"Why are mortgage locks so unreliable?"

One reason is that the adverse event that triggers the insurance – a rise in interest rates – affects every locked loan in lenders' pipelines. In contrast, the adverse event that triggers homeowner insurance is usually an isolated event. One house fire will not seriously damage a casualty insurance company, but a rise in interest rates can force a lender who is not adequately hedged into insolvency.

Most lenders hedge against a major hit to their profitability from rising rates. They hedge by executing transactions that will increase their profits when rates increase, offsetting their lock losses. A lender who is fully hedged would not be affected by a rise in rates, but since hedging is costly, few lenders are fully hedged.

A long period of declining interest rates weakens the lock system. Hedging during such a period is money down the drain, so lenders are tempted to do less of it. And a few may actually adopt a "go-for-broke" policy where they don't hedge at all. They look to make as much money as they can during the low-rate period, and go out of business when it ends, leaving failed locks behind. Indeed, a significant proportion of the failed locks in 2003 can be traced to one large lender who evidently pursued such a policy. When it closed its doors, hundreds of borrowers were left stranded.

Another weakness of the lock system is that some borrowers, especially among those refinancing, game the system. They lock the price with a lender, but if rates decline, they lock again with another lender. This practice raises the cost of locking, pushing lenders to find ways to protect themselves.

Some lenders try to protect themselves against this practice by charging a lock fee that is credited back to the borrower at closing but is not refundable if the borrower walks from the deal. Or the lender may insist that the borrower pay one or more fees, such as an appraisal fee, which the borrower would have to pay again if he went with another lender. These are fair conditions, but lenders who impose them place themselves at a competitive disadvantage, so they are far from universal.

A less savory practice that underlies many lock failures is to load the loan approval with conditions that allow the lender to back out. Every lock is conditioned on the borrower being approved for the loan, and approval is frequently subject to conditions. Most of these are completely reasonable, for example, the removal of a lien on the property. But some conditions are designed to allow the lender to exit the lock lawfully.

I recently heard of an interesting one from a puzzled borrower. His commitment letter stated that if the loan application, which the lender had approved, was rejected by the investor to whom the lender intended to sell the mortgage, the lender's lock was no longer valid. This borrower was alert, caught the condition, and asked me what I thought about it. I told him that it was the lender's responsibility, not his, to determine whether he met the investor's requirements. The lender removed the condition.

Many lenders would rather protect themselves with contractual escape clauses rather than charging a non-refundable fee because they know that most borrowers don't read contracts, but fees drive them away. Other things the same, smart borrowers should prefer lenders who charge a non-refundable lock fee. Lenders who protect themselves from being gamed in stable and declining rate markets are more likely to honor their locks in a rising rate market.