Technological Change and Employment in Economic History

 

Technological change has been the chief contributor to economic growth since the industrial revolution. Yet, technological change always seems accompanied by anxieties related to long-term unemployment despite increases in both total employment and per capita income over the last 150 years.  This anxiety continues  with the current onset and diffusion of assorted new technologies including AI, machine learning and quantum computing.  Yet the evidence suggests that despite over 150 years of rapid technological change, more jobs have been created than destroyed so that on net employment has continued to rise and matched or exceeded population growth.

 

My coauthor Olivia Di Matteo (UBC) and I have a paper on the program of the Social Science History Association Meetings in Toronto this week that looks at whether the past can inform the future when it comes to the impact of technology – quantum computing in particular – on the economy.  Our paper overviews the recent history of technological anxiety with comparison to actual outcomes, surveys the state of quantum computing and the challenges it faces, and then tries to extrapolate from current available metrics and past performance what the potential effects on employment and income might be.  The historical evidence suggests a positive and significant relationship between income, employment and assorted measures of technological change including computing measures.  Going forward there is no reason why future growth cannot benefit from new quantum technology, but much depends on having a measure of quantum computing to gauge its impact on income and employment.  Measuring the impact of quantum computing is more difficult given that new metrics apart from those obtained during the age of classical computing may apply.

 

The focus in this blog post is the historical evidence on employment performance in three countries at the forefront of technological change over the last 150 years: The United Kingdom (Figure 1), Canada (Figure 2), and the United States (Figure 3)[See note at end of post for data sources]. The United Kingdom’s experience as the first industrial nation revealed increases in both employment and the labour force as technological change both created and destroyed jobs but with substantial net job creation.  Indeed, using census records on employment in England and Wales since 1871 and Labour Force Survey Data from 1992, Stewart, De, and Cole (2015) show declines in occupations such as agricultural labourers, washers, launderers, telephonists, and telegraph operators both in absolute numbers and as a share of employment.  Meanwhile, these declines were accompanied by increases in other occupations such as accountants, bar staff, hairdressers, and other services. Overall, employment in the United Kingdom has trended steadily upwards since the mid 19th century irrespective of massive technological change as Figure 1 illustrates.

 

 

 

The picture is similar for Canada, as illustrated in Figure 2.   Between 1851 and 2021, in tandem with a population that grew from 2.4 to 38.3 million – a 16-fold increase – estimates of the Canadian labour force show growth from 762,000 to 20 million – a 26-fold increase in size.  Employment data is available from 1891, and over the period 1891 to 2021, employment in Canada grew from 1.6 to 18.0 million – a 11-fold increase – while the labour force over the same span also increased from 1.7 to 20 million – an approximately 12-fold increase.  The slowdown after 2017 in terms of labour force and employment can be attributed to the impact of the pandemic, and as the chart illustrates, there was recovery by 2022.  Evidence for the United States parallels that of the United Kingdom and Canada with respect to employment as illustrated in Figure 3. Again, from 1900 to 2022 – ostensibly a period of great technological change – total employment in the United States expanded six-fold while the population grew four-fold.  

 

 

 

So, why all the anxiety about technological change?  Well, despite the historical evidence to date, there is a background foreboding that much like mutual fund returns, the past may not be an indicator of the future if the onset of quantum information technologies, AI and machine learning together somehow represent a fundamentally different economic process that unlike the past will destroy more jobs than it creates. However, at this point these new technologies are still in their infancy and there is really no reason at this stage to expect the future to be that much different than the past, unless the relationship between technological change and its contribution to the economy itself shifts in some unforeseen fashion.

 

Sources/References

 

Data Sources for Figures 1-3: UK [  Data Source: A millennium of macroeconomic data for the UK, The Bank of England's collection of historical macroeconomic and financial statistics, Volume 3.1.], Canada: [Denton and Ostry (1961); Historical Statistics of Canada; Statistics Canada Catalogue 71-201 Annual, 1973 & 1989, Historical labour force statistics, actual data, seasonal factors, seasonally adjusted data; Statistics Canada, v102029212 Canada [11124], Labour force (Persons), Total, all occupations; Both sexes v102029368 Canada [11124], Employment (Persons), Total, all occupations, Both sexes]; USA:[ Historical Statistics of the United States (HSUS) from 1900 to 1945 and that of the Bureau of Labour Statistics (BLS) from 1948 to 2023.]

 

Stewart. I., D. De, and A. Cole (2015) Technology and People: The great job-creating machine. Deloitte.

Inflation, Productivity and Real Wage Stagnation: Canada 1960 to 2023

 

Today’s CPI inflation numbers have many breathing a sigh of relief with the expectation that with inflation below 2 percent, more interest rate relief is on the way and Canadians can resume their high personal borrowing lifestyle.   Lost in the short-term euphoria and celebration of expected lower borrowing costs is the long term cost that inflation has had on our standard of living given the low productivity gains of the last five decades.  Nowhere is this more evident than when one takes a look at how real wages have performed over time.

 

Figure 1 plots the average annual monthly hourly Canadian manufacturing wage – nominal and real – for the period from 1960 to 2023.  The nominal hourly manufacturing wage data and the All-City CPI data are both from the US Federal Reserve of St. Louis data sets [CPALCY01CAA661N; LCEAMN01CAM189S] with the real hourly wage data in $2015.   Why manufacturing wages?  Well, the manufacturing sector has generally been held up as the beacon for good quality and high paying jobs with a lot of hand wringing as manufacturing jobs have declined as a share of employment.  It sounds old fashioned but many still regard manufacturing jobs as the “high ground” of an economy in terms of value added to which I would also add the resource sector (including agriculture).

 

 

 

When nominal hourly wages are examined, their performance looks impressive.  The monthly nominal manufacturing wage in Canada in 1960 averaged $1.78/hr. By 2023, it was $30.66/hr and the average annual growth rate of real nominal hourly wages in manufacturing was 4.7 percent.  However, when adjusted for inflation using the All-City CPI for Canada with 2015 as the base year, real nominal wages barely double over the period going from $13.64/hr to $24.91/hr.  The average annual growth rate of real hourly manufacturing wages over this entire period was only 1 percent annually.  Given that at 1 percent annual growth it would take approximately 72 years for a quantity to double, we can expect real hourly wages in manufacturing to be double those in 1960 by 2032.

 

 

 

Figure 2 plots the annual average growth rate of real hourly manufacturing wages and adds a 5^th order polynomial smoothing plot.  When one examines both Figure 1 and 2, it becomes apparent that the stagnation in real wage growth really sets in during the 1970s.  There was a brief uptick in real wage growth in the wake of the FTA and NAFTA (in 1988 and 1994 respectively) but decline sets in again after the 2008-09 financial crisis.  When one combines the productivity decline that starts in the 1970s following the first oil price shock with the effects of inflation, the erosion of the standard of living – as captured by real wages – is dramatically illustrated.  It makes the case for why bringing inflation under control is so important and also why we need a productivity agenda to drive Canadian policy going into the next election.

Harris or Trump? For Canada, Post November 4th Is Going to Be a Challenge

 

As we move into the final sprint of the US election, it bears as always to pay attention to the economic implications for Canada.  Whatever one’s political priors or favorites may be in this election, in the end it needs to be realized that when it comes to US trade and economic interests, it does not matter whether Trump or Harris wins– American interests trump (no pun intended) Canadian ones.  And in the case of the economy and our trade relationship with the United States, be prepared for some tough bargaining.  While 2024 marked the 30^th Anniversary of NAFTA, it has since 2020 been replaced by the USMCA or CUSMA agreement with renewal talks beginning in 2026. 

 

Along with perennial sticking points like milk and dairy or softwood lumber, in the United States, despite what economists and evidence might say about economic growth and the benefits of trade in the wake of NAFTA and CUSMA, the debate will be shaped by the widespread belief that NAFTA in particular resulted in job losses and wage stagnation.   In the case of manufacturing, the accompanying graphic summarizes quite nicely why the Americans are going to be playing hardball.  In many respects, US manufacturing job losses did coincide with NAFTA. 

 

Figure 1 presents annual Canadian and American manufacturing employment from 1976 to 2023 using a dual scale since US employment and population in general is about ten times ours.  In 1994, there were nearly 19 million Americans employed in manufacturing and 1.8 million in Canada. In the decade afterwards, by 2005, US manufacturing employment fell to 16.2 million while Canadian manufacturing grew to 2.2 million.  In the wake of NAFTA, American manufacturing employment fell by 14 percent while Canadian manufacturing employment rose by 20 percent. 

 

 

 

 Since 2005, American manufacturing employment has declined slightly to 15.6 million while Canada’s declined to about 1.8 million where it has stabilized somewhat.  In other words, over thirty years, Canada has stayed flat in terms of total manufacturing employment (notwithstanding the rise and fall from 1994 to about 2010) while the US has seen a decline.  The good news is that since about 2019, as evidenced by the 5^th order polynomial smoothing line, both countries have seen a slight increase in manufacturing employment as a result of fallout from the pandemic, trade issues with China and the rise of onshoring production activities.

 

Yet those same polynomial smooths show a pretty consistent decline for the US since 1976 with Canada doing somewhat better.  True, Canada is not to blame for the decline in US manufacturing.  Both countries have seen a decline in manufacturing employment over time both in absolute numbers as well as a share of total employment.  It is not 1960 anymore.  There have been productivity issues in both countries as well as intense competition starting in the 1990s from China and other Asian economies as well as Mexico which is/was a part of CUSMA/NAFTA.  However, that does not matter.  For the United States, creating jobs in manufacturing will mean looking at all the players – including Canada.  It will not matter whether Harris or trump becomes President in this regard.  Notice has been served.

Changing Thunder Bay City Council: What Is The Path Forward?

 

I have already opined on the recent report putting forth options for the reform of Thunder Bay City Council, but the matter is so fundamentally important to the future of local democracy in our city as well as the effectiveness of municipal government that it is worth another post.  This especially requires another post given that the current council will soon move to deliberate and decide on which of the two options – if any – it is going to go with. 

 

To start, there are actually three options: The first, is four east-west wards running parallel from north to south numbered 1 to 4 that basically gives each ward a rural area, urban area as well as some industry and waterfront in the geographic and population composition of the ward.  Each ward would have two ward councillors for a total of eight.  As well, there would be two at-large councillors and a mayor.  This proposal is being recommended by city administration.  The second option is a full at large system with ten at-large councillors – no ward councillors - plus the mayor. One suspects that some type of ward structure will be retained with at large councillors “assigned” ward representation duties but how that might actually work is shrouded in fog. And third – while not explicitly referred to but lurking in the background – is simply the status quo option of seven ward councillors, five at large councillors and one mayor.

 

What remains important in the discussion is what the rationale for creating a new structure is.  First and foremost, whether warranted or not, there appears to be a widespread desire for change in city council’s composition driven by dissatisfaction with assorted aspects of municipal government in Thunder Bay.  A desire for change for the sake of change is never in of itself a good reason to change things but even recent albeit unscientific polls on TBnewswatch suggests that most respondents want change.  Past TBnewswatch  polls have suggested a desire for a smaller council with eight councillors plus a mayor garnering the most support.  When given the two proposed options plus the status quo as choice, the most recent poll finds that over 86 percent want change with only 13 percent supporting the status quo.  Moreover, when it comes to the two proposed options – about 46 percent support the 10 councillors plus mayor all at large option, and 40 percent support the four ward two councillor per ward plus two at large and a mayor format. 

 

 

What type of performance improvements to our municipal governance are these proposals supposed to make?  Anecdotally, observation suggests that Thunder Bay City Council meetings appear to be long and drawn out with detailed discussions and gridlock on minor and major matters alike.  Yet, how changing the composition of city council will address issues of dysfunction are not really obvious.  There is also a desire to save money but reducing council by two members saves at best $100,000 in salaries, benefits, and expenses on a tax funded budget of over $200 million.  Indeed, if cost savings are really what you are after, the savings do not come from the reduction in the size of city council but an improvement in the quality of councillors and decision-making on it.  There is nothing in either of the two proposed options that lend any evidence as to how a smaller council will be a better council when it comes to decision making. 

 

In the end the three options can be interpreted as follows: the status quo, a modified status quo which shrinks the council slightly but still includes a mix of ward and at-large councillors, and the ten councillors at large option which I would term the “mayor plus ten assistant mayors” option.  The attractiveness of an all–at large council to some members of the general public springs from the conviction that such councillors have the interests of the entire city at large while ward councillors are parochial nimbyists who block change.  My observations are that ward councillors seem stuck with the grunt work of dealing with specific ward and neighborhood issues while most  at-large councillors pick and choose what ward issues to advocate for based on the political benefits while behaving as pontificating prima donnas at meetings with speeches that chew up far too much time. 

 

In the end, an all–at large council favours those candidates with a lot of prior popular name exposure (it is like high school all over again) and/or the financial resources or special interest backing to mount a city-wide campaign.  The argument that somehow only at large councillors have the ability to see the whole city’s interests is spurious.  The average municipal politician usually puts their political interests first, and everything else is weighed and slotted towards meeting those interests – whether it is in the city’s interests or not.  Having no ward councillors at all is fundamentally at odds with the purpose of municipal government which is to provide municipal services to rate payers who either reside and/or operate businesses in wards.  Moving to an all–at large system reduces accountability to the mundane needs of ratepayers by creating a council of big picture overlords who will pass the buck on specific local issues when it suits them. 

 

Those that want an all–at large system should consider this analogy.  If at large councillors are really better able to see the “bigger municipal picture” than by extension having all MPs and MPPS elected from an at large list would be better at seeing provincial or national interests as opposed to parochial MPs or MPPs who just look after their ridings.  Perhaps, one might consider this a straw man argument and a deflection from the real issue but think about it carefully.  One may feel comfortable about all at large candidates with no wards because Thunder Bay is a relatively small city and ultimately everyone knows someone who knows someone who is friends with a councillor.  Everyone feels that they can have their voice heard.  And yet, without the institution of direct ward or riding representation, there is no guarantee that you will always have a direct opportunity to be heard.

 

So, what is the solution?  Well, my preferred option – which is of course not up for discussion in the current set of offerings - is eight ward councillors plus a mayor option.  The city would be divided not into four but eight wards thereby ensuring that some dedicated rural ward representation would remain.  It is now 50 years after amalgamation and the city has been united long enough that the average ward councillor should be able to see the forest for the trees.  If they cannot, it is more a function of the quality of the councillor rather than the ward system.  Going down to eight councillors plus a mayor would likely save several hundred thousand dollars – again a small sum compared to a budget in the hundreds of millions – but enough to increase resources available to perhaps attract better candidates. This does not necessarily mean raising salaries for councillors but could even involve providing funds so that they can hire some independent research support so they can better inform themselves on issues.

 

Of course, the other argument that has also been made in favour of an all–at large council is that there is a lack of Indigenous representation on the current city council.  Given that the Indigenous population in Thunder Bay is spread out across the city rather than concentrated in one ward, an all–at large system increases the odds that an indigenous candidate and by extension the growing indigenous population can gain a voice on city council.  However, there is no guarantee that having ten at large councillors will ensure an indigenous candidate gaining office.  Indeed, if having at large councillors is the best way to ensure better odds for an indigenous candidate getting elected to Thunder Bay City Council, then the best thing to do is to retain the current system with seven ward and five at-large councillors plus a mayor.  At least it will be the devil we know best.