No, that’s not a typo. The anatomy of an only-in-Washington credit snafu
By, Michael Grunwald
POLITICO MAGAZINE: If you’re curious just what kind of risks the US government is taking with its $3.3 trillion in loan programs — a portfolio considerably larger and significantly stranger than any private bank’s — the best place to start is the Federal Credit Supplement, an obscure batch of tables stashed in the back of the annual White House budget proposal.
When it was released earlier this month, the 95-page supplement revealed a wealth of data about the growing costs of federal credit programs, which deal with everything from historically black colleges to boll weevil eradication to Pacific ground fishermen. For instance, it projects that taxpayers will recover 46 cents on every dollar loaned to renovate rural apartment complexes, but will earn a 24-cent profit on every dollar loaned for Indian tribe land acquisition. Its suggests that risky U.S. loan guarantees for biorefineries and Ukraine are actually getting somewhat safer. It’s where I recently discovered a surprise $21.8 billion shortfall in the federal student loan program.
But the weirdest number lurking in this year’s credit supplement was in Table 3, where the White House budget office explains the assumptions behind its cost estimates for various loan programs. What caught my eye was the default rate for an Agriculture Department program called Broadband Treasury Rate Loans: 116.37 percent.
A default rate above 100%? Was that a typo?
It was not a typo.
So what was it? The average default rate for bank loans is about 3 percent. The troubled student loan program has lifetime default rates around 25 percent. How on earth could a credit program, even a risky one, get to 116 percent? Were the recipients defaulting en masse, then stealing an extra 16 percent from the Treasury?
The explanation I eventually got from the Obama administration was not that damning. But it wasn’t exactly comforting, either. The crazy number was apparently produced by flawed execution of a flawed model of a flawed program. In reality, the Agriculture Department expects to recover about 80 cents of every dollar it lends to telecoms to extend high-speed Internet to underserved rural areas. Administration officials couldn’t pinpoint the actual default rate, but it’s much lower than 116%. They say the main culprits for that wrong number were a radically overbroad definition of “default,” as well as some inappropriate double-counting.
Basically, it’s complicated, which is true of all federal credit programs—which is a problem with federal credit programs.
“Very few people pay attention to this stuff,” says Roman Iwachiw, a financial analyst whose Washington consultancy helps government agencies and banks grapple with federal credit programs. “If more people read the credit supplement, the government would have a better incentive to make the numbers make sense.”
The loan program in question, created during the George W. Bush administration, is part of a larger rural broadband effort the Agriculture Department describes as a 21st century version of the New Deal’s rural electrification, linking Americans who don’t live near cities to one of the necessities of modern life. The Obama administration says it has brought new or improved high-speed Internet to 1.5 million rural Americans since 2009, increasing their productivity, expanding their access to quality health care and education, and reducing the “digital divide.” The case for government assistance—and it’s largely a bipartisan case—is that it’s often unprofitable for the private sector to extend service to sparsely populated areas.
Unfortunately, loans to private firms pursuing unprofitable activities don’t always get paid back. Last year, a Government Accountability Office review of the broadband program’s first 100 loans, totaling almost $2 billion, revealed serious problems; 18 had already defaulted, and another 25 had been rescinded before any money went out. The feds had lost $78 million on a single 2008 loan to Open Range Communications, a wireless provider that went bankrupt in 2011. And the GAO found that the Agriculture Department had no idea why its loans were going bad or which kind of loans were likely to go bad in the future; it didn’t even have performance measures to determine whether the program was achieving its goals.
“It was clear they didn’t have a handle on why they had such a significant failure rate,” says Mark Goldstein, GAO’s director for physical infrastructure.
However, that “significant” failure rate wasn’t anywhere near 116 percent. That surreal number emerged from an Agriculture Department economic model for the broadband loan program. There is no standard cost model for the government’s 120 or so credit programs; the dozen Cabinet departments that oversee them develop their own models for their own programs. And Agriculture Department officials acknowledge that the model for their 2015 default rate had some glaring errors, which were apparently spotted by outside accountants at Ernst & Young.
“We had a faulty model that created some double-counting,” says USDA spokesman David Sandretti. “It was a mistake, but it’s been resolved.”
In fact, because of the tweaks to the model, the Obama budget projects a default rate of just under 70 percent for 2016. Which is still absurdly high—and which administration officials say isn’t quite right, either. Even if you set aside the double-counting problem that’s been fixed, the entry for broadband loans in the column headed “Default Rate” is really a combination of defaults and delinquencies. And it’s calculated by accounting for missed interest payments as well as the original loan amounts, which explains how it could have risen above 100 percent. Budget officials say the broadband program gives its borrowers unusual leeway to restructure loans if they’re having financial problems, so when late payments are counted as defaults, the default rate explodes.
“These numbers are exaggerated, and it’s important to be careful when you look at them,” one White House budget official told me.
My questions about the 116% led to an odd conversation with the White House officials, who at times suggested that the weird number in the credit supplement demonstrated their commitment to transparency, and at times suggested that it was silly to expect numbers in a supplement that hardly anyone reads to reflect reality. They sounded annoyed that I had cherry-picked an embarrassing statistic, and implied that if I were a budget expert I would have known the default rate was not a default rate.
Still, the rural broadband program does have a high default rate, although no one could tell me exactly how high; the program’s new “credit subsidy score,” a proxy for the estimated cost of its defaults over time, is 21.87 percent. The USDA’s Sandretti said the Obama administration has stepped up its monitoring of the program, checking in with every borrower every month. It is also making fewer loans—only eight in the first five years under Obama, versus ninety in the last five years under Bush—and none of them have defaulted. The larger point, he said, is that families who live in the country deserve online connections as fast as the ones in cities and suburbs, even if some of the broadband loans end up going bad.
“Clearly, we want to make sure we’re good stewards of taxpayer dollars,” Sandretti said. “But high-speed Internet access is as essential to economic health in this century as electricity was in the 20th century, and rural America isn’t going to get it through the private sector alone. If it were easy, someone would’ve done it already.”
Ultimately, the 116 percent figure in the Federal Credit Supplement may tell us less about broadband loans in particular than about the government’s massive loan portfolio in general. It’s hard for insiders and nearly impossible for outsiders to figure out how much various programs will cost, or what kind of assumptions lie behind those cost estimates. It’s not surprising that the government has to hire consultants like Roman Iwachiw and accountants like Ernst & Young to help interpret its own rules and its own numbers.
“From the outside, it definitely looks complicated and arcane,” Iwachiw says. “It’s not designed to make it easy for the general public to understand.”
Michael Grunwald is a senior staff writer for Politico Magazine where this story originally published and is reprinted here with POLITICO’S permission.