OverviewFuel consumption increased strongly in the last fifteen years in Brazil. Diesel oil consumption was 35 billion liters in 2000 and 59 billion liters in 2014. It corresponds to a 4% growth rate by year (EPE, 2014). As Brazilian diesel production grew at a slower pace, diesel imports almost doubled, from 6 billion liters in 2000 to 11 billion liters in 2014.The transport segment (freight and passengers) was the main determinant of diesel demand growth in Brazil. It increased at a 7.2% rate a year between 2000 and 2014. Due to government incentives, 312 thousand new trucks were sold in 2011 and 2012 (EPE, 2013).The increase of diesel imports is currently a major concern of Brazilian authorities. New refineries were planned but as Petrobras is in a deep financial crisis, projects were postponed or suspended. As refine activity is not attractive in Brazil, the domestic supply must not increase substantially in the next years.The objective of this paper is to analyze the determinants of diesel oil demand from 2000 to 2014 and to forecast demand from 2016 to 2030. We use state level annual data to estimate a dynamic panel data model. Our intent is to contribute for the Brazilian energy planning throughout a robust analysis of diesel demand, which can be useful for demand management and supply decisions.The paper is organised as follows: The next section describes the methodology used to forecast diesel demand and shows the dataset used in the paper. The third section presents the results of the forecast. Finally, the fourth section concludes the paper and provides some directions.MethodsDiesel oil demand was estimated through a dynamic panel data model. The demand specification can be represented by:ğ‘_(ğ‘–ğ‘¡)=ğœŒğ‘_(ğ‘–,ğ‘¡âˆ’1)+ğ‘‹â€²_(ğ‘–ğ‘¡)ğ›½+ğ‘¢_(ğ‘–ğ‘¡); ğ‘–=1,…,ğ‘ and ğ‘¡=1,…,𑇠(1)where ğ‘_(ğ‘–ğ‘¡) is the diesel demand at the state i and time t, ğ‘_(ğ‘–,ğ‘¡âˆ’1) refers to demand in time ğ‘¡âˆ’1, ğ‘‹â€²_(ğ‘–ğ‘¡) is a vector of other variables that influences demand and ğ‘¢ğ‘–ğ‘¡ is the regression error. The error can be decomposed in ğ‘¢_(ğ‘–ğ‘¡)=ğ‘_ğ‘–+ğœ€_(ğ‘–ğ‘¡), where the fixed effect ğ‘_ğ‘– is the non-observed specific features that are constant in time in each Brazilian state and ğœ€_(ğ‘–ğ‘¡) is the idiosyncratic compound. We suppose ğ‘_ğ‘–ï½ğ¼ğ¼ğ·(0,ğœ_ğ‘~2) and ğœ€_(ğ‘–ğ‘¡)ï½ğ¼ğ¼ğ·(0,ğœ_ğœ€~2).To handle the endogeneity problem that results from the inclusion of the autoregressive term we applied the Arellano and Bover (1995) and Blundell and Bond (1998) method. The System Generalized Moments Method (System GMM) considers a system with variables in level and in first differences.Based on this econometric model, diesel demand equation can be especified as:ğ‘_(ğ‘–ğ‘¡)=ğ›½_0+ğœŒğ‘_(ğ‘–,ğ‘¡âˆ’1)+ğ›½_1ğ‘ƒ_(ğ‘–ğ‘¡)+ğ›½_2ğ‘Œ_(ğ‘–ğ‘¡)+ğ›½_3ğ¹+ğ›½_4ğ‘ƒğ‘œ_(ğ‘–ğ‘¡)+ğ›½_5ğ·+ğ‘¢_(ğ‘–ğ‘¡) (2)where ğ‘ƒ_(ğ‘–ğ‘¡) is the diesel oil price in the state i and time t, ğ‘Œ_(ğ‘–ğ‘¡) is income, ğ¹ oil diesel fleetand, ğ· is aregional dummy that identifies North, South, Center Westand and Souteast regions.From the estimated equation (2), we simulate future demand of diesel oil until 2030 to evaluate impacts ofdifferent economic scenarios on supply and demanand balance. Those scenarios include oil diesel price, incomeand fleet.The data set was provided by the Oil, Natural Gas and Biofuels National Agency (ANP), by the Energy ResearchCompany (EPE) and by the National Traffic Department (Denatran). The data is annual and covers the periodfrom 2000 to 2014 for the 26 states and the Federal District.ResultsBy the econometric analysis, we conclude that diesel demand is inelastic to price variations. An 1% increase indiesel oil price will cause a 0.2% reduction on diesel consumption. This low price elasticity is due to the lack ofsubstitute fuels.Diesel demand is also inelastic to income variations. An 1% increase on consumer income will result in a 0.1%increase in diesel oil demand.In our diesel oil consumption forecast, it shows a moderate growth from 2015 to 2017. After 2019, consumptiongrowth accelerates as Brazilian macroeconomic situation improves. Table 1 shows the forecast of theconsumption and production of diesel fuel in the Brazilian market.In the demand and supply balance, we see that Brazil will be a net importer of diesel for most of the years. Weestimate exports only in 2017. After 2019, diesel oil production will not be enough to match the increasingdemand. So, diesel imports upsurge again.ConclusionsRegarding the relevance of oil diesel consumption for the fuel sector in Brazil, this research estimated thedemand determinants and developed a long term forecast model for oil diesel demand in the Brazilian transportsegment.The demand forecast for 2016-2030 indicates a strong demand increase in our reference scenario. So, Brazilianenergy policy must consider it as to provide security of supply of this .In our reference case scenario, the entry of the two new refineries (Abreu e Lima and Comperj) won’t be enoughto compensate diesel demand growth. As Brazilian economy recovers, diesel oil imports increases.
展开▼