The report, by the forecasting group Oxford Economics, concluded that adding a levy of 18p to 24p per litre to sugary drinks would have a negligible effect on calorie reduction but significant costs in terms of industry job losses. These findings were widely reported, with a coalition of other businesses subsequently joining the British Soft Drinks Association in opposing the levy.
Worryingly, many media outlets failed to disclose that the study was produced for the British Soft Drinks Association. But of greater concern is the way that it underestimates the benefits of the levy and overestimates its costs, potentially misleading public debate.
One of the main arguments put forward in the study is that the levy would only reduce sugar consumption by five calories per person per day. There are reasons to question this. Cross-price “elasticities” are used to calculate how the consumption of some goods are affected by a change in the price of others. The study takes its estimates of these from a recent paper in the British Medical Journal which looked at the market demand for different drinks.
However, in copying these estimates it ignored those elasticities that were statistically insignificant, thus distorting the results. It should have been acknowledged that the consumption of water and diet beverages could benefit from a hike in the price of sugary drinks, meaning that the study’s conclusions most likely underestimate the extent to which consumers would substitute sugary drinks for sugar-free alternatives. In addition, the study ignored the standard errors of these elasticities. This downplays the enormous amount of uncertainty associated with this kind of modelling.
Allied to this, the study takes no account of the signalling effect to consumers of the taxation – only the price effect. In other words, people may consume fewer sugary drinks not just because the price has gone up but also because the perception of them as “unhealthy” has been strengthened. Empirical research into the effect of taxation has shown that, in some instances, consumption of products such as alcoholic drinks do appear to have decreased beyond the level expected by the price increase alone.
And as pointed out by BBC Reality Check, the study assumed that no further changes would be made to the formulation of soft drinks, which is at odds with what the industry is currently doing. If the existence of the levy leads to more drinks being reformulated to contain less sugar (and thereby avoid some or all of the tax) then total sugar consumption from sugary drinks should fall even in the absence of product substitution.
The health hit
Not only may total sugar reduction be greater than predicted in the study, the beneficial health impacts could be better too. First of all, the reduction in total sugar consumption anticipated in the study is averaged out across the whole population. This ignores the fact that consumption of sugary drinks is concentrated among teenagers.
According to the National Diet and Nutrition Survey 2008-2012, people aged 11 to 18 consume on average 21g of added sugar per day from soft drinks. This is more than double the amount consumed by people aged between 19 and 64. For this reason the “five calories per person per day” conclusion is misleading. It will be much higher for teenagers – a key target group of the levy.
Moreover the study assumed that the only intended health effect of the levy is to help prevent obesity. But a reduction in the consumption of sugary drinks will also help prevent tooth decay and could reduce the incidence of Type 2 diabetes as well (independently of its effects on body weight). Tackling these two diseases has long been part of the levy’s purpose, as outlined by Public Health England, which formally proposed the tax last year.
Lastly, the study also treats all sugars the same. Naturally-occurring lactose in milk and fructose in processed juices are counted the same as the refined white sugar (sucrose) added to energy drinks and carbonated drinks. Since the study argues that consumers will switch from energy drinks and carbonated drinks to milk and juices, the overall calorie reduction seems limited. From a dietary point of view, however, milk and juices offer additional benefits (for example, vitamins and minerals, and no added caffeine) which these other drinks do not. Encouraging this switch may be considered a benefit of the levy independently of its effect on calorie intake.
One of the perceived risks of the levy is that it will lead to economic contraction and unemployment. The study estimates that lower sales will reduce the industry’s GDP contribution by £132m and result in 4,000 job losses, a figure subsequently cited in the headlines of City AM and The Sun among other newspapers.
However, the employment effects it examines are confined to the soft drinks industry and exclude the dairy industry, which, according to study itself, stands to benefit from a 3.7% increase in the volume of milk sold. Based on the figures provided in the study, if the effects on the dairy industry are included then this would reduce the GDP and employment effects by around three quarters – in other words, the predicted loss in GDP would be £33m and the reduction in employment more like 1,000 jobs. Taking into account the likely increase in water and diet drinks sales would further reduce the predicted losses, possibly even turning them into a gain.
As noted before, there is also a degree of uncertainty attached to forecast modelling, which again is not adequately communicated. Evidence from other countries has shown that when it comes to the effects of “health taxes” on employment, the taxes tend to be outweighed by other factors affecting the revenue, profitability and employment prospects of the industry in question. Anticipating all these in advance is inherently difficult.
Acknowledging this might have made the “can the tax” coalition more cautious when claiming about the levy that “we know that it will cause the loss of over 4,000 jobs”. Regrettably, as often happens when research becomes yoked to interest group lobbying, such nuance is all too easily lost.