on POST is significantly negative only for the change group (coefficient = −0.734; p≤ .01), whereas it is not significant for the no-effect subsample (coefficient = −0.031; p = .85). These results suggest that the significant and negative coefficient based on the entire sample is due to the change sample, mitigating the concern that our results are driven by some macroeconomic and regulatory changes.

Table 4.

Change in Investment Efficiency for the Change and No-Effect Samples—Unconditional Tests.

Panel A: Underinvestment Versus Normal Investment.
	(1) Full sample		(2) Change sample		(3) No-effect sample
Intercept	3.393	(2.84)**	3.301	(1.66)	0.917	(0.49)
POST	−0.441	(−3.30)***	−0.734	(−3.21)**	−0.032	(−0.15)
Analyst	−0.120	(−5.07)***	−0.093	(−2.42)*	−0.157	(−4.07)***
FINV	1.090	(0.19)	6.805	(0.96)	4.306	(0.33)
SIZE	−0.188	(−3.27)**	−0.139	(−1.50)	−0.018	(−0.18)
MB	0.137	(1.62)	−0.010	(−0.08)	0.174	(1.45)
ZSCORE	−0.140	(−1.25)	−0.213	(−1.11)	−0.155	(−0.87)
CYCLE	−0.023	(−0.24)	−0.021	(−0.13)	−0.119	(−0.74)
LOSS	0.302	(1.96)*	0.612	(2.44)*	0.389	(1.49)
TAN	−3.723	(−7.53)***	−5.367	(−6.04)***	−2.961	(−3.66)***
MEAN_LEV	1.994	(1.39)	0.591	(0.26)	4.111	(1.68)
CFO_SALE	−0.109	(−0.34)	−0.657	(−1.20)	−0.206	(−0.49)
VOL_OCF	2.058	(1.79)	2.746	(1.66)	1.767	(0.98)
VOL_SALE	1.243	(3.26)**	0.759	(1.23)	1.541	(2.46)*
VOL_INV	−1.247	(−1.14)	−0.317	(−0.19)	−3.555	(−1.85)
AGE	0.044	(5.82)***	0.036	(2.72)**	0.043	(3.42)***
DIV	−0.534	(−4.35)***	−0.591	(−3.00)**	−0.422	(−2.14)*
Industry and year dummies	Included		Included		Included
n	2,389		1,769		620
Adjusted R²	.227		.266		.257
Panel B: Overinvestment Versus Normal Investment.
	(1) Full sample		(2) Change sample		(3) No-effect sample
Intercept	−6.242	(−4.87)***	−5.223	(−2.70)**	−0.807	(−1.19)
POST	−0.277	(−2.57)**	−0.407	(−2.12)*	0.231	(1.29)
Analyst	0.077	(4.01)***	0.108	(3.47)***	0.092	(3.26)**
FINV	−4.509	(−0.62)	2.154	(0.27)	−7.120	(−0.66)
SIZE	0.049	(0.98)	0.054	(0.75)	0.073	(0.05)
MB	−0.037	(−0.46)	0.111	(0.83)	0.130	(0.19)
ZSCORE	0.241	(2.73)**	0.164	(0.94)	−0.184	(−0.26)
CYCLE	0.354	(2.61)**	0.288	(1.45)	−0.537	(−1.18)
LOSS	−0.598	(−3.43)***	−1.007	(−3.55)***	−0.686	(−2.34)*
TAN	5.609	(13.21)***	4.296	(5.79)***	3.035	(8.47)***
MEAN_LEV	0.090	(0.05)	0.416	(0.17)	−0.761	(0.06)
CFO_SALE	0.875	(2.15)*	0.231	(0.39)	0.929	(1.56)
VOL_OCF	0.249	(0.22)	0.639	(0.40)	−2.310	(−1.00)
VOL_SALE	0.843	(2.50)*	0.211	(0.42)	0.661	(2.62)**
VOL_INV	0.519	(0.60)	1.514	(1.16)	5.046	(0.88)
AGE	−0.076	(−5.94)***	−0.037	(−2.44)*	−0.090	(−4.16)***
DIV	0.360	(2.74)**	0.737	(3.84)***	0.326	(1.91)*
Industry and year dummies	Included		Included		Included
n	2,389		1,769		620
Adjusted R²	.227		.266		.257

Note. All variables are as defined in the appendix except UNDER_INVEST (or OVER_INVEST). We estimate a multinomial logistic regression that tests the likelihood that a firm might be in the extreme investment residual quartiles when POST is equal to 1. Following Biddle, Hilary, and Verdi (2009), we separately examine the likelihood of over- and underinvestment. UNDER_INVEST (or OVER_INVEST) is based on the unexplained portion (i.e., residual) of investment after running the following regression: INVEST_t = a+b×ΔSGROWTH_t−1. The residual of firm-year observations is quartile ranked. The bottom quartile of unexplained investment is classified as underinvestment (UNDER_INVEST), whereas the top quartile is classified as overinvestment (OVER_INVEST), and observations in the middle two quartiles are classified as normal (or more optimal) investments.

UNDER_INVEST (or OVER_INVEST)_i,t = α+β₁POST_i,t−1+β₂Analyst_i,t−1+β₃FINV_i,t−1+ Control variables +ϵ_i,t,

where all variables including control variables are defined as in Biddle et al. (2009).

, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively, in a two-tailed test.

The results of estimating Model (3b) with OVER_INV as the dependent variable are reported in Panel B. When the full sample is used, POST is also negatively associated with underinvestment (coefficient = −0.227; p = .001), suggesting that the adoption of IAS 27 promotes investments when the current level is lower than the optimal level based on sales growth. Similar to the results in Panel A, when Model (3b) is estimated for both the change group and the no-effect group, only in the change group is the coefficient on POST significantly negative (coefficient = −0.407; p = .03). The coefficient on POST is not significant when Model (3b) is estimated with the no-effect sample (p = .19).

Turning to the control variables, we explain the results when the dependent variable is OVER_INV as results based on UNDER_INV can be interpreted as opposite to those based on OVER_INV. We first find that the coefficient on AnalystCoverage is significantly positive, suggesting that stock analysts exacerbate the overinvestment problem. The coefficient on FINV is insignificant, implying that foreign investors do not significantly influence overinvestment. The extent of overinvestment is smaller for loss firms and older firms, while it is greater for financially healthy firms, firms with a longer operating cycle, firms with more tangible assets, firms with higher volatility of sales, and dividend-paying firms. These results are generally consistent with the literature.

In sum, consistent with our hypothesis, we find that IAS 27 adoption improves investment efficiency by reducing overinvestment or underinvestment toward optimal levels of investment. These results exist more strongly in the change sample, further strengthening our conclusion.

The Impact of IAS 27 Adoption on Foreign Investors

We also examine the impact of IAS 27 adoption on foreign investors. As we predict a positive impact of IAS 27 adoption on foreign investors’ shareholdings, we expect the coefficient on POST to be significantly positive when estimating the following Model (4):

Δ FIN V_{i, t} = α + β_{1} POS T_{i, t - 1} + β_{2} Δ SIZ E_{i, t - 1} + β_{3} Δ LE V_{i, t - 1} + β_{4} Δ Analys {t_{i}}_{, t - 1} + β_{5} Δ M B_{i, t - 1} + β_{6} Δ RO A_{i, t - 1} + β_{7} Δ EARNST D_{i, t - 1} + ε_{i, t} .

We also include control variables which could affect foreign investors’ shareholdings. Similar to the literature (e.g., Florou & Pope, 2012), we include a change in analyst coverage to control for the richness of the information environment. We expect that firms with more analyst coverage attract more foreign investment as such firms’ information environment is better, and thus the home bias of foreign investors is reduced. As firm size, SIZE, reflects several aspects of firm characteristics according to the literature, such as the information environment, risk, and liquidity, we have no specific prediction for the SIZE variable. Similar to Hansen, Miletkov, and Wintoki (2013), we also include changes in leverage, market-to-book ratio, and return on assets (ROA). Finally, a change in earnings volatility is included to control for foreign investors’ preference for a smoother earnings path. As foreign investors suffer from home bias due to information asymmetry between managers and foreign investors (Ahearne, Griever, & Warnock, 2004), foreign investors may prefer smoother earnings, which improve earnings informativeness (Tucker & Zarowin, 2006) and thus reduce information asymmetry.

Table 5 presents the results of estimating Model (4). Similar to our previous analysis, we first estimate Model (4) based on the entire sample. Consistent with our hypothesis, we find that the coefficient on POST is significantly positive (p≤ .01). We further estimate Model (4) for both subsamples: the change subsample and the no-effect subsample. Again, the positive impact of IAS 27 adoption on foreign ownership exists only in the change subsample. When estimating Model (4) for the change sample, the coefficient on POST is significantly positive (coefficient = 0.005; p < .01). In contrast, the coefficient on POST is insignificant in the no-effect sample. Coefficients of most control variables are not significant, possibly because of low variations between periods.

Table 5.

Pattern in Foreign Ownership Before and After Adoption of IAS 27.

The change in foreign institutional investors	(1) Full sample		(2) Change sample		(3) No-effect sample
Intercept	0.007	(0.61)	0.032	(1.95)	−0.022	(−0.53)
POST_it	0.004	(2.13)*	0.005	(1.94)*	−0.001	(−0.32)
ΔSIZE_it	0.000	(0.26)	−0.001	(−1.50)	0.000	(0.15)
ΔLEV_it	0.007	(1.36)	0.003	(0.33)	0.000	(0.01)
ΔAnalyst_it	0.000	(0.28)	0.001	(1.79)	−0.001	(−0.63)
ΔMB_it	0.000	(0.43)	−0.002	(−2.13)*	0.003	(1.74)
ΔROA_it	0.008	(0.68)	0.037	(1.24)	0.005	(0.14)
ΔEARNSTD_it	0.015	(0.68)	0.045	(0.96)	−0.022	(−0.65)
Industry and year dummies	Included		Included		Included
n	2,389		1,769		620
Adjusted R²	.047		.026		.012

Note. t statistics are given in parentheses. t statistics are calculated using clustered standard errors by firm and year (Petersen, 2009). For variable definitions, see the appendix. IAS 27 = International Accounting Standard No. 27.

, **, and *** indicate 10%, 5%, and 1% levels of confidence.

Sensitivity Analysis

Alternative Measures of Investment Efficiency for Unconditional Tests

Our primary results for the first hypothesis are based on the measure developed by Biddle et al. (2009). Although this is a popular measure in the accounting literature, it requires a strong assumption—that sales growth is a good predictor of future investment—which may be subject to a measurement error problem. To mitigate this concern, we also use the following alternative measures. For the first alternative measure, we consider the industry median of investment as the level of normal investment. When a firm has a higher (lower) investment than the industry median or when the difference between the firm’s investment and its industry median investment is positive (negative), it is considered more likely to overinvest (underinvest). Similar to our primary measures of UNDER_INV and OVER_INV, we quartile-rank each observation and label observations in the bottom (top) quartile as UNDER_INV (OVER_INV) = 1. Then, we estimate the same Models (3a) and (3b) which are used for the unconditional tests and report the results in Table 6.⁸ Panel A contains the results when the dependent variable is UNDER_INV. Similar to the results in Table 4, Panel A, when estimating Model (3a) with the full sample, the coefficient on POST is significantly negative (p≤ .001), suggesting that firms are less likely to underinvest after the adoption of IAS 27. Furthermore, when estimating Model (3a) with the change and no-effect subsamples, the coefficient on POST is significant only in the change subsample, while it is insignificant in the no-effect subsample. Panel B contains the results when the dependent variable is OVER_INV. Again, similar to the results in Panel B of Table 4, the coefficient on POST is significantly negative, implying that after adopting IAS 27, Taiwanese firms are less likely to overinvest. Again, the positive impact of IAS 27 adoption on investment efficiency through the reduction in overinvestment exists only in the change subsample, not in the no-effect subsample.

Table 6.

Sensitivity Analysis: Change in Investment Efficiency for the Change and No-Effect Samples—Unconditional Tests Using Other Measures of Investment.

Panel A: Underinvestment Versus Normal Investment.
	(1) Full sample		(2) Change sample		(3) No-effect sample
Intercept	3.715	(4.15)***	4.220	(2.66)**	1.029	(0.60)
POST	−0.335	(−3.05)**	−0.232	(−2.92)**	0.030	(0.13)
Analyst	−0.008	(−0.44)	0.016	(0.60)	−0.079	(−2.72)**
FINV	−1.834	(−0.31)	1.296	(0.17)	−7.410	(−0.60)
SIZE	0.034	(0.77)	−0.011	(−0.15)	0.070	(0.84)
MB	0.245	(4.68)***	0.050	(0.51)	0.106	(1.23)
ZSCORE	−0.378	(−5.10)***	−0.286	(−2.01)*	−0.082	(−0.63)
CYCLE	−0.574	(−7.17)***	−0.572	(−3.99)***	−0.340	(−2.10)*
LOSS	0.688	(5.28)***	1.084	(5.44)***	0.558	(2.28)*
TAN	−3.429	(−9.34)***	−3.880	(−6.33)***	−2.858	(−4.40)***
MEAN_LEV	−3.365	(−3.95)***	−3.529	(−2.87)**	2.006	(0.70)
CFO_SALE	−0.969	(−3.46)***	−1.056	(−2.23)*	−1.963	(−3.01)**
VOL_OCF	−0.247	(−0.27)	2.786	(2.04)*	−0.695	(−0.44)
VOL_SALE	−0.178	(−0.57)	−0.381	(−0.76)	0.227	(0.44)
VOL_INV	3.265	(3.90)***	3.380	(2.68)**	0.675	(0.48)
AGE	−0.027	(−4.27)***	−0.025	(−2.57)*	0.015	(1.16)
DIV	0.058	(0.57)	−0.242	(−1.50)	−0.489	(−2.51)*
Industry and year dummies	Included		Included		Included
n	2,389		1,769		620
Adjusted R²	.176		.2122		.1364
Panel B: Overinvestment Versus Normal Investment.
	(1) Full sample		(2) Change sample		(3) No-effect sample
Intercept	−5.424	(−5.83)***	−7.218	(−4.54)***	−2.921	(−1.72)
POST	−0.181	(−1.87)*	−0.302	(−1.90)*	0.197	(0.93)
Analyst	0.040	(2.31)*	0.024	(0.92)	0.048	(1.45)
FINV	−6.103	(−0.96)	−0.301	(−0.04)	−13.647	(−1.06)
SIZE	0.220	(5.36)***	0.222	(3.52)***	0.195	(2.22)*
MB	0.208	(3.43)***	0.189	(1.77)	0.181	(2.09)*
ZSCORE	−0.200	(−2.41)*	−0.001	(−0.01)	−0.210	(−1.56)
CYCLE	0.051	(0.62)	0.200	(1.36)	−0.460	(−2.80)**
LOSS	−0.578	(−3.64)***	−0.868	(−3.37)***	−0.629	(−2.19)*
TAN	2.676	(7.28)***	3.790	(6.29)***	2.816	(4.49)***
MEAN_LEV	−1.247	(−1.60)	0.145	(0.13)	−0.186	(−0.07)
CFO_SALE	0.454	(1.59)	0.478	(0.97)	0.305	(0.69)
VOL_OCF	−1.650	(−1.63)	−0.046	(−0.03)	−3.681	(−2.13)*
VOL_SALE	0.962	(3.08)**	1.106	(2.37)*	1.378	(2.52)*
VOL_INV	2.309	(2.81)**	1.514	(1.25)	4.276	(2.83)**
AGE	−0.058	(−8.73)***	−0.040	(−4.60)***	−0.072	(−3.90)***
DIV	0.879	(7.82)***	0.852	(4.89)***	0.619	(3.13)**
Industry and year dummies	Included		Included		Included
n	2,389		1,769		620
Adjusted R²		.176		.2122		.1364

Note. All variables are as defined in the appendix except UNDER_INVEST (or OVER_INVEST). We estimate a multinomial logistic regression that tests the likelihood that a firm might be in one of the extreme investment residual quartiles when POST is equal to 1. Following Biddle, Hilary, and Verdi (2009), we separately examine the likelihood of over- and underinvestment. UNDER_INVEST (or OVER_INVEST) is measured based on Cella (2014). The unexplained (or abnormal) portion of investment is measured as each firm-year investment minus the average investment of the industry to which each firm belongs. Then, the abnormal portion of the firm-year observations is quartile ranked. The bottom quartile of unexplained investment is classified as underinvestment (UNDER_INVEST), whereas the top quartile is classified as overinvestment (OVER_INVEST), and observations in the middle two quartiles are classified as normal (or more optimal) investments.

UNDER_INVEST (or OVER_INVEST)_i,t = α+β₁POST_i,t−1+β₂Analyst_i,t−1+β₃FINV_i,t−1+ Control variables +ϵ_i,t,

where all variables including control variables (reported in the appendix) are defined as in Biddle et al. (2009).

, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively, in a two-tailed test.

Our second alternative measure follows Titman et al. (2004), who use each firm’s average of the past 3 years’ investment as normal or the optimal investment. Thus, if this year’s investment is greater (lower) than the average of the past 3 years, it is considered overinvestment (underinvestment). Again, we quartile-rank the difference between this year’s investment and the average of the past 3 years, and then define OVER_INV and UNDER_INV similarly. We estimate Models (3a) and (3b) with these alternative measures and find that our primary results are robust to these alternative measures.

Results With Additional Control Variables

It is plausible that our results are driven by some factors that cause firms to be categorized as the change subsample or the no-effect subsample. Our main model has already included some control variables to address this possibility. For example, VOL_OCF is included as firms with high operating uncertainty tend to hide off-balance-sheet risks (Ketz, 2003). Thus, firms with high operating uncertainty may have greater incentives to avoid consolidation. Nonetheless, to further control for potential opportunistic motives and innate differences in firm characteristics between the change group and the no-effect group, we also include the following variables in Model (1): the significance of affiliated transactions (i.e., the amount of sales between the firm and its affiliated parties, the amount of purchases between the firm and its affiliated parties, the net loans between the firm and its affiliated parties, and the amount of guarantees between the firm and its affiliated parties), an audit quality measure, and various measures of the strength of corporate governance. We find that our (untabulated) results are robust to the inclusion of these additional control variables.⁹

Summary and Conclusion

Consolidated financial statements are the financial statements of a group presented as those of a single economic entity (Paragraph 4 of IAS 27). In preparing consolidated financial statements, one key term is “control.” In IAS 27, control is defined as the power to govern the financial and operating policies of an entity so as to obtain benefits from its activities. Although the definition seems straightforward, without a definite and comprehensive pronouncement on control, most firms in many countries, including the United States, use majority ownership as a general policy and manage the 50% threshold to omit some subsidiaries from consolidation. In contrast, IAS 27 defines consolidation using the broader and more comprehensive principle of whether an investing entity has effective control over financial and operating decisions of the voting entities. The recent literature (e.g., Hsu & Pourjalali, 2014) shows that the control concept in IAS 27 provides more useful information to stock investors, which is reflected in the FERC. However, our knowledge of its impact on various managers’ decisions, particularly managers’ investment decisions, is still limited. Our article attempts to fill this void by examining how the adoption of the broader and more comprehensive control concept in IAS 27 affects firm-level investment efficiency.

We use the change in Taiwanese accounting standards as a natural experiment to test the relation between the adoption of IAS 27 and investment efficiency. We hypothesize that adopting IAS 27 will improve investment efficiency as IAS 27 reduces the likelihood of abnormal intercompany transactions (Hsu & Pourjalali, 2014), which decreases managers’ ability to manage earnings through affiliated transactions, and thus improves the quality of consolidated financial statements. Similar to Biddle et al. (2009), we predict that the higher quality of consolidated financial statements under IAS 27 leads to investment efficiency. In addition, a broader definition of control can reduce managerial incentives to engage in value destroying activities such as empire building.

We compare investment efficiency between pre-IAS 27 and post-IAS 27 adoption periods and find that after the adoption of IAS 27, investment efficiency improves, with decreases in both over- and underinvestment. We also find that the results are more pronounced in firms which are directly affected by the adoption of IAS 27, mitigating the concern that our results may be driven by some unobservable macroeconomic or regulatory factors and further supporting our hypotheses. We also examine whether the adoption of IAS 27 leads to an increase in foreign investors’ shareholdings in Taiwan, as one plausible benefit from adopting the more comprehensive control concept. We find that overall, stock ownership by foreign investors significantly increases after IAS 27 adoption in Taiwan.

Given the sample and methodology used for this study, we provide strong evidence that a more inclusive definition of control reduces moral hazard and adverse selection problems, which in turn results in decisions that lead to more optimal investment levels. It has been very challenging to separate possible cultural, economic, and regulatory influences from the effect of IFRS adoptions (including IAS 27) on firms’ investment decisions for most prior studies utilizing cross-country variations. This study addresses this concern by using the natural exogenous shock of an accounting standard change within one country. Our findings also have significant implications for both FASB and IASB in their continual efforts to define what constitutes an accounting entity.

Footnotes

Appendix

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Aggarwal

Klapper

Wysocki

P. D.

(2005). Portfolio preferences of foreign institutional investors. Journal of Banking & Finance, 29, 2919-2946.

Ahearne

A. G.

Griever

W. L.

Warnock

F. E.

(2004). Information costs and home bias: An analysis of US holdings of foreign equities. Journal of International Economics, 62, 313-336.

Altman

E. I.

(1968). Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. Journal of Finance, 23, 589-609.

American Institute of Certified Public Accountants. (1959). Consolidated financial statements (Accounting Research Bulletin 51). New York, NY: Author.

American Institute of Certified Public Accountants. (1971). The equity method of accounting for investments in common stock (Accounting Principles Board Opinion 18). New York, NY: Author.

Atanasov

Boone

Haushalter

(2010). Is there shareholder expropriation in the United States? An analysis of publicly-traded subsidiaries. Journal of Financial and Quantitative Analysis, 45, 1-26.

Baik

Kang

J. K.

Kim

Lee

(2013). The liability of foreignness in international equity investments: Evidence from the U.S. stock market. Journal of International Business Studies, 4, 391-411.

Barth

M. E.

Landsman

W. R.

Lang

Williams

(2012). Are IFRS-based and US GAAP-based accounting amounts comparable?Journal of Accounting and Economics, 54, 68-93.

Bauman

M. P.

(2003). The impact and valuation of off-balance-sheet activities concealed by equity method accounting. Accounting Horizons, 17, 303-314.

10.

Biddle

Callahan

C. M.

Hong

H. A.

Knowles

R. L.

(2013). Does mandatory adoption of International Financial Reporting Standards increase investment efficiency? (Working paper). Pok Fu Lam: The University of Hong Kong.

11.

Biddle

Hilary

(2006). Accounting quality and firm-level capital investment. The Accounting Review, 81, 963-982.

12.

Biddle

Hilary

Verdi

(2009). How does financial reporting quality relate to investment efficiency?Journal of Accounting and Economics, 48, 112-131.

13.

Blacconiere

W. G.

Hopkins

P. E.

(2002). Teaching notes for general electric: Investment accounting and consolidations. Issues in Accounting Education, 17(3), 13-29.

14.

Bushman

Smith

(2001). Financial accounting information and corporate governance. Journal of Accounting and Economics, 31, 237-333.

15.

Cella

(2014, February). Institutional investors and corporate investment (Working paper). Stockholm, Sweden: Stockholm School of Economics.

16.

Chen

Young

Zhuang

(2013). Externalities of mandatory IFRS adoption: Evidence from cross-border spillover effects of financial information on investment efficiency. The Accounting Review, 88, 881-914.

17.

Covrig

V. M.

DeFond

M. L.

Hung

(2007). Home bias, foreign mutual fund holdings, and the voluntary adoption of international accounting standards. Journal of Accounting Research, 45, 41-70.

18.

Financial Accounting Standards Board. (1987). Consolidation of all majority-owned subsidiaries (Statement of Financial Accounting Standards No. 94). Stamford, CT: Author.

19.

Financial Accounting Standards Board. (2000). Accounting for transfers and servicing of financial assets and extinguishments of liabilities (a replacement of FASB Statement No. 125) (Statement of Financial Accounting Standards No. 140). Stamford, CT: Author.

20.

Financial Accounting Standards Board. (2003). Consolidation of variable interest entities—An interpretation of ARB No. 51 (FASB Interpretation 46(R)). Stamford, CT: Author.

21.

Financial Accounting Standards Board. (2009a). Accounting for transfers of financial assets an amendment of FASB Statement No. 140 (FAS 166). Stamford, CT: Author.

22.

Financial Accounting Standards Board. (2009b). Amendments to FASB Interpretation No. 46(R) (Statement of Financial Accounting Standards No. 167). Stamford, CT: Author.

23.

Florou

Pope

(2012). Mandatory IFRS adoption and institutional investor decisions. The Accounting Review, 87, 1993-2025.

24.

Francis

LaFond

Olsson

P. M.

Schipper

(2004). Costs of equity and earnings attributes. The Accounting Review, 79, 967-1010.

25.

Hansen

Miletkov

Wintoki

M. B.

(2013). When does the adoption and use of IFRS increase foreign investment? (Working paper). Blacksburg: Virginia Tech University.

26.

Hartgraves

A. L.

Benston

G. J.

(2002). The evolving accounting standards for special purpose entities and consolidation. Accounting Horizons, 16, 245-258.

27.

Hoshi

Kashyap

Scharfstein

(1991). Corporate structure liquidity and investment: Evidence from Japanese panel data. Quarterly Journal of Economics, 6, 33-60.

28.

Hosmer

D. W.

Lemeshow

(2000). Applied logistic regression (2nd ed.). New York, NY: John Wiley.

29.

Hoyle

J. B.

Schaefer

T. F.

Doupnik

T. S.

(2011). Advanced accounting (11th ed.). New York, NY: McGraw-Hill.

30.

Hsu

A. W.

Duh

R. R.

Cheng

(2012). Does the control-based approach to consolidated statements better reflect market value than the ownership-based approach?International Journal of Accounting, 47, 198-225.

31.

Hsu

A. W.

Pourjalali

(2014). The impact of IAS No. 27 on the market’s ability to anticipate future earnings. Contemporary Accounting Research. Advance online publication. doi:10.1111/1911-3846.12088

32.

International Accounting Standards Board. (2003). IAS 27: Consolidated and separate financial statements. London, England: Author.

33.

International Accounting Standards Board. (2004). SIC 12: Consolidation-special purpose entities. London, England: Author.

34.

International Accounting Standards Board. (2010). Exposure draft on conceptual framework for financial reporting—The reporting entity. Retrieved from http://www.ifrs.org/Current-Projects/IASB-Projects/Conceptual-Framework/EDMar10/Documents/EDConceptualFrameworkMar10.pdf

35.

International Accounting Standards Board. (2011). IFRS 10: Consolidated financial statements. London, England: Author.

36.

International Accounting Standards Board. (2012). IFRS 12: Disclosure of interests in other entities. London, England: Author. (Amended by Investment Entities, issued October 2012)

37.

Jung

Lee

Weber

(2014). Financial reporting quality and labor investment efficiency. Contemporary Accounting Research, 31, 1047-1076.

38.

Kang

J.-K.

Stulz

R. M.

(1997). Why is there a home bias? An analysis of foreign portfolio equity ownership in Japan. Journal of Financial Economics, 46, 3-28.

39.

Kang

J.-K.

Stulz

R. M.

(2000). Do banking shocks affect borrowing firm performance? An analysis of the Japanese experience. Journal of Business, 73, 1-23.

40.

Ketz

J. E.

(2003). Hidden financial risk: Understanding off-balance-sheet accounting. New York, NY: John Wiley.

41.

Lambert

Leuz

Verrecchia

(2007). Accounting information, disclosure, and the cost of capital. Journal of Accounting Research, 45, 385-420.

42.

Lamont

(1997). Cash flow and investment: Evidence from internal capital markets. The Journal of Finance, 52, 83-109.

43.

Lang

L. E.

Ofek

Stulz

(1996). Leverage, investment and firm growth. Journal of Financial Economics, 40, 3-29.

44.

(2010). Does mandatory adoption of International Financial Reporting Standards in the European Union reduce the cost of equity capital?The Accounting Review, 85, 607-636.

45.

Myers

Majluf

(1984). Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics, 13, 187-221.

46.

Petersen

M. A.

(2009). Estimating standard errors in finance panel data sets: Comparing approaches. Review of Financial Studies, 22, 435-480.

47.

Rajan

Servaes

Zingales

(2000). The diversification discount and inefficient investment. Journal of Finance, 55, 35-80.

48.

Scharfstein

D. S.

(1998, January). The dark side of internal capital markets II: Evidence from diversified conglomerates (NBER Working Paper Series, No. 6352). Boston, MA: Harvard Business School.

49.

Shin

Stulz

R. M.

(1998). Are internal capital markets efficient?Quarterly Journal of Economics, 113, 531-552.

50.

Stein

(2003). Agency, information and corporate investment. In Constantinides

Harris

Stulz

(Eds.), Handbook of the economics of finance (pp. 111-165). London, England: Elsevier.

51.

Thomas

W. B.

Herrmann

D. R.

Inoue

(2004). Earnings management through affiliated transactions. Journal of International Accounting Research, 3(2), 1-25.

52.

Titman

Wei

K. C.

Xie

(2004). Capital investments and stock returns. Journal of Financial and Quantitative Analysis, 39, 677-700.

53.

Tobin

(1969). A general equilibrium approach to monetary theory. Journal of Money, Credit, and Banking, 1, 15-29.

54.

Tucker

J. W.

Zarowin

P. A.

(2006). Does income smoothing improve earnings informativeness?The Accounting Review, 81, 251-270.