DO WELL‐CONNECTED BOARDS INVEST OPTIMALLY IN R&D ACTIVITIES?

• DOI: http://doi.org/10.1111/jfir.12228
• Date: 01 December 2020
• Published date: 01 December 2020

The Journal of Financial Research Vol. XLIII, No. 4 Pages 895–932 Winter 2020

DOI: 10.1111/jfir.12228

DO WELL‐CONNECTED BOARDS INVEST OPTIMALLY IN R&D ACTIVITIES?

Subramanian R. Iyer

University of New Mexico

Harikumar Sankaran and

New Mexico State University

Yan Zhang

Abstract

Researchers have argued that the uncertainty surrounding innovative activities

causes firms to either underinvest or overinvest in research and development

(R&D). We examine whether the information gained by boards through

directors’connections helps mitigate such distortions. We find that an increase

in directors’connections has an asymmetric impact on under‐and overinvesting

firms. R&D expenditures are shown to increase with board connections. Such

increases in R&D intensity exacerbate the extent of overinvestment, resulting

in a loss in future market‐to‐book value. The increase in R&D intensity,

however, reduces underinvestment only among firms with higher than average

R&D productivity. We find that increased director busyness is one cause of

overinvestment.

JEL Classification: G03, G32, O30

I. Introduction

In the absence of capital market frictions, managers maximize their shareholders’

wealth by investing in all projects that have a positive net present value (NPV)

(Modigliani and Miller 1958). Managers deviate from this investment rule when

there is a misalignment between their interests with those of shareholders

(Myers 1977; Jensen 1986) or in the presence of asymmetric information between

insiders and outside investors (Myers and Majluf 1984), resulting in a loss

in shareholders’wealth.

1

Several mechanisms shown to reduce such investment

inefficiencies include the design of complex contractual arrangements between the

We thank the participants in the University of New Mexico seminar, 2016 Southwestern Finance

Association annual meeting, 2017 World Finance Conference, 2018 European Financial Management

Association annual meeting, 2018 Multinational Finance Association annual meeting, 2018 Financial

Management Association conference, and 2018 Pacific Basin Finance, Economic, Accounting, and Management

conference (Rutgers University). We especially thank Shofiqur Rahman for his valued comments. All errors

are ours.

1

Aside from frictions due to moral hazard and adverse selection, chief executive officer (CEO) attributes

such as overoptimism (Heaton 2002) and overconfidence (Malmendier and Tate 2005, 2008) have been shown to

cause managers to either underinvest or overinvest, resulting in value loss.

895

© 2020 The Southern Finance Association and the Southwestern Finance Association

parties in conflict (Haugen and Senbet 1981; Barnea, Haugen, and Senbet 1981),

discipline by external creditors (Jensen 1986), increased transparency (Koo,

Ramalingegowda, and Yu 2017), and governance by activist shareholders

(Richardson 2006; Brav et al. 2008), among others. Although investment in

tangible assets such as plant and equipment or acquisitions lend themselves to

scrutiny, it is relatively more difficult to detect and monitor the efficiency of

investments in intangible assets (e.g., intellectual property).

2

In this article, we focus on firms’investments in research and development

(R&D). To remain competitive, managers need to have long‐term vision and

strategy, which enables them to continually adapt technologies and processes to

offer new products and services. Several studies show that R&D investments

indicate the strategic importance of innovation and constitute an important input in

the development of intangible assets, that is, intellectual property (Chauvin and

Hirschey 1993; David, Hitt, and Gimeno 2001; Mosakowski 1993). The impetus

for the strategic direction arises from an understanding of market trends,

technological changes adopted by other firms, repositioning of the product space

of mature firms, innovative ideas of growth firms, and potential sources of

financing to support investment in such intangible assets. Research shows that

corporate boards are best suited to provide this guidance because of their ability to

tap into their connections and obtain important strategic information that guides

corporate resources toward timely and valuable R&D investments.

3

We posit that

the wisdom corporate boards of directors acquire from their connections not only

induce managers to invest more in R&D but also enable them to advise managers

to cut back on investments that may no longer yield profits (reduce over-

investment) and commit resources to projects that are likely to strategically

position the firm for the future (reduce underinvestment). Our research uses a well‐

established methodology developed in the management literature (Knott 2008) to

detect under‐and overinvestment in R&D, and examines corporate boards’ability

to mitigate such distortions using the strategic information gained through their

directors’network connections.

Two key constructs in our analysis are an estimate of distortions in R&D

investments and board connectivity. Studies use regression models to estimate a

predicted level of total investment as a function of firm characteristics and then use the

negative (positive) regression residual as a proxy for underinvestment (overinvestment)

(Richardson 2006; Biddle, Hilary, and Verdi 2009).

4

The regression‐based prediction

approach, however, is not suitable for estimating distortions in R&D investments for

two main reasons. First, our study requires an estimate of under‐and overinvestment in

R&D. The residuals from regression methods or models pertain to total investments

2

See Holmstrom (1989) and Ahuja and Novelli (2017) for a discussion of under‐and overinvestment

in R&D.

3

See Chuluun, Prevost, and Upadhyay (2017) and Helmers, Patnam, and Rau (2017) for research on the

impact of board connections on R&D investments.

4

Total investment is expressed as the total of capital expenditure, cost incurred in acquisitions, R&D

expenditure minus sale of property, plant, and equipment divided by lagged total assets of a firm.

896 The Journal of Financial Research

and are, therefore, inaccurate proxies for distortions in R&D. Second, the innate

research productivity of technology within an organization, inherent uncertainty of

cash flows from R&D projects, long gestation period, and technological spillovers are

factors that influence investment in R&D but not tangible assets. These aspects of

R&D investments are difficult to model in regression specifications.

5

We rely on the empirical framework in Knott (2008) because it considers the unique

features of R&D investments and provides an estimate of a firm’s R&D productivity (i.e.,

research quotient, or RQ) for any given year.

6

RQ has been subject to several robustness

tests and found to be a complementary measure of a company’s innovation (Cooper, Knott,

and Yang 2019). The methodology in Knott (2008) facilitates the estimation of a firm‐level

benchmark for R&D expenditures in a given year (hereafter optimal R&D expenditures, or

R*).

7

Using the estimated value of R*, we identify underinvestment (overinvestment) in

R&D when the actual R&D expenditure is less (greater) than R*.

8

The second construct is a measure of board connections. The board of directors

of a corporation is meant to monitor and advise top management (Hermalin and

Weisbach 2003; Adams and Ferreira 2007; Coles, Daniel, and Naveen 2008). Evidence

from recent studies indicates that well‐connected directors favorably affect firms’

innovation output measured in terms of patents and patent citations (Chuluun, Prevost,

and Upadhyay 2017; Helmers, Patnam, and Rau 2017). The informational advantages

of board connections, however, may come at the cost of management oversight.

Several studies that examine the adverse effects of multiple directorships show that

well‐connected board members with multiple directorships do not have as much time to

perform their monitoring duties, which causes firm performance to decline (Fich and

Shivdasani 2006; Devos, Prevost, and Puthenpurackal 2009; Falato, Kadyrzhanova,

and Lel 2014; Ahn, Jiraporn, and Kim 2010; Hauser 2018).

9

Using the data from the

Corporate Board Member Magazine Director Database (BoardMag), we compute

different measures of how centrally directors are located in their network of

connections.

10

As in prior studies (Larcker, So, and Wang 2013), we assume that

5

The difficulty in measuring R&D effectiveness is also evident from the recent reports from consulting

firms such as the Boston Consulting Group (https://www.bcg.com/publications/2017/automotive-new-metric-

boost-automakers-research-development-efficiency.aspx) and McKinsey & Company (https://www.mckinsey.

com/business-functions/operations/our-insights/brightening-the-black-box-of-r-and-d).

6

RQ is the output elasticity of R&D. It is the percentage increase in revenue from a 1% increase in R&D.

RQ offers a universal, uniform, and reliable measure of a firm’s R&D productivity (Knott 2017, p. 8).

7

We obtain the estimates of RQ and R* from the Wharton Research Data Services (WRDS) database. The

estimation process to determine RQ and R* and its robustness checks are described in the user manual for the

WRDS RQ database. This comprehensive manual explains the theory underlying RQ, describes the functional

and operational form for all variables, as well as the logic behind those functional forms. It then compares

estimates for all variables in RQ estimation to those from four other versions of R&D production function

estimation, including attempts to control for the endogenous choice of inputs.

8

Because of the potential for measurement errors, we consider a broader window around R* and define a

firm as underinvested in R&D if R&D <90% of R* and as overinvested if R&D >110% of R*. Our results are

qualitatively unchanged from our main results.

9

We report the positive relation between multiple directorships and board centrality later in Panel C of

Table 1.

10

Specifically, we estimate Betweenness, Closeness, and Degree centrality measures. See Appendix A for

details.

897Do Connected Boards Invest Optimally?