Weekly Preliminary OS Climate Risk Review (5/13/21)

By May 20, 2021 No Comments

Wildfire season is right around the corner – or actually arrived in some states – and while California is typically the big headline grabbing state each year, this week we’ve plucked out a non-disclosing issuer located in the lesser known wildfire-risQy state of Utah. While this week’s issuer ranks at the 98th percentile nationally for wildfire-driven property value-at-risk, the issuer doesn’t mention ‘wildfire’ or ‘climate change’ even once in its preliminary official statement.

This week’s climate non-disclosure burn victim:

Board of Education of Tooele County School District, Utah — General Obligation School Building Bonds (46,340,000)

The Tooele County School District (risQ Score 2.4; Wildfire risQ Score 3.8) is located in northwest Utah and has a relatively low population density. While the District has a surface area six times the size of Rhode Island, its population of ~70,000 is less than 1/15th as large as the small New England state. The District’s northeast corner borders the Great Salt Lake and its western region is covered by the Great Salt Lake Desert, which is home to the Bonneville Salts Flats and Speedway. With the exception of land speed record attempts not a lot of population density there.

That’s probably for the best given historical wildfire activity in the area. According to the USGS Monitoring Trends in Burn Severity project (MTBS) there have been no less than 67 fires in the district between 2000 – 2018, with multiple areas in the District burning more than once. It’s no surprise that Tooele County (coterminous with the School District) ranks in the 94th percentile nationally amongst counties for wildfire insurance claims during the past 36 years.

We’re not asserting that it’s primarily the District’s responsibility to address its precarious exposure to wildfire. However, the fact that the issuer doesn’t mention its wildfire risk even once in its POS is eyebrow-raising, especially considering how the severity, frequency and location of wildfire activity is currently evolving within both the State and District. County seat Tooele City is home to roughly half of the District’s population, and the map below shows wildfire events during the 2000 – 2018 period. All fires shown in the right image below are within 17 miles of the City, with the closest proximity burn event — the 2014 Anaconda Fire — spreading less than 3.5 miles from the City’s northeastern boundary. The Anaconda fire resulted in 1,100 acres burned and caused some within the city to evacuate.

Image comes from the MTBS website showing the burn severities of fires in Tooele County School District (left) and around Tooele City (right) during 2000 – 2018. Tooele City is encircled in orange and the Anaconda Fire is indicated with an orange arrow.

Urban areas such as Tooele City will see increased wildfire activity as global warming perturbs foundational climatic parameters such as humidity, moisture and rainfall. Residents of Utah have already been affected by these impacts, as extreme drought has plagued the State in recent years. Governor Spencer Cox of Utah called a state of emergency  in March of this year given the statewide drought conditions (also not disclosed in the POS). As of the beginning of April, the U.S. Drought Monitor categorized 90% of Utah as experiencing “extreme drought”, with over 2.7 million people in the state feeling the impacts. Recent drought conditions continue an emerging trend for the State, wherein the summer of 2020 was logged as the driest in Utah since records started being kept 126 years ago. The State doesn’t appear to have the situation well under control, either — while Utah is considered the 2nd ‘driest’ state, just trailing Nevada, Utah also ranks 2nd just behind Idaho in terms of water-use per capita (source).

Utah’s unprecedented drought conditions are leading to an unprecedented early onset of wildfire activity in the state. As of early April 2021, over 126 wildfires covering 6,200 acres have already burned across Utah, well above the 5-year running average for this time of year of 46 wildfires covering 189 acres (source). While Utah has experienced wildfire activity in March in the past, the increasingly dry conditions have created an environment where wildfires spark more often and spread faster and farther.

You can’t quantify risk by looking into the rearview mirror. In an effort to measure evolving burn probabilities resulting from our changing climate, so we collaborated with the U.S. Forest Service (USFS) on a best-in-class climate conditioned wildfire model. The risQ-USFS wildfire model employs a methodology that combines the outputs of global climate models (GCMs) and historical observations into a novel stochastic modeling framework wherein:

  1. historical reanalysis data is compared with historical GCM experiments, providing for a quantitative measure of past biases in GCM models,
  2. past biases are then used to simulate thousands of new bias corrected GCM outputs that are independent of the original models, and
  3. a suite of projected time series provides us with robust statistics on how meteorological variables may change under future climate conditions.

The two ‘future climate conditions’ that risQ models are RCP 4.5, which assumes carbon dioxide emissions start declining by approximately 2045, and RCP 8.5, which assumes that emissions continue to rise through till the end of the 21st century. These represent reasonable goalposts based on current trajectories and aggregated global policy and action. The difference in wildfire risk increases the further out into the future we look. In 2030, the difference in expected property value at risk between the two climate scenarios is negligible; however by 2060 the delta between the two scenarios increases results in a 27% difference in the increase in wildfire risk. An RCP 8.5 path drives a material difference in risk for Tooele School District compared to the RCP 4.5. Today, Tooele ranks 4th amongst school districts in Utah with a Wildfire risQ Score of 3.8, only behind Tintic (4.7), Juab (3.9) and Washington (4.4).

The District’s material risks really start to shake out when we dig into the low probability – high impact wildfire events. The table below shows the wildfire ‘tail risk’ for property value replacement costs and GDP impairment in the District, rolled up over three time horizons. While the chances of a highly impactful ‘500-year’ wildfire event occurring in any given year is of course inherently low, the odds become orders of magnitude more yikes-inducing if we aggregate over a 10- to 20-year time horizon. In that context, there is a ~2% probability of a wildfire event resulting in 74% property losses in the School District within the next decade. That said, lets just recall just how many fires seem to be occurring in the district and ask yourself how those odds are feeling.

As the risks posed by wildfire events grow in materiality and pertinence to municipal debt market participants, risQ strives to give its clients a real-time view of these events as they’re unfolding. Each day we match the CUSIP-connected location database to fire perimeters sourced from the National Interagency Fire Center and flag any CUSIPs associated with those locations that are within or near the fire perimeter. Even if a wildfire event isn’t directly impacting a municipal asset in an investor’s portfolio, our Wildfire Alerts enable clients to keep an eye on events burning in close proximity to their portfolio of investments, and identify early warning signals of emerging wildfire exposure. Clients who have uploaded their portfolios can see any holdings that are affected by these fires under the My portfolio wildfire impact section of the Wildfire Alerts page.  For the 2021 fire season, those alerts will also include a more meaningful view into the potential impacts of wildfire events on the developed area of a location, as well as bringing in additional real-time localized wildfire data for the states that make those datasets publicly accessible. 

Tooele County School District’s POS covers everything from cyber to COVID-19 risk, despite the fact that the District has never experienced a material cyber breach and enrollment has increased 25% due to the increase in online learning as a result of COVID-19. The fact that neither wildfire nor drought are mentioned even once in the entire 168 page disclosure document leaves us wondering whether the Issuer has any clue of how severe these emerging risks are becoming. Of the $46 million of debt being issued, over $26 million of that is slated for maturity between 2031-2041 when climate change will be a thing and COVID-19 a distant memory, relatively speaking.

Honorable Mentions

The State of West Virginia (risQ Score 1.6; Flood risQ Score 3.6)

Among states, West Virginia has the 5th highest Flood risQ Score, trailing only coastal locations – Louisiana, Florida, Delaware and Maine – and trailing only Louisiana for inland flood risk specifically. That should be enough to warrant some discussion, but when you add on the local economy’s reliance on fossil fuels and the state’s electricity generation being 91% coal-fired, all the more reason to look at the POS balancing act (if there is any attempt at balance at all).

The quick analysis on the climate risk disclosure side. No mention of “climate change” across 658 pages of POS, whether in reference to physical climate risk or carbon transition risk. “Flood” is mentioned 4 times, and all around severe 2016 flooding that required a drawdown – fittingly – from the state’s rainy day funds. Beyond that, nothing. You really don’t need to look too hard to find flood events impacting West Virginia beyond just the noted 2016 event, but feel free to satisfy yourselves. By our count, a total of 1440 serially flooding properties have already been bought out state wide, and that doesn’t happen from just one event.

We also know on a national basis across all counties that both population change and property value growth can be correlated with flood risk. West Virginia has this across its counties and at the state level in aggregate. The POS itself records a 3.8% population decline since 2012. Our data shows that only 7 out of the state’s 55 counties experienced any population growth over the same period, while some experienced double digit percentage declines. In terms of property value growth, no county in the state is in the top half of counties nationally, and the majority are in the bottom quartile. Some might say that this is as much about loss of jobs rather than climate risk correlated, but the numbers on a national county cohort basis don’t lie.

All that said, the state has ample carbon transition risk as well. In contrast to “flood”, “coal” is mentioned 105 times across the POS and, by the document’s own acknowledgement, accounted for 1/3 of the state’s GDP growth from 2017 onwards. Our own Scope 1 emissions data for every county in the country only reinforces the state’s tenuous position. On an overall Scope 1 emissions per capita basis which takes into account electricity generation, transportation, industrial and residential/commercial sources, 7 out of 55 (13%) of West Virginia counties are in the highest emitting 6% of counties. This  makes the state overrepresented by a factor of two in the higher emitting counties. In aggregate, the state overall trails only Wyoming and North Dakota on a Scope 1 emissions per capita basis. In terms of absolute electricity production Scope 1 emissions alone, those same 7 counties are in the worst 3% of counties, making West Virginia counties overrepresented nationally by a factor of four in the highest emitters. The map below shows per capita Scope 1 electricity emissions by county (on log scale) with West Virginia counties highlighted in the blue circle. At the state level, West Virginia trails only Wyoming on Scope 1 electricity production emissions per capita, keeping in mind that the Wyoming has 1/3 the population.

We can take the exposure to fossil fuel risk even further back to the source. In our workplace demographic data, the only industry category with a job count above the 30th percentile nationally for the state is Mining, Quarrying, and Oil and Gas Extraction, and that’s a massive 86th percentile in absolute terms which is monstrous for such a small state. Again, only Wyoming and North Dakota are more reliant on primary resource extraction, and fossil fuel extraction specifically, for their economic wellbeing. West Virginia has 2.5-3 times the population of those comparables, making their absolute hole much deeper to dig out of.

That doesn’t bode well, but you won’t find any mention of “risk” in and around the secular trends away from fossil fuels. We tried to get creative looking for “carbon” (as in carbon transition risk) and didn’t find anything either.

Summarizing, no mention of all the flood risk and the need for mitigating that risk. No mention of carbon transition risk in any context. That’s a climate risk non-disclosure double-whammy.

Coral Gables, FL (risQ Score 4.1; Hurricane risQ Score 4.0; Flood risQ Score 3.4)

Maybe not as verbose on their climate (change) risks and actions as nearby Miami Beach but evidence both in the OS and elsewhere of awareness and action. The couple of paragraphs under “Climate Change Issues” are pretty serviceable although the specific actions being taken are a little behind where Coral Gables’ risk profile suggest they should be. Of note:

Scientific studies on climate change show that “sea levels will rise, extreme temperatures will become more common, and extreme weather events will become more frequent” (emphasis is ours). That’s a change from many issuers looking to dilute the risks and their attribution.

We’re probably beyond the time of talking and consulting and into a time of action and execution. Hurricane Irma in 2017 wasn’t so severe for the Coral Gables area relatively speaking but the city still lists $5.8 million of costs still being taken in the finances in 2020. In addition, a lot of focus is on sea level rise, and the combination of coastal flood and hurricane storm surge is indeed 55% of the risk. That said, 35% of the climate risk isn’t coastal – its hurricane precipitation and inland flooding – so a more holistic view of the risk might be wise when it comes to infrastructure spending and mitigation.

That said, there are concrete steps in terms of financial preparation. Page 11 of Appendix B lays out the 10.75% increase in user rates for the specific purpose of funding future infrastructure projects to address sea level rise. The goal is to raise an additional funds totaling $100 million by 2040. We can also see that Coral Gables has a rating of 5 in the NFIP’s Community Rating Survey (CRS). Only two jurisdictions in Florida – the Town of Cutler Bay, and the City of Palm Coast – rate better and both of those are 4’s.

Could the POS be better? Sure. That said, if every issuer met this standard we would be in good shape and, as you know, we’re far from that.

Volusia County School Board, FL (risQ Score 3.5; Flood risQ Score 4.0; Hurricane risQ Score 3.0)

First – and in direct contrast to the above – on page 82 you will find:

Scientific studies on climate change show that “sea levels may rise, extreme temperatures may become more common, and extreme weather events may become more frequent”. 

See what they did there? Setting that aside, from there you can find copy/paste language on potential impacts but nothing to suggest there’s anything material to worry about. However, a quick glance directly above to the Property Insurance section indicates that people who know a thing or two about risk (i.e. insurers) are a little more circumspect in their view of the risk. The school boards property insurance costs have climbed to such a degree recently that they’ve had to reduce their coverage. We wonder if that’s a clue that something could be up with this climate risk thing. 

Unlike for Coral Gables above, no mention of any local efforts around risk mitigation. Interestingly, however, Volusia County also has a 5 rating in the NFIP CRS. There have also been enough buy-outs of serially flooding properties to put Volusia at the 96th percentile nationally. Apparently someone out there thinks that the risk is worth addressing even if the school board doesn’t see the need to talk about it or maintain prior levels of flood insurance. Much of the local population feels the same way given that 30% of the flood risk to property is not insured, with all the associated implications on property value and recovery after a flooding event. Just let it ride and roll the dice.

Side note: for a school district, this is on the low end for percentage of population under 18, which seems like a metric to care about. 24th percentile in the state and 16th percentile nationally. We’ll see how this plays out demographically over time. 

Pasco County, FL (risQ Score 3.0; Flood risQ Score 3.9)

The last of this week’s Florida POS triplets, although by far the smallest in par value at $7.4 million, so we won’t belabor this. Four lines of text on page 26 summarize climate change, natural disasters, “flood”, “hurricane” and any other related risk you might care to name. Perhaps the attitude conveyed in the POS correlates with the local population’s concern: only around 25% of the flood risk to residential property is insured. That’s a problem for a GO bond, we would think. Even more so for a county at the 96th percentile nationally and right at the risQiest 1/3 of counties in the state of Florida.

Just as one additional note here, the risQ Score of 3.0 applies to 2031, but if we go all the way to the 2041 callable option, you can take that to 3.3 instead. A 1.0 point increase in risQ Score is a doubling of overall risk, so a 0.3 change is meaningful (if you think climate change is a thing, that is).

Leave a Reply