In Depth Analysis: CalculatedRisk Newsletter on Real Estate (Ad Free) Read it here.

Thursday, March 22, 2007

Private Mortgage Insurance for UberNerds II: Flows, Pools, Bulks, and Captives

by Tanta on 3/22/2007 10:24:00 PM

In our last episode, we found our hero, MI, throwing itself in front of our hapless lender, bravely absorbing the first impact of the real estate train wreck. Before we go on to talk about other issues, let’s make sure we understand what “loss position” means. If you’re used to thinking of insurance in terms of things like auto or health, you’re used to being "the insured." You're also in the first loss position: that’s what a “deductible” is. It means that in an insurable event, you will take the first loss up to your deductible amount, and only if the losses exceed that limit will the insurer step up and take the rest of it. MI doesn’t work that way (with the possible exception of “captive reinsurance” issues, which we’ll talk about below).

MI is designed to be the first loss position: it covers losses up to the coverage level amount, and only if losses exceed that amount will the lender (who is “the insured”) take the rest of it. It's rather like the reverse of a deductible. The lender/investor doesn’t spend its own money until the MI has spent all the money it is contractually obligated to spend first.

This may answer the questions some people had about how the insurer and the lender might be somewhat differently motivated when it comes to mitigating losses on a loan. The MI will be pushing for collection efforts, forbearance, modifications, and so on, whenever it’s prudent (in the MI’s perspective), because the MI pays first in a foreclosure and doesn't want to do so unnecessarily. The lender might want to foreclose sooner rather than later, because in an early foreclosure, most losses are likely to be covered by the MI. Who gets its way is a question of how the fine print in the policy reads. The lender/servicer is not obligated by law to follow the MI’s requirements. It is obligated by sheer self-interest: if you don’t follow the MI’s servicing rules, the MI won’t pay the claim.

Mortgage insurance policies can come in several flavors. The most common is the primary policy. This is a policy that insures some percent of the losses on an individual loan. Primary policies can be written on a flow basis or a bulk basis. “Flow” means that policies are written on a loan-by-loan basis, as they are originated. An originator could then take a pool of loans that have already been covered with flow primary policies, and securitize them. If the amount of coverage on the existing flow primary policies is sufficient, no further insuring of the pool of loans is necessary. “Bulk” means that policies are written to cover individual loans, but they are written all at once, after the loans have already been closed, on a big portfolio of loans.

Flow primary policies can be “borrower paid” or “lender paid.” Bulk primary policies are always “lender paid.” This part is confusing, because in real-world terms, the borrower is always paying for the MI somehow. The difference is really in how the borrower pays it (which can impact how well the risk is priced, as we’ll see).

When we use the term “borrower paid,” we mean that the mortgage insurance premium for the individual loan is set by the insurer before the loan is closed, according to the insurer’s underwriting guidelines, and is paid out of the borrower’s monthly loan payment (the “T&I” or “escrow” portion). Some premiums are paid monthly; the servicer takes the monthly part out of the borrower’s payment and remits it to the MI each month. Some are paid annually; the servicer puts the monthly part of the borrower’s payment into an escrow account, and then remits the annual premium to the MI when it is due (after sufficient funds have built up in the escrow account). The borrower never pays the MI premiums directly; this is important. The borrower is not the insured; the lender is. The servicer must collect premiums from the borrower and then send them to the MI. The borrower is not responsible for making sure that happens; the borrower is only responsible for making the MI part of the monthly payment to the servicer as specified in the loan closing documents. The MI’s job is to make sure the servicer is doing its job.

The industry term “lender-paid” refers to a situation where the borrower’s closing docs do not specify a monthly premium portion to be paid into escrow. Instead, the borrower just pays a higher interest rate on the loan, and the lender pays the MI premium out of its interest income, which (ideally) reduces the yield on a lender-paid loan to the equivalent of the yield on a borrower-paid loan. In this regard, it works like servicing fees: the servicer takes the required slice off the interest payment needed to pay the MI premiums, and then passes through the remainder to the investor. You can see why it’s still really paid by the borrower: it comes out of the interest payments the borrower makes. It’s called “lender-paid” because it does not come out of an escrow account funded directly by borrower T&I payments.

Lender-paid primary flow insurance (LPMI) got really popular for a while, because it was marketed as “tax advantaged.” (Countrywide’s product is actually called “TAMI,” Tax Advantaged Mortgage Insurance.) The idea is that mortgage interest is tax-deductible but mortgage premiums are not (although that’s changing); the borrower could make more or less the same monthly payment on the lender-paid plan, but deduct more on the tax return because it was interest instead of premium.

Some of us think this wasn’t always such a great deal for the borrower; it depends on the fine print in the closing documents. Specifically, the higher-rate lender-paid deal is OK as long as you get a nearly free option to modify your loan back to market rates once your LTV drops under 80%. (“Nearly free” would mean you paid only the cost of a new appraisal to verify LTV; you would have to pay that to cancel borrower-paid insurance, too.) If you aren’t given such an option, lender-paid MI can cost you more over time even with the deduction, since it’s costlier to get rid of it (you’d basically have to pay refinance costs and hope rates are low at the time). Anyone who wishes to insist that LPMI is a great deal for the consumer should ponder the fact that rating agencies and securitizers tend to prefer it to borrower-paid MI precisely because it is marginally more persistent, meaning that borrowers pay higher rates longer than they pay monthly insurance premiums. If and when prevailing rates rise substantially, some borrowers will find it difficult or unpleasant to try to “refi” out of LPMI.

Bulk policies, of course, are all LPMI (since the MI is added to the loans in bulk transactions, after they’re closed, they have to be LPMI instead of borrower-paid. You can’t go back to a borrower after you’ve closed the loan and say, hey, we decided you need to start paying MI.) There are two important issues for bulk: first, these are still loan-level policies; it’s really just a matter of writing a whole lot of individual policies at once, rather than on a flow as-the-loans-close basis. Second, from the insurer’s perspective, bulk policies are nearly always of weaker average credit quality than flow policies. It’s rather like the difference between pork loin and sausages. Bulk loan portfolios are the “sausages,” and they’ll always have some percentage of “filler.” As long as they still taste something like pork loin—that is, as long as the averages across the portfolio of loans is OK—then they’re still insurable. But the premium structure for bulk is different from flow, to take the credit quality differences into account.

What is called “pool insurance” is a pig of a different color. Pool insurance is a way of covering a big group of loans, like bulk, but unlike bulk, pool insurance is not a primary policy. A primary policy covers a maximum percent of the losses on an individual loan. It is not “cross-default” insurance, meaning that if you don’t lose the whole maximum claim amount on a given loan, you can’t apply the remaining claim amount to another defaulted loan. With pool insurance, you can apply the total policy maximum to different loans. Pool insurance can be applied to a pool of loans that already have primary policies, a pool of loans that do not have primary policies, or a mixture of the two. It generally has a maximum dollar amount or percent of pool balance that it will cover; individual loan losses are applied to that limit until it is reached, and then the remaining pool balance is uninsured. When there is a loss on a loan with a primary policy, the primary policy pays first to its maximum claim amount, and then the pool insurance kicks in for any remaining loss.

So pool insurance can either supplement or replace primary coverage, depending on the pool, the agreement, and the underlying loans. Pool insurance is often called “lender paid,” but it’s really mostly used in securitizations and so it’s more properly considered “bondholder paid.” It’s a form of credit enhancement, and like any other credit enhancement, its cost comes out of the yield of the underlying loans. It is cost-effective to the bondholder only to the extent that the underlying loans 1) throw off enough income to leave a reasonable net yield after the cost of the credit enhancement and 2) are of high enough credit quality themselves to be eligible for relatively inexpensive premiums.

Bulk and pool insurance can be like a belt added to the primary flow policy suspenders. You will sometimes hear of lenders—we’re hearing quite a bit of this lately—adding MI to a portfolio of loans that they’ve held for some time. The idea is that they sniffed the wind and decided perhaps more insurance coverage would be a good idea. This may make the shareholders of those portfolio lenders feel better.

Before getting too warm and fuzzy over it, though, shareholders might remember two things: MI companies can smell change in the air just as much if not more than lenders can; being on the hook for losses does tend to sharpen one’s wits. Buying insurance while you’re young and healthy (flow primary) tends to be cheaper than buying it when you’re sicker and older (bulk or pool policies acquired after “sniffing the wind”). Furthermore, “lender-paid” MI is a profitable proposition to the lender only insofar as the lender originally set the borrower’s interest rate at closing to a level high enough to cover the insurance costs. Having to go back and insure an originally uninsured or underinsured portfolio can mean having to spend interest income that did not originally “budget” for insurance. Ouch. So a lender who announces that it has just insured (or increased the insurance on) some part of its portfolio is telling you that 1) it thinks the outlook is worse than what it predicted when it closed the loans and 2) it just reduced its income on that portfolio and 3) it thinks that loss of income is worth it, which tells you, in essence, how much worse the outlook is.

Today’s final topic is this “captive reinsurance” thing. In short, that’s a setup in which a lender forms a subsidiary which “captures” part of the mortgage insurance premium. In exchange, it agrees to take some part of the MI’s exposure. A typical arrangement might involve the lender’s captive taking 25% of the premium in exchange for covering 25% of the first loss.

The devil, with these things, is always in the details; whether this is a good deal or bad deal for either party depends on which 25% the captive is taking. It could involve a split, meaning the captive pays 25 cents for every 75 cents the insurer pays, starting with the first dollar paid. Or, it could mean the captive pays the first 25% of losses, and then the MI pays any remaining losses up to 75% of the maximum claim. Or the MI could pay the first 25%, the captive the second 25%, and the MI the last 50%. You have to read the agreement; these things aren’t exactly standardized.

Suffice it to say that, in the boom years, they were a real money-maker for lenders, since losses were so low what with bubble prices and endless refis. Whether they will continue to be, and whether some lenders may yet re-discover the meaning of that old phrase about chickens coming home to roost, is yet to be seen. On the whole, my personal view is that if there’s a sucker at the table somewhere, it probably isn’t the MI. They are in the business of insuring against losses, and most of them have been at it long enough to have been there, done that, got scars to show for it. A few of these “new, quickly growing mortgage lenders!” who emerged more or less in the boom may not necessarily be pricing their risk quite exactly right, in my view.

So far, we haven’t addressed the big question of how, then, the MI’s underwriting guidelines stack up against everyone else’s. That’s a critical question, if you want to know how much risk the MIs have taken on, and how likely they are to withstand the losses they might have to take. But it’s also a subject for the next installment.