AI automation and consequential job loss may require taxing robots

The fourth industrial revolution brings a metamorphosis in the labor market, as the industrial revolution did more than a century ago. An era in which digital skills are critical to surf the wave could put potentially at risk up to 375 million workers who will lose their jobs to automation by 2030 across the globe. The idea of a Universal Basic Income (UBI) to level the playing field for those in need is not new and was highly discussed especially during the Covid-19 pandemic. But the key question is where do we find the money to fuel UBI? So, do taxing tech giants could fuel universal basic income to face job loss due to automation?

In the thirties Jon Maynard Keynes predicted that by 2030 people would work only 15 hours a week, due to technological unemployment, having more time to pursue leisure and other creative endeavors. Keynes was right when he predicted economic progress and growth, but he failed to predict the behavior of future labor markets. Remember he was lucky enough to see the changes the industrial revolution brought in productivity gains at the turn of the XIX century, so it was easy to imagine a future in which humans could work less and live better.

Today’s landscape looks more gloomy, since predicted job losses to robotization especially of low-skilled occupations will have a deep social impact in a world in which 659 million people still live under $2.15 dollars a day and 3.1 billion people cannot afford a healthy diet. Robots could replace 40% of the world’s jobs within 15 to 25 years. What does this mean to the low-skilled workforce? To those that are already in disadvantage when it comes to STEM fields such as women or people living in the global south? Will AI automation push families into poverty.?

Robotization, job loss and poverty

Accenture estimates that 40% of all working hours could be impacted by Large Language Models (LLMs) such as GPT-4, this could both mean that some activities will be replaced and others will remain, thus automation will have a diverse impact depending on the type of occupation. What’s more 80% of today’s jobs are likely to be affected by generative AI, impacting fundamentally their core tasks.

For instance, many clerical or secretarial jobs will decline quickly because of AI such as bank tellers, postal service clerks and data entry clerks, but it predicts a surge of 40% in the demand of machine learning specialists, 30% to 35% increase for roles such as data analysts and scientists, and 31% for information security analysts.

A recent survey conducted by the World Economic Forum to more than 800 companies around the globe estimates a net decrease of jobs of 14 million or 2% of current employment, this means a job growth of 69 million jobs and a decrease of 83 million jobs.

What is clear is that low-skilled populations, workers in clerical and secretarial roles – mainly women- and workers out of the technological fields are at disadvantage. For example, women make up just 16% of tenure-track faculty focused on AI globally, and only 26% of data and AI positions in the workforce. What’s more, women are 40% less likely to adopt new tech, while junior men are more likely to adopt Gen AI tools at work compared to women (69% vs. 48%).

The Global South undoubtedly lags behind developed economies when it comes to digital. Sub-Saharan Africa still has a limited internet penetration of 36% as well as South Asia with 43%, compared to a global internet penetration rate of 63%. The Oxford Insights Assessment of 181 countries found that the less prepared to use AI in public services are Sub-Saharan Africa, some Central and South Asian countries and some Latin American Countries.

The Network Readiness Index 2023 evaluated 134 economies based on factors related to their readiness to harness the benefits of the digital revolution – including technology, people, governance and impact -. The United States and Singapore ranked 1 and 2 respectively, while the last five positions were scored by Burundi, Chad, DRC, Mauritania and Mozambique, all except for Mauritania in Sub-Saharan Africa. ?

A research funded by OpenAI explored the impact of Large Language Models (LLMs) on the US labor market and found that the most impacted will be workers with Bachelor’s degrees, workers with Master’s degrees will also be impacted but less and even less those with high-school degrees. The report suggests that around 80% of the U.S. workforce could have at least 10% of their work tasks affected by the introduction of LLMs, while approximately 19% of workers may see at least 50% of their tasks impacted”.

A study quantifies the extent at which automation has contributed to income inequality in the US, which provides an idea of the correlation between automation, job loss, poverty and inequality. The study found that between 50% and 70% of the overall changes in the U.S. wage structure over the last four decades are driven by automation. They also estimated that automation has reduced the wages of men and women without a high school degree by 8.8% and 2.3% respectively.

Another research found that workplace automation significantly increases household energy poverty – and the positive relationship is very robust-, because it reduces people's income and work-related social capital, thus exposing households to higher risks of energy poverty. They also found that consequences are more prominent for rural households, less educated people, migrants, those without labor contracts, non trade-union members, and out-of-system workers. Have in mind that today 760 million people still live without electricity globally, while 2.3 billion people lack access to clean fuels and modern technologies for cooking, so potentially AI automation could drive millions of families into poverty.

Therefore, the more digital an economy becomes, the more risk of job loss to automation and the more likelihood of pushing vulnerable families into poverty – migrants, women-headed households, indigenous families, etc. ?

What is crystal clear from this picture is that some jobs will be lost and new jobs will be created to AI automation, and as macroeconomic principles have taught, net job destruction leads to increased unemployment, less income, less consumption, increased poverty and in turn less job creation. Spiraling into a vicious cycle.?

So, policy choices must be made now to ensure that in the future when AI automation widely spreads all over industries and sectors job transitions run smoothly and in cases in which this won’t be possible, social protection systems ensure access to a living wage and dignified quality of life for all. ?

Now let’s explore a little bit more the notion of social protection and Universal Basic Income or UBI.

Universal social protection is strongly needed in the digital age

Social protection systems help people in poverty and vulnerability to cope with crisis, shocks, find jobs, or access to health and education. Usually, these systems provide benefits to the poor and vulnerable and to individuals on the basis of risks faced across the lifecycle such as unemployment, sickness, death, disability, maternity, etc.

Universal social protection systems enshrine the notion of UBI, which are ultimately social protection floors “for all”, regardless of sex, age, race, location, ethnical background or religion. These systems require funding that either comes from social security contributions from wages or through general taxation. Today only 46.9% of people have access to some form of social protection benefits worldwide, which leaves a high proportion of people vulnerable to all kinds of shocks, especially economic ones.

AI automation will require important job transitions within the next few years, including skilling, upskilling and re-skilling efforts across industries. We already know that some occupations will be taken by robots, especially those related to repetitive tasks such as data entry clerks, also those needing vast amounts of information to be processed such as legal advisers, or even those that require more creativity such as journalists or marketers. If human capital development and investments in education is weak, we will inevitably see large numbers of unemployed people at risk of becoming poor, putting pressure to security conditions and potentially eroding democracy.

Keynes predictions in the thirties envisioned a future in which fewer working hours would mean more time for leisure and creative endeavors. This is only possible if all basic needs are met first – including income, housing, food, clothing, health, education, energy and water. Here’s when the concept of Universal Basic Income comes into play to ensure a social protection floor for everyone, regardless of any social, political or economic condition. In order to live happy, full and healthy lives people need to make choices and access to opportunity, this is the core concept of human development.

Being or not being unemployed must be a human choice, not an inevitable consequence of digitalization.

The ILO estimates that only 3.3% of annual GDP is needed to establish social protection floors in developing countries, which is feasible even with restrictive fiscal spaces. However the financing gap in low-income countries starkly reaches 52.3% of GDP annually.

Universal social protection, then, would require 10.6% additional spending in low- and middle-income countries of their annual government expenditure. But this task is quite tricky for low-income countries because the required extra spending should be 28 times their social protection expenditure.

A UBI could be the answer to job loss risks brought by AI automation, especially for countries in the Global South, either if unemployment is a reality by choice or by necessity. But finding the fiscal space needed to make it real is key and here’s where new sources of taxing come into play.

?

Why not a UBI funded by robots?

Today big technology companies have access virtually to every household and mobile device in the world, generating profits from operations across the globe even when their offices are concentrated in few locations. In fact, PwC estimates that the digital economy will contribute up to 25% of global GDP by 2025. The global ranking of the companies above the one trillion dollars market capitalization is in the hands of only six big technology companies namely ?Apple ($3.5T), Microsoft ($3.4T), Nvidia ($3.2T), Google ($2.3T), Amazon ($2.0T) and Meta ($1.3T).

But where do these companies pay taxes?

Tax avoidance and profit shifting of digital services is a common practice that is eroding both taxing and welfare systems. The OECD estimates $100-240 billion dollars in lost tax revenue annually, which is equivalent to 4-10% of global corporate tax income revenue. The transition towards inclusive international taxing systems requires global consensus and measures to secure that these companies pay fair taxes everywhere they operate. But this has multiple caveats.

First, multiple jurisdictions can impose a tax claim on the income earned by a multinational corporation, so the key question is how to attribute taxable profits of digital multinational enterprises to different jurisdictions?

Second, under international tax rules, primary taxing rights are allocated to source countries. So, there’s a fierce country competition to attract companies based on fiscal incentives.

Third, another key principle of taxation is avoiding double taxation.

And fourth, corporate taxing practices are diverse and strategies to avoid high taxing rates are common such as base erosion and profit shifting. For instance, companies may invest in real estate in one country and shift profits to a different country with a reduced tax rate.

During the Covid-19 pandemic many global discussions took place to find the extra budget to fund the recovery efforts. As a result, the EU proposed a digital levy, that was supposed to be implemented in January 2023. Digital service taxes became widespread around the globe, especially in Europe and Africa with different national taxing systems put in place. Digital service taxes are gross revenue taxes that mix gross receipts and transactions such as the sale of advertising space, provision of digital intermediary services or the sale of data collected from users. These are not income taxes or online sales taxes.

The OECD made a huge effort to reach consensus to standardize international taxation via the OECD/G20 Inclusive Framework on Base Erosion and Profit Sharing, so they asked for a halt of national digital service taxing initiatives. The OECD estimates new digital service tax revenues totaling about $100 billion a year, which is about 4% of global corporate income tax revenues. The model proposes two pillars. Pillar one, establishes that for a multinational corporation to be subject to tax in a market jurisdiction, it must have a nexus with that jurisdiction, based on a fixed market revenue – threshold of 1 million euros, and a 12-month nexus revenue threshold. And pillar two, establishes two mechanisms to ensure multinational corporations pay a 15% minimum tax regardless of where they’re headquartered or the jurisdictions in which they operate.

There’s still a long way ahead to reach the consensus needed to implement OECD’s inclusive framework for taxing, while humanity is facing a structural transformation of its labor market, that could yield great benefits but also disastrous impacts.

What is crystal clear is that “we need to tax robots” or let’s say the owners of those AI-fueled automated systems that will propel both huge occupational transitions and rise in unemployment.

Inclusive international taxing policies coupled with enhanced co-responsibility from technology companies are needed to find the fiscal space required by universal social protection floors in the digital age. States cannot do it by themselves. In fact, the value big tech companies generate is even bigger than most economies. For example, Apple’s market cap is larger than 96% of country GDPs, including in this list Brazil, Italy, Canada and Russia. Only seven countries in the world have a higher GDP. While Microsoft’s value is larger than the GDP of Brazil, Canada, Russia and South Korea.

Now more than ever big technology companies need to step forward and recognize the moral obligation to find avenues to solve the social problems they’re triggering, so in turn economies in which they operate can keep flourishing.? ?

?

?

?