Short sellers and regular Canadian investors have very different views of Canadian financials. These disparate views are easier to understand when you take into account how both groups view the investing world. One group views the world through the lens of probabilities, the other looks at how things are actually performing.
Short sellers (I include myself in this group) are from the trading and hedge fund world. In this world, market prices change as probabilities change. An example would be corporate bond trading. If a hedge fund buys a corporate bond and the probability of default decreases, the bond will go up in price and the hedge fund makes money. If the probability of default increases, the bond price drops and the fund loses money. Betting on changing probabilities is largely how traders and hedge funds make their money.
The accounting for Canadian financials does not operate in the world of probabilities. Lenders only reflect losses when defaults occur. There is very little provision for losses due to changes in the probability of defaults. While this difference of worldview - probability versus actual default - seems small, it is actually a huge deal.
An investor that does not view investing through a probability lens looks at the reported earnings from Canadian lenders and thinks ‘what’s the big deal?’. Earnings are at record highs, return on equity is great, defaults are low, and there is low provisions for losses. Great report.
An investor trained to view the world through the lens of probabilities looks at the same report and thinks: ‘Houses are so expensive that future losses seem inevitable - why aren’t they reserving more now?’ or ‘There is so much personal debt that the probability and number of defaults is likely to rise, why isn’t that reflected?’ or ‘If we look through a cycle of defaults and normalize earnings for higher losses the income statement would be much lower and ROE’s much less; why isn’t this riskiness being demonstrated in the accounting?’
The same financial statements - very different perspectives. In fact, these views are so different that the two groups can hardly find common ground to discuss.
Should accounting be based on actual losses, or reflect a more probabilistic approach? While I think it should be probabilistic, both approaches have some merit. Companies need the ability to look through cycles and weather out periods where markets have wildly swung too far from reality. Conversely, using actual losses can allow lenders to under emphasize risk, access capital they shouldn’t, and expose investors to more risk than they were aware of. An example of such glaring distortion in Canada is alternative lenders that focus on riskier borrowers (HCG and EQB). These lenders have 70% less set aside for defaults than the bigger banks who service better borrowers. Does that make any sense? This can only occur when a probabilistic approach is totally ignored.
To give a specific example of how misleading and catastrophic actual loss accounting can be, let’s take a brief look at the Bank of Ireland. In March 2008, Bank of Ireland reported earnings that had provisions for loan losses of 0.44%. Ireland had similar accounting to Canada no provisions had to be set aside as long as the original loan was being paid. That meant that Irish lenders made no adjustments for the lending risk they were taking as their housing bubble was inflating. In the 5 years after this March 2008 report, the bank lost €9bil, or 1.4x of its 2008 shareholder equity of €6.5bil. Non-performing loans were 19% of Bank of Ireland’s total loans! The stock went down 99% and would have been 100% without the government bailing them out.
This is what can happen when there is a build-up in risk that is not reflected in any way in the accounting of a bank. Shareholders that aren’t from the world of probabilities get blindsided.
I’m writing this after reading Home Capital’s earnings report where, despite all that happened in the last quarter, they dropped provisions for loan losses from 0.16% to 0.07%! Home Capital weren’t alone in dropping provisions in the face of increased risk. Genworth MI Canada (MIC) recorded a loss ratio of just 3% last quarter (over the past 12 years the loss ratio has averaged 25.25%) and as a result posted decent results. This low level of provisions comes from two companies with virtually no reserves for an increase in mortgage defaults. Both companies only have their equity as a risk buffer. Since both companies are levered more than 10x, that isn’t much of a buffer.
Let’s look at recent news and how mortgage default probabilities might change as a result:
Indicators for a decreased probability of defaults:
- House prices are still up quite a lot over the past couple of years. Any defaults now will have low loan losses as a result
- Economy has been reasonably strong lately
Indicators for an increased probability of defaults:
- House prices are dropping quickly in GTA since April peak
- Evidence of terrible, horrible lending at HCG (is 7bps enough to cover this risk?!?!?!?)
- Bad lending may not be just a HCG issue
- Overall increase and level of Canadian indebtedness
- Helocs total 14% of GDP, US was 3% at peak
- Increasing focus on housing in GDP (ie. huge job losses when house bubble bursts)
- BOC raising rates
- B20 changes making it more difficult to get a mortgage, if they are enacted
Collectively, I think these indicators suggest the probability of losses in the lending and mortgage sector has increased dramatically.
This risk is not reflected in the accounting of Canadian financials. I urge investors who are taking these accounts at face value to view these issues probabilistically. I think you may find the results aren’t as good as they appear and things are going to get really ugly.
Disclosure: short HCG, EQB and MIC