Measuring Canada’s GDPEE

Statcan plunges into sewer for cannabis figures

On Thursday, Statistics Canada put out a paper entitled “Wastewater-based estimate of cannabis consumption, March to August 2018.” The world, watching Canada conclude its Year of Cannabis, is beginning to clue in that one of the supreme benefits of legalizing the weed is that you can do proper science on it without legal interference or excessive social stigmatization.

This seems to be true even of scientific activity that would have been lawful beforehand. Statcan has embraced the new national project of doob science with restrained excitement, even passion. There is something new, an entire industry a-borning, to be measured!

How often, in a data-monger’s whole career, does this happen? Statistical troops: to your abaci!

If you share even a fraction of this nerd spirit, you cannot help be interested in this particular piece of the evolving cannabis-measurement apparatus. Statcan has recruited engineers at sewage treatment plants in Halifax, Montreal, Toronto, Edmonton and Vancouver to take measurements of THC-COOH, the most detectable metabolite of tetrahydrocannabinol. “Metabolite” is the technical term for what is left over after your body is done turning THC into an enhanced appreciation for The Electric Prunes and an extreme craving for Flamin’ Hot Cheetos. In short, Statcan is doing a piss test — but on entire cities.

The headline finding in the initial paper is that, among the five cities, Halifax’s pee yielded the heaviest point estimate of per-capita cannabis use. This will no doubt be a great help to the Trailer Park Boys-based component of the city’s tourism business. But the whole idea of estimating cannabis usage from metabolites in sewage is in its infancy here, and the bands of uncertainty surrounding the numbers are very wide.

Although there is some existing scientific literature on estimating mass cannabis consumption from sewage samples, it comes from Europe, and in matter like this you have to roll your own science, so to speak. So far Statcan only has six months of data. You can’t get a full picture of seasonal trends and environmental influences without several years of numbers. And if legalization changes the overall quantity of cannabis consumed, that will make it a little harder to correct for seasonal changes: these are measurements Statcan would ideally have been taking for many years before C-day.

Still, this kind of measurement has potential, even if it cannot be perfected.

It might turn out to be impossible to make city-to-city comparisons of cannabis intensity of the sort that StatCan is highlighting now.

Taking the samples will still allow the agency to make intertemporal, withincity comparisons. You may never know if people really smoke more pot in Halifax than in Edmonton, but if the level of cannabis metabolites doubles in one place and does not change in the other, you know something interesting is going on.

Moreover, we will have excellent mass sales data for legal cannabis in all of these cities.

If you compare the changes in sales of legal weed to the changes in peoples’ pee, you can probably get a really good estimate of the relative size of the black market.

So this is a terrific, promising statistical instrument, even assuming the worst about the natural limitations. Sewage testing for THC metabolites doesn’t raise any concerns about individual privacy.

Taking the measurements does not appear to cost much, if anything. And the method could, in principle, perhaps be extended to other recreational drugs.

No, you can’t beat data straight from the faucet — or the drain.

You may have noticed that on Monday Statcan officially backed down from its controversial proposal to demand personal account information from the big Canadian banks: the agency put out a notice that it has “put this project on pause” while the federal privacy commissioner subjects its senior people to light interrogative torture for a while.

Statistics Canada, as an institution, behaved in an almost unforgivable manner when it came to the plans for personal banking data: it has transparency duties, and specific obligations to the commissioner, that appear to have simply been forgotten.

But, in a way, Statcan’s frolicking in a river of piss goes to demonstrate why a scientist’s eagerness briefly got the better of the agency.

It is frustrating and expensive to measure a country and its economic activity by means of surveys, questionnaires and phone calls.

You are a constant prisoner of mathematical sampling assumptions and a hostage to your fellow citizens’ goodwill. It would be so much nicer to have a quantity you could measure in a test tube — “directly” — without all that worrying about personalities.

How glorious it would be if inflation or the GDP could be estimated from sewage. (Come to think of it, this is probably possible. It just wouldn’t be a very good estimate. I think...)

Compelling banks to surrender data on account activity would give Statcan economic information of astonishing power and fine resolution.

Of course, if you gained access to this data about your friends and neighbours, you might suddenly find yourself with a lot of power too ...

TERRIFIC, PROMISING STATISTICAL INSTRUMENT. — COLBY COSH