Wheat vs wages – AI’s impact on land and labour

The value of a year's labour rocketed from being worth 11 tonnes of wheat in 1900 to 200 tonnes in recent years. Might AI now begin to reverse that trend?

My July Craigmore Commentary updated some research tracking relative prices of lambs and laptops. Due to the China Shock and robotics, manufacturing outputs repriced downward. As a result, prices of scarcer farm produce rose, when measured in laptops.

This Commentary moves on to a possible new "AI Shock", exploring how AI may reshape economies, and asking what this may mean for farmland?

Experts' thoughts on AI

Some experts, such as the IMF, McKinsey and Goldman are bullish on AI. For example, Goldman predicts that global GDP could be 7% higher in 10 years as a result of AI, and McKinsey suggests manufacturing systems are already seeing significant gains.

Meanwhile the IMF predicts a major positive productivity shock in the service sector. Their graph below estimates that many professionals will benefit from the tools (pink bars), but that roles of eg. clerical support workers are threatened (red bars). Agricultural labour has a low exposure to AI (blue bar).

Employment share by exposure and complementarity

Source: UK Labour Force Survey, and IMF staff calculations

Daron Acemoglu, a Nobel prize winning economist from MIT, is less impressed by AI. He predicts that new tools will boost production by just 1% over ten years (0.1% p.a.). However, even he agrees with the bulls who predict that AI may lead firms to use less labour.

It seems fair to conclude a wide range of AI experts, bulls and sceptics, agree that AI will reduce demand for labour.

A shock theory

Might AI be about to write a chapter in an economic history of shocks that might run as follows:

1. First a 125 year "green revolution" saw agricultural employment shrink from 60% of workforces to now less than 2% in developed countries, while manufacturing and services sectors took up the slack,¹
2. Then a shock from China and robotics hit manufacturing, lowering relative prices of manufactures such as laptops, but leaving services not just unscathed but as a much higher proportion of the workforce,
3. Now AI seems likely to boost the productivity of both workers and professionals to the extent that firms may hire less labour (perhaps already visible in a reduction in graduate places offered by many of the large accounting and law firms). In which case might the relative value of labour services fall?

If the third "AI Shock" stage of this theory unfolds, this may reverse a remarkable trend in the opposite direction. While the average UK salary earned the value of only 11 tonnes of wheat in 1900, this had risen to 200 tonnes in recent years (see graph below).

Tonnes of wheat an average UK salary could purchase per year

Source: ONS, Clark (2003) "The Price History of English Agriculture", AHDB

A summary of AI's impact on agriculture

There are three ways in which AI may impact farmland:

1. Machine intelligence may increase the efficiency of farm operations ie. reduce costs of production. However, production levels of foodstuffs are unlikely to greatly increase as they are constrained by biological and physical limits,
2. If the predictions of McKinsey and the IMF are right, AI will lead to large productivity gains in the manufacturing and services sectors. A less dramatic AI productivity shock in agriculture, vs services, may finally reverse this trend ie. enhance agriculture's terms of trade (prices of wheat and lambs may rise when measured in wages, etc),
3. If AI creates unemployment, then resulting attempts by governments to use spending to address the problem could cause financial instability and/or inflation, as governments and central banks spend money, which they do not have.² As noted in my July piece inflation has traditionally been agriculture's friend.

Is AI already impacting labour markets?

The ratio between average prices of lambs and of salaries in New Zealand has been stable at around 500 lambs for a year's work over the past 25 years. Similarly in the UK salaries have stabilised at about 200 tonnes of wheat since 2000 (see graph above).

Does this suggest a century of out-performance of labour vs agricultural commodities may now be at turning point? If the analysis in this Commentary is correct, then in 10 years it may no longer cost the value of 200 tonnes of wheat to employ the average UK worker.

<sup>1. the value of land capitalised these productivity gains so, even though terms of trade of each tonne of produce fell, overall and prices continued to outperform inflation.
2. some turmoil off the back of AI should not be a surprise. It happens after other past positive productivity shocks. For example, the railroads, motor cars, internet and financials innovation led to the booms crashes and government interventions associated with substantial market correction in 1873, 1983, 1929, 2000 and 2009. Once again farmland, like gold, could be a defensive way to be invested during such turmoil.</sup>

Watch this space and keep in touch. If you have any questions about the report or any other farming matter, please contact us at the email addresses below.

Forbes Elworthy Founder [email protected]

Che Charteris CEO [email protected]

Nick Tapp Chairman CS LLP [email protected]