|
Article Excerpt
The proper design of price interventions in energy markets requires consideration of equity and efficiency effects. In this paper, budget survey data from 29,000 Indonesian households are used to estimate a demand system for five energy sources, which is identified by the spatial variation in unit values (expenditures divided by quantities). We correct for the various quality and measurement error biases that result when unit values are used as proxies for market prices. The price elasticities are combined with tax and subsidy rates to calculate the marginal social cost of price changes for each item. The results suggest that even with high levels of inequality aversion there is a case for reducing the large subsidies on kerosene in Indonesia, supporting the reforms that have been announced recently.
1. INTRODUCTION
Energy demand is rising rapidly in developing countries. The energy pricing policies of those countries are therefore increasingly important to the efficient use of the world's energy supply. Some developing country energy markets are highly distorted by consumer subsidies (IEA, 1999). For example, the Government of Indonesia spent over US$13 billion dollars on consumer fuel subsidies in 2005. These subsidies have a major effect on the overall energy balance in Indonesia because households account for about 45 percent of total energy consumption. There also are large fiscal effects, with about one-quarter of the government budget (and about five percent of GDP) going on fuel subsidies (Sen and Steer, 2005).
Dramatic reforms have been attempted by the Indonesian government in response to this escalating cost of fuel subsidies. In October 2005 the subsidised price of kerosene was raised 186 percent, from Rp 700 per litre (US 7 cents) to Rp 2000 per litre (US 19 cents). The prices for diesel and gasoline were raised by approximately 90 percent, following on from increases of 30 percent in March 2005 (Table 1). These earlier price increases did not apply to kerosene. Moreover, fuel subsidies were timetabled for a complete phase out by either the end of 2006 (gasoline and diesel) or the end of 2007 (kerosene). These energy subsidies are meant to be replaced with a set of targeted subsidies, whose benefits are to be designed so that they are restricted to low-income groups (Kompas, 2005; Jakarta Post, 2005).
It is unclear whether these ambitious plans for reform will be realised. First, despite the substantial price rises enacted in 2005, there is still a long way to go if Indonesian fuel prices are to be set at world levels. The kerosene price in October 2005 was only 31 percent of the world price, while gasoline and diesel prices were about two-thirds of the world level (Table 1). Second, many previous attempts at reforming energy price policy in Indonesia have failed because of the resulting political difficulties. Attempted reforms in 2003 were reversed after widespread protests while the price rises in 1998 are believed to have precipitated the downfall of the Suharto regime (BBC, 2005; Economist, 2005). Moreover, these subsidies have been long-term features of the Indonesian economy, dating back to the mid-1970s (Dick, 1980). The subsidization of especially kerosene has been seen as one feasible way of meeting equity objectives, because the poor are presumed to use kerosene as their main cooking fuel. (1) However there are debates about whether the poor are the main beneficiaries of kerosene subsidies (Sumarto and Saryahadi, 2001). Indeed, even though there was early evidence that a disproportionate share of the subsidy was being captured by richer urban households, the subsidy policy continued to be strengthened and kerosene prices were held below one-fifth of the world level as far back as 1980 (Pitt, 1985).
The aim of this paper is to provide empirical evidence to help assess whether the proposed reforms of energy price policy in Indonesia are likely to be welfare-enhancing. Specifically, the equity and efficiency effects of price changes in the household energy sector are analysed. To achieve this aim, the marginal social costs of indirect taxes and subsidies are calculated for five fuels and household energy sources: kerosene, gasoline, oil, LPG, and electricity. These marginal social costs depend on the rate at which household welfare falls as prices increase, and on the rate at which net public revenue rises (Ahmad and Stern, 1984). If a reform is optimally designed, the costs in terms of social welfare of the last Rupiah of government expenditure saved by cutting subsidies (or raising taxes) on each good should be equal. To obtain the two required parameters--the welfare derivative and the revenue derivative--information is needed on tax and subsidy rates, consumption patterns, and aggregate demand responses. These requirements are less onerous than for optimal taxation exercises because for small ('marginal') change in prices, the effect on each household's welfare can be measured by their current consumption of the good. (2) Thus, the analysis follows the recommendation of Newbery (2005) to use the basic principles of public finance to introduce order into discussions of how energy taxes and subsidies might rationally be set.
This empirical analysis is needed because the previous literature on energy demand in Indonesia does not provide clear guidance for evaluating the subsidy reforms. On the one hand, Pitt (1985) concluded that the price of kerosene should be increased on both equity and efficiency grounds. In part, this conclusion rested on an estimate from household survey data that the elasticity of kerosene demand with respect to its own price was -1.03, suggesting that price distortions would create large substitution effects. But other estimated elasticities are not nearly so large: Koshal et al. (1999) use time series data to estimate a long-run own-price elasticity of demand of only -0.17. If the demand for kerosene really is this price inelastic the efficiency losses from the subsidy might be evaluated as being less important than the presumed benefits in the form of transfers to the poor. (3) Consequently, once distributional concerns were taken into account by Yizhaki and Lewis (1996) they reached the opposite conclusion to Pitt (1985); specifically, Yitzhaki and Lewis concluded that there would be aggregate improvements in welfare from increasing the subsidy on kerosene.
The wide range in estimated own-price elasticities of demand for kerosene may result from the use of different data. Pitt (1985) used household surveys, with prices proxied by unit values (expenditures divided by quantities), and the identification of demand responses coming from spatial rather than temporal variation in prices. Advantages of these data are the larger sample sizes compared with short time-series (for example, McRae (1994) uses just 15 annual observations to estimate gasoline demand models for Indonesia and several other Asian countries) and that they can provide disaggregated elasticities for target groups such as rural households or the poorest quintile of households. However, unit values are not the same thing as prices, and it has been shown by Deaton (1990) that simply treating unit values as direct substitutes for prices (as Pitt did) may cause estimated elasticities to be too large in absolute terms, causing substitution possibilities to be overstated and tending to raise the calculated efficiency cost of subsidies. Therefore, in this paper we use the estimation methods developed by Deaton (1990) to correct the biases that result from using unit values.
The rest of the paper proceeds as follows. The next section discusses the theoretical framework underlying the calculation of the social costs of marginal tax and subsidy reform. In Section 3, we describe the data and econometric estimation methods. Section 4 presents the estimation results for the disaggregated household energy demand system. Section 5 looks at the implications of...
|
