# The case against Universal Basic Income

Recently, Universal Basic Income has received a lot of publicity and the general belief is that it’s the solution to technological unemployment. While I agree that technological unemployment is an issue that must be dealt with I believe that a Negative Income Tax is a much more intelligent partial solution. I shall start by presenting the fundamental problems with Universal Basic Income before introducing the advantages of a Negative Income Tax.

Three fundamental problems with Universal Basic Income:

1. UBI pilot studies are fundamentally flawed:

• Let’s define the variable Y, the actual change in behaviour under UBI
• Pilot studies measure a proxy variable X which is measurable in a setting where all recipients know that these handouts will only last a fixed amount of time
• The deviation between X and Y may be arbitrarily large as the handouts received during finite-time pilot studies are treated as capital investments whereas in the UBI scenario, these handouts are treated as a common good whose marginal utility diminishes with time.

For the above reasons, there’s no reason to believe that a person receiving regular cash benefits will behave in a particularly enterprising manner.

2. UBI is based on the assumption that a flat rate is a good idea. However, if we assume any fixed amount for UBI, the population total will be prohibitively expensive. Consider the UK population, and a basic monthly rate of 1000 pounds per month issued 10 months a year:

\begin{aligned} 6.10^7.10^3.10=6.10^{11} \end{aligned}

That’s .6 trillion pounds per annum, more than 15 times the UK military budget!

3. The impact of automation won’t be uniformly distributed. For this reason, the ‘universal’ assumption doesn’t make sense.

I’ve spent my last two summers in robotics labs in the UK and Hong Kong and from these experiences I’ve learned a few things. First, the task of defining a reward function for a robot on a continuous action space is highly nontrivial so we won’t have any fully autonomous robots anytime soon. Second, concept learning is equally hard so jobs that involve coming up with new concepts(ex. R&D, product development) won’t disappear anytime soon. However, anything that is procedural or mainly requires probabilistic inference might disappear in the next ten year. This means that work won’t disappear although certain jobs(ex. law, medicine,accounting) will become highly automated.

So that’s what must be said about UBI. It’s a terrible system that isn’t cost-effective and it disincentivizes work while making the false assumption that automation will have a uniform impact on the job market.

Negative Income Tax:

The basic idea is to use the mechanism by which we now collect tax revenues from people with incomes above some minimum level to provide financial assistance to people with incomes below that level.

-Milton Friedman

1. The NIT is a reasonable alternative to UBI which doesn’t create disincentives to work as citizens aren’t guaranteed a basic salary.
2. In its most basic implementation, the NIT is very simple:$\begin{cases} m = \text{minimum salary} \\ S = \text{actual salary}\\ \alpha = \text{tax rate}\\ \Delta S = S-m \end{cases}$If $\Delta S \geq 0$, then the citizen incurs a normal tax and when $\Delta S < 0$ the citizen gets paid $-\alpha \Delta S$.
3. NIT is much less expensive to implement:Let’s suppose half the UK population is living below the baseline salary of 10000 pounds and $\alpha = 0.5$. Then in the worst case scenario, the NIT program would cost:\begin{aligned} \frac{1}{4}.6.10^7.10^4=\frac{3}{2}.10^{11} \end{aligned}That’s 4 times less than the cost of implementing UBI! Moreover, I must add that this is a highly improbable scenario and we can also define much more sophisticated NIT systems with an adaptive tax rate.

Discussion:

From an economic perspective, the Negative Income Tax has the advantage that it’s economically feasible and it doesn’t create disincentives to work which is very important as a post-work future is at least several centuries away. However, an economic solution is only a partial solution to the problem of technological unemployment. There will also need to be robust programs for re-educating and retraining a significant fraction of the population. A concrete example of this is Xavier Niel’s free software engineering institution and South Africa’s free wethinkcode institution. This is something that governments should work on in partnership with companies that understand what the future of work will look like.

Conclusion:

In summary, we should take UBI pilot studies with a grain of salt, the implicit ‘universal’ assumption is dubious and any UBI program would be prohibitively expensive in practice. In comparison, the negative income tax which has none of these defects, is a much more reasonable economic solution although it shouldn’t be considered a complete answer to the problem of technological unemployment.