The controversy surrounding Am Trust Financial Services (NASDAQ:AFSI) began on Dec 12 2013 when noted short-seller GeoInvesting questioned the insurer's accounting practices. The issue was greeted by an astute response from John Hempton, a well-known investor in banks and insurance companies, who dismissed the premise of GeoInvesting short attack as "amateurish". Counter-responses from GeoInvesting and management at AFSI further complicate the controversy. Since Dec 12, the stock of AFSI has fallen about 26%. Maiden Holdings (NASDAQ:MHLD) is a victim of collateral damage because of its reinsurance agreements with Am Trust. MHLD 3Q 10Q states that "Maiden Bermuda will continue to reinsure losses at its proportional 40% share per the Reinsurance Agreement". The chairman of the Board at MHLD, Barry Zyskind, is also the President, CEO and director at Am Trust.
Because the involved parties failed in varying extents to accurately address the points they seek to refute, the controversy is shrouded in confusion. The harsh, emotionally-charged tones also provided little help in clarifying the confusion. Fortunately, the resolution does not require one to be a professional actuary. This article seeks to clarify the arguments made by the involved parties and explain why AFSI, in all likelihood, is not perpetuating fraud.
Issue 1: AFSI hid losses in a Luxembourg company. Therefore its profits should be lower than what AFSI claims on its financial statements.
GeoInvesting - The sum of the parts, which make up net losses or loss adjusted expenses (net L/LAE), did not match the sum of the net L/LAE reported in the 10K. The difference was interpreted as hidden losses. The sum of the parts was derived from numbers in reports by AM Best and Technology Insurance Company. Here's where the interesting part comes in: the difference between the sum of the parts and the sum in the 10K was roughly equal to ceded losses to ACHL (a Luxembourg subsidiary of AFSI) reported by Technology Insurance Company. AFSI failed to report these ceded losses so as to increase its profits.
Mr Hempton - Ceded losses are not real losses. A ceded loss is part of a reinsurance agreement, in which a reinsurer shares premiums and losses with the insurer. Such losses are known as "ceded losses" to the reinsurer. In this context, the reinsurer is ACHL (the Luxembourg subsidiary of AFSI). Since ceded losses are not actual money-losing transactions, ceded losses should not reduce profits.
GeoInvesting's counter to Mr Hempton - The key in reinsurance is to share BOTH premiums and losses. But AFSI ceded ONLY losses, not premiums. Look at AII Bermuda 2010 financial statements (AII Bermuda is a subsidiary of AFSI) that showed AII Bermuda ceded losses to ACHL. Since losses were ceded without premiums, ACHL should report a loss under net L/LAE. This loss should be consolidated into AFSI, but it was accounted for under the equity method (or "partial consolidation") in 2010.
AFSI - There has never been an off balance sheet company which is used to hide losses. The reported loss ratios is the fully consolidated actual loss ratio. Moreover, operations at Luxembourg made up only 2.5% of net profits between 2009 and 2013.
(If you are still following the argument, give yourself a pat on the back)
My opinion - Consider the premise of GeoInvesting's counter to Mr Hempton. Notice that the argument subtly changed from hiding 100% of losses in the original GeoInvesting argument to hiding less than 100% of losses in the counter-argument. In other words, AFSI did recognize the losses, but the recognition was just not done properly. Also notice that GeoInvesting could only prove its counter-argument in 2010. This is because AII Bermuda has public statements only in 2010. Even if the losses were not properly recognized in 2010, one cannot conclude that the same situation happens today. Lastly, what if there are other ways to prove that losses are ceded to ACHL without using AII Bermuda? There is one other way - ceded losses to ACHL were reported by Technology Insurance Company (in the original GeoInvesting argument). But the ceded losses do not exactly match the difference in the numbers between the 10K and the sum of the parts, implying that an extent of accounting consolidation might have occurred. Can one confirm that consolidation did occur? The best answer is one does not know, but one cannot say that the ceded losses to ACHL were definitely unaccounted for.
Issue 2: AFSI bought Luxembourg captives to maintain profit growth and conceal losses.
GeoInvesting - AFSI spent $730 million to buy captives (they are really just reinsurance companies, but I use this term to be consistent with GeoInvesting's article) with only $688 million of equalization reserves. The statement appeared to imply that the transactions were suspect because AFSI did not seem to profit by paying above-reserve values. The shady nature of the transactions led GeoInvesting to conclude that "AFSI is not recognizing these intra-company losses correctly - most likely by mixing and matching Luxembourg GAAP and US GAAP. In the following section we will show that magnitude of losses (that ACHL is absorbing and clearly not disclosing) by ….. "
Mr Hempton - no opinions on this issue
GeoInvesting's counter to Mr Hempton - Geoinvesting appeared to allow this issue to slip (because Mr Hempton did not mention it?).
AFSI - The quotes are from the business update conference call hosted by AFSI on Dec 16 2013. "Captive reinsurance companies are common among large multinationals" that elect to reinsure their own insurance business. This is because "the insurance terms available in the open market are unattractive or because these companies believe they will capture additional profit by self-insuring". The captive reinsurance companies set up 2 types of reserves: "standard reserves for known liabilities", and "an equalization reserve to protect against insurance volatility". At some point, "the company accumulates more reserves that it needs to protect against [insurance] volatility, either because better pricing is available in the market or the losses have fun favorably for several years".
My opinion - Consider GeoInvesting's argument. How does the argument make the jump from purchase of captives to improper recognition of intra-company losses? The conclusion is definitely not properly supported by its premise. I also believe that AFSI gave a reasonable explanation for its purchase of captives. Favorable loss rates imply superior management of the reinsurance business, which justifies a higher price tag above book value (not reserve value). How the purchase of superior captives relate to concealing of losses is a point that was not made clear by GeoInvesting.
Issue 3: AFSI uses improper assumptions in its accounting of life settlement contracts. If correct assumptions are used, AFSI will face losses in the write-downs of the contracts.
I'll break this argument down into a few parts. Before we delve into this subject, it is essential to understand what life insurance is, what life settlement contracts are and what goes into valuing these contracts.
A (very) short primer on life insurance and life settlement contracts.
Life insurance in general refers to the practice of paying premiums regularly (or a lump sum) in exchange for a sum of money upon the death (or terminal illness) of the insured person. When does the purchase of life insurance make sense? Similar to all businesses, it makes sense only when you receive more than you pay out. Hence the sum of money received upon death or illness (for simplicity, I'll call this death benefit from here on) must be greater than the sum of premiums one pays.
Life settlement contracts are arrangements in which life insurances are sold to a third-party such as an investor. So when do investors buy life settlement contracts? Only when investors think that the life settlement contract will provide a death benefit from greater than the cost of the life settlement contract. When an investor buys a life settlement contract, the investor assumes the premium payments. You may think about this as the "reverse" of a bond (pay rather than receive coupons) which does not have a specified maturity date. To value a life settlement contract, investors need to understand two things:
One, how long will the insured person live. Call this life expectancy. The life expectancy determines the maturity date. The shorter the expected life expectancy, the sooner the investor receives the death benefit, the higher the value of the contract. Second, the time value of money measured by the discount rate. The lower the discount rate, the higher the value of money today.
Issue 3.1: Assumptions for life expectancy were too low, thereby over-valuing the life settlement contracts.
GeoInvesting - AFSI relied on internal estimates for life expectancies, which were revised downward, in contrast to the standard industry practice of using third-party estimates for life expectancies, where were revised upward.
Mr Hempton - Look at the profile of the insured people under the life settlement contracts. AFSI revised the assumption for life expectancy down from 139 months to 133 months (11.1 years). This was reasonable given that the average age of the insured was close to 80 years old, implying that the average insured would live to 91.1 years old (80+11.1), which seemed like a reasonable estimate. In addition, AFSI only has 277 policies, a sufficiently small scope for which AFSI can get accurate data.
GeoInvesting's counter to Mr Hempton - According to 21st Service in January 2013, which is claimed by GeoInvesting to be the most commonly used life expectancy provider, life expectancies should increase by 19%. GeoInvesting went to claim that AFSI should increase its assumptions for life expectancies by more than 19% because AFSI held premium-financed life settlement contracts that are often held by the wealthy (and longer-living) crowd.
My opinion - The industry has questioned the accuracy of life expectancy assumptions made by 21st service. Rigi Capital Partners is an outspoken critic. The company was founded by Dr Chandra Poojari, a former quant at Bluecrest Capital (a very successful hedge fund), and managed $8.9 billion in investments in life insurance policies. Rigi believed that "there is too little transparency regarding data or methodology in the announcement from 21st services", and that "21st services has used a lot of experience data to redefine its underwriting [while] institutional investors have some time ago already started using larger and more granular databases for their work". Moreover, 21st services made changes within a week of the release of its new life expectancy assumptions without releasing sufficient details to the change. The comments do not appear to support 21st services as a well-respected provider of life expectancy assumptions. Another manager of life policy-based investment funds, SL Investment Management ($840m AUM), remarked that "the shift in [21st services Life Expectancy] was well overdue", another indication that 21st services may not be the leader GeoInvesting claimed it to be.
More interestingly, SL also pointed out that "we have always taken the approach that replying on a single LE is inappropriate and as such we insist on multiple life expectancies for each policy prior to acquisition". This means that even if 21st services were to increase assumptions for life expectancies, AFSI may not need to revise its assumptions for all policies but do so on a case-by-case basis.
Lastly, even if AFSI were to get it all wrong (very unlikely), the life settlement contracts account for only 1.5% of total assets.
Issue 3.2: The discount rate that AFSI used was too low (7.5%) compared to the peer average of 20%. A low discount rate overvalued the life settlement contracts.
GeoInvesting - If the peer average was used, the life settlement contracts would be overvalued by $90-135m and wiped out roughly 13-19% of AFSI's tangible equity.
Mr Hempton - 7% (it's really 7.5%, Mr Hempton might have overlooked this detail) seemed reasonable when you consider the specifics of AFSI's life settlement contracts. AFSI's bet was "on 80 years olds dying pretty soon". With the insured person having a high risk of death, a low discount rates seemed appropriate.
GeoInvesting's counter to Mr Hempton - 7.5% was not reasonable because it was not a market-based measurement. Accounting rules demand a market-based discount rate that was closer to 20%. Moreover, EEA Life Settlement Fund was criticized by Ernst and Young for using a discount rate of 10%, which was "significantly lower than would be used by market participants in an arm's-length transaction".
My opinion - I think confusion exists in differentiating between discount rate and effective discount rate, which I think it's the "real" market-based discount rate used. AFSI attempted to clarify the difference in the latest conference call. The discount rate is used to discount future cash flows excluding reserves while the effective discount rate takes reserves into account. The inclusion of reserves seem counter-intuitive because reserves are not cash flows, but they should be included because they adjusts cash flows for risk.
For example, say cash flows in the next year is $100 and the discount rate is 10%. What if you are not confident that cash flows in the next year will be $100? You will establish a reserve, say $10, that reduces your cash flows to $90. The present value of $90 is $82 (=90/1.10), which gives an effective discount rate of 22% (=100/82 -1) when calculated with the original $100 cash flow. The effective discount rate (22%) is higher than the discount rate (10%) because the effective discount rate is derived from "risk-adjusted" cash flows, which are just cash flows with reserves subtracted. This makes intuitive sense because the higher the expected risk, the higher the reserves, and the lower the future cash flows.
The above example almost describes what AFSI does. The future cash flows of the life settlement contracts were estimated at $779 million, which is reduced by $281 million of reserves to yield $498 million. Applying a discount rate of 7.5% to $498 million yielded $227 million, which gave an effective discount rate of 14.2% when calculated with the original $779 million cash flows.
The effective (OTC:REAL) discount rate of 14.2% at AFSI compares very favorably to the EEA life settlement fund of 10%. AFSI has 272 policies relative to EEA's 660. The smaller scope should represent lower risk because AFSI is more likely than EEA to have sufficient data to evaluate all policies. Where EEA should have used a higher discount rate to reflect its higher risk, AFSI uses a higher discount rate even with lower risk. This is a clear indication of the conservative accounting practice at AFSI.
But a discount rate of 14.2% is still low compared to peers. This may reflect AFSI's specific focus on a group of older individuals that are likely to have lower life expectancies than the norm. The evaluation of the large group of companies that hold life settlement contracts is beyond the scope of this article, but my argument is already sufficient to prove that the discount rate does not imply fraudulent practice.
Lastly (again), even if AFSI were to get it all wrong (very unlikely), the life settlement contracts account for only 1.5% of total assets according to the figures stated in AFSI's latest conference call.
Issue 3.3: More than half of AFSI's life settlement contracts have Phoenix Life Insurance (PLI), which is facing bankruptcy, as the issuing carrier/counterparty.
GeoInvesting - Tiger Capital LLC, an AFSI subsidiary, said that it held 136 PLI life settlement contracts. However, AFSI disclosed that only 81 PLI life settlement contracts it held had no value. This implies that a minimum of 55 PLI contracts (=136-81) had no value. Marking these contracts to true market value will result in a significant impairment.
Mr Hempton - I agree that PLI is involved in the money-losing business of selling underpriced insurance policies, but PLI looks like it will survive. Regulators allowed PLI to pay almost $30 million in dividends to the parent company, which is publicly-listed Phoenix Cos (NYSE:PNX). This could not have happened if regulators had discomfort about valid policyholders being paid.
GeoInvesting's counter to Mr Hempton - The dividends were not paid by PLI, but by PLIC (Phoenix Life Insurance Company). Dividends paid by a sister company did not imply that PLI is financially-sound. To be precise, Tiger Capital held contracts issued by PHL variable, not PLI (note the change here). PHL variable is thinly-capitalized relative to the risks of its life insurance policies, and is likely to go bankrupt. AFSI will have to write down the policies issued by PHL variable sooner or later.
My opinion - Note that in GeoInvesting's original argument, there was no mention of the magnitude of the expected loss should contracts be written down. In the latest conference call, AFSI said that its net carrying exposure to Phoenix was approximately $20 million. Since AFSI has about $1.4 billion of equity with only $560 million of debt, the company, in all likelihood, should be able to weather a write-down of $20 million with ease.
Issue 4: A few executives in AFSI management team have sketchy histories.
GeoInvesting - Two executives, CFO Ronald Pipoly and COO Michael Saxon, were executives at Credit General, whose CEO Robert Lucia was charged with stealing $30 million from the company. Both were named as defendants in lawsuits brought by the Ohio Director of Insurance. Even Am Trust, which entered into a management agreement with Credit General in 2000, was also sued by the Ohio Director of Insurance.
Mr Hempton - No response to this issue.
My opinion - The Ohio Director of Insurance sued a total of nine officers and directors of Credit General for breaching their fiduciary duties, ultimately leading to the downfall of Credit General in 2001. After 13 years, none of the officers and directors (apart from the CEO) have been charged with any wrongdoing. To claim that the downfall of Credit General is a result of CFO Ronald Pipoly and COO Michael Saxon having shady characters is misleading.
Issue 5: The stock of AFSI is massively overvalued relative to its peers
GeoInvesting - AFSI traded at a price-to-book of 4.7x relative to its peer average of 1.2x. Even if the accounting issues were not present, the stock price should fall and trade closer to its peer average.
Mr Hempton - No response to this issue.
My opinion - This is perhaps the most absurd claim in the entire argument. A higher-than-average multiple can never be the only reason for the stock to fall. To be precise, a stock falls when the multiple is too high relative to its prospects for growth. Without discussing growth prospects in the context of the stock's multiple, one will never know whether the multiple is too high or too low. GeoInvesting never went into a discussion about the core businesses or their growth prospects at AFSI, which is in the business of insuring specialty risk and extended warranties, and small commercial business and reinsuring personal lines.
In addition to shedding light on the convoluted arguments, this article also hopes to remind readers that the founding shareholders of AFSI own approximately 58% of the outstanding shares of the company. The vote of confidence given by the two largest shareholders, Leah Karkunkel and George Karfunkel, who owned a combined 26% of AFSI is reflected by the purchase of approximately 1 million shares, equivalent to an increase of about 5% of their holdings. The explicit display of support should convince the shareholders of AFSI to align their confidence in the company with that displayed by the founding shareholders.