The Effects of Pollution and Energy Taxes across the European Income Distribution. Economics Department, Keele University, UK

Transcription

1 The Effects of Pollution and Energy Taxes across the European Income Distribution E.J.Symons, S.Speck and J.L.R.Proops* Economics Department, Keele University, UK Regional Environmental Center for Central and Eastern Europe, Budapest, Hungary *School of Politics, International Relations and the Environment, Keele University, UK Introduction Recent proposals from the European Union are designed to promote environmental improvement, or at least reduce the continuation of environmental decay. One such proposal was for a carbon/energy tax in The latest proposal on the EU level was for re-structuring the taxation of energy products; i.e. to increase the excise duties on hydrocarbon oils and to introduce minimum excise duties for other energy products (natural gas, electricity, coal, etc.). This paper examines the likely impact of such environmental taxes on households in a number of European countries, and analyses how the effects differ across the income distribution; i.e. we consider the equity implications of EU environmental taxation. Much air pollution (e.g. CO 2, SO 2 and NO x ) derives from the burning of fossil fuels, and it is on this fossil fuel-pollution nexus that we concentrate in this paper. Now energy and (air) pollution taxes will affect the price of fossil fuels and, hence, consumer prices, both directly for fuels and indirectly for manufactured goods. These price changes will, in turn affect the effective purchasing power of households, unless compensating changes in other taxes are made. Our analysis is conducted in three stages. First, we use an input-output framework to assess the likely impact of pollution/energy taxes, via increases in the costs of using fossil fuels, upon the prices of consumer goods. Second, we investigate the expenditure patterns of consumers. Third, the price changes deriving from pollution/energy tax increases, are linked to consumer demand, and the implication for consumers is estimated in terms of the extra tax paid. A key feature of our analysis is that we are able to examine how this extra tax payment is distributed across the income distribution, so the effects of pollution/energy taxes on equity can be addressed directly. EuroStat data 1 22/02/01

2 In this paper we assume that consumers do not respond to the change in relative prices, so the estimated effect is that directly impacting on consumers. Extending this approach, by incorporating behavioural responses by consumers to the increased prices, is a topic for further work. Our analysis extends the existing literature in a number of directions. Barker and Köhler (1997) analyse consumer expenditure on environmentally sensitive goods for EC Member States, and investigate the implications of increasing excise duties on these consumption goods. Our paper goes beyond their analysis, by incorporating indirect effects of environmental taxation on producers, with the associated effect on consumer prices. Symons, Proops and Gay (1994) and Cornwell and Creedy (1996) examine the distributional effects of carbon taxes on the UK and Australia respectively, using micro-simulation analysis. The current paper applies the ideas in these latter papers to five EU countries, namely Germany, Italy, Spain, France and the UK, using grouped household expenditure data. Energy Use and Pollution Emissions by Households and in Production In order to examine the scope for implementing such pollution and energy taxes in the EU, one needs a modelling framework detailing how fossil fuels and energy are used in an economy, and how they relate to the corresponding pollution emissions. The incidence of a pollution/energy tax will be related to consumer behaviour in two ways. Firstly, individuals purchase fossil fuels and energy directly (e.g. petrol, heating oil, coal, gas) which will carry the pollution/energy tax; this household purchasing of fuels constitutes directconsumption demand. Secondly, households purchase goods, the production of which has entailed the use of fossil fuels and energy (e.g. motor car manufacture entails large amounts of fossil fuel use in the production process); this constitutes indirect consumption-demand. This too will carry with it a certain incidence of the environmental tax. The indirect consumption-demand for fossil fuels, and the corresponding energy use and air pollution, can be identified with the derived demand for these fuels through production; thus we may identify the indirect consumption-demand with the 'production demand' for fossil fuels. This can be further subdivided into direct and indirect production demand through the recognition of inter-industry trading. Do we have to describe in more detail the difference between direct and indirect production demand? Our analysis therefore seeks to identify the demand for fossil fuels and energy by producing sector and by fuel, decomposed into the direct consumption-demand, the direct production-demand and the indirect production-demand. In this way, we can effectively impute all EuroStat data 2 22/02/01

3 fossil fuel and energy demand to the final demand for goods and services, and identify the content of any energy/pollution factor in the goods purchased. These energy/pollution intensities will be used to assess the impact on consumer prices of a tax on energy use and pollution related to the use of fossil fuels. The ideal tool for such sectoral analysis, and decomposition of fuel use and pollution, is inputoutput analysis (Gay and Proops, 1992; Proops et al., 1993). The technique has been successfully applied in Symons et al. (1994) for the analysis of taxes on carbon dioxide emission for the UK. The effect of a pollution tax will be to raise the cost of manufactured goods delivered to final demand (i.e. consumers). The increase in price is determined by the pollutant content of the good and the incidence of the tax. The usual short-term assumption in such an exercise is that the ultimate tax burden is on consumers, i.e. producers pass on any tax related increase in cost in the form of higher prices. This could be a critical assumption in the international context, where it is more likely that some of the burden of the tax will be borne by the owners of energy resources rather than the energy users. One might imagine, for example, a scenario where an energy or CO 2 tax placed on a single country would lead to a loss of competitiveness if passed on fully in the form of higher prices. In this situation firms may have an incentive to absorb the tax in the form of lower profits thereby sharing the incentive of the tax between consumers and producers. This requires investigation in more detail (Proops et al., 1993:chap. 8). It should be noted that the use of an input-output framework does not directly allow for substitution possibilities in production (though see Proops et al. (1993: chap.11) for a discussion of the effects of altering input-output coefficients). This gives rise to two opposing biases in the model. If substitution possibilities do exist in production, but are omitted by the analysis, the post tax pollution level is overstated, since the tax would cause a shift of techniques such that goods produced were less pollution intensive. However, substitutability in production would also reduce the incidence of the tax on consumer prices, and therefore on consumer behaviour, thus lessening the effect of the tax on pollution reduction. The overall effect of a tax with substitution in production would therefore be to reduce the pollution intensity of the goods but also to cause a smaller change in demand for these goods than in the no substitution case. The results presented in this must be regarded as being derived from a short-term model where substitutability in production is limited. Pollution and Energy Intensities EuroStat data 3 22/02/01

4 While the calculation of pollution intensities can be accomplished for a range of air pollutants, we concentrate on CO 2, as this is the gas currently most discussed for targeting as part of EU environmental reform. Details of the method of calculation of these intensities are to be found in Gay and Proops (1992), and Proops et al. (1993: chap. 8). The sources used for the input-output tables, and the energy and pollution data for these calculations, were as follows. Germany: Statistisches Bundesamt (1994a, 1994b, 1995). Italy: Istituto Nazionale Di Statistica (1988a, 1988b). Spain: Instituto Nacional de Estadistice (1990, 1991). UK: Central Statistical Office (1993), Department of Energy (1990, 1991, 1993). For France, because directly comparable data was not available, the French calculations were based on modified German and UK data (see Appendix for details). Table 1: Pollution and Energy Intensity of Commodity Groups Commodity Group CO 2 Intensity (kg CO 2 /ECU) Energy Intensity (joules/ecu) Italy UK Spain France Germany Italy UK Spain France Germany Food Alcohol Tobacco ??? ??? Clothing Hh Services Fuel Hh Goods PGS Transport Vehicles Petrol L Goods L Services Housing Table 1 shows the total (including direct consumption and production) intensity figures for CO 2 and energy for 15 commodity groups for the five countries we consider. The group intensity figures are calculated as weighted means of 32 more disaggregated components. The weights are calculated from each country s expenditure and represent the average expenditure weight across all household in the country, and are expressed respectively as kg or joules per ECU spent (at June 1996 prices). EuroStat data 4 22/02/01

5 Experiments indicated that these do not differ markedly across the income distribution so the approximation is quite good. Separate categories for alcohol and tobacco are not available for Germany, so the figures given for alcohol represent a combined intensity. There are some interesting differences across the countries. Notably, energy intensities for Germany are higher than for the other countries. Germany has high energy intensity for all components of fuel (Elec, Gas, Oil, Coal). Spain has a high energy intensity for gas but this forms a small proportion of total fuel use. Italy s energy fuel intensities are all low except for coal so overall, Germany is the outsider. The other countries have different intensities for fuel components and different expenditure weights but, when combined, the fuel intensities are similar. This pattern is repeated in CO 2 intensities. Other consumption goods that have high energy and CO 2 intensities that are not shown separately in the table are china/glass, communication, plastic products, transport and public transport. The Effects of a Tax Imposing a tax on energy and CO 2 emission will affect consumer prices, p i,via the intensity figures in the following way: p i = t i = αi i. Here: t i is the incidence of the tax on the ith commodity group; α is the tax on pollution (ECUs/kg or ECUs/joule); I i is the pollution intensity for group I. Thus t represents the change in price in our analysis, since there are no exogenous price changes. Therefore, an energy or CO 2 tax will increase the price of goods according to their energy or CO 2 intensity, and the amount of tax paid will be t i *expenditure i on each commodity group. Household Expenditure Data The paper uses budget survey data from EuroStat (1992) for the analysis of the distributional effects for France, UK, Spain and Italy. This data provides the distribution of expenditure patterns across different equivalent expenditure groups (i.e. total expenditure for each household is divided by the equivalence scale for that household). This is subsequently divided by the average equivalent expenditure, to yield a ratio indicating how that household compares with the average. Households are ranked according to this relative expenditure level and the mean expenditure pattern reported for EuroStat data 5 22/02/01

6 6 groups within each group (groups generated according to relative expenditure in the following levels: >0.4, , , , , >1.6). No directly comparable data is available for Germany and we use the German survey (Statistisches Bundesamt, 1994). This data provides the expenditure patterns for households according to their absolute income not their relative expenditure. Smith (1992) indicates that, at least for the UK, distribution results are the same for ranking by income or expenditure or equivalent expenditure so general comparisons across can be made despite the difference in ranking method and the fact that EuroStat data is for adult equivalents. An important limitation to the EuroStat data is that neither the number of households in each survey, nor the proportion in each expenditure group, are given. We do not know how representative of the nation the data is, but Barker and Köhler (1997) imply that the EuroStat categories concentrate on low expenditure groups and bunch up to 30% of the households in the top category. Our analysis requires total expenditure for each income/expenditure group as well as budget shares. In this paper we estimate annual total expenditure for each group by finding the proportion of average total expenditure in all households that each group attains from the national budget survey. Further work will make more extensive use of national budget surveys, where available, to overcome this shortcoming. Köhler, Luchmann and Wadeskog (1999) give a lengthy discussion of the similarities of national consumption patterns and how budget shares differ across income groups for energy and vehicles. The main point of interest is that low expenditure groups tend to spend a larger proportion of their budget on household fuel 1 (the UK is an exception to this which differs from the expenditure patterns observed using the national household survey, Family Expenditure Survey). For example, in France the lower income group spends 10% of their income on fuel, this falls as equivalent income rises to 3% for the highest income group. The corresponding figures for Italy are 7.9% and 3.8%; Spain 5% and 2.6%; and 10% and 4% for Germany. Average expenditure shares differ across countries. Notably, the UK devotes a higher proportion of the budget to household fuel (6.3%) than the other countries examined in this paper (Germany 6%; France 5.7%; Italy 5.5%; Spain 3.4%). The proportion spent on petrol is increasing across the income distribution for all countries (except for the highest income group) and the amount differs across countries, eg. UK (4.2%), France 1 These figures differ from those in Köhler et al because their figures are budget shares for all expenditure; these exclude some categories mainly.. EuroStat data 6 22/02/01

7 (4.9%), Italy (6.4%), Spain (4%) and Germany (2.5%). These differences in expenditure patterns all contribute to the differing effects on each country of energy/carbon taxes but are not the sole factor determining whether pollution taxes will be regressive or progressive. The Impact of Energy/Carbon Taxes on Households Welfare Two fiscal reforms that are indicative of a general pattern for any environmental taxes that may be introduced are considered in this paper. Potential policy issues will be considered in more detail in future work. The two reforms examined are: (i) a CO 2 tax of 0.1ECU per kg emission of CO 2 for each country; (ii) an energy tax that raises the same revenue as the CO 2 tax within each country. This paper analyses the first round effects of increased taxes across the income distribution. It does not consider what the government may do with the revenue generated from the tax. Obviously these revenues could be used to mitigate any regressive effects but, in order to do this, we need to know the extent of the regressivity of the tax without any additional effects. The increase in price is calculated for each commodity group. For EuroStat data, consumers expenditure is calculated as the weight multiplied by total expenditure in ECUs published in EuroStat for the survey year. This is reflated to 1996 prices by the national consumer price index. German expenditure figures are similarly converted to ECUs using the official exchange rates used by EuroStat and then reflated to 1996 prices. Total expenditure and budget shares are assumed fixed in each reform, so the price elasticity is implicitly assumed to be unity. Any increase in price is exactly offset by a reduction in quantity, so the expenditure on each commodity group is fixed. Thus the increase in price caused by the tax imposed is a money metric measure of the loss to welfare of the household. Figure 1 Increased Tax Burden for Germany % total income range co2 energy EuroStat data 7 22/02/01

8 Figure 1 shows the effect of each tax on German households. The horizontal axis shows the income group: Total is the average for all households; 1 is the average for households with income below 1000DM/month; 2 for DM; 3 for DM; 4 for DM; 5 for DM; 6 for DM; and 7 for monthly income between DM. The vertical axis shows the increase in tax paid as a percentage of total expenditure. Both taxes are regressive, energy taxes slightly more so than CO 2. Overall households pay over 6%. This is about 8% for the lowest income group and falling to just above 5% for the highest. Figure 2 Increased Tax Burden for Carbon Tax 0.1ecu/kg 14 % total < >1.6 Relative Equivlent income group UK France Italy Spain Figure 2 shows the effect of the CO 2 tax on the other countries. The horizontal axis shows the equivalent expenditure relative to the mean of all households group: Total is the average for all households. The vertical axis is the same as before. The average tax burden is much higher for the UK and Italy at 10% and Spain at 8%. The tax is regressive for France and Spain (though only slightly for Spain) but neutral for Italy and progressive for UK except for the highest income group. The effects of the equivalent energy tax are shown in Figure 3. The results show the same general picture as the carbon tax. The range of results, from progressive to regressive, are in line with those found by other techniques (Smith 1992, Pearson 1995). EuroStat data 8 22/02/01

9 Figure 3 Increased Tax Burden for Equivlent Energy Tax % UK France Italy Spain total < >1.6 Relative Equivlent income group The results need to be looked at in more detail to investigate what contributes to these differences. Looking at each country in turn we can see which commodity groups contribute to the progressivity/regressivity of the overall result. Germany: the additional tax on fuel accounts for over 70% of the overall increase in both taxes while only about 50% for the average of all households. This is because fuel accounts for 10% of the low income households total expenditure but only 6% on average. This dominates the progressive nature of the tax on other commodities, e.g. petrol, household goods, clothing and leisure goods and services. Furthermore the energy tax is slightly more regressive than the carbon tax and the richer households do relatively well under the carbon tax. This reflects the relative differences between energy and carbon intensities. UK: Expenditure patterns do not follow any simple progressive or regressive pattern, so neither do the increased tax burden for commodity groups. Fuel forms a high proportion of household expenditure but this increases with equivalent income until the highest income group and then drops dramatically. The increased tax burden does the same. The increased tax on food contributes regressively overall but this effect is offset by all the other main commodity groups which are initially progressive at low income groups but the highest income group always pays proportionately less. Spain and Italy: In both countries petrol and clothing contribute progressively to the overall tax but the effect is dominated in Spain by the regressivity of fuel and food although it is not so apparent as in Germany. The poorer households in Italy do much worse than those in Spain because they spend a higher proportion of their income on fuel and the intensity for fuel is approximately the same in EuroStat data 9 22/02/01

10 both countries. However, the intensity for petrol in Italy is approximately double that in Spain so the progressivity of expenditure is more important in determining the overall effect for each income group and outweighs the regressivity of the tax on fuel. France: The main sources of increased tax are fuel and petrol. Petrol expenditure weights are similar across all income groups so the regressive nature of fuel dominates accounting for 60% of the tax increase at the bottom of the income distribution and 32% at the top. Overall, the results do not indicate that pollution taxes will have the same effect on all European countries. The results for the UK are not consistent with Symons et al. (1994) and Cornwell and Creedy (1996) for Australia. Both these papers show an increase in inequality following the imposition of a carbon tax. Here we can only examine the mean tax burden for each group since we are not using household survey data, but our results here indicate that inequality is decreasing. The cause for this is likely to be different data and the fact that the bottom income groups are overrepresented in EuroStat. An investigation to examine these differences in detail is the subject for further work. Barker and Köhler (1997) is not directly comparable, because they allow disposable income to increase through reduced national insurance contributions. As a result, total expenditure is not fixed. Also, budget shares are able to respond to the tax changes, so it is difficult to see whether the pollution taxes are regressive or progressive. Conclusions This paper examines the impact of an energy or CO 2 tax on five European countries, France, Italy, Germany, Spain and the UK. This involved estimating the energy and CO 2 intensity for consumer expenditure categories, including both the intensity for direct consumption of the good and the intensity reflecting energy use or CO 2 emissions generated during production. A tax on energy use or CO 2 emissions will raise the price of consumption goods in direct relation to the intensity of that good. Consumers therefore face an increased tax burden from consumption. This will vary according to the proportion of total expenditure allocated to each good and will differ across the income distribution. The paper estimated the increase in tax burden using the EuroStat Family Budgets household consumption data based on surveys carried out in 1988 and a comparable German data set. The results do not show similar patterns emerging. In Germany, France, and slightly in Spain the imposition of the taxes was regressive. This was not the case for the UK and Italy. Further work needs to be done to investigate the cause of this using national household surveys to allow full income distribution and different household types to be examined, and by EuroStat data 10 22/02/01

11 allowing consumer behaviour to respond. This will allow a full distributional analysis to be performed. EuroStat data 11 22/02/01

12 Appendix The unavailability of suitable French input-output data led to the French energy and carbon intensities being proxied as related to the German and UK intensities. The proxying process was in two stages. First, the German and UK intensities were averaged for each commodity group, to provide a first (rough) approximation of the French intensities. Second, the first estimate was refined, so that when these intensities were multiplied by the corresponding final demands for each commodity group, and then summed, the corresponding total energy use (or carbon emissions) were correct for the French economy (these figures were known). To make this adjustment, the first estimates for the French intensities were multiplied by the ratio: R = 1 2 E E Y F G G / Y F E + Y U U See Eurostat (1997) Yearbook 1996, Luxembourg. Here: E X is the energy/carbon for country X. Y X is the GDP for country X. F represents France, U is the UK and G is Germany. EuroStat data 12 22/02/01

13 References Barker.T, J.Köhler, (1998) Equity and Ecotax Reform in the EU: Achieving a 10% Reduction in CO2 Emissions Using Excise Duties, Fiscal Studies 19 (4) Central Statistical Office (1993) Input-Output Tables 1989, HMSO, London. Cornwell.A, J.Creedy, (1996) Carbon Taxes, Prices and Inequality in Australia, Fiscal Studies 17 (3) Department of Energy (1990, 1991, 1993) UK Digest of Energy Statistics. HMSO, London. Department of Trade and Industry (1989) Purchases Inquiry. HMSO, London. EuroStat (1992) Family Budgets: Comparative Tables Belgium, Germany, Spain, Italy, Portugal. EuroStat Publications, Luxembourg Gay, P.W., J.L.R. Proops (1992) Carbon dioxide production by the UK: an input-output assessment, Applied Energy 44 (2) Istituto Nazionale Di Statistica (ISTAT) (1988a) Tavola intersettoriale dell economia a 44 branche. Rome. Istituto Nazionale Di Statistica (ISTAT) (1988b) Consumi di energia 44 brancha. Rome. Instituto Nacional de Estadistice (INE) (1990) Cuentas Nacionales y Tabla Input-Output Contabilidad Nacional de Espana, Madrid. Instituto Nacional de Estadistice (INE) (1991) Tabla Input-Output de la Energia de Espana Contabilidad Nacional de Espana, Madrid. Köhler, Luchmann and Wadeskog (1999) Expenditure on Environmentally Sensitive goods and Services: Household Spending in Europe Working Paper 9903, Department of Applied Economics, University of Cambridge. Pearson, M. (1995) 'The Political Economy of Implementing Environmental Taxes', International Tax and Public Finance 2: Proops, J.L.R., M. Faber, G. Wagenhals (1993) Reducing CO 2 Emissions: A Comparative Study for Germany and the UK. Springer-Verlag, Heidelberg. Smith, S (1992) The Distributional Consequences of Taxes on Energy and Carbon Content of Fuels, European Economy Special Edition 1992 No. 1 the Economics of Limiting CO 2 Emissions EuroStat data 13 22/02/01

14 Statistisches Bundesamt (1994a) Einkommens- und Verbrauchsstichprobe (Income and Consumption Survey) Fachserie 15, Heft 5: Aufwendungen privater Haushalte fur den Privaten Verbauch, Wiesbaden Statistisches Bundesamt (1994b) Volkswirtschaftliche Gesamtrechnungen: Input-Output-Tabellen. Fachserie 18, Reihe 2, Metzler Poeschel Verlag, Stuttgart. Statistisches Bundesamt (1995) Umwelt: Umweltökonomische Gesamtrechungen - Material- und Energieflußrechnugen. Fachserie19, Reihe 5, Metzler Poeschel Verlag, Stuttgart, Symons,E.J, J.Proops, P.Gay, (1994) Carbon Taxes, Consumer Demand and Carbon Dioxide Emissions: a simulation analysis for the UK, Fiscal Studies 15 (2) EuroStat data 14 22/02/01