Fairfax: no clues from the past

With bank accounting and words such as HTM (held-to-maturity), AFS (available-for-sale), and unrealized losses becoming a part of every investor's daily vocabulary from the verge of a couple of large failing banks, I've found there's an erroneous tendency to treat all financial institutions the same. On the face of it, you get why investors do it. Financial institutions hold money for other people. They then take that money, called float, and invest it in relatively liquid assets, earning the difference until the money will eventually have to be paid back. The source of the funds varies between types of institutions. For banks, it's mostly depositors. For insurers, it's mostly policyholders. But the game is the same: managing money and earning the difference between what comes in and what comes out.

The source of the float and the methods to manage it mostly distinguish a bank from an insurer. Unlike banks, insurers bear solvency risk on both sides of the balance sheet, in the risks they underwrite and the assets they buy. Where banks know the running cost of their liabilities, sloppy underwriting at an insurer can nullify or worsen the returns generated by investing the float, but can also turn an advantage if the underwriting is profitable. And unlike banks, recent news being case in point, insurers aren't subject to bank runs. When bank depositors want their money quickly, there's little a bank can do but sell its fixed income portfolio to meet redemptions, and if there's a large unrealized loss on bonds that the bank initially booked to hold to maturity, a run can punch a hole in the bank's statutory capital. Insurers can be subject to tragedies of equal proportions from catastrophes, but generally employ less leverage and can reinsure for protection. Besides, insurance liabilities are more predictable. If an insurer has 5-year claims on average, that's 20% of the portfolio that needs to be liquidated at max (if there's a 15% unrealized loss on the portfolio, that would only be a 3% hit). Both banking and insurance come down to term matching, but because of the insurer's added visibility, insurers are generally more flexible with how they can invest their float.

But just like banking isn't an inherently great business, insurance isn't either. The insurance industry has low entry barriers and marginal differentiation among a large number of underwriters, any of whom may onboard recklessly priced risk at any given time. Still, some factors can turn individual banks and insurers into good businesses. Since this writeup is about insurance, I'll put banking aside for now and focus on what makes a good insurance company.

From a risk management perspective, you can build a mental model of insurance through two dimensions:

1) Opportunistic vs non-opportunistic insurers: When investors think about opportunistic insurers, they often assume that some companies are better at cycle management than others, expanding when markets are hard and pulling back when conditions soften and competition increases, as if the industry in aggregate forsakes underwriting discipline and overly focuses on top-line growth even as premium rates decline. This is often boilerplate. All insurance companies boast that they have a culture of "opportunistic" underwriting discipline, but that's table stakes to be in business. Some insurers are better than others, whether it be from experience or courage. But when I say "opportunistic", I mean on the investment side.

On the investment side, the vast majority of the industry falls in the non-opportunistic camp. This isn't necessarily bad. "Opportunistic" is usually not a word you want to be passed around about you in the money management business. Non-opportunistic insurers (think State Farm, Allstate, or Nationwide) focus almost entirely on underwriting, either operating profitably by writing carefully selective policies or ramping up leverage to earn a satisfactory ROE by investing the float in low-risk, low-return investments that match their liabilities—an ancillary activity typically outsourced to 3rd-party managers. On the other hand, opportunistic insurers (think Berkshire, Markel, or Fairfax) not only allocate a significant part of their float to equities but also buy non-insurance businesses outright. Because this approach, when done right, may bring outsized returns compared to non-opportunistic insurers, opportunistic insurers may give more leeway as to how profitably the underwriting side of the business operates, as long as the money lost is less (meaning cheaper) than equal opportunities to raise money. Both models work within different instances and competences, but investors tend to favor those that get more of their income from consistently profitable underwriting than those that get most of their income from the float—the reason being that so few insurers manage to underwrite profitably every year that doing so indicates a hard-to-copy advantage, whereas anyone can look smart for a while by gearing up the float or reaching to yield. In other words, an investment edge is harder to verify than an underwriting edge. The counterargument, of course, is that <90% combined ratios aren't the holy grail either, since that could mean the insurer isn't optimizing for float size, forsaking an important income stream.

2) Short-tail vs long-tail risks: The "tail" is the time required to learn of and settle claims. It's an important consideration when an insurer manages its reserves. Short-tail claims (e.g., most property, auto, general liability, or travel) are settled quickly. Each loss event is small, and the law of large numbers applies, making reserving and pricing more synchronized to maintain profitability. But the same factors make short-tail underwriting a price-driven commodity and a game of reaching the lowest-cost advantage. Long-tail claims (e.g., asbestos, workers' compensation, or product liability) take years or decades to settle and are subject to more uncertainty in reserving. In extreme cases, like asbestos, long-tail claims may take upwards of 40 years to settle, continuously lurking on the insurer's balance sheet, occasionally requiring painful revisions to cover sins of the distant past. With something short-tail and granular like auto, claims patterns are risky but not uncertain. Long-tail is more uncertain but isn't as price-sensitive, making it prone to reckless underwriting since consistent under-pricing can take years to uncover.

This brings us to why insurance (or banking) isn't an inherently great business. An inherently great business can be run by anyone without inept management inflicting mortal damage. That isn't insurance. In insurance, the ability and temperament of management are all-encompassing because it's too easy to bet the company, and there are lots of temptations to do the wrong thing, either excessively reaching for yield, causing term mismatch, or writing loads of bad premiums. Such short-termism is insurance's kryptonite. Because insurance is, by and large, a commoditized necessity, like fresh air, there's no substitute other than for the uninsured to mechanically chip away at their paycheck, stuff the money under the mattress, and pray for good weather. Reckless management at an insurer can easily tap into that need and achieve outsized growth by aggressively pricing policies or accepting risks that prudent competitors are wise enough to avoid. So growth in itself is far from a reliable measure in insurance. If anything, abnormal growth in premiums written is more often a cause for concern than something to celebrate.

That is with four unlesses: as an insurer, you can rationally outgrow competitors even in a soft market if you either 1) have a low-cost competitive advantage, perhaps stemming from superior technology or sales channels, 2) cushion your growth with consistently prudent reserving, 3) have an investment edge that can afford a higher combined ratio, and/or 4) have a customer-centric culture, translating into higher retention rates and LTV (on a unit basis, a newly-written policy is typically money-losing in year one but gets increasingly below a 100% combined ratio in later years). All four characteristics feed on each other. With a low expense ratio, you're more likely to think longer term and avoid imprudent reserving to boost quarterly numbers. You can invest for the long term and provide superior service to your policyholders. Insurance brokers appreciate that and are more likely to direct volumes your way.

Fairfax Financial has all four characteristics. Commonly called the 'Canadian Berkshire' due to its value investing principles (but a fluke at 1/40th of the size in terms of market cap), it falls right into the opportunistic camp. 15 years ago, if you'd asked an insurance analyst about Fairfax, they'd tell you that it isn't an insurance company, but an investment company masquerading as an insurance company. And 15 years ago, they'd have been right. Fairfax was never your general insurer focused on maximizing underwriting profits, limiting volatility, and protecting dividends in any given year. And compared to Berkshire, which mostly writes short-tail risk, Fairfax leans more to the long-tail and was never afraid of a little extra volatility on both sides of the balance sheet, trusting that the Hamblin Watsa investment committee's acumen would more than make up for it. Fairfax's first 20 years of operation: