Description

American Coastal Insurance Corp (ACIC)

An Incredibly High-Quality Insurer Trading at forward ~5.5x P/E

In August of 2023, frostybluebird pitched American Coastal on VIC. Soon after publishing his pitch, we reached out to him, looking to get insights on certain parts of the company that we did not fully understand. The insights we gained from frostybluebird sparked a multi-month long journey researching the company’s ins and outs. We suggest that everyone read the original pitch that frostybluebird created as frostybluebird’s research formed the entire foundation of our analysis below. frostybluebird deserves full credit for finding and researching, in our opinion, one of the best pitches in VIC in recent years. We hope our research below can be useful to those looking for additional information/insights on American Coastal.

Executive Summary: To provide a quick summary of American Coastal (AmCo), it is a small-cap property and casualty insurer based in St. Petersburg, Florida. It is almost exclusively focused on garden-style condominiums with  above-average risk characteristics, selling windstorm insurance policies to the HOAs that represent them. Over the last decade, AmCo has been able to average combined ratios below 70% with no years of unprofitability. This has been possible due to AmCo’s underwriting superiority, which is derived from its exclusive relationship with AmRisc, strong underwriting discipline, and focus on niche markets. Until a few quarters ago, AmCo had been consolidated with UPC, a lower-quality personal lines insurer, whose losses in the past have been subsidized by AmCo’s underwriting profits (loss sharing agreement). However, UPC recently entered runoff, leaving AmCo as a highly profitable standalone company. Again, for those looking for a more comprehensive overview of the pitch, please refer to frostybluebird’s VIC writeup.

We believe that ACIC trades at a forward PE of ~5.5x, with a payback period of just 4.5 years and strong growth thereafter. Not accounting for the potential of a new MGA that AmCo plans to create to compete in the personal lines brokerage market, conservative top line growth, and a modest 10x PE exit multiple, we get an intrinsic value per share of ~$27. This leads us to the belief that AmCo is the most undervalued insurer in Florida.

For this write-up, our goal is to present our research on a few key topics  that we believe to be especially important and insightful. These topics are: hurricane risk, reinsurance, the hard market environment, management, equity offering, and NY personal lines exposure. The mediums through which we conducted our research were interviews with multiple P&C insurance experts, review of the most recent quarterly earnings and filings, asking management questions during the Q3 earnings call, discussions with HOAs, extensive simulations, academic papers, calls with fund managers who own ACIC, and interviews with management (refer to Daikoku Capital’s substack for a summary of our conversation with Brad Martz, the CFO of AmCo).

Reinsurance

We had the opportunity to speak to Professor Woollams of Columbia University about reinsurance. He is an expert in the management of commercial property and casualty insurance claims, spending nearly two decades at AIG as President of Global Commercial Claims. In addition, we spoke to Shiwen Jiang, an insurance actuary at Berkshire Hathaway Specialty Insurance who has over three decades of experience in the actuarial sciences for P&C insurance. Shiwen was also able to provide general insight on Berkshire’s P&C reinsurance strategy.

High Reinsurance Capacity with Low Attachment Points

A major determining factor in the risk of an insurance business is the quality of their reinsurance stack. We were of the belief that the two most important numbers to pay attention to when analyzing the reinsurance stack (Figure 2) was the size of the stack at the upper end (approx. $1 billion for ACIC) and the risk of the stack being depleted (1 in 167 year hurricane event for AmCo). However, after further due diligence, we have identified new metrics that can be used to evaluate the quality of a reinsurance stack.

AmCo has an attachment/retention point of $10 million per event. This means that for each event, AmCo pays the first $10 million in claims and then the reinsurance kicks in. In most years, we can expect AmCo to spend their full $10 million retention. For example, this year management has stated that even with the impacts of Idalia, which crossed over Florida in August 2023, current loss estimates are well below the annual reinsurance retention limit.

An actuarial expert suggested that we compare each insurance company’s attachment/retention limit to their policyholder/equity surplus. Currently, AmCo, even with a low equity base because of the loss-sharing agreement with UPC, has a 12x equity to retention limit, which is far above average in Florida but still below the very best catastrophe insurance companies in the US. Figure 1 below shows the equity to retention limits for various competitors in the Florida P&C market. We included Palomar as it is a high-quality comp on the West Coast that insures against earthquakes and hurricanes.

Figure 1: Equity to Retention Table

(source: Arcata Capital, Daikoku Capital, and Company Filings)

Figure 2: American Coastal Reinsurance Stack

(source: ACIC IR Presentation)

Additionally, Shiwen Jiang explained to us the 3-module actuarial simulation used for insurance and reinsurance to calculate the 1 in 167-year probability in Figure 2. The first module is the event/hurricane simulation, which allows actuaries to model the formation, movement, and dynamics of hurricanes. The simulation varies the diameter, wind speed, pressure, and location of hurricanes over many years (>100,000 in some cases). The data used in these simulations is almost entirely historical with minimal future considerations. Since reinsurance agreements are only in place for 1-year at a time, past data has a very strong correlation with hurricane risk for the next year. Therefore, the 1 in 167 year number only accurately indicates the risk of the reinsurance stack being fully utilized for the next few years. This number should not be used to extrapolate any further than a few years as the effects of climate change along with other unforeseeable factors will introduce variability into the model. Shiwen Jiang instead suggested that we compare this 1 in 167 number across competitors to evaluate how AmCo’s stack stacks up. For the competitors we looked at, the least comprehensive stack implied a 1 in 100 probability while the most comprehensive stack (Palomar) implied a 1 in 250-year probability. AmCo sits comfortably in the middle. In addition, AmCo’s superior performance during large hurricane events and disciplined insurance strategy (remember that AmCo has never turned a loss in their 15-year operating history due to their focus on garden style condominiums) leads us to believe that management is potentially being conservative on AmCo’s exposure to catastrophic hurricane risk.

Reinsurers consider a variety of factors when selling reinsurance to specific P&C insurance companies. These considerations, simulated using the 3-module model, include the construction type of the insured properties, location, etc. In addition, reinsurers have close relationships with insurers. While these relationships tend to be stronger the larger the client as a percentage of total insurance book, reinsurers will often have the same actuaries underwriting the same insurance companies year after year. For example, one actuary may be tasked with just managing the insurance book for AmCo and a handful of other insurers. The fact that reinsurers likely have an established relationship with AmCo means that AmCo will get reinsurance rates that consider the company’s low total value insured at risk.

Coming into reinsurance negotiations, management often has a plan as to what type of reinsurance stack they want. For example, they will likely have a target max size, retention/attachment point, percent quota share versus excess loss, and approximate price. If AmCo is not able to negotiate the exact terms that they want the first time around, they can always renegotiate at a higher price. If after final negotiations, AmCo does not have a reinsurance stack that sufficiently covers the risk they are exposed to, management always has the option to underwrite lower levels for that year. Because of the long-track record of underwriting discipline that Dan Peed has shown, his firm understanding of risk created through founding and running AmRisc, and low combined ratios, we believe that Dan Peed would decrease gross premiums written if a poor reinsurance stack was negotiated. The quality of the insurance stack has been integrated into our financial model in both direct and indirect ways. Directly we have assumed 1 named event per year causing 12.5 million in retention to be used up. Additionally, we have modeled private XOL reinsurance expenses as a percent of GPW increasing from 19% in 2024 to 23% in the out year. Using Crystal Ball, we have simulated both retention points and private XOL reinsurance costs across 1,000+ scenarios with even 3 standard deviation results leading to extremely strong upside (indicates lack of model sensitivity to these rates). 

Dynamic Rate Adjustability in Response to Reinsurance Market Trends

One concern that we had with AmCo was how non-Florida catastrophic events would affect reinsurance capacity in Florida (e.g., hurricane in Japan). Leading on from this question, we were concerned that this could lead to a set of circumstances where overall reinsurance costs rise but rates do not rise in Florida.

The first answer is that a significant portion of American Coastal’s reinsurance stack is obtained from the Florida Optional Reinsurance Assistance Program (FORA) and the Florida Hurricane Catastrophe Fund (FHCF). They have inured reinsurance from FORA and FHCF. The FHCF reinsurance is particularly significant given that it covers them for up to $665.2 million in losses in excess of $305.1 million in losses ($665.2m xs $305.1m) visualized in Figure 3. Both these reinsurance programs are inherently Florida-focused and thus are not impacted by CAT events in the rest of the world. 

Figure 3: FORA/FHCF Reinsurance Stack 

(source: ACIC IR Presentation)

One of the concerns Professor Woollams brought up was whether programs like FORA/FHCF are sustainable both in terms of durability and ability to pay after a black swan event. To address that concern, we spent more time looking at how FHCF is funded and maintained.

The FHCF is a tax-exempt trust fund created by the State of Florida in 1993 that is designed to be self-supporting and self-funded. The FHCF is administered by the State Board of Administration of Florida under Section 215.555 of the Florida Statutes. All participating insurers, like AmCo, pay the FHCF annual reimbursement premiums as consideration for the reimbursement coverage that FHCF provides. The reimbursement premiums are based on insured values of covered properties (as reported annually to the FHCF).

The annual reimbursement contract provides for reimbursement of a percentage of an insurer’s residential hurricane losses in excess of its retention which is determined under a statutory formula. Reimbursement is provided at one of three percentage levels (90%, 75%, or 45%) which is selected in advance by the insurer seeking coverage. This means that once an insurer’s losses from residential hurricane damage exceed their retention level, the FHCF will cover either 90%, 75%, or 45% of the additional losses, depending on what level the insurer has chosen. 

The FHCF obtains its funding from the following available potential sources:

1. Accumulated and current year reimbursement premiums

2. Recoveries from reinsurance and other risk-transfer mechanisms 

3. Pre-event bond proceeds and other pre-event liquidity resources

4. Proceeds of post-event revenue bonds or bank loans issued

5. Investment earnings or accumulated reimbursement premiums 

It is important to note that the actual and potential obligations of the FHCF are limited by statute. For the contract year June 1, 2023 – May 31, 2024, the maximum potential liability of the FHCF is $17 billion, with projected available total liquid resources of approximately $7.7 billion. These liquid resources consist of $4.2 billion of project year-end fund balance and $3.5 billion of pre-event bond proceeds.

Figure 4: FHCF Statutory Limits & Estimated Claims-Paying

(source: FHCF Filings) 

Verisk and RMS are companies that provide catastrophe modeling services. Their models are used to estimate potential losses from hurricanes, while using company-by-company data which includes analysis based on model results by ZIP code and type of business and each individual company retention, company limit, and coverage selection. The data shown in figure 5 below is for the approximately 150 participating insurers where each insurer has its own retention and coverage limits, and therefore each participating insurer has its own unique probabilities of triggering its FHCF coverage and reaching its FHCF coverage limit.

Figure 5: FHCF Projected Return Times & Ground Up Losses

 (source: Arcata Capital, Daikoku Capital, FHCF Filings) 

In the context of Figure 5, the “Return Times” column is the estimated number of years between hurricane events that cause losses equal to the corresponding number in the column named “FHCF Layer Loss”. The “Ground Up Losses for Average Verisk” is the predetermined amount of loss that the average insurance company is responsible for covering before the FHCF’s coverage begins.