Do CEOs Matter? Evidence from Hospitalization Events

• Date: 01 August 2020
• DOI: http://doi.org/10.1111/jofi.12897
• Author: MORTEN BENNEDSEN,FRANCISCO PÉREZ‐GONZÁLEZ,DANIEL WOLFENZON
• Published date: 01 August 2020

THE JOURNAL OF FINANCE •VOL. LXXV, NO. 4 •AUGUST 2020

Do CEOs Matter? Evidence from Hospitalization

Events

MORTEN BENNEDSEN, FRANCISCO P ´

EREZ-GONZ ´

ALEZ,

and DANIEL WOLFENZON∗

ABSTRACT

Using variation in firms’ exposure to their CEOs resulting from hospitalization, we es-

timate the effect of chief executive officers (CEOs) on firm policies, holding firm-CEO

matches constant. We document three main findings. First, CEOs have a significant

effect on profitability and investment. Second, CEO effects are larger for younger

CEOs, in growing and family-controlled firms, and in human-capital-intensive indus-

tries. Third, CEOs are unique: the hospitalization of other senior executives does not

have similar effects on the performance. Overall, our findings demonstrate that CEOs

are a key driver of firm performance, which suggests that CEO contingency plans are

valuable.

IN THIS PAPER,WE USE variation in firms’ exposure to their chief executive officer

(CEO) resulting from hospitalization to estimate the effect of CEOs on firm

performance. The basic premise behind this test is that hospitalizations affect

managers’ ability to perform their jobs as they are physically away from the

office or convalescing from a medical condition. Econometrically, this test is

attractive because it allows us to assess how firm outcomes change as firms’

exposure to their CEO varies, holding the firm-CEO match constant.

To investigate this issue empirically, we use firm, CEO, and hospitaliza-

tion data from 12,753 Danish firms from 1996 to 2012. Denmark provides a

∗Morten Bennedsen is with the University of Copenhagen and INSEAD. Francisco P´

erez-

Gonz´

alez is with ITAM. Daniel Wolfenzon is with Columbia University. Previous versions of this

paper circulated under the titles “Do CEOs Matter?” and “Estimating the Value of the Boss: Evi-

dence from Hospitalization Events.” We thank the Editor for his valuable input and the referees

for their helpful comments. We thank seminar and conference participants from numerous places

in United States, Europe, and Asia for excellent comments and suggestions. We are grateful to

the Danish National Research Foundation, the Danish Social Science Research Foundation, the

EPRN network, and the Hoffmann Research Foundation for financial support. Francisco P´

erez-

Gonz´

alez thanks Asociaci´

on Mexicana de Cultura, A.C., for financial support. We are also grateful

to the Danish Commerce and Companies Agency and the Research Office in Statistics Denmark for

providing us with data, and to Shruti Sheth, Jihye Jang, and Jiacheng Yan for excellent research

assistance. We are particularly grateful to Pernille Bang and Kamilla Heurlen for their help with

accessing various data sources. In accordance with the disclosure principles of The Journal of

Finance, we state that none of the three authors has anything to disclose.

Correspondence: Daniel Wolfenzon, Columbia Business School, Columbia University, 3022

Broadway, New York, NY 10027; e-mail: dw2382@gsb.columbia.edu.

DOI: 10.1111/jofi.12897

C2020 the American Finance Association

1877

1878 The Journal of Finance R

near-ideal setting to run our analysis for four reasons. First, reliable finan-

cial information is available for the universe of limited liability firms. Second,

firms are required to report the name of their CEO to government agencies.

Third, and importantly for our purposes, detailed information about individu-

als’ hospitalizations (e.g., length and diagnosis) is available from the National

Patient Registry (NPR), and such data can be matched with CEO and firm

records. Lastly, given that more than 95% of hospitals spending in Denmark

are financed through general government expenditures, this setting mitigates

concerns that CEO hospitalizations may affect the performance directly as a

result of increased medical bills.1

We find that CEOs have an economically and statistically large effect on firm

performance. In particular, firms underperform when their chief executives

are hospitalized but otherwise exhibit similar performance as other firms. Our

results indicate that a 10-day hospital stay reduces firm operating profitability

by 5.8% from its mean (or by 0.5 percentage points).

To shed additional light on the magnitude of the above effect, we first ex-

amine the relation between hospitalization length and absence days, a more

direct measure of firm exposure to the CEO. This evidence indicates that even

short hospitalization spells lead to prolonged absences from the workplace. For

example, while an employee with no hospitalizations is absent an average of

seven days per year, the number of absences increases to 39 days per year when

an employee is hospitalized from four to six days.

We next examine whether our main result captures a “pure” CEO effect

or the combined effect of the chief executive’s absence and the organizational

response to this absence. We find no unusual behavior in terms of hiring or

firing of senior managers around CEO hospitalizations. That is, in terms of

aspects of the organization that we can observe, there does not seem to be a

large response to CEO hospitalizations, which suggests that our test is able to

isolate the CEO effect.

In addition to profitability, we examine the effect of CEO hospitalizations

on other corporate and financial outcomes. We show that, following a CEO

hospitalization event, firm investment and asset growth decrease, while the

likelihood of financial distress increases. These results further support the

view that CEOs have a meaningful effect on their organizations.

We next try to replicate our main tests for non-CEO senior managers. We

find that the magnitude of the effect is about half that for CEOs. These results

highlight the importance of CEOs relative to other senior employees, and justify

the attention that chief executives receive in both the media and the academic

literature (see survey by Bertrand (2009)).

When we analyze individual, firm, and industry settings in which CEOs

are particularly valuable, we find that CEO effects are larger for younger and

highly educated CEOs. We also find that CEO effects are significant in settings

in which the value of managerial discretion is high (i.e., in rapidly growing

environments and in industries with highly educated employees), and that

1Source: stats.oecd.org based on data from 2005 to 2009.

Do CEOs Matter? 1879

large CEO effects obtain in firms with concentrated power such as family firms

and firms with no boards.

In additional analysis, we show that CEO turnover probability is higher after

hospitalizations. While it is possible that convalescing CEOs voluntary leave

their position, another interpretation of this result is that CEO turnover events

represent a revealed preference test of the basic premise of this paper. Specifi-

cally, if CEO hospitalizations undermine managers’ productive potential, then

the firm’s owners or the board of directors act on this information by replacing

the CEO.

Finally, despite the sharp effect of hospitalizations on managerial turnover,

we show that the hospitalization results presented in this paper are not driven

by those firms that replace their CEOs. In particular, we show that the effects

of CEO hospitalizations hold in the subsample of firms that do not experience

a managerial turnover.

Taken together, our results demonstrate that CEOs are valuable for the

organizations they lead. As such, our paper is consistent with a growing line

of research in economics and finance that stresses the unique contribution

of managers to firm outcomes (Bertrand and Schoar (2003), P´

erez-Gonz´

alez

(2006), Bennedsen et al. (2007), Bloom and Van Reenen (2007), Bloom et al.

(2013), among others).

We address several challenges to our identification strategy. The first is that,

while we claim that causality runs from hospitalization length to firm perfor-

mance, it could be the case that superior performance causes shorter hospi-

talizations. For example, if the presence of the CEO is more valuable during

periods of higher profitability, this may lead executives of superior-performing

firms to try to leave the hospital as soon as possible. To address this concern,

we instrument the length of a CEO’s hospital stay using the fraction of lengthy

hospital stays per medical condition for the entire country. Using this instru-

mental variable (IV) approach, we show that our previous results continue to

hold.

The second identification challenge is that hospitalizations can coincide with

periods of poor performance without causing them. First, to the extent that

some illnesses are stress-related, hospitalizations may be triggered by poor

firm performance. In this case, firms on a downward performance trajectory

would be more likely to have a hospitalized CEO. We address this concern by

focusing on the subsample of diagnoses that medical research identifies as not

being stress-related. Our results continue to go through. In addition, we find

no relation between future hospitalization and current performance. Second,

it could be the case that CEOs wait for the right moment to seek treatment,

choosing to do so when performance is poor, perhaps because firm activity and

demands on the CEO are lower. We address this concern by restricting the

sample to only the deadliest diseases for which CEOs are unlikely to wait to

seek treatment. We again find similar results.

Our results extend this literature in several important dimensions. First,

we provide an estimate of the value of CEOs using firm-CEO fixed effects.

These empirical specifications allow us to isolate the consequences of shifting