How COVID-19 Could Impact Post-Hurricane Loss Inflation in Florida: AIR Worldwide
By the end of March 2020 people were wondering what the impact of COVID-19 on the upcoming hurricane season would be. There is no historical event to provide evidence for how the scenario might unfold, so we need to start with the basics. Loss inflation includes economic demand surge, increased loss adjustment expenses, increased losses from delays in claims settlement for building losses, and business interruption.
Loss inflation for these scenarios is a disturbance to an economic market equilibrium. There is variable supply and demand for loss adjusters, materials, and construction labor. How COVID-19 affects these individual economic markets will determine the extent of loss inflation. To understand the possible outcomes, we can use Hurricane Irma from the 2017 hurricane season along with present-day economic data to get some insight.
Figure 1 shows some construction metrics that can tell us about what was going on in Florida just before Hurricane Irma made landfall in September 2017. The permits per worker provide a measure of how much construction labor is in demand. Because this ratio is forward-looking, we need to study the prior year to get a good measure of what to expect the following year.
For 2016 the value of permits per worker was 2.23, but to give you some perspective the value in 2003, the year before hurricanes Charley, Frances, and Jean (CFJ), it was slightly above 4. Coincidentally, the number of employed construction workers was about the same in both seasons (94.8K vs. 95.5K employees). This tells us that the demand for work was twice as high during CFJ.
If we look at the last row in Figure 1, we see that the permits per worker increased almost 20% in three years. While the workload increased also, it is still 40% lower than the 2003 metric. The highest concentrations of construction employees in 2016 and 2019 were in Miami-Dade, Orange, and Broward counties along with the highest growth rates in the sector, suggesting that this is where most of the construction is happening.
Figure 1. Florida economic statistics:
(Source: AIR, data: Xactware)
At this time there has been no suggestion of a shelter in place order in high hurricane activity states such as Florida that includes construction workers as non-essential. The executive order for shelter in place in Florida specifically exempted construction for the state, and the local guidance in Miami-Dade is similar.
How can COVID-19 impact the supply of labor? The most obvious way is when workers get sick and the infection spreads throughout the construction site. From AIR Pandemic Model assumptions for COVID-19 we know that about 10% will get severely sick without any mitigation. This is a population assumption, but it is the best data we have now and we can apply it to the supply of construction workers.
The result—assuming nothing else changes—is that the permits per worker will go up about 10% in this case as well. But we also know from new reports that personal protective equipment (PPE) is required on job sites, and this order is being enforced. This may not have the intended impact, however, if PPE and social distancing guidelines are not followed off the job site. The effect of this requirement will be discussed a little later.
Demand for construction labor is also likely to be affected but the mechanism isn’t quite so straightforward. Construction projects are chosen because the profit margins make business sense; the main channel where we can see an impact is a reduction in the number of permits being issued. Over the past few months we have seen the economy brought to a standstill and heard anecdotally of a new urban migration being driven by those who are still employed. The question we are asking is how will this affect demand for future housing permits?
The permit data for all types of construction work in Florida is available from Verisk Buildfax. While we can’t draw conclusions from one data point, the data from the second quarter of 2020 show a significant drop compared to the same period in 2019 for some of the more populous counties in southeastern Florida. This is something we are going to be keeping an eye on to see if it continues to trend downward. Another possible outcome is that construction on currently permitted projects ceases because they are no longer economically viable. There are also early indications that only existing construction sites are active and start dates for current permitted work have been postponed.
The other important piece we need to focus on is construction labor and materials cost inflation. Cost inflation data are shown in Figure 2 for the years prior to the 2004, 2017, and 2020 hurricane seasons. The first row shows the change in costs for Q4 2003 over the prior year and all rows compare the national and Florida trends for labor and materials costs. For labor, the cost inflation for Florida is the same or less than the national trend for all three seasons. The materials cost inflation is similar but, in this case, it was negative prior to the 2004 hurricane season.
Figure 2. Pre-hurricane season cost inflation:
(Source: AIR, data: Xactware)
Figure 3 shows the same trend after the hurricane events for those same seasons. For example, in 2004 the change in labor costs in the year after Q4 was 19% in Florida compared to 7% nationally. This is a striking contrast to what happened after the 2017 season. While there was some local labor cost increase within Florida from economic demand surge in 2017, it was nowhere near as extreme as in 2004. What is striking is that the materials cost increase in 2004 following the hurricane season is muted compared to labor. In 2017 it falls, which could mean there is better supply chain management compared to the earlier season or that the mix of materials used for repairs after the season looks very different from the materials driving the higher national inflation.
Figure 3. Post-event cost inflation
(Source: AIR, data: Xactware)
Construction cost inflation in the U.S. has been on the high side the last few years, so it is more important than ever to have accurate replacement values to understand losses. For those who keep replacement values up to date, I do not expect anything out of the ordinary from the above causes and I do not at this time think there will be a separate COVID-19 effect.
As with post-catastrophe repair work, there is market equilibrium for loss adjusters. Again, let’s use Hurricane Irma as an example of what could happen in Florida. In Figure 4, the loss adjusters from the Quarterly Census of Employment and Wages (QCEW) are used to construct metrics to understand the supply and demand of workers. Because there is a six-month delay getting data, we can only see data for part of the year in 2019.
Using Florida claims data from Property Claims Services we can calculate the claims per worker to estimate the demand. Between Q2 and Q4 there was an increase in the number of claims adjusters, presumably to meet the workload. Assuming Irma happened in 2019, or two years later than it did, we can see the number of adjusters prior to the season had increased a couple of percent over 2017. This is not significant, but it is enough to lower the number of claims per worker if Hurricane Irma had happened that year.
Figure 4 Florida claims adjuster statistics:
(Source: QCEW)
Again, we ask the question how could COVID-19 affect this market? Loss adjusters are part of the list of exceptions to the shelter in place order in Florida because they are necessary to maintain the safety and sanitation of homes and other buildings. As we did with construction workers, we must consider that all the adjusters are exposed to the virus and the population assumptions we used above apply here as well. There could be an impact on the workload, but it is not likely to be big.
The use of technology to adjust claims has changed since 2017 and could have a significant impact going forward. Services such as ClaimXperience from Xactware are being used to speed up the process and reduce claims-handling costs and will have a big impact on the residential market, which accounted for the majority of claims after Hurricane Irma. The effect will be to increase the supply of experienced adjusters to handle more complicated commercial claims.
We mentioned earlier the importance of PPE and the requirement for it to be worn by workers on all job sites for any type of repair where filtration media is currently needed. This fixed cost is going to be new for all claims for the foreseeable future. It is unexpected in the sense that it was probably not considered in the cost of repairs when policies were priced, but we know that it is required now and can plan for it.
Xactware has been tracking the cost of PPE and, as might be expected, the cost has gone up significantly since the beginning of the COVID-19 pandemic. The impact of this increase is going to change depending on the size of the claim and whether it requires on-site workers to make repairs. Content losses and BI/ALE-related claims will not likely be affected by this. The cost of PPE should be fixed and should vary only by the number of workers on each job and the number of days required to make repairs. The cost is unlikely to move the total cost significantly if most workers are wearing bandanas or simple paper masks.
Without prior experience with a worldwide pandemic during hurricane season, making assumptions about what is going to happen with regard to repair costs takes us back to before there were catastrophe models. It is certainly appealing to use our “gut” instincts, but without data to back them up, we would be guessing, just as many had to do before Hurricane Andrew in 1992.
The economic fundamentals for the construction market in Florida suggest nothing extraordinary. There is no housing bubble as there was in 2004, the demand for construction is currently moderate (and may be dropping off because of COVID-19), and cost inflation is no different than it was three years ago. For economic demand surge, post-event cost inflation, and loss adjustment costs you should expect what you would in any other season. The impact on BI/ALE may be measurable but there are supply chain models that can provide more insight into the complicated relationships that will be affected.
This post originally appeared on AIR Worldwide’s blog and was reprinted with permission.