Does Government Environmental Expenditure Reduce Residential Energy Consumption in Canada? Evidence from Provincial Panel Data

Abstract

The residential sector has the potential to significantly impact overall energy consumption and emissions due to the long lifespan of residential buildings and the difficulty of outsourcing residential emissions to other nations. This research primarily aimed to investigate the impact of government expenditure on environmental protection in reducing household energy consumption, thus contributing to Canada’s environmental quality. We utilized panel data from ten Canadian provinces covering the period from 1995 to 2020. To estimate household energy demand, both conceptual and empirical models were developed. Advanced second-generation econometric techniques were applied, including appropriate unit root and co-integration tests. Long-term relationships were analyzed using the fully modified ordinary least squares (FMOLS) and dynamic ordinary least squares (DOLS) methodologies. The long-term estimates indicate that all explanatory variables align with theoretical expectations and are highly significant. The findings reveal that government funding for environmental protection, along with province-specific environmental policies, contributes to reducing per capita residential energy consumption, which is essential for enhancing energy efficiency. Additionally, factors such as weather, income, demographic shifts, and energy prices are found to influence household energy consumption trends. These outcomes highlight the importance of increased government funding for environmental protection and underscore the need for provinces to implement context-specific environmental policies. This approach is essential in the effort to reduce household energy consumption across Canada.

1. Introduction

Escalating global energy consumption has heightened the strain on environmental pollution and global warming. The Paris Climate Conference (COP21) agreement aspired to curtail the rise in global average temperature to below 2 °C. To attain this objective, a reduction in energy consumption and greenhouse gas emissions is imperative across all economic sectors. The residential energy sector plays a pivotal role in affecting emission reductions, being responsible for 25% of worldwide energy consumption and 17% of global CO₂ emissions [1]. Given factors like population growth and heightened demand for indoor comfort [2], this sector holds significant potential to instigate environmental change in response to climate issues. Implementing measures to curtail residential energy consumption can yield noteworthy impacts on both overall energy consumption and the trajectory of global climate change. Notably, the enduring lifespan of residential buildings and the intricate challenges in transferring residential emissions to other nations underscore this sector’s importance [3], rendering residential energy policies potentially more efficacious than those targeting other sectors.

Much like other sectors, energy consumption within the residential domain carries negative externalities in the form of environmental pollution, which the market cannot rectify. As such, government intervention becomes imperative. Typically, three measures are adopted to address environmental pollution [4]. Firstly, governments may impose taxes on energy consumers to discourage consumption. Secondly, governments can establish appropriate property rights and oversee negotiations between polluters and pollution victims, aiming to reach optimal solutions. Thirdly, increased government environmental expenditure can incentivize and guide households toward environmentally friendly energy-use decisions.

Over time, governments worldwide have progressively allocated a significant portion of their national income to diverse activities, including environmental protection. Total government spending ranged from 25% to 59% of Gross Domestic Product (GDP) across OECD countries in 2021 [5]. This spending capacity influences numerous aspects of the economy. Shifting government spending toward public goods can mitigate pollution, yet the impact of increased total government spending remains equivocal [6]. Expenditure on environmental protection emerges as a vital tool for local governments to enhance environmental quality [4]. Despite the considerable economic ramifications of fiscal spending levels and composition, the link between fiscal spending policies on environmental protection and residential energy usage remains notably unexplored in the literature.

Governments can foster an environment conducive to energy efficiency and sustainability through policies, programs, and initiatives. Firstly, they can mandate building codes, standards, and appliance labeling to contribute to lowered residential energy consumption. Secondly, governments can allocate resources for the development and enhancement of energy-efficient infrastructure, including investments in renewable energy systems, smart grids, and energy-efficient buildings. Such measures reduce energy usage within residential areas while optimizing energy distribution. Thirdly, governments can provide financial incentives like tax credits, rebates, and subsidies to encourage the adoption of energy-efficient technologies and practices. These incentives offset initial costs, rendering energy-saving measures more appealing and accessible to homeowners. Lastly, government spending can support educational programs and awareness campaigns, further influencing residential energy consumption.

Canada has launched various programs and policies to enhance environmental protection. Major initiatives include the Low Carbon Economy Fund, which finances energy efficiency projects; the Nature Smart Climate Solutions Fund, a ten-year, CAD 1.4 billion program for ecosystem restoration; and the EcoAction Community Funding Program, which supports local environmental projects. Other initiatives, such as the Environmental Damages Fund, the Canada Nature Fund, the Canada Greener Homes Initiative, and the Canada Water Agency, further reflect Canada’s strong environmental commitment.

Household energy consumption in Canada has shown marked improvement over time. The average energy consumption of a Canadian home fell from 144 gigajoules in 1995 to 90.5 gigajoules in 2019 [7]. This progress can be attributed to the introduction of energy-efficient appliances and heating systems, mandated by building standards and labeling requirements. However, substantial disparities in household energy consumption persist across the provinces. For example, households in Alberta (124.6 gigajoules), Saskatchewan (120.5 gigajoules), Manitoba (115.7 gigajoules), and Ontario (94.8 gigajoules) report higher-than-average consumption, while households in other provinces exhibit lower-than-average levels.

A portion of government spending in Canada is dedicated to improving environmental quality, which is classified under the functions of government. This spending varies across provinces and years. Given the household sector’s significant role in overall energy consumption, evaluating the effectiveness of such expenditures and their impact on residential energy use is essential. Furthermore, some provinces have introduced explicit environmental policies to reduce greenhouse gas emissions, introducing additional heterogeneity. For instance, British Columbia implemented a carbon tax, Quebec adopted a cap-and-trade system, and Ontario and Nova Scotia pursued the closure of coal and pulp plants.

Given this backdrop, the central research question of this study emerges: do government expenditure on environmental protection and province-specific environmental policies contribute to reducing residential energy consumption in Canada? Accordingly, this study’s objective is to investigate whether government expenditure on environmental protection and explicit environmental policies influence reductions in household energy consumption and contribute to energy efficiency and environmental quality in Canada. Thus, this study extends the existing literature on environmental governance by analyzing the correlation between government expenditure on environmental protection and residential energy consumption. It offers a distinct contribution to the current body of knowledge by highlighting that government environmental expenditure negatively impacts household energy consumption and underscores the role of province-specific environmental policies in reducing residential energy use. This study also compiles a dataset detailing Canada’s government expenditure on environmental protection by province and year, an aspect of potential independent interest.

The subsequent sections of this paper are structured as follows: Section 2 presents the literature review. Section 3 outlines the methodology, encompassing variables, data, the model, and empirical techniques. Descriptive statistics and results are discussed in Section 4. Section 5 offers a comprehensive discussion, while conclusions and policy implications are presented in Section 6.

2. Literature Review

In this review, we meticulously assess three distinct categories within the existing literature. These categories encompass the following: (i) the impact of environmental policy, including financial incentives and energy performance standards, on residential energy consumption; (ii) the dynamics between government expenditure and environmental quality; and (iii) the influential drivers underpinning residential energy consumption.

2.1. Environmental Policy and Residential Energy Consumption

Research on the influence of environmental policies on residential energy consumption has been relatively limited. Several studies—including those by Tsemekidi et al. [8], Bertoldi and Hirl [9], Bertoldi and Mosconi [10], and Horowitz and Bertoldi [11]—have used panel data and diverse indicators to evaluate the impact of EU policies on energy conservation in the residential sector. Employing an econometric energy demand model, Filippini and Hunt [12] examined the effects of US residential energy efficiency policies, while Filippini, Hunt, and Zoric [13] applied a stochastic frontier approach to assess the effectiveness of EU energy efficiency measures in households. Their findings highlight that financial incentives and energy performance standards play crucial roles in driving energy efficiency investments, whereas information-based measures tend to have more limited impacts.

Tsemekidi et al. [8] quantified the EU’s energy savings potential, demonstrating that current policies can meet the 2030 climate targets. They also investigated the effectiveness of EU energy efficiency policies—including eco-design, energy labeling, building codes, and directives—in reducing residential energy consumption. Bertoldi and Hirl [9] and Bertoldi and Mosconi [10] highlight the importance of policy flexibility, stability, and robust monitoring for the successful implementation of EEOSs. Horowitz and Bertoldi [11] reinforce the economic case, showing that energy efficiency investments are more cost-effective than supply-side solutions. Collectively, these studies underscore the significant role of energy efficiency measures, particularly Energy Efficiency Obligation Schemes (EEOSs), in achieving substantial energy savings and advancing the EU’s climate objectives.

Further studies have explored the impact of specific policies on residential energy use. For example, Aydin and Brounen [14] assessed the influence of mandatory energy efficiency labels for household appliances and building standards, concluding that both measures contributed to lower residential energy consumption. Broin et al. [15] analyzed EU space heating energy efficiency policies by categorizing them into financial, regulatory, and informative measures, highlighting that regulatory measures were particularly effective.

A body of research highlights the role of fiscal and financial incentives in promoting energy efficiency (EE) and green building adoption. Simpeh and Smallwood [16] present a framework for South Africa that integrates financial incentives with recognition and support to align stakeholder interests. Blommestein and Daim [17] emphasize that while financial savings drive consumer adoption of energy-efficient devices, environmental awareness and product performance are also influential. In Portugal, Koengkan et al. [18] find that higher income does not necessarily lead to better energy performance, but access to consumer credit and fiscal incentives like tax rebates significantly boost the adoption of high-efficiency homes. On a broader scale, Sloot and Scheibehenne [19] show that financial incentives modestly reduce total energy use (−1.83%) but substantially cut peak electricity demand (−10%), particularly when paired with enabling technologies like smart meters. Studies in South Africa reveal that existing financial incentives often fail to effectively stimulate energy efficiency. A study of major listed businesses showed that although tax incentives are available for energy efficiency and renewable energy investments, they were not perceived as sufficiently motivating; non-tax factors were more influential in decision-making, and compliance burdens were seen as deterrents, according to Dippenaar [20].

2.2. Government Expenditure and Environmental Quality

Government intervention to combat environmental pollution typically involves imposing taxes on energy use, establishing property rights to facilitate negotiations, and increasing public spending to promote environmentally friendly behavior [4]. Studies indicate that government spending improves environmental quality in Northern African countries but has a negative effect in Southern African countries [21]. In China, environmental investments in one province positively influenced neighboring provinces, demonstrating a spillover effect [22]. Furthermore, the composition of government spending plays a crucial role, with environmentally focused spending leading to better environmental outcomes [23].

However, a unanimous consensus on the efficacy of government expenditure in enhancing environmental quality has yet to emerge. For instance, Donkor et al. [21] reported a positive link between government expenditure and environmental quality in Northern African countries, contrasted with a negative relationship in Southern African countries. Bernauer and Koubi [24] found a positive association between government expenditure and sulfur dioxide (SO₂) emissions across 42 countries, suggesting that spending may not always lead to pollution reductions. Halkos and Paizanos [25] observed the significant but varying impacts of government expenditure on SO₂ and CO₂ emissions, contingent on countries’ income levels. In contrast, Lopez and Palacios [26] identified a substantial negative correlation between overall government expenditure (and public goods spending) and SO₂ and ozone emissions in wealthier European nations. Finally, Islam and Lopez [27] demonstrated that centralized government spending reduced air pollution across 50 US states, while state-level spending had minimal effects.

Environmental expenditure theoretically yields positive outcomes for environmental quality; however, empirical findings have presented a nuanced picture. Lopez et al. [23] identified a negative link between public goods expenditure and water pollution. Lin et al. [22] unearthed a connection between substantial environmental expenditures and lower pollution levels in highly technologically advanced regions in China, highlighting the spillover effect of such investments. Galinato and Galinato [28] established a tangible link between public goods spending, deforestation, and CO₂ emissions in 12 countries. In a study focusing on 14 MENA countries, Gholipour and Farzanegan [29] discerned improved air quality with higher environmental expenditures, conditioned on organizational structure and governance quality. Similarly, He et al. [30] discovered a positive correlation between environmental expenditures and air quality indices in China’s most polluted cities. Fan et al. [4] found that local environmental protection expenditure correlated with reduced industrial pollution emissions in various Chinese cities, displaying spillover and heterogeneous effects.

Overall, the impact of general government and environmental expenditure on environmental quality is nuanced and influenced by factors such as region, income levels, specific environmental indicators, and the emphasis on public spending. While some studies have demonstrated the affirmative impact of such expenditures on environmental quality, others reported mixed or inconsequential results, reflecting the intricate relationship between public expenditure and environmental outcomes.

2.3. Drivers of Residential Energy Consumption

A substantial body of the literature has elucidated the determinants of residential energy consumption across diverse countries [8,31,32,33,34,35]. Notably, Belaid [35] explored the effect of dwelling attributes, household characteristics, climate, and behavior in France, finding that dwelling characteristics wielded a potent direct influence on energy consumption. Romàn-Collado and Colinet [31] singled out per capita income as a key driver of residential energy consumption. Borozan [36] confirmed the significance of socio-economic and contextual variables, such as disposable income and climatic conditions, as determinants of energy consumption. Brounen et al. [32] concluded that building characteristics chiefly determined residential gas consumption, while electricity consumption was more closely tied to household composition, particularly income and family size. Lévy and Belaid [34], examining household and building profiles, pinpointed demographic features as exerting a substantial impact on energy consumption. Otsuka [33] estimated Japanese residential energy demand, identifying population density and electrification rates as pivotal factors in residential energy efficiency. Similarly, Reuter et al. [37] emphasized that comfort and behavioral factors in EU households heightened energy consumption, counteracted by enhanced efficiency measures. Notably, socio-economic attributes have demonstrated a moderate effect on per capita energy expenditures, with dwelling characteristics, especially household size, wielding a far more pronounced impact in the UK [38].

In developing nations, various factors shape residential energy consumption, including per capita GDP, population size, and electrification ratios [39,40,41]. Balarama et al. [42], utilizing Bangladesh’s data, found significant disparities in electricity price elasticities. In China, Liu et al. [43] conducted a study across 30 provinces from 1995 to 2012, finding that as incomes increased, residential electricity consumption remained stable, with an income elasticity of approximately one.

Notably, government policies and environmental expenditures often have long-term objectives. Given the household sector’s pivotal role in total energy consumption, it becomes essential to evaluate the efficacy of these policies and their impact on residential energy consumption. This review of the literature underscores the gap in exploring the interplay between government expenditure on environmental protection and residential energy consumption. The present study endeavors to bridge this gap by investigating the relationship between these variables. While the evaluation of this relationship may pose challenges due to the intricate nature of time-series data on government environmental protection expenditures, a compiled panel dataset offers a unique opportunity for this study’s execution.

3. Methodology and Data Source

3.1. Specification of the Variables and Data

Government spending on environmental protection is classified within the Canadian Classifications of Functions of Government as one of ten categories. It is quantified as a percentage share of total government spending, typically directed toward enforcing and overseeing environmental regulations, as well as helping businesses, households, and non-profit sectors in their endeavors to reduce pollution. This allocation seeks to shape stakeholders’ behavior and motivate them toward pro-environmental decisions and actions. Government expenditure on environmental protection is specified as a share of GDP. Alongside this expenditure, certain provinces have implemented specific policies to curtail greenhouse gas emissions, contributing to a landscape of diversity. For instance, British Columbia introduced a carbon tax in 2008, while Quebec adopted a cap-and-trade policy in 2013. Conversely, some provinces have enacted plant closure policies; for instance, Ontario ceased coal-based electricity generation in 2014, and Nova Scotia closed its Minas Basin pulp and power mills in 2012 to enhance environmental conditions. Due to this variability and the limited samples of each policy type, a policy dummy variable is introduced to encompass market-based environmental policies across different provinces. A comprehensive description of each variable, including the policy dummy, is presented in

Table 1. Name, specification, and source of each variable.

Variable	Name	Specification and Source of Each Variable
Per capita residential energy use (in mwh)	pcreu	Per capita residential energy use is estimated by dividing total residential energy use by total population as of July of each year. Statistics Canada Table: 25-10-0029-01 (formerly CANSIM 128-0016) for energy use and Table: 17-10-0005-01 (formerly CANSIM 051-0001) for population (estimated on 1st July of each year).
Per capita GDP	pcgdp	Per capita GDP is calculated by dividing GDP by population as of July 01 in CAD at the 2012 price.
Share of government expenditure on environmental protection	sgexp	Estimated as a percentage of provincial GDP. Statistics Canada Table: 10-10-0005-01 (formerly CANSIM 385-0041) for 2008 to 2020; data before 2008 were collected from Statistics Canada Table: 10-10-0041-01 (formerly CANSIM 385-0003).
Share of renewable energy	sren	Share (in %) of renewable energy in total energy use (based on Statistics Canada data); hydro, nuclear, wind, biomass, etc., are considered renewable energy. Statistics Canada Table: 25-10-0030-01.
Share of population age 65 years and above	Spop65	Share (in %) of the population age 65 years and above in total population. Statistics Canada Table: 17-10-0005-01 (formerly CANSIM 051-0001).
Heating-degree days	hdd	Number of heating-degree days in the year based on the monthly average of all weather stations in the province. Environment Canada Monthly Climate Summaries: https://climate.weather.gc.ca/prods_servs/cdn_climate_summary_e.html (accessed on 1 June 2022).
Cooling-degree days	cdd	Number of cooling-degree days in the year based on the monthly average of all weather stations in a province. Environment Canada Monthly Climate Summaries: https://climate.weather.gc.ca/prods_servs/cdn_climate_summary_e.html (accessed on 1 June 2022)
Price of electricity	pelect	Average electricity price index (in 2002 base). Statistics Canada (price of electricity is used as a proxy of energy price) Table: 18-10-0005-01 (formerly CANSIM 326-0021).
Policy dummy	pd	The value is one if the year is greater than 2007 for BC, if the year is greater than 2013 for Ontario, if the year is greater than 2012 for Quebec, or if the year is greater than 2012 for Nova Scotia; otherwise, it is zero.

.

This study harnesses panel data spanning ten provinces of Canada from 1995 to 2020, amassing a total of 260 observations. The data are derived from reputable sources, including Statistics Canada and Environment Canada, with specific references outlined in Table 1.

3.2. The Model

In a conceptual context, the demand for household energy consumption can be derived by optimizing the household utility function subject to budget constraints. However, the interplay of household and dwelling attributes, weather patterns, household behavior, and environmental attitudes also impact household energy consumption [44]. Consequently, factors shaping household energy consumption can be segmented into two categories based on source: (i) internal and (ii) external determinants. Internal factors encompass elements such as energy prices, income, behavior, and motivation, as well as household and dwelling specifics. Conversely, external factors incorporate weather conditions and fiscal and regulatory environmental policies, which significantly influence their energy-related attitudes and behaviors. The following model is proposed to estimate Canada’s household energy consumption, utilizing panel data across provinces:

pcreu_it = f(sgexp_it, pelect_it, pcgdp_it, spop65_it, hdd_it, cdd_it, pd_it)

(1)

where

pcreu_it is the per capita residential energy use for province i in year t;

sgexp_it is the share of governmental expenditure on environmental protection in total government expenditure for province i in year t;

pelect_it is the price per unit of energy in the province in year t;

pcgdp_it is the per capita real income for province i in year t;

spop65_it is the share of the population age 65 and above in the total population for province i in year t;

hdd_it is the number of heating-degree days for province i in year t;

cdd_it is the number of cooling-degree days for province i in year t; and

pd_it is a policy dummy for the presence or absence of an explicit environmental policy for province i in year t.

With log transformation, the model can be written as follows without subscripts:

$$lpcreu={\beta }_0+{\beta }_{1}Pd+{\beta }_{2}sgexp+{\beta }_{3}pelect+{\beta }_{4}lpcgdp+{\beta }_{5}spop65+{\beta }_{6}lhdd+{\beta }_{7}lcdd+\epsilon$$

(2)

Note that sgexp, pelect, and spop65 are expressed as percentages and, therefore, do not require a logarithmic transformation. Additionally, pd is a discrete variable taking a value of 0 or 1 and does not need to be logged.

An increase in income and/or heating- to cooling-degree days is expected to elevate residential energy consumption. Conversely, energy prices, government expenditure on environmental protection, and the explicit environmental policy dummy are anticipated to exert a negative influence, promoting energy efficiency. Notably, household and dwelling attributes are excluded due to the paucity of annual time-series data by province. The impact of changes in the elderly population on household energy consumption can be mixed. While more time at home can drive up energy use, budget constraints and environmental awareness can offset or amplify these effects.

3.3. Empirical Techniques

Given the utilization of panel time-series data, employing appropriate panel techniques is crucial to discerning the relationship between residential energy consumption and its determinants, encompassing government expenditure on environmental protection and explicit environmental policies.

3.3.1. Unit Root Test

To assess the stationary nature of variables, conducting panel unit root tests is imperative. Detecting cross-sectional dependence is the preliminary step, as the validity of first-generation unit root tests hinges on addressing this aspect [45,46]. Pesaran’s [45] test, evaluating cross-sectional dependence, is adopted in this study to ascertain its presence. The null hypothesis for this test is the absence of cross-sectional dependence. If cross-sectional dependence is detected in the series, the second-generation unit root test, specifically, the CIPS test by Pesaran [47], is preferred due to its enhanced robustness under cross-sectional dependence and heterogeneity conditions.

3.3.2. Co-Integration Test

This section identifies long-term co-integration using both first- and second-generation tests, following the analysis of variable stationarity. While first-generation co-integration tests like the Kao, Pedroni, and Johansen tests can be unreliable in the presence of cross-sectional dependence, the second-generation test employs bootstrapping co-integration, emerging as a robust alternative. Both first- and second-generation co-integration tests are applied using the Pedroni [48] and Westerlund and Edgerton [49] bootstrapping methods. Notably, these tests necessitate the stationarity of all variables at their first difference, I (1).

3.3.3. Long-Term Estimation

If the variables being examined exhibit non-stationarity in their level form but demonstrate stationarity upon first differencing and display a co-integration relationship, the long-run relationship can be effectively estimated using the fully modified ordinary least squares (FMOLS) technique. The FMOLS method was originally introduced by Phillips and Hansen [50], with further refinement by Pedroni [51]. Given its precision and the ability to address issues of endogeneity and autocorrelation, this method has been selected for our analysis. Moreover, to ensure the robustness of our findings, this model is also estimated using the dynamic ordinary least squares (DOLS) method.

4. Results

4.1. Descriptive Statistics

Table 2. Descriptive statistics of the variables used in the study.

	Obj.	Mean	Std. Dev.	Min.	Max.
pcreu (mwh)	260	10.99	1.98	6.28	28.85
pd	260	0.15	0.36	0	1
sgexp (%)	260	0.51	0.16	0.19	85.76
pelect (2002 = 100)	260	119.56	26.26	65.9	191
pcgdp (constant $)	260	49,810	13,710	28,992	85,756
spop65 (%)	260	14.64	2.61	9.62	22.29
hdd	260	3808.24	547.19	2628.43	5281.89
cdd	260	107.86	57.24	12.79	314.26

presents an overview of the descriptive statistics for all variables included in the analysis. Ahead of regression, it is prudent to assess the interplay between variables and the variance inflation factor (VIF) for each regressor, which has been duly computed.

Table 3. Pairwise correlation coefficients between variables used in the study.

	pcreu	pd	sgexp	pelect	pcgdp	spop65	hdd	cdd
pcreu	1
pd	−0.29 (0.00)	1
sgexp	−0.27 (0.00)	0.22 (0.00)	1
pelect	−0.30 (0.00)	0.42 (0.00)	0.12 (0.05)	1
pcgdp	0.64 (0.00)	−0.06 (0.34)	−0.32 (0.00)	0.20 (0.00)	1
spop65	−0.53 (0.00)	0.46 (0.00)	0.13 (0.03)	0.74 (0.00)	−0.17 (0.00)	1
hdd	0.56 (0.00)	−0.33 (0.00)	−0.49 (0.00)	−0.17 (0.00)	0.57 (0.00)	−0.25 (0.00)	1
cdd	−0.21 (0.00)	0.14 (0.03)	0.26 (0.00)	0.05 (0.44)	−0.36 (0.00)	0.17 (0.01)	−0.37 (0.00)	1

illustrates the pairwise correlation coefficients, accompanied by their corresponding significance levels in parentheses. In terms of VIF, the computed values span from 1.22 to 3.32, indicating that the issue of multicollinearity among the regressors is not a prominent concern. The detailed VIF outcomes are accessible upon request. Notably, the policy dummy (PD) and the share of government expenditure on environmental protection (sgexp) are pivotal variables in this study, warranting a closer examination of their relationships with per capita government expenditure (pcgexp). The pairwise correlation analysis reveals a noteworthy negative association between these two key variables and per capita government expenditure (as presented in Table 3). This negative trend is further depicted in Figure 1, illustrating the inverse relationship between pcreu and sgexp. An insightful comparison is presented in

Table 4. Differences in the means of per capita residential energy use with and without explicit policy intervention.

Provinces/Periods	Obj	Mean	Std. Dev.
Provinces where there was no explicit envir. policy	156	11.45	2.28
Provinces where there was explicit envir. policy	104	10.31	1.12
Periods for the provinces where there was no explicit envir. policy	65	10.71	1.042
Periods for the provinces where there was explicit envir. policy	39	9.64	0.906

, showcasing the divergence in pcreu with and without explicit environmental policies. Within periods and provinces characterized by explicit environmental policies, the mean per capita residential energy consumption was notably lower at 9.64 mwh, in contrast to 10.71 mwh in instances where there were no explicit environmental policies.

4.2. Results of Unit Root Tests

To commence, we evaluated whether the variable series manifested cross-sectional dependence, employing the Pesaran [45] cross-sectional dependence (CD) test. As presented in

Table 5. Results of cross-sectional dependence and CIPS unit root tests.

Variables	CD Test Statistics	p-Value	CIPS Test Statistics
			Level	1st Difference
lpcreu	8.44	0.000	−2.09	−4.89 ***
sgexp	8.15	0.000	−2.16	−5.55 ***
pelect	30.92	0.000	−2.50	−3.76 ***
lpcgdp	32.42	0.000	−2.18	−3.93 ***
spop65	31.53	0.000	−1.14	−3.70 ***
lhdd	18.14	0.000	−2.19	−5.51 ***
lcdd	13.51	0.000	−2.15	−6.05 ***

Note: (1) For the CD test, H0 is cross-sectional independence, and for CIPS, H0 is homogeneous non-stationarity. (2) *** indicate statistical significance at less than 1%. (3) lpcreu: log of per capita residential energy use; sgexp: share of government expenditure on environmental protection; pelect: electricity price index; lpcgdp: log of per capita GDP; spop65: share of elderly population in total population; lhdd: log of heating-degree days; lcdd: log of cooling-degree days.

, the outcomes indicate cross-sectional dependence, as evidenced by the rejection of the null hypothesis. This discovery necessitated the Pesaran [47] cross-sectional Im, Pesaran, and Shin (CIPS) unit root test. Notably, the CIPS unit root test is advantageous not only in scenarios of cross-sectional dependence but also when dealing with heterogeneity. The test comprises the Levin, Lin, and Chu method; the Im, Pesaran, and Shin method; and the Augmented Dickey–Fuller method. Table 5 enumerates the outcomes of the CIPS test, encompassing both level and first-difference forms. The results reveal that while all variables in their level form exhibit non-stationarity, they attain stationarity upon the first difference (I(I)).

4.3. Results of Co-Integration Tests

Following the unit root tests, the investigation of the long-term relationship among the variables commenced through the initial application of the Pedroni [48] co-integration test. However, it is important to note that the metrics of this test are reliant on within-dimension and group-based method statistics. While this test offers insights, the more precise and dependable examination for panel co-integration emerged by using the Westerlund [52] approach, which incorporates the bootstrap methodology. Consequently, both the Pedroni and Westerlund tests were executed to assess the potential co-integration of variables. The outcomes of both tests are shown in

Table 6. Pedroni and Westerlund panel co-integration test results.

Type of Test	Test Statistics	p-Value	Conclusion
Pedroni
Modified Phillips–Perron t	−2.7597	0.003	Co-integration exists.
Phillips–Perron t	−3.0391	0.001	Co-integration exists.
Augmented Dicky–Fuller t	−3.1162	0.001	Co-integration exists.
Westerlund:
Variance ratio	−1.9825	0.021	Co-integration exists.

. The results of both tests converge, demonstrating that all variables are indeed co-integrated, underscoring the presence of a sustained long-term relationship between the variables.

4.4. Short-Term Estimation

The outcomes of the unit root tests underscore the non-stationary nature of all variables at that level apart from their attainment of stationarity upon first differencing. This lays the groundwork for estimating the regression model in the first difference, allowing us to explore the short-term relationship between per capita residential energy consumption and its determinants. The results of both pooled ordinary least squares (OLS) and panel random effect estimations are exhibited in

Table 7. Results of Pooled OLS and Panel Random Effect (xtre) estimation (dependent variable: d.lpcreu).

Regressors	Pooled OLS Estimation in 1st Diff. Form			Panel Random Effect Estimation (xtre) in 1st Diff. Form
	Coefficient	t-Stats	p-Value	Coefficient	t-Stats	p-Value
pd	−0.011	−1.13	0.259	−0.011	−1.14	0.253
d.sgexp	−0.052	−1.28	0.202	−0.052	−1.42	0.156
d.pelect	−0.0003	−0.53	0.597	−0.0003	−0.55	0.584
d.lpcgdp	0.101	0.90	0.370	0.101	0.81	0.420
d.spop65	−0.042	−1.99	0.048	−0.042	−2.37	0.018
d.lhdd	0.518	11.11	0.000	0.518	11.01	0.000
d.lcdd	0.015	2.15	0.032	−0.015	−1.92	0.055
constant	0.009	1.38	0.167	0.008	1.40	0.161
F-tats or Wald chi2	F (7242) = 22.64 (p-value = 000)			141.95 (p-value = 000)

Note: lpcreu: log of per capita residential energy use; sgexp: share of government expenditure on environmental protection; pelect: electricity price index; lpcgdp: log of per capita GDP; spop65: share of elderly population in total population; lhdd: log of heating-degree days; lcdd: log of cooling-degree days.

. Notably, all variables in these two estimations are presented in the first difference except for the policy dummy variable. While the model was also estimated using the panel fixed-effect method, the Hausman test favored the random effect approach over fixed effects. The outcomes of both estimations align closely, showing expected signs across all variables. Notably, the policy dummy (pd), government expenditure on environmental protection (sgexp), energy price (pelect), and the share of the elderly population (spop65) all demonstrate a negative relationship. Conversely, variables such as weather (hdd and cdd) and income (pcgdp) show a positive relationship. Among these explanatory variables, weather (hdd and cdd) and the share of the elderly population exert a statistically significant influence on residential energy consumption in the short term. However, other variables, such as the policy dummy (pd), government expenditure on environmental protection (sgexp), energy price (pelect), and per capita GDP (pcgdp), do not exert a significant impact on residential energy consumption in the short term.

4.5. Results of Long-Term Estimation

Given the established co-integration between per capita residential energy consumption and its determinants—encompassing the policy dummy, government expenditure on environmental protection, energy price, income, demographic characteristics, and weather—a panel co-integration regression was pursued to scrutinize the long-term relationship. This was achieved through the application of the fully modified ordinary least squares (FMOLS) and dynamic ordinary least squares (DOLS) techniques. The outcomes of the FMOLS and DOLS estimations are congruent and are presented in

Table 8. Results of fully modified OLS (FMOLS) and dynamic OLS (DOLS) regressions (dependent variable: lpcreu).

Regressors	FMOLS Estimation			DOLS Estimation
	Coefficient	t-Stats	p-Value	Coefficient	t-Stats	p-Value
pd	−0.01	−6.23	0.000	−0.01	−5.43	0.000
sgexp	−0.12	−3.52	0.000	−0.11	−2.41	0.000
pelect	−0.001	−17.36	0.000	−0.001	−3.64	0.000
lpcgdp	0.05	2.30	0.000	0.08	1.35	0.001
spop65	−0.06	−39.33	0.000	−0.06	−8.86	0.000
lhdd	0.41	29.76	0.000	0.39	10.52	0.000
lcdd	−0.01	−0.37	0.54	−0.001	−0.14	0.85

Note: lpcreu: log of per capita residential energy use; sgexp: share of government expenditure on environmental protection; pelect: electricity price index; lpcgdp: log of per capita GDP; spop65: share of elderly population in total population; lhdd: log of heating-degree days; lcdd: log of cooling-degree days.

, indicating robust and consistent results.

The coefficients of all explanatory variables align with a priori expectations and exhibit high significance levels, often at less than 1%, except for cooling-degree days. As outlined in Table 8, the policy dummy (pd), government expenditure on environmental protection (sgexp), energy price (pelect), and the share of the elderly population (spop65) exhibit a negative relationship with long-term per capita residential energy consumption (lpcreu). Conversely, per capita GDP (lpcgdd), heating-degree days (lhdd), and cooling-degree days (lcdd) display a positive relationship with per capita residential energy use (lpcreu). Specifically, the coefficient of the policy dummy (pd) stands at −0.01, suggesting that per capita residential energy consumption is approximately 1% lower during years and in provinces where explicit environmental policies were instituted than in periods and regions lacking such policies. Notably, the effect of government expenditure on environmental protection on residential energy consumption is statistically significant. The sgexp coefficient indicates a −0.12, implying that a 1% increase in the share of government expenditure on environmental protection leads to a 0.12% reduction in per capita residential energy use (Table 8, column 2).

Energy price influences per capita residential energy use negatively; however, its coefficient is relatively small, suggesting the highly inelastic price elasticity of household energy demand. Furthermore, residential energy consumption displays a significant correlation with income. A 1% rise in per capita GDP is associated with a 0.05% increase in household energy consumption per capita. Additionally, an increase in the share of the elderly population in the total population negatively impacts household energy use, leading to a 0.06% decline for every 1% rise in the share of the elderly population. The coefficient of heating-degree days (lhdd) is 0.41, indicating that a 1% escalation in heating-degree days corresponds to a 0.41% rise in household energy consumption.

5. Discussion

Our findings shed light on the substantial roles played by government expenditure on environmental protection (sgexp) and province-specific explicit environmental policies in Canada, both of which significantly contribute to the reduction of per capita residential energy consumption in the long run. Notably, a 10% increase in the share of government expenditure in GDP dedicated to environmental protection corresponds to a 1.2% decrease in per capita residential energy use. This estimate indicates that, on average, government environmental expenditure reduces residential energy consumption by 437,456 mwh per year across the ten Canadian provinces. Assuming coal-fired power generation, this reduction translates into approximately 0.06% of Canada’s annual CO₂e emissions.

Similarly, provinces implementing explicit environmental policies exhibit a 1% lower per capita household energy consumption than provinces without such policies. These results underscore the efficacy of government expenditure earmarked for environmental protection and other environmental policies in encouraging households to adopt pro-environmental behaviors, thus reducing household energy consumption. This aligns with the existing literature, where studies such as that of Aydin and Brounen [14] found that mandatory energy efficiency labels for household appliances and building standards have reduced household electricity consumption across Europe. Similarly, Fan et al. [4] demonstrate that local environmental protection expenditures in China contribute to improved environmental quality by curbing pollution.

Our investigation also delves into the significance of various determinants that shape residential energy use. Drawing from the available literature, these determinants encompass energy price, income, demographic characteristics, and climate. Our estimates affirm that the price elasticity of demand for residential energy is negative and highly significant, yet its magnitude is notably low (−0.001), indicating high inelasticity. This finding is in accordance with the existing literature [14,39,53], which also underscores the inelastic nature of price elasticity for residential energy consumption. For instance, Aydin and Brounen [14] estimated the price elasticity of demand for residential energy use to be −0.07.

Household energy is considered a necessity in modern households for essential activities like heating, cooling, cooking, and lighting. As a result, even when energy prices change, households find it challenging to significantly reduce their consumption in the short term because of the limited availability of substitutes and the essential role energy plays in daily life. This does not indicate that energy saving is not connected to the energy price level but, in the long run, it may be relatively elastic.

Higher income levels typically prompt households to acquire more energy-consuming equipment and appliances, contributing to increased energy use. Thus, a positive correlation is anticipated between income and residential energy use. In harmony with the literature, our estimates reveal that residential energy consumption is significantly influenced by income. Specifically, our calculated income elasticity for household energy consumption stands at 0.05, a value consistent with previous studies in developed countries. However, this value is low compared with the estimated values for European countries, which fall within a range of 0.19 to 0.216 [14]. The income elasticity of electricity, as reported in the literature, varies from 0.2 to 0.4 for G7 countries [54] and is 0.05 for the US [55].

In Canada, the lower income elasticity of residential energy consumption reflects the country’s cold climate, which makes space heating essential regardless of income. Stringent building codes and widespread energy-efficient technologies mean that energy use does not increase much as incomes rise, unlike in other countries. Regional climate differences further dampen overall income elasticity. As a result, price-based policies like carbon taxes may have limited direct effects on household energy consumption. Instead, policies should prioritize direct interventions—such as building retrofits, better insulation, and efficient heating systems—while ensuring equitable access to energy efficiency upgrades for lower-income households.

The composition of the population, particularly the share of elderly individuals, significantly affects household energy use. While older individuals typically spend more time at home and use more appliances—leading to increased energy consumption—factors like budget constraints and environmental awareness can mitigate these effects. Our estimates reveal that a 1-percentage-point increase in the elderly share results in a slight decline (−0.06%) in per capita energy use, contrasting with some studies that report a positive relationship. Estiri and Zagheni [56] attribute rising energy use to housing size, while Inoue et al. [57] found that Japan’s aging population increased household energy consumption by 12% from 1995 to 2015. A study in innovation in aging [58] similarly suggests that aging influences energy use via changes in household size and consumption. These findings stress the importance of demographic trends in shaping energy policy.

Heating-degree days (hdd) and cooling-degree days (cdd) are essential gauges for determining the heating and cooling energy demands of dwellings, particularly in response to climate and weather conditions [8]. Notably, our findings align with other studies [14], indicating that a 1-percent increase in heating-degree days correlates with a 0.41% increase in residential energy demand. This correlation can be attributed to intensified heating system use during colder days. Regarding cooling demands, our analysis shows that the number of cooling-degree days does not significantly impact residential energy use in our sample of Canadian provinces, where the use of air conditioning is limited. Households demonstrate significant responsiveness to both heating- and cooling-degree days in the short term, with coefficients higher than those observed in the long run.

Taking all factors into account, our results show that heating-degree days have the strongest influence on per capita residential energy use, followed by government environmental expenditure. By contrast, energy price has the weakest influence on per capita residential energy use in Canada.

The dataset used in this study lacks detailed information on households, dwellings, and policy implementation. Future research could address these limitations by collecting more granular household-level data through surveys, including energy consumption patterns, appliance ownership, building characteristics, and energy-saving behaviors. Additionally, incorporating specific policy data (like rebates and retrofitting incentives) and granular weather and climate variables would help capture broader environmental influences. Longitudinal data would also enhance our understanding of how households adapt over time, including variations by income, age, and household composition.

6. Conclusions and Policy Implication

The residential sector holds a substantial share of total energy consumption, presenting an opportunity to significantly impact overall energy use and emissions due to the long lifespan of buildings and the challenge of relocating emissions abroad. Government investment in environmental protection has the potential to lower residential energy consumption, enhancing environmental quality. Through policies and initiatives, governments can foster an energy-efficient and sustainable environment to reduce residential energy use. While Canadian household energy consumption has improved, substantial inter-provincial variations persist. Therefore, this study aimed to assess whether government expenditure on environmental protection influences reduced household energy consumption and environmental quality enhancement in Canada.

We employed panel data encompassing ten Canadian provinces from 1995 to 2020, sourced from Statistics Canada and Environment Canada. A conceptual model and an empirical model were developed to estimate household energy consumption, classifying determinants into internal factors (energy price, income, behavior, motivation, household, and dwelling characteristics) and external factors (weather, fiscal, and regulatory environmental policies). The dependent variable was residential energy use (per capita), while the explanatory variable of focus was government expenditure on environmental protection (specified as a share of GDP).

Appropriate panel techniques were applied, including second-generation unit root (under cross-sectional dependence) and co-integration (Westerlund bootstrapping) tests. The unit root tests confirmed that all variables were non-stationarity in the level form and stationarity in the first differences, with a co-integration relationship between variables, indicating a long-run relationship. Consequently, the FMOLS and dynamic OLS methods were employed to estimate the long-run relationship. Short-run relationships were also evaluated using pooled OLS and random effect panel methods, maintaining first difference forms for all variables.

The long-run analysis revealed that all explanatory variables were highly significant, aligning with expectations, except for cooling-degree days. Variables such as the policy dummy (pd), the share of government spending on environmental protection in GDP (sgexp), energy price (pelect), and the share of the elderly population (spop65) displayed a negative correlation with per capita residential energy consumption (lpcreu). Conversely, per capita GDP (lpcgdp) and heating-degree days (lhdd) exhibited a positive relationship with per capita household energy consumption. Notably, a 10% increase in the share of government expenditure in GDP targeting environmental protection led to a 1.2% reduction in residential energy use, while provinces implementing explicit environmental policies demonstrated 1% lower per capita household energy use.

Given the demonstrated effectiveness of province-specific environmental policies in significantly reducing household energy consumption, it is important to advocate for the adoption of explicit environmental policies in provinces currently lacking them. Implementing tailored measures such as carbon taxes, building retrofits, or energy efficiency incentives can provide substantial reductions in household energy demand, contributing to both climate action and economic resilience. These policy efforts will help ensure that all provinces participate in the transition to sustainable energy use, leveraging the clear benefits already seen in provinces with robust environmental initiatives.

This study also found that both price and income elasticities reflect inelastic responses of residential energy consumption to changes in price and income. By contrast, an increase in the share of the elderly population was associated with reduced household energy consumption, while heating-degree days emerged as the most significant factor influencing energy use. Specifically, heating-degree days exert the greatest impact on per capita residential energy consumption in Canada, followed by government environmental expenditure, with energy price having the least influence. Based on our estimates, government spending leads to carbon reductions ranging from 4.3 kg to 5.75 kg CO₂e per dollar spent. These findings highlight the effectiveness of government expenditure on environmental protection in encouraging pro-environmental behavior and reducing household energy consumption.

We recommend prioritizing dedicated government spending on environmental protection as a key strategy to curtail household energy consumption and improve environmental quality. The role of province-specific environmental policies in influencing household energy use highlights the significance of adopting such policies across all provinces. Although this study makes substantial contributions, it also identifies areas for future research, such as assessing the relative effectiveness of different types of province-specific environmental policies and exploring the influence of government environmental spending on residential energy use efficiency and energy use types. Additionally, due to data limitations, we did not delve into the impact of household and building characteristics, an avenue that future research could explore.

Future climate challenges—such as increased heating and cooling demands—underscore the importance of tailoring government spending and policies to promote energy-efficient technologies and building retrofits. Investing in insulation improvements, smart thermostats, and adaptive technologies will help households to manage their rising energy needs more sustainably. Policy incentives can also target energy equity, ensuring low-income households have access to these upgrades. Additionally, integrating climate resilience strategies—such as shading, passive cooling designs, and renewable heating solutions—into building codes and government programs will further prepare residential energy systems for a warming climate.