RATING CHANGES AND COMPETING INFORMATION: EVIDENCE ON PUBLICLY TRADED INSURANCE FIRMS

• DOI: http://doi.org/10.1111/jori.12181
• Author: Leon Chen,Steven W. Pottier
• Date: 01 September 2018
• Published date: 01 September 2018

RATING CHANGES AND COMPETING INFORMATION:

EVIDENCE ON PUBLICLY TRADED INSURANCE FIRMS

Leon Chen

Steven W. Pottier

ABSTRACT

We examine the predictive ability of three competing sources of financial

information—rating changes, profit changes, and excess stock returns. We

find the following significant relations between current and lagged values of

these three endogenous variables: (1) rating changes are positively related to

past excess stock returns, (2) profit changes are positively related to past

excess stock returns, and (3) profit changes and excess stock returns aremean

reverting. In addition, profit changes are substantially more predictable than

rating changes or excess stock returns, and past values of profit changes

account for most of the observed ability to predict current profit changes. In

contrast, past profit changes have little predictive ability in relation to excess

stock returns or rating changes.

INTRODUCTION

Rating changes, profitability, and stock returns are competing sources of information

on the performance and prospects of insurance firms. A general research question is

whether one source of information is a substitute for another for purposes of

predicting the future direction of firm financial performance. While the existing

literature on insurance firms examines limited aspects of the information content and

predictability of rating changes, stock returns, and profit changes, ours is the first

study to undertake a comprehensive analysis controlling for competing information.

In theory, the stock market should respond only to new information contained in

rating changes or earnings announcements (Kothari, 2001). But, to what extent is the

market response to an information event due to the information event versus

Leon Chen is Associate Professor at the Department of Finance, College of Business, 150 Morris

Hall, Minnesota State University, Mankato, MN. Chen can be contacted via e-mail:

yilin.chen@mnsu.edu. Steven W. Pottier is Associate Professor at the Department of Insurance,

Legal Studies, and Real Estate, Terry College of Business, 206 Brooks Hall, University of

Georgia, Athens, GA. Pottier can be contacted via e-mail: spottier@uga.edu.

© 2016 The Journal of Risk and Insurance. Vol. 85, No. 3, 811–842 (2018).

DOI: 10.1111/jori.12181

811

other contemporaneous publicly available information? Investors, regulators, and

academics

1

are interested in questions of information content and incremental

predictive ability for several reasons, including the potential to profit from market

inefficiencies, the impact of market behavior on the firm’s financial prospects, and

gauging the relative importance of competing and costly information.

Rating agencies and stock investors evaluate financial information about insurance

firms to form expectations regarding future cash flows. Stock returns reflect investors’

assessments of firm value. Ratings reflect the rating agency’s assessment of firm

performance and financial prospects. Insurer financial ratings are related to the

insurance firm as a whole, while bond ratings are related to a specific bond issue.

Insurer rating agencies claim that their assessments are based on both publicly

available information about the rated insurer as well as private information from

company managers. On the other hand, equity market participants must, in general,

rely on publicly available information to form their expectations about future profits

(i.e., earnings), cash flows, and market risks. Accounting profit measures are

generated within the firm and eventually reflect almost all aspects of firm

performance. In sum, all three of these measures, ratings—stock returns, and

profits—are comprehensive financial measures.

In this study, we examine the relation between lagged val ues of our three financial

measures—rating chang es, excess stock return s, and profit changes—and their

current value. The main part of our empirical ana lysis consists of three different

sets of regressions; one set corresponds to each of these three financial measures as

the endogenous/depen dent variable (rating changes, excess stock ret urns, and

profit changes). In essen ce, we apply the Granger ca usality (GC) concept to

investigate the empiric al relations between cur rent and lagged endogenous

variables and the relative predictive ability of ea ch set of lagged endogenous

variables. While we do provide some evidence on con temporaneous relatio ns, this

is not the purview of GC ana lysis. The use of lagged en dogenous and lagged

exogenous variables to predict the current val ues of rating changes, ex cess stock

returns, and profit chan ges allows us to more care fully control for information

available to market par ticipants before current values are known. C learly, investors

and other parties are inter ested in stock return and ear nings forecasts, and our

study provides evidence on forecasts in terms of predictive abi lity. In general, GC

analysis and event studi es are both interested in t he information conten t of some

metric(s). As a multiva riate regression approach, GC is better suited to control for

other information even ts preceding the announ cement of the dependent va riable,

while event studies that use s hort event windows, in prin ciple, have fewer

confounding informat ion events to control for . The central contributi on of our

study is the investigat ion of these three broad financial measures in a sing le study.

This allows us to use a con sistent sample and set o f variables. As discussed next,

the importance of the info rmation content of these t hree financial measures i s

1

Francis, Schipper, and Vincent (2002) investigate whether competing information, primarily

analyst reports, reduces the usefulness of earnings announcements, and find that analyst

reports do not substitute for earnings announcements.

812 THE JOURNAL OF RISK AND INSURANCE

recognized in the academ ic literature, but due to d ifferences in sample or m ethod

or variables, are unable to a nswer questions about th e incremental informat ion

content of the lagged endogenous and exogenous financial variables used in t he

present study.

Prior studies in the insur ance literature on the relationship between ra ting changes

and stock returns are limit ed to event studies. Halek an d Eckles (2010) examine the

stock market response to chan ges in insurer financial strength ratings and find

negative stock market resp onses to rating downgrades, but no significant res ponses

to rating upgrades. Chen, Gaver, and Pottier (2016) examine stock returns during

a 25-month event window centered on the month of the rating change. They find

that stock returns move in t he direction of insurer rat ing changes in the 12-month

period prior to the announcement. There is an addi tional stock price resp onse

following the announce ment of a downgrade, but no response to upgrade

announcements.

Epermanis and Harrington (2006) analyze premium growth surrounding insurer

rating changes for property–casualty insurers, and find evidence that premiums

decrease following rating downgrades. These authors do not consider stock market

response because most of their sample firms are not publicly traded. Their study

does not directly address insurer profitability and does not address the issue of

incremental information content.

Our main findings are as fo llows. First, positive (negative) abnormal stock returns

precede rating upgrade s (downgrades). While ex cess stock returns move in the

same direction as upcomin g rating changes, rating changes are difficu lt to predict, a

result we surmise is likel y due to the low frequency of rating changes. Second,

abnormal stock returns a re higher (lower) precedi ng increases (decreases) in

profits; however, their a bility to predict profit ch anges is inconsequent ial after

controlling for past profi t changes. Third, past pro fit changes have weak predict ive

power in relation to both cu rrent abnormal stock ret urns and current rating

changes. Fourth, profit changes and exce ss stock returns are mea n reverting,

consistent with the liter ature on publicly traded fi rms generally. Fifth, ex cess stock

returns are negatively (p ositively) related to prior threshold rating down grades in

the full sample period (fin ancial crisis subperiod ). Regardless of these ob served

relations, excess stock ret urns are substantially mo re difficult to predict than pr ofit

changes and exhibit low au tocorrelation. Furth er, contemporaneous ex cess stock

returns are positively related to rating changes and profit chan ges, suggesting that

the latter two information sources are partly reflected in stock returns in the same

time interval. Lastly, we d o not find convincing evidence that the stock market is

inefficient.

The remainder of this article is structured as follows. In the next section, we survey

relevant literature and develop hypotheses. The third section discusses our sample

and data selections. The fourth section describes the variables we use in our analysis.

The fifth section presents summary statistics of the variables. The sixth section

explains the application of Granger causality analysis in the context of the present

study. The seventh section presents regression analysis results and the last section

concludes this study.

RATING CHANGES AND COMPETING INFORMATION 813