1. INTRODUCTION

Fossil fuel prices deserve particular attention because of the influence they have on the well-being of an economy. This is the case because given the consumption of energy and transportation, they indirectly impact economic activity and the public sector’s account, which in turn influence the country’s debt and its credibility in the eyes of investors. There are direct, as well as indirect, impacts on the environment, which are damaging to the population’s health. Direct impacts emerge by way of energy costs and demands, in addition to the incentive to invest in renewable energy directly. Indirectly, there are impacts on economic growth, and a positive relationship between growth and fuel consumption (^{Pereira and Pereira, 2014}). As such, economic policies should ensure that there is a sufficient supply of energy in order to guarantee the smooth functioning of the markets, while avoiding excessive use to prevent harming people’s health and the environment (^{OECD, 2015}).

Various countries around the world have energy subsidy policies, which incentivize the consumption of petroleum-based products, natural gas, and electricity. The International Energy Agency estimates (^{AIE, 2016}) indicate that in 2014, 493 billion dollars were granted in subsidies for fossil fuels globally.

The Organization for Economic Co-operation and Development (OECD) defines a subsidy as any method of maintaining prices below market levels for consumers or producers, or which reduce costs for consumers or producers (^{OECD, 2015, p. 114}). Accordingly, subsidies are represented by any difference in price between the amount paid by the end consumer and the international reference price, minus transportation and distribution costs. Subsidies after taxes seek to account for social costs also generated by the negative externalities of energy consumption, like contamination, climate change, and congestion (^{Barany and Grigonytė, 2015}, p. 4).

Taking into account negative externalities, the ideal price of fuel would be above the market equilibrium level, reducing the demand for energy. According to ^{Coady et al. (2015}), prices for consumers should be composed of the following: the cost (or opportunity cost) for the energy producer; a Pigouvian tax, to reflect the cost of negative externalities; and a consumption tax.

Different countries around the world have policies that function to the contrary, maintaining prices artificially below equilibrium prices, in an effort to stimulate economic growth or benefit the lowest income sector of the population, and hoping that energy would then have less clout in the consumer basket.

Despite this, many studies question the possibility of achieving such goals by way of energy subsidies, arguing that their effects might be contrary to what they intend. ^{Clements et al. (2013}) assert that these policies may impede growth since they can increase public debt, discourage investment in more efficient and cleaner energy, take the place of other investments that could incentivize growth, encourage smuggling, and decrease competitiveness in the private sector.

The aforementioned study also notes that subsidies are regressive, demonstrating that on average, 20% of the high-income population collect 43% of subsidies, while 20% of the poorest only receive 7%. According to the world average, gasoline is the item that shows the most regressivity, while kerosene shows progressivity.

^{Ellis (2010}) provides the primary economic, environmental, and social impacts of fossil fuel subsidies. Among the economic impacts cited are: the increase in energy consumption, and consequently, the disincentive to use more efficient forms of energy; the possibility of reducing exchange earnings; increased dependence on importation; the incentive to use energy intensively in production, to the detriment of labor; and the possibility of reducing producer’s profits, thus creating shortages of the subsidized product. The environmental impacts consist of emissions not only of carbon gas, but of other gases like sulfur dioxide and nitrogen oxide, which can cause harm to human health (such as flu, pneumonia, and asthma), structures, and agriculture. Some social impacts derive from the foregoing, given that such disadvantages are more detrimental to the lower income population.

Accordingly, fuel subsidy policies can cause fiscal, environmental, and health problems, while also reducing innovation within the sector, and damaging economic growth. In addition, they probably benefit the high-income population more than the low-income one.

As such, it can be acknowledged that reforms to end subsidy policies are necessary. These reforms should be well planned, given that an increase in energy prices causes some disadvantages on the short term. First, there would be a direct inflationary impact, since energy is significantly represented in the consumer basket, and an indirect one, given that energy is an input for the majority of goods. There is also the possibility of a decrease in the country’s international competitiveness since there would not be simultaneous reforms in other countries, thus causing comparatively high production costs for those that do raise their prices. The lowest class population would suffer more from the impacts of the policy given that, despite regressivity, the removal of subsidies would significantly impact their budgets. ^{Clements et al. (2013}) argue that an increase of USD 0.25 for a liter of fuel could increase the price level by 5.5% for the poorest families.

In the case of Brazil, between 2011 and 2014 the international price of petroleum was higher than what was seen domestically, and Petrobras continued to sell at a price established by the government, in order to prevent a transfer of the increase onto the domestic market. The state-owned company accordingly provided subsidies for petroleum-based fuels upon importation, and resold them at a lower price on the domestic market, to the clear detriment of the business. This situation was also exacerbated because the company did not stop making investments projected in the 2012-2016 Business Plan.

According to the Brazilian Infrastructure Center’s estimates (^{CBIE, 2014}), the price of gasoline in Brazil reached a level 33% lower than in the Gulf of Mexico (the international reference price), while diesel hit a 29% discrepancy.

Fiscal losses for the public sector are not only derived from the sale of fuels at a lower price that than that of cost. Revenues from the Income Tax and the Social Contribution on Net Profits, which have an impact on Petrobras’ profits, also decrease. Additionally, the revenue from taxes related to the price of making fuels is lowered. The government ended the Contribution for Economic Strength Intervention [Portuguese CIDE] on fuels during the period from June 2012 to February 2015, further decreasing revenue.

Maintaining fuel prices low by artificial means generates distortions in a variety of the economy’s markets. For example, in a direct way, the low price of gasoline causes less demand for alcohol by consumers, which discourages its production. Producers of sugar cane, then, substitute production of alcohol for that of sugar, causing the latter’s price to decrease for exportation (^{ANTT, 2014}). ^{Di Bella et al. (2015}) calculate that, between 2011 and 2013, subsidies cost 1.2% of Brazil’s GDP.

Beyond this Introduction, this study is comprised of five other sections. In section 2, a brief overview of the literature on the economic effects of fuel prices is provided. Section 3 describes the economic model and section 4 calibrates it to the data on the Brazilian economy. Section 5 analyzes the results obtained from the removal of the subsidy in Brazil, accompanied by a variety of tax policies; lastly, in section 6 the final conclusions are presented.

2. FOSSIL FUEL SUBSIDIES

The study of fossil fuel subsidies has received substantial attention in recent years in literature on the topic. Institutions such as the OECD, the IMF, the International Energy Agency [Spanish AIE] and the World Bank have determined the value of this policy and evaluated its social and economic impacts. Such works note that the elimination of subsidies reduces the government’s budget deficit, causing a more efficient distribution of resources and increasing potential economic growth (^{Barany and Grigonytė, 2015}, p. 1).

^{Parry and Small (2005}) use a closed economy static model to calculate the second-best tax for gasoline in the United Kingdom and United States. The study suggests that the tax should reflect congestion, accidents, and air contamination. The tax should also include a Ramsey tax, which reflects the appropriate equilibrium between the consumption tax and the labor tax, in order to fund government spending. The authors claim that the United Kingdom maintains a high tax rate and as such, in addition to minimizing negative externalities, provides the government with a substantial income without notably affecting economic efficiency, since fuels have a low price elasticity of demand.

Between 2003 and 2005, international fuel prices increased substantially; in Central Africa, Gabon implemented a policy similar to the one Brazil enforced from 2011 to 2014, maintaining fuel prices constant. ^{Leigh (2006}) calculates that the subsidy costs in that country ranged from 1.6 to 2% of the GDP in 2005 and 2006, respectively. The author concludes that 10% of the richest population benefitted from a third of the subsidies, while 30% of the poorest received only 10% of the policy’s costs.

^{Lin and Li’s article (2012}) uses a general equilibrium model to simulate the impacts of removing energy subsidies in China, and compares different strategies that could mitigate the effects of the policy’s withdrawal. The results show that the reduction of subsidies would have global effects by way of trade. China would then lose competitiveness in the face of other countries that did not eliminate their subsidies. There would also be effects on energy consumption, which would be transferred to other countries, so as to decrease the Asian country’s carbon emissions while increasing emissions in other parts of the world.

^{Glomm and Jung (2012}) study the effect of an energy subsidy reduction with a general dynamic equilibrium model, calibrated for Egypt. The authors conclude that a reduction of 15% in energy subsidies could increase or decrease the country’s GDP by 3%. The direction of this effect depends on the policy used by the government to substitute the subsidies. If investment in infrastructure were undertaken, the country’s economy would grow. If the policy were substituted for lower taxes, the extra income and excess of energy would be exported at international prices, diminishing the GDP.

^{Anand et al. (2013}) evaluate the impact of a reform designed to withdraw energy subsidies in India on the public budget and the population’s well-being. This is a country that has elevated spending on subsidies, and increased its GDP by 0.8% between 2009 and 2010 and by 1.9% between 2011 and 2012. The authors argue that the benefit is seven times greater for 10% of the population’s richest families than for the poorest 10%. They also calculate that the removal of these subsidies would have a negative impact of 4% on the low-income group’s budget and of 5% for the highest income group. The article suggests that the following reforms be adopted: utilization of international petroleum prices as a reference for national prices; gradual removal of subsidies; and the implementation of a transfer income policy, as an offset for individuals from the lowest class.

^{Coady et al. (2017}) estimate the worldwide cost of fuel subsidies as the difference between what should be paid at discounted prices, the environmental costs and consumption tax, and what is actually paid in fuel consumption. The quantity calculated is more than USD 4.9 billion for 2013, and is estimated at USD 5.3 billion for 2015, the equivalent of 6.5% of the world’s GDP.

^{Dartanto (2013}) argues that subsidy policies reduce the government’s budget for sectors that could promote economic growth. The removal of subsidies and investment in infrastructure and education could reduce poverty in the long term, though it could increase in the short term. The author uses a computable general equilibrium model applied to the Indonesian economy to determine the relationship between subsidy policies and fiscal equilibrium, the impact of subsidy withdrawal on poverty, and the effectiveness of a policy to protect low-income families from the effects of subsidy withdrawal in the country. The results demonstrate that a 25% reduction in subsidies could increase poverty by 0.253%. Nonetheless, relocating 50% of the budget designated for the subsidy to transfer income policies could cancel out this increase in poverty.

Another approach is that of ^{Plante (2013}), who built a small open economy model and applied it to fuel export and import economies. The goal was to ascertain how subsidies affect macroeconomic variables in the long term, what the role of the subsidy financing method is in those results, and if the distinction between import and export economy is in fact representative. For the import economy model, the author used a lump-sum tax, a labor tax, or a consumption tax for non-fuel goods. In the export economy, the subsidy consists of the sale of fuels at a price lower than the international one. The results show that the subsidy reduces the aggregate economic well-being. This loss of well-being happens because of the distortion of relative prices and not as a result of the way subsidies are financed. The subsidy can also decrease the consumption of non-fuel goods, alter the allocation of work between sectors, and distort relative prices.

In the Brazilian case, ^{Almeida et al. (2015}) calculate that between 2011 and 2014, losses on gasoline, diesel and liquified petroleum gas importation were BRL 21 billion; the resulting income loss from fuel sales at a price lower than the international one was a total of BRL 98 billion. Between 2010 and 2013, Petrobras’ market value decreased by 43%. The authors argue that the subsidy policy adopted by the government hampered investments by other businesses in the national petroleum refinement market, which became unattractive. To improve this situation, policies should be more transparent. For example, a stabilization fund could be created to buffer the elevation of fuel prices, if they are causing economic disequilibrium.

^{Bistafa (2016}) uses a computable dynamic general equilibrium model to study the macroeconomic impacts of petroleum exploration and gasoline price control policies in the ethanol sector, during the period from 2011 to 2014. The author concludes that the gasoline price control policy provoked a 7% reduction in ethanol production, as compared to the scenario without the policy.

^{Cunha (2015}), on the other hand, seeks to analyze the influence of diesel and gasoline prices on inflation in Brazil, calculating the direct effects, which consist of the influence of fuels on the inflation index, and indirect ones, which occur due to increases in input and freight costs.

The author utilized a product matrix input to evaluate the interdependence between productive sectors and as a consequence, identifies how readjustments in fuel prices affect the final price of other goods. Between 2005 and 2013, the 1% increase in the price of diesel caused an increase between 0.03 and 0.04% of the general index of market prices (GIMP). During the same period, a 1% increase in the price of gasoline elevated the General Consumer Price Index (GCPI) by values between 0.04 and 0.06%. The author concludes that the diesel price control policy contributed to inflation control, measured by the GIMP for the years 2007, 2009, 2010, 2011 and 2012. Gasoline price control contributed to GCPI control during the same years, in addition to in 2013.

This article aims to make an innovative contribution to the literature by introducing a model for calculating the changes that occur in consumption and production of fuels and other goods in the Brazilian economy when subsidies for fuel prices are withdrawn, and offset by a decrease in taxation. This work helps us, then, to understand the negative effects of the price control promoted in Brazil between 2011 and 2014, in addition to the potential benefits to the economy of removing subsidies.

3. METHODOLOGY

The model proposed by ^{Plante (2013}) was used here, adapted to the Brazilian economy, with the addition of taxation and a fuel production sector, which are not included in the original model. The model examines three sectors of the economy: fuels, tradable goods, and non-tradable goods. The government and families are also included in the model.

The Brazilian economy will be analyzed using data from 2013, evaluating what impact the subsidy withdrawal of 2014 could have on the economy. Accordingly, the new stationary state will be calculated in the absence of the subsidy, keeping in mind that when the government uses the subsidy, it must be financed with a tax increase. Additionally, the difference in relative prices of petroleum-derived goods also affects the stationary state of the economy.

4. DATA AND CALIBRATION

5. RESULTS

Having calculated all of the values, the fuel price value was changed so as to reduce the effect of taxes on the amount spent with the subsidy. What is seen, then, is a neutral change in revenue. There is also a simulation of eliminating the subsidy without changing tax rates.

The subsidy represented 0.45% of the GDP in 2013. This value was calculated from the difference between the subsidized price and the non-subsidized price, multiplied by the amount of fuel consumed. This amount was subtracted from the revenue of each tax to calculate the new contribution.

It was assumed that the price of tradable goods is internationally fixed, such that it was not affected by the changes in subsidies and taxation. Table 3 shows tax contributions when the subsidy policy was in use, and after it was eliminated.

Tables 4 and 5 demonstrate the changes effected by the selected variables when the fuel price subsidy is withdrawn and the aliquot of each tax is diminished to offset it. To calculate utility variations, the intertemporal elasticity of substitution values and the wage elasticity in relation to job supply from ^{Plante’s article (2013}) were applied. The values are τ=0.5 and are μ=1 Constants are k ₁ and are k ₂ were taken to be equal to 1.

6. CONCLUSIONS

This work aimed to evaluate the possible economic impacts caused by the removal of fuel subsidies utilized in the Brazilian economy between 2011 and 2014. For this purpose, ^{Plante’s (2013}) macroeconomic model was adapted to the Brazilian economy.

The results suggest that the removal of the fuel subsidy, compensated by a reduction in consumption, capital, or labor taxes, decreases fuel consumption and increases consumption and production of other goods.

When the removal of the subsidy is offset by a reduction of fuel taxes, consumption and productivity of all other goods increases, leading to a positive economic situation, though the model doesn’t consider the negative externalities caused by the increased consumption of fuel energy.

Until a certain point, this was the strategy used by president Dilma Rousseff’s government in 2011, when the CIDE tax on fuel was reduced to zero, as a way to avoid fuel price increases. The salient point is that afterwards, the only option was to reduce federal taxes, and the government explicitly assumed control of prices. This situation was worsened in 2015, when the federal and state governments promoted a tax increase on gasoline, along with the removal of subsidies. What this paper suggests is that a more appropriate path would have been the opposite one: to remove subsidies while decreasing taxes. Despite this, the worsening of the fiscal situation was ultimately more important, and taxes were raised.

The model has limitations in relation to the number of variables that can be analyzed, like the percentage of imported fuel and government spending. Environmental impacts are not analyzed either. For future work, an analysis of the variables not studied here is recommended, given the limitations of the model, such as the impact of removing subsidies for different social classes of Brazilian society. It would also be interesting to adopt a model which allows for an analysis of the economy’s transition to a scenario without subsidies. The contribution of subsidies to the increase of negative externalities caused by fuels could also be verified, like increases in transportation and contamination.