Government Productivity After Four Decades of Information Revolution
Why isn't government getting better at delivering services?
Beginning with the previous post I’ve begun experimenting with adding AI-generated voiceovers for those of you who prefer audio. Please do let me know whether you love the idea or hate it - or whether you couldn’t stop laughing right the way through. (Just wait ‘till you hear how it pronounces “Shawinigan”.)
If you’d never set foot in an actual office, you’d probably expect more workers to produce more output. “Many hands make light work” and all that. But in the real world things don’t always work that way. Especially in the public sector. Here’s where we try to measure the fraught relationship between government bloat and productivity.
Just how bloated is our federal government and is the bloat getting worse? At least from the perspective of employment levels, the government itself provides us with plenty of good data.
The actual proportion of federal employees to Canada’s total population has been noticeably growing over the past five years. Because of the scale I chose for its y-axis, the graph below does exaggerate the changes a bit, but you can see how the federal workforce grew from around 0.72% of the total population in 2017 to 0.9% in 2023. That means about one out of every 59 employed Canadians now works for the federal government.
Over most of those same years, the proportion of executives (at all levels, from EX-01 to EX-05) to non-executives has also grown from its 2018 low of 2.44%, to 2.55% in 2023. Such changes will obviously drive up salary expenses, but they also risk sacrificing actual productivity for “process management busywork”.
Some might argue that a larger government workforce is necessary for expanding and improving government services. But in a world where automation and artificial intelligence are supposed to be making everything more efficient, should that be true?
How can we measure worker efficiency? When it comes to large economies, one popular approach is to track labor productivity - a measure of real gross domestic product per hour worked. Predictably, Statistics Canada publishes a labor productivity index where the 2017 value is set to “100” and the performance of all other years is measured against 2017.
As you can see from the graph below, the greatest improvements in the Canadian economy’s labor productivity took place between 1994 and 2014 - which just happened to have more or less been the years of highest software and networked systems adoption. (The graph conveniently skips 2020, when COVID lockdowns turned everything upside down.)
So efficiency in Canada’s general economy rose consistently. How did the government do through that same period? Labor productivity in government can be a bit harder to measure. After all, it’s not like they produce a physical product or sell services directly to customers.
The best way I can come up with to rate government worker productivity is through the size of the workforce. That is, government provides a set of services that doesn’t change dramatically from year to year. Sure, as the population grows they’ll need to process more passport applications and oversee the movement of greater numbers of international travelers across our borders. But task complexity shouldn’t increase faster than the population itself.
I should note that the government frequently adds new programs involving financial entitlements and tax credits. But those are simply disbursements and shouldn’t require significant increases in labor.
Assuming government labor productivity remains the same, the proportion of government employees to Canadian residents should remain steady. However, as we’ve seen, not only was productivity (i.e., the worker:population proportion) not steady, but it actually dropped by 25%.
Which is strange. After all, the government hasn’t been shy about taking advantage of digitization. I’m old enough to remember how, each spring, millions of Canadians would mail their physical income tax forms to vast processing warehouses in places like Shawinigan, Quebec, where clerks would open, read, check, and then process the contents.
Handling even simple tax forms would have demanded at least 15 minutes of labor. And there were millions of them. You young ‘uns should have seen ‘em all piled up in the local post office waiting to be postmarked. Today that entire process probably requires not much more than a handful of developers and a mid-sized team of network engineers.
So then why do we need 25% more federal workers per Canadian when so many of the basic services have been automated?
Theres a really good book called “Recoding america: why government is failing in the digial age and how we can do better” that does a good job on why digitizing govt doesn’t automatically improve things. One of her explanations was that govt digitizing doesn’t automatically improve productivity is that although they may digitize some parts of govt, they’re not improving the underlying processes so govt is still having to go through many hoops to deliver services. Another thing she mentioned is that policy making is considered a high status job but implementing them is considered low status grunt work. She also said that policy makers and the people implementing policies don’t communicate directly. And when they do, it tends to be a one way street (policy makers -> implementers).
Although she wrote this book about the US, im guessing the explanations are also relevant for canada
Sir, I offer two comments.
First, as always, a very interesting read.
Second, you solicited comments about your voice over. First, the AI "voice" certainly is monotone; good for a chuckle by listeners but the "voice" certainly wouldn't understand that we are laughing at it. As for the basic idea of a aural version of the column, I am much more interested in the written version for two reasons.
Reason one: I always find a point in a column (any column, any publication) where I say something like, "Who is that?" "What is their position/expertise/etc." "Wasn't the opposite asserted above here?" and so forth and I then go back in the column to see what I erroneously glossed over, just plain well missed, etc. Much easier with the written format than the aural format.
Reason two: I am an old dog and hearing loss has set in so my listening skills are pretty inferior.
Oh, and by the way, I also remember those tax forms. In fact, as an accountant in the old days I was "the one" completing X gazillion of the things (first by hand and then by computer to be printed) and sending them to the various tax centres. By the end of my work life we had graduated to online filing; a mixed blessing, to be sure.