Murray Rothbard came up with a pretty good argument (the inalienability of moral agency) as to why the performance of a promise should be unenforceble if the person changed their mind. Coercing performance would amount to contractual slavery. But he considered defaulting on debt to be a form of fraud at the time of the agreement (an exchange of values including future payment by one party), and therefore believed the payment of debt could be compelled.

IMO blogger "quasibill" at The Bell Tower has made some pretty good arguments, on the same moral agency basis as Rothbard's argument against contractual slavery, as to why a creditor should have no power to compel payment beyond the right to seize assets at the time of default.
http://tinyurl.com/622brg

Since "fraud" requires assumption of a state of mind that is almost impossible to prove, garnishing wages or otherwise coercing payment of the debt amounts to coercing specific performance of a promise. In order to legitimately do so, the lender has the burden of proof to show deliberate deception at the time of agreement. If not, the agreement to pay amounts to just another promise of the kind that Rothbard said was unenforceable.

It is the responsibility of the lender to take adequate care to verify the ability of the borrower to pay, and to take proper precautions (like making sure the loan is adequately secured by collateral) for minimizing loss in the event of default. As quasibill argued in a post to LeftLibertarian2, the actual contract is complete at the time the property changes hands:

"In actuality, a loan agreement involves a transaction whereby lender gives property in exchange for a promise of future performance. The contract is "complete" at that moment. The lender is in complete control of this, BTW - he can refuse to transfer the property until he is subjectively assured of the value of the promise. Some possibilities, including what you have mentioned, are insurance, or bonds, or co-signers..."

The borrower's half of the contract is his promise to pay, coupled with whatever assurances and safeguards he can provide at the time. It is up to the lender to make sure that the safeguards are sufficient before letting the property change hands.

In the case of a mortgage, the lender was protected twice over against default by such safeguards (by the property itself as collateral, and by the high risk premium in the interest rate); if they failed, it's the lender's fault. Normally, in a mortgage, the property itself is considered the collateral against default. If the collapse of an asset bubble (a bubble deliberately inflated by the mortgage brokers themselves) means that this turns out not to be enough, well they can put it on their TS list. In addition, the interest rate the borrowers are paying constitutes another precaution, as part of it amounts to an insurance premium against default. If somebody declares bankruptcy on a 30% credit card debt, the risk of such a default on unsecured debt is precisely what they were paying such high interest for in the first place.