Do government spending and taxation really reduce inequality, or do we need more thorough measurements? A response to the World Bank researchers

11th February 2016 By: Econ3x3

Do government spending and taxation really reduce inequality, or do we need more thorough measurements? A response to the World Bank researchers

World Bank staff and consultants claim that South Africa’s progressive taxation and pro-poor social spending reduce the Gini inequality coefficient from 0.77 to 0.59. But their data and methodology are deficient: their research ignores large areas of government spending and taxation that may significantly increase inequality. Thus their conclusion that fiscal policy is redistributive is overhasty and unfounded – whilst it is prone to be used, or misused, to promote a budget-cutting political agenda.

‘South Africa can claim to have one of the world’s most redistributive public purses,’ claims Business Day associate editor Hilary Joffe (2015), drawing upon World Bank research findings (Inchauste et al. 2014, Inchauste et al. 2015, World Bank 2014), which were recently reiterated by Woolard et al. (Econ 3x3, October 2015). In response to the question: ‘How much is inequality reduced by progressive taxation and government spending?,’ their answer is that, when they adjust household incomes ‘comprehensively’ for the impact of government revenue and expenditure, the Gini coefficient (which measures income inequality) is reduced from 0.77 to 0.59. This claim is regularly repeated by economists and high-profile commentators, often in support of fiscal austerity (see Bond 2015).

Although I have no easy answer in rebuttal, there is a significant problem. The research ignores large areas of state spending that would probably raise the Gini if included. As a brief glance at the national budget shows, the types of expenditure that the World Bank researchers took into account ignore other categories of expenditure that also have strong distributional effects. Likewise, their selection of taxes omits important components of corporate taxation – including the plethora of tax allowances, credits, loopholes and incentives embedded therein, all of which increase capital gains to holders of company shares. There also are other problems, e.g. regarding the quality of services (see below).

The Bank’s research on this vital matter is not only deficient due to its incompleteness, but is politically biased in a way that largely supports the status quo’s perspective on inequality, namely that there is little fiscal space for further redistribution. Although a key member of the research team has acknowledged the validity of my critical questions and the limitations of the methodology (Lustig 2015), World Bank staff and consultants go on repeating their findings, as in the Econ3x3 article – and most recently in a draft document for a collaborative World Bank project (2016: 22) to assess poverty and inequality in SA.

World Bank researchers acknowledge weaknesses, but . . .

The fiscal tools examined by the Bank researchers (i.e. Inchauste et al.) were only those related to household taxation, social spending, and municipal services – and even there they made assumptions that are dubious in the South African context. To be sure, while repeatedly claiming ‘comprehensive’ coverage, the Bank’s staff and consultants have felt compelled to admit the following data and methodological ‘limitations’:

However, if they fail to address the biases implicit in these shortcomings, it is impossible to conclude that ‘(n)ot only are South Africa’s main fiscal instruments progressive overall, the degree and structure of progressiveness is such that these instruments achieve significant reductions in income inequality’ (Woolard et al., Econ3x3, October 2015: 7). Consider some of these drawbacks (for more detail, see Bond, forthcoming).

Ignoring quality in state spending: the case of education

Education, the single largest budgetary commitment, illustrates how dubious the alleged social spending benefits for recipients can be. Most public schools produce extremely low-quality education, thus locking in inequality with regard to life chances.

The World Economic Forum’s (2015) Global Competitiveness Report 2015–16 rated South African education as the worst of 140 countries in terms of science and mathematics training, and 138th in overall quality. As Spaull (2013: 10) remarks after studying the 1994–2011 outcomes: ‘South Africa has the worst education system of all middle-income countries that participate in cross-national assessments of educational achievement.’ The  OECD (2010: 248) notes that ‘in 2008, only 1.4% of working-age Africans held a [university] degree, compared to almost 20% of working-age Whites. This proportion for Africans has hardly increased since 1993, while the proportion for Whites has grown by 5.4%.’

Given these outcomes, it could just as easily be argued that inequality is amplified by the manner in which public education is provided to the low-income majority. This story is fairly typical of maldistributed state resources; similar concerns have been raised regarding the quality of health services to the poor.

As National Treasury senior official Andrew Donaldson acknowledges (in a 2014 Econ3x3 article): ‘In areas such as education, health care and urban transport, service provision tends to evolve in differentiated ways […] the result is a fragmented, unequal structure in which the allocation of resources and the quality of services diverge.’ Combined with semi-privatised systems, such public spending, he admits, ‘entrenches inequality between rich and poor.’

Spending ignored by the World Bank analysts

The social spending items taken into account by the World Bank researchers account for 43% of total government spending (for fiscal year 2010/11). The October 2015 medium-term budget statement anticipates R1.45 trillion in 2016–17 spending, of which R219 billion (15%) is allocated to basic public education subsidies and R170 billion (12%) to public health (with the quality variances across income groups a main caveat for both these types of spending, as discussed above).

But other major areas were not analysed by the Bank researchers, though they have vital distributional implications:

No one (myself included) has done the difficult work required to put numbers to the distributive effect of the biases within these spending categories. But, without having done so, Bank staff and consultants should not make expansive claims about a ‘post-fiscal’ Gini coefficient improvement.

Likely regressive state spending: a first look  

Although the World Bank (2014: 21) claims to ‘comprehensively assess the distributional impact of government taxation and spending’ using a ‘comprehensive fiscal incidence analysis’ (Inchauste et al, 2015: 9), its researchers ignore major items that appear to subsidise the accumulation of capital gains to wealthier households. Consider the following:

1. State subsidies to capital/corporations/corporate shareholders (‘corporate welfare’):

As would be required in properly assessing the value of education and healthcare investment, the Bank and other researchers could estimate the distribution of the benefits from these investments across the income and wealth spectrum.

It is vital for researchers who investigate inequality to assess how much such state subsidies (corporate welfare) add to firms’ longer-term asset bases. This translates into capital gains on shares, a benefit accruing mainly to the rich. Shares on the JSE represent a vast component of household wealth.

2. Other state spending (superficially) considered by the Bank’s researchers includes the provision of household water and electricity, whether by an Eskom or municipalities.

3. Treasury regulations also have significant distributional effects. For example, its deregulatory attitude to transnational corporate profit expatriation has allowed a great deal of income to flow to firms’ overseas financial headquarters (thus supporting future capital gains for wealthy households). Global Financial Integrity estimates the average annual illicit outflows at $21 billion for 2004–2013 (Kar & Spanjers 2015). Such tax laxity towards ‘base erosion and profit shifting’ by multinational corporates is a most important negative-redistributive aspect of fiscal policy not measured by the Bank.

4. Whether fiscal policy favours the wealthy as opposed to the (long-term) interests of the poor also depends on ‘natural capital accounting’, i.e. putting a value on non-renewable resource depletion associated with corporate extraction of minerals. The World Bank (2011: 193) estimated the impact of natural capital shrinkage on South African gross national income in 2005 to be negative 9%. Not to tax mineral wealth is a distributional fiscal policy choice which allows the proceeds of the depletion of non-renewable resources to shift from society as a whole to wealthier shareholders (even though workers in mines, smelters, transport and other downstream beneficiation industries also benefit).


The World Bank researchers have meticulously measured something. But what they have measured is not the whole picture. While selected elements of state taxation and spending have an inequality-reducing effect on a selected component of well-being (household income) – as measured using a selected quantitative measure, the Gini coefficient – this is not a satisfactory basis from which to draw overarching distributional conclusions.

Bank staff and researchers have ignored systemic state-induced inequalities that shape distributions of income and of wealth (including capital gains) and broad human welfare. Indeed, the so-called ‘market distribution’ or ‘pre-fiscal’ distribution of income is already the systemic outcome of an inequality-producing economy that is substantially shaped and supported by state action that has long favoured wealthier people and the corporations in which they invest.

As the influence of the World Bank researchers’ project grew in the past year, I queried the work and received a series of (ultimately bureaucratic) emails from its officials (Bond 2015). Fortunately, upon asking the main consultant, Nora Lustig of Tulane University, why more accurate assessments of the state’s pro-corporate fiscal benefits were not attempted so as to offset the bias from only considering social spending, she took up the challenge with honesty: ‘Your questions are very valid. Regretfully, we have yet to figure out a solid methodological approach to allocate the burden/benefit to households of the list of interventions you list’ (Lustig 2015).

The question is: should Bank staff and allied researchers have been so hasty to publish research findings that had major implications for fiscal and budgetary policy when the methodology is evidently too limited to thoroughly answer the question they have posed? This is especially important in 2016 when even more pressure has risen from financial markets and credit rating agencies to reduce social spending – following the 2015 Budget of Finance Minister Nhlanhla Nene that cut the real value of welfare grants.

Beyond the critical flaws in measurement, the main risk of the World Bank research on ‘post-fiscal’ South African inequality is political bias: it promotes, or can be misused to promote, the budget-cutting agenda of the most regressive faction of corporate capital, i.e. financiers.

To mitigate this damage, the research task ahead for the World Bank and other inequality researchers is surely to attempt to properly measure the distributive impact of all government spending and all taxation (including allowances). That would make a real contribution to the battle against inequality and could inform appropriate tax and expenditure policies and projects (including planned ‘white-elephant’ mega-projects).

Until that research has been accomplished, would it not be better for economists, when asked about the distributional effect of fiscal policy in South Africa, to simply say: we do not yet know enough to answer that question?

Written by Patrick Bond, Econ3x3