AN EVALUATION OF THE WELFARE EFFECTS OF REDUCING PETROLEUM PRODUCTS SUBSIDES IN NIGERIA

1. AN EVALUATION OF THE WELFARE EFFECTS OF REDUCING PETROLEUM PRODUCTS SUBSIDES IN NIGERIA Being a Paper Presented at the 2013 Annual Conference of the Nigerian Energy Economic Society at Lagos held from 22-23 April, 2013 By ADAGUNODO Mathew Department of Economics College of Management and Social Sciences Osun State University, Okuku Campus Okuku, Osun State Phone No: +234-851134214 E-mail address: adamat2003@yahoo.com

2. OUTLINE OF THE STUDY • The Problem • Objective • Justification • Scope • The Background • Literature Review • Theoretical Framework and Methodology

3. THE PROBLEM The downstream sub – sector has been constrained by the poor state of the refineries, which have been producing at less than 40 percent of installed capacities in the past few years, despite huge expenses incurred on turn – around maintenance of the crisis ridden refineries. The downstream sector of Nigeria’s petroleum industry is characterized by supply uncertainty, fuelled by the mismanagement of the refineries, endemic corruption, lack of transparency, direct government interference and bureaucratic processes. .This has led to massive importation of petroleum products by government and major oil marketers. Petroleum products subsidies have long been features of the Nigerian economy since early 70s in form of implicit subsidies. The market is a regulated monopoly in which NNPC serves as a regulator, a distributor, producer in the industry. The demand and supply situation is subjected to a subsidy and price fixing effect. Fuel demand and supply in Nigeria are inelastic. These subsidies have a number of perverse consequences These include among others: they send false price signal that encourage overuse of resources, they inhibit the development of substitutes that are more environmentally friendly and divert scarce financial resources from other social purposes.

4. THE PROBLEM CONTD These include among others: they send false price signal that encourage overuse of resources, they inhibit the development of substitutes that are more environmentally friendly and divert scarce financial resources from other social purposes. Subsidies discourage private investors in refineries, place burden on state budget, persistent fuel shortage, smuggling and adulteration of products. Energy price increases are usually announced with short notice, and with limited attempts at explaining the rationale behind the changes. Social unrest was a common response: violence and protests followed price rises in Egypt (1977), Morocco (1981, 1984), Tunisia (1984), Jordan (1989, 1996) and Ghana (2005).

5. OBJECTIVES OF THE STUDY The main objective of this study is to assess the equity and efficiency of petroleum products prices reform Nigeria. Hence, this study will specifically: • Examine the structure of petroleum product demand and pricing in Nigeria. • Quantify the magnitude of petroleum products prices subsidies • Estimate price and cross elasticity of demand for petroleum products • Assess the impact of subsidy reform on the welfare of households.

6. RESEARCH QUESTION From the foregoing discussion, several questions emerge that: • What are the trends and patterns of petroleum products demand and pricing in Nigeria? • What is the magnitude of petroleum products prices subsidies in Nigeria? • What are the welfare implications of its removal?

7. JUSTIFICATION FOR THE STUDY This study identifies some gaps in the literature reviewed, which it proposes to address. For instance, though some of the earlier studies examined the effect of subsidy reform on household (Gibson & Oliver, 2008; Andriamihaja & Vecchi, 2007), however, they did not consider Nigeria in their study Also, some studies in Nigeria have written on effect of energy pricing (Adenikinju 2000; Iwayemi & Adenikinju 1996; Nwafor et al 2006) however, their studies focus on macroeconomic impact using CGE model. The macroeconomic benefit does not automatically transmit to household benefit. Hossain(2003) just provided framework for petroleum products pricing and considers two years in his study, it did not consider its impact on the household. The study intends to bridge this gaps by using marginal social cost approach to estimate the impact of subsidy reform on household’s welfare This study also extends the previous work reported in Iwayemi (1994) by estimating impact of subsidy on each petroleum products. Analysis of equity and efficiency of petroleum products subsidy removal on household is imperative in an economy such as Nigeria that is undergoing substantial reform in its petroleum sector. It is also important to examine the distributional impact across the households with particular emphasis on the poorest households.

8. SCOPE OF STUDY This study will consider the impact of Petroleum Products Pricing reform in Nigeria using partial equilibrium approach. It focuses on efficiency of such reform in Nigeria and welfare implications. The study will combine both household data and time series to estimate the equity and efficiency of petroleum pricing reform. The descriptive analysis will cover from 1978 to 2011. The empirical analysis will make use of the recent household survey otherwise conduct its survey to capture the impact of the reform on household

9. BACKGROUND Nigeria has proven reserve of crude oil of 37.2 billion barrels as at the end of 2010, the tenth largest in the world and the second largest in Africa behind Libya. Nigeria with a daily production averaging about 2.4 million barrels is the 8th largest exporter in the world and largest in Africa. Nigeria has been trapped in almost three decades of petroleum products shortage and importations. Nigeria’s four refineries (all producing below 40 per cent of installed capacity while the one in Kaduna is not producing at all) and twenty two depots are in comatose for effective refining and distribution of her 2.4million bpd produces daily. This cannot be compared with Venezuela with 14 refineries refining 1.28 million bpd produces daily. Saudi Arabia has nine refineries, Malaysia has six refineries, and Libya also has six refineries. This explains why the price of PMS in Nigeria is one of the most expensive among oil producing countries, although cheapest among neighboring countries. A litre of petrol in Kuwait is N30.66k ; Qatar, N33.12k; Saudi Arabia N17.52k; Bahrain N39.42k; Egypt N46.72k; UAE N54.02k; Iran N58.40k; Malaysia N73.00; Mexico N81.76k; Indonesia N81.14,Russia N90.52k,Venezuela N5.00k and Libya N15.92

10. BACKGROUND CONTD The table 2.3 therefore provides statistical information on the domestic consumption of the major petroleum products in Nigeria within 1977 and 2010. The prices of petroleum products in Nigeria should theoretically be derived from International Crude oil prices since the marginal supply (litres) comes from import, it should therefore reflect import price. The Petroleum Products Pricing Regulatory Agency (PPPRA) was established in the year 2003 as an autonomous Agency to determine the pricing of petroleum products in Nigeria. It is also important to point out that the pricing template which PPPRA uses in computing the amount of subsidy is biased in favour of importers and overstates the level of subsidy.

11. BACKGROUND CONTD • the import prices used in the template are generally higher than the international (spot) market prices (after adjusting for the difference between “fob” and “cif”); • the template provides for unnecessary financing charges, port and storage charges and margins for importers, transporters, dealers, distributors and retailers • Since 1973, the uniform prices of PMS (same products price irrespective of location) fixed by the government have been adjusted twenty-five times (see table 2.5) to adjust the gap between domestic petroleum product prices and international prices.

12. BACKGROUND CONTD The domestic petroleum products prices have been set administratively in Nigeria since 1973, as in most oil exporting countries. However, when international prices began to rise in 2004, low domestic petroleum product prices became increasingly out of line with the market value of oil (see table 2.6A and 2.6B). Oil prices have increased dramatically since beginning of 2003. Figure 2.3 shows that, between January 2003 and December 2010, international prices rose significantly (+13 per cent per year), if not steadily. From $28.77 a barrel in 2003,the price of crude oil peak in August 2010 at $100.60, and closed at $81.07 in December2010. High domestic inflation and exchange rate deregulation contributed further to erode domestic petroleum prices vis –a-vis international benchmarks.

13. BACKGROUND CONTD The deregulation of exchange rate in 1999 and the resulting naira depreciation also accentuated a growing disparity between domestic and international petroleum product prices. • As international oil prices approached US $110 per barrel and f.o.b. gasoline prices hovered $1 per litre, Nigeria’s domestic price of US$0.59 per litre of gasoline was clearly out of touch with reality, unsustainable, and unjustifiable by any economic theory (see table 6c). • The subsidy level in 2008 alone was 150% of capital expenditure of the Federal Government in that year. • In 2006, subsidy payment on fuel products was 50% the size of Federal Government expenditure. • The estimated subsidy payment is about 400% the budgeted capital expenditures for Human Capital Development.

14. BACKGROUND CONTD December 31, 2011 before the partial withdrawal of subsidy, it stood at 76 naira per litre but was reduced to 44 naira when the official price of petrol was pegged at 97 naira per litre.Oil subsidy has moved from being implicit to explicit from 1999. The figure 2.4 indicates that for each litre of petroleum products sold, N19.4 was spent on subsidy in 2003. This implied a subsidy of N153 billion or 1.42% of GDP. By the end of 2008with subsidy shooting up to N450 billion (see table 2.6C), it went up to 3% of GDP It increased to N654.762 billion (3.43% of GDP) in 2008. The government paid N1, 173.2 billion (US$9.7 billion) as a subsidy between 2006 and 2008. The payment made in the second quarter of 2009 was N406.134 billion (3.43% of GDP). The subsidy payment in 2010 was amounted 1.25 billion. Nigeria spent 19 percent of her budget on subsidy payment in 2011. (table 2.6c) These figures exceed total allocation to priority sectors of our economy. The recent development has revealed that most of the claims and payments are frauds.

15. LITERATURE REVIEW • There is rapidly growing literature on energy pricing and subsidies. For clearityof exposition, the review here is organised into four parts, namely, conceptual issues, theoretical, methodological and empirical literature. They are discussed in that order. 3.1 CONCEPTUAL ISSUES IN ENERGY SUBSIDY: A subsidy can be categorized into implicit, explicit and cross subsidies. Three methods of quantifying the magnitude of subsidies are in the literature (Koplow, 2009; UNEP, 2008;IEA 1999; World Bank 2010).These are price gap approach, the program-specific approach and the measure of producer or consumer subsidy equivalent. (table3.1) Energy subsidies should be assessed by their relative efficacy, sector efficiency and cost effectiveness. Komives et al (2005) proposes two dimension of subsidy performances (i) Benefit incidence (how well the subsidy targets benefits to poor household as opposed to other household) Beneficiary incidence (What proportion of poor households as a whole receive the subsidy). Two types of elasticity are important when analysing energy market: own price elasticity of demand; and cross price elasticities of demand.

16. THEORETICAL LITERATURE There has been considerable discussion about the traditional view that inefficiencies result from subsidized energy prices. Economic theory suggests that subsidies are inefficient because, in the absence of market imperfections and with convex indifference curves, the value of the subsidy to the consumer will be less than its cost to the government (see Katz and Rosen, 1994). When the price deviates from this point of static equilibrium, resource allocation is inefficient since the benefit to consumers from the last unit of energy consumed are smaller than the costs involved in supplying the energy service. The most desirable economic policy is the policy yielding the highest level of welfare. Hick (1942) introduced measure of consumer’s surplus based on compensating and equivalent variations in income and total expenditure. Lau and Stoker (1980, 1981, and 1982) have developed methods of constructing indirect utility functions and expenditure functions for a population of consumers.

17. THEORETICAL LITERATURE CONTD Since the pioneering work of Atkinson (1979) and Kolm (1969) the measurement of social welfare has been based on explicit social welfare function introduced Atkinson and Kolm are defined on the distribution of income rather than the distribution of individual welfare. Muellbauer(1974) has shown that measures of social welfare based on individual welfare if and only if preferences are identical and homothetic for all consumers. Roberts (1980b) has derived restriction on preferences under which measures of social welfare introduced by Muellbeuer(1974) are independent of the prices faced by individual consuming units. In the absence of restrictions on social welfare functions, individuals must have identical homothetic preferences. Most of this studies violate the concavity of social orderings over optimal commodity allocation. This study will improve on reviewed studies by specifying welfare function with heterogeneous marginal utilities to assess welfare and distributional impact of price reform.

18. METHODOLOGICAL LITERATURE Partial-equilibrium as well as general equilibrium model have been used to study the impacts of fossil fuel subsidy reform. Partial-equilibrium models consider only the product market in which subsidy reform is occurring (in this case, the energy market) and estimate price, demand and production changes in fossil fuel as a result of subsidy removal based on simply supply-and-demand curves and economic assumptions (Von Moltke et al 2004). Partial equilibrium analyses provide a valuable input into policy dialogue and reform, especially when combined with qualitative discussion of the likely efficiency effects of reforms. For example, switching reform to reduction of subsidy on products with inelastic demands or negative social externalities (such as petroleum products) when these are initially large can be expected to increase the overall efficiency of the reform system. (table 3.2) EMPIRICAL LITERATURE There have been studies on subsidies and economy, and many researchers have directed the focus of their studies on subsidies as well as household both within and outside Nigeria. Several findings emanate from various empirical investigations on energy price reform and petroleum products subsidy in particular and its effect of removal on household welfare. (table 3.3)

19. THEORETICALFRAMEWORK Specifically following the treatment of Ahmed and Stern (1984), we assume that there are H households in the economy and denote by Xi h (q), the quantity of commodity i purchase by household h. We write this quantity as a function of consumer prices only, because we assume factor income to be fixed. Intuitively, we assume that the producer prices Pi …. PN of N commodities is fixed and there are no pure profits. Subsidy reduction increases the consumer prices. (section 4.0) Pi = qi+si(1) Aggregate consumption is then given by Xi= ihi=1 . . . . .N (2) The government realised certain amount of revenue through subsidy reform W=W(U1, . . . ,UH)=W(V1(e1, p) , . . . ,VH(eH , p)) , (4) (3) R= The starting point is the social welfare function, which aggregates individual welfare levels. Define the social welfare function over h=1. . . H households (5) = Xi + k Let we consider the consequences of a marginal change in the subsidy. This change will affect government revenue.

20. THEORETICALFRAMEWORK CONTD It will also have an effect on social welfare: h = h =W*(V1(e1,p) , . . . ,VH(eH ,p))-W(VI(e1 ,p) , . . . ,VH(eH,P))/ = i (6) The approach is based on that pioneered by Bergson (1937). Before deriving the distributional characteristics, first consider the impact on social welfare from a change in household expenditure. h h (7) h • Thus, following Newberry (1995), the distributional characteristics, di for ith good is In money terms, household h is worse off by the quantity consumed, or in utility terms is worse off by , xih hxih. (8) di Where is the average of social utility weighs over all households and H ih h is the aggregate consumption of the ith good (9) = hxih.

21. THEORETICALFRAMEWORK CONTD The change in social utility for a small change in price (e.g., a tax or subsidy on quantities) in equation 6 can be approximated to provide a numerical measure of change in social welfare. The welfare impact of a price change as result of reform is seen to depend on two factors: the distribution of consumption, given social weights and the level of consumptions among household, given by xih(Newbery 1995). And are total expenditure and size of household h, and is the coefficient of inequality aversion. , h h/ h)- = h h • / i = (10) Goods with low marginal cost ratios are those that are candidates for either a tax increase or a subsidy reduction. When all the ratios are the same there is no further scope beneficial reform. This approach can be implemented by noting that welfare derivative (the numerator of (10)) is a pure distributional measure for good i and it is the ratio of two average budget shares: = • (11) • / i i ]/ h h The denominator of the marginal social cost ratio represents the efficiency aspect of subsidy induced price change. A given price change will produce a larger net revenue effect, the greater is total consumption of the good and the less the substitution away from taxed goods (Olivia and Gibson 2006).And using (5) and (6) • (12)

22. THEORETICALFRAMEWORK CONTD The denominator of the marginal social cost ratio represents the efficiency aspect of subsidy induced price change. A given price change will produce a larger net revenue effect, the greater is total consumption of the good and the less the substitution away from taxed goods (Olivia and Gibson 2006).And using (5) and (6) h i /( Xi + ih k ) • (13) Multiply numerator and denominator of equation (13) by price qileads to an expression which can be operationalised easily: k) (14) h I kisk ih) *(pk / i k, ( / ) ki Where refers to the uncompensated price elasticity and sk*= */ the subsidy reduction rate as a fraction of consumer price. The equation (14) is the combination of subsidy factor and cross elasticities. (15) i

23. THEORETICALFRAMEWORK CONTD i The first term of the denominator in equation (15) measures the own –price distortionary effect of the subsidy. If it is large and positive, as would be the case for a heavily subsidised and price elastic good, the term will contribute to a small marginal social cost (msci) and would indicate the low cost of saving fiscal expenditures from decrease in the subsidy on this good. The last term is the sum of the subsidy factors multiplied by cross price elasticities, and captures the effects on other goods(and resulting revenue changes) from subsidy reform on good i. The use of the complete expression in equation (15) integrates equity and efficiency consideration. To illustrate the trade- off between efficiency and equity, it seems interesting to rewrite (15) as (16) i Let us first consider the case where we would neglect all changes in consumption pattern. If the assume that Equation (16) then reduce to ki (17) ih/ h i i The marginal welfare cost of subsidy reform for any commodity, then coincides with the distributional characteristics of that commodity, as introduced by Feldstein (1972). This concept summarises the variation of consumption pattern across income classes by weighting the market shares of the different households in the consumption of commodity i,using the as weights. h’s

24. THEORETICALFRAMEWORK CONTD If equity does not matter at all, in that case equation (16) reduces to i (18) Equation (18) concentrates on efficiency aspects of the reform.

25. MODEL The empirical model applied in this study to get price responses needed for the marginal reform calculation in equation (15) is almost ideal demand systems (AIDS) model. The model is adopted due to the following reasons:( a) it is relatively easy to estimate and interpret.(b) It satisfies the axioms of choice exactly.(c) it is as flexible as other locally flexible functional forms but it has the added advantage of being compatible with aggregation over consumers. It can thus be interpreted in terms of economic models of consumer behaviour when estimated with aggregated (macroeconomic) or disaggregated (household survey) data (Glewwe, 2001).(d) It is derived from a specific cost function and therefore corresponds with a well-defined preference structure, which is convenient for welfare analysis. (e) Homogeneity and symmetry restrictions depend only on the estimated parameters and are therefore easily tested and/or imposed. (f) AIDS provides also an arbitrary first-order approximation to any demand system. (g) It aggregates perfectly across consumers without invoking parallel linear Engel curves and finally, it has a functional form which is consistent with known household-budget data. i The AIDS Model for petroleum products demand can be expressed it t/ ik i k) it iti=1,……..,n (19) and where, in observation t; it is the budget (expenditure) share of the ithproduct is the nominal price of the kthproduct; t

26. MODELCONTD is total expenditure; t is the random error term it is the translog price index defined by; t 0 k = kt k t it t=1… T (20) it This price index makes the system non –linear, which normally complicates the estimation process. In order to overcome this problem, Deaton and Meulbaurer(1980) suggest using another price index. The only difference between the AIDS and its linear version, the LA/AIDS, lies in the specification of the price index. Several authors, include Green and Alston(1991);Moschini(1995);Asche and Wessels(1997), have discussed the relationship the linear and nonlinear specifications. In several of these studies, Monto Carlo studies were used to show that the use of differential functional forms of the index in the LA/aids provides results that compare reasonably well to the AIDS model (Asche &Wessels 1997). The stone’s price index, as suggested by Deaton and Meulbauer (1980), which can be used to replace the translog price index, is defined as follows: it LogP= i,t (21)

27. MODEL CONTD Eales & Unnevehr (1998) showed that the substitution of the Stone’s price index for the translog price index causes a simultaneity problem, because the dependent variable it) also appears on the right-hand side of the LA/AIDS. They suggested using the lagged share (wi t-1) for equation (19) with the lagged shares into Equation (21) yields LA/AIDS, given by: i it ( it ik i ( kt (22) ik Equation 22 can then be applied to the empirical data, where after the anticipated parameters can be used to calculate the required elasticities. Compensated and uncompensated elasticities will be calculated using the formulas. (23) = eit wk/wi) )=- I,j=1,2,…….,N. t (wk/wi) it/w) (24) Where otherwise. The average expenditure shares are represented are RSUR parameter estimates for the LA/AIDS model. by wt whereas, and it t The formula used to calculate the expenditure elasticities can be written as: (25) 1+ (βi/wi) t=

28. DATA USED AND STATISTICAL PROPERTIES OF THE VARIABLES The annual data from Central Banks and household data from National Bureau of Statistics (NBS) will be used to estimate the model. The variables will be subjected to statistical tests, including: Univariate properties of the data, structural breaks, separability and exogeneity of the expenditure variable. LIKELY CHALLENGES Assessing the magnitude of fossil- fuel subsidies and its impact on household is a task challenged by poor data quality, limited data availability and lack of data comparability as there is no harmonised or consistent reporting structure for fossil fuel subsidies in Nigeria.

29. EMPIRICAL RESULTS AND CONCLUSION • Table 5.1 presents expenditure elasticity. The coefficient for AGO and PMS are positive indicating luxury goods whose budget share rise more than proportionally as household expenditure rises.10 percent increases in total expenditure leads to 2.3 percent increases in budget share for PMS. • Table 5.1 EXPENDITURE ELASTICITY Source: Author’s estimation Also, 10 percent increase in total expenditure brings about 0.2 percent increase in budget share for AGO. The total expenditure elasticity for DPK is -0.31 indicating that kerosene is a necessity. This implies households in Nigeria have larger proportional increases in demand for this product as their income rises. Table 6, contains the estimate own-and cross-price elasticities. The own-price as well as cross – price effects are well determined. All estimated own price elasticities are negative as they should be. The estimated own price elasticity of demand found to be -0.2110.

30. Table 5.2 OWN – PRICE & CROSS PRICE ELASTICITIES Source: Author’s estimation Specifically, 10 per cent increase in price of PMS will bring about 2.1 percent decline in the quantity consume by the Nigerian household. This suggests that on efficiency ground, the subsidy on PMS should be reduced. The own price elasticities are large for diesel and kerosene, justified the removal of subsidy on AGO and high price of DPK caused a lot of substitution into these products.

31. Table 5.3 shows the efficiency effects of subsidy reduction on each of the petroleum products, distinguishing between the terms in the denominator of marginal social cost formula. The first column indicate subsidy factor (difference between world and domestic price), while the second column shows the own price elasticities of quantity and quality together. The products of the first and second columns which is shown as the third column, gives own contribution of price distortion that would be caused by a marginal increase in price. TABLE 5.3 Efficiency of energy price reform in Nigeria Sources: Authors’ estimation The table 5.3 above presents efficiency of reduction in energy subsidies. As it can be seen, the own effects of all energy products are large with the exception of DPK. The reduction or removal of energy subsidy on PMS would save the largest amount from government budget.

32. TABLE 5.4a Equity effects of energy price reforms in Nigeria Authors’ estimation The table 5.4a presents equity effects when coefficient of equality aversion is zero ( ). That is, when there is no distribution concern, the marginal social cost of reducing subsidies on PMS is lowest. AGO has highest marginal social cost among the energy source. However, marginal social cost for all energy products are extremely low suggesting reduction of subsidies for all petroleum products.

33. TABLE 5.4b Presents situation of high inequality aversion Source: Authors ‘estimation • The larger coefficient of inequality aversion attached larger value to the product often consumed by the poor and a relatively smaller value to those consumed by household that are better off. For, the lower social cost of reduced government expenditure (or equivalent additional revenue) would come from removing subsidies on AGO followed by aggregate petroleum products price. Government will incur high social cost if prices of electricity and DPK were raised. It is advisable that subsidy on kerosene and electricity should be retained for equity consideration.

34. 6.0 SUMMARY AND CONCLUSION This study combines time series and household survey data from 5,000 Nigerian household data using stratified multi stages sampling techniques to estimate a demand system for petroleum products. This study use marginal social cost approach to evaluate equity and efficiency implications of petroleum products subsidy reform in Nigeria. The marginal social cost approach originated with Feldstein (1972) and Ahmed and Stern (1984) and is based on the optimal commodity taxation rules derived by Ramsey (1927) and Samuelson (1986). This study concludes that reduction or removal of subsidy on PMS will save largest amount from government budget. Where there is no distribution concern, the marginal social cost of reducing subsidies on PMS is lowest. AGO has highest marginal social cost among petroleum products. However, marginal social cost for all energy products are extremely low suggesting reduction of subsidies on all energy products in Nigeria

35. THANK YOU