We’ve commented in the past that geospatial specificity is critical to understanding climate risk and socioeconomic factors for a given municipal bond issuer. To often, state- or county-level information is used by cities, school districts, utility districts as well as the diaspora of revenue bond issuers across the bond issuing universe. This week’s Preliminary Official Statement from the Channelview Independent School District (Harris County, TX) is a good example of how a combination of selective and geospatially non–specific information can be “brought to life”.
Some quick statistics on Channelview ISD to get you started. The Property VaR – the critical metric for a GO issuer – is 95th percentile nationally and 91st percentile in Texas. All the flood components contribute – inland, hurricane precipitation, storm surge and coastal – with garden variety inland flood contributing around 56% of the overall risk to 2030. Indeed, hurricane precipitation represents 22% of the overall Property VaR once annualized probabilities are taken into account, while the expected equivalent property loss from a one-time hurricane precipitation event is 19%. You would have been hiding under a rock for a few years not to have some idea of Harris County’s Hurricane Harvey experience and have a general idea of the risk profile. However, there are 21 school districts that overlap with Harris County, and Channelview ISD is in the worst three for both Property VaR and GDP Impairment Risk. While Harris County’s Cumulative Property VaR to 2030 is 16.4%, the range across the school districts spans a range from 12.7% to 28.1%, with Channelview ISD at 22.1%. If you use Harris County to judge risk to Channelview ISD, you’re significantly underestimating the risk.
In the OS there is the obligatory language regarding Harvey and the impact on southeast Texas on page 34, but with a specific declaration that the district received minimal damage and reimbursable FEMA costs of $30,000. There’s plenty of evidence that damage to the district was much higher than $30,000, and that this number most likely corresponds to the school districts own facilities versus the district in its entirety. A side-by-side images below that only shows the Channelview area on the San Jacinto itself, damage is substantial and reports abound of a large number of residences entirely being washed away, and residents choosing to leave versus rebuild. Still more discussed how a proximate superfund site – linked to the areas history petrochemical activity – impacted the Channelview area. Beyond the questionable minimalizing of past events, the OS makes no mention that the risk of such events are likely to increase or have the potential to do so. Nor is there mention of risk mitigation efforts specific to the district. Finally, no mention of the risk that climate events might have for the underlying ad valorem property tax base in the body of the OS. That said, the observant amongst you would see in the Selected Financial Information that the Taxable Assessed Value dipped by 9% the year after Harvey. No attribution or alternate explanation was provided in the OS that we could see.
Beyond all this we can always point to out usual metrics for Harris County: NFIP claims/capita/year (worst 1% nationally), flood insurance gap (worst 10% nationally) and mismatch of FEMA flood risk to overall flood risk (less than 25% of the risk is accounted for). Given Channelview ISD’s higher risk profile it is not hard to imagine the implications. In addition, Channelview ISD is in the bottom quarter of the 21 Harris County school districts on Per Capita Income, Average Owner Occupied Housing Value and even Total Population. Resiliency to future hurricane events is going to be weaker as a result. The climate change conditioned probability of any hurricane hitting the district by 2031 (the earliest call date) is 21%, and 57% by 2046 (the latest maturity). Even the odds of a Category 3 hurricane (or higher) are a robust 11% and 40% at the same dates. In parallel, a 100 year event inland flood with 9.6% and 22% probabilities for 2031 and 2046 has a 16.5% Property VaR, quite separate (and additive to) the probability and impact of hurricane events. There is plenty of literature available on what repeated flooding events do to the likelihood of people staying, and the property values and insurance costs of those that do. None of this is good. None if it is discussed in the OS.
Coming back to the concept of geospatial specificity required to understand risk, its also possible to view various MUDs, WCIDs and FWSDs that overlap with the district and see where how the risk is distributed. Recalling that Channelview ISD’s Cumulative Property VaR to 2030 is 22%, its FWSD 47 (34%), FWSD 6 (29%) and MUD 285 (26%) that have outsized risk. These all sit on the periphery of the ISD. In contrast, MUD 53 (21%), WCID 21 (17%) and WCID 84 (16%) are more central. Lets not forget that Harris County overall has a 16% Cumulative Property VaR to 2030, just for context versus the aforementioned county-wide statistics.
From a carbon transition perspective, we can point to indicators in the OS as well as using risQ data that point to outsized risk. The largest tax payers in the district are topped by Lyondell Chemical while others in the same petrochemical space also feature further down. Supporting this reliance, 26% of the jobs in the district come from within the petrochemical value chain (oil & gas, manufacturing and logistics) versus 17% for Harris County as a whole. The district has an exposure to prior manufacturing activity as indicated by the large superfund site, but it has present day (and therefore future) financial risk as well.
A lot to chew on in this, but the bottom line is that the OS was abjectly poor in characterizing the climate risk to the district or even showing a basic understanding of the implications of it. In addition, those that are willing to use larger jurisdictions as proxies for risk will draw bad conclusions that reward higher risk issuers that don’t disclose effectively, and even risk penalizing issuers which outperform by making geospatially naive assumptions.