You are on page 1of 16

Energy Policy 150 (2021) 112145

Contents lists available at ScienceDirect

Energy Policy
journal homepage: http://www.elsevier.com/locate/enpol

Tension in Mexico’s energy transition: Are urban residential consumers in


Aguascalientes willing to pay for renewable energy and
green jobs?☆,☆☆,☆☆☆
Adán L. Martínez-Cruz a, b, c, Héctor M. Núñez b, *
a
Department of Forest Economics, and Centre for Environmental and Resource Economics (CERE), Swedish University of Agricultural Sciences (SLU), Sweden
b
Department of Economics, Centro de Investigacion y Docencia Economicas (CIDE), Mexico
c
Centre for Energy Policy and Economics (CEPE), ETH-Zurich, Switzerland

A R T I C L E I N F O A B S T R A C T

Keywords: Via a discrete choice experiment (DCE), a sample of urban residents that contribute to their household electricity
Renewable energy sources bill in Aguascalientes, Mexico, has been asked to choose from among four electricity contracts —a status quo
Green jobs alternative, and three alternatives described in terms of type of renewable energy source (RES), % of RES in
Just energy transition
current electricity mix, new jobs in RE sector, and % increase in self-reported bimonthly electricity bill. Re­
Discrete choice experiment
Aguascalientes
spondents report a positive willingness to pay (WTP) for both RES and new jobs in RE sector, and a higher WTP
Mexico for solar energy in comparison to biomass energy. These results are timely as there is a tension in Mexico due to
the energy strategy pursued by the current federal administration —which in practice has slowed down the
energy transition initiated in 2015. This paper’s findings suggest that respondents to our DCE would support a
just energy transition aiming to boost both RES and the creation of green jobs.

1. Introduction transmission tariffs, indefinite suspension of critical trials for new pro­
jects involving renewable energy sources (RES), and a new energy
While pursuing energy transitions, societies worldwide are facing program that discourages the pursuing of RES projects (SENER, 2020).
challenges imposed by tensions among existing and new participants of The change in priorities has been received with concerns from both
energy markets (Baker and Phillips, 2019; Chen et al., 2019). Case in national and international stakeholders due to its negative implications
point, several stakeholders in Mexico are currently pursuing measures to for climate change goals, its potential economic inefficiency, and the
counterbalance federal administration’s actions that have slowed down uncertainty it generates. Several state administrations, renewable en­
Mexico’s energy transition. ergy companies, and NGOs have pursued legal action and have been able
Mexico’s commitment to energy transition was expressed in 2015 to temporarily stop implementation of the federal administration’s de­
through the Energy Transition Law which aims to opening the Mexican cisions. For instance, Mexican justice has postponed the suspension of
energy market to competition and adopting clean energy technologies the critical trials policy and the new energy program —particularly, the
and targets —including decarbonization strategies for the electricity plans to reduce RES′ participation (Garcia, Karol, 2020; El Universal,
sector (SENER, 2018c). Since 2018, however, a change in federal 2020b).
administration has led to new priorities. The focus is now on bringing The economy of the State of Aguascalientes embodies well the ten­
the energy industry back under government control (SENER, 2020). sions that federal energy policies have created. Aguascalientes’ economy
Four specific actions represent in practice a backlash to Mexico’s energy heavily relies on foreign investment due to the exports from its car
transition: cancellation of clean energy auctions, increase in manufacturing industry. Consequently, employment and new


This work was supported by the Fondo Sectorial CONACYT-SENER Sustentabilidad Energética, México [projects No. Cluster Biocombustibles Gaseosos-247006
and 245750].☆☆ A previous version of this manuscript was circulated under the running title “Stated Willingness to Pay for Residential Renewable Electricity and
Green Jobs in Aguascalientes, Mexico”.☆☆☆ The authors gratefully acknowledge constructive comments from two anonymous reviewers, and the assistance of Joel
Vazquez, Eduardo Robles, Humberto Dena, Jimena Bravo, Emiliano Huerta, Victoria Ramírez, and Miguel Mojica.
* Corresponding author. Circuito Tecnopolo Norte 117, Aguascalientes, AGS, CP 20313, Mexico.
E-mail addresses: adan.martinez.cruz@slu.se (A.L. Martínez-Cruz), hector.nunez@cide.edu (H.M. Núñez).

https://doi.org/10.1016/j.enpol.2021.112145
Received 10 April 2020; Received in revised form 4 January 2021; Accepted 6 January 2021
Available online 23 January 2021
0301-4215/© 2021 Elsevier Ltd. All rights reserved.
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

investments can be severely affected by policies that generate uncer­ in an emerging economy.
tainty to the private sector, including the energy sector. In addition, There is a well-established literature on stated preferences of urban
located in a semi-arid region with high potential and installed capacity residents for RES. Findings from this literature have been consolidated
to generate solar energy, it makes sense for Aguascalientes to invest in in one literature review (Oerlemans et al., 2016), and four meta-analyses
RES. Indeed, the administration of Aguascalientes has committed to (Ma et al., 2015; Soon and Ahmad, 2015; Sundt and Rehdanz, 2015;
continuing RES projects to counterbalance federal energy policies (see Pokhrel, 2016).3 These synthesizing papers have documented a strong
Sanchez, 2020). focus on developed countries —particularly, U.S., Canada, EU countries,
In this tense context, preferences of energy end-users seem to have China, South Korea, and Japan. A few studies have documented stated
been overlooked. Households are as important as suppliers because preferences for RES among urban residents of non-developed countries
changes in their preferences, choices, and behaviors will trigger a suc­ —e.g. Kenya (Osiolo, 2017), South Africa (Oliver et al., 2011), Crete
cessful energy transition (Steg et al., 2018). According to INEGI (2018b), (Zografakis et al., 2010), and Chile (Aravena et al., 2012).4
77% of the Mexican households are expecting an increase in the share of Previous DCE studies have dealt with preferences for creation of jobs
electricity that is generated with RES in the following years.1 SENER together with preferences for RES.5 Table 1 breaks down the attributes
(2018d) reports that residential consumers represent about 28% of the considered in previous DCE studies into three categories —attributes
total electricity consumption in the country. Hence it seems reasonable capturing RES features, price attribute, and other attributes. Within the
to wonder whether urban residential consumers would support RES other attributes’ category, we describe in bold letters how previous
further. studies have incorporated attributes reflecting creation of jobs. In esti­
In addition, as energy transitions are expected to displace jobs in mating costs and benefits from externalities created by energy renew­
industries that rely on fossil fuels, exploring whether residential elec­ able investments in Scotland, Bergmann et al. (2006) uses a DCE that
tricity consumers are willing to pay a premium in exchange of creation includes new local long-term jobs as one of its attributes. Departing from
of new green jobs is useful to design policies that aim to simultaneously the fact that RES impose societal-environmental trade-offs and with
increase generation of RES and creation of jobs —which is essential for a focus in Finland, Kosenius and Ollikainen (2013) include number of new
just energy transition (Rosemberg, 2017). local jobs resulting from producing energy with RES. Longo et al. (2008)
This paper analyzes data gathered through a discrete choice exper­ explore preferences for programs promoting production of RES in Bath,
iment (DCE) which was implemented in November 2019 on urban res­ England, and include change in the number of employees in the electricity
idents that contribute to paying their household electricity bill in the sector as an attribute in their DCE. Soliño et al. (2012) document pref­
city of Aguascalientes. This city hosts 1.2 million residents (about 68% erences for electricity generated from forest biomass in Spain and
of the state population), it is fully covered by the electricity grid, and its include creation of jobs in rural areas as an attribute. Yoo and Ready
residential consumers represent 87% of total electricity demand in the (2014) document preferences for RES in Pennsylvania, U.S., and list
sate of Aguascalientes (Secretaria de Energía, 2016). Respondents to our impacts on jobs in Pennsylvania as attribute of their DCE.6
DCE report positive willingness to pay (WTP) for both RES and creation Table 1 also offers insights about how previous studies have incor­
of new green jobs —results that, we argue, highlight their interest on porated attributes that capture RES. One set of studies have documented
policies aiming to boost both RES and green jobs. As this sample was preferences for RES treating them in a generic manner (e.g. Alberini
gathered prior COVID-19 hit, we discuss whether these findings may et al., 2018; Bergmann et al., 2006; Longo et al., 2008). For instance,
hold or not for post-pandemic times. Alberini et al. (2018) explore preferences for policies aiming to reduce
To the best of our knowledge, no previous study has documented CO2 emissions in Italy and Czech Republic, and have inferred prefer­
stated preferences for both RES and green jobs among urban residents of ences for RES by presenting an attribute that describes the goal of the
a city located in an emerging economy.2 Previous DCE studies have policies as either pursuing energy efficiency or promoting RES. A second set
mostly focused on preferences of residents in cities located in developed of studies presents the RES attribute as a categorical variable that
countries (see Oerlemans et al., 2016; Ma et al., 2015; Soon and Ahmad,
2015; Sundt and Rehdanz, 2015; Pokhrel, 2016). Our focus on an
emerging economy is not only a matter of geographical location. Pre­ 3
Due to their period of analysis, these studies do not cover the recent arrival
vious DCE studies have placed their contribution in contexts in which of vignette designs to this literature. These designs have been implemented to
RES are positioning themselves into the energy market as a result from explore preferences for electricity mixes (that may include RES) in post-
serious efforts pursuing an energy transition —in particular, studies Fukushima Japan (Rehdanz et al., 2017), and residential consumers’ prefer­
focused on EU countries and specific states in U.S. (e.g. Bergmann et al. ences for RES and supply reliability in Germany and Great Britain (Merk et al.,
2006, Borchers et al. 2007, Longo et al. 2008, Susaeta et al. 2011, Gracia 2019).
4
et al. 2012, Kosenius and Ollikainen 2013, Yoo and Ready 2014, Vec­ A related literature documents stated preferences of rural residents for
chiato and Tempesta 2015, Botelho et al. 2018, Alberini et al. 2018, electricity generated with RES (e.g. Graber et al., 2018; Abdullah and Mariel,
2010). This literature is related to our study when it comes to its focus on
Mamkhezri et al., 2020). In contrast, this study’s focus on Aguasca­
preferences for RES and its methodology (i.e. their use of DCEs). We do not
lientes is motivated by the uncertainty around whether Mexico will
cover this literature here because it considers attributes that are not directly
seriously and strongly pursue an energy transition in the near future. relevant to inform this study’s DCE. For instance, Graber et al. (2018) focuses
on a context in which there is a lack of continuous electricity service, and
2. Review of empirical studies on preferences for renewable consequently it includes the number of night hours with electricity as one of the
energy sources and green jobs relevant attributes. Our focus on urban residential consumers that have access
to a continuous electricity services make us focus mostly on previous studies
This section places this paper within a literature documenting stated documenting urban residential preferences.
5
preferences for both RES and green jobs among urban residents. It also While contingent valuation (CV) and discrete choice experiments (DCE) are
emphasizes that there are no previous studies focusing on a city located the two preferred methods to collect stated preferences for residential RES (see
Oerlemans et al., 2016; Ma et al., 2015; Soon and Ahmad, 2015; Sundt and
Rehdanz, 2015; Pokhrel, 2016), we have chosen a DCE design because it is
better fitted to incorporate variation in several attributes at once.
1 6
Survey by INEGI (2018b) inquires specifically about the perception of the A few studies have pursued an alternative strategy to capture a social
future situation of use of solar and wind resources for residential electricity. attribute that indirectly encompasses creation of jobs. For instance, Vecchiato
2
For the specific case of Mexico, Reyes-Mercado and Rajagopal (2014) and Tempesta (2015) describe their social attribute as certified origin of biomass
document attitudes and perceptions towards the adoption of domestic water and Gracia et al. (2012) phrase it as region of origin of RES; and Kim et al. (2018)
solar boilers but stopped short from documenting willingness to pay for RES. describe it as utility company’s social contribution.

2
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

Table 1
Attributes included in selected previous DCE studies exploring preferences for RES.
Study Aim Location RES attribute Price attribute Other attributes

Alberini et al. Explore preferences for Italy and Czech None Annual cost to household Goal of policy (energy efficiency v.
(2018) policies seeking to reduce Republic during 10 years s. promotion of RES); mechanisms
CO2 emissions Italy: 25, 50, 100, 300 (incentives, regulation, taxes on
Euros; Czech Republic: 400, fossil fuels, information-based
800, 2,000, 5000 CZK approaches); and annual reduction
in CO2 emissions (0.25, 0.50, 1.00,
1.65 tons)
Bergmann Estimate costs and benefits 8 council districts of None Annual increase in Impacts on landscape (none, low,
et al. from externalities created Scotland household electricity bill (0, moderate, high); impacts on
(2006) by RES investments 7, 16, 29, 45 GBP) wildlife (slight improvement, no
impact, slight harm); air pollution
(none, slight increase); and new
local long-term jobs (1–3, 8–12,
20–15)
Botelho et al. Compare welfare impacts Portugal Energy source: wind power, Increase in monthly Significant effect on landscape
(2018) from wind, solar, and hydropower, solar photovoltaic electricity bill (4, 8, 12 (present, absent); significant effect
hydropower Euros) on fauna and flora (present,
absent); glare effect significantly
affecting population (present,
absent); and noise significantly
affecting population (present,
absent)
Borchers Explore references for New Castle County, Energy source: wind, solar, biomass, Increase in monthly None
et al. voluntary participation in Delaware, USA farm methane, generic green energy electricity bill (5, 10, 15,
(2007) green energy electricity source; and % of respondent’s 20, 30 USD)
programs monthly electrical usage: 10%, 25%
Cicia et al. Explore preferences for Italy Energy source: wind, solar, biomass, Changes in bimonthly None
(2012) different RES and nuclear electricity bill: 0, 5, 10, 15,
20, 30 Euros for all sources
but nuclear; and 0, -5, -10,
− 15, − 20, and − 30 Euros
for nuclear source
Gracia et al. Explore preferences for Zaragoza, Spain % of respondent’s electrical usage Price per kWh (0.17, 0.21, Region of origin (Aragon,
(2012) mix of RES from wind (16%; 18%; 21%; and 0.24, 0.28 Euros) unknown)
26%), from solar (6%; 10%; 14%;
and 18%); and from biomass (2%;
3%; 5%; and 6%)
Kim et al. Explore preferences for Residential areas Share of electricity usage covered via Additional monthly Smart meter (installed, not
(2018) RES with the highest energy sources: maintaining current electricity bill (1, 3, 5 USD) installed); number of blackouts per
population in South shares; significantly increasing share year (1, 2, 3); duration of each
Korea of nuclear energy; significantly blackout (10, 30, go minutes); and
increasing share of fossil fuels; utility company’s social
significantly increasing share of RES. contribution (passive, active)
Kosenius and Valuation of Finland Energy source: wind, crop, wood, Additional payment in the Impact on biodiversity in the
Ollikainen environmental and societal hydro electricity bill on yearly proximity of energy production (no
(2013) trade-offs from RES basis (5, 30, 80, 160 Euros) change, deterioration,
improvement); change in carbon
emissions (− 60%, − 70%, − 90%,
− 95%, − 97%, − 99%); and number
of new local jobs resulting from
producing energy with RES (20,
100, 400, 500, 800, 1,400, 2,500,
5000)
Longo et al. Explore consumers’ Bath, England None Increase in yearly Annual reduction in GHG emissions
(2008) preferences for programs electricity bill (6.5, 16, 25, due to RES increase (1%, 2%, 3%);
that promote the 38 GBP) annual length of electricity
production of RES shortages in minutes (30, 60, 120);
and change in number of
employees in the electricity
sector (- 1,000, 0, + 1000)
Murakami Explore consumers’ 4 states in USA and Fuel Mix: fossil fuels (coal, oil, gas), Monthly electricity bill Reductions in air emissions (Nox,
et al. preferences for nuclear Japan Nuclear, RES (solar, wind, biomass, USA: 90, 100, 110, 120 USD SO2, CO2): No reduction, 20%
(2015) and renewable electricity geothermal, etc.), and hydroelectric; Japan: 7,000, 8,000, 9,000, lower, 40% lower, and 60% lower
after the Fukushima and ratio of energy generation: 0%, 10,000 JPY
nuclear plant accident 10%, 20%, and 30%
Soliño et al. Explore consumers’ North West region None Two versions:annual (30, Reduction in CO2 emissions (7%,
(2012) preferences for electricity of Spain 60, 90, 120 Euros); and 14%); contribution to lowering
generated from forest bimonthly (5, 10, 15, 20 depletion of natural resources
biomass Euros) (contributes, does not contribute);
decrease in risk of fires (25%, 50%,
75%); and creation of jobs in rural
areas (3,000, 6000)
Su et al. Explore consumers’ Kaunas region, Installation costs (1,500, 3,000,
(2018) preferences for renewable Lithuania 4,500, and 6500 Euros); length in
(continued on next page)

3
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

Table 1 (continued )
Study Aim Location RES attribute Price attribute Other attributes

micro generation Renewable microgeneration Average monthly bill for warranty period (2, 5, 10, 13
technologies technologies: solar thermal, biomass energy consumed at house years); special requirements for
boiler, solar panel, micro wind (16, 30, 35, 38 Euros) operation of equipment (weather,
fuel, noise, none); and possibility
for sharing (very low, low,
moderate, high)
Susaeta et al. Explore consumers’ 3 states in USA None Increase in price of Two versions for two attributes:
(2011) preferences for woody electricity per kWh in reduction in CO2 emission: 1–5%,
biomass based electricity monthly bill (0.01, 0.02, 6–10%, 11–20%; and 1–60%,
0.03, 0.04 USD) 61–70%, 71–90%; and %
improvement of forest habitat by
reducing wildfire risk and
improving forest health: 1–20%,
21–50%, 51–65%; and 1–25%,
26–50%, 51–75%
Vecchiato Explore consumers’ Veneto, Italy RES: solar photovoltaic, biomass Increase in monthly Size of the power plant (big, small);
and preferences for electricity from agriculture, biomass from forest electricity bill (0, 2, 5, 10 distance of power plant from
Tempesta contracts including RES Euros) interviewees’ living area (1, 3, 10
(2015) kms); and certified origin of the
biomass (Veneto Region, other
areas).
Yoo and Explore heterogeneity in Pennsylvania, USA % of electricity generated from RES: Additional cost to Impact on jobs in Pennsylvania (-
Ready consumers’ preferences for wind (2.8%–4.6%); solar (0.5%– household through higher 3,000, 0, + 3000)
(2014) RES 1.4%); biomass (1.5%–2.8%); and monthly electricity bills
other RES (3.2%–4.7%) and/or taxes (0–25 USD)

captures alternative sources (e.g. Cicia et al., 2012; Kosenius and Olli­ to efficient co-generation and nuclear power and the remaining 72% to
kainen, 2013; Vecchiato and Tempesta, 2015). For instance, Cicia et al. RES (SENER, 2019; 2018d). RES power increased by 1.0% over the same
(2012) presents an attribute with four levels, one for each RES under period, while its contribution to the electricity supply decreased from
consideration —wind, solar, biomass, and nuclear. A third set of studies 18% to 16%. These two patterns are explained by the reduction in hy­
disentangle preferences for specific RES and the share of electricity by dropower generation. At the same time, the installed capacity of other
RES (e.g. Borchers et al., 2007; Gracia et al., 2012; Murakami et al., sources (i.e. wind, geothermic, biomass, solar, and biogas) has
2015; Yoo and Ready, 2014). For instance, Borchers et al. (2007) in­ dramatically increased. For instance, solar and biogas installed capacity
cludes an attribute that captures five RES (wind, solar, biomass, farm went up from 461 MW to 89 MW in 2017 to 1647 MW and 217 MW in
methane, and generic green energy source) and an attribute that varies 2018, respectively (SENER, 2018e). Their potential is still quite large
the share (10%, and 25%) of respondent’s monthly electrical usage that —solar and bioenergy estimated installed capacity will be 11,661 MW
is generated with any RES. and 1478 MW by 2032, respectively SENER (2018c).7
When it comes to the way price has been incorporated in previous However, despite formally pursuing the goals set in 2015, the current
DCE studies, Table 1 illustrates that most studies phrase it as absolute federal administration (2019–2024) has slowed down Mexico’s energy
value increases in electricity bill —either on a yearly basis (e.g. Berg­ transition, and instead is encouraging fossil fuels over RES. This slow­
mann et al., 2006; Longo et al., 2008; Soliño et al., 2012), or bimonthly down has resulted from four actions. First, the cancellation of what
(e.g. Cicia et al., 2012), or monthly (e.g. Botelho et al., 2018; Borchers would have been Mexico’s fourth clean energy auction. Second, the
et al., 2007; Kim et al., 2018; Murakami et al., 2015; Vecchiato and significant increase in transmission tariffs to independent plants —most
Tempesta, 2015). A few studies have expressed the price attribute as of which generate energy with RES. Third, the indefinite suspension of
price per kWh in absolute value ]Gracia2012,Susaeta 2011. critical trials for new projects involving RES —which initially were
An additional insight from Table 1 is the wider scope of attributes planned to be launched by end of 2020. Fourth, a new energy program to
covered by previous DCE studies exploring preferences for RES. In rule the energy sector for the period 2020–2024 which has been inter­
particular, a number of studies have documented preferences for re­ preted as a discouragement of RES —in this respect, the current federal
ductions in emissions of greenhouse gases (e.g. Alberini et al., 2018; administration argues that RES represent a problem for the electricity
Kosenius and Ollikainen, 2013; Longo et al., 2008; Murakami et al., system due to its intermittence; that this problem worsens during this
2015; Soliño et al., 2012; Susaeta et al., 2011). Another topic that this economic recession; and that RES projects only benefits the private
literature has covered is the trade-off between renewable energy sector (SENER, 2020).
investments/infrastructure and impacts on nature components such as In this context, there are pending actions to set the institutional
landscape and wildlife (e.g. Bergmann et al., 2006; Botelho et al., 2018; conditions untapping Mexico’s energy transition. For instance, high
Kosenius and Ollikainen, 2013; Soliño et al., 2012; Susaeta et al., 2011). emission power plants, such as coal-fired power plants, must shut down
A final insight from Table 1 that we wish to highlight is the absence —and it is expected that federal and state authorities make sure that new
of studies that explore preferences of urban residents for both RES and green jobs are able to compensate job losses from closed plants. Also, it is
generation of jobs in emerging economies. necessary to set a competitive market across the supply chain in the
electricity market. The first step was to let private companies enter the
3. Tension in Mexico’s energy transition power market and compete with the state-owned utility company (CFE).
The following transition steps should allow private companies to
In 2015, Mexico committed to an energy transition by setting the compete in the rest of the supply chain, including the retail market for
ambitious goal of producing 30% of its electricity from RES by 2021 the qualifier users (annual demand ≥ 20 GWh). A change in tariff
–-and 35% by 2024 (Mexican, 2015 “Energy Transition Law”). Conse­
quently, from 2014 to 2018, Mexico increased production of clean
electricity at a rate of 4.8% annual average. During the same period, it 7
It is worth noticing that solar insolation in Mexico is larger than 5 kWh/m2/
reached a 23.1% share into the electricity mix of which 28% corresponds day, one of the largest worldwide (Hancevic et al., 2017).

4
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

scheme is needed to allow private retailers enter the residential market highest utility is expressed as the monetary value of the utility derived
and offer a competitive portfolio to consumers with a large variety of from Umax
i
characteristics —for instance, a tariff scheme that would include shares
Uimax
of RES as part of the electricity that residential consumers use. WTPi = (3)
βp
National and international stakeholders are pursuing measures to
counterbalance the federal administration’s actions that have slowed
where WTPi is individual i’s WTP; and βp is the price preference
down Mexico’s energy transition. For instance, a few state administra­
parameter. Under the assumption that indirect utility is linear in attri­
tions are i) considering carbon taxes to power inputs; ii) providing legal
butes, including income, βp is the negative of the marginal utility from
support to RES companies; and iii) exploring the decentralization of the
electricity grid for some private companies (see El Universal, 2020a; income. Under the assumptions embedded in equation (1), a researcher
Romo, Patricia, 2020). In addition, other states and municipalities may cannot observe Uimax as defined in equation (2). A researcher can only
take further steps to support RES and creation of green jobs in their make statements in terms of expected utilities which are calculated over
regions. Actions within their reach include but are not limited to prop­ the error term εij
erty tax exemptions to renewable plants and/or their tier companies, ( )
E Uimax = Eε [max{Vi1 , Vi2 , ……., ViJ }] (4)
training human capital on energy issues at state universities, and waste
recovering and recycling policies. Conversely, states and municipalities Under the assumption that εij follow a type I extreme value distri­
in Mexico cannot take direct measures to counterbalance decisions of bution, the expected maximum utility can be calculated through the log
the state-owned utility company (CFE). sum formula8
In particular, the administration of the State of Aguascalientes has
( ) ∑
J
( )
committed to encourage RES further. Actions in this direction include i) E Uimax = ln exp Vij (5)
to encourage human resources training at state universities; ii) to j=1

mandate that street lighting relies on RES; iii) to subsidize small busi­
Accordingly, statements involving welfare measures are made in
nesses to install clean energy technologies for up to MXN 150,000; and
expected terms. For a before (b) and an after (a) situations —where after
iv) to launch a new waste processing facility for power generation and
implies a change in the available alternatives—, the expected value of
other uses (see Becerril, 2018; El Heraldo, 2020; Gobierno del Estado de
the compensation variation (CV) due to the change in individual i’s
Aguascalientes, 2020; Lider Emprersarial, 2018).
utility is expressed as
The stakes are high for Aguascalientes when it comes to pursuing an
energy transition. Located in a semi-arid region, Aguascalientes has a 1 ( ( ))
Eε (CVi ) = Eε (Uimax,a ) − Eε Uimax,b
high potential and an installed capacity for solar and a medium-high − βp
potential for biomass (IRENA, 2018; SENER, 2018a, d). Aguascalientes
(
( ) ( )
) (6)
1 ∑J ∑J

is home of two 250 MW solar parks and two 4 MW bioenergy plants


a b
= ln exp Vij − ln exp Vij
− βp
already connected to the grid and will host about additional 570 MW
j=1 j=1

from solar photovoltaic (SENER, 2019). These projects are already The marginal willingness to pay (MWTP) can be derived from
approved but in stand-by due to the current federal policy. The State of equation (4.1) as follows. Assume attribute q changes in a non-marginal
Aguascalientes is estimated to have an annual power generation po­ fashion across all alternatives -i.e. qa = qb + Δq is the level of q after Δq
tential of 1350 TJ from urban solid waste and wastewater, which is a has been added to qb . Introduce the change in q in equation (4.1) and,
high potential for a small region (SENER, 2018a). because such a change occurs across all alternatives, factor it.9 The ex­
pected CV can be expressed as follows
4. Theoretical and empirical approaches
βq
Eε (CVi [Δq]) = − Δq (7)
4.1. Random utility model βp

The Random Utility Model (RUM) provides theoretical support to the where βq is the marginal utility from q. Equation (7) reduces to the WTP
empirical analysis of discrete choice experiments (see Train, 2009). The for a marginal change across alternatives when Δq = 1 —i.e. when the
departure point of the RUM is that, when faced to J mutually exclusive change in q is marginal, and
alternatives, individual i chooses the alternative that provides him/her βq
with the highest utility. An individual’s indirect utility from each Eε (MWTPi ) = − (8)
βp
alternative is denoted as Uij for i = 1, 2, …, I and j = 1, 2, …, J. The in­
dividual is assumed to know his/her own utility function with certainty. Equation (8) can be interpreted as the ratio of the marginal utility
The researcher, however, cannot fully observe each Uij . Thus, from the from the attribute that changes and the negative of the marginal utility
researcher’s point of view and once a linear indirect utility function is from income.
assumed, Uij can be expressed as

(1) 4.2. Econometric model



Uij = Vij + εij = β xij + εij

where Vij is the component observed by the researcher; xij is a (M +1)X1 Empirical estimations of the parameters required in the calculation
column vector denoting M alternative-specific attributes and the
of the expected MWTP (i.e. ̂β q and ̂
β p ) can be obtained via a conditional
alternative-specific intercept; β is a (M +1)X1 column vector repre­
logit econometric specification. The departure point of this empirical
senting the alternative-specific intercept, and the preferences for the
model is the same as to establish the theoretical expectations of the
alternative-specific attributes; and εij represents the purely random
welfare measures under discrete choice modelling —i.e. εij is distributed
heterogeneity that the researcher is unable to observe.
according to a type I extreme value distribution. Under this assumption,
If an individual chooses the alternative associated with the highest
utility, then the individual i chooses Umax
i , where

Uimax = max{Ui1 , Ui2 , ……., UiJ } (2) 8


Pioneer derivations of the log sum formula were independently developed
by Ben-Akiva (1973) and McFadden (1973).
The willingness to pay (WTP) for the alternative associated with the 9
Further details can be found in Haab and McConnell (2002).

5
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

the probability that individual i chooses alternative j is expressed as By including both attributes, our DCE is designed to empirically test
follows which one of these three cases hold in our sample. Borchers et al. (2007),
[ ] Gracia et al. (2012), and Yang et al. (2016) are instances of previous
Pij = Pr Vij + εij > Vik + εik ∀k ∕
=j
[ ] studies that have also included both RES and share of RES as attributes in
= Pr εij > Vik − Vij + εik ∀k ∕ (9)
=j

their DCE.
=∑
eVij
=∑
eβ xij The first attribute in Table 2 refers to RES. WTP of residential users
has been documented to vary depending on the source of the renewable
Vik ′
k∈J e k∈J e
β xik

energy. In particular, stated WTP is higher for solar and wind energies
A conditional logit (CL) specification faces two limitations to model
and lower for biomass and hydropower. These empirical results have
empirical discrete choice data (Train, 2009). First, a CL can represent
been reported both by documents consolidating the relevant literature
systematic variation (i.e. taste variation that is related to observed
(Ma et al., 2015; Sundt and Rehdanz, 2015), and by individual studies
characteristics) but not random taste variation (i.e. differences in tastes
focusing, for instance, on Spanish (Gracia et al., 2012), Danish (Yang
that cannot be linked to observed characteristics). Second, the estima­
et al., 2016), American (Borchers et al., 2007), and Italian consumers
tion of the CL probabilities implies proportional substitution across al­
(Vecchiato and Tempesta, 2015; Cicia et al., 2012).
ternatives —more flexible, more realistic patterns cannot be fitted with a
In our DCE, the renewable energy originates from either solar source,
CL model.10
or biomass, or a 50/50 mix. This way of controlling for energy source
The random parameters logit (RPL) results from adapting the CL
follows strategies pursued by, for instance, Cicia et al. (2012), Kosenius
model to incorporate non-systematic heterogeneity in preferences and
and Ollikainen (2013), and Vecchiato and Tempesta (2015). Studies
discard the proportional substitution across alternatives. The RPL turns
including similar sources than ours have documented a larger WTP for
out to be a highly flexible model that can approximate any random
solar over wind and biomass (e.g. Borchers et al., 2007; Gracia et al.,
utility model (McFadden and Train, 2000).
2012; Cicia et al., 2012; Vecchiato and Tempesta, 2015), and when
The RPL probabilities are the integrals of standard logit probabilities
hydropower is included, hydropower is preferred to the rest (Botelho
over a density of parameters. That is, keeping in mind equation (4.2), a
et al., 2018).
RPL is a model whose choice probabilities can be expressed in the
The second attribute in our DCE refers to the share of RES in current
following form
electricity mix − 10%, 20% or 30%. These values are relevant under the
∫ ′
eβ xij light of the Mexican Energy Transition Law (ETL). Specifically, the ETL
Pij = ∑ ′ f (β)dβ (10) mandates that 10.9% of electricity consumption in Mexico comes from
β xik
k∈J e
clean energies for large consumers. In addition, 30% of total electricity
generation in Mexico must come from clean energies by 2021. Indeed,
where f(β) is a density function. The RPL probability is a weighted
the current administration claims that Mexico has already an installed
average of the logit probabilities evaluated at different values of β, with
capacity to generate up to 31% from RES (Forbes, 2020).
the weights given by the density function f(β). In statistical terms, the
The third attribute in our DCE refers to the number of new jobs
weighted average of several functions is called a mixed function.
created in the RES sector —which we call green jobs. This attribute takes
Consequently, a RPL is a mixture of the logit function evaluated at
values 100, 1,000, or 2000. The jobs would be created in the state of
different β′ s with f(β) as the mixing function.
Aguascalientes, including the capital city. Strictly speaking, the creation
of jobs is not an attribute of the electricity service. If respondents’ utility
5. Survey methods and data
positively depends on the number of green jobs, then pursuing the
adoption of RES may yield a double dividend —generation of less
5.1. Discrete choice experiment
greenhouse emissions per kWh and the creation of jobs that are valued
by urban residents. A positive WTP for new jobs has previously been
Table 2 lists the four attributes of our discrete choice experiment
documented by DCE studies (Kosenius and Ollikainen, 2013; Bergmann
(DCE) and their corresponding levels. These attributes are i) renewable
et al., 2006; Soliño et al., 2012).
energy source (RES), ii) % of RES in current electricity mix, iii) new jobs
The fourth attribute in our DCE is the price attribute. In contrast to
in RE sector, and iv) % increase in self-reported bimonthly electricity
studies reviewed in Table 1, which describe the price attribute as
bill. The inclusion of RES as well as share of RES is key for this study’s
changes in absolute values, our DCE presents it as a percentage increase
purpose. Respondents may be willing to pay a premium for renewable
in respondents’ self-reported bimonthly electricity bill. This pivot design
electricity because they value it regardless of its source. It is also possible
has been pursued by, for instance, Amador et al. (2013) and it has the
that this WTP arises from respondents’ preferences for a specific RES. A
advantage of generating more variation in the price attribute —and this
third alternative is that consumers have preferences for a larger share of
variation remains within a reasonable range of absolute values.
renewable electricity while they attach a premium to a specific source.
Our price attribute takes values 5%, 20%, and 40%. These range of
values is within the ranges presented in previous DCE studies (e.g. Longo
Table 2 et al., 2008; Kosenius and Ollikainen, 2013). Also, an increase of 40% is
Attributes and levels in discrete choice experiment. on average 61 MXP per month, which represents 1.35% of the average
Attributes Levels household head’s income in Aguascalientes (INEGI, 2018a). When
incorporated in the empirical analysis, we calculate the corresponding
Renewable energy source (RES) Solar, biomass, and mix (50/
50)
hypothetical bimonthly electricity bill by applying the percentage
% of RES in current electricity mix 10%, 20%, and 30% change to the self-reported bill.
New green jobs (new jobs in RE sector) 100, 1,000, and 2000 DCE implicitly frames the decision as opting in by including a status
% increase in self-reported bimonthly electricity 5%, 20%, and 40% quo in the design. Hence, the hypothetical increase in bill can be
bill
interpreted as an opt-in extra fee (holding tariffs unchanged) that would
compensate for an increase in the share of electricity generated with RES
—which we deem more realistic in a context where tariffs are heavily

10
A third limitation is that a CL is not fitted to capture correlation over time
(Train, 2009).

6
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

and horizontally subsidized in Mexico.11 (22% versus 21%). Our sample includes smaller shares of respondents i)
The scenarios of our DCE were designed according to an orthogonal between 40 and 50 years old (16% versus 23%); ii) between 50 and 60
main effects strategy (see Aizaki, 2012). The DCE contains nine choice years old (14% versus 20%); and iii) older than 60 years old (6% versus
sets, all of which were presented to respondents. These nine scenarios 24%).
yield a D-efficiency of 95.2%. Each choice set includes three alternatives A bigger proportion of our respondents report higher incomes at the
described in terms of four attributes, and a status quo alternative.12 The household level in comparison to official statistics at the household head
respondents were asked to choose one alternative in each choice set. level. While the official proportion for income below 8000 Mexican
Fig. 1 illustrates a choice set.13 pesos (MXP) is 83%, this proportion is 39% in our sample; for income
between MXP 8000 and MXP 15,000, the official statistic is 10% and our
5.2. Data collection sample’s is 33%; and for income above MXP 15,000, the official number
is 7% and our sample’s is 28%. The proportion of respondents with a
Face-to-face implementation of our DCE was conducted through full-time job is closer to the official statistic for household heads in
September to November 2019. Respondents were approached randomly Aguascalientes − 48% versus 41%. Around 72% of our respondents
in public spaces such as shopping malls and the main square in down­ report affording the full amount of the electricity bill —which implies
town of Aguascalientes city. We made sure that respondents were adults that a share of respondents affording the electricity bill hold part-time
residing in Aguascalientes city that contribute to paying the electricity jobs.
bill.14 Two averages at the household level in our sample are almost
The survey instrument was composed by three sections. In the first identical to official numbers. In our sample, around 70% of respondents
one, respondents were explained the mean of RES and what each of them live in a house that is owned by a household member and the average
imply in terms of generation of electricity. In the second section, re­ number of household members is 3.93. These numbers in the official
spondents were presented to the DCE. In the third section, we gather statistics are 69% and 3.91, respectively.
socioeconomic characteristics of respondents and their households. In comparison to the official statistic, a smaller proportion of re­
spondents pay electricity on a bimonthly basis − 61% in our sample
5.3. Descriptive statistics versus 89%. The average self-reported electricity bill is MXP 455 —on a
bimonthly basis. The average hypothetical bimonthly bill implied by the
Once missing values have been dropped, our sample is composed of DCE scenarios is MXP 659. The average self-reported bill is bigger than
199 respondents. Table 3 reports mean, standard deviation, minimum the official statistic —MXP 455 versus MXP 306—, and a two-tailed t-
and maximum of variables describing respondents’ and his/her house­ test rejects the null hypothesis that the difference between these
hold’s characteristics. To put our sample’s characteristics in context, numbers is zero with 99% of confidence —with a p-value of 0.0019 that
Table 3 reports in its last column the official statistics for household corresponds to a t-test statistic of 3.15. We highlight that our sample is
heads in Aguascalientes (see INEGI, 2018a). In comparison to official similar to official statistics when it comes to household’s characteristics
statistics of household heads in Aguascalientes, our sample is composed such as the number of household members and whether the house is
by a higher share of females (42% versus 26%), and a smaller proportion owned by a household member.
of married people (44% versus 58%). At the individual level, our sample is composed by a smaller share of
In terms of range of age, respondents of our DCE are younger than married people; and bigger shares of females, and younger respondents
household heads in Aguascalientes. The proportion of respondents that who belong to higher incomes households. As our respondents of in­
are younger than 30 years old is higher in our sample (43% versus 12%). terest are adults that contribute (at least partially) to paying the elec­
The proportion of respondents between 30 and 40 years old in our tricity bill and not necessarily household heads, the differences in
sample is almost identical to the official proportion of household heads individuals’ characteristics are not unexpected. The higher self-reported
electricity bill and household’s income imply that our sample is
composed by households richer than the average one in Aguascalientes.
11
Historical evidence shows that residential electric subsidy in Mexico is In section 9, we discuss implications of these differences in terms of
difficult to be removed due to its high political cost. On the other hand, Han­ external validity of our results and conclusions.
cevic and Lopez-Aguilar (2019) find that electricity consumption in the country
is quite inelastic to price (− 0.21) and to income (0.19). So, in the remote 6. Results
possibility that electric subsidy were removed, the payment of this extra fee
would be affected, but not significantly. 6.1. Econometric specifications
12
Concerns have been raised about incentive compatibility of discrete choice
experiments. There is evidence suggesting that lack of incentive compatibility
Table 4 reports estimates from six random parameters logit (RPL)
in DCE yields overestimation of stated WTP for public goods. For private goods,
specifications.15 These specifications assume uncorrelated normal dis­
as the one under analysis in this paper, evidence suggest that estimates in non-
incentive compatible contexts are within the 95% confidence interval of esti­ tributions for all parameters with exception of the price parameter
mates in which incentive compatible mechanisms are in place. The reader is which is assumed fixed. The first specification (I) is estimated on the
referred to Johnston et al. (2017), a contemporary guidance for stated prefer­ entire working sample (199 respondents). The second specification (II)
ence studies, and to Taylor et al. (2010), who report evidence supporting the is estimated on a sample that includes respondents in households with
claims in this footnote. eight members or less (195 respondents) to explore whether results are
13
The choice set in Fig. 1 illustrates four types of feedstocks for biomass driven by preferences of respondents in relatively large households
including agricultural and forestry residues, and municipal solid waste and —with an average of 3.93 household members, our working sample
wastewater. includes respondents in households with up to 15 members (see
14
We have not incentivized participation of our respondents and this may
Table 3). To explore price parameter sensitivity, the third specification
generate potential sample selection. However, potential respondents were not
(III) excludes 33 choice sets for which hypothetical bills fall below or
invited to participate on the premise that the survey was about RES or new jobs
in the RES sector. In this sense, the self-selection is not expected to be associated
above the 1% tails —the resulting sample contains 195 respondents that
with interest in those issues. Instead, respondents were invited to participate on are not the same as in specification (II). The fourth specification is
the premise that the survey was about electricity consumption and preferences.
Thus, our respondents may be more likely to be interested in energy matters
15
than non-respondents. We thank an anonymous reviewer for requesting this Random parameters logit specifications reported in this work are estimated
clarification. in Stata with the commands mixlogit and mixlogitwtp for the WTP-space case.

7
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

Fig. 1. Example of a choice set.

estimated on answers from 191 respondents that pass exclusion criteria We highlight four features of Table 4. First, price parameter is
(II) and (III). The fifth specification (IV) excludes 193 choice sets for negative and significant at 99% across specifications, and its absolute
which hypothetical bills fall below or above the 5% tails, leaving a value is higher when implementing exclusion criteria. Second, point
sample with 177 respondents. The sixth specification is estimated on a estimates of all but one parameter are similar and statistically significant
sample of 174 respondents which results from implementing exclusion across specifications —the exception is the parameter associated with
criteria (II) and (IV). 20% of RES, which is insignificant across specifications. Third, estimates

8
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

Table 3 respondents have no extra willingness to pay for a share of 20% in


Sample’s summary statistics (n = 199), and official statistics for household heads comparison to a share of 10%. Taking 100 new green jobs as reference
in Aguascalientes. category, the positive signs of parameters associated with 1000 and
Variable Mean Std. Min. Max. Aguasca- 2000 new green jobs, respectively, imply that respondents prefer the
Dev. lientesa creation of 1000 and 2000 green jobs over the creation of 100 jobs.
Respondent’s characteristics
1 if female 0.417 0.494 0.000 1.000 0.260
6.2. Welfare estimates
1 if married 0.442 0.498 0.000 1.000 0.580
1 if younger than 30 years 0.427 0.496 0.000 1.000 0.120
old For each attribute level in our DCE, we have estimated the marginal
1 if 30–40 years old 0.216 0.413 0.000 1.000 0.210 willingness to pay (MWTP) —expressed as bimonthly, 2018 thousand
1 if 40–50 years old 0.156 0.364 0.000 1.000 0.230 Mexican pesos (MXP). Table 5 reports six sets of MWTP estimates and
1 if 50–60 years old 0.136 0.343 0.000 1.000 0.200
1 if older than 60 years old 0.065 0.248 0.000 1.000 0.240
their 95% confidence interval. Each set of MWTP estimates corresponds
0.477 0.501 0.000 1.000 0.410 to one specification in Table 4.
1 if full-time job Our attention is on the smallest WTP estimates —i.e. those reported
1 if affords full amount of 0.719 0.451 0.000 1.000 – in the last column of Table 5, which arise from the specification that
electricity bill
excludes both the 5% (upper and lower) tails of the hypothetical bill and
Household’s characteristics
1 if monthly income up to 8 0.392 0.489 0.000 1.000 0.830 the respondents in households with more than 8 members.17 We do so
K (MXP) because stated preferences methods may overestimate WTP measures.
1 if monthly income 0.332 0.472 0.000 1.000 0.100 By focusing on the most conservative WTP estimates, we aim to provide
between 8 K and 15 K implications that can more likely inform public policies.
(MXP)
Thus, according to estimates reported in the last column of Table 5,
1 if monthly income above 0.276 0.448 0.000 1.000 0.070
15 K (MXP) respondents’ loss in utility for remaining in the status quo situation is
0.704 0.458 0.000 1.000 0.690 valued at MXP 899 bimonthly. When it comes to solar energy, re­
1 if house is owned by a spondents’ WTP is MXP 34 on a bimonthly basis —zero is included in
household member
95% confidence interval, and the zero null hypothesis can only be
Number of household 3.930 1.890 1.000 15.000 3.910
members rejected at a 10% of significance. Respondents report a bimonthly loss of
1 if bill is paid on bimonthly 0.613 0.488 0.000 1.000 0.890 MXP 51 when it comes to biomass energy in comparison to a 50/50 mix
basis of biomass and solar energies. The WTP for a share of 20% of RES is not
0.455 0.667 0.004 7.500 0.306b statistically significant. In contrast, the WTP for a share of 30% is MXP
Self-reported electricity
74 on a bimonthly basis. The WTP for the creation of 1000 new green
bill (thousand MXP)
Hypothetical bimonthly jobs is MXP 40 on a bimonthly basis; and MXP 59 for the creation of
electricity bill (thousand 0.659 0.920 0.008 7.875 – 2000 new green jobs.
MXP)
a
Source: INEGI (2018a). 6.3. Robustness checks
b
A two-tailed t-test rejects at 99% the null hypothesis that the difference
between sample mean and population parameter is zero – p-value is 0.0019 for a We have explored whether our results are robust to treating both
t-test statistic of 3.15.
share of RES and the number of green jobs as continuous variables.
Results of this section are reported in the online appendix. The param­
of standard deviation of most parameters are similar across specifica­ eter estimates from six RPL specifications that treat those attributes as
tions —the exceptions are those of biomass and 2000 new green jobs. continuous, and their corresponding welfare estimates, are reported in
Fourth, standard deviations associated with status quo, solar, and 30% Tables A3 and A4, respectively.18 Focusing our attention on WTP esti­
of RES are statistically significant across specifications. We deem the mates reported in the last column of Table A4, we highlight that most
first and second features as evidence that all attributes included in our point estimates are similar in magnitude to those in Table 5. For
DCE are relevant to respondents; and the third and fourth features, as instance, the negative WTP for biomass is MXP 54 (versus MXP 51 in
evidence that unobserved heterogeneity is at place —and, consequently, Table 5); and MWTP for solar is MXP 30 (versus MXP 34 in Table 5).
RPL specifications are preferred over CL ones.16 MWTP for an increase in 1000 new green jobs is MXP 30, or for 2000
When it comes to interpreting the sign of specific parameters, the jobs is MXP 60, which are close, respectively to the MXP 40 and MXP 59
negative sign of the status quo parameter implies that respondents reported in Table 5. For the case of status quo, the negative WTP is
dislike their current situation. Taking solar-biomass mix as reference estimated at MXP 751 (versus MXP 899 in Table 5). MWTP for a 1%
category, the positive sign of the solar energy parameter implies that increase in the share of RES is MXP 5 which implies a WTP of MXP 150
solar is preferred over the mix; and the negative sign of the biomass for a share of 30% —around twice as much as the MXP 74 reported in
parameter implies that the mix is preferred over biomass. This ordering Table 5.
of preferences is consistent with results from previous studies that have Once we have checked that results are robust to inclusion of attri­
also documented a disutility from biomass ( e.g. Cicia et al., 2012; Gracia butes as continuous variables, we have decided to base our main WTP
et al., 2012; Yoo and Ready, 2014). As Gracia et al. (2012) have pointed estimates on specifications that include share of RES and the number of
out, this disutility may be explained by a lack of knowledge on how to
use biomass sustainably, and/or may be associated with negative ex­
ternalities from its use in the past. Taking 10% of electricity from RES as 17
The smaller estimates are direct consequence of the higher absolute value of
reference category, the positive sign of the parameter associated with the price parameter yielded by the corresponding specification. From equation
30% share implies that a share of 30% is preferred over a share of 10%. (8) and given the similarity across specifications of all but the price parameter,
Similarly, the insignificant parameter associated with 20% implies that a higher absolute value of the price parameter results in a smaller MWTP.
18
For completeness, we have also estimated the corresponding conditional
logit specifications. These specifications and the marginal willingness to pay
estimates are reported in Tables A5 and A6, respectively. Random parameters
16
For completeness, conditional logit specifications and welfare measures logit specifications reported in Table A3 are estimated in Stata with the com­
arising from them are reported in tables A1 and A2 at the Online Appendix. mand mixlogit.

9
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

Table 4
Random Parameters Logit specifications on stated choices —assuming price parameter is fixed, and all non-price parameters are normally distributed and uncorrelated;
and including % of RES and new green jobs as sets of binary variables identifying attributes’ levels.
Entire sample (l) Excluding > 8 h h members (II) Excluding 1% tailsd (III) (II)+(III) Excluding 5% tailsd (IV) (II)+(IV)

Mean
1 if status quo option − 4.891*** − 5.101*** − 4.937*** − 5.401*** − 4.969*** − 5.028***
(0.781) (0.678) (0.654) (0.724) (0.686) (0.687)
a
1 if solar 0.188** 0.204** 0.217** 0.210** 0.206** 0.192**
(0.0831) (0.0856) (0.0851) (0.0879) (0.0948) (0.0955)
1 if biomassa − 0.235** − 0.243*** − 0.236** − 0.251*** − 0.269*** − 0.284***
(0.0727) (0.0714) (0.0732) (0.0761) (0.0782) (0.0818)
1 if 20% of electricity − 0.0288 − 0.0353 − 0.0368 − 0.0490 − 0.0615 − 0.0643
comes from RESb (0.0700) (0.0713) (0.0714) (0.0738) (0.0757) (0.0768)
1 if 30% of electricity 0.407*** 0.423*** 0.413*** 0.417*** 0.425*** 0.413***
comes from RESb (0.0877) (0.0903) (0.0933) (0.0943) (0.0982) (0.0997)
if 1000 new green jobsc 0.196** 0.211** 0.209** 0.226** 0.220** 0.224**
(0.0705) (0.0712) (0.0718) (0.0734) (0.0770) (0.0775)
1 if 2000 new green jobsc 0.308*** 0.313*** 0.328*** 0.335*** 0.353*** 0.329***
(0.0695) (0.0701) (0.0712) (0.0736) (0.0763) (0.0804)
Hypothetical bimonthly electricity − 2.465*** − 2.652*** − 3.420*** − 3.720*** − 5.545*** − 5.592***
bill (thousand MXP pesos) (0.273) (0.287) (0.324) (0.343) (0.456) (0.459)
SD
1 if status quo option 4.528*** 4.093*** 4.087*** 4.700*** 4.827*** 5.162***
(0.632) (0.536) (0.489) (0.623) (0.732) (0.614)
1 if solara 0.715*** 0.743*** 0.734*** 0.779*** 0.843*** 0.847***
(0.0956) (0.0948) (0.0896) (0.0969) (0.111) (0.106)
1 if biomassa 0.237 0.116 0.150 0.288** 0.214 0.333**
(0.196) (0.220) (0.248) (0.137) (0.230) (0.144)
1 if 20% of electricity 0.0418 0.0995 0.0790 0.154 0.0899 0.0645
comes from RESb (0.139) (0.140) (0.130) (0.128) (0.176) (0.158)
1 if 30% of electricity 0.829*** 0.867*** 0.906*** 0.940*** 0.947*** 0.951***
comes from RESb (0.0982) (0.101) (0.103) (0.104) (0.107) (0.110)
1 if 1000 new green jobsc 0.0577 0.0127 0.0420 0.0620 0.0899 0.0150
(0.0991) (0.107) (0.107) (0.118) (0.114) (0.112)
1 if 2000 new green jobsc 0.195 0.163 0.228* 0.275** 0.256* 0.371***
(0.130) (0.147) (0.127) (0.130) (0.144) (0.104)
Respondents 199 195 195 191 177 174
Observations 7164 7020 7032 6888 6390 6287
ll − 1928.4 − 1873.1 − 1865.4 − 1811.6 − 1679.1 − 1644.1
AIC 3886.8 3776.2 3760.8 3653.3 3388.2 3318.1
BIC 3989.9 3879.1 3863.7 3755.8 3489.7 3419.3

Standard errors in parentheses.* p < 0.10, ** p < 0.05, *** p < 0.001.
a
Reference category: 50/50 combination of solar and biomass.
b
Reference category: 10% of electricity comes from RES.
c
Reference category: 100 new green jobs.
d
It refers to both lower and upper tails of the distribution of hypothetical bimonthly electricity bill.

Table 5
Marginal willingness to pay (bimonthly, 2018 thousand Mexican pesos) and 95% confidence intervals resulting from Random Parameters Logit specifications reported
in Table 4.
MWTP for Entire sample (I) Excluding > 8 h h members (II) Excluding 1% tails (III) (II)+(III) Excluding 5% tails (IV) (II)+(IV)

Status quo − 1.984 − 1.923 − 1.443 − 1.452 − 0.896 − 0.899


Lower Bound − 2.904 − 2.687 − 1.979 − 1.973 − 1.212 − 1.208
Upper Bound − 1.271 − 1.342 − 1.007 − 1.016 − 0.625 − 0.633
Solar 0.076 0.077 0.063 0.056 0.037 0.034a
Lower Bound 0.007 0.011 0.013 0.009 0.003 0.000
Upper Bound 0.150 0.147 0.117 0.106 0.072 0.069
Biomass − 0.095 − 0.091 − 0.069 − 0.068 − 0.049 − 0.051
Lower Bound − 0.157 − 0.148 − 0.112 − 0.109 − 0.077 − 0.080
Upper Bound − 0.042 − 0.043 − 0.030 − 0.030 − 0.023 − 0.024
20% of RE − 0.012 − 0.013 − 0.011 − 0.013 − 0.011 − 0.011
Lower Bound − 0.068 − 0.066 − 0.052 − 0.052 − 0.038 − 0.038
Upper Bound 0.048 0.043 0.033 0.028 0.017 0.017
30% of RE 0.165 0.159 0.121 0.112 0.077 0.074
Lower Bound 0.097 0.095 0.069 0.065 0.044 0.040
Upper Bound 0.254 0.243 0.182 0.169 0.116 0.114
1000 new green jobs 0.080 0.080 0.061 0.061 0.040 0.040
Lower Bound 0.024 0.028 0.021 0.023 0.013 0.013
Upper Bound 0.144 0.140 0.107 0.104 0.069 0.069
2000 new green jobs 0.125 0.118 0.096 0.090 0.064 0.059
Lower Bound 0.068 0.065 0.055 0.051 0.037 0.030
Upper Bound 0.199 0.186 0.145 0.137 0.096 0.092
a
Zero null hypothesis can be rejected with p < 0.10.

10
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

green jobs as a series of binary variables —i.e. those reported in Table 5. —to the point that respondents older than 60 report non-statistically
In deciding so, we have followed the premise that our preferred speci­ significant positive WTP.21 Similarly, WTP for 30% share of RES in­
fication is the one delivering the smallest WTP estimates —when creases from thousand MXP 0.032 among respondents younger than 30
assuming continuous attributes, WTP for 30% RES is estimated to be to thousand MXP 0.048 among respondents between 50 and 60 years old
twice the value estimated when including attributes as binary variables —with respondents older than 60 reporting a WTP that is not different
and thus we prefer to focus on results from Table 5. from zero. Fig. 2 also illustrates that WTP for solar energy is the largest
We have also checked that our estimates are robust to assumptions among respondents that belong to the 30–40 years old category (thou­
about the distribution and correlation of the price parameter. Table A7 sand MXP 0.073), and then for those between 60 and 60 years old
reports results from six RPL specifications that assume a log-normal (thousand MXP 0.062) but no different from zero for those in other age
distribution for the price parameter and normal distributions for all categories.
non-price parameters, and treat all random parameters as uncorrelated. The second take-away is that variation in WTP for status quo, solar
Table A9 reports results from six RPL specifications estimated in WTP energy, and creation of jobs are associated with income. Fig. 3 illustrates
Space which computationally makes it easier to add correlation among that disutility from status quo monotonically increases with income
random parameters to the specifications in Table A7. Specifications in —from thousand MXP -0.407 among respondents with less than thou­
Table A9 treat parameters associated with biomass, 20% RES share, and sand MXP eight to thousand MXP -0.615 among those with more than
1000 new jobs as fixed parameters19; all other parameters are treated as thousand MXP 15. Similarly, point estimates of WTP for solar energy,
normally distributed with exception of the price parameter which is creation of 1000 jobs and 2000 jobs increase monotonically across in­
assumed log-normally distributed. Signs and statistical significance of come categories.
parameters yielded by these models suggest robustness of the simpler We have not focused our main attention on documenting observed
model we have focused on. However, WTP estimates arising from heterogeneity because Bayesian Information Criterion (BIC) strongly
models in Table A7, and reported in Table A8, are at least three times suggests that specifications in Tables A10 and A11 are statistically
larger than those of our simpler specification. For the case of WTP es­ outperformed by specification in Table 4 —while BIC of the specification
timates yielded by the WTP-Space specifications, magnitudes are around modeling only unobserved heterogeneity is 3,419, specifications in table
larger than those of our simpler specification but never twice as much. A10 are 4181 and 3615; 4150 and 3,569, for those in Table A11.22
Following the same reasoning as above, we keep our focus on specifi­
cations that yield the smallest WTP estimates. 7. Comparison of results with previous studies
We have explored implications from observed heterogeneity.
Tables A10 and A11 report RPL specifications estimated on the sample To put this paper’s welfare estimates in context with respect to
that excludes respondents in households with more than 8 household previous papers’ estimates, it is convenient to carry out comparisons in
members and choice sets with hypothetical bill above and below the 5% terms of WTP per kWh under specific scenarios. Thus, based on the after-
tails. Specification (I) in table A10 takes the specification in column six tax self-reported electricity bill, we first retrieve electricity consump­
of Table 4 —i.e. a RPL model assuming price parameter is fixed and all tion, measured in terms of kWh. In Aguascalientes, consumers face a
non-price parameters are normally distributed and uncorrelated— and three-tier block increasing tariff. At the time of the survey, first tier
interact each attribute with socioeconomic characteristics reported in charge was MXP 0.823 to consumption up to 150 kWh; the second tier
Table 3 —i.e. categorical variables capturing respondent’s age and in­ charged MXP 0.996 to consumption above 150 kWh and up to 280 kWh;
come, and whether he/she is female, married, with full-time job, affords and the third tier charged 2.912 for consumption over 280 kWh.23
full electricity bill, and he/she or other household member owns their Correspondingly, electricity consumption is inferred by inverting the
house; and an integer variable capturing the number of household tariff formula.
members. The second specification in Table A10 keeps statistically sig­ Once electricity consumption is inferred, estimated bimonthly WTP
nificant interactions from specification (I). Table A11 reports two RPL is added to the self-reported bimonthly bill, and the total amount is
specifications that follow the same logic than those in Table A11 but divided by the inferred consumption. This estimation of WTP per kWh
instead assumes that price parameter is log-normally distributed and all assumes that i) consumers know and keep in mind the three-tier block
non-price parameters are normally distributed and uncorrelated tariff when it comes to deciding electricity consumption; ii) consumers
—similar to specifications reported in table A7.20 pay attention to before-tax amounts and not to after-tax amounts; and
Specifications in Tables A10 and A11 suggest that results that we iii) had consumers paid a premium on top of their current bills, they
have focused on in this section (i.e. those reported in the sixth columns would not change consumption levels.
of Table 4) are robust to observed heterogeneity. Sign and statistical Table 6 reports WTP per kWh under nine scenarios that we deem of
significance of the average values of parameters remain similar; and the public policy interest. All nine scenarios assume a 30% share of RE, and
statistical significance of parameters reflecting unobserved heteroge­ the corresponding WTP is calculated based on MWTP numbers reported
neity (i.e. standard deviation of random parameters) also remains in last column of Table 5. The first scenario assumes that this share is
similar. achieved with biomass energy (I). Notice that estimation of WTP under
We report two take-aways from specifications modeling observed this scenario implies the subtraction of MXP 51 from MXP 74 which is
heterogeneity. First, variations in WTP for status quo, solar energy and the utility that 30% RES share produces —the subtraction is due to the
30% share of RES are associated with age. Fig. 2 illustrates that disutility disutility from biomass energy in comparison to a 50/50 biomass and
from status quo (expressed as negative WTP) is largest among re­
spondents younger than 30 and it decreases non-monotonically with age

21
Point estimates in Figs. 2 and 3 and their corresponding standard errors are
reported in Tables A12 and A13, respectively.
19 22
Specifications reported in Table 4, A3, A7 yield insignificant standard de­ Akaike Information Criterion (AIC) suggests that specifications that include
viations for these parameters which suggest that they can be assumed fixed to the full set of interactions are outperformed by the simpler specification (3318
increase convergence success of WTP-Space models. vs. 3378 and 3338), but the specifications retaining significant interactions
20
We have tried the estimation of WTP-space specifications as those in table outperform the simpler specification (3318 vs. 3291 and 3251).
23
A9 but have run into convergence issues that remain after several trails that If annual consumption exceeds 3000 kWh, the household is reclassified as a
assume different initial parameter values. high-consuming household (Demanda de alto consumo) and the corresponding
tariff is significantly higher than tier-3 tariff. We are not able to identify any of
our respondents in the high-consuming category.

11
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

Fig. 2. Point estimates of marginal willingness to pay (bimonthly, 2018 thousand Mexican pesos) for selected attributes by respondent’s age category.

Fig. 3. Point estimates of marginal willingness to pay (bimonthly, 2018 thousand Mexican pesos) for selected attributes by respondent’s income category.
Note: Calculations are based on results reported in table A10. Standard errors are reported in table A13.

solar mix. The second and third scenarios assume (I) and, respectively, (III). The eighth and ninth scenarios assume (III) and, respectively, the
the creation of 1000 and 2000 new green jobs. The fourth scenario as­ creation of 1000 and 2000 new green jobs. Table 6 first reports the
sumes that the 30% share of RES is reached with solar energy, but no bimonthly WTP under each of the nine describes scenarios —both in
premium is paid for this source of energy (II). This no premium 2018 MXP and USD, respectively.24 Then Table 6 reports the bimonthly
assumption aims to take into consideration that, according to estimates WTP as a proportion of the average bimonthly bill —both self-reported
in the last column of Table 5, MWTP for solar energy is statistically and hypothetical. In its last two columns, Table 6 reports bimonthly
significant only with p < 0.10. The fifth and sixth scenarios assume (II) WTP per kWh —in 2018 MXP and USD (cents), respectively.
and, respectively, the creation of 1000 and 2000 new green jobs. The A 30% share of renewable electricity generated through biomass is
seventh scenario assumes that the 30% share of RES is reached with valued at USD 1.20 on a bimonthly basis. When 1000 new green jobs are
solar energy and a premium of MXP 34 is paid for this source of energy created, consumers’ stated benefits are USD 3.28; and when 2000 new

24
Assuming an exchange rate of 19.22 MXP/USD which was the average
closing price in 2019 according to Macrotrends (2020).

12
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

Table 6
Bimonthly willingness to pay for RES share/RES/green jobs scenarios —calculated based on estimates reported in last column of Table 5.
As proportion of average bimonthly bill

Bimonthly WTP self-reported hypothetical per kWhb


a
Scenarios (2018 MXP) USD (MXP 455) (MXP 659) (2018 MXP) USD (cents)a

Assuming desutility of 51 MXP from biomass


30% share from biomass (I) 23 1.20 0.05 0.03 0.59 3.09
(I) + 1000 new green jobs 63 3.28 0.14 0.10 0.83 4.33
(I) + 2000 new green jobs 82 4.27 0.18 0.12 0.95 4.93
Assuming no premium for solar
30% share from solar (II) 74 3.85 0.16 0.11 0.90 4.68
(II) + 1000 new green jobs 114 5.93 0.25 0.17 1.14 5.93
(II) + 2000 new green jobs 133 6.92 0.29 0.20 1.25 6.52
Assuming a 34 MXP premium for solar
30% share from solar (III) 108 5.62 0.24 0.16 1.10 5.74
(III) + 1000 new green jobs 148 7.70 0.33 0.22 1.34 6.99
(III) + 2000 new green jobs 167 8.69 0.37 0.25 1.46 7.59

As described in section 7, based on after-tax self-reported bill, i) consumers are first assigned to their corresponding tariff; ii) consumed kWh are inferred; iii) for a given
scenario, bimonthly WTP (first column in Table 6) is added to the self-reported bill; and iv) assuming that kWh remain unchanged, MXP per kWh are calculated under
each scenario.
a
Assuming an exchange rate of 19.22 MXP/USD which was the average closing price in 2019 (see Macrotrends, 2020).
b
Sample mean of implied bimonthly WTP per kWh, assuming consumed kWh remain unchanged if a scenario is in place.

green jobs are added, stated benefits are estimated at USD 4.27. These instance, the volatility in oil prices make investments in state-owned oil
numbers represent, respectively 5%, 14%, and 18% of the average self- company even riskier (Davis, 2019; Stromsta, 2019; Martin, 2019). In
reported bill; or 3%, 10%, and 12% of the hypothetical bill. In terms of addition, these investments may deviate Mexico from achieving climate
USD per kWh, these numbers are equivalent, respectively, to 3.09 cents, change goals. These concerns have become stronger as discussions begin
4.33 cents, and 4.93 cents. on how to overcome the COVID-19 recession (Castañeda-Morales, 2020;
For the three scenarios involving solar energy under the assumption Levy, 2020).
that no premium is paid for this energy source, the bimonthly WTP is In this context, findings from this study suggest that a policy simul­
USD 3.85 for the 30% share; and USD 5.93 (USD 6.92) when 1000 taneously boosting renewable energies and the creation of green jobs
(2000) are added. In terms of cents per kWh, these numbers translate will likely have the support of urban residents in Aguascalientes that
into 4.68, 5.98, and 6.52, respectively. When a premium for solar energy contribute to paying their household electricity bill. We discuss the
is added, the bimonthly WTP becomes USD 5.62, USD 7.70, and USD external validity of this conclusion in section 9. As Aguascalientes’
8.69, respectively. In terms of WTP per kWh, these numbers are equiv­ administration has expressed its intentions to counterbalance federal
alent to 5.74, 6.99, and 7.59, respectively. administration’s strategy, potential support from urban residential
Our per kWh welfare estimates fall well within the range of values electricity consumers is an essential element when designing and
reported by previous studies. The studies summarizing the previous implementing state level policies.
literature provide useful information to carry out a comparison. Ma et al. Our findings suggest that a policy achieving a 30% share of renew­
(2015) find that a majority of WTP values falling between − 10 cents and able biomass energy and the creation of 1000 new green jobs is valued
+10 cents and Sundt and Rehdanz (2015) report an average of 3.18 by electricity consumers at USD 3.28 on a bimonthly basis − 2.1 of which
cents. Our lowest per kWh estimates (3.09) for 30% share produced with correspond to the creation of the jobs. To put this number in context, the
biomass is not only close to the average reported by Sundt and Rehdanz federal administration provides USD 195 on a monthly basis to people
(2015), but also along with our highest bimonthly per kWh estimate between 18 and 29 years old who enroll in training programs (STPS,
(7.59) for 30% share produced with solar, both fall within the 2020). If the trainee were going to learn skills that facilitate his/her
[ − 10, +10] interval reported by Ma et al. (2015). participation in the energy renewable sector, a household in our sample
In terms of monthly household WTP, our estimates on a monthly would contribute on average around 1% of the USD 195 that the federal
basis fall well below USD 13.13 (average) and USD 11.67 (median) re­ administration transfers to a trainee. Equivalently, the contribution
ported by Sundt and Rehdanz (2015). Our lowest monthly household from 100 households would cover the USD 195 on a bimonthly basis. In
WTP is equivalent to 0.60 cents, and our highest is USD 4.34. This a similar way, the levelized cost of 1 kWh of bioenergy is estimated at
relative magnitude in the household WTP should not be surprising as USD 5.7 cents for the State of Aguascalientes (SENER, 2018b). This
most of previous studies have focused on developed countries and means that a household in our sample would contribute on average
Mexico has only recently being classified as upper middle income. about 50%–75% to cover these costs (i.e. USD 3.09 cents per kWh for
30% share and USD 4.33 cents per kWh for 30% share and 1000 jobs in
8. Conclusions and public policy implications Table 5).
A policy encouraging re-training of individuals to transition to the
The national business sector has expressed concerns about the cur­ renewable energy industries would also provide a just transition for
rent federal administration’s energy strategy due to its negative impli­ workers currently employed in fossil-fuel based industries —a just
cations for the competitiveness of the energy sector and potential transition encompasses policies that support workers and families whose
spillovers to other economic sectors (Gonzalez, 2019; Nava, 2020). This jobs, incomes, and livelihoods are at risk as the world pursues sustain­
apprehension is shared by international stakeholders and discussions able pathways (Rosemberg, 2017). In addition, this policy would help to
have been hold among diplomats of eight developed countries to explore sustainably recover from the COVID-19 recession. As many people is
the best way to communicate such concerns to the federal administra­ expected to lose their jobs, a compensation during the re-training period
tion (Proceso, 2020). would help in avoiding that displaced workers transition to jobs in the
Scholars and analysts have also expressed that the planned renova­ informal sector. A transition to the informal sector is undesirable
tion of state-owned oil and electricity companies may be too costly —for because i) it would decrease wages in the informal sector —negatively

13
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

impacting not only newcomers but also those who permanently hold an is not enough to unambiguously claim external validity of our sample.
informal job –; and ii) it implies a loss in returns from investments made We have explored whether observed heterogeneity in preferences is
by individuals and the society as a whole on skills that are not useful in associated with socioeconomic characteristics in our sample. The evi­
the informal sector —a waste from which a recovery is difficult once dence is ambiguous: while WTP for creation of jobs seems to be asso­
workers transition to the informal sector because as they do not return to ciated with income and WTP for solar energy and share for renewable
the formal sector easily (Levy, 2020). energy seems to be associated with age, specifications that control for
When designing a policy that boosts both renewable energies and observed heterogeneity are statistically outperformed by specifications
creation of green jobs, it is important to keep in mind that the potential that do not control for socioeconomic variables.
number of new jobs differs depending on the source of energy. For The implication from the previous caveat it that our conclusions are
instance, IRENA (2015) estimates that while one Gigawatt (GW) of solar supported by average parameters arising from our sample and extrap­
photovoltaic power can create on average 8250 new jobs, one GW of olation to the universe of urban household heads in Aguascalientes is not
biomass power can create 31,000 new jobs. Another element to keep in warranted. While this does not discredit our results, it is necessary to
mind is that jobs created by biomass energy generation are mostly carry out further research to learn whether conclusions from this study
permanent jobs but those created by solar energy generation are in hold at a wider scale —ideally, we would want to obtain conclusions
general temporary ones. that hold at a national scale.25
Indeed, our estimates arise from a sample obtained in 2019 and may A second caveat refers to the absence in our DCE of attributes that
not reflect post-pandemic preferences. We highlight that previous evi­ capture potential negative externalities or disadvantages arising from
dence suggest that WTP for renewable energies suffers slight, if any, the generation and use of renewable energies. In some contexts, the
changes once an economic recession has occurred. For instance, Soon environmental benefits associated with transition from fossil fuels to
and Ahmad (2015) shows that WTP for renewable energies remained renewable energies may have trade-offs in terms of non-negligible
positive before and after the 2008 crash, and in several cases it became negative environmental impacts. Previous DCE studies have docu­
larger after the crisis —e.g. for the U.S: USD 12 in 2006 (Borchers et al., mented these trade-offs in the context of landscape impacts, impacts on
2007) vs. USD 15 in 2008 (Aldy et al., 2012); for EU: USD 10 in 2007 the fauna and flora, and noise (e.g. Botelho2018, Kosenius 2013).
(Bigerna and Polinori, 2014) vs. USD 17 in 2010 (Kontogianni et al., Recently, the president of Mexico has emphasized that his administra­
2013). Also, Kinoshita (2020) finds that renewable energies are tion will prioritize potential environmental damages when granting
preferred over conventional electricity after the occurrence of hazards. permits to build energy renewable projects (Infobae, 2020).
In addition, there is evidence in developed countries that public support The implication from not including trade-offs in terms of environ­
of policies encouraging a transition to a renewable energies portfolio has mental externalities is that welfare estimates reported in this study may
actually translated into the payment of a premium for green electricity be smaller once these externalities are incorporated. This paper does not
(e.g. Knapp et al., 2020). deal with a specific project to which a clear negative externality can be
In this context, we want to highlight that the justification of a policy attached. Thus, we deem appropriate to abstract from negative exter­
as the one described here does not need to rely on the interpretation of nalities in this application.
estimated welfare measures as the willingness of consumers to actually A few studies have explored consumers preferences in contexts in
pay for the creation of new green jobs or for an increase in the share of which higher shares of renewable energies may imply lower-quality
renewable energies. This clarification is particularly pertinent for post- services in terms of the intermittent nature of renewable energies
pandemic times because priorities of residential consumers in Aguas­ (Longo et al., 2008; Merk et al., 2019). In this respect, we highlight that
calientes may have changed —at least temporarily— and it may be the existence of a trade-off between share of renewable energies and
unrealistic to expect them to monetarily contribute to the creation of stability has recently been challenged by Diesendorf and Elliston (2018).
green jobs. However, this temporary condition should not take the focus Also, for the Mexican case, Vidal-Amaro and Sheinbaum-Pardo (2018)
away from medium- and long-term goals such as the reduction of have proposed a transition strategy to a system with a share as high as
greenhouse emissions and, ultimately, the slowdown of climate change. 75% of renewable energies in the electricity portfolio. Thus, we also
Thus, a just energy transition policy is reasonable not only on the deem appropriate to abstract from potential changes in the stability of
grounds of the stated benefits reported in the current document but also the services.
in terms of wider benefits to the Mexican and international societies.
We also wish to highlight that a policy as the one suggested here can CRediT authorship contribution statement
be designed and implemented by state and municipality administra­
tions. This is a direct consequence of the federal system in Mexico which Adán L. Martínez-Cruz: Conceptualization, Methodology, Software,
provides to states and municipalities with a degree of autonomy when it Formal analysis, Writing - original draft, Writing - review & editing.
comes to industrial and labor policies. Héctor M. Núñez: Conceptualization, Funding acquisition, Investiga­
tion, Formal analysis, Writing - original draft, Writing - review &
9. Limitations editing.

This paper’s findings come with two caveats. A first limitation refers
to the external validity. We have been careful in phrasing our conclu­ 25
In carrying out future research, we suggest to keep in mind an issue that will
sions as holding for urban residents in Aguascalientes that contribute to help clarify the representativeness of findings: interviewing only household
paying their household electricity bill because we cannot unambiguously heads may not provide a clear picture about households’ energy-related de­
claim that our results hold for the universe of urban households heads in cisions. Charlier and Martinez-Cruz (2020) argue that there is evidence from
the city of Aguascalientes. While we highlight that our sample is similar both engineering and psychological literature suggesting that household heads’
to official statistics when it comes to households’ characteristics such as characteristics and preferences are not necessarily the most important ones
when it comes to a household’s energy consumption and efficiency. Although
the number of members and whether the house is owned by a household
this claim can be controversial for our case, we believe that in a context in
member, we also recognize that at the individual level our sample is
which more than one household member pay the electricity bill, it is plausible
composed by a smaller share of married people; and bigger shares of to think that the decision of paying a premium for renewable energy may be
females, younger, and higher incomes respondents. We argue that, taken by vote and the final decision may not reflect the household head’s
because our respondents are adults that contribute (at least partially) to preferences. As pointed out by Charlier and Martinez-Cruz (2020), this is an
paying the electricity bill, these differences with respect to household issue whose potential implications have been overlooked so far in economic
heads official statistics should not be surprising. However, this argument research.

14
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

Declaration of competing interest ucional-contra-Acuerdo-emitido-por-Sener-sobre-energias-renovables-20200616-


0079.html. (Accessed 1 July 2020).
Gobierno del Estado de Aguascalientes, 2020. Convocatoria pública para acceder a los
The authors declare that they have no known competing financial apoyos del programa de proyectos productivos de MIPYMES y emprendedores 2020.
interests or personal relationships that could have appeared to influence https://www.aguascalientes.gob.mx/sedec/pdf/marzo2020/ENERGIAS-LIMPIAS
the work reported in this paper. -2020.pdf. (Accessed 11 August 2020).
Gonzalez, S., 2019. Propuestas de CFE afectarían inversiones privadas: Concamin. https
://www.jornada.com.mx/ultimas/economia/2019/12/29/propuestas-de-cfe-atent
Appendix A. Supplementary data an-contra-inversiones-privadas-concamin-3102.html. (Accessed 6 April 2020).
Graber, S., Narayanan, T., Alfaro, J., Palit, D., 2018. Solar microgrids in rural India:
consumers’ willingness to pay for attributes of electricity. Energy for Sustainable
Supplementary data related to this article can be found at https:// Development 42, 32–43.
doi.org/10.1016/j.enpol.2021.112145. Gracia, A., Barreiro-Hurlé, J.s., rez, L. Pé rez y Pé, 2012. Can renewable energy be
financed with higher electricity prices? Evidence from a Spanish region. Energy Pol.
50, 784–794.
References Haab, T.C., McConnell, K.E., 2002. Valuing Environmental and Natural Resources: the
Econometrics of Non-market Valuation. Edward Elgar Publishing.
Hancevic, P.I., Lopez-Aguilar, J.A., 2019. Energy efficiency programs in the context of
Abdullah, S., Mariel, P., 2010. Choice experiment study on the willingness to pay to
increasing block tariffs: the case of residential electricity in Mexico. Energy Pol. 131,
improve electricity services. Energy Pol. 38 (8), 4570–4581.
320–331.
Aizaki, H., 2012. Basic functions for supporting an implementation of choice experiments
Hancevic, P.I., Nuñez, H.M., Rosellon, J., 2017. Distributed photovoltaic power
in R. J. Stat. Software 50 (c02), 1–24.
generation: possibilities, benefits, and challenges for a widespread application in the
Alberini, A., Bigano, A., Ščasný, M., Zvěrinová, I., 2018. Preferences for energy efficiency
mexican residential sector. Energy Pol. 110, 478–489.
vs. Renewables: what is the willingness to pay to reduce CO2 emissions? Ecol. Econ.
INEGI, 2018a. Encuesta Nacional de Ingresos y Gastos de los Hogares (ENIGH), Instituto
144, 171–185.
Nacional de Estadística, Geografía e Informática. https://www.inegi.org.mx/progr
Aldy, J.E., Kotchen, M.J., Leiserowitz, A.A., 2012. Willingness to pay and political
amas/enigh/nc/2018/. (Accessed 6 February 2020).
support for a us national clean energy standard. Nat. Clim. Change 2 (8), 596–599.
INEGI, 2018b. Encuesta Nacional sobre Consumo de Energéticos en Viviendas
Amador, F.J., González, R.M., Ramos-Real, F.J., 2013. Supplier choice and wtp for
Particulares (ENCEVI), Instituto Nacional de Estadística, Geografía e Informática.
electricity attributes in an emerging market: the role of perceived past experience,
https://www.inegi.org.mx/programas/encevi/2018/. (Accessed 6 June 2020).
environmental concern and energy saving behavior. Energy Econ. 40, 953–966.
Infobae, 2020. AMLO criticó la instalación de aerogeneradores en La Rumorosa porque
Aravena, C., Hutchinson, W.G., Longo, A., 2012. Environmental pricing of externalities
producen “contaminación visual”. https://www.infobae.com/america/mexico/202
from different sources of electricity generation in Chile. Energy Econ. 34 (4),
0/03/29/amlo-critico-la-instalacion-de-aerogeneradores-en-la-rumorosa-porque-pro
1214–1225.
ducen-contaminacion-visual/. (Accessed 7 April 2020).
Baker, L., Phillips, J., 2019. Tensions in the transition: the politics of electricity
IRENA, 2015. Renewable Energy Prospects: Mexico. Technical Report. International
distribution in South Africa. Environment and Planning C: Politics and Space 37 (1),
Renewable Energy Agency.
177–196.
IRENA, 2018. Power Generation Costs In 2017, Volume Abu Dhabi. International
Becerril, Dinorah, 2018. Aguascalientes acelera en el sector energético. https://www.
Renewable Energy Agency.
eleconomista.com.mx/estados/Aguascalientes-acelera-en-el-sector-energetico-
Johnston, R.J., Boyle, K.J., Adamowicz, W., Bennett, J., Brouwer, R., Cameron, T.A.,
20180319-0135.html. (Accessed 11 March 2019).
Hanemann, W.M., Hanley, N., Ryan, M., Scarpa, R., et al., 2017. Contemporary
Ben-Akiva, M.E., 1973. Structure of Passenger Travel Demand Models. Ph. D. thesis.
guidance for stated preference studies. Journal of the Association of Environmental
Massachusetts Institute of Technology.
and Resource Economists 4 (2), 319–405.
Bergmann, A., Hanley, N., Wright, R., 2006. Valuing the attributes of renewable energy
Kim, J., Park, S.Y., Lee, J., 2018. Do people really want renewable energy ? Who wants
investments. Energy Pol. 34 (9), 1004–1014.
renewable energy ?: discrete choice model of reference-dependent preference in
Bigerna, S., Polinori, P., 2014. Italian households’ willingness to pay for green electricity.
South Korea. Energy Pol. 120 (April), 761–770.
Renew. Sustain. Energy Rev. 34, 110–121.
Kinoshita, S., 2020. Conjoint analysis of Japanese households’ energy-saving behavior
Borchers, A.M., Duke, J.M., Parsons, G.R., 2007. Does willingness to pay for green energy
after the earthquake: the role of the preferences for renewable energy. Energy
differ by source? Energy Pol. 35 (6), 3327–3334.
Environ. 31 (4), 676–691.
Botelho, A., Lourenço-Gomes, L., Costa Pinto, L.M., Sousa, S., Valente, M., 2018.
Knapp, L., O’Shaughnessy, E., Heeter, J., Mills, S., DeCicco, J.M., 2020. Will consumers
Discrete-choice experiments valuing local environmental impacts of renewables: two
really pay for green electricity? comparing stated and revealed preferences for
approaches to a case study in Portugal. Environ. Dev. Sustain. 20, 145–162.
residential programs in the United States. Energy Research & Social Science 65,
Castañeda-Morales, J.A., 2020. Frente a la crisis: un golpe de timon. https://nexos.com.
101457.
mx/?p=47396. (Accessed 6 April 2020).
Kontogianni, A., Tourkolias, C., Skourtos, M., 2013. Renewables portfolio, individual
Charlier, D., Martinez-Cruz, A.L., 2020. Do Habitual Energy Saving Behaviors of
preferences and social values towards res technologies. Energy Pol. 55, 467–476.
Household Heads Impact Energy Consumption in Their Own Dwelling? An
Kosenius, A.K., Ollikainen, M., 2013. Valuation of environmental and societal trade-offs
Exploration in the French Residential Sector. CERE working paper series. https
of renewable energy sources. Energy Pol. 62, 1148–1156.
://papers.ssrn.com/sol3/papers.cfm?abstract_id=3577223.
Levy, S., 2020. Superemos juntos la emergencia. https://www.nexos.com.mx/?
Chen, B., Xiong, R., Li, H., Sun, Q., Yang, J., 2019. Pathways for sustainable energy
p=47405. (Accessed 6 April 2020).
transition. J. Clean. Prod. 228, 1564–1571.
Longo, A., Markandya, A., Petrucci, M., 2008. The internalization of externalities in the
Cicia, G., Cembalo, L., Del Giudice, T., Palladino, A., 2012. Fossil energy versus nuclear,
production of electricity: willingness to pay for the attributes of a policy for
wind, solar and agricultural biomass: insights from an Italian national survey.
renewable energy. Ecol. Econ. 67 (1), 140–152.
Energy Pol. 42, 59–66.
Ma, C., Rogers, A.A., Kragt, M.E., Zhang, F., Polyakov, M., Gibson, F., Chalak, M.,
Davis, Lucas, 2019. Mexico goes backward on renewables. https://energyathaas.wordpre
Pandit, R., Tapsuwan, S., 2015. Consumers’ willingness to pay for renewable energy:
ss.com/2019/08/12/mexico-goes-backward-on-renewables. (Accessed 10 March
a meta-regression analysis. Resour. Energy Econ. 42, 93–109.
2020).
Macrotrends, 2020. US Dollar Peso Exchange Rate (USD MXN) —Historical Chart. htt
Secretaria de Energía, 2016. Evaluación rápida del uso de energía en las ciudades,
ps://www.macrotrends.net/2559/us-dollar-mexican-peso-exchange-rate-historica
mediante la implementación de trace en ciudades de la república mexicana. https:
l-chart. (Accessed 5 April 2020).
//www.gob.mx/cms/uploads/attachment/file/171177/2__Aguascalientes.pdf.
Mamkhezri, J., Thacher, J.A., Chermak, J.M., 2020. Consumer preferences for solar
(Accessed 11 March 2019).
energy: a choice experiment study. Energy J. 41 (5).
Diesendorf, M., Elliston, B., 2018. The feasibility of 100% renewable electricity systems:
Martin, J., 2019. How AMLO Is Undermining mexico’s Clean Energy Goals. https://www
a response to critics. Renew. Sustain. Energy Rev. 93, 318–330.
.worldpoliticsreview.com/articles/28370/how-amlo-is-undermining-mexico-cli
El Heraldo, 2020. Parque fotovoltaico proveerá energía para alumbrado público. http
mate-change-goals. (Accessed 11 March 2020).
s://www.heraldo.mx/parque-fotovoltaico-proveera-energia-para-alumbrado
McFadden, D., 1973. Conditional logit analysis of qualitative choice behavior. Frontiers
-publico/. (Accessed 11 August 2020).
in Econometrics 105–142.
El Universal, 2020a. Aguascalientes dará respaldo legal a empresas de energías
McFadden, D., Train, K., 2000. Mixed mnl models for discrete response. J. Appl. Econom.
renovables. https://www.eluniversal.com.mx/estados/aguascalientes-dara-respa
15 (5), 447–470.
ldo-legal-empresas-de-energias-renovables. (Accessed 1 July 2020).
Merk, C., Rehdanz, K., Schrö der, C., 2019. How consumers trade off supply security and
El Universal, 2020b. Juez frena programa energético de la 4T. https://www.eluniversal.
green electricity: evidence from Germany and great britain. Energy Econ., 104528
com.mx/nacion/juez-frena-programa-energetico-de-la-4t. (Accessed 9 September
Murakami, K., Ida, T., Tanaka, M., Friedman, L., 2015, jul. Consumers’ willingness to pay
2020).
for renewable and nuclear energy: a comparative analysis between the US and
Lider Emprersarial, 2018. Habrá Centro de Valorización de Residuos Sólidos Urbanos.
Japan. Energy Econ. 50, 178–189.
https://www.liderempresarial.com/habra-centro-de-valorizacion-de-residuos-solido
Nava, D., 2020. Concamin advierte sobre proyecto de la CRE que busca modificar reglas
s-urbanos. (Accessed 11 August 2019).
sobre suministro electrico. https://www.elfinanciero.com.mx/economia/concami
Forbes, 2020. México genera del total 31\% de energía renovable: Sener. https://www.
n-advierte-sobre-proyecto-de-la-cre-que-busca-modificar-reglas-sobre-suministro-ele
forbes.com.mx/economia-mexico-energia-renovable-sener/. (Accessed 11 August
ctrico. (Accessed 6 April 2020).
2020).
Garcia, Karol, 2020. México sufrirá frenazo de energías renovables este año. htt
ps://www.eleconomista.com.mx/estados/Jalisco-presenta-controversia-constit

15
A.L. Martínez-Cruz and H.M. Núñez Energy Policy 150 (2021) 112145

Oerlemans, L.A., Chan, K.-Y., Volschenk, J., 2016. Willingness to pay for green Soliño, M., Farizo, B.A., Vázquez, M.X., Prada, A., 2012. Generating electricity with
electricity: a review of the contingent valuation literature and its sources of error. forest biomass: consistency and payment timeframe effects in choice experiments.
Renew. Sustain. Energy Rev. 66, 875–885. Energy Pol. 41, 798–806.
Oliver, H., Volschenk, J., Smit, E., 2011. Residential consumers in the cape peninsula’s Soon, J.-J., Ahmad, S.-A., 2015. Willingly or grudgingly? a meta-analysis on the
willingness to pay for premium priced green electricity. Energy Pol. 39 (2), 544–550. willingness-to-pay for renewable energy use. Renew. Sustain. Energy Rev. 44,
Osiolo, H.H., 2017. Willingness to pay for improved energy: evidence from Kenya. 877–887.
Renew. Energy 112, 104–112. Steg, L., Shwom, R., Dietz, T., 2018. What drives energy consumers?: engaging people in
Pokhrel, K.R., 2016. Consumer Willingnes to Pay for Renewable Energy: a Meta Analysis. a sustainable energy transition. IEEE Power Energy Mag. 16 (1), 20–28.
Norwegian University of Life Sciences, Å s. Master’s thesis. (Accessed 16 March STPS, 2020. Programa Jovenes Construyendo el Futuro. https://jovenesconstruyendoel
2020). futuro.stps.gob.mx/. (Accessed 6 April 2020).
Proceso, 2020. Revelan reunión de diplomáticos de ocho países inquietos por la política Stromsta, K., 2019. WoodMac: Mexico Set to Miss Clean Energy Target Under AMLO. htt
energética de AMLO. Proceso. In: https://www.proceso.com.mx/621119/politica-e ps://www.greentechmedia.com/articles/read/woodmac-mexico-set-to-miss-clean-e
nergetica-amlo. (Accessed 6 April 2020). nergy-target-under-amlo. (Accessed 6 April 2020).
Rehdanz, K., Schröder, C., Narita, D., Okubo, T., 2017. Public preferences for alternative Su, W., Liu, M., Zeng, S., Štreimikienė, D., Baležentis, T., Ališauskaitė-Šeškienė, I., 2018.
electricity mixes in post-fukushima Japan. Energy Econ. 65, 262–270. Valuating renewable microgeneration technologies in Lithuanian households: a
Reyes-Mercado, P.V., Rajagopal, 2014. Consumer preferences for green power in Mexico. study on willingness to pay. J. Clean. Prod. 191, 318–329.
Int. J. Bus. Innovat. Res. 8 (2), 210–224. Sundt, S., Rehdanz, K., 2015. Consumers’ willingness to pay for green electricity: a meta-
Romo, Patricia, 2020. Jalisco presenta controversia constitucional contra Acuerdo analysis of the literature. Energy Econ. 51, 1–8.
emitido por Sener sobre energías renovables. https://www.eleconomista.com.mx/e Susaeta, A., Lal, P., Alavalapati, J., Mercer, E., 2011. Random preferences towards
stados/Jalisco-presenta-controversia-constitucional-contra-Acuerdo-emitido-po bioenergy environmental externalities: a case study of woody biomass based
r-Sener-sobre-energias-renovables-20200616-0079.html. (Accessed 1 July 2020). electricity in the Southern United States. Energy Econ. 33 (6), 1111–1118.
Rosemberg, Anabella, 2017. Strengthening Just Transition Policies in International Taylor, L.O., Morrison, M.D., Boyle, K.J., 2010. Exchange rules and the incentive
Climate Governance. https://stanleycenter.org/publications/pab/RosembergPABSt compatibility of choice experiments. Environ. Resour. Econ. 47 (2), 197–220.
rengtheningJustTransition417.pdf. (Accessed 11 March 2020). Train, K.E., 2009. Discrete Choice Methods with Simulation. Cambridge university press.
Sanchez, Axel, 2020. Cómo esquivar a la CFE… o a Bartlett. https://www.elfinanciero. Vecchiato, D., Tempesta, T., 2015. Public preferences for electricity contracts including
com.mx/opinion/jonathan-ruiz/como-esquivar-a-la-cfe-o-a-bartlett. (Accessed 1 renewable energy: a marketing analysis with choice experiments. Energy 88,
July 2020). 168–179.
SENER, 2018a. Atlas nacional de zonas con alto potential de energías limpias. http Vidal-Amaro, J.J., Sheinbaum-Pardo, C., 2018. A transition strategy from fossil fuels to
s://dgel.energia.gob.mx/azel/. (Accessed 6 December 2019). renewable energy sources in the mexican electricity system. Journal of Sustainable
SENER, 2018b. Programa de Desarrollo del Sistema Eléctrico Nacional 2018-2032. Development of Energy, Water and Environment Systems 6 (1), 47–66.
Secretaría de Energía. Yang, Y., Solgaard, H.S., Haider, W., 2016. Wind, hydro or mixed renewable energy
SENER, 2018c. Prospectiva de Energias Renovables 2018-2032. Secretaría de Energía. source: preference for electricity products when the share of renewable energy
SENER, 2018d. Prospectiva del Sector Electrico 2018-2032. Secretaría de Energía. increases. Energy Pol. 97, 521–531.
SENER, 2018e. Reporte de Avance de Energías Limpias Primer Semestre 2018. Secretaría Yoo, J., Ready, R.C., 2014. Preference heterogeneity for renewable energy technology.
de Energía. Energy Econ. 42, 101–114.
SENER, 2019. Programa de Desarrollo del Sistema Eléctrico Nacional 2019-2033. Zografakis, N., Sifaki, E., Pagalou, M., Nikitaki, G., Psarakis, V., Tsagarakis, K.P., 2010.
Secretaría de Energía. Assessment of public acceptance and willingness to pay for renewable energy sources
SENER, 2020. PROGRAMA Sectorial de Energía 2020-2024. Secretaría de Energía. in C rete. Renew. Sustain. Energy Rev. 14 (3), 1088–1095.

16

You might also like