The cornerstone of good science is reproducibility. Otherwise, the evidence is just anecdotal. Unfortunately, the pressures of the present academic climate make replication studies uncommon. Generally speaking, journals prefer unique, novel findings, and as tenure track positions dwindle, there’s added pressure on early-career academic researchers to make a name for themselves with more novel studies.
So, it’s unsurprising that the most cited study on firm governance in the face of internationalization, a 1998 study by W.G. Sanders and M.A. Carpenter, “Internationalization and firm governance: The roles of CEO compensation, top team composition, and board structure,” has not had its key findings verified by other researchers. Even as digitization has introduced new analytic tools and new modes of organization, the evidence is constrained by Sanders and Carpenter’s small sample size, a couple of hundred firms.
But as the University of Arkansas’s Oleg Petrenko and doctoral candidate Tsutomu Doiguchi point out, the business ecosystem is very different three decades on. CEO pay has increased 514%, and researchers have found stronger explanations for the structure of CEO compensation packages than internationalization. Moreover, there is a greater emphasis today on diversity among boards and executives since the ‘90s, so firms hire more international executives and board members, who then bring specialized knowledge about conducting business in international settings.
All of these changes mean that it is necessary to revisit Sanders and Carpenter’s findings to ensure they are still robust in the early twenty-first century, argue Petrenko and Doiguchi. To do so, Petrenko, Doiguchi, and their coauthor, Jeffery Chandler, sought to replicate the findings of the earlier research. Only, they did not simply replicate the study; instead, they extended it by analyzing more firms (about 5x as many) during a longer timeframe, 1999 – 2019. The team of researchers employed the same tests and specifications to test the generalizability of Sanders and Carpenter’s earlier work.
Petrenko, Doiguchi, and Chandler published their article, “Revisiting the Effect of Internationalization on Firm Governance: A Replication and Extension Study,” in Management International Review. Their study produced results that did not totally support all of the conclusions Sanders and Carpenter came to in their 1998 paper. Petrenko, Doiguchi, and Chandler caution us however that this is not an indictment of the earlier research. Instead, they say the data “point[s] to an interesting development” over the past 30 years.
The Data Show Times are Changing
To replicate the study, Petrenko, Doiguchi, and Chandler adopted the same variables Sanders and Carpenter had used. But instead of using the same data, they used a new population to directly test the generalizability of Sanders and Carpenter’s key findings. In a second step, the team of researchers used more rigorous analysis and additional control variables to again better test the generalizability of the earlier study.
Petrenko, Doiguchi, and Chandler’s replication study showed mixed results, although they do say it “finds general support for Sanders and Carpenter’s general premise” since about half of their findings were supported. Specifically, the researchers confirmed that internationalization seems to prompt firms to split the CEO and chairman of the board positions. Also confirming Sanders and Carpenter’s earlier study, the researchers also found that internationalization predicts more independent boards. Contrary to the earlier study, however, Petrenko, Doiguchi, and Chandler found that internationalization predicts smaller boards, which are far more nimble and offer better returns.
Crucially, they do not find any support that firms cope with internationalization through long-term CEO compensation or higher compensation packages. This aligns with more recent studies on CEO pay that show other social pressures, such as a director’s political ideology, better account for those changes.
There are, however, still some firms that seem to cope with the stresses of internationalization with larger CEO compensation packages. Petrenko, Doiguchi, and Chandler found that corruption in the countries of firm operation better accounts for this than internationalization per se. That is, CEO pay becomes an anti-corruption mechanism, and indeed, the researchers found a measure of a country’s corruption correlated with CEO pay. So, companies use higher CEO pay when operating in corrupt countries to cope with internationalization, but countries with low corruption levels do not impose the same pressures.
The researchers suggest that the reason firms no longer seem to use higher CEO pay to address internationalization is that in the last 30 years, it has become an expectation of a CEO to monitor a firm’s international divisions. And to this point, their data shows evidence that companies are hiring more CEOs with international experience. It is increasingly becoming the case that C-suite executives eying the CEO role need international experience to be competitive.
The Next Thirty Years
Petrenko, Doiguchi, and Chandler have helped usher Sanders and Carpenter into the next generation of scholarship by performing the crucial scientific work of replication. And the researchers see ample space for future research to further explore some of their initial observations. For example, their study shows that a country of operation’s corruption is a boundary case where internationalization does seem to directly increase a CEO’s pay, so a better exploration of a board’s ideological stance and the CEO’s compensation may be due.
Similarly, while the study shows a relationship between internationalization and board size, it does not explain the mechanism. The researchers identify this need for an explanation as a promising place for new research since their study disagrees with Sanders and Carpenter on whether internationalization puts upward or downward pressure on board size. It may be, they suggest, that international firms need executives with specific sets of skills, which may have appeared in Sanders and Carpenter’s analysis as a demand for larger boards.
The team of researchers also invite others to conduct studies to further extend Sanders and Carpeneter. Petrenko, Doiguchi, and Chandler offer plenty of other ways firms may cope with internationalization based on their relationship with their CEOs that haven’t yet been fully explored. They suggest, for example, firms may use more visible CEOs, or firms may pick CEOs with particular personality profiles that fit the image of an ideal leader.
But a degree of caution is certainly due. While the researchers corroborated and generalized aspects of Sanders and Carpenters’ study, their replication study did not find support for about half of the earlier study’s findings in the new circumstances. Even practices that seem to intuitively make sense to us can have little strong evidence to back them up. The scientific process is a recursive one, and skepticism is one of its cardinal virtues.
