Policymakers need to be cautious about evidence from fiscal-food policies, John Gibson writes.
Fiscal-food policies use taxes to alter relative prices for food and drink to induce a shift to healthier diets. Discussion of such taxes implies that countries in the Pacific can ‘have their cake and eat it too’ because it is claimed these taxes benefit both health and fiscal revenue.
Traditionally, economists see trade-offs between these objectives. Efficient items to tax from a revenue standpoint have own-price inelastic demand; the pre-tax and post-tax equilibrium quantities are similar so taxes entail little distortion. Thus, an unhealthy food is unlikely to be efficient to tax for both health and revenue reasons, since one objective needs own-price elastic demand and the other needs inelastic demand.
Yet belief in elastic demand for unhealthy food and drink pervades discussion of fiscal-food policies at the highest levels. For example, the World Health Organization (WHO) advocates taxes on sugar-sweetened beverages at a rate high enough to raise the retail price of these drinks by at least 20 per cent. WHO suggests this will cause proportional reductions in consumption, implying an own-price elasticity of quantity demand of minus one for these drinks.
However, most elasticity estimates used to support fiscal-food policies are exaggerated. They fail to account for consumers downgrading quality when prices rise. If quality responses are ignored, they wrongly get bundled in with any quantity responses, overstating the rate at which price rises can moderate intakes.
These overstated elasticities are studied in our recent paper in Asia & the Pacific Policy Studies that looks at the demand for soft drinks in Papua New Guinea and the Solomon Islands.
Taxes on sugar-sweetened beverages (or ‘soda taxes’) are the most frequently discussed fiscal-food policy and have recently been introduced in several countries, such as Mexico.
A problem with most elasticity studies is they use calculations for standard, undifferentiated goods. Yet they use household survey data, which cover a mix of different brands, package sizes, container types and so forth.
For example, in a single supermarket in Port Moresby, one can find 33 different specifications of fizzy drink, with a 5:1 range in price per litre from the dearest to the cheapest. In contrast, the dearest and cheapest areas of Papua New Guinea have only a 2:1 price difference, when doing a like-with-like comparison of soft drink prices.
The standard approach to price elasticity estimation relies on price variation across areas, and sometimes over time, but ignores quality-related price variation seen on the shelves of any store.
The scope for bias due to ignoring quality is even worse outside PNG and the Solomon Islands, because most other countries have much less inter-area price variation, due to having better infrastructure and easier topography. Moreover, in most countries, there is a far wider range of quality variation within soft drinks. For example, the supermarket nearest my home has 160 different fizzy drink specifications with a 15:1 price ratio from dearest to cheapest.
About one-third of the consumer response to soft drink price variation in the Solomon Islands is on the quantity margin, while two-thirds is on the quality margin. In Papua New Guinea, consumer responses are split evenly between the quantity and quality margins.
If the quality response is wrongly treated as a quantity response to price – as it is in most of the literature – the elasticity of soft drink demand with respect to own-price is exaggerated by a factor of two in Papua New Guinea and a factor of three in the Solomon Islands. Thus, any fiscal-food policy taxes based on the wrong elasticities will be far too optimistic in expecting that it would just take small taxes to reduce fizzy drink consumption.
Where spatial price variation is less than in these two Melanesian countries, and product variety greater, the bias from ignoring quality will be even greater. For example, when the own-price elasticity of quantity demand for soft drinks in Mexico is estimated in a way that allows responses on the quality margin, it shrinks from around −1.2 to just −0.3. This causes forecast health benefits of Mexico’s soda tax to shrink by a factor of four.
In order to give policymakers more reliable evidence on price elasticities, researchers should work with databases that have measures of food and drink demand (either quantity or budget shares), measures of quality (such as the unit value, which is the ratio of expenditure on a food or drink group to the quantity purchased), and measures of market prices.
Surprisingly few countries have the required data. In the absence of these comprehensive databases, and of more reliable elasticity estimates, policymakers should be cautious about the evidence produced by advocates in support of fiscal-food policies.
This piece is based on the authors’ paper in Asia & the Pacific Policy Studies, Fiscal-food policies are likely misinformed by biased price elasticities from household surveys: evidence from Melanesia. The full paper is free to read and download.