How CPI common, a flawed metric, blunted the Bank of Canada to inflation

Open this photo in gallery:

A core inflation gauge that was once the Bank of Canada’s favourite has become its problem child – at a time when the central bank desperately needs what that measure had promised, then sorely failed, to provide.

In a speech this month in Halifax, Bank of Canada Governor Tiff Macklem acknowledged that CPI-common – one of a trio of measures that the bank has relied on since 2016 to get a read on underlying inflation pressures affecting the broad economy – is too broken to be useful as we approach a pivotal phase in the bank’s fight against inflation. While the bank will lean heavily on its other two core measures (called CPI-median and CPI-trim) to gauge how well its sharp interest rate increases are working, he said, “We are reassessing CPI-common.”

By last week, when he spoke with reporters from the International Monetary Fund and World Bank meetings in Washington, he omitted CPI-common entirely when discussing the importance of core inflation readings. CPI-common has quickly become the Voldemort of inflation indicators.

The rapid rise in inflation in the past year and a half exposed serious flaws in a statistical tool that, on paper, looked fantastic to the central bank when it began looking at better ways to track core inflation nearly a decade ago.

The “common component” measure, as the bank called it then, is a mathematical model that separates from each component of the Consumer Price Index the portion of its price change that it shares with all other components, and discards the portion that is unique to that single product or service. The portion that remains indicates the underlying inflationary pressures affecting the entire economy.

Crucially, the common component showed a strong correlation with the output gap – the measure of how close the economy is to running at its full capacity. Inflation stemming from the closing of the output gap – when demand catches up to or even exceeds supply – is exactly the kind of inflation central banks worry about.

It’s no wonder, then, that the Bank of Canada talked fondly and often about the common component measure when it approached its 2016 decision to adopt new gauges for core inflation. But the biggest knock against it at the time was that it’s highly technical and opaque – a pretty tough sell to the general public.

In the end, the bank decided to adopt three core measures rather than just the one, with the other two – CPI-median and CPI-trim – offering more straightforward ways of looking at the underlying inflation trend. Officially, the Bank of Canada insisted that it looked at all three equally; but there was a lingering suspicion that CPI-common was the unspoken favourite. It was the magical black box with the potential to spit out the most helpful data.

That is, until this year – when CPI-common spit out dangerously misleading garbage.

CPI-common’s statistical model is built on historical CPI data, and the steep rise in inflation in the past year and a half has twisted those historical averages pretty severely. The more inflation has surged, the more the foundations of the model have shifted. As a result, the CPI-common readings have been revised sharply upward as the year has progressed. It now appears that the figures from earlier in the year badly underestimated core inflation.

The Bank of Canada didn’t realize this at the time – and, indeed, appeared to lean on still-low CPI-common readings as evidence that Canadian inflation wasn’t yet being driven by a closing of the output gap.

In a speech last March, just after the bank started raising rates, Mr. Macklem said the CPI-common was running at a modest 2.3 per cent in January – even though the overall inflation rate had topped 5 per cent, CPI-trim had hit 4 per cent and CPI-median was 3.3 per cent. The only slightly elevated CPI-common reading was one of the indicators that suggested to the bank that the output gap had closed, but only just; the other, higher inflation readings were being skewed by the effects of supply bottlenecks on goods from overseas.

“The elevated inflation we are experiencing today is not the result of too much demand in the economy,” Mr. Macklem asserted.

But according to the subsequent revisions, CPI-common last January was not 2.3 per cent, but a much more problematic 3.8 per cent. It had been north of 3 per cent for six months.

Had Mr. Macklem and his colleagues been looking at the revised CPI-common figures rather than the original underestimates, they would have received a much earlier signal that the output gap had closed and that we had moved into a state of excess demand. That would have argued for the bank to start raising interest rates at least a half a year earlier than it did – thus avoiding the aggressive pace of hikes of the past six months.

While the bank was looking at a whole range of indicators in that decision, the deeply flawed CPI-common readings must have been a substantial contributor to its hesitance.

Last week, Mr. Macklem was again talking about the importance of gauging the output gap, as the bank raises interest rates to try to reverse excess demand and cool inflationary pressures. Core inflation measures will be critical tools in tracking the bank’s success, and determining the path of interest rates.

But the bank will have to sort that out without CPI-common. As a vehicle to help guide policy, it steered the bank into a very costly ditch.