Don’t tax productivity

In the late seventeenth century, England’s King William III tried to fix his budget by taxing windows.  It wasn’t completely illogical, wealthier citizens had bigger homes with more openings.  But not surprisingly, this led to dark rooms as people made the obvious decision to board up their windows.  England was, quite literally, taxing the sun.

We can laugh, but today there are similar and serious discussions going on around the world about taxing AI, risking companies boarding up the very productivity we now urgently need.  As governments grapple with budget deficits, they are worrying about jobs being displaced by technology, particularly AI, and are pre-emptively looking to new sources of income.

Last year in “Demographics is not destiny”, I argued that a housing crisis was the unexpected consequence of poor productivity growth combined with an ageing population.  A failure to embrace automation in a misguided attempt to protect jobs could add new regressive taxes to the mix as well.

Every politician knows that you get elected by creating jobs.  But they are worrying about the wrong problem.  Over the next 25 years, the share of the global population in the workforce is projected to fall from 64% to 57%.  To simply maintain today’s standard of living, productivity will need to rise by 12% just to offset this demographic shift.  That’s before we even consider the additional gains required to fuel economic growth that we’ve all come to expect.

In Japan and Germany, the demographic shift is already well underway.  Germany faces the same global 2050 trend in as little as 10 to 15 years while Japan will need to lift productivity by even more in the same timeframe just to stay still.

Countries like the US and Australia are sidestepping their demographic issues through immigration.  This is often hiding per capita declines in economic growth.  Worse, the most productive immigration (that is bringing in the best skills) comes at the expense of the source countries that are giving up these highly sought-after workers.  Immigration is a great thing, as former US Secretary of State, Dr Condoleezza Rice, has said “Globalisation is a fact, not a policy”.  But any country that relies on it to cover a failure to prepare for the future is setting itself up for future intergenerational harm.

Politicians have to stop worrying about the number of jobs and worry about their quality instead.  In a digital world, work falls into three types: for the machines (such as delivery drivers dispatched by apps), with the machines (such as clinicians advised by AI) and on the machines (including programmers, engineers and anyone training the algorithms).

Job satisfaction, financial rewards and productivity are all aligned with moving up this value chain, working “on” the machines rather than “for” them.  AI is actually making it easier to make this shift, with low and no code solutions becoming readily available to codify insights and strategic direction in almost every field.  And in a world of constant change, skills can’t be front-loaded in early life, they must be updated continuously through lifelong learning.

This is a major shift in the way we educate our population.  A decade ago I wrote about the disruption of universities.  What was important then is critical now as we have to adapt to new ways of working shaped by technology and driven by productivity and demographic pressures.

Which brings us back to incentives and taxation.  Far from creating unemployment, AI and technology-enabled automation mitigates our global demographic crisis.  That’s why we should resist calls from the EU parliament, Bill Gates, or Anthropic CEO Dario Amodei, for taxes on robots or an AI token tax.

We don’t need new taxes, we need smarter use of the existing ones.  A progressive income tax system can capture rising wages from more productive workers, maintain government revenue, and still reward employers who invest in automation.  Governments can maintain income tax revenue if the reduced pool of workers is each earning more.  Individuals are motivated to maintain their skills if they can see the rewards flow through to their hip pockets.  And employers are still benefiting in this balanced future from reduced labour bills as a result of automation.

This ideal future is achievable, but it demands smart policy in a time of dizzying technology change.  In fact, change doesn’t happen at a constant pace, it comes in waves.  During surges governments need to hold on for the ride and use periods of consolidation to recalibrate, finding the balance between tax revenue, rewarding workers and shareholder returns.  Let’s hope today’s policymakers show more wisdom than a king who taxed the windows and darkened his nation’s future.

2 Replies to “Don’t tax productivity”

  1. This is a disingenuous article.

    It’s right that immigration to replace capital is a false move but the characterisation of digital work is essentially reductive and wrong ( it’s more diffuse than that) but to then claim that as we move up that hierarchy of IT work value chain, which is underpinned by AI, is fallacious.

    This is a standard consulting ruse to offer a solution that suits the interests of the clients: Hey, we understand the future and work and AI is gonna be good for you.

    So, as the piece moves on, we have to adapt and we will be happier and richer. Yes, well capitalism taught us that but where is the detail?

    The answer is vacuous: Because AI fills in the productivity gap we can’t tax it. But AI has had a tax free ride e.g. network externalities and the environment for too long. This means that capital earns more so the tax is actually on society, leaving govt with a shortfall.

    The argument hinges on this : ”tax system can capture rising wages from more productive workers”. The evidence of the last 30 years is that workers are more productive but have had a lower share of the gains than capital. And tax experts may find racket creep comes in there too which negates the proposed effect.

    At best, this argument is sophistry, or casuistry.

  2. This is a disingenuous post which is really sophistry.

    The characterization of digital work which is essentially wrong but is made worse with the claim that as labour move up the IT value chain underpinned by AI, so AI is making things better for us.

    This is the hack consulting fallacy to offer a solution that suits the interests of the client: hey, AI is gonna be good for you.

    The answer is hopeless: Because AI fills in the productivity gap we can’t tax it. But AI has had a tax free ride for too long e.g. network externalities and the environment are clear examples. This means that capital earns more so the tax is actually on society, leaving govt with a shortfall.

    His argument hinges on this : tax system can capture rising wages from more productive workers. The empirical evidence of the last 30 years is that workers are more productive but have had lower share of the gains than capital. The argument ed tended here is that capital should not be taxed when it will earn super profits and not pay for the wider costs.

    Need to do better than that.

Leave a Reply to Harry Shand Cancel reply

Your email address will not be published. Required fields are marked *