For the last 18 months, we’ve been providing climate risk analytics for the entire municipal bond universe. We have rich underlying data to look at property value at risk, GDP impairment risk, and carbon transition risk, and have taken this high granularity data and simplified it down to a single 0 to 5 risQ Score, where a 5 is worst. Our US Fixed Income clients are now using this to drive investment decisions.
In parallel, we have developed and the same capabilities across mortgage-backed securities. Those products are now available, including the risQ Score at the loan, pool and portfolio levels. In agency RMBS the implications are alarming, to say the least. For agency CRTs, the differences are stark on both percentage of UPB at risk, as well as the dollar values of UPB at risk. The idea that agency CRT instruments are all sufficiently large and homogeneous that there is no difference to worry about is a total myth. Some key data you should know:
- On a %UPB basis, the highest risk CRT pools have 5.9% of loans with a risQ Score greater than 4.0, amounting to $1.3 billion, while for the lowest risk pools have only 0.3% in this very high risk cohort, or less than $30 million in UPB within those pools.
- Kicking that up a notch to the extremely high risk cohort of loans with a risQ Score greater than 4.5, the highest risk pools have greater than 3% of loans to worry about, versus only 0.1% for the least risky pools.
So depending on your climate risk threshold, that’s a 20 to 30 x difference in risk profile.
Why should you care? Firstly, there is a direct correlation between climate risk and loan delinquency, validated across over 26 million agency loans from 2005-2018. The summary risQ Score data for 12 million Freddie loans is shown here.
Secondly, much of that risk is not insured, averaging around 80% across all the securitized mortgages, but with variability on a pool-to-pool basis. For securitized loans – or any loans for that matter – already at the higher end of the risk spectrum for non-climate reasons, its not hard to imagine a climate event tipping point putting that debt literally and figuratively under water.
Thirdly, climate risk is being concentrated into securitized pools versus the overall loan universe.. The data below shows the securitization rate of conforming loans from 2009 to 2019 versus the Hurricane risQ Score component.
Thirdly, this will all flow through to home price appreciation, whether through escalating home insurance costs that home buyers will factor into their decisions, or through an outright change in demand for high risk properties as buyers become more informed and risks become more apparent.
Finally, climate change is a thing, which means the proportion of properties (and their securitized debt) in high risk locations will only keep growing bringing more potential delinquency into play, and the severity of climate risk for those already at risk will only get worse only compounding the severity that’s already been in evidence.
This is starting to catch the eye of both regulators and investors, given the recent request by the Federal Housing Finance Agency’s for input on climate risks, and the responses they received, in one case by our partners at DeltaTerra Capital. The FHFA make those public but in advance of that, Al Yoon at Debtwire reviewed the submission and wrote a piece summarizing much of that content, and providing further commentary. Some excerpts from Al’s article include:
- Fannie Mae, Freddie Mac, the Federal Home Loan banks and credit risk transfer bond investors are facing potential losses approaching USD 50bn based on their exposures to hurricanes, floods and wildfires, DeltaTerra founder David Burt wrote in a 19 April letter answering the Federal Housing Finance Agency’s January request for input on climate risks. Those losses would exceed USD 107bn in DeltaTerra’s “bear loss” scenario.
“The “greatest and most imminent threat to the agencies” is that many home markets are priced as if the costs of natural disasters are not rising, despite past evidence of inadequate insurance and increasing damages.”…”Some 20% of homes in the continental US are in communities impacted by mispricing of flood and wildfire risk”Dave Burt, CEO of DeltaTerra Capital
- Investors and the government-sponsored enterprises aren’t properly accounting for rising climate risks, in light of the storms that have walloped parts of the US in recent years and expectations around increasingly severe weather.
“Risk officers at large mortgage companies are “scrambling” to understand what those costs will do to profits, some in anticipation of regulatory scrutiny”Clifford Rossi, Professor at the University of Maryland’s Robert H. Smith School of Business.
- A report this month from the Climate Cost Project put a spotlight on the under-appreciated costs of climate risks. Among the findings, the authors discovered that 81% of 127 Horry County, South Carolina, properties surveyed flooded at least two times between 2016 and 2020, and a third of those were in flood zones designated as low-risk areas by FEMA flood maps. Nearly two-thirds of homeowners surveyed had to go into debt to cover losses, according to the report.
“We were surprised to see so much disparity in climate risk in these pools, and variability of flood insurance the property owners carry”Chris Hartshorn, Chief Commercial Officer of risQ
Let us know if you would like the full pdf of that article.
Agency RMBS investors have to get in front of the climate risk that is facing them now. Investors are blind to that risk as a result of lack of disclosure about the locations of loan pool collateral in general, and also the gross underrepresentation of risk that Special Flood Hazard Area status actually indicates. Finally, some investors are stubbornly and willingly ignoring the risk through a combination of myths: that the pools aren’t that different; and that climate risk doesn’t matter anyway. Such outmoded conventional wisdom is becoming decreasingly conventional and certainly are not wise given the wealth of data and insight now available.
risQ Scores don’t lie.