Proponents of a basic income grant (BIG) tend to overstate the simplicity of its implementation, its pro-growth outcomes, and how cost effective it might be. Modelling from a recent paper published by the IZA Institute of Labour Economics makes a strong case against such a policy for poverty relief.
Consistent with our arguments and our modelling work on the issue, the paper shows that the potential of a BIG to reduce poverty depend greatly on the design and implementation of such a policy. The authors show that the effects of a BIG depend on how existing social social security and taxation policies are adjusted when a BIG is implemented. While some BIG designs could reduce poverty, they show that a BIG is always very expensive, which undermines the long-term feasibility of such a policy. Their conclusion is worth highlighting, that ‘the simplicity of BI [grants], however, tends to be vastly overstated’.
While the study is calibrated for developed economies where the tax base is much larger, the general point they make is applicable in South Africa too: that the cost of a BIG would be very big. The authors show that a BIG equal to the poverty line in the Netherlands would cost almost 90% of total government tax revenue (or 25% of GDP). By comparison, BIG proposals currently being debated in South Africa range between around 7 to around 40% of GDP, depending on the number of potential beneficiaries and value of the grants. The authors draw two conclusions based on their microsimulations:
- Higher grants do not necessarily reduce poverty more
- A large proportion of the population would be worse off financially
These conclusions reflect the size of the required adjustments to other forms of social support and the increased taxation required to fund a BIG. In a South African context, the required adjustments to finance a BIG would likely be larger than in the study, given our more limited social security system, higher cost of public borrowing, limited tax base and relatively high top marginal tax rates. Our modelling shows that the required tax increases to fund a BIG in South Africa would end up reducing total employment and the size of the economy.
There may be alternative income support policies that might be more cost effective than a BIG, such as extending existing automatic stabilisers through enhanced unemployment insurance, increasing support for job creation through public programmes or government subsidies, or reforms to the existing social security framework. It would be useful for studies to compare the dynamic macroeconomic impacts of such policies to extensions of social transfers to help assess what options the South African government might have to support job creation and reduce poverty.
The South African cost estimates mentioned are summarised in Table A.1 and the literature review in our paper.
I am grateful for comments and suggestions from Hylton Hollander and Roy Havemann.