Time to Get Personal With Insurers

This column was originally published on RealMoney on March 27 at 3:34 p.m. EST. It's being republished as a bonus for TheStreet.com readers.

I favor financial companies with strong balance sheets and high business quality. Over a full market cycle, they will usually outperform the average lower-quality company.

There is an exception, though: When credit spreads are abnormally high (as from October 2002 to January 2004), and the credit cycle turns, it pays to buy low-quality companies that have survived the crisis. This would be one of the few times I would be tempted to lever up and buy all the momentum I can.

Simple screening techniques aren't generally applicable to financials, for the reasons that I outlined last week in the first part of this two-part series. Two relatively similar institutions may write business with decidedly different risk profiles. The one writing the more risky business will typically trade at lower price-to-earnings (P/E) and price-to-book (P/B) ratios, though free cash flow, could it be fairly measured, might show the company writing the higher quality business to be the cheaper one. Also, earnings quality will be higher for the company writing better-quality business and the likelihood of negative surprises will be lower.

I use screens as a starting point. Typically, I use a metric that compares P/B to prospective return on equity (for more on my screening methods click here). This works best when comparing companies in very similar industries. I have nine models like this screening the subsectors of the U.S.-traded insurance universe.

That gives me a reasonable idea of what is rich and cheap. Then the hard step comes: overlaying the simple metric with knowledge of the quality of business written by the companies in question. Sometimes it leads me to buy companies that are technically rich and sell short companies that are technically cheap, because the value of quality business practices often outweighs apparent cheapness.

Let me give you an example from one of my subsectors, personal lines insurers. I set a market-cap limit of $200 million when I defined my group.

Personal lines insurers are facing pricing deterioration, but not as sharp as that of commercial lines insurers. Pure premium rates are falling in the low single digits, on average, but that varies from state to state. There are two relatively large companies leading the wave of price-cutting: St. Paul Travelers(STA) and Geico. Competitors allege that both are underwriting at an expected loss.

Competition is even sharper among the companies that underwrite "nonstandard" risks, i.e., bad driving record, no driving record, etc. The premiums are higher and many companies think that they have a better marketing/pricing model than the competition. Companies writing nonstandard business usually trade at a discount to those that write standard risks.

The ability to trust the published financials is relatively high in the personal lines space because of the "short-tailed" nature of the liabilities. When people suffer losses, they know and report them quickly, usually.

Given that backdrop, consider this graph of P/B to prospective ROE for 19 companies in the personal lines sector:

Gauging Valuations Graphing price-to-book vs. ROE for the companies in the personal lines sector gives a sense of relative valuations

Source: David Merkel

The line going through the center of the dots is the "best fit" or "regression" line. Companies below the line may have P/B ratios too low for their ROEs, meaning that they are seemingly cheap; companies above the line may be overvalued. That gives me a starting point for analysis.

The regression line also can be used to predict price-to-book value ratios going forward. If the companies earn what analysts expect them to earn, we can project what the proper price-to-book value for the company should be relative to where other companies in the same industry trade. From that, it is simple to calculate the expected percentage price increase/decrease from the projected P/B ratio. Big caveat: The regression makes the assumption that earnings growth, required return and dividend policy will remain constant. This simplifying assumption must be tested qualitatively when analyzing companies.

Now consider this table of values from the regression:

Personal Property and Casualty Insurers
Symbol	Name	Market Cap	P/E	P/E (next year)	Est. ROE	P/B	Est. P/B	Possible Gain/Loss
AFFM	Affirmative	$197.3M	8.67	7.72	13%	1.03	1.37	33%
ALFA	Alfa	1.338B	13.74	12.97	14	1.79	1.68	-6
ALL	Allstate	35.184B	20.73	8.94	20	1.75	2.45	40
BRW	Bristol West	571.7M	11.02	11.65	15	1.69	1.59	-6
CGI	Commerce Group	1.796B	7.34	10.11	13	1.36	1.71	26
DRCT	Direct General	346.8M	9.41	8.32	18	1.47	1.66	13
ERIE	Erie Indemnity	3.196B	15.86	14.57	17	2.53	1.98	-22
FAC	First Acceptance	588.3M	24.65	N/A	10	2.53	1.39	-45
HMN	Horace Mann Educator	795.7M	11.10	10.84	13	1.36	1.55	14
IPCC	Infinity Property and Casualty	871.5M	8.21	12.73	11	1.38	1.48	7
KFS	Kingsway Financial	1.126B	8.11	8.35	18	1.50	1.82	21
MCY	Mercury General	3.000B	11.90	12.16	15	1.87	1.88	1
MLAN	Midland	650.6M	10.40	11.29	12	1.37	1.50	9
PGR	Progressive	21.115B	15.22	15.68	22	3.43	2.48	-28
SAFC	Safeco	6.520B	9.79	10.23	16	1.59	2.01	26
SAFT	Safety Insurance	729.2M	7.66	10.75	17	1.86	1.76	-6
STFC	State Auto Financial	1.372B	10.91	10.32	17	1.77	1.84	4
THG	Hanover Insurance	2.803B	N/A	12.52	12	1.44	1.69	17
TW	21st Century Insurance	1.376B	15.79	16.04	10	1.66	1.53	-8
Source: David Merkel

Once I get to this point, I take a deeper look at the 20% in a sector that seem to be the most underpriced and the 20% that are seemingly the most overpriced. In personal lines, that boils down to the top and bottom four.