Blessed by the Potato

The debt-to-income ratio in Canada has been breaking new records for the past few years, and a little while ago surpassed the lofty level reached in the US before their housing bust. There have been a few slightly alarmist articles in the press about that, but even more that brush it off. The latest (and what made me decide to write this post) comes from Robb at B&E:

“The rising debt-to-income ratio makes for splashy headlines, but all it means is wages have been flat for a few years while cheap borrowing rates fueled a huge increase in low interest mortgage debt.”

Robb makes light of the record reading in debt-to-income, and I think I know why: it’s a difficult metric to wrap your head around, as evidenced by the fact that the rest of the post talks about matters on the single-household level. The debt-to-income measure is however a population measure, so it’s hard to interpret thinking about it as though it were any given household.

Indeed, with the measure at “only” 164%, it sounds downright low if you approach it with the right mindset. If you’re at the stage of your life where you purchased a house at a fairly prudent 4X your income and put 20% down, your debt-to-income from the mortgage alone would stand at 320%, so it’s hard to see why a national figure at half that is alarming.

To interpret it, you must first remember that it is a population measure. It includes those people earlier in their lives with massive mortgage debt loads as well as those who are near retirement, and should have peak earnings with minimal debt. Even then, saying that any one level has meaning is quite difficult, as you then have to parse the demographics, figure out how many old people you have and what their debt-to-income should be, how many young people, etc.

Fortunately, the second important thing to keep in mind with population measures like this is that changes over time are often more important than the absolute level. And that is what makes the recent readings in the debt-to-income measure alarming. Just 8 years ago (2005) the metric sat at 120%. That implies that Canadians have, on average, increased their debt loads by 37% in that time, or about 4% per year (relative to incomes, or in real terms).

On top of that, we have to consider how we would have expected the ratio to change over that time. One big factor is demographics: as the baby boomers near (and enter) retirement, we should expect a larger portion of our population to become debt-free — leading to an expectation that since 2005 that debt-to-income measure should have been trending down, not up. That means that we have some combination of seniors getting much closer to retirement with debt or even entering retirement with debt, and young people taking on disproportionately more debt, so much that it is swamping the demographic effect (mostly the latter).

Another influence over the past few years has been decreasing and record-low interest rates. People have suggested that it’s fine for people to be taking on more debt relative to measures like income because low rates have lowered the servicing burden. On the other hand, if people had been prudent, the low rates could have meant the same payments on the old debt would retire it even faster, another factor that might lead us to expect another incremental decrease in the population measure of debt-to-income. An increase instead means that we now have more debt that is even more interest-rate sensitive. Sure, it’s fine and dandy and easy to manage as long as interest rates stay low, but it implies an added risk to the population should rates increase in the future.

In the context of a housing bubble, this is even more meaningful. Much of the increased debt load has been mortgage debt, used to pay for houses that have increased in price. This is debt that will be long-lived, giving rates lots of opportunity to increase and make repayment difficult. If boomers are carrying debt into retirement, it may be due to a plan to hold onto more real estate — and the associated mortgage and HELOC — for the time being and downsize later, building more pressure for a future crash. The point of it all is that debt-to-income is a measure of risk.

Some have asked why another metric isn’t used, such as debt-to-assets, or reformulated, debt-to-equity. And the reason is that it doesn’t give the same warnings on risk. Yes, if you have assets to support your liabilities, you could in theory sell your assets to settle the debt. However, people in general don’t do that so debt must ultimately be repaid with income. It’s tougher to use debt-to-assets to identify risk (or changes in risk over time). In asset bubbles the debt remains after the assets correct, so you could be refinancing appreciating assets to use more debt to buy more assets, keeping the same amount of equity on the way up. Unfortunately once the asset stopped appreciating, the debt remains, and only after the fact would you see the debt-to-equity ratios move. By way of example, you could save up $25k on a $50k salary and buy a $250k house with 10% down, giving you a 90% reading on debt-to-assets and 450% debt-to-income reading — high on a population measure, but not outrageous for a single, young household. If the house appreciated to $350k, you could sell it and move up to a $1.25M house and still have 10% down, giving you the same 90% debt-to-assets measure. Yet now your debt-to-income is a whopping 2250% — it is clear that you will never pay that mortgage back on a $50k salary.