Is tech change really accelerating?

Accelerating or decelerating? Just about all the media commentary about the future of work says that tech change is accelerating, like this example, originally published in the Listener. But what does the data tell us?
It’s now six years since Frey & Osborne’s claim that 47% of total US employment was at risk of automation over the following 10–20 years. Replicating their methodology for New Zealand, the NZIER and Chartered Accountants Australia and New Zealand estimated that 46% of New Zealand jobs were at risk. If they were right, we’d expect to see some strong signs of job displacement in the data by now. So, I went looking… and didn’t find much. Or more accurately, what I did find is directly contradictory to a “technological disruption of the labour market” story.
There’s no single measure of the rate of tech change1 but if that rate is accelerating, we’d see the impact on firms, the economy and the labour market. Acceleration would show up in many statistics, including:
- Job displacement – higher rates of job displacement and more frequent exits from the labour market
- Job churn – higher rates of both job creation and destruction
- Job switching – more people changing jobs
- Firm dynamism – higher rates of firm creation and destruction
- Productivity – higher rates of productivity growth
We might also expect to see changes in types of work arrangements, especially if tech change was changing business models and ways of working. Many have forecast an increase in casual, part-time, short-term and other “precarious” forms of employment.
So here are a few charts that might surprise you:
Involuntary unemployment as a share of the labour force 2
Job displacement rates in New Zealand have been cyclical, and relatively low since the late 1990s. Over the last decade (ie, since the 2008 recession), displacement rates have declined.
Annual job creation and job destruction as a share of total jobs 3
Rates of job destruction have actually declined in New Zealand since 2000. Recent rates of job creation are at a similar level to rates in the early 2000s. (In Australia, job creation has been declining.)
Share of New Zealand workers who switched jobs, 2000–17 4
The share of New Zealand workers who switch jobs is at similar levels to the early 2000s – it decreased after the 2008 recession but has since increased.
Firm creation and destruction as a percentage of total firms 5
The rate of firm creation was generally declining between the early 2000s and 2013, although it has increased slightly since. Firm destruction has been relatively stable, sitting around 6%. (Rates of firm creation in Australia and the US have also been declining.)
Labour productivity growth in OECD countries 6
Labour productivity growth rates in many developed countries have fallen to very low levels.
As for changes in work arrangements, the share of the labour force in casual or fixed term work has not been increasing in New Zealand and there does not appear to have been an increase in contract-based work for main jobs. It is possible that more people are choosing to do gig work as a second job – New Zealand doesn’t have great data on second jobs, though we do know that the proportion of workers who hold multiple jobs has been declining.
What to make of all of this? My take outs:
- Recent trends show that people are not being displaced from work more frequently than previous decades, and workers are staying in jobs for longer, not shorter, periods.
- Slowing productivity growth, job churn, and firm creation and destruction, all point to declining dynamism in New Zealand and other countries.
- Based on these indicators, tech change does not appear to be accelerating, and the data is more consistent with decelerating change.
Does that mean we won’t see disruptive technological change in the future? Well, as John MacCormick will explain in his upcoming romance-themed post, trends can change! And based on what I’ve been reading in Dave’s blog posts job-loss predictions, technology predictions, AI tech forecasts, diverging forecasts and the perils of straight-line forecasting, I wouldn’t rush to make predictions. But it would be sensible to keep monitoring changes in these indicators in New Zealand, and in countries that develop and adopt tech more quickly than us.
Notes
1. In this post, I’m using “tech change” as a shorthand for the labour market effects of new technologies, particularly automation technologies.
2. Based on Stats NZ Household Labour Force Survey data.
3. Based on Longitudinal Employer-Employee Database (LEED) data.
4. Coleman and Zheng (forthcoming) and based on micro data from LEED. The share of workers who switch jobs is measured by the percentage of workers who changed employers within a 12-month period (data is for March of each year).
5. Based on Longitudinal Business Database (LBD) data from the spreadsheet of Stephenson (2019).
6. Based on OECD data.
Have your say
Comments
Ben Reid 17/09/2019, 5:15 pm (4 years ago)
Interesting article, ties in with much of our thinking at the AI Forum. However, unfortunately your appropriation of the term "tech change" to refer just to labour market effects is extremely misleading when used in the title. Technological change itself is by many measures clearly accelerating (ref. Kurzweil) - the article should really be renamed "Is tech-driven labour market change really accelerating?"
Simon Smelt 18/08/2019, 9:18 am (4 years ago)
Technological leaps may not map onto economic impact neatly. The pace of techno change may fall as it is digested and economic impact grows. Look at aviation:
1901 Wright brothers
1908 Bleriot crosses channel
1919 First cross Atlantic flight
1933 US coast to coast commercial flight
1939 Regular trans-Atlantic commercial flights
1969 Boeing 747 and Concorde prototypes
1970s on. Big tech leaps lacking as technology is refined. Massification of air travel (passenger and freight) has growing economic and employment impact.
Essentially the commercial age followed the pioneer age. Tech change was for a period more rapid than expected, comercialisation less rapid than expected. Then it switched.
anthony gibbons 16/08/2019, 12:24 pm (4 years ago)
Great piece. Frustratingly confounding
No one has commented on this page yet.
RSS feed for comments on this page | RSS feed for all comments