Actuarial Independence and Managerial Discretion

• DOI: http://doi.org/10.1111/jori.12199
• Published date: 01 December 2018
• Date: 01 December 2018

ACTUARIAL INDEPENDENCE AND MANAGERIAL DISCRETION

Shinichi Kamiya

Andreas Milidonis

ABSTRACT

Appointed actuaries are res ponsible for estimating the la rgest liability

on property–casualty insu rance companies’ balance s heet. Actuarial

independence is crucial in safeg uarding accurate estimate s, where this

independence is self-regul ated by actuarial profession al institutions.

However, professional con flicts of interest arise when appo inted actuaries

also hold an officer positio n within the same firm, as offic er actuaries

also face managerial incentive s. Using a sample of U.S. insurers that

employ in-house appointed a ctuaries from 2007 to 2014, we fin d evidence

that officer actuaries have dif ferent reserving practices th an nonofficer

actuaries. This difference i n reserving is associated with tax s hielding and

earnings management incent ives. Results are consistent wit h managerial

discretion dominating actu arial independence; they are eco nomically

significant and should be of co ncern to regulators and pro fessional

institutions.

INTRODUCTION

Safeguarding auditor independence is crucial in minimizing management of financial

statements. To limit potential conflicts of interest that might arise between auditors

and audited firms, the Sarbanes–Oxley Act of 2002 puts in place new auditor

requirements. One of the requirements restricts auditing firms from providing

nonaudit services (e.g., consulting) for the same clients. However, this restriction does

not apply to similar professional bodies like the actuarial profession. Actuaries

are responsible for accurately estimating insurance loss reserves (valued at almost

Shinichi Kamiya is at the Division of Banking & Finance, Nanyang Business School, Nanyang

Technological University, S3-B1B-64, 50 Nanyang Avenue, Singapore 639798; Tel. þ65 6790

5718. Kamiya can be contacted via e-mail: skamiya@ntu.edu.sg. Andreas Milidonis is at the

Department of Accounting & Finance, School of Economics & Management, University of

Cyprus, P.O. Box 20537, CY-1678 Nicosia, Cyprus; Tel. þ357 22 89 3626. Milidonis can be

contacted via e-mail: andreas.milidonis@ucy.ac.cy. We would like to thank the two anonymous

referees, Yevgeniy Davydov (discussant), Georges Dionne, Mary Kelly, Andros Kourtellos,

Tyler Leverty, Kevin Shaver (discussant), and participants at the following conferences for their

comments: 2014 American Risk and Insurance Association, 2014 European Group of Risk and

Insurance Economists, and 2014 Taiwan Risk and Insurance Association annual meeting.

© 2016 The Journal of Risk and Insurance. Vol. 85, No. 4, 1055–1082 (2018).

DOI: 10.1111/jori.12199

1055

US$500 billion in 2013),

1

set aside to pay future unpaid contract claims. However,

actuarial independence might be compromised when actuaries are also provided

with managerial incentives.

The actuarial profession is largely self-regulated. Each property–casualty (PC)

insurance company has an appointed actuary, who is responsible of evaluating the

estimation of loss reserves.

2

Appointed actuaries are either external actuarial

consultants or in-house employees of the firm. In-house appointed actuaries can also

hold an officer position in the firm, which means that they are actively involved in

firm’s managerial or policy-making decisions (NAIC, 2011). Such managerial duties

are typically coupled with a more convex compensation structure that rewards better

firm performance. Clearly, a direct conflict of professional interest arises between

actuarial independence and managerial discretion for appointed actuaries who also

hold an officer position.

In this article, we focus on revisions in loss reserve estimates produced by in-house

appointed actuaries of U.S. PC insurance companies. We manually separate in-house

appointed actuaries, into officer actuaries (OA; in-house appointed actuaries who

also hold an officer position) and nonofficer actuaries (NOA; in-house appointed

actuaries who do not hold an officer position). Using data from 2007 to 2014, we

estimate 5-year reserve errors as the difference between an originally reported reserve

estimate and the revised estimate 5 years later, and test whether reserve errors differ

between OA and NOA. Reserve errors are expected to be zero on average. However,

we observe deviations from originally reported reserves in both cases. Specifically,

OA have a lower proportion of reserve errors being zero and a higher proportion

in the negative region (underreserving) during the sample period.

3

This study

investigates whether the officer role of the appointed actuary is associated with

systematic differences in loss reserve errors.

Actuarial independence can be compromised when appointed actuaries also hold an

officer position, for two reasons. First OA face a more convex compensation structure

than NOA, such as performance-related bonuses (Eastman et al., 2014; D.W. Simpson

& Company, 2015; XpertHR, 2015), which encourage improving short-term

profitability for the firm, with a parallel short-term compensation benefit for OA.

Second, the duties of OA go beyond those of NOA (i.e., to oversee the accurate

estimation of loss reserves) as they are also part of the management team, which faces

incentives that could affect loss reserving practices. In summary, OA face stronger

1

Authors’ calculations from www.snl.com.

2

Appointed actuary is defined to be: “...an actuary who is appointed or retained in accordance

with the provisions of law, regulation, or contract or other arrangement, as the designee to

issue a statement of actuarial opinion” (Actuarial Standards Board, 2000). The appointed

actuary for each firm is appointed by the firm’s board of directors.

3

The means of 5-year reserve errors for OA and NOA during our sample period are 1.02 and

2.19, respectively (Table 2, Panel B). The mean difference between the two mutually exclusive

groups of actuaries is statistically significant at the 1 percent significance level (two-sample t-

test with unequal variance). A plot of the reserve error distributions (for the full sample and

also by the type of actuary) is shown in the Supplementary Material.

1056 THE JOURNAL OF RISK AND INSURANCE