The initiation of audit committee interlocks and the contagion of accounting policy choices: evidence from special items

For example, an audit committee chair of a NASDAQ firm stated that “The audit committee is absolutely involved. We focus on any prior year restructuring reserves, particularly focusing on where we put the reversal of any reserves. We focus on the allowance for doubtful accounts, inventory reserves, and income/deferred income estimates. We also look at impairment of intangibles” (Beasley et al. 2009, p. 95).

Classified boards are common in US firms (Yerak 2012). Since a director may be removed from the audit committee without being removed from the classified board (Arthaud-Day et al. 2006), our focus on audit committees could help mitigate the entrenchment effect associated with directors on a staggered board (Bebchuk and Cohen 2005).

Our definition of an audit committee interlock considers the situation where an inside director of one firm joins the audit committee of another firm. Supported by the empirical evidence that executives have greater access to inside information than audit committee members do (Ravina and Sapienza 2010), this research design considers all cases in which a new director uses information obtained from the contagious firm to influence audit committee decisions of the suspect firm. As the initiation of an audit committee interlock is a one-time event, we adopt this broader definition to increase the size of our sample and the power of our study.

Han et al. (2017) find that Chinese firms with interlocked directors have similar inventory and depreciation methods. As a choice between FIFO and weighted average cost, the inventory cost method in China exhibits little variation, as 71.92% of the firms use the weighted average cost method as of December 31, 2009 (http://​www.​atlantis-press.​com/​php/​download_​paper.​php?​id=​9403). Given the stickiness of inventory methods, their findings are likely driven by endogenous ex-ante similarities between interlocked firms rather than by the asserted information transfer.

This one-year requirement helps us establish the sequence of events without ambiguity.

Since special items are expenses reported in the income statement, we scale special items, negative special items, and special items components by sales in our paper (McVay 2006; Fan et al. 2010). Our main results in Tables 3, 4, 5, 6 and 7 are robust when, alternatively, we deflate special items, negative special items, and special items components by average total assets.

Consistent with Elliott and Hanna (1996), we assign a zero to special items if this data item is missing in Compustat.

Bebchuk et al. (2009) track six provisions: staggered boards, limits to shareholder amendments of the bylaws, supermajority requirements for mergers, supermajority requirements for charter amendments, poison pills, and golden parachute arrangements. They construct an entrenchment index (E-Index) by assigning a score from zero to six based on the number of these provisions that the company has in a given year.

We follow the Fama-French 12 industry classification (http://​mba.​tuck.​dartmouth.​edu/​pages/​faculty/​ken.​french/​Data_​Library/​det_​12_​ind_​port.​html).

As reported in Johnson et al. (2011), the SI distribution is skewed because the number of firms reporting negative special items is significantly greater than the number of firms reporting positive special items. This is most likely driven by managerial incentives to classify expenses or losses into SI, which is often excluded in executive compensation contracts and analyst forecasts (e.g., McVay 2006). Per the central limit theorem, OLS estimates are asymptotically normally distributed in large enough samples (Wooldridge 2013), which our sample is. Thus, a strictly normal distribution of the error is not necessary for the validity of our OLS regressions. Nonetheless, in untabulated analyses, we find that our main results in Tables 3, 4, 5, 6 and 7 are robust with bootstrapped standard errors.

Restructuring costs include “a write-down of assets to their estimated realizable value; operating losses to be incurred prior to the sales; a provision for severance and other employee termination costs; and a provision for the cost of redeployment of assets” (Elliott and Shaw 1988, p. 117).

Note that our asset impairment measure includes write-downs for tangible assets and impairment of goodwill for intangible assets, as Cready et al. (2012) find that earnings increases are greater for restructuring charges than for asset write-downs or goodwill impairment charges.

The other four special items components – extinguishment of debt, acquisition/merger, settlement (litigation/merger), and in-process research & development – are less likely to be discretionary accounting choices. While the timing of extinguishment of debt is under management’s control, the amount is less likely to be manipulated because the extinguishment payment by cash or by a new debt is objective. Representing the portion of R&D considered to be “purchased,” the in-process research & development item is written off immediately upon acquisition if the R&D items are deemed not to have an alternative use.

We find that the correlation of restructuring costs between suspect firms and contagious firms is significant before the new interlock formation, which may be because firms with similar restructuring activities tend to select similar directors; i.e., ex-ante similarities exist. However, the significant increase in the correlation (β₂ is significant for all models in Table 7) due to interlock formation clearly shows that information also is transferred via interlocks. Therefore, the significant correlation post interlock formation between restructuring costs of interlocked firms is a result of both ex-ante similarities and information transfer.

Note that the joint probability of independent events is their product.

“A falsification hypothesis is a claim, distinct from the one being tested, that researchers believe is highly unlikely to be causally related to the intervention in question” (Prasad and Jena 2013, p. 241).

Our results are robust if we define new audit committee interlocks in their first 2 years of existence.

Hiring and firing directors is a significant corporate decision that requires voting by shareholders. It seems rather unlikely that firms hire and fire directors in sync with their desires for a single, specific accounting policy, as accounting expertise should, in general, fall across a much larger spectrum.