II. The Economics of Taxing Alcohol: The Example of WineFootnote ¹

The simplest way of modeling the effects of consumer taxation of wine (or beer, spirits, or any other tradable product) in a small open national economy is to use a partial equilibrium diagram, assume the country is a price-taking exporter, and examine the effects on such things as domestic prices, quantities consumed domestically and exported or imported, and national economic welfare. In Figure 1(a), it is assumed further that there are no externalities associated with producing, consuming, or trading wine, so the marginal private and social benefits (MSB_f and MPB_f) coincide as do the marginal and social costs of domestic production (MSC_f and MPC_f). If P_f is the free-trade price, then, in the absence of government intervention, O_fQ_f units are produced, O_fC_f units are consumed domestically, and C_fQ_f is exported. An ad valorem tax on domestic consumers of 100t percent (or an equivalent volumetric tax) would lower domestic consumption (and raise exports) by C_f’C_f units, raise government revenue by area acmn, but reduce consumer welfare by area admn. Hence there would be a net reduction in national economic welfare of area acd.

Source: Anderson (Reference Anderson2010).

Figure 1 The Market for Wine in a Small, Open, Wine-Exporting Economy

Figure 1(a) may well apply to the fine wine market. Indeed, it may understate the national welfare cost of such taxation if, as suggested by extensive reviews of the health science literature, moderate wine consumption can have net positive health externalities depending on the social setting. That possibility is ignored in Figure 1(a), where the demand curve is assumed to be unaffected by the consumer price of basic wine (zero elasticity of substitution between fine and basic wine).

The market for basic wine, by contrast, is illustrated in Figure 1(b), in which (i) the domestic demand curve (MPB_b) is more elastic than in Figure 1(a) because non-premium beer and spirits are assumed to be substitutes for basic wine, especially for binge drinkers simply wanting alcohol (see, e.g., Gallet, Reference Gallet2007; Srivastava and Zhao, Reference Srivastava and Zhao2010; Srivastava et al., Reference Srivastava, McLaren, Wohlgenant and Zhao2014; Yang, Zhao, and Srivastava, Reference Yang, Zhao and Srivastava2016), and (ii) the marginal social benefit curve (MSB_b) is increasingly below MPB_b because of the negative externalities on society associated with excessive alcohol consumption.Footnote ² If P_b is the free-trade price for basic wine, then with no government intervention, O_bQ_b units are produced, O_bC_b units are consumed domestically, and C_bQ_b units are exported in this example. An ad valorem tax on domestic consumers of 100t percent would lower domestic consumption (and raise exports) of basic wine by C_b’C_b units, raise government revenue by area eijk, and reduce consumer welfare by area gijk, but it would reduce the externality on the rest of the society by area ighe. Hence there would be a net improvement in national economic welfare in area ghe from this tax on basic wine.

If the tax on consumption of basic wine also applied to fine wine, the national welfare gain from the taxing of basic wine would be reduced by the welfare loss in the fine wine market, namely area acd in Figure 1(a). If both types of wine were taxed at the same ad valorem rate t (as in several countries, see below), the dollar tax per litre of wine or alcohol would be higher on the higher-priced fine wine and so increase the probability of a net loss from such a wine tax. The likelihood of an overall loss to society is higher the bigger are the ratios of P_f to P_b and C_f to C_b. One way to reduce that loss prospect is to have a lower ad valorem rate for fine wine but, if that is too politically difficult to introduce (e.g., because only richer people drink fine wine—see Srivastava and Zhao, Reference Srivastava and Zhao2010), then a change from a common ad valorem tax to a common volumetric tax would be a more covert way of achieving a similar outcome.Footnote ³

The numerous factors affecting the supply and demand curves, especially the rates of tax on non-wine beverage consumption, vary across countries and over time. There is thus no reason in principle to expect the optimal wine tax rates to be the same across countries, or to change in the same way as national income growth, urbanization, and other structural changes occur at different rates.

Also, the CTE in ad valorem (percentage) terms vary with the price of wine in those many countries with specific (volumetric) tax rates in place. If the specific tax is in local currency units per litre of beverage, the CTE per litre of alcohol also varies with changes in average alcohol content—which has been rising per litre of wine in recent years (Alston et al., Reference Alston, Fuller, Lapsley, Soleas and Tumber2015)—and with changes in exchange rates per U.S. dollar.

Countries also differ in the extent to which they are “small” in the sense of being price takers in the international market for wine or other beverages. Fine wines especially, tend to be differentiated products, so a country's export demand curve for them would be somewhat downward sloping, rather than horizontal at P_f as in Figure 1(a). Altering that assumption would not affect the above qualitative conclusions regarding the optimal consumer tax, but it would affect the outcome quantitatively for producers because the tax would shift more sales to the export market and thereby depress the price received for them. That means a tax reform that replaced a uniform ad valorem tax on all domestic wine consumption with a uniform volumetric tax (whose ad valorem equivalent was, therefore, higher than t for basic wine but lower than t for fine wine in Figure 1) would raise relative returns to producers of fine wine and hence encourage grape growers and winemakers to upgrade the quality of their products.

Consumer prices also could be raised by subsidizing wine exports, which is the equivalent of a consumption tax and a domestic production subsidy at the same rate. In the absence of any positive production externalities, excessive production would be encouraged, adding to the welfare cost of government intervention in this market.

More commonly, countries that are net importers of wine (not shown in Figure 1) can and often do tax their consumers of alcohol with a tariff or non-tariff barrier to beverage imports. Like an export subsidy, an import tariff is the equivalent of a consumption tax and a domestic production subsidy. It is applied before any excise tax is imposed on those imports. Where a nation's climate rules out any likelihood of stimulating domestic wine production, the tariff could serve as an exact substitute for a wine excise or sales tax.

The earlier analysis applies equally to beer and spirits, using Figure 1(a) for craft products and Figure 1(b) for standard products. Since the production of both beer and spirits is not climate-dependent, the use of trade instruments such as an import tariff is likely to encourage excessive production in addition to curtailing domestic consumption.

Most countries also now have a VAT or GST system. That instrument is unlikely to discourage excessive alcohol consumption if the same rate applies to all consumer products, but it is a further contribution to differences across countries in beverage retail price.

III. Empirical Methodology

There are various ways to report consumer taxes on wine and other beverages. Since measures include ad valorem taxes as well as specific taxes (either of the beverage or of alcohol per litre), and since prices to which they apply and the alcohol content per litre of beverage vary between beverage types, it is helpful for comparing across countries and over time to present the wine CTEs in two formats: U.S. dollars per unit of alcohol (to which changes in foreign exchange rates also have an influence), and as a percentage of the wholesale pre-tax price at representative price points and alcohol percentages. It is also helpful to express the taxes in relative terms, so we show as well the ratios of the tax on wine to the tax on beer and on spirits.

Specifically, we express the CTE at the following average wholesale pre-tax prices for still wines: non-premium ($2.50 per litre), mid-range commercial premium ($7.50 per litre), and super-premium ($20 per litre). Taxes are often different for sparkling wine, so their CTE is expressed separately, at $20 per litre. The alcohol content of wine in volume terms is assumed to average 12% in 2008 and 12.5% in 2018. Throughout, wine refers just to grape wine. Since rice wine has a higher alcohol content than grape wine (or beer), it is considered as part of spirits.

The beer and spirits industries are now following the wine industry in offering premium products at much higher prices than for standard products, and the production of craft beer and craft spirits is booming in many countries, albeit from very low bases (Garavaglia and Swinnen, Reference Garavaglia and Swinnen2017; Swinnen and Briski, Reference Swinnen and Briski2017; Cockx, Meloni, and Swinnen, Reference Cockx, Meloni and Swinnen2019, Reference Cockx, Meloni and Swinnen2020). The premium portion of those industries is thus still relatively small in terms of volume of sales, so it is not considered separately in what follows. Rather, the focus is on just standard-quality beer and spirits, at representative wholesale pre-tax prices of $2 and $15 per litre of beverage, respectively. Their alcohol contents in volume terms are assumed to average 40% for spirits and 4% for beer in 2008 and 4.5% in 2018.

When the CTE is defined as the percentage by which the pre-tax wholesale price has been raised by beverage taxes, that ad valorem CTE would be the same at the retail level if the wholesale-to-retail margin (like the VAT/GST) was ad valorem. If, in fact, the ad valorem equivalent of those margins is inversely related to the product price, however, then our wholesale-level ad valorem CTE will be an overestimate of the impact on consumers at the retail level. Similarly, the tax per litre of alcohol is an underestimate of the specific tax at the retail level to the extent that the wholesale-to-retail margin is positively related to the product's price and more so the higher the ad valorem VAT/GST rate.

An ad valorem estimate of the combined set of taxes paid by the retail consumer can be found as follows, assuming the domestic product is a close substitute for the imported product and the wholesale-to-retail margin is ad valorem:

$${\rm CTE} = \lpar {1 + {\rm m}} \rpar \lpar {1 + {\rm t}} \rpar \lpar {1 + {\rm v}} \rpar$$

where m is the import tariff, t is the excise tax, and v is the value-added tax. This is the method used to generate Table 3.

IV. Data Sources

The primary sources for excise tax data are the European Commission (2008, 2018) and the OECD (2008, 2018), plus national government websites. Here the focus is on the latest data (2018) and on rates a decade earlier. Data on import tariffs are from the WTO (2019) for 2018 and from the World Bank (2019a) for earlier years. Unfortunately, we do not have access to the tariff equivalent of non-tariff import restrictive measures on these beverages. Export subsidies have been used in the recent past by the European Union (EU)—and in Australia in the inter-war period, see Anderson (Reference Anderson and Aryal2015, Table 20)—but they have been only a minor source of market intervention in the EU and so are ignored here.Footnote ⁴ Some countries impose ad valorem import tariffs on beverages, while others impose specific taxes per litre of product or per litre of alcohol in the product. The specific rates are converted to ad valorem rates at specified prices and alcohol contents.

Excise taxes are shown in Appendix Tables A1 to A4. Import tariffs are in Appendix Table A5. Shares of each of four wine types in the volume of wine consumed in each country are shown in Appendix Table A6. Those shares are used to calculate the national weighted average tax on overall wine consumption. Shares of wine, beer, and spirits in the total volume of alcohol consumption are provided in Appendix Table A7: they are used to calculate weighted averages of taxes on all alcohol. The VAT/GST and nominal exchange rates are in Appendix Tables A8 and A9, respectively.