Features and assumptions

This model relies mostly on neoclassical microeconomics assumptions. First, the model assumes homogeneous production across firms, which means that industries produce a single good from their production process. Second, the model performs under a perfect competition environment, so in equilibrium, markets clear because of agents’ price-taking behavior. We also represent production and utility with homogeneous functions of degree to ensure constant-returns-to-scale technologies and homothetic consumer preferences. This results in industries’ production growing proportionally as the number of production factors increases. Similarly, households’ utility would change commensurate with the increase in final good consumption.

The model has other, less standard features suggested by previous research (see, e.g., Lysy and Taylor Reference Lysy, Taylor, Taylor, Bacha, Cardoso and Lysy1980; Dervis, De Melo, and Robinson Reference Dervis, De Melo and Robinson1982; Robinson Reference Robinson, de Janvry and Kanbur2006; Zalai and Révész Reference Zalai and Révész2016; Rada Reference Rada, Foster-McGregor, Alcorta, Szirmani and Verspagen2021). The model works under the small open economy assumption. This implies that this economy’s demand or supply cannot affect international prices. Furthermore, the model incorporates some short-run Keynesian closures (Robinson Reference Robinson, de Janvry and Kanbur2006; Zalai and Révész Reference Zalai and Révész2016; Rada Reference Rada, Foster-McGregor, Alcorta, Szirmani and Verspagen2021):

• External sector: The nominal exchange rate is assumed to be exogenous and fixed due to the full dollarization of the Ecuadorian economy. Therefore, the adjustment occurs in external savings, which depend on the financing needs of the private and public sectors.

• Government: Real government spending and tax rates are exogenous. Hence, the adjustment is made through an endogenous fiscal deficit.

• Factors: Real payments are flexible and determined through the marginal product of each factor. Factors are assumed to be perfectly mobile across sectors.

• Savings investment: The marginal propensities to saving of private institutions, the real investment and the capital endowment are kept fixed. Labor employment is left free to ensure the savings-investment balance.Footnote ¹

Figure 1 illustrates the economic flow assumed by the model. In this economy, the twenty-two firms listed in Table 2 make their optimal decisions for intermediate consumption, value-added, and production allocation. Households interact with the domestic and global markets to choose their good consumption. In the model, five households are representative of each income quintile in the Ecuadorian economy. For a more thorough revision of agents’ behavior, the mathematical development is explained in the following sections.

Figure 1. Ecuadorian economic flow assumed in the model. Blue boxes identify economic agents. Red boxes represent agents’ preferences and technologies. Arrows identify economic flows and interactions between agents.

Table 2. Economic activities included in the model

Firms

Each firm pursues its own profit maximization by first minimizing its costs and following a zero-profit condition due to technologies with constant returns to scale. The technological restriction consists of a three-tier nesting: intermediate consumption, value-added, and final production.

Intermediate consumption (box 1 in Figure 1) assumes that intermediate goods are perfect complements, so one unit of a final good will require at least a fixed share of each intermediate good to be produced. A Marx-Leontief technology is used to model this behavior:

(1) $${I{C_j} = \mathop {\min }\limits_{1 \le i \le n} \left\{ {{{{X_{ij}}} \over {{\gamma _{ij}}}}} \right\},\;}$$

where $I{C_j}$ is the aggregate intermediate consumption of firm $j$ , ${X_{ij}}$ is firm $j$ ’s demand for intermediate good $i$ , and ${\gamma _{ij}}$ identifies its fixed share of $I{C_j}$ .

Value-added (box 2 in Figure 1) employs a classical production function of two inputs: capital and labor. A Cobb-Douglas function is used to model a fixed substitution equal to 1 for these factors of production:Footnote ²

(2) $${V{A_j} = A_j^{va}{K_j}^{{\alpha _j}}{L_j}^{1 - {\alpha _j}},}$$

where $V{A_j}$ is the aggregate value added of firm $j$ , $A_j^{va}$ is its total factor productivity, ${K_j}$ and ${L_j}$ identify its capital and labor demands, respectively, and ${\alpha _j}$ identifies its capital share of $V{A_j}$ .

Final production (box 3 in Figure 1) aggregates value-added and intermediate consumption. Again, we assume a Marx-Leontief functional form for this behavior to model a fixed share for each aggregate:

(3) $${{Q_j} = min\left\{ {{{I{C_j}} \over {\delta _j^{ic}}},{{V{A_j}} \over {\delta _j^{va}}}} \right\},}$$

where ${Q_j}$ is the final production of firm $j$ , $\delta _j^{ic}$ and $\delta _j^{va}\;$ are the intermediate consumption and value-added fixed shares of ${Q_j}$ , respectively.

External sector

The external sector is modeled as a single entity: rest of the world. There is no distinction by destination or origin.

Exports (box 4 in Figure 1) are modeled by a technology with constant elasticity of transformation (CET). Thus, each firm decides between selling its final production on the domestic market ( ${D_j}$ ) or exporting it ( ${E_j}$ ) based on the following function:

(4) $${{Q_j} = A_j^q{{\left( {{\phi _j}E_j^{{{1 - {\rho _j}} \over {{\rho _j}}}} + \left( {1 - {\phi _j}} \right)D_j^{{{1 - {\rho _j}} \over {{\rho _j}}}}} \right)}^{{{{\rho _j}} \over {1 - {\rho _j}}}}},}$$

where, for each firm $j$ , $A_j^q$ is an efficiency parameter, ${E_j}$ and ${D_j}$ identify the final production allocated on foreign and domestic markets, respectively, ${\phi _j}$ identifies the share of exports, and ${\rho _j}$ is the elasticity of transformation.

On the other hand, imports (box 5 in Figure 1) are modeled using a technology with constant elasticity of substitution (CES). Hence, aggregate supply for national consumption of good $j$ ( ${S_j}$ ) equals the sum of the quantity of final production of firm $j$ allocated to the domestic market ( ${D_j}$ ) and the imported quantity of this good ( ${M_j}$ ), as shown in the following equation:

(5) $${{S_j} = A_j^s{{\left( {{\beta _j}M_j^{{{{\sigma _j} - 1} \over {{\sigma _j}}}} + \left( {1 - {\beta _j}} \right)D_j^{{{{\sigma _j} - 1} \over {{\sigma _j}}}}} \right)}^{{{{\sigma _j}} \over {{\sigma _j} - 1}}}},}$$

where, for each good $j$ , $A_j^s$ is the efficiency parameter, ${M_j}$ stands for imports, ${D_j}$ is the final production allocated to the domestic market, ${\beta _j}$ represents the share of imports, and ${\sigma _j}$ is the elasticity of substitution.

Households

Households (box 6 in Figure 1) are classified by income quintiles. Each household selects its optimal basket for utility maximization, constrained by its disposable income. The utility function is a two-tier nested specification that follows a CES form for low and high-octane gasoline, and a Cobb-Douglas form for the rest of the goods. This nesting is used to capture the substitution between both types of gasoline available in Ecuador.

Formally, the representative household’s utility is stated as follows:

(6) $${{U_h} = \mathop \prod \limits_{i \notin O} C_{hi}^{{\theta _{hi}}}\;{{\left[ {\mathop \sum \limits_{k \in O} {\eta _{hk}}C_{hk}^{{{\omega - 1} \over \omega }}} \right]}^{{\omega \over {\omega - 1}}\;\overline {{\theta _h}} }},\;\mathop \sum \limits_{\rm{i}} {\theta _{hi}} + \overline {{\theta _h}} = 1,}$$

where ${U_h}$ represents the utility of household $h$ , ${C_{hi}}$ identifies its final consumption of good $i$ , ${\theta _{hi}}$ is the $ith$ good’s share in final consumption ( $\overline {{\theta _h}} $ for gasoline), ${\eta _{hk}}$ is the share of the $kth$ octane gasoline in gasoline basket, ω is the elasticity of substitution for gasoline, $p_i^S$ is the consumer price of good $i$ , and $O$ is the gasoline basket (low and high-octane gasoline).

The income constraint establishes that the total expenditure for each household is equal to the available income for consumption as follows:

(7) $${\mathop \sum \limits_k {p_k}{C_{hk}} = \;\varphi _h^{dsp}\left( {\varphi _h^L{w_L}L + \varphi _h^K{w_K}K + t{r_h}} \right),}$$

where, for each household $h$ , $\varphi _h^{dsp}$ is the marginal propensity of consumption, $\varphi _h^L$ is the share of the compensation to the employees ( ${w_L}L)$ , $\varphi _h^K$ is the share of the gross capital returns ( ${w_K}K)$ , and $t{r_h}$ are the net cash transactions received from or paid to others private agents or government.

Government

Government intervention is assumed exogenous and consists of two main activities: tax collection from direct and indirect rates and distribution of taxes through public consumption, investment, transfers, and subsidies.

Official data from the SAM identifies three main indirect tax rates which affect the price system: production tariffs ( $\tau _i^{tar}$ ), import rates ( $\tau _i^m\;$ ) and value-added rates ( $\tau _i^{va}$ ). Also, prices are modified by specific subsidies ( ${\psi _i}$ ). Therefore, the relationship between prices before and after fiscal intervention is described below:

(8) $${p_i} = {\left( {1 + \tau _i^{va}} \right)}{\left( {1 + \tau _i^m\;} \right)}{\left( {1 + \tau _i^{tar}} \right)} {p_i^{'}} - {\psi _i},$$

where, $p_i^{'}$ is the price of the $ith$ good before fiscal intervention and ${p_i}$ is the price of the $ith$ good after tax and subsidy policies.

Equilibrium

The economic system shown in Figure 1 has two equilibrium equations. The first one occurs in the goods market:

(9) $${S_i^* = \mathop \sum \limits_{j = 1}^n X_{ij}^* + \;\mathop \sum \limits_{h = 1}^m C_{hi}^* + {G_i} + {I_i},}$$

where, for each good $i$ , $S_i^*$ identifies its aggregate supply for national consumption, $X_{ij}^*$ represents the aggregate intermediate consumption by firm $j$ , $C_{hi}^*$ identifies the aggregate final consumption by household $h$ , ${G_i}$ represents the government consumption, and ${I_i}$ is the investment.

The second equilibrium equation occurs in the factors market:

(10) $${\mathop \sum \limits_{i = 1}^n {L_j} = \overline L,\;\;\mathop \sum \limits_{i = 1}^n {K_j} = \;\overline K,}$$

where $\overline L$ and $\overline K$ are the fixed total factor endowments of labor and capital, respectively.

These equilibrium equations consider the behavior of all agents above, that is: households maximize their income-constrained utility, firms minimize their total cost and obtain zero-profits locally and abroad, and government spending equals tax collection plus public financing needs. The model solves only for relative prices, with the nominal wage serving as numéraire and being set as 1.

Calibration

The general equilibrium model was calibrated on a social accounting matrix (SAM) with data from 2014.Footnote ³ This matrix provides information about the real economy, which includes intermediate consumption, production, trade balance, value-added, income generation, distribution and allocation, taxes, and subsidies. Also, we used official information from the 2014 Supply and Use Tables, published by the Central Bank of Ecuador, to disaggregate diesel, gasoline, and other fuel products that was originally in one merged sector in the SAM. The consistency of the disaggregated matrix with official data was verified through the RAS technique, as proposed by Bacharat (Reference Bacharat1970).

Data limitations

Although the absence of recently updated official data is a challenge that cannot be readily surmounted, it is a limitation worth mentioning for the model calibration. Because liquid fuel subsidies were more strongly discussed after 2019, the 2014 SAM is not admittedly the most accurate approximation of the Ecuadorian economic context of these subsequent years. Between 2014 and 2019, the macroeconomic environment suffered from a collection of not-minor shocks, two of the most significant being the fall in oil prices in 2015 and the earthquake in 2016, whose implications may result in an overestimation of the effects of simulation scenarios, mainly on fiscal accounts.

In fact, during these years, the structure of the fiscal budget was drastically adjusted according to official data. The fall in Ecuadorian oil prices below USD 30 per barrel sharply reduced fiscal oil revenue by up to 38% in 2015. Consequently, the government adjusted its budget by getting unilateral credits, issuing sovereign bonds, and cutting public investment by almost 18%. The oil industry’s strong linkages with other sectors rapidly transmitted the shock to the whole economy, so gross domestic product (GDP) decelerated up to 3 percentage points. More importantly for the model, this complex environment affected subsidy spending because the fall in the oil price market also reduced the prices of oil derivatives. This caused a partial benefit for the fiscal budget due to the transitory reduction in more than half the amount of subsidy spending to USD 2.3 billion from USD 5.6 billion. Although oil prices started their recovery in 2016, the government was not able to fully take advantage since the public deficit reached up to 10% of GDP after that year’s earthquake in the country.

Therefore, the data used for calibration shows a stronger economic environment than the time when reforms started, and it obscures the real dimension of the 2015–2016 recession and its long-lasting repercussions in the subsequent years. If the official SAM is updated to more recent data, the contraction of oil revenue in 2015 and 2016 as well as its slow recovery until 2019 would likely result in a base scenario with a higher fiscal deficit and lower subsidy pressure in the government budget in comparison to the one seen in the 2014 SAM. Thus, it is crucial to acknowledge that the model results are just a reference level with a highly probable overestimation bias due to the dissimilitude in fiscal budget structures from 2014 to 2019.

Background

Ecuador’s fuel subsidy policy was first implemented in 1974 with the main purpose of improving access for those who were not able to afford international market prices (Andrade Reference Andrade2011; Carrillo-Maldonado, Díaz-Cassou, and Tejeda Reference Carrillo-Maldonado, Díaz-Cassou and Tejeda2018; Jara et al. Reference Jara, Chun Lee, Montesdeoca and Varela2018). The policy framework for diesel, gasoline, and LPG consisted of fixed prices—established by presidential decrees—so that the difference between costs and returns of imported fuel resulted in the amount of government subsidy. In 1977, the first Ecuadorian oil refinery started the manufacturing of domestic fuel products that also were included in the subsidy scheme (Acosta Reference Acosta2001). Because of the lack of correlation with international prices, those subsidies started a fiscal gap that many administrations have tried to reform.

During the 1980s the Ecuadorian economy suffered the impacts of a collapse of its oil revenues; thus, some discretionary adjustments were made to gasoline and LPG subsidies. Nevertheless, those changes responded only to fiscal pressures, not to a systematic pass-through of international movements of prices (Carrillo-Maldonado, Díaz-Cassou, and Tejeda Reference Carrillo-Maldonado, Díaz-Cassou and Tejeda2018). The absence of compensation measures triggered some national mobilizations, which intensified in the 1990s, when more attempts at reform and the economic instability led to a governance crisis in which seven different presidents held office in ten years (1996–2006; most were overturned by riots due to fuel subsidy-related decisions).

Throughout the first five years of the decade 2010–2015, the Ecuadorian government sought an energy transition to hydro-oriented sources (Carrillo-Maldonado, Díaz-Cassou, and Tejeda Reference Carrillo-Maldonado, Díaz-Cassou and Tejeda2018). Although gasoline and diesel subsidies remained the same, a progressive elimination of LPG subsidy was planned to be executed as of 2016. For this purpose, the government designed the Efficient Cooking Program (PCE, in its Spanish acronym), which consisted of incentives for the manufacturing and purchasing of low-cost induction cooktops. The ultimate goal was to reduce LPG consumption so that its subsidy can be eliminated without harming vulnerable sectors and thus avoiding social reluctance (Chanatásig and Salazar Reference Chanatásig and Salazar2018). However, both supply and demand slowed due to the economic recession of 2015–2016; therefore, the reform to LPG subsidy was left behind to focus on diesel and gasoline subsidies in recent years.

Incentives to a gasoline and diesel subsidy reform

Ecuador is a full-dollarized, small, open economy with high level of subsidies (Ríos-Roca, Garrón, and Cisneros 2007; Mendoza Reference Mendoza2014; Marchán, Espinasa, and Yépez-García Reference Marchán, Espinasa and Yépez-García2017). During the last commodity boom between 2011 and 2014, the country spent yearly almost 4% of its GDP by fixing fuel prices while its fiscal deficit reached an average of 5%. Most pressures came from subsidizing diesel, which has been historically sold up to third the international market price and represented around of half the total amount of subsidies spending, as Figure 2 shows.

Figure 2. Diesel subsidy insights. Source: Central Bank of Ecuador. The shaded area between 2011 and 2014 underscores the commodity boom period.

Regarding gasoline, Ecuador had two types of subsidized products based on quality: high octane (super) and low octane (extra or ecopaís). While the first type has been mainly used by the richest quintiles, the low-octane gasoline has been widely consumed despite household income (Jara et al. Reference Jara, Chun Lee, Montesdeoca and Varela2018). Hence, there has been an unequal distribution of the direct benefits apparently biased in favor of the high-income households.

In 2015–2016, the Ecuadorian economy suffered several negative effects on its balance of payments. External shocks like the decline of its terms of trade (–31% annual in February 2015) mainly due to the fall of its oil revenues (–67% annual in February 2015), the appreciation of its real effective exchange rate (14% annual in August 2015) influenced by the sharp depreciation of the nominal exchange rate of Colombia and Perú (–37% and –13% annual in August 2015, respectively), and some materialized natural risks such as El Niño phenomenon and the magnitude 7.8 earthquake in April 2016 passed through the external and real accounts until the government balance.

After applying short-term fiscal and monetary impulses to alleviate the impacts on the real sector, the Ecuadorian economy shrank by –1.2% of GDP in 2016. As a result of the debt expansion to cover financial needs between 2014 and 2016, the external debt had increased almost 10 points to 27% of GDP, and the Central Bank’s liquid assets had reduced by 17 points to 50% of total assets. Therefore, some structural austerity measures started to be part of the economic agenda of the Ecuadorian government in the last months of 2017 in order to decrease sovereign risk.

Recent policy changes

Regarding the fuel subsidy policy, reforms began in 2018. Through the Presidential Decree 2018-490, the price of high-octane gasoline increased by 42%. Arguing that this type of gasoline is mostly consumed by expensive cars owned by the economic elite, the government was able to avoid the political risks of the reform, so it was successfully implemented. With the same argument, the second reform was launched a few months later in Presidential Decree 2018-619. However, this reform not only eliminated all the amount of subsidy for high-quality gasoline but also increased the price of low-octane gasoline (by up to 25%) and laid the path for upcoming social unrest.

In March 2019, Ecuador signed an Extended Fund Facility (EFF) agreement with the IMF to solve the government’s financial needs. The letter of intention claimed a reform had to be carried out in the same year. The reform was launched in October with Presidential Decree 2019-883, and it consisted of liberalizing diesel and gasoline prices. Due to the sudden elimination of subsidies, some riots against the change occurred. After eleven days of nationwide mobilizations, political leaders started conversations with the main groups against the reform. The bargaining process resulted in the Ecuadorian government not only derogating the reform but also replacing it with a new decree that called for further rationalization. The latter decree prompted policymakers to revisit and study new targeting mechanisms, and it stated that forthcoming cuts to fuel subsidies must mitigate the impact and focus their benefits on the most vulnerable sectors.

Throughout 2020–2022, the COVID-19 pandemic severely affected Ecuador’s financial accounts. The international shock in oil prices motivated more austerity measures and four new fuel subsidy reforms in less than three years. Among new policies, the government applied a transitory band system to low-octane gasoline and diesel prices so that subsidies activated only if their costs of production or import increase by a monthly rate greater than 5% (Executive Decree 2020-1054; later reduced to 3% for diesel in Executive Decree 2020-1222). The uncertainty surrounding the effects of this new, apparently complex scheme and the lockdown due to COVID-19 initially shut down any possibility of social mobilizations. However, as the international oil price gradually rose throughout 2021, the accumulated change in fuel prices reached more than 50%. After fourteen months with the band system, the government released Executive Decree 2021-231, by which it returned to the traditional fixed-price scheme. This reform was sustained for seven months until June 2022, when new riots urged political leaders for reducing prices. In the aftermath of eighteen days of social unrest, the Ecuadorian government yielded and cut USD 0.15 gasoline and diesel prices through the Executive Decree 2022-467. Again, this decree was launched with the compromise of studying new targeting mechanisms.

Scenario delimitation

Bearing in mind the context previously discussed, we chose four scenarios to carry out contrafactual assessments. Two of the scenarios involve government-adopted reforms aimed at the partial or total elimination of fuel subsidies, and the remaining scenarios assess two proposals for targeting.

Scenario 1: Decree 619, subsidy elimination for high-octane gasoline and reduction for low-octane gasoline. The first scenario aims to explore the current context of fuel subsidy policy.Footnote ⁴ It is based on the government’s decision launched in December 2018 in Executive Decree 619 (Table 3): a 40% reduction in low-octane gasoline subsidies (for a 25% increase in the final price) and the total elimination of high-octane gasoline subsidies. ^{Footnote 5}

Table 3. Summary of the fuel subsidies reform in the Executive Decree No. 619

Source: Agency for Hydrocarbon Regulation and Control of Ecuador

Scenario 2: Decree 883, elimination of subsidies for gasoline and diesel. The second scenario seeks to estimate the upper-bound effects of eliminating fuel subsidies without any targeting or compensating measure. Therefore, it summarizes the attempt to liberalize low-octane gasoline and diesel prices through Executive Decree 883 (Table 4), which was subsequently repealed. With the subsidies’ removal, the diesel price increased by 121%, while gasoline prices rose by 29%.

Table 4. Summary of fuel subsidies reform in the Executive Decree No. 883

Source: Agency for Hydrocarbon Regulation and Control of Ecuador

Scenario 3: Targeting household final consumption. This scenario seeks a targeting alternative that eliminates subsidies depending on the type of fuel and household income quintile. Targeting encompasses households in the poorest quintile for diesel and the two poorest quintiles for the low-octane gasoline (Table 5). The nontargeted quintiles suffer a 100% elimination of subsidies so that this scheme seeks to eliminate around 90% of subsidies’ benefits received by the richest quintiles.

Table 5. Proposal for targeting household final consumption

Scenario 4: Targeting household final consumption and intermediate consumption. Scenario 4 targets household final consumption in the same way as scenario 3 but also incorporates firms’ intermediate consumption to the subsidies’ targeting. It keeps diesel subsidies for strong-linked economic activities, according to their Leontief multipliers; meanwhile, it considers a partial 50% reduction in diesel subsidies for weakly linked economic activities. This scheme is applied to avoid a systemic output contraction due to the higher costs of intermediate goods (Table 6).

Table 6. Proposal for targeting intermediate consumption

Note: SF: strong forward. SB: strong backward. Despite diesel’s weak linkage with the rest of activities, it is excluded from the reform to avoid spillover effects from self-consumption.