South African Institute of Race Relations NPC
Submission to the Economics Tax Analysis Chief Directorate, National Treasury, regarding the Policy Paper on Taxation of Sugar-Sweetened Beverages of 2016
Johannesburg, 22nd August 2016
The Economics Tax Analysis Chief Directorate of the National Treasury (the Treasury) has invited public comment on the Policy paper on the Taxation of Sugar-Sweetened Beverages published on 8th July 2016 (the policy paper).
Public comment is due not later than 22nd August 2016. However, this period is too short to meet the constitutional requirement for proper public consultation. It would also have been helpful if the policy paper had been accompanied by an initial socio-economic assessment of its likely economic and other ramifications, as envisaged in the government’s new Socio-Economic Impact Assessment System (SEIAS).
This submission is made by the South African Institute of Race Relations NPC (IRR), a non-profit organisation formed in 1929 to oppose racial discrimination and promote racial goodwill. Its current objects are to promote democracy, human rights, development, and reconciliation between the peoples of South Africa.
The proposed tax and its rationale
The proposed tax on sugar-sweetened beverages (SSBs) is to be levied on producers and importers as an indirect excise tax. It will be levied, in general, at the rate of 2.29c per gram on the added and intrinsic sugar contained in all soft-drinks other than 100% fruit juices and milk products. According to the policy paper, ‘this rate roughly equates to a 20 per cent tax incidence for the most popular soft drink, ie Coca Cola, averaging 35g per 330ml’.
According to the Treasury, the primary rationale is not to increase the government’s tax take, but rather to help overcome the problem of obesity within South Africa. Obesity is becoming a major problem in many countries around the world. In South Africa, according to the Global Burden of Disease 2013 Obesity Collaboration, the obesity prevalence rate stands at 13% among adult men and at 42% among adult women.
Both in South Africa and elsewhere in the world, women are generally more prone to obesity than men. Prevalence is also generally higher among those who have less education and a lower socio-economic status.
The many and complex causes of obesity
According to a 2014 study by the McKinsey Global Institute (the McKinsey study), obesity is a ‘critical global issue’ and has many and complicated causes. Says the McKinsey study: ‘The root causes of rising obesity are highly complex, scanning evolutionary, biological, psychological, sociological, economic, and institutional factors’. It is not surprising, thus, that one study by the United Kingdom (UK) government on the reasons for obesity identified ‘more than 100 variables that directly or indirectly affect obesity outcomes’.
As the McKinsey study stresses, ‘obesity is the result of a multitude of factors, and therefore no single solution is likely to be effective in tackling it’. A range of interventions is needed to ‘encourage and empower individuals to make the required behavioural changes’.
Sugar taxes are ineffective, but enjoy the most media attention
The McKinsey study concentrated on 44 interventions that are being used to counter obesity in various countries around the world. It found that a 10% sugar tax was among the least effective interventions, as there is no direct evidence that such a tax has any impact on body weight at all. That a tax of this kind will bring about positive changes in sugar intake or obesity is simply an unproven assumption, which is based on nothing more than supposedly logical deductions from indirect evidence of possible consumption shifts.
The McKinsey study also found that the media nevertheless gave far more coverage to sugar tax than to other interventions that are far more successful in countering obesity. In one year, for example, there were 930 media articles referring to the benefits of a sugar tax, as opposed to 182 on the intervention likely to be most effective: a reduction in portion size.
Though the McKinsey study does not spell this out, this skewed media coverage is likely to skew policy interventions in favour of a sugar tax. In doing so, it risks downplaying and distracting attention from alternative interventions with much greater prospects of success.
Countries that have tried, and/or rejected, sugar taxes
The policy paper says various countries have implemented sugar taxes and thereby reduced consumption of taxed food or drinks. However, the causal link between higher prices and decreased consumption of taxed products is patchy and unconvincing. Moreover, even if consumption of taxed products does indeed go down, consumers may instead switch to cheaper brands, cheaper outlets, or other products with as much sugar or more. Because of these substitution effects, overall sugar intake is unlikely to decline and obesity is unlikely to be reduced.
The policy paper fails to acknowledge these weaknesses in the case for a sugar tax. It also brushes over the experience of countries which have either abolished sugar taxes, or declined to introduce them. Such countries include Denmark, Iceland, and Romania, while Ireland repealed its tax on soft drinks in 1992. In addition, Norway introduced its sugar tax as far back as 1981, which means the tax is unlikely to have affected the frequency of soft drink consumption in the period from 2001 to 2008, as cited in the policy paper.
The policy paper also cites the experience of Mexico as confirming the positive effects of an SSB tax. Mexico introduced a 10% tax (1 peso per litre) on SSBs in January 2014. There is some evidence (contradicted, however, by sales data in 2014 and 2015) that this tax reduced consumption of SSBs by 6% during 2014. However, assuming that the tax did indeed bring about a 6% reduction in the consumption of taxed profits, the average reduction in SSBs was nevertheless a mere 11.5ml per person per day. This is too little to have any impact on obesity.
The ‘mathematical model’ in the Manyema article
The policy paper puts great emphasis on ‘a mathematical model’ developed by various South African authors (including researcher Mercy Manyema and Professor Karen Hofman of Wits University) which supposedly provides evidence that a 20% SSB tax will reduce the prevalence of obesity in South Africa by 3.8% among men and by 2.4% among women, yielding a total reduction in the number of obese people of 222 660.
However, the Manyema article cites no evidence at all in support of its conclusions, which are based entirely on a series of assumptions which may not in fact hold true. It expressly recognises at least some of the weaknesses in its methodology, including a ‘lack of SA-specific own- and cross-price elasticity data’ (ie data dealing with likely price increases both in taxed products and in possible substitute products). It also acknowledges that ‘its model does not include the substitution effect of other sweetened drinks such as coffee, tea, and hot chocolate’, whereas ‘other studies have shown that the demand for tea and coffee…goes up with SSB price increases’. It further admits that it made no attempt to ‘account for substitution of SSBs with other sugar-sweetened food items’, instead simply assuming that such substitution would not be significant.
In addition, the estimates provided in the Manyema article in fact vary widely. It recognises, for example, that the reduction in obesity prevalence among men could be as low as 0.4% (not 3.8%), which would result in reduced obesity for only 16 060 males. It also acknowledges that the reduction in prevalence could be as low as 0.3% among women, which would result in reduced obesity for 16 550 females. On this basis, some 32 610 people would experience a reduction in obesity, rather than the much higher total (222 660) it emphasises.
The Manyema article also acknowledges that, even if the average (3.8%) is in fact achieved among men, this will ‘equate to 0.5 percentage point change’ in the prevalence of obesity among males. Moreover, if the average of 2.4% is indeed achieved among women, this will be ‘equivalent to a 0.8 percentage point change’ in the prevalence of female obesity.
Such low numbers, speculative as they are, hardly justify a substantial tax on a wide range of arbitrarily-selected products, especially when this tax could also have significant negative economic ramifications.
Similar studies also unconvincing
The ‘mathematical model’ reflected in the Manyema article – and on which the policy paper sets so much store – is similar to various studies elsewhere, which also make unproven assumptions about sugar taxes and reduced obesity. As the Institute for Economic Affairs (IEA) in the United Kingdom has pointed out, a systematic review of some 880 studies dealing with the assumed link between food and SSB taxes and the prevalence of obesity has found little evidence of this causal connection.
Writes the IEA: ‘There is a striking contrast between theoretical studies, which generally predict that such taxes “work”, and studies of hard data in places that have actually implemented them, which generally show the opposite. Lacking real world evidence that sugar taxes are effective as health measures, campaigners continue to cite findings from crude economic models which do not adequately account for the ability of consumers to choose cheaper or discounted brands, to shop at cheaper shops, or to switch to alternative high-calorie food and drink products.’
Likely economic consequences of the proposed tax in South Africa
It is difficult to predict the likely economic consequences of the proposed SSB tax. However, if Manyema article is correct in predicting the extent to which SSB consumption will be reduced as a result of the tax, these consequences are likely to be considerable.
The likely yield from the SSB tax, based on the number of taxable soft drinks sold in South Africa in 2015, would be R10.5bn. If consumption is reduced in the way the Manyema article predicts, the soft drinks industry would be hard hit. Its contribution to gross value added (GVA), a proxy for GDP, could decrease from roughly R60bn to around R47bn, a decline of some 22%. An estimated 42 000 direct, indirect, and induced jobs (out of current total of 185 000 jobs) could be lost. The industry’s contribution to tax revenues could drop from R17bn to R14bn. Moreover, this decrease of R3bn would be only partially offset by the yield from the new tax (which, on the reduced consumption thus envisaged), would amount to some R7.6bn.
The SSB tax would also have a negative impact on the entire value chain, from agriculture (where some 79 000 people are directly employed in the sugar sector alone) through to utilities, construction, the hospitality industry, the wholesale and retail sector, the transport industry, and the financial and business services sector, among others. In each of these spheres, current contributions to GVA would be diminished, jobs would be lost, and tax payments would come down – but without any offsetting contribution in these spheres from the new SSB tax.
The retail sector would be particularly hard hit. It could well lose another 15 000 jobs (over and above those earlier identified), while between 10 000 and 20 000 anticipated jobs would become much more difficult to generate. Within the retail sector, the impact would fall most heavily on the informal and often home-based ‘spaza’ shops which currently employ some 360 000 people and which rely on soft drink sales for some 15% to 20% of their revenue. The SSB tax would significantly reduce the sales and profit margins of these small enterprises, which could lead to the closure of between 6 500 and 11 500 of these outlets. Major supermarkets and discount stores would find it easier to absorb the impact, but spaza shops – along with other local and traditional retailers – would be particularly badly affected.
The South African economy already stands on the brink of recession and the downgrading of its international credit ratings to sub-investment (junk) status. The country also confronts a major crisis of joblessness, with the unemployment rate already standing (on the broad definition which includes discouraged would-be workers) at 36% in general and at 67% among young people aged 15 to 24. No other country that has seen fit to introduce a sugar tax has confronted an equivalent burden of joblessness. In circumstances such as these, South Africa simply cannot afford to put between 55 000 and 60 000 jobs at risk – and especially not when the claimed health benefits of the proposed tax are most unlikely to be realised.
If the SSB tax is introduced, demand for taxed products is likely to prove more inelastic than the Manyema article assumes. In this event, the consumption of taxed SSBs will not decrease as sharply and the negative impact on GVA, jobs, and tax revenues within the soft drinks industry will be much reduced. However, the health rationale for the SSB tax will also then be shown to be deeply flawed, raising yet further questions as to why the tax should be imposed at all.
A regressive tax, with major impact for consumers
In addition, the SSB tax will be a regressive measure with a particularly negative impact on low-income households, which spend a significantly higher proportion of their overall income on food and beverages than wealthier households do. The policy paper claims that this negative impact will be offset by health gains, but since SSB taxes have no proven, or even likely, impact on obesity, no such gain is likely to be realised.
The SSB tax will also put further strain on consumers, who are already face a 6.3% inflation rate, a heavy debt burden, high interest rates, and rising unemployment (with some 500 000 jobs already lost this year). Yet consumer spending is vital to the economy, contributing some 60% to GDP. With the country already facing major economic malaise, South Africa cannot afford the further economic blows the SSB tax could well generate.
The cheapest, not the most effective, option
The policy paper describes the proposed SSB tax as the most ‘cost-effective’ option available to help counter obesity. What the Treasury really means, however, is that the tax is the cheapest way the government can (purport to) tackle the obesity problem.
As the policy paper notes, implementing a fiscal measure like the SSB tax is likely to cost only R0.20 per head to implement (according to 2010 data). By contrast, regulating food advertising would cost R0.90 per head, while food labelling would cost R2.50 per head, worksite interventions would cost R4.50 per head, mass media campaigns would cost R7.50 per head, school-based interventions would cost R11.10 per head, and physician counselling would cost R11.80 per head. In other words, implementing an SSB tax will (not surprisingly) be cheaper for the government to implement than any other measure, but whether it will be effective in countering obesity is doubtful.
Across the world, food and beverage taxes, as the McKinsey study has shown, are among the least effective ways of overcoming obesity. Moreover, because obesity has so many and such complex causes, it can be countered only through a wide range of interventions. The policy paper implicitly acknowledges this when it claims that the SSB tax will be part of a ‘comprehensive package of measures’ aimed at stimulating ‘healthy food choices’, promoting physical activity, educating and mobilising communities, and establishing ‘a surveillance system’ to strengthen monitoring and evaluation.
Whether any such additional measures are in fact taken remains, however, to be seen, for the government lacks the means to embark on the more expensive interventions the policy paper outlines. Nor is the additional revenue from the SSB tax likely to be used for any such purpose. In reality, SSB tax revenue will be paid into the National Revenue Fund, which has no way of earmarking the funds paid into it for specific health (or other) purposes.
A stealth tax, rather than a health tax
The SSB tax is essentially a stealth tax, not a health tax. The government urgently needs to bring in more revenue, but is reluctant to embark on the policy reforms which would be effective in doing so by stimulating investment, growth, and jobs. The Treasury also knows that personal income and corporate tax rates cannot easily be raised, whereas the VAT rate, at 14%, is low by global standards.
South Africa thus has scope to increase the VAT rate, but this would be highly politically unpopular. However, if enough stealth taxes can be introduced, the government will be able to attain the equivalent in increased revenue (R21bn a year) that a one percentage point increase in the VAT rate would yield.
If the SSB tax does indeed yield R10.5bn, the revenue it generates will take the government almost half way towards its R21bn goal. In addition, once the SSB tax is in place, the government plans to proceed with other food taxes. These could enable it to bring in another R10bn in consumption taxes – and without it ever having to acknowledge that what it is (effectively) doing is to increase the VAT rate from 14% to 15%.
Sugar probably not a key cause of obesity
Sugar is probably also not the key culprit in rising obesity, as the Treasury should acknowledge. In South Africa – as in the UK, Australia, and Canada, for example – sugar consumption has gone down in recent decades, even as obesity has increased.
The South African National Health and Nutrition Examination Survey (SANHANES-1), which was carried out by the Human Sciences Research Council (HSRC) and Medical Research Council (MRC) in 2013, has also shown that 80% of South Africans already consume sugar in moderation.
The SANHANES-1 survey assessed dietary sugar intake among South Africans on a scoring scale of 0 to 8. A score of 0-2 implies a person is accustomed to a low sugar intake, 3-5 indicates a moderate sugar intake, and 6-8 shows a high sugar intake. The mean score for South Africans was 3.0, which is at the low end of the moderate band. Among young people, the mean was slightly higher, at 3.47. A similar bias is seen towards urban formal areas, where the mean score was 3.36, and towards white consumers, who scored 3.44. Sugar intake is lowest among black Africans, and in rural and informal areas. These scores indicate that South Africans on average do not consume too much sugar.
In total, SANHANES-1 found that only 19.7% of South Africans have a high sugar consumption score, 38.2% consume moderate amounts, and the largest share of the population, 42.1%, has a low sugar intake. This distribution is similar to the distribution of fat intake among South Africans: only 18.3% have a high fat intake, 46.4% consume fat in moderate amounts, and 35.3% eat fat only in low amounts.
What this means, in short, is that more than 80% of the population does not have a high sugar (or fat) intake. If the SSB tax is nevertheless introduced, it will have a major impact on people who generally do not need to reduce their sugar consumption – and especially on the urban and the rural poor.
The Treasury seeks ‘evidence’ not ‘speculation’
The Treasury has rejected warnings that the SSB tax could reduce investment, growth, and jobs, saying this is simply ‘scare-mongering’. It says it will only rely on ‘evidence’, not ‘speculation’. But ‘the mathematical model’ in the Manyema article, on which it strongly relies as the rationale for the new tax, provides no evidence at all that the 20% SSB tax it urges will reduce obesity in any significant way.
If the government is serious about countering obesity, it must recognise (as McKinsey study points out) that sugar taxes are among the least effective interventions, even though it is sugar taxes that (absurdly) are given by far the most media attention.
The best way forward
Further research into the ramifications of the proposed SSB tax is needed as part of the socio-economic impact assessment system (SEIAS) that applies to all new policies and laws. KPMG, an audit firm and management consultancy, recommends that a full SEIAS investigation should be carried out so as to examine ‘the impact of the sugar tax in terms of potential unintended consequences’. Such consequences could include ‘job losses and unnecessarily burdening consumers’, while the proposed tax could also have a negative impact on poverty alleviation and the reduction of income inequality.
Says KPMG: ‘As part of the broader [SEIAS] assessment, an analysis of the effects of the sugar tax on the entire economy in terms of economic growth, household expenditure, investment, tax revenue and employment levels’ is needed too. In addition, the SEIAS assessment should investigate ‘alternative options for achieving the desired policy outcomes’.
In weighing how best to proceed, the first imperative must be to do no harm. People who consume sugar in moderate amounts are unlikely to suffer adverse health effects from their sugar intake – and this is precisely what some 80% of South Africans already do.
In addition, only one in four people are in South Africa is obese or suffers from diabetes, while healthy life years lost to these conditions stands at 2.9% and 1.6% respectively. By contrast, unsafe sex, interpersonal violence, alcohol harm, and tobacco smoking together account for more than half of all healthy life years lost.
This is not to discount the problem of obesity or to suggest that effective action against it is not needed. However, what this comparison underscores is that the damage from obesity is relatively limited, compared to that from unsafe sex and high levels of violence, among other things. This in turn means that the government has time to weigh the different options and concentrate on finding those with the best chances of success. The proposed SSB tax does not rank among these and needs to be abandoned, rather than pursued.
The government also needs to recognise that no simple quick fix for obesity is available. In weighing up possible anti-obesity interventions, it should concentrate on those identified by McKinsey as having the most impact on obesity in international experience.
As the McKinsey study further emphasises, the government also needs to acknowledge the importance of drawing all stakeholders into a broad-based initiative against obesity. A public-private partnership, as urged by the Beverage Association of South Africa (BevSA), is thus likely to have far better outcomes than the proposed tax.
The government should also recognise the success that BevSA has already had in voluntarily taking steps to reduce sugar intake through reformulation and other measures. These interventions have already achieved a greater reduction in calorie consumption than the SSB tax is likely to generate.
As BevSA notes, ‘reformulation and smaller pack-sizes’ have already brought about a reduction of ‘53kJ per litre since 2010’. By contrast, the proposed SSB tax will bring about only a marginal reduction in overall daily calorie consumption – if it in fact affects this at all, after substitution effects have been taken into account.
The key focus for the future should be on working together with the soft-drinks industry – which is willing to act further on reformulation and communication campaigns – in developing a multifaceted and far more effective strategy to counter the obesity problem.
South African Institute of Race Relations NPC (IRR) 22 August 2016