Sage Journals: Discover world-class research

Abstract

Using a discrete choice experiment, owner occupants of non-apartment dwellings in Victoria, Australia were asked to evaluate alternative electricity contracts with features including increased renewables, different technologies, and demand-side management policies. We assert that a respondent’s financial literacy is correlated with whether they perceive the contract features as investments. To test this, respondents completed a financial literacy quiz, with their corresponding scores used as an indicator of financial literacy. These scores were incorporated within a hybrid mixed logit model to evaluate whether financial literacy explains the variation in choices. Results indicate that respondents with higher levels of financial literacy had a higher willingness to pay for the contract features. This case study is relevant for an environment of rising electricity prices and the transition to greater reliance on renewables.

JEL Classification: Q40, Q41, Q51

Keywords

demand-side management renewable energy community batteries financial literacy Australia

Introduction

Internationally, residential electricity prices have increased significantly over the previous decade. The factors driving these increases include domestic market structures, aging infrastructure and growth in renewables to achieve long-term climate policy objectives (International Energy Agency 2022). Recently electricity prices in the UK have increased sharply in the wake of BREXIT and increased demand for electricity post-lockdown with demand outstripping supply (Bolton and Stewart 2024). There were sharp increases in electricity prices, for EU countries dependent on Russian oil and gas for example, Germany, following the Russian invasion of Ukraine in March 2022 and subsequent sanctions (Prisecaru 2022). For countries such as Australia, households in some states faced double-digit increases in residential electricity prices in 2022 (Reserve Bank of Australia 2022). The previous decade had seen Australia move toward a sustainable energy generation mix (Csereklyei, Anantharama and Kallies 2021), nationally representing 32.5 percent of total generation (Clean Energy Council 2022). In the wake of market fluctuations, we explore the potential for infrastructure investment and demand-side management using a Discrete Choice Experiment (DCE), evaluating different energy contracts and obtaining Willingness to Pay (WTP) estimates for various contract characteristics, using the retail energy market of Victoria, Australia as a case study. A DCE allows for exploration of preferences for investment in infrastructure and the estimation of non-market values, for example, household preferences for a greener electricity mix. This method also allows for the evaluation of new or improved contract features which cannot be assessed using existing market data.

Rising prices present a challenge for achieving policy objectives focused on minimizing costs, maintaining a reliable energy system, and transitioning toward a low-carbon economy. These objectives make up the three competing elements of the traditional Energy Trilemma (Pliousis et al. 2019; Song et al. 2017; Tocock et al. 2024; Tocock et al. 2023). Although the urgency of the renewable generation transition remains, other demand side management measures or contract features could provide options for residential consumers, retailers, and regulators especially in the wake of recent market fluctuations.

This study provides insights into household preferences on different ways to manage price increases, reliability, and growth in renewables including demand-side management. Consumer contracts with demand-side features are potentially challenging for consumers to assess. One potential source of variation in household preferences is the financial literacy of the household. This form of literacy could influence the perceived benefits of various contract features and may be correlated with an individual’s propensity to focus on the upfront costs of a feature and discount the benefits realized over time. There is some evidence that this is the case for decisions evaluating alternative energy-efficient appliances (Kallbekken, Sælen and Hermansen 2013). If financial literacy does influence the choices made between alternative features of an electricity contract, then it could explain why some features fail to gain public support even when the feature is justified as an investment leading to future benefits, for example lower energy prices. This can also create problems for regulators who may believe the features are not valued as much as they would otherwise be, leading to underinvestment in new technologies. In this study, respondents completed a financial literacy quiz as part of a survey. Since financial literacy is inferred rather than observed, the respondents’ quiz score is used as a proxy indicator of their financial literacy. We use this data and choices made in a series of residential contracts to estimate a hybrid choice model. The model accounts for potential measurement errors in this proxy and allows for correlations between the socio-demographic characteristics of a respondent with their quiz score.

The DCE was designed to support decision-making processes, for example informing firms whether their consumers will co-fund investments in energy infrastructure, or perhaps assist regulators in determining whether households’ value similar contract features (Khosroshahi et al. 2021). Our results suggest that a respondent’s financial literacy influences how they evaluate contracts within the DCE. Lower levels of financial literacy were found to lead to lower stated WTP for all contract features.

The rest of the paper is structured as follows. First, a summary of the Australian electricity market and related studies are provided. Next, the methods section discusses the survey design, including the sampling approach and experimental design utilized for the choice experiment as well as how financial literacy was measured. The econometrics of the hybrid choice model is then described followed by the results and discussion. Finally, the conclusions of the study are provided alongside possible avenues for future research.

Background

Investments associated with the introduction of Australia’s Renewable Energy Target (RET) have led to mixed results with respect to the effective price households pay per kWh of electricity consumed. This result is in part due to the expected merit order effect being offset by other factors. Between 2008 and 2018 electricity prices increased by over 5 percent per annum on average (Australian Competition and Consumer Commission (ACCC) 2018, 2019), despite an average 1 percent per capita reduction in consumption during the same period (Department of the Environment and Energy 2019). Between 2019 and 2021 prices were starting to fall (ACCC 2022), however this reduction is forecast to be short-lived. A reduction in wholesale prices driven by the merit-order effect (Figueiredo and da Silva 2019; Sensfuß, Ragwitz and Genoese 2008) has not been realized in Australia in part due to increases in the price of Liquefied Natural Gas (LNG) and exit of coal-fired generation (McConnell and Sandiford 2020; Simhauser 2018). The recent spike in the price of liquified natural gas or LNG required the Australian Energy Market Operator to temporarily suspend the wholesale spot market and has called on household to restrict appliance use during peak periods in June 2022 as a demand-side measure (Australian Energy Market Operator (AEMO) 2022).

In Australia retail electricity prices are regulated, where retailers are required to justify changes in retail prices (Australian Energy Regulator 2020). Regulators in the United Kingdom and Australia have increasingly been focusing on how households perceive the value of services provided by utilities (ACCC 2019; Essential Services Commission 2016). This is, in part, achieved by retailers offering retail electricity contracts in line with household preferences. Regulators have been hesitant to approve new contracts offered by retailers unless it is clear that they are in the best interests of consumers over time. One objective of this study is to evaluate whether existing service offerings represent value for money from the average household’s perspective.

Over the last twenty years, there have been numerous studies analyzing consumers’ willingness to pay for policies that reduce CO_2-e emissions, including increased investment in renewable energy generation technologies (Alberini et al. 2018; Roe et al. 2001). One of the issues related to increased renewable energy generation is that they are often intermittent in nature, which is especially problematic when these sources of energy generation replace baseload generation technologies (Edenhofer et al. 2013). One solution to this problem is the installation of battery storage to allow low marginal cost renewable energy to be stored and sold in the peak period when demand is at its highest (Hanser et al. 2017; Ratnam, Weller and Kellett 2015; Sioshansi 2010).

In addition to supply-side investments, demand-side management policies could reduce or shift demand away from peak periods in the evening when generation is sourced from peak-load generators (Broberg and Persson 2016). Numerous market-based instruments have been developed over time, with mixed results regarding their efficacy in reducing or shifting aggregate consumption (Richter and Pollitt 2018; Strbac 2008). One specific problem relates to customer inertia, in that consumers tend not to actively review and change their electricity contracts to reduce costs (Giraudet 2020; Gleerup et al. 2010; Hortaçsu, Madanizadeh and Puller 2017). These observed behaviors are one potential explanation for the observed energy-efficiency gap (Gillingham and Palmer 2014; Jaffe and Stavins 1994). Some studies have observed that consumers do value having additional information about their electricity consumption, even if it does not prompt them to search for cheaper contracts (Gerpott and Paukert 2013; Kaufmann, Künzel and Loock 2013; Richter and Pollitt 2018).

There is a growing literature on consumers’ willingness to be flexible in their consumgption to reduce electricity costs in exchange for the disutility associated with changing habits (Kubli, Loock and Wüstenhagen 2018). This research, however, is largely focused on “prosumers”, households that both produce and consume electricity (Bergman and Eyre 2011; Parag and Sovacool 2016). As of 2021 this represents around 30 percent of all Australian households who have rooftop solar photovoltaics installed (Australian Renewable Energy Agency 2024). Another theme in the literature on habits, Tjørring et al. (2018) found that households were willing to shift their use of cleaning appliances to different times. Finally, Ruokamo et al. (2019) found that households required compensation for having their electricity consumption affected during the evening.

When households enter into electricity contracts, the price paid covers more than the wholesale cost of generation. Investments in infrastructure that delivers electricity as well as funding the RET are just some of the additional costs that are paid. Electricity that is renewable and reliable may represent an investment for households, in that more is paid today for future benefits to be enjoyed in the future. We hypothesize in this study that financial literacy may be a determinant influencing how households evaluate different contracts, representing different investment opportunities. Financial literacy is important since it determines an individual’s capacity to evaluate financial information. It has been shown to be correlated with effective retirement planning and positive savings behavior (Behrman et al. 2012; Clark, Lusardi and Mitchell 2015). It has also been shown as a predictor for determining the ability of individuals to make investment decisions (Stolper and Walter 2017). Meier and Sprenger (2013) investigate the relationship between financial literacy and time preferences. They find that individuals who participate in financial education programs have relatively lower discount rates when making investment decisions. Sutter et al. (2013) designed an incentivized experiment eliciting children’s and adolescents’ time preferences, with their results suggesting a link between impatience (higher discount rates) and poor saving choices. This link is also suggested in Lührmann, Serra-Garcia and Winter (2018), where a statistically significant relationship was found between financial education and the proportion of time-consistent choices. A paper by Brent and Ward (2018), the only similar study to our knowledge, finds that financial literacy influences the consistency of choices made when evaluating alternative hot water systems. Respondents who were less financially literate were more likely to purchase systems that had a lower upfront cost, but higher running costs over time. In this study our contribution to this literature is the expansion on the idea that financial literacy may be important to account for when households evaluate alternative electricity contracts, specifically influencing the willingness to pay for the features associated with each contract.

Methods

Survey Design

The survey used in this study was developed as part of a larger, multi-treatment project, investigating various aspects of consumer preferences for alternative electricity contracts. We use one treatment in this analysis. After reading a general information screen about the project, the households who provided informed consent (Ethics Clearance H0016832), proceeded to view information and answer questions across a four-part survey. The financial literacy quiz was the first part of the survey. The second part of the survey described how Australian retail electricity prices have increased across the country over the last ten years, as well as describing some reasons why this has occurred. The third part explained the contracts to be evaluated and their associated features. A cheap talk script was provided following the feature descriptions as a reminder to respondents to make choices as if they really had to pay. Following this reminder, respondents were asked to look at the combination of features and make a choice. Respondents then completed eight choice tasks selecting from three different electricity contracts. The fourth part of the survey involved answering attitudinal and socio-demographic questions. In the development of the survey, focus groups and in-depth interviews were conducted with paper copies of the survey to test wording and survey length.

The contracts that would be offered by retailers included a combination of the following features, namely changes in the amount of power sourced from renewable energy generation, limiting appliance use during the evening peak period, installation of batteries to store electricity that can be accessed by the community, and finally the installation of smart meters that would provide households with more frequent updates about the cost of powering their home.

Each feature was described as having the potential to lower future electricity prices within the national electricity market in the long run. During the time period when the survey was in-progress, renewable energy generation made up just over 15 percent of the national energy mix, with the majority of electricity generated from burning coal.

Peak energy consumption for households in Victoria, on average, occurs between the hours of 5pm and 8pm. During this time, peak load generation leads to substantially higher prices being demanded by generators. One way to lessen the need to access these higher priced sources is to restrict how much power can be consumed during this peak period by households. By default, to keep prices constant, all electricity retailers would restrict two of three possible consumption activities. The three activities mentioned included cooking, cleaning, and entertainment. A list of common appliances associated with each activity was also detailed to provide context. As part of the feature description, households would get to choose which activities they would restrict. Alternative contracts would allow for one or both sets of restrictions to be lifted, however, these options never had a zero-cost associated with them. Several questions queried whether respondents understood what these restrictions would mean for their electricity consumption habits.

Another feature presented was related to community storage. At the time of the survey, there were few large-scale batteries operating nationally. A major battery project had been built in South Australia which was widely reported (Sonali 2017), therefore, respondents are likely to have been aware of the potential for such battery projects. The description of this feature was based on the assumption that batteries would be usable as a replacement source of power during a blackout as well as a substitute for peak-load power sources in the evening.

The final feature described whether households could access their consumption information more frequently through the installation of smart meters. Currently, most Australian households only receive this information with their quarterly bill, whereas the smart meters would allow consumption information to be received either daily or in real-time, depending on the type of meter and frequency households prefer. For each of the contracts offered, except for the status quo contract, there is a cost imposed on the household. The cost was defined as an increase in the fixed component of the household’s electricity bill, paid every quarter for five years.

There was a business-as-usual, status quo contract, with no additional cost being imposed on households. This option resulted in no increased investment in renewable energy or storage activities (as a proportion of the current energy infrastructure mix), consumption information being provided quarterly, and two consumption activities being restricted for each household. An example of the choice tasks is shown in Figure 1. The list of features and their associated levels are shown in Table 1. Between each respondent, the order of contract features, apart from costs, was randomized to eliminate potential order effects.

Figure 1.

Example choice task.

Table 1.

Description of Features and Levels.

Features	Levels (Non-status quo)	Levels (Status quo)
Proportion of generation from renewable sources	15 percent, 30 percent, 45 percent, 60 percent	15 percent
Consumption restrictions	Two restrictions, One restriction, No restrictions	Two restrictions
Consumption information	Quarterly, Daily reminders, Real-time reminders	Quarterly
Community storage	0 MWh, 20 MWh, 40 MWh, 60 MWh	0 MWh
Fixed cost increase per quarterly for five years to your household	$20, $25, $30, $35, $40, $45, $50, $55, $60, $65, $70, $75	$ 0

Survey Sample and Experimental Design

A stratified random sampling method was utilized across Victoria based on gender, age, and rural versus urban location. The survey was administered online in June 2019 by Online Research Unit (ORU; http://theoru.com/), an Australian panel provider who continuously refresh their panel using a combination of techniques to reduce issues of self-selection bias. All data collection occurred in May to June 2019. The data collection was not impacted by COVID-19. Retail electricity prices increased post-covid and the complex trade-offs became even more relevant for households impacted by increasing electricity prices.

An efficient design, based on a multinomial logit model, was initially developed using priors obtained from the literature, specifically priors relating to renewable energy investments (Brennan and Van Rensburg 2016; Ozbafli and Jenkins 2016). For the other features, namely consumption limits, storage, and consumption information, no priors were available. The expected sign for these parameters was positive, so they were calibrated to ensure utility balance and no dominated alternatives (Bliemer and Collins 2016). All designs were generated using Ngene version 1.2 (ChoiceMetrics 2018). All designs estimated included the main effects only.

The initial design was piloted, and a basic model was estimated with the significant estimates included as priors in a Bayesian D-efficient design. For the parameters related to billing information, storage, and no consumption limits, the estimated coefficients were not significant in the simple multinomial logit model estimated using pilot data (n = 56). It is not always possible to know if a statistically insignificant parameter is due to the small size of the pilot or due to the feature being unimportant to respondents. These parameters were set to be positive to balance utility and ensure no dominated alternatives. The final design included forty-eight choice tasks divided into six blocks with a simulated Bayesian D-efficient error of 0.003448.

Measurement of Financial Literacy

Several methods have been developed to measure a respondent’s financial literacy. The most prevalent approach involves test-based measures (Hung, Parker and Yoong 2009), whereby responses to a set of questions pertaining to specific financial concepts are often aggregated. Other methods include estimating effects using indicator variables (Gathergood 2012; Jappelli 2010). Self-assessed measures of financial literacy are another approach, whereby financial literacy is inferred from a respondent’s self-assessment (Stolper and Walter 2017).

A set of questions assessing each respondent’s knowledge of financial investments were used to measure financial literacy. The quiz was designed for Australian respondents with experts in financial planning to assess the likelihood that a consumer could make sound, independent financial investment decisions (Louviere et al. 2016). Other studies that have focused on energy-related financial literacy questions, for example Blasch et al. (2021) and Filippini, Kumar and Srinivasan (2020). These indexes were used to test the link between financial literacy and the purchase of energy-efficient durable goods. We opted for the use of investment-specific measures of financial literacy. We hypothesized that households who have relatively more knowledge of concepts related to financial investments would also perceive the contract features as investments and therefore have a higher willingness to pay.

Seven questions from the quiz related to financial investments were included in the survey and were answered by all respondents. The questions asked are shown in Table 2 below.

Table 2.

Financial Literacy Questions (Multiple Choice).

Question	Possible answers (Order randomized across respondents)
Normally, which of these assets exhibits the highest fluctuations over time?	(a) Savings accounts (b) Stocks (c) Bonds
Which of the following is ALWAYS true when dividend payments are received?	(a) The payment would be the same dollar amount every year (b) The payment would vary from year to year (c) The payment would be a fixed percentage of the share price (d) The payment only rises and falls with interest rates
It is usually possible to reduce the risk of investing in the share market by buying a wide range of shares.	(a) True (b) False
If you invest $1,000 in a managed fund, is it possible to have less than $1,000 when you withdraw your money?	(a) Yes (b) No
Is an investment with a high return likely to be high risk?	(a) Yes (b) No
If a friend inherits $10,000 today and her sister inherits $10,000 three years from now, who will be richer in three years because of the inheritance?	(a) Your friend (b) Her sister (c) They will be equally rich
If you own shares in an Australian company, which ONE of these statements is true about the tax you will pay on dividend income?	(a) The dividend income is taxed at a fixed rate of 15 percent (b) If the dividend carries franking credits, you are eligible for a tax offset for the company tax already paid (c) If the dividend carries franking credits, you pay no tax on the dividend (d) The dividend income is not taxed (e) Don’t know

The order of the questions was randomized between respondents to eliminate potential ordering effects. Each question was worth one point, with a maximum score of seven points, representing the highest level of financial literacy. Although higher scores imply higher levels of financial literacy, the scores themselves are ordinal. Each question has an unequal weighting in terms of their importance in assessing an individual’s financial literacy. Consequently, including the scores as interaction terms would lead to measurement error. The choice of modeling these scores as part of a hybrid choice model can address the measurement error issue (Vij and Walker 2016). In addition, it allows for the linking of socio-demographic factors with the respondents’ scores. It was assumed that a high level of financial literacy may be necessary to evaluate electricity contracts involving multiple features associated with investments in energy infrastructure and policy. The non-status quo contracts incurred costs every quarter for five years. The benefits of increased renewables and battery storage, however, are long-term. For example, increased renewable energy generation leads to less reliance on fossil fuels, reducing the associated negative externalities over time. The costs of installing smart meters could be offset through cost savings as consumption information allows households to understand how to lower their future electricity bills. Finally, paying for the removal of restrictions could better align the high costs of generation during the peak period with the final prices paid by households.

Model Specifications

A random utility framework underpins our model of consumer preferences for electricity contracts. With this framework, each decision maker n is expected to obtain a level of utility from a specific alternative j for choice task c equal to $U_{njc}$ . The most common functional form assumed is a linear relationship among attribute levels for each alternative and their respective parameters, as shown in equation (1):

U_{njc} = β x_{njc} + ε_{njc} \forall j, j = 1, 2, . ., J

(1)

Based on this specification, $x_{njc}$ is a vector of attribute levels (hereinafter referred to as contract features), $β$ represents a vector of parameters to be estimated, and $ε_{njc}$ is the unobserved error term. Following the approach in McFadden (1974), equation (2) shows that the probability of individual n selecting alternative j in choice task c as:

P_{njc} = Prob (V_{ncj} + ε_{ncj} > V_{nci} + ε_{nci}) \forall j \neq i

(2)

The basic multinomial logit model is restrictive in the assumptions made regarding preferences and decision heuristics. To account for preference heterogeneity, a mixed multinomial logit specification was employed utilizing both random parameters and an error component. Preference heterogeneity for each contract feature is modeled through the fixed preference parameter is now expressed as shown in equation (3):

β = β^{m} + Ω υ_{n}

(3)

where the first component $β^{m}$ represents a vector of mean preference estimates for each feature. Random taste variation is represented by a vector of zero-mean random variables $υ_{n}$ as well as a lower-triangular matrix $Ω$ that estimates the covariances between random parameters. For every non-cost feature, a normal distribution is specified for the error term $υ_{nk}$ . A lognormal distribution is specified for the cost attribute’s error-term to avoid the issues associated with defining the ratio of two normally distributed parameters (Bliemer and Rose 2013). Full correlation is modeled between the random parameters, with the correlation matrix available from the authors upon request.

For this study, a three-alternative unlabeled choice experiment has been created with one of the alternatives specified as the status quo. To control for left-right bias and/or status quo bias, three additional parameters have been included in the model. Two alternative-specific constants are estimated (as random parameters) for the status quo contract and the third contract (Option C). Further, a zero-mean normally distributed error component has been included for the two non-status quo alternatives to account for potential differences in substitution patterns between the status quo and non-status quo alternatives (Herriges and Phaneuf 2002; Scarpa, Willis and Acutt 2007).

In this study, the choice model is directly estimated in Willingness to Pay Space (WTPS; Train and Weeks 2005), as shown in equation (4):

U_{njc} = α (δ_{AC} + ω x_{njc} - p_{njc} + κ_{nBC}) + ε_{njc}

(4)

where $α$ is the estimated cost parameter and $p_{njc}$ represents the cost attribute and $ω$ is defined as the ratio $(\frac{β}{α})$ . The cost parameter is perfectly confounded with scale within the model. The last set of parameters represent the alternative specific constants for options A and C ( $δ_{AC}$ ) and the error component ( $κ_{nBC}$ ) shared between options B and C, also expresses in WTPS.

To model how financial literacy influences willingness to pay an interaction parameter is included in the specification of the cost parameter, as shown in equation (5):

α = \exp (σ + ϕ Fi n_{n} + τ υ_{n})

(5)

where $σ$ is the mean estimate of the cost/scale parameter, the product $τ υ_{n}$ estimates the random variation in the cost parameter, and $ϕ$ is an interaction parameter with the financial literacy latent variable for individual n. We hypothesize that $ϕ$ will be positive, suggesting that the larger the financial literacy variable the larger the WTP for each contract feature.

Financial literacy cannot be directly observed as it is a latent variable and can therefore only be indirectly measured as responses to questions designed to measure the latent variable. To allow for the latent nature of financial literacy, a hybrid choice model can be used to simultaneously measure the latent variable analyzed as well as its effect on the discrete choice experiment (Ben-Akiva et al. 2002; Daly, Hess and Train 2012). This modeling approach can also address issues related to measurement error with latent variables (Budziński and Czajkowski 2022). Figure 2 shows an overview of how the latent variable model fits in with the current discrete choice model.

Figure 2.

Overview of hybrid mixed logit model.

Financial literacy is modeled using a structural equation estimating the correlation between literacy and socio-economic characteristics, as shown in equation (6):

Fi n_{n} = ρ z_{n} + η_{n}

(6)

where, $ρ$ is a vector of estimated coefficients for the vector $z_{n}$ of socioeconomic characteristics for respondent n, and $η$ is a normally distributed error term. For this study, the characteristics modeled for each respondent include their age, gender, education level, and income. Education is recorded as three separate effects-coded variables, with high school the base level which is compared against higher levels of education including TAFE (post-high school, technical skills training), undergraduate and postgraduate (Honours year, Masters, and Ph.D.) studies. Finally, income has been included in a similar manner as education, with an annual income of less than $41,000 set as the base level. Other levels, including for those respondents who preferred not to state their income, were included to test whether there were differences in financial literacy for said income levels relative to the base level. With this structural equation, correlations among the financial literacy score and the socio-demographic characteristics of the respondents can be estimated.

The latent variable is also linked to an indicator function, in this case, the individuals’ score for the financial literacy quiz via equation (7):

I_{m}^{*} = μ + γ Fi n_{n} + ζ_{m}

(7)

with $I_{m}^{*}$ representing the individual’s score on the quiz, $μ$ is a vector of threshold parameters, $γ$ is a vector of coefficients to estimate and zeta is an extreme value distributed error term.¹ As the financial literacy score is an ordinal variable, a linear specification linking the score to socioeconomic characteristics would lead to a misspecification error. Therefore, an ordered logistic function is used to calculate the probabilities for each level of the indicator variables.

Both the choice model and latent variables are solved simultaneously. The log-likelihood function is solved using simulated maximum likelihood, with the combined solution to equations (4) and (7) being shown as:

LL = \sum_{n = 1}^{N} \ln [\frac{1}{R} \sum_{r = 1}^{R} [Π_{j = 1}^{J} Π_{c = 1}^{C} {(P_{njc}^{r})}^{y_{njc}} \times Π_{m = 1}^{M} P (I_{m}^{r})]]

(8)

where, $y_{njc}$ denotes the actual choices made and r the number of draws used for simulation. The draws are taken for the errors mentioned above, with the probabilities calculated for each set of draws. The draws were sampled using Modified Latin Hypercube Sampling (Hess, Train and Polak 2006) with 5,000 draws to ensure the stability of parameter estimates. The final model was estimated using Python Biogeme (Bierlaire 2016) with supporting code from Rose and Zhang (2017).

Results and Discussion

In total 18,250 invites (random sample of the ORU panel stratified by urban/rural, age, and gender) were sent out with reminders in three waves to obtain 518 respondents in total. Socio-demographic targets were used in estimating the required number of observations, but not imposed as quotas. To be eligible to complete the survey, the respondent had to be owner-occupier (with or without mortgage) of a single-family, detached house and responsible for the electricity bill (thus excluding renters and occupants of apartments, townhouses, etc.). The sample respondent characteristics reported in Table 3 are compared with state proportions based on the 2016 census.² Although we did not see any statistically significant difference between the sample and census proportions of age and gender there are differences in terms of education and income. In our sample there was an oversampling of respondents who had stated their highest stated level of education was a bachelor’s degree or higher. There was also an under sampling of lower-income households. Consequently, there is a limitation as to how generalizable the reported results may be. We expand on this point further at the end of the paper.

Table 3.

Comparison of Sample to Population Proportions for Victoria.

Age	Rest of Victoria				Greater Melbourne
	Men		Women		Men		Women
	Sample (%)	Census (%)	Sample (%)	Census (%)	Sample (%)	Census (%)	Sample (%)	Census (%)
18–29	2.12	1.84	2.12	1.75	8.49	8.37	8.30	8.24
30–44	3.09	2.59	2.51	2.70	11.58	11.34	11.58	11.43
45–59	3.67	3.01	3.47	3.17	8.11	9.06	8.49	9.53
60+	4.83	3.94	3.47	4.28	8.69	8.69	9.27	10.06
Education			Sample (%)	Census (%)	Income		Sample (%)	Census (%)
Year 12 or below			18.92	50.92	$1–$51,999		21.43	29.81
Tafe			22.39	24.19	$52,000–$90,999		21.62	21.61
Bachelor’s and above			58.69	24.89	$91,000–$155,999		26.06	22.64
					$156,000 or more		16.80	13.92
					Prefer not to say		14.09	12.02

For the financial literacy quiz, each respondent completed seven questions as shown in Table 4. When compared to the results reported in Louviere et al. (2016), the percentage of correct answers is comparable for most questions. There was an exception with one question regarding the tax consequences of dividend income where we added “I do not know” as a response, which was not the case in Louviere et al. (2016). This response was added based on feedback received in focus groups. These responses were then randomly assigned to the remaining results in a similar proportion of correct responses.

Table 4.

Financial Literacy Percentage of Correct Answers Comparison.

Question	% Correct
	Current survey	Louviere et al. (2016)
Normally, which of these assets exhibits the highest fluctuations over time? [Multiple Choice]	82.8	85.6
Which of the following is ALWAYS true when dividend payments are received? [Multiple Choice]	76.8	76.8
It is usually possible to reduce the risk of investing in the share market by buying a wide range of shares. [True/False]	85.7	85.5
If you invest $1,000 in a managed fund, is it possible to have less than $1,000 when you withdraw your money? [True/False]	84.6	79.9
Is an investment with a high return likely to be high risk? [Yes/No]	92.5	94.0
If a friend inherits $10,000 today and her sister inherits $10,000 three years from now, who will be richer in three years because of the inheritance? [Multiple Choice]	69.7	61.9
If you own shares in an Australian company, which ONE of these statements is true about the tax you will pay on dividend income? [Multiple Choice]	44.4	58.9

The choice frequencies for each of the contract alternatives are shown in Table 5. More than a third of all choices involved selecting the status quo contract. This result suggests that there is a significant proportion of respondents selecting the status-quo contract despite having two consumption restrictions imposed on the household. Of all the respondents who completed the survey, 26 (five percent) selected a contract that involved at least one restriction being removed for every choice task.

Table 5.

Proportion of Alternatives Selected.

Alternative selected	Proportions (%)
Option A (Status quo)	35.67
Option B	35.76
Option C	28.57

The socio-demographic factors included in the structural equation model are all statistically significant, as shown in Table 6. Older, male, and more educated respondents were more likely to be financially literate. In terms of the relative magnitude of the education parameters, we do not have sufficient statistical evidence to suggest any difference between respondents who have an undergraduate versus postgraduate degree as their highest level of education. The negative coefficient for gender (female = 1) is consistent with past studies (Bucher-Koenen et al. 2017; Klapper, Lusardi and van Oudheusden 2015). The sign on education is also consistent with past studies (Christelis, Jappelli and Padula 2010; Lusardi, Mitchell and Curto 2010). The positive relationship between age and financial literacy is less clear, with some studies suggesting a non-linear relationship, with financial literacy increasing up until a certain age, then falling (Finke, Howe and Huston 2017; Gamble, Juliusson and Gärling 2009). A quadratic term for age was included in an alternative model, however, the final log-likelihood for this model was inferior. Income was found to be statistically significant for the highest levels (household income exceeding $156,000 per annum). This result is broadly in line with past studies, see for example Lusardi and Mitchell (2014) and Mauldin et al. (2016). Finally, for the measurement equation the gamma parameter is positive and significant, suggesting that higher levels of the latent variable lead to a greater likelihood of scoring higher on the financial literacy quiz. The threshold parameters are all significant except for two levels, which correspond to lower scores. This is not surprising given the relatively few respondents who correctly answered a maximum of up to four questions.

Table 6.

Structural and Measurement Equation Results.

Variable	Coefficient (Robust standard error)
Age	0.0549*** (0.0161)
Gender	−0.4593*** (0.1638)
Trade qualification (TAFE)	0.1981 (0.2425)
Undergraduate education	0.9207** (0.3712)
Postgraduate education	0.7685** (0.3244)
Income $41,000–$77,999	−0.1409 (0.1696)
Income $78,000–$103,999	−0.1944 (0.2127)
Income $104,000–$155,999	0.3701* (0.1932)
Income $156,000 or More	0.7125** (0.2844)
Income prefer not to say	0.0674 (0.2112)
$μ_{1}$	−2.4944*** (0.5881)
$μ_{2}$	−1.3169*** (0.4576)
$μ_{3}$	0.2525 (0.3835)
$μ_{4}$	1.7595*** (0.4327)
$μ_{5}$	3.3140*** (0.5937)
$μ_{6}$	5.0870*** (0.8533)
$Γ$	1.2069** (0.5324)

***

1 percent significance. **5 percent significance. *10 percent significance.

The results of the Hybrid Mixed Logit Model are reported in Table 7. Starting with the mean parameters, all features are statistically significant and have the expected sign. The alternative-specific constants are negative, suggesting that there is unobserved heterogeneity that leads respondents to select contracts that are not the status quo or option C. In terms of the relative importance of features, removing consumption restrictions was the most important for respondents.

Table 7.

Hybrid Mixed Logit Model (WTPS).

Variable	Coefficient (Robust standard error)
	Mean parameters	Standard deviation parameters
Daily reminders	0.6085*** (0.1004)	3.1429*** (0.1442)
Real-time reminders	0.5152*** (0.1179)	6.9941*** (0.1202)
One consumption restriction	3.4617*** (0.1461)	8.0493*** (0.1333)
No consumption restrictions	6.229*** (0.1573)	8.2149*** (0.0972)
Renewable generation	0.2556*** (0.0039)	0.4119*** (0.0053)
Storage	0.0456*** (0.0020)	0.1762*** (0.0039)
Household cost ($/year)	−3.2175*** (0.4759)	1.7303*** (0.2164)
ASC (Status quo)	−35.6777*** (0.2537)	0.1102 (0.2881)
ASC (Option C)	−4.0322*** (0.1780)	7.8867*** (0.1191)
Error component		39.0713*** (0.4582)
Financial literacy cost interaction ( $ϕ$ )	0.5715*** (0.1274)
Diagnostics
No. of observations		4,144
Log-likelihood		−3,069.790
BIC		6,483.329
McFadden pseudo R²		0.326

***

1 percent significance. **5 percent significance. *10 percent significance.

For the remaining features, it was assumed that more frequent information would allow households to better optimize their consumption (Filippini, Hirl and Masiero 2018). Gerpott and Paukert (2013) found that German households were willing to pay a one-time levy for consumption information, but not an ongoing monthly fee. Kaufmann, Künzel and Loock (2013) found that Swedish households were willing to pay up to 1.55 CHF on top of their existing tariff for real-time information shown on an in-home display. The difference in mean estimates for the billing reminders suggests that it is better to have more information, however the perceived differences in the value of daily versus real-time information is minimal. The small WTP values in our study may also be due to the perceived low economic benefits and privacy concerns associated with smart meters identified in previous studies (Léautier 2014; von Loessl 2023).

In terms of consumption restrictions, the results suggest that consumers would rather have no restrictions on their consumption activities. This being the first study to estimate this specific type of restrictions, there are no direct comparisons. Kavousian, Rajagopal and Fischer (2013) identified numerous factors that influence household consumption based on smart meter data. They confirm previous studies concluding that energy consumption is driven by behavior rather than perceived efficiency gains (Cramer et al. 1985; Gouveia, Fortes and Seixas 2012). These findings suggest that households are opting out of consumption restrictions to preserve habits, rather than considering the efficiency gains from shifting some activities to other times of the day. The information regarding the type of tariff the household incurred was not collected in this study, although it could be a useful piece of information for any future research analyzing consumption restrictions with price-incentives.

The WTP estimate for the community storage feature has limited comparability with previous studies. In part, this is due to how we described the feature as a technology that reduced the duration of blackouts. Goett, Hudson and Train (2000) estimated that households were willing to pay a value of $1.21 cents per kWh to reduce outages from four 30-minute periods to two of the same length. Carlsson and Martinsson (2008) estimated that Swedish households are willing to pay 7.81 SEK to reduce power outages to a maximum of four hours over the weekend, although they noted significant heterogeneity with respect to the WTP, depending on when the outage occurs, as well as the length. The estimate in our study is closer to the lower end of the studies identified, with households found to be willing to pay an additional $ 0.1824 per year for each MW of battery storage installed. Using the Hornsdale Power Reserve battery capacity of 100 MW, the average annual WTP would be $18.24, which when considering the 30,000 homes that could be powered would lead to a benefit $547,000 per annum in addition to the other commercial benefits the battery provides (Global Infrastructure Hub 2019).

One explanation for the low WTP estimate is the fact that 44 percent of the cost of electricity bills goes toward supporting existing infrastructure and ensuring already high levels of system reliability (Australian Energy Market Commission 2020). Another potential reason is that batteries are not new sources of generation, but rather allow for energy to be stored and discharged later. Baseload renewable energy sources of generation such as pumped hydroelectric dams already exist in Australia and may be considered suitable substitutes. Finally, the relationship between large-scale battery storage and electricity prices may not be well-understood by consumers. Energy storage technologies, excluding pumped hydro storage, make up a small proportion of most countries’ existing energy infrastructure (International Renewable Energy Agency 2017). In Australia, the only large-scale battery storage in operation at the time of the survey was the Tesla-Neoen 100 MW lithium-ion battery, however, new installations in operation in 2021 total to 260 MW, with an additional 18,660 MW proposed for construction (AEMO 2021)

Finally, for the renewables WTP feature, we can compare the results with other studies focusing on renewable energy technologies, independent of the type of technology. Ivanova (2012) estimated that Queensland households are WTP an additional $28 AUD on top of their quarterly electricity bill to support renewable energy sources. Roe et al. (2001) found that US households were willing to pay an annual $21 US premium on top of their existing electricity costs. Our results are relatively lower in real terms. A general finding by Sundt and Rehdanz (2015) is that estimated WTP is generally lower in studies which do not specify the type of renewable energy technology. Another factor that may explain this result is that we expressed the change in renewable energy technologies as a percentage change, which previous studies did not. Our WTP estimates could be useful for policy evaluation, when combined with work by Nelson, Nolan and Gilmore (2021) where changes in the proportion of renewable energy and the reductions in Australia’s emissions profile have been modelled. Nelson, Nolan and Gilmore (2021) estimated that for every 1 percent change in renewable energy emissions have been reduced by 1.635 million tons each year. Although caution must be applied in assuming the same linear relationship going forward, the results in this study suggest that households are willing to pay more to reduce emissions.

For every feature, the associated standard deviation parameters are statistically significant, implying preference heterogeneity within the sampled population (Hensher, Rose and Greene 2015). A statistically significant error component suggests that respondents are trading off between the non-status quo alternatives differently relative to the status quo. The financial literacy cost interaction parameter suggests that the sociodemographic characteristics of the sample, which are correlated with the responses to the financial literacy quiz, also influence the willingness to pay for each of the contract features.³ In terms of financial literacy and its influence within a discrete choice experiment, the only other paper that can be compared is the recent study by Brent and Ward (2018). Utilizing a similar methodology, they find that people who are financially literate make more consistent choices. There are however two important points to note when comparing results. In the Brent and Ward (2018) study, they estimate a Generalized Mixed Logit without the hybrid specification. They conclude that financial literacy is a source of scale heterogeneity, whilst in our study we find evidence that suggests that financial literacy may influence the willingness to pay for the contract features presented.

From a policy perspective, accounting for a respondent’s financial literacy can be important when considering policies that represent financial investments, such as investments in energy infrastructure. These investments are gradual, with the benefits sometimes not realized until many years after the costs have been incurred. This study finds evidence that the valuation of the features included in this choice experiment are correlated with the financial literacy of the respondent. Although the measure of financial literacy modeled in this paper is a proxy, nonetheless, it provides additional evidence that financial literacy does influence willingness to pay. If respondents do not consider the benefits associated with the features evaluated, they may only focus on the cost or complete preference elicitation exercises in a manner reflecting indeterministic preferences. One question that could be the focus of future research is whether financial literacy plays a role in explaining the choice heuristics utilized by respondent when evaluating contracts, and potentially whether financial literacy influences the scale parameter.

Conclusions and Future Research

Over the past decade, Australian households have been facing rising electricity costs. This study shows that there is still broad support for increasing investments that improve the existing network despite the rising energy prices for consumers. From a policy perspective in Australia, there appears to be support for additional renewable energy generation. In Australia, this is achieved through the long-run renewable energy target. In addition to funding more renewables, households are also willing to pay to support investments in battery storage. In this choice experiment, storage was presented as one way that would increase the reliability of the network. Battery storage, however, could also be used to partially address intermittency issues associated with renewable energy generation, thereby improving their perceived value within the national energy mix. From a regulator’s perspective, the significant WTP estimates suggest that consumers perceive battery storage as value for money.

The choice experiment explored the potential for consumption restrictions as well as the installation of smart meters. The results suggest that households do not want to change their habits during the peak evening consumption period at least prior to the latest large price increases. There is scope for rerunning this type of survey following the experience of some large price increases and calls for consumers to restrict usage at different times of day. In addition, there is also support for smart meters to monitor consumption habits. The benefits of smart meters relate to monitoring real-time consumption through daily reminders.

Finally, this study also looked at how financial literacy influenced the willingness to pay for the contract features presented. The results suggest that those respondents who were less literate, as proxied by the low scores on the financial literacy quiz, had a lower willingness to pay for each of the contract features. It is important to state some limitations regarding the generalizability of the results. The results reflect the responses of homeowners (with or without a mortgage) of non-apartment dwellings. Our results reflect the views of this subset of households and care should be taken in generalizing to a general population. In addition, the oversampling of respondents with bachelor’s or postgraduate degrees and under sampling of low-income households may have biased the measured impact of financial literacy on contract choice. Despite these limitations our results provide some evidence that improving the financial literacy of respondents could lead to the wider acceptance of contract features that reflect investments in the future, as opposed to features that merely reflect costs to be borne by households today. Related to this idea, for future studies, it may also be worth exploring a household’s financial literacy with their knowledge of other facets of the electricity network, or environmental problems more generally. Exploring these relationships may provide richer insight into the other factors that determine which households make consistent decisions. Another area for future research could focus on exploring in more detail the link between financial literacy and specific contract features.

Footnotes

Appendix

Table A2.

Discrete Choice Model Financial Literacy Interactions Other Contract Features.

Variable	Coefficient (Robust standard error)
	Mean parameters	Standard deviation parameters
Daily reminders	−1.1384** (0.519)	2.8314*** (0.1259)
Real-time reminders	2.3763*** (0.4281)	7.1874*** (0.1204)
One consumption restriction	2.6156*** (0.3225)	7.6489*** (0.1014)
No consumption restrictions	6.1442*** (0.3695)	8.3544*** (0.0863)
Renewable generation	0.1051*** (0.0139)	0.3848*** (0.007)
Storage	0.0602*** (0.0061)	0.1749*** (0.0044)
Household cost ($/year)	−1.5377*** (0.1504)	2.0116*** (0.2102)
ASC (Status quo)	−34.8593*** (0.3326)	2.0999*** (0.1795)
ASC (Option C)	−3.7961*** (0.2021)	8.3237*** (0.1617)
Error component		39.3989*** (0.5392)
Financial literacy interactions
Daily reminders × Financial literacy	0.6255*** (0.1036)
Real-time reminders × Financial literacy	−0.6446*** (0.0837)
One consumption restriction × Financial literacy	0.2920*** (0.0876)
No consumption restrictions × Financial literacy	0.2701** (0.1194)
Renewable generation × Financial literacy	0.0356*** (0.0044)
Storage × Financial literacy	−0.0036* (0.0020)
Diagnostics
No. of observations		4,144
Log-likelihood		−3,068.510
BIC		6,512.019
McFadden pseudo R²		0.326

10 Percent significance; **5 Percent significance; ***1 Percent significance.

Acknowledgements

The authors are grateful to Jasmine Tocock for useful comments on financial literacy provided early in the development of this study.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding for this study was provided by the University of Tasmania.

ORCID iD

Mark Tocock

1

For all the error terms included in the hybrid model specification it is assumed that error terms are mutually independent (Vij and Walker 2016).

2

Results for the sample test of proportions are reported in the . We note for comparing education and income some aggregation is required to facilitate comparisons with census numbers, details of the aggregation are also mentioned in the Appendix.

3

In the we have estimated an alternative model whereby each feature other than cost is interacted with financial literacy. We opted not to include in the manuscript due to a relatively poorer model fit.

References

Alberini

Bigano

Ščasný

Zvěřinová

2018. “Preferences for Energy Efficiency vs. Renewables: What is the Willingness to Pay to Reduce CO₂ Emissions?”Ecological Economics 144: 171–85. doi:10.1016/j.ecolecon.2017.08.009.

Australian Competition and Consumer Commission. 2018. “Restoring Electricity Affordability and Australia’s Competitive Advantage.” Retail Electricity Pricing Inquiry—Final Report. June. https://www.accc.gov.au/system/files/Retail%20Electricity%20Pricing%20Inquiry%E2%80%94Final%20Report%20June%202018_0.pdf.

Australian Competition and Consumer Commission. 2019. “Inquiry into the National Electricity Market.”November. https://www.accc.gov.au/system/files/Inquiry%20into%20the%20National%20Electricity%20Market%20-%20November%202019%20Report.PDF.

Australian Competition and Consumer Commission. 2022. “Inquiry into the National Electricity Market.”May. https://www.accc.gov.au/system/files/Inquiry%20into%20the%20National%20Electricity%20Market%20-%20May%202022%20report%2813880615.1%29.pdf.

Australian Energy Market Commission. 2020. “2020 Retail Energy Competition Review.”June. https://www.aemc.gov.au/sites/default/files/documents/2020_retail_energy_competition_review_-_final_report.pdf.

Australian Energy Market Operator. 2021. “Energy Explainer: Big Batteries.”https://aemo.com.au/-/media/files/learn/energy-101/batteries-explainer.pdf.

Australian Energy Market Operator. 2022. “AEMO Suspends NEM Wholesale Market.”https://aemo.com.au/newsroom/media-release/aemo-suspends-nem-wholesale-market.

Australian Energy Regulator. 2020. “State of the Energy Market 2020.”June. https://www.aer.gov.au/system/files/State%20of%20the%20energy%20market%202020%20-%20Full%20report%20A4_0.pdf.

Australian Renewable Energy Agency. 2024. “Solar Energy.”https://arena.gov.au/renewable-energy/solar/.

10.

Behrman

J. R.

Mitchell

O. S.

Soo

C. K.

Bravo

2012. “How Financial Literacy Affects Household Wealth Accumulation.” American Economic Review 102 (3): 300–304. doi:10.1257/aer.102.3.300.

11.

Ben-Akiva

McFadden

Train

Walker

Bhat

Bierlaire

, et al. 2002. “Hybrid Choice Models: Progress and Challenges.” Marketing Letters 13 (3): 163–75. doi:10.1023/A:1020254301302.

12.

Bergman

Eyre

2011. “What Role for Microgeneration in a Shift to a Low Carbon Domestic Energy Sector in the UK?” Energy Efficiency 4: 335–53. doi:10.1007/s12053-011-9107-9.

13.

Bierlaire

2016. “Python Biogeme: A Short Introduction.” Report TRANSP-OR 160706, Series on Biogeme, Transport and Mobility Laboratory, School of Architecture, Civil and Environmental Engineering, Ecole Polytechnique Fédérale de Lausanne, Switzerland.

14.

Blasch

Boogen

Daminato

Filippini

2021. “Empower the Consumer! Energy-related Financial Literacy and its Implications for Economic Decision Making.” Economics of Energy & Environmental Policy 10 (2): 149–80. doi:10.5547/2160-5890.10.2.jbla.

15.

Bliemer

Collins

2016. “On Determining Priors for the Generation of Efficient Stated Choice Experimental Designs.” Journal of Choice Modelling 21: 10–14. doi:10.1016/j.jocm.2016.03.001.

16.

Bliemer

Rose

J. M.

2013. “Confidence Intervals of Willingness-to-Pay for Random Coefficient Logit Models.” Transportation Research Part B: Methodological 58: 199–214. doi:10.1016/j.trb.2013.09.010.

17.

Bolton

Stewart

2024. “Domestic Energy Prices.” Research Briefing UK House of Commons Library. https://researchbriefings.files.parliament.uk/documents/CBP-9491/CBP-9491.pdf.

18.

Brennan

Van Rensburg

T. M.

2016. “Wind Farm Externalities and Public Preferences for Community Consultation in Ireland: A Discrete Choice Experiments Approach.”Energy Policy 94: 355–65. doi:10.1016/j.enpol.2016.04.031.

19.

Brent

Ward

2018. “Energy Efficiency and Financial Literacy.”Journal of Environmental Economics and Management 90: 181–216. doi:10.1016/j.jeem.2018.05.004.

20.

Broberg

Persson

2016. “Is Our Everyday Comfort for Sale? Preferences for Demand Management on the Electricity Market.”Energy Economics 54: 24–32. doi:10.2139/ssrn.2615874.

21.

Bucher-Koenen

Lusardi

Alessie

van Rooij

2017. “How Financially Literate are Women? An Overview and New Insights.”The Journal of Consumer Affairs 51 (2): 255–83. doi:10.1111/joca.12121.

22.

Budziński

Czajkowski

2022. “Endogeneity and Measurement Bias of the Indicator Variables in Hybrid Choice Models: A Monte Carlo Investigation.”Environmental & Resource Economics 83: 605–629. doi:10.1007/s10640-022-00702-0.

23.

Carlsson

Martinsson

2008. “Does it Matter When a Power Outage Occurs? — A Choice Experiment Study on the Willingness to Pay to Avoid Power Outages.”Energy Economics 30 (3): 1232–45. doi:10.1016/j.eneco.2007.04.001.

24.

ChoiceMetrics. 2018. Ngene 1.2 User Manual & Reference Guide.

25.

Christelis

Jappelli

Padula

2010. “Cognitive Abilities and Portfolio Choice.”European Economic Review 54 (1): 18–38. doi:10.1016/j.euroecorev.2009.04.001.

26.

Clark

Lusardi

Mitchell

O. S.

2015. “Financial Knowledge and 401(k) Investment Performance: A Case Study.”Journal of Pension Economics & Finance 16 (3): 324–47. doi:10.1017/S1474747215000384.

27.

Clean Energy Council. 2022. “Clean Energy Australia Report 2022.”https://apo.org.au/sites/default/files/resource-files/2022-04/apo-nid317318.pdf.

28.

Cramer

J. C.

Miller

Craig

Hackett

B. M.

Dietz

T. M.

Vine

E. L.

Levine

M. D.

Kowalczyk

D. J.

1985. “Social and Engineering Determinants and Their Equity Implications in Residential Electricity Use.”Energy 10 (12): 1283–91. doi:10.1016/0360-5442(85)90139-2.

29.

Csereklyei

Anantharama

Kallies

2021. “Electricity Market Transitions in Australia: Evidence Using Model-Based Clustering.”Energy Economics 103: 105590. doi:10.1016/j.eneco.2021.105590.

30.

Daly

A. J.

Hess

Train

2012. “Assuring Finite Moments for Willingness to Pay in Random Coefficients Models.”Transportation 39 (1): 19–31. doi:10.1007/s11116-011-9331-3.

31.

Department of the Environment and Energy. 2019. “Australian Energy Update 2019.”September. https://www.energy.gov.au/sites/default/files/australian_energy_statistics_2019_energy_update_report_september.pdf.

32.

Edenhofer

Hirth

Knopf

Pahle

Schlömer

Schmid

Ueckerdt

2013. “On the Economics of Renewable Energy Sources.”Energy Economics 40 (Supplement 1): S12–23. doi:10.1016/j.eneco.2013.09.015.

33.

Essential Services Commission. 2016. “A Practical Application of the PREMO Framework.”https://www.esc.vic.gov.au/sites/default/files/documents/KPMG-A-practical-application-of-the-PREMO-framework.pdf.

34.

Figueiredo

N. C.

da Silva

P. P.

2019. “The “Merit-Order Effect” of Wind and Solar Power: Volatility and Determinants.”Renewable and Sustainable Energy Reviews 102: 54–62. doi:10.1016/j.rser.2018.11.042.

35.

Filippini

Hirl

Masiero

2018. “Habits and Rational Behaviour in Residential Electricity Demand.”Resource and Energy Economics 52: 137–52. doi:10.1016/j.reseneeco.2018.01.002.

36.

Filippini

Kumar

Srinivasan

2020. “Energy-Related Financial Literacy and Bounded Rationality in Appliance Replacement Attitudes: Evidence from Nepal.”Environment and Development Economics 25 (4): 399–422. doi:10.1017/S1355770X20000078.

37.

Finke

M. S.

Howe

J. S.

Huston

S. J.

2017. “Old Age and the Decline in Financial Literacy.”Management Science 63 (1): 213–30. doi:10.1287/mnsc.2015.2293.

38.

Gamble

Juliusson

E. A.

Gärling

2009. “Consumer Attitudes towards Switching Supplier in Three Deregulated Markets.”The Journal of Socio-Economics 38 (5): 814–19. doi:10.1016/j.socec.2009.05.002.

39.

Gathergood

2012. “Self-Control, Financial Literacy and Consumer Over-Indebtedness.”Journal of Economic Psychology 33 (3): 590–602. doi:10.1016/j.joep.2011.11.006.

40.

Gerpott

T. J.

Paukert

2013. “Determinants of Willingness to Pay for Smart Meters: An Empirical Analysis of Household Customers in Germany.”Energy Policy 61: 483–95. doi:10.1016/j.enpol.2013.06.012.

41.

Gillingham

Palmer

2014. “Bridging the Energy Efficiency Gap: Policy Insights from Economic Theory and Empirical Evidence.”Review of Environmental Economics and Policy 8 (1): 18–38. doi:10.1093/reep/ret021.

42.

Giraudet

L.G.

2020. “Energy Efficiency as a Credence Good: A Review of Informational Barriers to Energy Savings in the Building Sector.”Energy Economics 87: 104698. doi:10.1016/j.eneco.2020.104698.

43.

Gleerup

Larsen

Leth-Petersen

Togeby

2010. “The Effect of Feedback by Text Message (SMS) and Email on Household Electricity Consumption: Experimental Evidence.”The Energy Journal 31 (3): 113–32. doi:10.5547/ISSN0195-6574-EJ-Vol31-No3-6.

44.

Global Infrastructure Hub. 2019. “Showcase Project: Hornsdale Power Reserve Project.”September. https://cdn.gihub.org/umbraco/media/2765/gih_showcaseprojects_hornsdale_2019_web_art.pdf.

45.

Goett

Hudson

Train

2000. “Customers’ Choice among Retail Energy Suppliers: The Willingness-to-Pay for Service Attributes.”The Energy Journal 21 (4): 1–28. https://www.jstor.org/stable/41322898.

46.

Gouveia

J. P.

Fortes

Seixas

2012. “Projections of Energy Services Demand for Residential Buildings: Insights from a Bottom-up Methodology.”Energy 47 (1): 430–42. doi:10.1016/j.energy.2012.09.042.

47.

Hanser

Lueken

Gorman

Mashal

, and The Brattle Group. 2017. “The Practicality of Distributed PV-Battery Systems to Reduce Household Grid Reliance.”Utilities Policy 46: 22–32. doi:10.1016/j.jup.2017.03.004.

48.

Hensher

Rose

Greene

2015. Applied Choice Analysis. 2nd ed. Cambridge: Cambridge University Press.

49.

Herriges

J. A.

Phaneuf

2002. “Inducing Patterns of Correlation and Substitution in Repeated Logit Models of Recreation Demand.”American Journal of Agricultural Economics 84 (4): 1076–90. doi:10.1111/1467-8276.00055.

50.

Hess

Train

K. E.

Polak

J. W.

2006. “On the Use of a Modified Latin Hypercube Sampling (MLHS) Method in the Estimation of a Mixed Logit Model for Vehicle Choice.”Transportation Research Part B: Methodological 40 (2): 147–63. doi:10.1016/j.trb.2004.10.005.

51.

Hortaçsu

Madanizadeh

S. A.

Puller

S. L.

2017. “Power to Choose? An Analysis of Consumer Inertia in the Residential Electricity Market.”American Economic Journal: Economic Policy 9 (4): 192–226. doi:10.1257/pol.20150235.

52.

Hung

Parker

A. M.

Yoong

2009. “Defining and Measuring Financial Literacy.” RAND Working Paper Series WR-708. https://www.rand.org/content/dam/rand/pubs/working_papers/2009/RAND_WR708.pdf.

53.

Ivanova

2012. “Are consumers’ Willing to Pay Extra for the Electricity from Renewable Energy Sources? An Example of Queensland, Australia.”International Journal of Renewable Energy Research 2 (4): 758–66. https://www.ijrer.org/ijrer/index.php/ijrer/article/view/387/pdf.

54.

International Energy Agency. 2022. “Electricity Market Report, January 2022.”https://iea.blob.core.windows.net/assets/d75d928b-9448-4c9b-b13d-6a92145af5a3/ElectricityMarketReport_January2022.pdf.

55.

International Renewable Energy Agency. 2017. “Electricity Storage and Renewables: Costs and Markets to 2030.”https://www.irena.org/-/media/Files/IRENA/Agency/Publication/2017/Oct/IRENA_Electricity_Storage_Costs_2017.pdf.

56.

Jaffe

A. B.

Stavins

R. N.

1994. “The Energy-Efficiency Gap: What Does it Mean?” Energy Policy 22 (10): 804–810. doi:10.1016/0301-4215(94)90138-4.

57.

Jappelli

2010. “Economic Literacy: An International Comparison.”The Economic Journal 120 (548): F429–51. doi:10.1111/j.1468-0297.2010.02397.x.

58.

Kallbekken

Sælen

Hermansen

E. A. T.

2013. “Bridging the Energy Efficiency Gap: A Field Experiment on Lifetime Energy Costs and Household Appliances.”Journal of Consumer Policy 36 (1): 1–16. doi:10.1007/s10603-012-9211-z.

59.

Kaufmann

Künzel

Loock

2013. “Customer Value of Smart Metering: Explorative Evidence from a Choice-Based Conjoint Study in Switzerland.”Energy Policy 53: 229–39. doi:10.1016/j.enpol.2012.10.072.

60.

Kavousian

Rajagopal

Fischer

2013. “Determinants of Residential Electricity Consumption: Using Smart Meter Data to Examine the Effect of Climate, Building Characteristics, Appliance Stock, and Occupants’ Behavior.”Energy 55: 184–94. doi:10.1016/j.energy.2013.03.086.

61.

Khosroshahi

Crase

Cooper

Burton

2021. “Matching Customers’ Preferences for Tariff Reform with Managers’ Appetite for Change: The Case of Volumetric-Only Tariffs in Australia.”The Australian Journal of Agricultural and Resource Economics 65 (2): 449–71. doi:10.1111/1467-8489.12417.

62.

Klapper

Lusardi

van Oudheusden

2015. “Financial Literacy around the World: Insights from the Standard & Poors Ratings Services Global Financial Literacy Survey.”https://gflec.org/wp-content/uploads/2015/11/Finlit_paper_16_F2_singles.pdf.

63.

Kubli

Loock

Wüstenhagen

2018. “The Flexible Prosumer: Measuring the Willingness to Co-Create Distributed Flexibility.”Energy Policy 114: 540–48. doi:10.1016/j.enpol.2017.12.044.

64.

Léautier

T. O.

2014. “Is Mandating “Smart Meters” Smart?”The Energy Journal 35 (4): 135–38. doi:10.5547/01956574.35.4.6.

65.

Louviere

J. J.

Bateman

Thorp

Eckert

2016. “Developing New Financial Literacy Measures to Better Link Financial Capability to Outcomes.”July, 3–58. Sydney, Australia: Financial Literacy Australia. https://opus.lib.uts.edu.au/rest/bitstreams/ad2f42b7-3b82-48f2-8aa2-2eaed82fc663/retrieve.

66.

Lührmann

Serra-Garcia

Winter

2018. “The Impact of Financial Education on Adolescents’ Intertemporal Choices.”American Economic Journal: Economic Policy 10 (3): 309–32. doi:10.1257/pol.20170012.

67.

Lusardi

Mitchell

O. S.

2014. “The Economic Importance of Financial Literacy: Theory and Evidence.”Journal of Economic Literature 52 (1): 5–44. doi:10.1257/jel.52.1.5.

68.

Lusardi

Mitchell

O. S.

Curto

2010. “Financial Literacy among the Young.”The Journal of Consumer Affairs 44 (2): 358–80. doi:10.1111/j.1745-6606.2010.01173.x.

69.

Mauldin

T. A.

Henager

Bowen

C. F.

Cheang

2016. “Barriers and Facilitators to Saving Behavior in Low-to Moderate-Income Households.”Journal of Financial Counseling and Planning 27 (2): 231–51. doi:10.1891/1052-3073.27.2.231.

70.

McConnell

Sandiford

2020. “Impacts of LNG Export and Market Power on Australian Electricity Market Dynamics, 2016–2019.”Current Sustainable/Renewable Energy Reports 7 (4): 176–85. doi:10.1007/s40518-020-00164-2.

71.

McFadden

1974. “Conditional Logit Analysis of Qualitative Choice Behaviour.” In Frontiers in Econometrics, edited by Zarembka

, 105–142. New York, NY: Wiley. https://eml.berkeley.edu/reprints/mcfadden/zarembka.pdf.

72.

Meier

Sprenger

C. D.

2013. “Discounting Financial Literacy: Time Preferences and Participation in Financial Education Programs.”Journal of Economic Behaviour & Organization 95: 159–74. doi:10.1016/j.jebo.2012.02.024.

73.

Nelson

Nolan

Gilmore

2021. “What’s Next for the Renewable Energy Target – Resolving Australia’s Integration of Energy and Climate Change Policy?” The Australian Journal of Agricultural and Resource Economics 66 (1): 136–63. doi:10.1111/1467-8489.12457.

74.

Ozbafli

Jenkins

G. P.

2016. “Estimating the Willingness to Pay for Reliable Electricity Supply: A Choice Experiment Study.”Energy Economics 56: 443–52. doi:10.1016/j.eneco.2016.03.025.

75.

Parag

Sovacool

B. K.

2016. “Electricity Market Design for the Prosumer Era.”Nature Energy 1 (4): 16032. doi:10.1038/nenergy.2016.32.

76.

Pliousis

Andriosopoulos

Doumpos

Galariotis

2019. “A Multicriteria Assessment Approach to the Energy Trilemma.”The Energy Journal 40 (1_suppl): 143–66. doi:10.5547/01956574.40.SI1.apli.

77.

Prisecaru

2022. “The War in Ukraine and the Overhaul of EU Energy Security.”Global Economic Observer 10 (1): 16–25. http://www.globeco.ro/wp-content/uploads/vol/split/vol_10_no_1/geo_2022_vol10_no1_art_002.pdf.

78.

Ratnam

Weller

S. R.

Kellett

C. M.

2015. “Scheduling Residential Battery Storage with Solar PV: Assessing the Benefits of Net Metering.”Applied Energy 155: 881–91. doi:10.1016/j.apenergy.2015.06.061.

79.

Reserve Bank of Australia. 2022. “Statement on Monetary Policy – August 2022, Box A: Recent Developments in Energy Prices.”https://www.rba.gov.au/publications/smp/2022/aug/box-a-recent-developments-in-energy-prices.html.

80.

Richter

Pollitt

2018. “Which Smart Electricity Service Contracts Will Consumers Accept? The Demand for Compensation in a Platform Market.”Energy Economics 72: 436–50. doi:10.1016/j.eneco.2018.04.004.

81.

Roe

Teisl

M. F.

Levy

Russell

2001. “US Consumers’ Willingness to Pay for Green Electricity.”Energy Policy 29 (11): 917–25. doi:10.1016/S0301-4215(01)00006-4.

82.

Rose

J. M.

Zhang

2017. “An Introductory Guide to Estimating Discrete Choice Models Using Python Biogeme.” BIDA Working Paper 17-01. Business Intelligence and Data Analytics Research Centre, University of Technology Sydney.

83.

Ruokamo

Kopsakangas-Savolainen

Meriläinen

Svento

2019. “Towards Flexible Energy Demand–Preferences for Dynamic Contracts, Services and Emissions Reductions.”Energy Economics 84: 104522. doi:10.1016/j.eneco.2019.104522.

84.

Scarpa

Willis

K. G.

Acutt

2007. “Valuing Externalities from Water Supply: Status Quo, Choice Complexity and Individual Random Effects in Panel Kernel Logit Analysis of Choice Experiments.”Journal of Environmental Planning and Management 50 (4): 449–66. doi:10.1080/09640560701401861.

85.

Sensfuß

Ragwitz

Genoese

2008. “The Merit-Order Effect: A Detailed Analysis of the Price Effect of Renewable Electricity Generation on Spot Market Prices in Germany.”Energy Policy 36 (8): 3086–94. doi:10.1016/j.enpol.2008.03.035.

86.

Simshauser

2018. “Garbage Can Theory and Australia’s National Electricity Market: Decarbonisation in a Hostile Policy Environment.”Energy Policy 120: 697–713. doi:10.1016/j.enpol.2018.05.068.

87.

Sioshansi

2010. “Welfare Impacts of Electricity Storage and the Implications of Ownership Structure.”The Energy Journal 31 (2): 173–98. https://www.jstor.org/stable/41323286.

88.

Sonali

2017. “South Australia Taps New Gas Plant, Batteries to Fix Power Crisis.”https://www.reuters.com/article/australia-power-tesla-idUSL3N1GQ32M.

89.

Song

Zhou

Lai

K. K.

2017. “Measuring National Energy Performance via Energy Trilemma Index: A Stochastic Multicriteria Acceptability Analysis.”Energy Economics 66: 313–19. doi:10.1016/j.eneco.2017.07.004.

90.

Stolper

O. A.

Walter

2017. “Financial Literacy, Financial Advice, and Financial Behavior.”Journal of Business Economics 87: 581–643. doi:10.1007/s11573-017-0853-9.

91.

Strbac

2008. “Demand Side Management: Benefits and Challenges.”Energy Policy 36 (12): 4419–26. doi:10.1016/j.enpol.2008.09.030.

92.

Sundt

Rehdanz

2015. “Consumers’ Willingness to Pay for Green Electricity: A Meta-Analysis of the Literature.”Energy Economics 51: 1–8. doi:10.1016/j.eneco.2015.06.005.

93.

Sutter

Kocher

M. G.

Glätzle-Rützler

Trautmann

S. T.

2013. “Impatience and Uncertainty: Experimental Decisions Predict Adolescents’ Field Behavior.”American Economic Review 103 (1): 510–31. doi:10.1257/aer.103.1.510.

94.

Tjørring

Jensen

C. L.

Hansen

L. G.

Andersen

L. M.

2018. “Increasing the Flexibility of Electricity Consumption in Private Households: Does Gender Matter?” Energy Policy 118: 9–18. doi:10.1016/j.enpol.2018.03.006.

95.

Tocock

Hatton MacDonald

Rose

J. M.

2024. “Risk Preferences, Bill Increases and the Future Reliability of Electricity Networks in Australia.”Energy Research & Social Science 118: 103763. doi:10.1016/j.erss.2024.103763.

96.

Tocock

Tinch

Hatton MacDonald

Rose

J. M.

2023. “Managing the Energy Trilemma of Reliability, Affordability and Renewables: Assessing Consumer Demands with Discrete Choice Experiments.”The Australian Journal of Agricultural and Resource Economics 67 (2): 155–75. doi:10.1111/1467-8489.12509.

97.

Train

K. E.

Weeks

2005. “Discrete Choice Models in Preference Space and Willingness-to-Pay Space.” In Applications of Simulation Methods in Environmental and Resource Economics, The Economics of Non-Market Goods and Resources, edited by Scarpa

Alberini

, 1–16. Dordrecht, Netherlands: Springer. doi:10.1007/1-4020-3684-1_1.

98.

Vij

Walker

J. L.

2016. “How, When and Why Integrated Choice and Latent Variable Models are Latently Useful.”Transportation Research Part B: Methodological 90: 192–217. doi:10.1016/j.trb.2016.04.021.

99.

von Loessl

2023. “Smart Meter-Related Data Privacy Concerns and Dynamic Electricity Tariffs: Evidence from a Stated Choice Experiment.”Energy Policy 180: 113645. doi:10.1016/j.enpol.2023.113645.

Financial Literacy and Electricity Contracts

Abstract

Keywords

Introduction

Background

Methods

Survey Design

Survey Sample and Experimental Design

Measurement of Financial Literacy

Model Specifications

Results and Discussion

Conclusions and Future Research

Footnotes

Appendix

Acknowledgements

Declaration of Conflicting Interests

Funding

ORCID iD

1

2

3

References