Why doomsday housing forecasts have proven wrong

The housing market across Canada collapsed in March after governments imposed a pandemic-mandated lockdown. Sales tumbled, prices fell, and new listings almost disappeared.

It was all doom and gloom, all the time.

The uncertainty about COVID-19’s impact made the future of housing markets look uncertain, if not bleak. Then came the pessimistic forecasts of a weakening housing market. Leading the charge in May was Canada Mortgage and Housing Corp. (CMHC), which projected average housing prices would fall between nine and 18 per cent.

However, November housing stats released by Canadian Real Estate Association (CREA) paint a different picture altogether. Housing markets have been resilient - indeed, buoyant.

Actual sales in November were up 32.1 per cent from the year before. The qualityadjusted home price index (HPI) was up 11.6 per cent from November 2019.

Did the forecasters get the markets wrong? Or is it too early to celebrate?

In addition to CMHC, others, including the National Bank of Canada and Fitch Ratings Inc. have forecasted that housing markets will decline in 2021.

Forecasting is more of an art than science. George Box, a famed statistician who devised innovative forecasting tools, has warned that all (statistical) models are wrong, yet some are useful. But distinguishing wrong forecasts from the useful ones is even more challenging than forecasting.

Essentially, statistical forecasts project the future by relying on past trends. Sometimes, the predictions depend solely on the past realizations of observed indicators, such as housing prices. At other times, forecasters use additional relevant data to inform their statistical models. For example, someone might also include information about mortgage and unemployment rates when determining future housing prices or sales.

Forecasts based solely on previous observations assume the past holds the key to the future and nothing else is required. If other variables inform the model, the forecaster must make additional assumptions about what the future might hold for those variables. The latter type of model would forecast the future based on the past (housing prices or sales) and the assumptions made about supporting indicators such as the mortgage rates.

If this isn’t complex enough, consider that forecasts also differ based on the statistical method used. Projections are often based on partial equilibrium models, where a small set of variables is assumed to capture economic or social behaviour in a market. Others are estimated by general equilibrium models, which try to analyze the entire economy.

The problem, perhaps, is not with the forecasts, but with the forecasters. They seldom reveal the assumptions they have made, the motivations behind those assumptions, the statistical tools deployed or the variables they included or excluded. Also, forecasters often release point forecasts, a number, rather than a range of numbers or the uncertainty surrounding the estimates, known as the confidence interval.

With so much subjectivity involved in estimating a statistical model, it’s little wonder that forecasters seldom agree, or that their forecasts pan out. Remember, all models are wrong. So, what to do?

Going by first principles might be a prudent way to think about future outcomes. With interest rate cuts planned as a response to the expected slowdown in the economy, Doug Porter, BMO Financial Group’s chief economist, in March warned that rate cuts were likely to “put (Canadian) housing market on steroids.” His forecast contradicted the gloomy outlooks others had projected.

With interest rates at record-low levels, home ownership under COVID-19 has become even more accessible to those whose employment prospects are secure. Hence, Porter’s forecast has been on the mark.

How housing markets will evolve in 2021 depends upon the economy, markets and consumer preferences. COVID-19 has hit some economic sectors hard, but it has left others unscathed. Some have even thrived.

For example, sales of electronic equipment are up even though small retailers have been forced to shut their doors. Condominium sales are struggling, but suburban housing and cottage country homes are experiencing unprecedented demand.

Dismal housing forecasts perhaps assume that the economy will weaken when stimulus packages expire. One can also see that certain economic sectors and their associated labour markets might not experience such a downturn. Hence, those with secured economic prospects may continue the homebuying mania made possible by ultra-low mortgage rates.

Murtaza Haider is a professor of Real Estate Management at Ryerson University. Stephen Moranis is a real estate industry veteran. They can be reached at www.hmbulletin.com.