Monday, September 26, 2016

Basic Income Experiment in Kenya



Interest in a basic income has grown tremendously in the last few years. Whether the large RCTs of BI about to take place in Finland, interest from Jeremy Corbyn of the British Labour Party, or private experiments like Y Combinator, it's clear that the idea of basic income is intriguing to a variety of political groups, and a lot of data about the consequences of implementing a BI in many environments is about to become available.

The general time frame considered in this blog is around 2025-2035, an estimated window of when synergistic advances in AI and robotics will produce the conditions where a large fraction of jobs will be automated at a pace exceeding that which new jobs for humans are created, if they will be at all. BI in this context acts as a redistributive mechanism that keeps the economy working smoothly by channeling dollars from the winners of this economic disruption to the displaced.

The line of reasoning that I see supporting a basic income in that scenario of rapid first world jobs disruption is very straightforward--the alternative responses (or lack of response) to job displacement are inferior. For instance, limiting technological advancement decreases potentially massive gains in productivity and human freedom, and non-intervention that allows wages to sink so low that even the cheapest robot is less cost effective than a desperate human would create near slaves with no economic security. Although I think that some form of BI is the best policy, determining that exact form is much less straightforward and will probably be the result of a long series of experimental trials and theoretical developments.

The use of basic income today as a tool for poverty alleviation seems to me less straightforward in justification, as there are numerous other policies such as educational support, jobs support, etc. that work to the same end, but the potential is enormous. Researchers at GiveDirectly will begin a 3 treatment group trial in Kenya looking into the effects of a BI on over 200 villages, with some villages being guaranteed an income for over 12 years.

Business Insider goes more into detail about the plan as well, highlighting a quote from GiveDirectly "Comparing the first and second arms [the first treatment group gets a two year guarantee, the second a twelve year] will shed light on how important the guarantee of future transfers is for outcomes today..."

Indeed, it's the impact of just this kind of very long term support that it's hard to be certain about until it's been tried. Optimists (myself included) think it will encourage entrepreneurship, as the risk of a failed venture no longer likely means destitution, but a guiding principal of economics going into the 21st century must surely be that our intuitions and theories are never good enough to be above testing.

I think though one metric should stand above all others in evaluating these policies--did this policy reduce human suffering? We know from many studies that the biggest marginal utility gain of a dollar occurs in the very poor: going from an income of $10,000/year to $20,000/year in the US makes a much larger difference than going from $200,000/year to $210,000/year. Even if a theoretical goal of "economic development," is not realized via basic income, and we see something like people more or less still working the same jobs in the same villages 12 years from now, but the thousand diseases and maladies of poverty are reduced or eliminated, would we not consider it a successful policy to be implemented elsewhere?