The idea in a nutshell : Informed analysis from industry experts suggest that around 20% of jobs can be automated by AI. Even if that’s only partly true, the effect on the share price of the companies which implement AI projects is likely to be significant. It may double the market capitalisation of many. It might makes sense for you to buy some shares.
Evidence in support of that claim
There are a lot, lot, lot of estimates of the impact of Artificial Intelligence on the economy. Mostly they predict the effect in terms of it’s impact on unemployment. Wild assed guesses vary and often suggest that AI will displace between 30% to 50% of the workforce.
There are some concerns over how holistically these reports have been considered. Without going too far in to areas covered by other articles, few of these reports name a time line for the impacts they postulate. Only one that I know of talks about the jobs created by AI, as well as those which we will shed.
As a result, the net impact of AI on the economy as a whole is hard to judge. (Although some critical elements do seem disturbingly clear).
If, however, for now, we limit ourselves to consideration of the impact of AI on the companies in the economy (as opposed to the economy as a whole) the impact artificial intelligence will likely have, particularly on share prices, becomes interesting.
McKinsey suggest that 20% of jobs can be automated (on average)
McKinsey’s key point in their report on the subject was that it’s more useful to talk about the proportion of jobs which will be automated than the number of jobs that will be displaced.
Overall, having studied some real life examples, McKinsey suggest that, on average, 20% of people’s work will be automated by 2020.
Following through the mathematical implications of that
If this is even partly true, economies around the world seem set to boom.
- Let’s use the lowest of all the numbers in all the reports. Let’s say that 20% of work is automated by 2020.
- Let’s say that the number is an exaggeration. It’s 50% out. AI only delivers a 10% productivity improvement between now and the end of the decade.
- If we do all of those things, that’s still a ( roughly ) additional 3% per year productivity gain between now and 2020. That’s huge.
The AFR describes a brief, historical, productivity gain of 2.5% as ‘spectacular’. The longer term productivity trend over the last 16 years is described as ‘persistently sluggish.’ It (productivity) often goes backwards over the period they reviewed.
As such, it seems reasonable to presume that a productivity improvement of 3% and which is sustained for 3 years or thereabouts would be equally ‘spectacular’.
If this is true, share prices could double
Share price growth is a multiple of companies’ earnings. Share values are factors of the price / earnings the P/E ratio of a company. Current P/E ratios on the Dow are between 12 & 22.
That means that if productivity goes up 1%, it would not be unreasonable to multiply that 1% by between 12 & 22 times before you apply it to the share price.
So, if we reduce the productivity gains from AI, in the way I did in the section above: Even if we say that the lowest estimate forecast of the improvement in productivity is twice what it should be and use the lowest P/E ratio to determine the share price benefit, share prices double in the next few years.
And that’s on average – what about those who own the platform ?
It is the latest trite phrase to talk about the platform as being where the money is. Air BNB are the biggest hotel in the world and they own no rooms! Uber are the biggest limo service and they own no cars! Everyone wants to be the platform.
Google, Facebook, Microsoft and others are all investing in AI tools and services, starting with bots for reasons I have speculated on separately. As ‘normal’ companies – of the sort I am suggesting will gain from this 20% improvement in productivity by 2020 – try to implement AI projects, they will be using the platforms that the tech giants have set up.
Each ‘normal’ company then gets the benefit of AI. Each of the share price of each of the platform owners seems likely to go bonkers.
This is about specific companies, not the whole economy
This is a blog. It has to contain an idea which is likely to be notionally right. However, I should concede :
There’s not a lot of evidence supporting all this: It’s early days with AI rollouts and decent reasonable stats on the economic effects of them. This is one McKinsey report I am extrapolating a lot from. We’re taking a bit of data and applying it to every company in it. That might be a reasonable assumption if AI affects every industry. But then it might not.
This is not about the whole economy, it is about some companies in the whole economy. For this to be true of the whole economy, the people displaced by AI would have to reinvest their time. Then we need to ask: Is the ‘new time’ as productive as their ‘old time’? The law of diminishing returns might disagree with that assumption. If 20% of people were displaced by AI, the overall economy might go backwards in terms of growth. That sort of change would about double the dependency ratio and would require some new tax measures.
It’s time to buy technology shares
McKinsey’s nuanced point about the proportion of a job which is replaced rather than the number of jobs replaced is both reassuring and insightful. But it’s way too early to tell if this forecast is true.
As the benefits of AI projects accrue to those organisations it seems likely that they will accentuate existing levels of income inequality. Some people will get very rich and some will be displaced in to poverty and potentially, national unemployment rates which could be disturbing for us all, not just those put out of work. We should hope decent tax legislation ameliorates the social effects of this.
As usual, the best thing to do is often to plan for the worst and hope for the best. I, for one, will be assuming the government do nothing helpful and personally buying a lot of shares in technology companies the next few years. The opportunity of AI impacting companies and tech companies in particular seems to put the balance of risk / reward in favour of companies.