A Discussion of “Universal Basic Income vs. Targeted Transfers”
Universal Basic Income (UBI) is a hot issue— it even played a role in the 2020 Democrat presidential primary as a key feature of Andrew Yang’s platform.
I see a lot of political, ethical, and philosophical reasons given for it, but what do the economists have to say? Not much apparently! I found only one article in the top journals of the AEA. It showed that as far as economics goes, UBI may not be the best course of action.
Harvard Professor, Rema Hanna, teams up with Benjamin Olken of MIT to examine if poverty is best combated with programs targeted only to an eligible group or with UBI where everyone receives a payment. To tackle such a daunting question, they are using data from two countries, Peru and Indonesia.
They start with the assumption a country would have to pay for either type of anti-poverty program from their tax revenue because money from foreign aid and international institutions is too little to cover the costs of these programs.
The authors address two arguments for UBI.
- UBI has a high ease of implementation and low administrative costs since there is no need to screen and verify eligibility.
- UBI does not distort the supply of labor since the payment does not go up or down based on how much you choose to work.
However, they show both arguments are lacking.
For the first one, think about a recent example in the US. While the CARES act payment was not universal since there were income limits, it still highlights some issues facing a government trying to send money to the taxpayers. The IRS struggled with issues like having the wrong bank account information that caused errors and delays. And what about the “unbanked” who have no accounts for automatic deposit? A check has to be mailed to them, but again, the government did not have the correct addresses for every taxpayer. A developed country like the US has more information on its taxpayers than most developing countries and still faced this battle.
Second, they show graphically that a UBI applied to a progressive income tax system raises the post-tax income at the bottom of the income distribution as they receive the UBI payment and lowers the post-tax income at the top. The higher the UBI payment, the more taxes have to increase. Microeconomics would predict that higher taxes can lower work effort for higher income earners whether or not the UBI payment alone is enough to discourage work effort for some or all. Thus, there is reason to be concerned that the supply of labor could be altered by a UBI policy.
And the problems with these 2 arguments for UBI are even worse in developing countries.
Many developing countries do not have a full list of people, their addresses, or bank account information so sending out the UBI payment would be challenging.
And, a large percentage of economic activity in developing countries is in what they call the “informal sector.” People working for themselves or in small companies, people working for cash, and bartering are all ways that economic activity will not be recorded and reported to the government.
In addition, developing countries will have more people whose income is exempt from taxes because it is too low to meet the minimum taxable income threshold. Specifically for Peru and Indonesia, Hanna and Olken find that 79% of the employed population in Peru and 87.5% in Indonesia have income below the taxable level. (p. 205)
They have already said UBI will make the higher income earners face higher marginal tax rates to cover the UBI payment. With fewer people paying taxes like in Peru and Indonesia, this will place the total cost onto a small group of people. If a UBI policy could be enacted in such conditions, it would probably be a low payment to keep the cost on the taxpayers lower and thus not help those in poverty as much as is needed.
Since UBI is not looking promising, the authors then turn their attention to some existing anti-poverty policies in Peru and Indonesia that are targeting those who most need help rather than being universal. A major problem for targeting programs in developing countries — they cannot directly observe income, particularly of the lowest income earners. This leads the governments to constructing “proxy-means tests” to determine who is eligible and who is not.
Typically, the governments will use data from periodic, large scale censuses of the population to gather information on assets and other available survey data.
The government creates a proprietary statistical estimation method to use the data collected to predict income and establish eligibility. Those whose income is estimated to be below the established maximum income can receive the anti-poverty transfer payments.
Thus Hanna and Olken use the data from Peru and Indonesia to answer the question: which is optimal, a targeted program like they currently use or UBI?
They utilize the survey data from each country that is used to target eligible people in each country. Because it is a prediction, there is a difference in most cases between what is predicted and what is actual, and this introduces potential errors.
The authors use regression analysis to illustrate these errors. Essentially there are four outcomes.
- The household will be included correctly because their actual income makes them eligible.
- The household will be excluded correctly because their actual income makes them ineligible.
- The household will be included incorrectly — Inclusion Error: Their actual income is too high to be eligible but the predicted income showed them as eligible.
- The household will be excluded incorrectly — Exclusion Error: Their actual income is low enough to be eligible but the predicted income showed them as ineligible.
We don’t want to omit payments to those who should get it (Exclusion Error) and give it to those whose income is too high (Inclusion Error).
In fact, there is a trade off between Inclusion Error and Exclusion Error.
Looking at the extremes, you can make sure Inclusion Error is 0% by not making any payments, but then no one is helped, and Exclusion Error is 100% of the people below the poverty line.
Or, you can make sure you do not exclude anyone by paying everyone, a UBI, but then the inclusion error is 100% of the people above the poverty line.
When they apply the analysis to the two countries they are examining, they find that to reach 80% of the intended people below the poverty line (which implies a 20% exclusion error),
- Peru has a 31% Inclusion Error, and
- Indonesia has a 22% Inclusion Error. (p. 211)
What is the significance of this? In reality, developing countries have a fixed budget for their anti-poverty programs. As more payments go to the ineligible (Inclusion Error increasing), fewer dollars per capita are being paid out to each recipient.
Taken to the extreme, while a UBI has an exclusion error of 0%, the reality of a fixed budget means a much lower dollar per capital reaching those who are below the poverty line.
But wait! The UBI does not require all this data collection, income estimating, and other administrative costs. Surely that would make a difference! All the money saved could be used to increase the per capita payments.
Sadly, they find, not really. Using the data, they find a 1.7% savings of the overall budget in Peru and a 0.8% savings of the overall budget for Indonesia. (p. 213) The amount of increase to the per capita payment from these savings is less than the the decrease caused by the higher number of recipients for the UBI.
Next they turn their attention to “social welfare” which is econ-speak for, “Is society better off or worse off in aggregate?”
Essentially, for any given budget amount, society as a whole still gets some positive value for Inclusion Error recipients, those who should not have qualified. And, there is some negative value for society for all of the Exclusion Error recipients, those who should have qualified for payments. And there needs to be some value placed on the fact that per capita payment goes up as the income eligibility goes down, since fewer people are sharing the same total budget.
They ran their estimation of social welfare multiple times to find what scenario gives the highest social utility (maximum satisfaction after all of the trade-offs). Using the data for Peru and Indonesia, they could determine what is better for society: a targeted payment program or UBI.
Turns out, a targeted payment was best for both countries.
For Indonesia:
- The socially optimal program targets 19% of the people to be recipients.
- The Inclusion Error is 7.4% and the Exclusion Error is 58.2%
For Peru:
- The socially optimal program targets 18% of the people to be recipients.
- The Inclusion Error is 6.4% and the Exclusion Error is 52.4%
That means, social welfare peaked for these programs in these countries where Inclusion Error is 7.4% and 6.4% — a far cry from the 100% Inclusion Error of a UBI.
So it appears from this analysis, that targeted anti-poverty programs may be the best bet to help those in poverty in developing countries. Overall, not a good finding for UBI proponents! Still, this is a review of two specific countries, and it is focused on developing countries.
An argument advanced for UBI in developed countries is that our increase in productivity, perhaps due to AI, will so increase per capita output that fewer people will need to work, and we will still be able to pay a UBI to everyone from this bounty. Sounds like the Star Trek economics discussion from years ago. It would be a nice world to live in if it ever came to be.
References:
Hanna, Rema and Benjamin A. Olken. 2018, “Universal Basic Income versus Targeted Transfers: Anti-Poverty Programs in Developing Countries.” Journal of Economic Perspectives, 32 (4): 201–226.
By Ellen Clardy, PhD on .
Exported from Medium on December 15, 2022.
