1 by Muhammad Akimaya PhD candidate Colorado School of MinesSimulation of Price Controls on Different Grades of Gasoline: A case of Indonesia by Muhammad Akimaya PhD candidate Colorado School of Mines

2 Background (1) Oil-rich producing countries implement gasoline subsidy to distribute welfare. Good: Lower transportation costs, high affordability of energy Bad: Inefficiency in gasoline market

3 Background (2) Indonesia provides gasoline subsidy since 1960’s in the form of underpricing. Since 2002, Opened up premium gasoline market to multinational companies Only provide subsidy to regular gasoline Violation of Ramsey’s tax rule (1927) Further inefficiency?

4 Background (3) Further problems:Higher opportunity costs because of significant rise in crude oil price Rising domestic gasoline demand

5 Background (4)

6 Background (5) Solution? Remove the subsidy!Strong resistance from the public Public protests, sometimes led to violent protests Riots

7 Contribution In the literature: Effects on gasoline subsidy removalSocial welfare (Adam and Lestari 2008) Incidence of poverty (Dartanto 2013) Aggregate price level & real output (Clements et al. 2007) Alternative policy of reducing gasoline subsidy cost and inefficiency by price control on premium gasoline

8 Mathematical Model Objective: Decision variable:Minimize gasoline subsidy spending Decision variable: Subsidy rate on premium gasoline

9 Mathematical Model (2) min 𝑠 2 𝐺 = 𝑠 1 𝑄 83 + 𝑠 2 𝑄 87𝑄 83 =𝑓 𝑃 83 , 𝑃 87 , 𝑃 𝐴𝑂𝐺 ,𝐼 𝑄 87 =𝑓 𝑃 83 , 𝑃 87 , 𝑃 𝐴𝑂𝐺 ,𝐼 The first order condition gives us: 𝑑𝐺 𝑑 𝑠 2 = 𝑠 1 𝑑 𝑄 83 𝑑 𝑠 2 + 𝑠 2 𝑑 𝑄 87 𝑑 𝑠 2 + 𝑄 87 =0

10 Mathematical Model (3) 𝑑 𝑠 2 =−𝑑 𝑃 87 𝑑𝐺 𝑑 𝑠 2 =− 𝑠 1 𝑑 𝑄 83 𝑑 𝑃 87 − 𝑠 2 𝑑 𝑄 87 𝑑 𝑃 87 + 𝑄 87 𝑑𝐺 𝑑 𝑠 2 =− 𝑠 1 𝑑 𝑄 83 𝑑 𝑃 87 𝑃 87 𝑃 87 𝑄 83 𝑄 83 − 𝑠 2 𝑑 𝑄 87 𝑑 𝑃 87 𝑃 87 𝑃 87 𝑄 87 𝑄 87 + 𝑄 87 𝑑𝐺 𝑑 𝑠 2 =− 𝑠 1 Ε 83,87 𝑄 83 𝑃 87 − 𝑠 2 Ε 87 𝑄 87 𝑃 87 + 𝑄 87 𝑠 2 ∗ = − 𝑠 1 Ε 83,87 𝑄 83 𝑃 87 + 𝑄 87 Ε 87 𝑄 87 𝑃 87

11 Methodology Derive demand using Translog unit cost function calibration by Rutherford (2002) Assuming homotheticity, equivalent to uncompensated demand function Simulation and sensitivity tests

12 Calibration Setup *Akimaya (2016) **Hastings & Shapiro (2013)

13 Simulation 17.6% savings that amounts to 19.9 trillion rupiahs(roughly 1.6 billion USD)

14 Simulation (2)

15 Sensitivity Analyses Setup:

16 Sensitivity Analyses (2)Setup: *overall gasoline **regular gasoline

17 Sensitivity Analyses (3)Effect of income elasticity 83-octane

18 Sensitivity Analyses (4)Effect of income elasticity 87-octane

19 Sensitivity Analyses (5)Effect of cross-price elasticity 87-AOG

20 Welfare Positive: Negative: Assume perfectly elastic supplyReduce regular gasoline demand and also inefficiency in regular gasoline market Negative: Introduce distortion in premium gasoline market Assume perfectly elastic supply

21 Welfare (2) Regular gasoline market

22 Welfare (3) Premium gasoline market

23 Welfare (4) * represents value in billion USD Overall, 2.7% reduction in inefficiency that amounts to 135 million USD.

24 Conclusion Provides a new alternative to gasoline subsidy removalUnder benchmark scenario: 17.6% cost reduction (1.6 billion USD) 2.7% reduction in Deadweight Loss (135 million USD) Magnitudes of savings are mainly driven by cross-price elasticity between regular and premium gasoline

25 Limitations Assumption on substitutability between AOG and gasolineLimitations on translog model Consumers’ responsiveness to change in price Premium gasoline market inefficiency in the future Implementation costs Welfare on supply side

26 Thank you Q&A

27 Background (6)

28 Calibration ln 𝐶 𝑝 ≡ ln 𝑏 0 + 𝑖=1 𝑁 𝑏 𝑖 ln 𝑝 𝑖 𝑖=1 𝑁 𝑗=1 𝑁 𝛼 𝑖𝑗 ln 𝑝 𝑗 ≡ ln 𝑏 0 +𝐿 𝑝 𝑖=1,2,…𝑁-goods 𝑗=1,2,…𝑁-goods From the unit cost function, the compensated demand function is derived 𝑥 𝑖 𝑝 = 𝜕𝐶 𝑝 𝜕 𝑝 𝑖 = 𝜕 ln 𝐶 𝑝 𝜕 𝑝 𝑖 𝐶= 𝐶 𝑝 𝑝 𝑖 𝑏 𝑖 + 𝑗=1 𝑁 𝛼 𝑖𝑗 ln 𝑝 𝑗

29 Calibration 𝑖=1 𝑁 𝑏 𝑖 =1 𝛼 𝑖𝑗 = 𝛼 𝑗𝑖 ∀𝑖,∀𝑗 𝑗=1 𝑁 𝛼 𝑖𝑗 =0 ∀𝑖 The calibration process involves finding the  and b parameters as shown below. 𝛼 𝑖𝑗 = 𝜃 𝑖 𝜃 𝑗 𝜎 𝑖𝑗 𝐴 −1 , 𝑖≠𝑗 𝛼 𝑖𝑖 =− 𝑗≠𝑖 𝛼 𝑖𝑗 ,∀𝑖 𝑏 𝑖 = 𝜃 𝑖 − 𝑗=1 𝑁 𝛼 𝑖𝑗 ln 𝑝 𝑗 , ∀𝑖 𝑏 0 = 𝐶 𝑒 −𝐿 𝑝 𝜃 𝑖 𝑖𝑠 𝑡ℎ𝑒 𝑠ℎ𝑎𝑟𝑒 𝑜𝑓 𝑔𝑜𝑜𝑑 𝑖

30 Calibration According to Rutherford (2002), the term 𝜎 𝑖𝑗 𝐴 measures the responsiveness of the compensated demand for one input for a change in one input price 𝜎 𝑖𝑗 𝐴 = 𝜎 𝑖𝑗 𝐶 𝜃 𝑗 𝜎 𝑖𝑗 𝐶 is the compensated price elasticity (Mas-Colell et. al 1995) which is defined as 𝜎 𝑖𝑗 𝐶 = 𝜎 𝑖𝑗 + 𝜃 𝑗 𝜎 𝑖𝑦