The New York Times reported last week that the Obama administration was considering a proposal to “allow millions of homeowners with government-backed mortgages to refinance them at today’s lower interest rates, about 4 percent.”

The measure’s supporters tout this as an almost cost-free way to stimulate the economy, boost the housing market and reduce foreclosures. But universal refinancing is far from free, and is poorly designed to stimulate either the economy or the housing market. Certainly, Fannie Mae and Freddie Mac, the huge finance companies now under government control, will have to allow some mortgage modifications in order to reduce their foreclosure losses. That calls for smart, selective policies, not universal refinancing.

This wouldn’t be the first time that the government has gotten involved in the loan-modification and refinancing business. The Home Affordable Modification Program and the Home Affordable Refinance Program provided incentives to loan servicers willing to refinance or modify the loans of struggling homeowners. These were modest, sensible programs, but they have helped fewer homeowners than some might wish.

The most extreme proposal now being floated involves refinancing at current low rates all the mortgages insured by Fannie Mae and Freddie Mac. One of the big appeals is that it seems like free stimulus. As the Times reported, “proponents say the plan carries little risk because the mortgages are already guaranteed by Fannie Mae and Freddie Mac.”

Cost Billions

But even supporters of the plan, including my fellow Bloomberg View columnist Ezra Klein, emphasize that refinancing would mean that “Fannie Mae and Freddie Mac eat billions by giving up their ability to challenge these reps and warrants” given by mortgage issuers in the first place.

While forgoing the right to make banks bear the burden of carelessly vetted mortgages is a real cost, the largest price tag of any refinancing program is the direct one: Mortgages now paying investors 6 percent would pay only 4 percent. Someone would lose those interest payments, which Klein suggests may amount to as much as $85 billion annually.

My quick calculation based on the Times’s interest-rate graphic came to about $35 billion per year over many years. If the program works, agency-insured mortgages would be paid off early and bondholders would take losses essentially equal to the benefits gained by borrowers.

MBS Investors

As nobody loves a mortgage-backed security investor, this may seem like free stimulus, but, like it or not, we are all MBS investors now. When we’re talking about reducing payments to Freddie and Fannie mortgage holders, we’re really talking about reducing payments to ourselves. Our government owns a massive amount of agency-insured debt and would take a significant part of the hit if debtors reduced their payments.

The Federal Reserve System owns almost $900 billion of securities insured by Fannie Mae and Freddie Mac. Those two finance companies, which are now squarely part of our government, have retained mortgage portfolios of $730 billion and $680 billion, respectively. The Treasury still had about $80 billion of these securities on the books as of July. This $2.39 trillion portfolio is roughly equal in size to the Times’s estimate of “$2.4 trillion in mortgages backed by Fannie and Freddie” that “carried interest rates of 4.5 percent or higher.”

This rough equivalence doesn’t mean that private investors won’t bear much of the cost of prepayment, but some significant share of the revenue losses from lower mortgage payments would surely be paid by taxpayers.

Stimulus Effect

What about the program’s benefits? Refinancing makes little sense as stimulus, whose goal is to provide a temporary benefit that induces more spending today. Instead, state-supported refinancing is a benefit that pays off year after year for as long as three decades. As the loss to investors is experienced immediately, while the benefit to credit-constrained borrowers is spread over time, the net effect on the economy may well be negative.

It’s also hard to see why housing markets would be significantly strengthened by lowering the interest payments for existing homeowners. My work suggests that the link between interest rates and housing market is relatively modest, and this refinancing effort won’t do anything to reduce borrowing costs for new buyers.

Lower Payments

The most natural way in which widespread refinancing for troubled homeowners could support the housing market is by reducing foreclosures, and empirical work typically finds weak evidence that lower mortgage payments reduce the likelihood of defaults.

My colleague Christopher Foote, along with Kristopher Gerardi, Lorenz Goette and Paul Willen, estimate that a 20 percent reduction in the initial interest-rate cost of a mortgage would lead the monthly default rate to fall by about 0.4 percentage point for subprime borrowers and less than 0.1 percentage point for standard borrowers. They find that changes in the unemployment rate or housing prices have a significantly bigger impact on defaults than changes in the interest rate.

After all, if refinancing saved so much money by reducing defaults, private lenders would surely find mortgage modification far more appealing than they do.

Government support that enables refinancing among borrowers who couldn’t refinance in the private market on their own is a public transfer to these borrowers.

Rewarding Borrowers

Many people may wonder why the federal government should reward people who borrowed heavily to buy during the boom and whether such generosity doesn’t encourage future risk-taking. Others will surely ask why this aid isn’t means-tested or focused just on the elderly or families with small children, who might suffer most from a move. Still others will question whether it is wise to try to keep people in houses they never should have bought and still can’t afford.

Yet Fannie Mae and Freddie Mac are left with their obligation for large numbers of defaulting and potentially defaulting mortgages. Managing this portfolio makes some mortgage modifications appropriate, as long as these measures are motivated by a sensible desire to reduce taxpayer outlays, rather than high-flying claims about the widespread economy-wide benefits from subsidized refinancing.

Fannie and Freddie also have the problem of managing their stock of already foreclosed homes, and the Times discussed a second proposal to rent some of those homes so that they can be sold slowly.

Rental Proposal

Exploring options makes sense, because selling homes too quickly means low prices, especially if you are trying to move thousands of foreclosed homes at once. Yet the government needs to be quite careful here as well, because renting homes that were meant to be owned is never easy.

The case for slow sales was made in a classic paper by David Genesove and Chris Mayer, which found that Boston condominium sellers with high loan-to-equity levels sold their units more slowly and got substantially higher prices. Steve Levitt and Chad Syverson found that real estate agents, who presumably know more about housing markets than ordinary sellers, typically take 9.5 days longer to sell their homes and receive 3.7 percent higher prices, holding everything else constant.

By contrast, my colleague John Campbell, along with his co-authors Stefano Giglio and Parag Pathak, estimate a general forced-sale discount of 18 percent and a foreclosure discount of 28 percent.

Lost Value

Selling slowly offers some chance of reducing taxpayer losses on these foreclosed sales and keeping housing prices stable, but renting these homes would be tricky. We should also expect significant losses in value from renting, because renters have less incentive than homeowners to take care of the property. A significant literature shows that homes lose value - - as much as 1 percent per year -- when they are rented. The maintenance problem is likely to be exacerbated if, as I expect, the Feds end up trusting someone else to handle the rental contract.

The Federal Housing Finance Agency, the regulator of Fannie Mae and Freddie Mac, is soliciting ideas from the public and investors. I hope that the government doesn’t end up subsidizing private investors to turn landlord and that any landlords would post a significant maintenance deposit to cover damage to the property.

A massive refinancing effort is likely to have little impact on the economy or foreclosures or housing prices. What it would do, however, is hurt our government’s already precarious balance sheet by reducing the payments on its vast mortgage portfolio.

Fannie and Freddie should focus on reducing public losses with sensible, limited mortgage modifications and by selling foreclosed homes reasonably slowly.

(Edward Glaeser, an economics professor at Harvard University, is a Bloomberg View columnist. He is the author of “Triumph of the City.” The opinions expressed are his own.)

To contact the writer of this article: Edward Glaeser at eglaeser@harvard.edu.

To contact the editor responsible for this article: Max Berley at mberley@bloomberg.net.