Ira Sohn 2018 Investment contest finalist - Short Genworth MI Canada

I had the honour of being among four finalist for the Ira Sohn investment idea contest, as part of the Ira Sohn NYC conference.

While I was certainly disappointed not to win, it was a great experience and I met some investors who I have always wanted to meet, namely David Einhorn, Larry Robbins, and Nathaniel August.

My idea was short Genworth MI Canada (ticker MIC), and I think it is a compelling short.  You can see my presentation slides here, and my full report below.  Don't hesitate to reach out with any questions or thoughts you may have.   

Full disclosure: I am short Genworth MI Canada

Summary

Genworth MI Canada (TSE: MIC) is a Canadian mortgage insurer.  I am short for the following reasons:

  1. I believe the Canadian housing bubble has burst.  Housing prices are down 5% nationally from the peak last spring, and 14% in the Greater Toronto Area. This trend should continue as sales volumes are declining and inventories are rising.  Furthermore, as of Jan 1st, 2018, a new regulation called B-20 has gone into effect reducing housing affordability by 20%.  
  2. Their balance sheet is risky.  The company has $218bil of outstanding mortgages insured ($90bil for which effective LTV is greater than 75%).  The company has loss reserves of only $119mil, and $3.9bil of equity.
  3. The company aggressively books revenues.  Most premiums are recorded as revenue in the first 5 years.  However, the insurance lasts until the mortgage is repaid. As house prices stop rising, more losses will occur and revenue recognition should extend.  Lower premium volumes and slower revenue recognition will mean much weaker earnings in the future.

If the bubble truly has burst and the housing market crashes, balance sheet risks will likely wipe out their equity.  If we have the mythical ‘soft-landing’, then future earnings will be materially weaker.

The stock trades at 0.96x book, and ROE was 13% in 2017 (or more like 10% if you normalize the loss ratio).  The short is asymmetric; there is little downside to being short (20% to 30% if the stock trades at 1.2x to 1.3x book), with a potential upside of 100%.

In this article, I will first give a brief background on Canadian mortgages and mortgage insurance.  Then I will describe the current state of the Canadian housing market to demonstrate its precarious nature.  Finally, I will examine MIC’s risky balance sheet, look at the weakening income statement, and point out other areas of concern.

Background on Canadian Mortgages and Mortgage Insurance

Mortgages in Canada have a fixed interest rate period, typically 5 years, and the mortgage amount is amortized over a longer period, 25 to 30 years.  Once the fixed period is over, the interest rate is reset and the buyer has to renew their mortgage - it does not simply rollover.

Mortgage insurance is required for any borrower that puts down less than 20%. Roughly half of all Canadian mortgages have mortgage insurance.  This insurance is provided by 3 companies in Canada. Genworth Canada is the only publicly traded insurer and has a 27% market share. The other notable provider is a Canadian crown corporation, CMHC, which has a 65% market share.  In the case of duality of default (i.e., the borrower defaults and so does MIC), the bank bears the next 10% of losses and the government steps in to absorb the rest. However, if the mortgage is insured by CMHC, the government covers all losses even if CMHC defaults.  As a result, banks need to put up more capital for mortgages that aren’t insured by CMHC. CMHC-insured mortgages have a 0% risk-weighting, whereas for MIC risk-weighting is ~3.5%. Given banks’ preference for CMHC insurance, MIC is likely insuring a weaker pool of borrowers.

Canadian mortgage insurance is regulated, thus insurers don’t set the price.  Insurers only charge 4% for down payments of 5 to 9.99%, 3.1% for down payments of 10 to 14.99%, and 2.80% for down payments of 15 to 19.99%.  These prices increased in March 2017; they were more than 50bps lower previously. Keep in mind, mortgage insurers in Canada are responsible for 100% of the losses when a borrower defaults.

On the regulation side, there have been major mortgage regulation changes, namely the B-20.  Borrowers have to prove they can afford a mortgage at interest rates 2% higher than current mortgage rates.  This rule went into effect on Jan 1st, 2018. Simple math suggests that a buyer now can afford ~20% less house than they could prior to the rule.  Given the unaffordability of houses at current prices, this should reduce prices by ~20%.

To summarize, MIC has no pricing power and can’t set higher insurance premiums for higher risk mortgages, they face adverse selection as banks would much rather have the CMHC insure their loans, and regulatory changes point to house price drops that will cause losses when defaults occur.

Canadian Housing Market

Since 2012, house prices in Canada have appreciated at an annual rate of 6.6%, but household income has only grown at 1.6% and population has grown at just 1% annually.  In Canada, the median home is worth $500k, yet the median family income is only $80.9k. This house price-to-income ratio of 6.25x is well in excess of the peak price-to-income ratio of 3.8x in the USA before the crash.  

Debt in Canada has exploded.  For example, household debt-to-disposable income in Canada is 171% (this measure peaked at 128% in the USA), and household debt-to-GDP is 114% (peaked at 94% in the USA).  The most interesting example is HELOC debt, which has grown at an annual rate of 13.9% since 2000! The total amount of HELOC debt in Canada now stands at $235bil, which is 12% of GDP (in the USA, this peaked at 4.5%).  According to the Bank of Canada, 40% of HELOC borrowers do not make regular payments that cover both interest and principal. Keep in mind the interest on HELOC debt is floating rate.

Scary macro statistics are not a predictor of future market direction.  But in this case, the market has already turned. House prices made parabolic moves last spring.  In a four-month period, Dec 2016 to April 2017, prices rose 11% nationally and 28% in Toronto. Since then, detached home prices are down from their peak; 5% nationally, 14% in Toronto, and more than 20% in some Toronto suburbs.  The only segment of housing that isn’t down in Canada is condos. Condos are trading at sub 2% cap rates in Toronto and Vancouver. Landlords in those cities lose $500 and $1,600 per month, respectively, owning and renting out a unit.  Regulatory changes have restricted rental increases to 1.8% or less and have restricted the ability to use Airbnb. Landlords have no ability to generate enough rent in the future for these condos to become cash flow positive and thus it is only a matter of time until this segment also sees large price declines.

Genworth Analysis

Balance sheet risks

MIC has an estimated outstanding transactional insured mortgage balance of $118bil.  Outstanding mortgages underwritten since 2011 total $97bil and I estimate the effective LTV to be 90%.  Total equity plus loss reserves is only $4.08bil. My personal view is that losses from defaults will wipe out the equity of MIC’s shareholders.  

There are three important considerations for calculating prospective losses: LTV, default rate, and house price decline.  

Regarding LTV, it is important to note that MIC is liable for much higher losses than its reported LTV would suggest.  For defaults, MIC has to pay the bank any missed interest payments. It also has to pay all the legal fees and repair costs associated with foreclosing and selling the house.  MIC has estimated that these extra costs, on average, are 15% of a house’s value. In my analysis I add 15% to MIC’s reported LTVs to arrive at what I call an effective LTV. An interesting point is that MIC’s newly originated mortgages are underwritten at an average LTV of 92 or 93%.  Given the extra losses incurred upon default, it is effectively underwriting new mortgages at LTVs of 107 or 108%.

On the default side, most people will tell you that Canadian default rates are very low, and this has been true historically.  In the last housing cycle, in the early 1990s, defaults peaked in the 1% range. However, I think these historic defaults are not predictive this time.  The home price appreciation and debt build up are so much higher than what Canada has experienced in the past that I think we are likely to see much higher defaults as a result.  To use the USA experience as a benchmark, Fannie suffered defaults in the 7% range. Most US mortgage insurers saw defaults north of 15%. In my analysis, I present scenarios for defaults in the 5 to 15% range.

The last component for calculating prospective losses is how much house prices will decline.  Here, I chose a range of 20 to 40%. On the low end, I use 20% because that is the decreases in affordability due to B-20.  On the high end, I consider a 40% reduction, which would simply return the price-to-income ratio of Canada to the USA peak of 3.8x.

 

Table 1 summarizes estimated losses that MIC could face under various house price and default rate scenarios. As you can see, losses anywhere near those experienced by the US mortgage insurers would see huge losses to equity, if not a complete wipeout.  Even in the lowest loss scenario, MIC would wipe out all its excess regulatory capital of $400mil. This would hamper MIC’s ability to write new business and it may force them to sell equity when their stock is already under pressure.

Table 1: Losses for varying default rates and housing price decreases (CAD$mil)

Housing price decrease (%)

Default rate (%)

5

10

15

20

30

40

531

976

1,485

1,062

1,952

2,970

1,593

2,928

4,455

Total equity plus loss reserves:  4,080

I think these calculations are conservative for several reasons.  The analysis does not take into account fraud (which has been documented in the Canadian mortgage market), it does not include any losses from insurance currently being underwritten, and it also does not include any losses from their portfolio insurance business.

Income statement risks

Some people may see my balance sheet views as overly bearish.  However, in the instance the housing market has a slow decline, the income statement should also take a large hit.

MIC has an aggressive revenue recognition model. While the insurance lasts 25 to 30 years, MIC recognizes revenues in accordance with its historical pattern of loss.  Since we have been in a decades-long bull market, the model sees very little risk to a mortgage that is over 5 years old. As a result, MIC books over 80% of its revenues in the 5 years after the premium is received.  This explains why the company only has $2.1bil of unearned premiums reserves against their total mortgage book of $218bil (which is an estimate, the original insured amount is $495bil).

So far in this downturn, MIC’s earnings have already taken a hit. Transactional premiums, its main business, were down 14% year-over-year.  Portfolio insurance premiums, where it insures a portfolio of mortgages for banks, was down 51%. Earnings were higher year-over-year, only because its loss rate was 10%, 17 points lower than the 10 year average loss rate of 27%.  

MIC’s income statement is impacted in many ways in a down cycle.  If you were to normalize premium volumes, with 30% less volume (historical norms) and house prices 20% lower (conservatively), and extend revenue recognition periods to 10 years (which still feels aggressive), earned premium revenues would be only $39mil annually.  If you simply amortize their current unearned premiums over 10 years, total yearly revenue would only be $249mil. If expenses and investment income were held at current rates, and loss ratios were at the 10-year average of 27%, after tax earnings would only be $166mil.  That is a far cry from the more than $400mil of earnings they currently report. It is not hard to see how MIC generates much less income going forward, even if losses don’t increase.

Other MIC risks

There are other miscellaneous risks to be aware of with MIC.

Its investment book is risky and highly correlated to its operations.  Of its $6.2bil investment portfolio, $1.2bil is invested in the debt of other Canadian financials.  It also owns $546mil of preferred shares; $329mil of these prefs are in Canadian financials. These shares are very risky, and yield only 3.9%.

Genworth Financial (GNW) owns 57.1% of MIC’s shares.  GNW has been in financial trouble since 2009. There has been speculation for years that GNW may be forced to sell its MIC stake to raise capital.  This large holding creates a overhang on MIC stock limiting its upside potential.

Despite MIC having only $400mil more capital than its regulatory minimum, it has raised its dividend and bought back shares.  In my view, this is lighting needed capital on fire.

Risks to being short

MIC’s shares could appreciate from ~1x book, to 1.2 or 1.3x book given its reported (but in my eyes dubious) ROE of 13%.  This however, is a small downside risk, compared to the upside return of 100%. The most likely downside scenario is the short bleeds away money from the dividend yield of 4.6% and the cost to borrow the shares of 2.5%.  It should also be noted that when Canadian financials suffered financial stress in the spring of 2017, the borrow rate for HCG and EQB stocks went inexplicably high, roughly 100%. A similar rise in borrowing costs would hurt the upside potential of this idea.

Conclusion

MIC has a very risky business that is dependent on rising home prices.  Canadian home prices are dropping as the bubble appears to have burst. My view is that the debt build up in Canada will mean that the price drop in houses will be precipitous and defaults numerous.  Likely, this will result in a complete wipe-out of MIC’s equity. However, if I am wrong and Canadian housing experiences a soft landing, MIC’s earnings could drop more than 50%. Either way the stock should trade much lower than its current price of 1x book.

Ira Sohn 2018

 

 


5 comments

  • Re: “The most likely downside scenario is the short bleeds away money from the dividend yield of 4.6% and the cost to borrow the shares of 2.5%.“

    So basically, you have to fund 7% (assuming cost of borrow stays flat) in the hopes that your thesis plays out and the entire financial world wakes up to your point of view? Not sure this is the correct position to run, but hey, it’s your money,

    Dan
  • Sorry for the late comments!

    Brion – great to hear from you and thanks!

    Davo – you are right. Also, Genworth Australia already extended their revenue recognition periods. I think its just a matter of time till MIC does as well…

    Alicia – the simplest way to see the 20% is to try out a mortgage calculator for yourself and change the interest rate by 2%. This is a good one https://www.ratehub.ca/mortgage-payment-calculator
    While you are right the fixed period is 3 to 5years, the mortgage is amortized over 25 years. That means it is much more like a 25yr bond than a 3yr bond in terms of interest rate sensitivity. Roughly speaking every 100bps higher in interest rates means that the price should drop 10%. Households cannot lengthen amortization period. For mortgage insurance the amortization period is not permitted to be longer than 25 years. MIC is responsible for all losses on the insured amount, but you are right they do not insure homes worth more than $1mil. MIC’s deals are roughly 40% from the Big 6, 24% from regional lenders and credit unions, and 36% from mortgage finance companies. They do much more business with non-big 6 lenders than people would expect. Hope that helps and let me know if you have any other questions.

    Tim
  • Nice write up. You thesis also applies to another housing bubble country mortgage insurer, Genworth Mortage Insurance Australia, GMA on the Australian exchange. Laws differ of course, but the scenario is similar, the financial viability of an insurer to pay claims following the deflation of a housing bubble.

    Davo
  • Interesting piece. Can you share your calculation for the 20% decline on home price due to the new B-20 guidelines? I heard banks typically use a lower 3-yar rate to assess perspective home-buyers so I wonder whether the impact of 20% is overestimated. Also, won’t the households be able to lengthen amortization to pass the stress test?

    “Keep in mind, mortgage insurers in Canada are responsible for 100% of the losses when a borrower defaults” → I’m not sure about MIC but CMHC insures up to CAD$1mn (though that’s already a lot).

    Are the borrowers MIC deals with mainly Big Six? Any private mortgage lenders (hence with potentially higher mortgage default rate)?

    Alicia
  • Tim,

    Great writeup, thanks for posting and congratulations on the Sohn recognition!

    Brion

Leave a comment