Our latest extensive trip across the US (California, Arizona, Illinois, Washington, New York, Florida, Washington DC) highlighted the strength of the domestic labour market and the resulting resilience of the consumer as evidenced by busy restaurants, service-related shops, construction activity and more. Everyone who wants to work (and is largely employable) has a job. As previously discussed, the greatest segment of market “tightness” (i.e. where employers are having the greatest difficulty in hiring workers) is in the low skilled part of the labour market. This is illustrated in the chart below.
Why is the low skilled end of the market in such demand?
A fully automated process occurs when technology components are linked together to create a product or a service that has no direct human intervention. Automation is driving organisational efficiencies in many areas of task repeatability and is able to replace mid-skilled workers. However, process automation is not currently sufficiently agile and customise-able to compete with the range of contextual-task provided by low skilled workers. The competitive advantage that low skilled workers offer is that they perform a wide range of tasks that are highly specific to the context of the task before them. Examples of this segment include gardeners, cleaners, painters, construction and maintenance workers and restaurant hands.
The tightness of the low skilled labour market differs from prior cycles. The 1990s cycle was about globalisation and outsourcing, and the rise of the internet played a big role in both of these themes. These momentum drivers created the strongest employment growth in the mid and high skilled areas. The 2000s were a continuation of the globalisation and technology theme, and the US housing and global finance boom also came into play – collectively these were a bonanza for high skilled roles. This cycle has seen a hallowing out of mid-tier roles. This has been most pronounced in finance and insurance industries where large scale deleveraging has occurred/is continuing, but is happening in varying degrees across all industries.
Looking forward, we expect the trend of the past decade to continue and mid skilled jobs to be eliminated at a much faster rate. This includes all professional areas where the knowledge of the whole profession is being comprehensively coded. The quality of the decisions made by inter-connected system can therefore be shown to outperform human workers from an effectiveness, consistency and cost perspective. Some examples where the entire profession is being comprehensively coded includes law, medicine, agriculture, finance, banking, insurance and accountancy. Particular tasks in these professions that are rapidly being automated include consumer enquiry via chat-bots, decision making, compliance activity, report generation and even more traditional collaboration activity where contextually-aware automated work-flows are able to piece together the set of tasks across the value chain. This will continue to provide opportunities for knowledge creation professions but will significantly decrease demand for implementation/execution-based expertise.
For now though, the low skilled segment of the labour market will be a key support for consumption and more broadly the US economy as we head into 2020. Wages for these workers is grinding higher (also evidenced on the ground first hand) and importantly the recipients of these increased wages spend each incremental dollar. Should this segment of the labour market start to weaken then this would be of concern and a potential de-risk signal – but importantly for now we are not about to de-risk.
Whilst I agree with your assessment, Moore's law also applies to AI and robotics. Moore's law is exponential and will accelerate unemployment in all categories sooner than we think - e.g. the iphone is only 12 years old. Mass unemployment will mean falling human income tax revenues and force governments to tax robot income generating tasks and functions. The prices of goods and services will fall dramatically as the marginal cost of production approaches zero. All else being equal (capital, taxation, energy, raw material access, etc.) an AI robot can produce goods and services at the same price regardless of location. This will have global consequences as countries with low wages will no longer be the default for labour intensive industries. China may well lose her status as the world's factory as comparative advantages shift and decentralisation occurs. We live in interesting times.