Rising Canadian Core Inflation and Whether Common Component CPI is of Any Use

• Average core inflation continues to climb…

• …and put further pressure on the BoC…

• …to move into restrictive territory with another +75bps move

• Large, ongoing revisions to common component CPI…

• …keep raising average core inflation estimates…

• …and question the usefulness of the measure

• CDN CPI m/m % NSA // y/y %, July:
  
• Actual: 0.1 / 7.6
• Scotia: 0.4 / 7.8
• Consensus: 0.1 / 7.6
• Prior: 0.7 / 8.1
• Average ‘Core’ CPI: 5.3% y/y (5.2% prior, revised up from 5.0%)

Core inflation continues to creep higher to new records each month as inflationary pressures rise and broaden (chart 1). That lends itself toward expectations for another large hike of 75bps at the September 7^th meeting that is mostly but not fully priced in OIS markets.

I think the market reaction was generally sensible. GoC two-year yields have climbed about 10bps and the C$ appreciated by about a quarter of a cent on a USDCAD basis. The new information incrementally reinforces a strong tightening bias.

The Bank of Canada won't care about the headline softening. They'll be more concerned about ongoing upward pressure upon core measures. The data lends itself to a 75bps move on September 7th that would bring the policy rate closer to being in very mildly restrictive territory given estimates of the neutral policy rate range of 2–3%. Unlike the Fed, there isn't much on the agenda between now and then that could be incrementally impactful. This is the last CPI reading before that meeting and the next jobs report arrives two days after the decision. Plus, the September meeting is a statement-only affair with a day-after presser to be hosted by an as-yet undisclosed Governing Council member, sans forecast (unlike the Fed that will also update the SEP in September but on the 21^st and hence after the BoC). There are no scheduled BoC speakers between now and the blackout period that commences one week prior to the decision.

IS COMMON COMPONENT CPI OF ANY USE?

Average ‘core’ inflation—an average of the three central tendency measures—moved materially higher to 5.3% and with the prior month revised up two-tenths to 5.2% y/y. The prior month’s revision is because they keep messing with common component CPI that was revised up from 4.6% to 5.3% which is large. This issue is worth exploring further before turning toward other details.

The challenges with common component CPI run much deeper. Chart 2 shows the impact of cumulative upward revisions to this measure dating back to May 2021. The original reading for January 2022 was 2.3% y/y and cumulative revisions have taken that up to 3.6 which in the world of inflation estimates is huge. For February it’s now 3.9% versus the 2.6% original reading. For March it’s 4.3% versus 2.8, for April it’s 4.6% versus 3.2% and for May it’s 5.1% versus 3.9%.

So what’s going on here? I don’t know of another country that is messing with its core inflation readings like this while offering little useful explanation other than a remark that the black box changed again:

"In the case of CPI-common, revisions are due to the statistical technique used, as the factor model is estimated over all available historical data."

Wonderful, just sweep it under the rug. One possible explanation is that common component CPI has been getting revised higher as the mean and variance of headline CPI have been moving higher which serve as model inputs to the constantly re-estimated common component CPI inflation factor model. This raises the issue of how useful a measure of ‘core’ inflation is if it simply follows headline CPI in lagging fashion through model revisions. I’m unsure of the degree to which this explanation is valid for chronic upward revisions and so there should be further explanation offered by either StatCan or the Bank of Canada if they wish this measure to have any credibility with market participants and policymakers especially since there is still a segment of market opinion that (wrongly imo) believes the BoC pays more attention to common component CPI than the other measures.

In the meantime, I would advise placing greater emphasis upon other core readings like trimmed mean (5.4% y/y), weighted median CPI (5.0% y/y), and simpler measures dating back to before the BoC over complicated things to no apparent benefit with its central tendency gauges.

The hot core readings are also seen in simpler gauges and there remains an ongoing case for sticking with those gauges in keeping with many other countries. Simple CPI ex-food-and-energy was up 6.1% m/m SAAR which picked up from 5.2% the prior month. CPI excluding the 8 most volatile items was up 5% m/m SAAR and matched the prior month’s estimate. CPIX was up 5% m/m SAAR from 5.9% the prior month. In all of these cases, the rates have cooled from their peaks in March when the economy was reopening from omicron restrictions, but they are all vastly higher than the BoC’s 2% headline CPI target over the medium-term that uses the core readings as an operational guide. The fact that these m/m SAAR core readings are still running at the hot rates that existed on a trend basis before omicron reopening effects should give little to no comfort to doves.

OTHER DETAILS

Breadth remains very high (chart 3). 

Both services and core goods inflation continue to push higher (chart 4). The driver of weaker all-in goods inflation is weaker commodities, but underlying inflationary pressures remain hot.

Chart 5 shows that housing is cooling as an inflationary driver, at least the way it’s captured in Canada. Canada captures housing inflation using the house-only component of builder prices (ie: ex-land) as a driver of homeowners’ replacement cost and builder price inflation has been ebbing. Canada does not use owners’ equivalent rent like the US BLS does. 

Chart 6 shows other housing-related contributions to CPI inflation. Mortgage interest cost is rising with rate hikes and with much greater upward pressure forthcoming from this reading, but the BoC clearly looks through the directly measured effects of their rate hikes on inflation. Rent has cooled at least for now, but weakening housing markets may keep upward pressure upon this component. Housing’s influences upon CPI inflation nevertheless run much deeper given direct and indirect effects not well captured by just looking at builder prices.

Across other components, airfare continues to soar (chart 7). 

Auto price inflation remains disconnected from CAD changes because of supply chain issues that have made it a seller’s market (chart 8), but even if supply chain issues continue to gradually improve the next upward leg could come from the lagging effects of CAD weakness upon the next couple of years of new models introduced around late summer into the Fall.

Food inflation remains hot (chart 9). 

The recreation/reading/education category saw big gains in travel-related categories (chart 10).

Chart 11 shows a breakdown of the CPI basket in y/y terms and chart 12 does the same thing but in terms of weighted contributions to inflation by category.

Chart 13 shows a breakdown of the CPI basket in m/m NSA terms and chart 14 does the same thing in terms of weighted contributions.

Please also see the accompanying table that provides a detailed breakdown of the basket including various other measures and micro charts.