Description

Zenkoku Hosho is a company that could only exist in Japan.

On the surface, it's a commodity mortgage insurer -- like MBIA and Ambac. This is a pretty crappy business and I wouldn't normally give it a second look. However, Zenkoku operates in a very unusual set of circumstances that actually makes it an extremely interesting and attractive business.

Zenkoku insures mortgages, with about a 6.7% market share. In Japan, 80% of mortgages are "insured" at the outset. The borrower is responsible for shopping for mortgage insurance before he can take out a loan. Zenkoku is the dominant supplier of "third party" insurance. It primarily competes against the internal mortgage insurance departments of client banks -- particularly regional banks and co-ops. Of course, this is pretty weird. In effect, the banks are writing mortgage insurance against their own loans -- which is not insurance at all. This exercise simply acts to create price opacity and hassle for the borrower, without distributing any of the risks at all. Like I said: only in Japan.

Banks like having Zenoku around for several reasons -- capital efficiency, giving appearance of consumer choice, the speed of quotes given (vs other 3rd party insurers), and the option of offloading certain loans to Zenkoku for strategic reasons. I gather a big use case is when a regional bank expands to an geographical area which it's not familiar with. It will seek to offload a big portion of the risk from its balance sheet.

If a loan defaults, Zenkoku makes the bank whole. It then forecloses and sells the property, and takes a loss on the difference between the sale price and the principal outstanding.

Now, here's where it gets more interesting. The internal underwriting models of these banks, particularly regionals, appear to be terrible. In fact, they mostly offer mortgage insurance on a flat rate basis. So, no matter your creditworthiness, you will be quoted the same price. Zenkoku on the other hand actually has a credit model, generated from a mix of their own data (it's been around for 30 years and has underwritten a huge number of loans) and conventional credit bureau/public data. It's able to price based on risk -- from 8bps to maybe 50bps annualized. They are not forced to price or compete for every loan. So they can reject especially non-credit-worthy customers. Needless to say, this is a huge advantage. I don't think their credit model is terribly sophisticated when you line it up against U.S. standards, but it's definitely better than the competition.

Why don't the Japanese banks do this? It makes no sense. But like I said, it's Japan. These banks have always done business this way and they don't seem to want to change.

Even more absurdly, Zenkoku generates massive float when it grows. This is because it gets the entire amount of the mortgage insurance upfront. So, let's say you pay 10bps per year for a 10 year loan. Zenkoku will receive 10bps x 10 year x average principal outstanding, or roughly 50bps of the face value of the loan. This huge float means that liquidity risk is not a major issue.

Given that Zenkoku is not regulated by the banking authorities, they voluntarily keep a healthy amount of capital on hand -- equal to average losses over the last 10 years * 10. They also stress test by taking the highest loss rate over the last 10 years * the highest loss severity over the last 10 years * 120%. Under this "worst case", they can cover single year loss about 6x with capital on hand. This seems overkill to me, but they do it to keep a high credit rating and to re-assure banks that they can pay for losses. In reality, credit losses has never been a significant issue.

The Japanese mortgage market is likely in secular decline of about 1% per year. Zenkoku is a share gainer in a declining market. They are slowly taking a larger share of their traditional regional bank / Co-op base, but they are also signing new clients and particularly penetrating the tier 1 banks.

Nevertheless, I don't particularly like declining markets. On the plus side, the valuation is pretty cheap at about 14x NTM. They also pay a decent dividend. In fact, compared to what they can (and should) get away with, they are massively overcapitalized and pay way too little to shareholders. I think it's possible they get more aggressive here in the next several years, as ROEs decline from historical levels thanks to excess cash on balance sheet.

Catalyst

none