Cities Are Buying Bond Insurance That May Be Giving Them Nothing
Around noon one Wednesday in July, two school districts from California’s Central Valley auctioned off their bonds to Wall Street underwriters.
Both had the same credit rating. The deals were of similar size. They were being issued for the same purpose of refinancing higher-cost debt, and each gave banks the option to include the cost of insuring the bonds against default to help get them the lowest possible interest rates.
There was one key difference, though. The Washington Unified School District’s debt was sold without insurance. But the winning bidder for Stanislaus Union School District’s bonds included a $17,800 policy from Build America Mutual Assurance Co. to guarantee against the remote risk that the money won’t be paid back. That extra layer of security for investors was supposed to save the district money by making buyers willing to accept lower yields.
It didn’t.
When the results of the auctions came in, the Stanislaus district wound up with slightly higher rates than Washington on every single bond with the same maturity, suggesting it got little or nothing in return for its money.
The sales underscore doubts about the business model of the bond-insurance industry, which once backed more than half of the nation’s municipal debt issues until the biggest companies were roiled by the financial crisis more than a decade ago. A study by researchers from Pennsylvania State University and the University of Georgia found that buying such insurance is a bad deal for the vast majority of governments: Between 2009 and 2016, those that insured their general-obligation bonds paid a total of about $260 million more in borrowing costs than their model suggested they should have, even before the costs of the policies are factored in.
“We tortured the data every which way to Tuesday trying to find some evidence that this insurance wrap was lowering offering yields and we just can’t,” said Kimberly Cornaggia, an assistant professor of finance at Penn State’s Smeal College of Business and one of the study’s authors.
Build America Mutual and Assured Guaranty Ltd. disputed the study’s results, saying the analysis was flawed because it failed to account for market movements between bond pricings and included AAA and AA rated governments that rarely buy insurance. Both said the results are in conflict with the savings that lower-rated local governments and their financial advisers see from buying bond insurance.
“BAM members choose insurance because it generates net savings on their transactions, period,” said Grant Dewey, the head of Municipal Capital Markets at Build America Mutual. “Their underwriters and municipal advisers calculate and confirm that every day by evaluating spreads on comparable transactions and by comparing investor demand for insured and uninsured bonds.”
He said the comparison of the two California districts doesn’t adequately capture the savings, since Stanislaus’s sale, unlike Washington’s, included some longer-maturing securities that typically get the most benefit from insurance. Nathalie Wells, chief business official for the Stanislaus district, didn’t respond to requests for comment.
The bond insurance industry has shrunk to a shadow of its former self since the credit crisis, when MBIA Inc., Ambac Financial Group Inc. and Assured Guaranty were stripped of their AAA ratings because of losses tied to toxic mortgage-backed securities. The major credit rating companies also adjusted their methods in response to an outcry from public officials who said they had been consistently rating states and cities too low, leaving them buying guarantees from insurance companies that were at far bigger risk of defaulting.
At its peak in 2005, insurance covered 57% of new municipal bond issues. This year, insurers guaranteed about 6% of the $265 billion of fixed-rate debt issued. Assured Guaranty and Build America Mutual, the industry’s two main companies, are rated AA by S&P Global Ratings, the third-highest rank.
The results of the study by Cornaggia, her Penn State colleague Giang Nguyen and the University of Georgia’s John Hund suggest the business should be even smaller. About two-thirds of insured general-obligation bonds issued between 2008 and 2016 had A ratings — a level high enough that they didn’t reap any savings from buying insurance over that time, according to the authors’ calculations. About 8% had even higher grades.
The researchers studied a sample of more than 700,000 general-obligation bonds issued between 1985 and 2016.
Over the 30-year period, municipalities saved $496 million on their interest bills by paying for insurance, since for much of that time they benefited from the high ratings of the insurance companies they paid. Still, that savings was a fraction of the estimated $17 billion in premiums collected from state and local governments by just two big insurers, MBIA and Ambac, from 1995 to 2008, according to the scholars.
Prior to the financial crisis, “if you bought insurance from a triple-A rated insurance company, it got priced like a triple-A bond, but as soon as those insurance companies are single-A or even double-A rated that’s not triple-A certification and the investors just aren’t as interested,” Cornaggia said.
The study’s authors determined how much insurance costs or saves a municipality by using a model that factored in characteristics like bond size, issue size, state, credit rating, use of proceeds, maturity, and coupon to predict what the yield on a particular insured bond would be without insurance.
If the yield on the insured bond was higher than the yield on an uninsured bond with the same characteristics predicted by the model, the municipality was considered to have lost money, according to the authors.
Yet some of the findings are puzzling. For example, it shows that insurance, on average, lowered borrowing costs for Aa2 rated bonds over 30 years, but didn’t save money for lower-rated Aa3 and A1 bonds. Cornaggia said she couldn’t explain why.
“We are agnostic empiricists,” she said.
Municipalities aren’t the only ones buying overpriced insurance, Cornaggia said, comparing it to fabric stain coverage sold at furniture stores or warranties on televisions. The difference with bond insurance is that the beneficiaries of the policy — investors — aren’t paying the premiums.
“The person making the decision about whether to insure the bond isn’t actually the taxpayers, it’s an agent making a decision on behalf of those principals,” she said. “I do think there’s a risk aversion on the part of the municipal officer making the decision.”