4. Partial Equilibrium Analysis
In the following we use a partial equilibrium framework to estimate the impact of mandatory biofuel blends out to 2012. We assume the introduction of a mandatory bio-fuels target behaves like a tax.
There are two elements of a bio-fuels target which do not behave precisely like a tax normally would in the context of partial equilibrium analysis. A classical partial equilibrium approach to assessing the effects of taxation would require assessing the productive and consumptive costs of the tax alongside the benefits to consumers and producers from government expenditure which draws on revenue from the tax. When we model the mandatory biofuels targets as taxes, there is no corresponding revenue to the government.7
The second element is that our reference good is consumption of transport fuels, petrol and diesel. Introduction of biofuels would, all other things held constant, result in an increase in consumption of transport fuels where the energy content of biofuels is less than the energy content of traditional transport fuels. In order to account for this effect we adopted an approach which views existing consumption of transport fuels not as consumption of petrol or diesel itself but of an input to transport as a good. As such, our base case of, for instance, 2,000,000 litres of petrol consumed, represents 2,000,000 of our transport good. When a blend is imposed we assume that all of the efficiency reduction is passed on as a price increase.8
The measurement of the impact of taxation in partial equilibrium frameworks has received considerable attention in the research literature. This research has looked at the impact in areas such as choice among consumption goods; willingness to work; willingness to save; the production pattern in society; and the use of inputs by particular industries (Diewert and Lawrence, 1994). Therefore, a well established framework exists in terms of measuring the total excess burden of a proposed tax. Figure 7 below is useful as a stylised representation of the impact of a tax in terms of efficiency.9 The market demand curve is traditionally downward sloping and we have assumed that competitive conditions exist in the market, such that the supply curve can be taken to be horizontal.10
The current (pre-tax) situation is represented by the point C, where the price is P0 and the quantity is Q0.11 At price P0 (and indeed at any price level on the demand curve) some people pay less for the product than what they would be willing to. All we reveal when we buy a product is that it is worth at least the price being charged. Thus, the demand curve shows us that there are some who would be willing to pay a price close to D, and by paying only P0, they receive a "surplus" equal to (D- P0). It is the loss of this consumer surplus that explains the excess burden (or deadweight loss) of a tax.
Figure 7: The Excess Burden of a Tax
Source: NZIER
In other words, the original value of consumer surplus (triangle DC P0) has been broken up into three parts: the area of triangle DAP1 that is still consumer surplus; the area of rectangle P1AB P0 that is tax revenue collected by the Government; and the area of triangle ACB that is lost. Thus, the area ACB is an approximate measure of the excess burden of the tax. The total burden of the tax is the sum of revenue collected and the excess burden: the area of P1AC P0.
In the case of mandatory biofuel blends the government does not collect revenue, so alternative interpretation needs to be applied for the area of rectangle P1AB P0. In this context the area of rectangle P1AB P0 is the resource cost of providing the transport fuel blend. It is the value of productive resources that is being diverted to biofuel production and blending over and above that which would be required for the production of transport fuels in the absence of government intervention in the market.
In our partial equilibrium model the change in consumer behaviour in response to the tax determines the size of the excess burden. Here, consumer behaviour is reflected in the elasticity of demand. For instance, in the figure above, if the demand curve was steeper (demand was less elastic- meaning it is relatively unresponsive to changes in price) the triangle ACB would be smaller. This means the excess burden would also be smaller and less distortion would result from the imposition of the tax. If demand were perfectly inelastic (a vertical line), there would be no distortion and therefore no excess burden. The tax would simply transfer part of the surplus being earned by consumers to the producers of bio-fuel blends.
While the triangle ACB, in Figure 7, measures the excess burden of a move from P0 to P1, what we are interested in here is a series of such movements over a period of time, as the proposed "tax" may not be "one-off" in nature. Figure 8 below shows the effect on the excess burden of subsequent increases in the tax. When the price increases from P1 to P2 the deadweight loss is now the triangle A'B'C, which includes the original triangle ACB plus the trapezium A'ABB'. What this means is that the size of the excess burden increases at a greater rate than the increase in the tax. In geometric terms, the additional excess burden as a result of successive price increases (resulting from an increase in tax) is the sum of rectangles and triangles, rather than just the triangles. In this report's estimates it is the sum of these successive price increases that is measured.
Figure 8: Excess Burden from Successive "Tax" Increases
Source: NZIER
Our estimates crucially depend on the elasticity measure used in calculating the response of consumers to an increase in price. There are three major reasons why demand for petrol/diesel would be inelastic. The first is that it is a derived demand: few (if any) consumers purchase fuel for its own sake. Fuel is used as an input into production of other goods and services, the demand for which determines the demand for fuel. As a necessary input into production, this makes the responsiveness of demand to price less than would otherwise be the case.
The second reason is that there are few available substitutes for petrol/diesel. This lack of substitution opportunities essentially means that even large increases in price will not discourage consumption of petrol/diesel, as few viable alternatives are available, at least in the short run.12 In addition, a large proportion of the diesel that would be subject to the tax is likely to be for commercial use, and demand therefore would be less elastic (i.e. less discretionary) than domestic use.
Finally, the proportion of total income spent on petrol is relatively low. This means that the price of petrol could rise considerably without greatly affecting the quantity demanded. Contrast this with for instance, a car or a refrigerator, which accounts for a larger share of total income, and hence if these prices rise, consumers are more likely to be put off purchasing such items. While there are a large number of items for which the proportion of total income spent on them is less than petrol, there are also a number where more is.
There is no singularly accepted empirical estimate of the price elasticity of demand for fuel (petrol and diesel), though it is generally acknowledged that demand is inelastic.13 In a New Zealand context, Assendelft and Gale (1989) use a price elasticity of demand for petrol of 0.18 and 0.10 for the corresponding diesel elasticity.14 These estimates are based on work done previously by Hughes et al, at the then Ministry of Energy.
Assendelft and Gale posit that in the time between the work of Hughes et al and their work, some 9 years, the effects of structural change and the large reductions in the real price of petrol may mean that demand for both petrol and diesel is more inelastic than even these estimates. They cite an overseas study that estimates an average international price elasticity of demand for petrol of 0.11. More recent government energy forecasts in New Zealand have used long run price elasticities of demand of 0.19 for petrol and 0.13 for diesel, with corresponding income elasticities of 0.62 and 1.26.15
Goodwin (1992) and Oum et al (1992) present evidence of the low demand elasticity for automobile transport, as well as the consumption effects of price changes in petrol. Both these studies confirm that demand elasticities for transport (automobile) and demand for petrol to fuel that transport are low. Goodwin (1992) however, surveys empirical estimates of petrol price elasticities (generally thought to lie in the range 0.1-0.4) and finds that these traditional estimates may understate the true elasticity. He estimates the unweighted mean value of 120 elasticities of petrol consumption with respect to price is 0.48. This is still regarded as evidence of an inelastic demand, despite the apparent rise in elasticity relative to conventional wisdom.
For the sake of comparability with MED's forecasts of mineral fuel consumption we have utilised the estimates of short run price elasticities published in their Energy Outlook publication (MED, 2003). These elasticities are -0.05, for petrol consumption and -0.08 for diesel consumption.
4.1 Baseline Forecasts
Figure 9: Transport Fuel Consumption Forecasts: Millions of Litres of Diesel (RHS) and Petrol (LHS)
Source: NZIER
In order to assess consumptive and productive impacts using our partial equilibrium framework we have forecast consumption for each year out to 2012 and assessed impacts in terms of their deviation from baseline forecast in each year.
In producing our forecasts we have used MED's forecasts of transport fuel consumption, adjusting them as we deemed necessary given the time since the last Energy Outlook was published in October 2003.
Our forecasts have been split between diesel consumption and petrol consumption. Figure 9 shows our forecast for the path of petrol and diesel transport fuel consumption out to 2012 in millions of litres of fuel consumed.
On top of these forecasts we have assumed a constant conversion of fuel to CO2, using conversion factors from MED (2004). Different factors apply for regular and premium petrol and diesel as follows:
- Regular petrol emits 66.2 kt of CO2 per PJ of fuel consumed.
- Premium petrol emits 67.0 kt of CO2 per PJ of fuel consumed.
- Diesel emits 69.5 kt of CO2per PJ of fuel consumed.
Because we have not discriminated between regular and premium petrol in our analysis we have taken an average of the conversion factors for regular and premium petrol weighted by the proportion of fuel used of each type in New Zealand in the year to June 2004. This yielded a conversion factor of 66.38 kt of CO2 per PJ.
Our baseline forecast of CO2 emissions is presented in Figure 10.
Figure 10: BAU Forecast of Transport CO2 Emissions by Source: Kilo Tonnes of CO2
Source: NZIER
4.2 Assumptions
Like all modelling work, these estimates required some assumptions to be made. In the interests of tractability, we assumed that the full amount of the tax is passed on to consumers in the form of higher prices.16
We also assume the elasticity of demand for petrol/diesel is uniform across all consumer types. In reality, demand responsiveness is likely to differ across sectors and different types of consumers. Despite this, we believe the approach to be valid for a broad level assessment of what are the economy-wide implications of the "tax".
Finally, we have forecast the deadweight loss based on total transport diesel and petrol consumption. This ignores the fact that some consumers of transport fuels and a large number of diesel consumers are in fact businesses who will, typically pass the costs of fuel price increases on to final consumers/households. As such, the most appropriate analysis to conduct would have been to separately assess the impact of transport fuel price increases on consumers and on producer output prices. However, this would have required a time consuming analysis of the elasticities of consumers for the full range of consumer goods. Furthermore, it would have required a parallel assessment of price pass through by industry and by product. This sort of analysis is possible in the context of a general equilibrium model, but not feasible for the sort of partial equilibrium analysis conducted above.
Our "second-best" approach to evaluating the dead-weight loss from mandatory biofuel blends would be expected to underestimate the full extent of the deadweight loss. This is because consumers' demand for goods other than transport fuels is typically less inelastic than for transport fuels. As such, increases in fuel prices passed on to consumers through prices of other goods would result in greater deadweight losses than have been estimated here.
Indeed, the literature suggests that our estimates of the deadweight loss are, due to methodological issues, on the conservative side. Creedy (2004, 2003) provides evidence that the excess burden or deadweight loss from taxation increases by a magnitude over and above that which we have estimated. Indeed Creedy suggests that a doubling of "tax" rates results in a (approximate) quadrupling of the excess burden or deadweight loss imposed by the tax.
Back to Top