«July 7, 2015 World Risk and Insurance Economics Congress Munich, Germany Leon Chen Assistant Professor, Minnesota State University at Mankato Phone: ...»
We then classify our final sample into four groups based on Best’s explanation of the causes of downgrades and their information used, as shown in Table 1. Group 1 has 50 downgrades and they all have Best’s press release news with suspected Best’s private information and/or opinion, and the causes of these downgrades are all related to deterioration of firm’s fundamental performance, earnings and/or financial prospects. Group 2 has 27 downgrades and they are either stated that the downgrade is caused by certain public information, or have no suspected Best’s private information identified in press release news. Similar to Group 1, the causes of Group 2 downgrades are also related to deterioration of firm’s fundamental performance, earnings and/or financial prospects. Recall that we hypothesize in a short-run event window, shareholders should respond negatively to Group 1 downgrades, but not as much, if any, to Group 2 downgrades. Group 3 has 18 downgrades and they are all mainly caused by increased firm risk due to managerial decisions, such as increased financial leverage, new product launching, business expansion, increased investment or asset risk, aggressive pricing, increased business concentration, and acquisitions or spin-offs. Recall that we argue Group 3 downgrades can potentially be good news to shareholders due to the effect of wealth transfer from policyholders to shareholders.
Table 2 presents the rating downgrade matrix for the full sample. Most downgrades (129 out of 158) are one-level downgrades, while 29 are multi-level downgrades. Figure 2 shows the time distribution of Best’s rating downgrades for publicly-traded insurers.
3. Research method We conduct short-run daily abnormal return estimations using the event study method. Stock daily return data are obtained from the CRSP (Center for Research in Security Prices) for the downgraded firms.
Brown and Warner (1985) find that in a short estimation period, both market model and market-adjusted model 3 generates well-specified estimation results and produce no significant mean bias. 4 However, more recently, Ahern (2009) suggests that short-term models may also exhibit statistically significant biases, especially if the researcher uses data that are nonThe market model uses ordinary least squares (OLS) with a market index (i.e., 1-factor CAPM) to estimate expected returns, while the market-adjusted model uses the market index as expected returns.
Mean bias is a measure of statistical error in an event study model. If we choose random securities from the sample and assign random event dates, there should be no abnormal returns on average because the respective benchmark portfolio mean is the “normal return” for insurers in that portfolio and the average deviation from the mean is zero.
If mean abnormal returns are not zero for random securities and random event dates, then the model is said to have mean bias (Brown and Warner, 1985).
representative of the overall market, such as firms from a single industry. He finds that characteristic-based benchmark model, such as the one using the matched size-return portfolio of control stocks, displays the least bias of all models. Since our study uses data from one industry,
we use an industry-benchmark model to calculate the cumulative abnormal returns during the (day period surrounding the downgrade events. More specifically:
where Ri,t benchmark represents the expected return for firm i in day t, calculated by the mean of all publicly-traded insurer returns during that day. 5 ARi,t represents the difference between the expected return and the real return Ri,t, for firm i in day t. The cumulative abnormal return CARi,T is the sum of the abnormal stock return of each day surrounding the rating downgrade events for firm i during the short-run event window T. The cumulative average abnormal return CAARn,T is the mean of CARi,T for all n firms during the event window T.
4. Empirical results We show our results in Table 3 and Figure 3. Our cumulative abnormal returns are calculated in 6 different short-run event windows during the 21-day period surrounding the downgrade event.
We focus the following discussion on the Day 0 (the event day) results.
During the event day, downgrades overall generate statistically significant negative CAAR of -2.83 percent for the full sample. The Group 1 sample shows more negative CAAR of Publicly-traded insurers are identified in CRSP with SIC codes between “6311” and “6411.”
-5.89 percent, while the Group 2 sample does not show significant negative CAAR during the event day. Recall that both Group 1 and 2 downgrades are related to deterioration of firm’s fundamental performance, earnings and/or financial prospects, while Group 1 downgrades represent those with suspected Best’s private information and/or opinion and Group 2 downgrades do not. For all event windows in our study, the Group 1 abnormal returns are consistently lower (i.e., more negative) than the Group 2 sample, and the differences in abnormal returns between the Group 1 and 2 samples are consistently significant. The difference between Group 1 and 2 abnormal returns exceeds 15 percentage points in the (-10, +10) event window.
Our results suggest that market participants respond to rating downgrades differently, depending on whether the information is new and whether it is value relevant. When a rating downgrade is based on private information and the cause is deterioration in an insurer’s fundamental financial prospects, the market response is substantially more negative.
Group 3 downgrades do not show significant negative abnormal returns during most of the selected event windows, and actually shows positive 0.60 percent abnormal returns during the event day. The differences in abnormal returns between Group 1 and 3 are also significant during almost all selected event windows. Recall that Group 3 downgrades are mainly caused by increased firm risk due to managerial decisions, and the results suggest that these downgrades are not necessarily bad news to shareholders, regardless of whether rating agency’s private information or opinion is used. Although an increase in firms risk tends to increase equity values according to option pricing theory, a countervailing influence is that any rating downgrade potentially damages future profitability and premium growth. Group 3 abnormal returns are generally less negative than Group 2 abnormal returns, but the difference is not statistically significant. Our Group 4 sample represents those downgrades with no press release news identified, and the CAARs for that group are similar to those in the full sample for most event windows. This is expected since Group 4 downgrades likely would have represented a mix of Group 1-3 news, if all press release news information had been identified.
Figure 3 shows the comparison of cumulative average abnormal returns for Groups 1-3 during the 21-day event period surrounding the rating downgrade events, and we can clearly see in this figure the differences of abnormal returns between Group 1 and the other two groups.
Interestingly, Figure 3 also seems to suggest some private information/opinion leakage several days before the downgrade events.
5. Conclusion The information content of rating agency’s rating action news has generated great interest to academia and practitioners over time. Existing studies generally do not consider the differences of the downgrade causes and none has differentiated whether rating agencies have used private information, public information, or both.
Using hand-collected press release news of A.M. Best rating downgrade actions for a sample of publicly-traded insurance firms during the period of 1996-2012, we are able to differentiate the market impact according to whether the rating downgrade is based on the rating agency’s private information or opinion. This differentiation we believe is important in understanding whether rating agencies provide incrementally useful information to market participants.
Since our study uses data from one industry, we use an industry-benchmark model to calculate the average abnormal returns during the (-10, +10) 21-day period surrounding the downgrade events. We find that downgrades overall generate statistically significant negative CAAR of -2.83 percent during the event day, and -7.76 percent in the 3-day event window (-1, +1). More negative CAAR of -5.89 percent during the event day and -11.46 percent in the (-1, +1) event window are found for downgrades with the presence of suspected Best’s private information and/or opinion and with deterioration of firm’s fundamental financial prospects as related causes of downgrades,. For the event day, no statistically negative abnormal returns are found for downgrades where there is no Best’s private information/opinion, or where the downgrades are related to increased firm risk due to managerial decisions. And even for event windows wherein abnormal returns are significantly negative for these latter two groups, such returns are still significantly less negative than for downgrades with both private information indicated and deterioration in financial prospects as the downgrade cause.
Our empirical results suggest that shareholders do not respond to all types of downgrades in similar ways, and that the rating agency’s private information or opinion contained in the rating downgrade news is of great value to them. Also, a downgrade due to managerial decisions to increase business risk may not necessarily be bad news to shareholders, since such decisions do not necessarily decrease equity value and the potential effect of wealth transfer from policyholders to shareholders may actually be good news to shareholders. We also find some evidence suggesting some private information leakage several days before the downgrade events.
These results are consistent with our initial hypotheses, and we believe this paper may be interesting to both academics and practitioners, and fills an important gap in this strand of literature regarding shareholders’ responses to rating action news.
Ahern, Kenneth R., 2009, Sample Selection and Event Study Estimation, Journal of Empirical Finance, 16: 466-482.
Best, A.M., 1997-2012, Best’s Key Rating Guide, Property-Casualty Edition (Oldwick, NJ: A.M.
Best Co., Inc.).
Best, A.M., 1997-2012, Best’s Key Rating Guide, Life-Health Edition (Oldwick, NJ: A.M. Best Co., Inc.).
Brown, Stephen J., and Jerold B. Warner, 1985, Using Daily Stock Returns: the Case of Event Studies, Journal of Financial Economics, 14: 3-31.
Chen, Leon, and Steven W. Pottier, 2014, An Investigation of the Short-Run and Long-Run Stock Market Response to Insurer Rating Changes, working paper.
Cowan, Arnold R., 2007, Eventus 8.0 User’s Guide, Standard Edition 2.1. Cowan Research LC, Ames, Iowa Dichev, Ilia D. and Joseph Piotroski, 2001, The Long-Run Stock Returns Following Bond Ratings Changes, Journal of Finance, 56 (1): 173-203.
Epermanis, Karen, and Scott E. Harrington, 2006, Market Discipline in Property/Casualty Insurance: Evidence from Premium Growth Surrounding Changes in Financial Strength Ratings, Journal of Money, Credit, and Banking, 38 (6): 1515-1544.
Fama, Eugene F., 1970, Efficient Capital Markets: A Review of Theory and Empirical Work, Journal of Finance, 25 (2): 383-417.
Fama, Eugene F., 1991, Efficient Capital Markets: II, Journal of Finance, 46(5): 1575-1617.
Fan, Dennis K.K., Raymond W. So, Jason J. Yeh, 2006, Analyst Earnings Forecasts for Publicly Traded Insurance Companies, Review of Quantitative Finance and Accounting, 26:105-136.
Goh, Jeremy C. and Louis H. Ederington, 1993, Is a Bond Rating Downgrade Bad News, Good News, or No News for Stockholders? Journal of Finance, 48(5): 2001-2008.
Griffin, Paul A. and Antonio Z. Sanvicente, 1982, Common Stock Returns and Rating Changes:
A Methodological Comparison, Journal of Finance, 37(1): 103-119.
Halek, Martin, and David L. Eckles, 2010, Effect of Analysts’ Ratings on Insurer Stock Returns:
Evidence of Asymmetric Responses, Journal of Risk and Insurance, 77 (4): 801-827.
Hand, John R. M., Robert W. Holthausen and Richard W. Leftwich, 1992, The Effect of Bond Rating Agency Announcements on Bond and Stock Prices,” Journal of Finance 47(2): 733-752.
Figure 1: Sample Creation Process
Figure 2: Year Distribution of Rating Downgrades Table 3: Cumulative Average Abnormal Stock Returns Based on Different Information Sets and Disclosed Causes of Rating Downgrades
1. Group 1 events represent those publicly-traded insurer downgrades that have Best’s press release news with suspected Best’s private information and/or opinion. Group 2 downgrades are mainly caused by certain public information, or have no suspected Best’s private information identified from Best’s press release news. Both Group 1 and 2 downgrades are related to deterioration of firm’s fundamental performance, earnings and/or financial prospects. Group 3 downgrades are mainly caused by increased firm risk due to managerial decisions, such as increased financial leverage, new product launching, business expansion, increased investment or asset risk, aggressive pricing, increased business concentration, and acquisitions or spin-offs. Group 4 downgrades have no identified Best press release news.
2. The t-stat is the cross-sectional standard deviation test described on page 86 of Cowan’s Eventus 8.0 User Guide. Specifically, t-stat during the event window (T1, T2) equals to the mean of the cumulative abnormal return CAR(T1, T2) divided by the standard deviation of the sample mean of CAR(T1, T2).
Eventus also reports the number of securities with positive and negative cumulative abnormal returns, and sign Z statistic represents the significance level of a generalized sign test, as described on page 88 of the Eventus 8.0 User Guide. The symbols $, *, **, and *** denote statistical significance at the 0.10, 0.05,