Last week I was invited by the South African Revenue Service (Sars) to give a talk on the fourth industrial revolution and taxation in preparation for the Brazil, Russia, India, China and South Africa (BRICS) Summit. The fourth industrial revolution, which is powered by artificial intelligence (AI) will reduce the world of work and thus potentially reduce the amount of collectible tax.
AI is a computer technique that allows us to bring certain aspects of human intelligence into machines. According to the World Economic Forum, 70% of the jobs that exist today will not exist in 30 years. Any task that typically does not involve more than one minute of thinking to complete, will be automated.
This era of the fourth industrial revolution will be a post-work era because the need for humans in the workforce will be curtailed. It will change the face of labour. AI robots will populate factories and cannot belong to a trade union.
Our participation in this revolution will not be optional. Either we participate or as a country we are economically obliterated to the “dustbin of history”.
The social consequence of the fourth industrial revolution will be extensive. Those with financial capital will simply buy these AI robots and produce goods and services to maximise profit.
So the concept of the poor getting poorer and the rich getting richer will be exacerbated.
The Gini coefficient, which is a measure of inequality in society, will increase and this will threaten the existence of the notion of the nation state.
Now if people are going to be put out of their jobs, who will buy these goods that these robots will produce? Where will these customers get the money to buy these goods?
Leading capitalists have proposed a solution. Bill Gates the founder of Microsoft, proposes the taxation of AI robots. South African-born technology billionaire Elon Musk proposes introducing the universal basic income. Perhaps the first port of call is to understand the impact of these revolutionary changes on taxation. Tax collection is the best way in which government generates its revenue. Prior to the final audit, in South Africa the tax collected in the year 2016/2017 was R1216 trillion. This tax is used to run the state, to pay civil servants, to run our public schools and hospitals to maintain our roads as well as to pay social grants.
These are public goods as they are our investment into stability and prosperity.
If our tax collection is lower than our expenditure then we run a budget deficit. A persisting deficit in a country ultimately renders the government bankrupt.
The big question that needs to be answered is what will intelligent automation do to tax collection?
Suppose the auto-manufacturing plant in the City of Tshwane has 100 people who each contribute R5000 per month to Sars as tax. If this plant decides to automate the jobs that these 100 employees are doing then the tax revenue for Sars will drop by R500000. The only way this tax revenue will not drop is if Sars increases corporate tax for this company by R500000.
Corporate tax is the tax that is charged to companies. The concept of increasing corporate tax because of automation is effectively taxing AI robots. Can we realistically automate an entire economy?
To answer this question we should study the work of Nobel laureate Arthur Lewis from St Lucia in the Caribbean islands who became the first person of African descent to win the Nobel Prize in Economics.
He proposed the Two Sector Economy theory. He said by moving labour from the agricultural sector to the industrial sector and using the resulting profits to expand industrial production, a country can be moved from a developing to a developed economy.
He proposed that this can be done until it is no longer economical to move labour from agricultural to industrial sector and this point is called the Lewis turning point. After the Lewis turning point has been reached the cost of labour in that economy starts rising.
In our book Artificial Intelligence and Economic Theory: Skynet in the Market, we apply Lewis’s theory to automation and define the limits of automation. We divide the economy into two parts and these are the AI machine and the human economy. Labour will move from the human economy to the AI machine economy until it no longer makes economic sense to migrate labour from humans to AI machines. This might be because the tasks involved are too complicated or too human to be automated.
For example, if one needs to see a doctor perhaps this patient-doctor relationship is too human for people to prefer machine doctors over human doctors. So the limit of automation is a point at which it is more expensive to automate a job than to use humans mainly due to the difficulty of deploying technology.
Going back to tax collection, in 1974, American economist Arthur Laffer proposed a theory showing the relationship between the effective tax rate and the tax revenue collected by government. These ideas popularised by Arthur Laffer were not new and were described as early as the 14th century by Arab scholar Ibn Khaldun.
What does our government do to ensure that we collect enough taxes?
First, we should have dynamic and world-class economic scientists who are sufficiently skilled in modelling economic, social and political phenomena.
Second, we need to train people who understand automation and artificial intelligence.
Third, we should use AI robots to detect and prevent tax evasion.
Fourth, we should develop a framework to tax companies that are domiciled overseas but make their money in South Africa.
In conclusion, the fourth industrial revolution should not lead to a drop in the standard of living. Aristotle foretold this when he said: “The end of labour is to the gain of leisure”
To ensure that this is realised we should ensure that we manage the delicate balance amongst labour, automation and taxation.
* Marwala is the vice-chancellor and principal of the University of Johannesburg and the author of the book Economic Modelling Using Computational Intelligence. He writes in his personal capacity.