Employment numbers will be released Friday, and economists are expecting Statistics Canada to say that about 20,000 jobs were created last month. But what’s in a forecast, and how accurate are these predictions, anyway?
A look at the past year of forecasts shows plenty of reason to be skeptical. In three months of the past 12, the economists were almost bang on. But for other nine months, forecasts have been off by as much as 54,000 jobs. The average “miss” in Canada is 21,000.
The findings were calculated by comparing average forecasts compiled by Bloomberg between September, 2010, and September, 2011, with actual results. Forecasters were too pessimistic in seven months and too optimistic in five.
Why does this matter? Because investors and traders use these forecasts to form market expectations. A huge miss can easily sway currency markets, and to a lesser extent bond and stock markets too.
In the U.S., economists have plenty more indicators on labour market trends. There are monthly ADP reports on the private sector and Challenger surveys about layoffs. The U.S. Labour Department publishes information on jobless claims on a weekly basis; in Canada, EI reports are released with a two-month lag. And unlike Canada, the U.S. tracks whose benefits have expired without finding work.
That leaves people like Andrew Gretzinger, managing director and portfolio manager at Manulife Asset Management, with little else to go on but examining historical trends and looking at current cycles.
“It really is as much an art as a science – you look at the trends, figure out where you are in the cycle, and follow your guts,” said Mr. Gretzinger, named by Bloomberg as the country’s most accurate forecaster on employment.
He thinks Friday’s numbers will show a gain of 10,000 to 15,000 jobs. The past year has shown big job gains are typically followed by weaker numbers, and September was a very strong month, with the economy unexpectedly churning out 61,000 jobs. Mr. Gretzinger sees job growth moderating through the rest of this year.
StatsCan’s labour force survey is crunched by gathering information from 56,000 households across the country. It’s a huge sample size, but because it doesn’t cover the entire population, it too is subject to sampling error.
In October, economists also think the jobless rate rose to 7.2 per cent. The unemployment rate in September was 7.1 per cent – the lowest since December, 2008.
All of which is to say – job forecasts may give an interesting pointer on where the labour market may be headed. But they are best taken with a boulder-sized grain of salt.