As shown in the graph below, the incidence of VAT for the poorest decile is 9%. For the highest income quintile, the incidence is 12%. This means that the poorest pay 9c in VAT on every rand of disposable income and the richest pay 12c in VAT in each rand of disposable income. The challenge is that, though VAT is progressive, 9c/rand is worth a lot more to the poorest than 12crrand is for the richest. VAT, however, is an efficient tax that raises a lot of revenue easily and the costs of administration are low.

Figure 1: Incidence of indirect taxes in South Africa

Source: World Bank Group: Distributional Impact of Fiscal Policy in South Africa

A number of commentators have suggested that consideration be given to expanding the basket of goods that is zero-rated. They argue that this will cushion the impact of the increase in VAT on the poor. The ANC NEC reached a similar conclusion in its deliberations on the Budget, and has recommended that the extension of the zero-rating architecture be investigated.

The problem with zero-rating is that items that are zero-rated are consumed by all – the poor and the rich. While it is possible to identify goods that are consumed mainly by the poor, it is not possible to exclude higher income groups from consuming these goods and receiving a subsidy for doing so. Furthermore, while poor households spend a larger proportion of their income on basic items such as food, high income households spend significantly more on food than poor households and so they benefit more from zero-rating, in absolute terms, than low income households. As a result, significant potential VAT revenue from high income households is lost, thus undermining the revenue collection objectives of VAT. In essence, zero-rating is a subsidy paid to consumers who consume products that are zero-rated. Zero-rating has two effects: a desirable outcome of subsidising consumption by low-income households, and an undesirable outcome which subsidizes the consumption of the zero-rated good by high-income households. Ideally, the basket of zero-rated goods should seek to maximise the first outcome and minimize the second outcome. This trade-off can be demonstrated by assessing the revenue losses, by income decile, of goods that are currently zero-rated, as shown in the table below. This analysis was conducted for the Davis Tax Committee.

Table 1: Revenue foregone (in rand millions, 2012 prices) on zero-rates goods, by consumption decile

The table above shows the revenue losses, in 2012 prices, by decile for 15 items that are currently zero-rated. In total, zero-rating does result in a significant gain for the poor: the incidence of VAT is lower for lower income groups than it is for high income groups for many consumption items. The problem is that the subsidy to high-income deciles is significantly greater in absolute terms than it is to low-income decides. A good example of the problem is highlighted by zero-rating milk. The zero-rating of milk results in a revenue loss R980-million. 21% of the benefit of the zero-rating accrues to the wealthiest group who benefit to the tune of R208-million. In contrast, the poorest 10% of South African households only ‘get’ 3% of the subsidy, R32 million in absolute terms. The subsidy to the wealthiest 10% of households exceeds the subsidy to the poorest 40% of households. The impact of zero-rating fruit and vegetables also results in a subsidy that skewed toward high income groups. In short, though zero-rating does benefit the poor, its impact is significantly diluted because significant subsidies accrue to higher income earners.

It is important to clarify that while high-income groups do benefit from the subsidy, zero-rating is an effective instrument at protecting the poor, even though it may be less effective at addressing inequality. Should zero-rating be extended? It really depends on what our appetite is for subsidising the rich in order to protect the poor. We might say that we are happy to subsidize the rich as long as the poor benefit to a greater extent. The table below assesses the benefit ratio, between deciles 1-7 and 8-10, of the current goods that are zero-rated. If we adopt a ratio rule of 3 (that is the benefit to Q1-7 should be at least three times that of Q8-10), zero-rating of rice, bread, maize meal, dried beans, canned pilchards, cooking fat and paraffin is justified. On this rule, the zero-rating of mealie rice, milk, eggs fruit and vegetables is not justified.

Table 2: The benefit ratio of zero-rating

The balance of the evidence on zero-rating suggests that the zero-rating policy has been extended as far as possible. The inclusion of additional items will result in significantly greater subsidies to high income deciles and thus exacerbate inequality. In earlier research with Debbie Budlender and Daniela Casale, we identified three additional items that would pass the 3-times rules – children’s clothing, poultry and basic personal care items. However, except for poultry, it would be very difficult to differentiate between which products should be zero-rated and those that would not.

A proposal to protect the poor – why we should consider VAT

Tax authorities often tax about the three Es in tax policy – equity, ease of administration, and efficiency. We have dealt with the equity consideration above: VAT is not regressive. From an ease of administration perspective, VAT is a transparent tax (we all know how much VAT we pay). From an efficiency perspective, there are two considerations – how much money does the state collect for a small change in the rate of tax, and how much of what the state should collect does it actually collect (ie how easy is it to evade the tax). VAT has great advantages on both these scores. Studies that have been done show that the gap between what should be collected and what is actually collected is very small indeed – its much smaller in South Africa than in comparable countries. For these reasons, VAT is an extremely attractive option when it is important to quickly and cost-effectively raise revenue. This is especially so if there is a well-functioning VAT system, with a zero-rating provision that protects the most vulnerable citizens from the pernicious characteristics of VAT.

In addition to the three Es, one of the most notable benefits of VAT is that it raises a lot of money from the rich. An increase in VAT, from 15% to 16%, will raise R22.9-billion. In total, deciles 1-9 will pay about R9.26-billion (43.34%) while decile 10 will pay R13.63 billion. The poorest decile, in contrast, will pay just over R1-million. Clearly, the rich will pay a lot more of the VAT bill in absolute terms. In our view, a more effective proposal than expanding zero-rating would be to increase the VAT rate to 16% and design a tax credit system that would allow government to pay a credit to low-income groups. A further 1% increase in VAT will result in additional revenue of R22-billion (assuming that there are no behavioural responses to the increase in VAT).

Table 3: Progressivity of indirect taxes in South Africa

The R22-billion raised by the increase VAT increase from 15 to16% can be transferred back to households in deciles 1-9 only – so that the full burden of the second increase is borne only by decile 10, and the proceeds are distributed equally among households in deciles 1-9. The payoff scheme for such a proposal is shown in the table below. Deciles 1-6 benefit in net terms, with the payoff being very progressive. The costs of the entire VAT, over the two stages is borne by deciles 7-10. The costs are spread in an extremely progressive way.

Table 4: The progressive potential of VAT

The question that arises is how could the subsidy from be paid back to households? This could easily be administered by SARS through the current tax system. SARS could work with Home Affairs to compile a complete database of all citizens, and the cash transfer should be administered by SARS in this way. The main problem with the current tax rebate system in South Africa is that only people who are already income tax payers enjoy a tax rebate. This means that many low or zero earners are excluded from the redistributive elements of the tax system (although they do benefit on the expenditure side of the Budget). The tax credit system we are proposing would bring all individuals into the tax system so that they can benefit from cash transfer tax credits. The most appropriate system for South Africa is a targeted cash tax credit scheme that is not contingent on work status. There are already variations of this in place around the world.

The UK has an earned-tax credit, where all taxpayers are paid a credit. The credit is netted off against any tax burden that the individual accrues. So, the tax authority collects money from those whose account is in a net debit (usually the rich) and then makes a payment to those who are in credit (usually the poor). The benefit will be paid monthly, in cash, into each individual’s bank account. In many ways, this is similar to the tax rebate, but it is extended to all citizens, and takes the form of an actual payment for those whose tax credit exceeds their tax liability. While there are some particular issues with this system as it is implemented in the UK (unitive conditions and long lead times for payment), the principle of a transfer-based tax credit is a sound one. The United States has, since 1975, operated an earned income tax credit, which aims to support low and moderate-income individuals and families. The UK experience with Universal Credit has shown that the particulars of the design and implementation of a tax credit scheme matter.

Here we have presented a thought experiment investigating the merits of zero rating (a revenue-side intervention) with a system of pro-poor tax credits (an expenditure-side intervention). We have made two particularly important points. The first is that VAT is an efficient way to raise large amounts of revenue very easily. The second is that state spending is highly redistributive: unlike expanding zero-rating, the redistribution of VAT revenue through government spending can be targeted to be highly progressive. This could be accomplished through a tax credit system. Indeed, government spending on education, social assistance, and healthcare is progressive in and of itself. Coupled with the current zero-rating system, this makes a compelling case for VAT as a device to raise additional revenue.

Going back to our earlier paper, the thought experiment shows us the full effect of the proposal to increase VAT from 14% to 15% on the assumption that the funds are fully spent on the expenditure side of the Budget in a manner that it equally benefits decides 1-9, with quintile 10 not benefiting at all. On this scenario, the proposal to increase VAT from 14-15 %, significantly benefits poor households in South Africa. DM

Photo by Igor Ovsyannykov on Unsplash