Thirty years ago, nearly nobody was homeless in Vancouver. Today, at least 2,500 people are homeless, according to the latest GVRD count, and this is more than double what it was a decade ago.

Not surprisingly, homeless people are more likely to face problems other than lack of shelter, such as drug addiction, mental illness, family breakdown, victimization, poverty and lack of marketable skills and education. But, none of those things are new, so why is there so much homelessness now?

A great paper by John Quigley and Steve Raphael at the University of California at Berkeley offers part of an answer. They examine a vast amount of data on shelter usage in U.S. cities over many years. They find that variation in homeless shelter usage across cities and over time is driven by high rents and low incomes, and not by the usual suspects of drug addiction, mental illness, family breakdown and criminal victimization.

Increasing the incomes of poor people is easy: write them cheques. We already have a huge array of transfers to low-income people, targeting mainly those who work and have children. These transfers could easily be increased, and/or expanded to those without children and without work.

Rents are a tougher nut. My own research suggests that the key culprit here is that rents for the spaces that poor people typically occupy (small spaces, apartments in older buildings) have increased way faster than the rents of richer people. There are three main reasons for this. First, the federal and provincial governments stopped building public housing in the early 1990s and early 2000s, respectively. Second, new private sector building has almost solely been for owned, rather than rented, accommodation. Third, new private sector building has overbuilt older housing that poor people would previously have rented. All this adds up to fewer rental units available for poor households than 30 years ago, and fewer units means higher rents.

Policy action to reduce homelessness has to address the supply of rental units. First, we have to avoid policies that reduce this stock, like rent control, which makes landlords want to sell their buildings to owners and condo developers. Second, we have to act directly in building housing for the rental market. And finally, we have to focus on the number of units built, even at the expense of their quality. B.C. has taken some action lately, but much of it has been in improving the current stock, rather than on increasing the number of units available to low income people.

Krishna Pendakur is a professor of economics at Simon Fraser University.