Industry expertise on corporate boards

U.S. SEC, 17 CFR Parts 229, 239, 240, 249 and 274. Release # 33-9089, p. 29.

Such connections also potentially enhance incoming R&D spillover because the exposure of industry-experienced directors to other executives in the industry and their intimate knowledge of industry dynamics can improve a firm’s ability to benefit from technological advances at other industry firms (see, for example, Bernstein and Nadiri 1988; Chen et al. 2013; Jaffe 1986). We are grateful to an anonymous referee for pointing this out.

Consistent with this possibility, Chan et al. (2015) find that reductions in R&D spending are associated with positive long-term excess stock returns. Their result suggests that maturing firms optimally cut R&D investments in response to declining investment opportunity sets. See also Chambers et al. (2002).

For most tests, we report results using the first measure to conserve space. All results are comparable when using the other two measures.

Our prediction for firm size based on organizational complexity notwithstanding, it is plausible that smaller firms have a greater need for industry expert directors because they may be less able to have this required expertise internally. Thus, the effect of firm size on the demand for industry expert directors is unclear.

As in prior studies (Bushee 1998; Cheng 2004; Gunny 2010), we do not focus on analysts’ forecasts because real activities earnings management occurs throughout the year, before managers know the final forecasts of earnings.

Nevertheless, we perform the additional tests described in this sub-section using the sample of firms with non-zero R&D spending to sidestep the potential pooling problem.

As an example, consider Intel Corp. and Texas Instruments (TI), two same-industry firms in our sample (SIC code 36xx). Intel’s head office is located in Santa Clara, California (ZIP code 95054), while TI’s head office is located in Dallas, Texas (ZIP code 75243). During our sample period, 29 other firms with SIC code 36xx share the same 3-digit ZIP code with Intel (i.e., ZIP code 950xx) while only 4 such firms are located in the same 3-digit ZIP code as TI. Consistent with the argument that our instrument is relevant, 23.3% of Intel’s independent directors are industry experts while TI has no industry experts on its board. Nevertheless, it is less likely that Intel’s location in ZIP code 950xx in and of itself confers any significant advantage in R&D relative to TI because TI has 6 wholly owned subsidiaries in California, including 2 that are located within 5 miles of Intel’s headquarters.

The correlation between 2009 and 2000 values of the proportion of board industry experts is 0.63.

In untabulated tests, we also examine the effect of board industry expertise on innovation efficiency. Following prior studies (e.g., Hirshleifer, et al. 2013; Chan et al. 2001) we define innovation efficiency as the ratio of patents granted or citations per patent (natural log) to R&D capital, where R&D capital for year t = R&D _t−1 + 0.8*R&D _t−2 + 0.6*R&D _t−3 + 0.4*R&D _t−4 + 0.2*R&D _t−5. This measure captures the number of patents granted or their importance as a function of R&D capital invested. We find that industry expert directors are associated with greater innovation efficiency.

Diagnostic tests confirm the validity of these instruments in the Tobin’s q model. The Kleibergen–Paap test rejects the null of under identification while the Cragg–Donald Wald F-statistic for weak instruments is larger than all Stock–Yogo critical values for weak identification (Stock and Yogo 2005). Finally, Hansen’s J test of over-identification does not reject the null hypothesis that the instruments are exogenous in the second stage.

We are unable to conduct a similar analysis for R&D investments and patents due to sample size issues. For these variables, only 41 observations satisfy the sampling requirements.

NYSE and Nasdaq submitted their governance reform proposals to the SEC in August 2002 and October 2002, respectively. The proposals were approved in November 2003 and became fully mandatory in 2005.

We also perform a 2SLS analysis for our compensation variables using the instruments in our R&D tests. Instrumented industry expertise is positive and significant in the regression for stock option ratio and negative but insignificant in the regression for salary ratio. We do not tabulate these results due to space considerations.

An alternative interpretation is that industry expertise diminishes board effectiveness in CEO termination decisions. This could be because industry expert directors are more beholden to the CEO although it is not clear why that would be the case. It is also possible that industry expert directors identify with the CEO based on similar professional backgrounds. As shown by Tajfel and Turner (1979), individuals evaluate others more favorably when they regard those others as belonging to the same group(s) as themselves. However, similar industry experience is not likely to generate strong in-group identity because the group of individuals with similar industry backgrounds is potentially large and loosely defined. Furthermore, our pay–performance sensitivity result is inconsistent with an in-group bias explanation since such bias would predict the decoupling of CEO compensation and firm performance.