Effects of Economic Growth,Food Price Inflation and Population Explosion on Food Consumption Related Greenhouse Gas Emissions: A Panel Data Analysis

Abstract

This study aims to analyze the effect of economic growth, food price inflation, and population growth on GHG emissions from household food consumption by employing panel data from 170 countries. For empirical analysis, the study employs a two-way fixed effects model and a system GMM estimator. The results reveal that real per capita income has a significant positive impact on food consumption-related GHG emissions, whereas the square of real per capita income has a significant negative impact. This suggests the presence of a threshold beyond which further income growth reduces GHG emissions, this supporting the Environmental Kuznets Curve (EKC) hypothesis. However, the EKC hypothesis holds for high-income countries but not for low-income countries. Moreover, as food prices rise, GHG emissions from household food consumption also increase. Likewise, population growth leads to higher GHG emissions from household food consumption, with similar results observed in both low and high-income countries. These findings align with the Ehrlich–Commoner theory. Furthermore, the study identifies specific consumption patterns that influence GHG emissions. Per capita consumption of rice and fish mitigates emissions, while increased consumption of wheat, sugar, pulses, fruits, and meat exacerbates them. Finally, the choice of cooking fuel significantly impacts GHG emissions. Solid fuels such as coal and charcoal increase emissions, while gaseous fuels like natural gas and liquefied petroleum gas (LPG) reduce them. However, opposite trends are observed in low and high-income countries. These findings underscore the urgent need for emission reduction strategies in food consumption at both national and global levels.

Introduction

Greenhouse gas (GHG) emissions contribute to climate change and post an extensive, irreplaceable, and unalterable repercussions on every area of our lives (Hui et al., 2022; D. Liu et al., 2019). Climate change further causes the destruction of ecosystems and biodiversity, an increase in the frequency of life-threatening events (famines, floods, and cyclones), insecurity in agriculture production, variations in the accessibility of drinking water, health problems associated with extreme climate conditions, sea-level escalation, and other types of harm (Hasegawa et al., 2018; Holden et al., 2014). Though GHG emissions come from every economic sector, the food system is of particular significance because it accounts for about one third (34% [range 25%–42%]) of all anthropogenic emissions worldwide (Crippa et al., 2021; Shukla et al., 2019). Food-related emissions are produced by the entire food system, including methane (CH₄) from enteric fermentation in ruminant animals and rice farming, nitrous oxide (N₂O) from fertilizer use and soil management, and carbon dioxide (CO₂) from the combustion of fossil fuels (Aguiar et al., 2021). The difficult dilemma of how to feed a greater population while simultaneously reducing GHG emissions arises due to the strong correlation between food production and GHG emissions as well as the anticipated rise in global population to around 9.7 billion by 2050 (FAO, 2017; UN Population Fund, 2022). Many policymakers suggested numerous production models to address this issue, with a focus on the application of new technologies and management techniques. Other policymakers used various consumption strategies because they thought that without future increases in GHG emissions, the supply of food would not be able to fulfill demand (Aguiar et al., 2021).

Global food consumption produces 22% to 37% of all anthropogenic emissions (Charles, 2021), while agricultural production and land use change account for most emissions. Besides, only 2.4 GtCO2eq and 1 GtCO2eq are produced during the post-production and post-sale periods (Rogissart et al., 2019). Moreover, waste management in food production produces the least amount of emissions. Subsequent stages in the food system, such as packaging, retail, transport, processing, food preparation, and waste disposal, produce just 5% to 10% of world GHG emissions (Garnett et al., 2016). From 2010 to 2020, on a global scale, 1,015,777 kt of CO₂, 10 kt of N₂O, and 111 kt of CH₄ were emitted from household food consumption annually (Shabir et al., 2023). Regional differences exist in the proportion of overall GHG emissions. Asia is responsible for roughly 61% of the global CO₂, 70% of the global N₂O, and 70% of the global CH₄ emissions from household food consumption from 2010 to 2020 (Crippa et al., 2021). America is responsible for approximately 17% of the global CO₂, 14% of the global N₂O, and 9% of the global CH₄ emissions (FAO, 2022). Around 15% of the world’s CO₂, 12% of the world’s N₂O, and 15% of the world’s CH₄ emissions are attributed to Europe. Africa is responsible for approximately 6% of global CO₂, 3% of global N₂O, and 5% of global CH₄ emissions (FAO, 2022). Finally, Oceania is responsible for approximately 1% of the global CO₂, 2% of the global N₂O, and 0.2% of the global CH₄ emissions from household food consumption (FAO, 2022; Mrówczyńska-Kamińska et al., 2021). International differences exist as well, especially across wealthy nations. About 31% of all GHG emissions in the EU-25 are attributable to food consumption. The average American household contributes 15% of the nation’s emissions to the total yearly emissions from food consumption. In Sweden, the food system contributes more than 20% of all GHG emissions, compared to Australia’s 28.3% total GHG emissions (Li et al., 2016).

This increasing trends in GHG emissions from food consumption shows how the existing pattern of food consumption around the world is increasingly thought to be unsustainable. On the one hand, meeting the basic human need for nutrition is important, while on the other hand, meeting the basic human need for nutrition creates serious environmental risks (Ingram, 2011; Islam & Wong, 2017). Therefore, a revamped research agenda is necessary to ensure food and nutrition security with minimal environmental impact. For this purpose, a well-managed and improved food supply system is needed (Notarnicola et al., 2017). The “EAT-Lancet Commission (EATLC),” comprised of 37 researchers from sixteen nations, was formed in 2019. The main aim of the commission is to provide information to policymakers that will further help them establish universal targets addressing the crucial prerequisite for a “fundamental transformation of the worldwide food system.” The researchers at EATLC focused on food production and final consumption targets in order to develop a worldwide food system capable of feeding a population of 10 billion people by 2050 and enabling the UN’s SDGs while maximizing the adverse health impacts of diets and diet-related GHG emissions. In 2019, the EATLC published its report on “Healthy Diets from Sustainable Food Systems.” This report revealed that considerable dietetic and food production changes are required to prevent environmental damage from existing global food consumption and production patterns, as they have already exceeded the maximum GHG emissions boundary (EAT, 2019; Mertens et al., 2020). Based on the findings and suggestions of the EAT-Lancet Commission’s report, Gren et al. (2019) proposed that to keep global warming below 2°C, over pre-industrial levels, GHG emissions reductions are required in the food sector.

The increase in GHG emissions from food consumption on a global scale has drawn the attention of researchers to this important issue. Researchers used a lifecycle assessment (LCA)-based approach to a basket of food products to estimate the GHG emissions in order to analyze the environmental impact of food consumption (Bassi et al., 2022; Notarnicola et al., 2017: Silva et al., 2023). These studies employed household-level data and estimated GHG emitted from various food items. Other researchers quantified the life cycle GHG emissions of the food system (Mohareb et al., 2018). Some researchers determined the effects of dietary change on GHG emissions and found changes in GHG emissions when dietary patterns changed over time (Hawkins et al., 2018; Hitaj et al., 2019). Previous studies also analyzed the environmental cost of food consumption (GHG emissions) (Aguiar et al., 2021; Eberle & Fels, 2016; Schmidt et al., 2019). They found a significant positive association between food consumption and GHG emissions. Researchers developed a proper link between food consumption, GHG emissions, and global warming and found that GHG emissions from food consumption significantly increase global warming (Ivanovich et al., 2023; Kassem et al., 2022). Green et al. (2018) and Naja et al. (2018) evaluated the environmental footprints of various food products. Bjelle et al. (2021) investigated the impact of income on food-related GHG emissions at a global scale. They found a minor 1% reduction in total GHG emissions emitted from food consumption. This outcome is mostly driven by lower emissions in the BRICS and other regions of the world. Chen and Zhong (2022) looked at diet-related agricultural GHG emissions in China. The study’s findings indicate that food demand and diet-related agricultural GHG emissions will rise in the short term, but total diet-related agricultural GHG emissions may fall in the long-run due to a lower increase in animal-based food and a greater decrease in staple foods. The casual link between agricultural corban emissions, prices of factors of agricultural production, and the consumer price for food was assessed by Pang et al. (2021). They found that GHG emissions are highly influenced by food prices and the prices of factors of agricultural production.

Some researchers in the past analyzed the effects of economic growth, food prices, and population growth on CO₂ emissions. For an instant, Ssali et al. (2018) examined the impact of economic growth, energy use, and population growth on CO₂ emissions in sub–Saharan Africa. Results of the study revealed that an increase in energy use and population growth causes increase in CO₂ emissions. For the same region, Adzawla et al. (2019) tested the ECK hypothesis. Results of the study indicated a monotonic decreasing relationship between economic growth and GHG emissions in the long-run. In the same year, S. Khan et al. (2019) investigated the impacts of financial development, economic growth, and energy consumption on CO₂ emission. Results of the study confirmed the ECK hypothesis for the global panel. Similarly, Qiao et al. (2019) examined the effects of the agriculture, economic growth, and renewable energy on CO₂ emissions within the framework of the EKC. Results of the study confirmed the EKC hypothesis for full sample and for developed countries.

In the next year, Nabi et al. (2020) examined the linkages between population growth, price level, poverty, and CO₂ emissions. Results of the study showed a positive relationship between changes in price level, poverty rates, and carbon emissions. In the same year, Ronaghi et al. (2020) examined the relationship between governance, economic performance, and CO₂ emissions. Results of the study showed that the governance index, GDP growth, inflation rate, exports, imports, foreign investment, and employment have significant impacts on CO₂ emissions. One year later, Muhammad et al. (2021) examined the impact of foreign direct investment, natural resources, renewable energy consumption, and economic growth on environmental degradation. Results of the study indicated that natural resources and economic growth are the main factors that boost the environmental degradation in BRICS, developing, and developed countries. After 1 year gap, Gao et al. (2022) analyzed the impact of tax policy on total factor carbon emission efficiency. Results of the study showed that tax reforms enhanced the total factor carbon emission efficiency in China. Recently, Acevedo-Ramos et al. (2023) examined the validity of the ECK for Colombia. Results of the study supported the existence of an EKC relationship between GDP and methane emissions. In the next year, Alfaisal et al. (2025) looked at how economic performance, tourism, renewable energy, and energy efficiency affect carbon emissions in the emerging economies of BRICS during 1990 to 2021. The study found that economic performance is the significant driver of higher emissions levels in the BRICS countries. In the same year, Somoye et al. (2025) investigated the impact of CO₂ emissions on food production in the United States. Results of the study showed that CO₂ emissions, GDP, population, and inflation have a significant positive impact on food production in the long-run.

However, most of the previous studies used household-level cross-sectional data; a few studies used small panel data consisting of a few countries (Crippa et al., 2021; Mrówczyńska-Kamińska et al., 2021), while other studies used time-series data (Hawkins et al., 2018; He et al., 2021; L. C. Liu et al., 2011). Besides, the studies that used proper econometric methodologies employed a limited number of economic and demographic variables while ignoring the consumption of various food items and types of cooking fuels as explanatory variables. Owing to the complexity of food systems, which consist of environmental, economic, racial, and geopolitical subsystems, the linkages and connections between individuals, their culture, their economic status, and the environment can be understood by seeing them as socio-economic-ecological systems. Since the food system and climate change (caused by rising GHG emissions) are fundamentally socio-economic and ecological issues, it is crucial to take economic and demographic factors into account when analyzing the effects and available mitigation options. Climate change and food together stimulate cultural norms and identities, which in turn influence behavior. When studying how dietetic behaviors and patterns have changed, it is critical to take economic and demographic aspects into account in order to analyze discrepancies and the potential for policy or behavior changes that may possibly impact diets’ ability to adapt to climate change (Bjelle et al., 2021). However, studies carried out on a macroscale are critical in providing decision-makers with knowledge for transitioning to more sustainable food consumption patterns by decoupling environmental impacts from human requirements while ensuring economic growth (Notarnicola et al., 2017). With this context in mind, the current study seeks to analyze the effect of economic growth, food prices, inflation, and population explosions on GHG emissions from household food consumption by employing panel data from 170 countries. The main contribution of this study is that it estimates an empirical model using a rich dataset of 170 countries, exploring economic and demographic variables as well as consumption of various food items and types of cooking fuels as explanatory variables that may impact GHG emissions from food consumption.

Material and Methods

Theoretical Framework

The theoretical framework of this study is based on the Environmental Kuznets Curve (EKC) hypothesis. According to the EKC hypothesis, economic growth and food-related GHG emissions have a quadratic relationship, such as at the early stages of economic development as the per capita income of households increases the GHG emissions from their food consumption increases but after a certain period when the income of the household increases beyond a certain threshold, with further increase in households’ income decrease their food related GHG emissions as shown in Figure 1 (De Bruyn et al., 1998). The higher per capita income rises the purchasing power of the households, they start purchasing larger quantities of animal-based, processed, and convenience foods, which are often associated with higher GHG emissions. However, as households’ income reaches to a certain threshold, the trends between income and food-related GHG emissions begins to change. At this stage, households may become more aware of the environmental impact of their food consumption; therefore, they adopt more sustainable food consumption patterns (Galeotti et al., 2006; Stern, 2004).

Figure 1.

Environmental Kuznets curve.

Data and Its Sources

For analyzing the study objective, we used data from the Food and Agriculture Organization of the United Nations (FAO), the World Bank, and the World Health Organization (WHO). The data on GHG (CO₂, N₂O, and CH₄) emissions from household food consumption, food price index, and per capita consumption of various food commodities is extracted from FAOSTAT (FAO, 2024). The data on real per capita GDP and population is taken from World Development Indicators (WDI), published by the World Bank (2024). Finally, country-wise data on types of cooking fuels is taken from the WHO website (World Health Organization, 2024). The study extracted data from the above sources for panel of 170 countries (see Table A1) from 2010 to 2020, covering an 11-year time span. Initially, we selected the entire 196 countries for our analysis but we found that data of various variables mentioned above for 26 countries are missing for certain years. Therefore, we did not include these countries into our data set. Therefore, our final dataset consists of 170 countries. Similarly, since the data on per capita consumption of various food items is available from 2010 to 2020 while data on GHG (CO₂, N₂O, and CH₄) is available up to 2020; therefore, we construct our dataset from 2010 to 2020. The study used panel data, since panel data is often favored over time-series and cross-sectional data because they comprise superior sampling variability and degrees of freedom, which increases the validity and accuracy of parameter estimation (Fong et al., 2020).

Econometric Model

In order to bridge a quadratic relationship between per capita income and household food-related GHG (CO₂, N₂O, and CH₄) emissions, we follow Fong et al. (2020) and developed the following regressions models:

$$\begin{array}{l}ln\left(C{O}_{2it}\right)={\alpha }_{1i}+{\delta }_{1t}+{\beta }_{1}ln\left(y_{it}\right)+{\beta }_2{(ln{y}_{it})}^2+{\gamma }_1\ln \left(p_{it}\right)\\ +{\theta }_{1k}ln\left(N_{it}\right)+{\vartheta }_{1j}\sum ln\left(c_{it}\right)+{\phi }_{1j}\sum ln\left(f_{it}\right)+{\mu }_{1it}\end{array}$$

(1)

$$\begin{array}{c}\multicolumn{1}{c}{ln\left(N_2{O}_{\mathrm{it}}\right)={\alpha }_{2i}+{\delta }_{2t}+{\beta }_{12}ln\left(y_{\mathrm{it}}\right)+{\beta }_{22}(ln{y}_{\mathrm{it}}{)}^2+{\gamma }_2\ln \left(p_{\mathrm{it}}\right)}\\ \multicolumn{1}{c}{+{\theta }_{2k}ln\left(N_{\mathrm{it}}\right)+{\vartheta }_{2j}\sum ln\left(c_{\mathrm{it}}\right)+{\phi }_{2j}\sum ln\left(f_{\mathrm{it}}\right)+{\mu }_{2\mathrm{it}}}\end{array}$$

(2)

$$\begin{array}{c}\multicolumn{1}{c}{ln\left(C{H}_{4\mathrm{it}}\right)={\alpha }_{3i}+{\delta }_{3t}+{\beta }_{13}ln\left(y_{\mathrm{it}}\right)+{\beta }_{33}(ln{y}_{\mathrm{it}}{)}^2+{\gamma }_3\ln \left(p_{\mathrm{it}}\right)}\\ \multicolumn{1}{c}{+{\theta }_{3k}ln\left(N_{\mathrm{it}}\right)+{\vartheta }_{3j}\sum ln\left(c_{\mathrm{it}}\right)+{\phi }_{3j}\sum ln\left(f_{\mathrm{it}}\right)+{\mu }_{3\mathrm{it}}}\end{array}$$

(3)

where $C{O}_{2\mathrm{it}}$ , $N_2{O}_{\mathrm{it}}$ , and $C{H}_{4\mathrm{it}}$ are emissions of CO₂, N₂O, and CH₄ for country i in year t, $y_{\mathrm{it}}$ is per capita GDP for country i in year t, $p_{\mathrm{it}}$ is food price index for country i in year t, $N_{\mathrm{it}}$ is the total population of country i in year t, $c_{\mathrm{it}}$ is per capita consumption of food commodity of country i in year t, and $f_{\mathrm{it}}$ is the types of fuel for cooking. ${\mu }_{\mathrm{it}}$ stands for the error terms, ${\alpha }_i$ and ${\delta }_t$ represent the country and year fixed effects, respectively. $t=1,2,3,\dots \dots \dots \dots ,11$ is the time span of the data where $i=1,2,3,\dots \dots \dots \dots ,170$ is the number countries in the panel. $\beta ,\gamma ,\theta ,\vartheta$ , and φ are the models’ coefficients to be calculated. These coefficients may take positive or negative values and can be statistically significant at the 1%, 5%, or 10% levels, or they may be found to be insignificant (Hayat et al., 2025). In our models, the dependent variables are CO₂, N₂O, and CH₄ emissions. The explanatory variables are income (GDP per capita and its squared term; Acevedo-Ramos et al., 2023; Adzawla et al., 2019; S. Khan et al., 2019; Mrówczyńska-Kamińska et al., 2021), food prices (food price indices base year 2015; Nabi et al., 2020; Pang et al., 2021; Ronaghi et al., 2020; Somoye et al., 2025; Ssali et al., 2018), population (total population; Mrówczyńska-Kamińska et al., 2021; Nabi et al., 2020; Ronaghi et al., 2020; Silva et al., 2023; Somoye et al., 2025; Ssali et al., 2018), per capita food consumption (per capita consumption of wheat, rice, sugar, pulses, fruits, milk, eggs, meat, and fish; Chen & Zhong, 2022), and type of fuel for cooking (population with primary reliance on biomass, charcoal, coal, electricity, gas, and kerosene for cooking; Wright et al., 2020).

Estimation Techniques and Diagnostic Tests

For estimation of models (1), (2), and (3), we used fixed effects estimators that account individual heterogeneity. Fixed effects capture country-specific features that remain constant over time, such as environment and topography, while controlling for regular temporal external shocks that might influence the models variables, such as financial crises or the introduction of, or changes to, regional policies (Fong et al., 2020). For the standard EKC, here models are estimated using country FE and year FE because fixed effect models control for panel heterogeneity (Clune et al., 2017). Furthermore, fixed effects estimation facilitates the assessment of country-specific influences that the pooled OLS approach fails to capture (Ward & Leigh, 1993). Alternative estimation techniques, such as System Generalized Method of Moments (GMM) is also used in this study, but the method of system GMM may deemed unnecessary for the current study because the fixed effects model adequately addressed endogeneity concerns related to unobservable firm-specific factors. However, for robust analysis we also estimate our models via System GMM method (Oganda, 2023). For ensure the validity and reliability of the regression models we applied various diagnostic tests. For checking the problem of multicollinearity, we have employed variance inflation factor (VIF) and tolerance (TOL; Gujarati, 2009). The Wooldridge test is performed for detecting the issue of serial correlation in the fixed effect models. For detecting the problem of endogeneity in the fixed effect models, we use Durbin-Wu-Hausman test.

Descriptive Statistics

Table 1 summarizes the descriptive statistics of key factors influencing household food consumption-related GHG emissions. It is observed that from 2010 to 2020, every selected country included in the analysis on average emits 5,950 kt of CO₂, 0.057 kt of N₂O, and 0.646 kt of CH₄ annually. The global food system is responsible for nearly half of CH₄ emissions, two-thirds of N₂O emissions, and 3% of CO₂ emissions.

Table 1.

Descriptive Statistics of Key Factors Influencing Household Food Consumption-related GHG Emissions.

Variables	Definition	Mean and standard deviation
Dependent
CO2 emissions	Carbon dioxide (CO2) emissions from households’ food consumption (kg)	5,950,000,000 (691,000,000)
N2O emissions	Nitrous oxide (N2O) emissions from households’ food consumption (kg)	57,049.04 (8,548.4)
CH4 emissions	Methane (CH4) emissions from households’ food consumption (kg)	646,792.8 (102,319.7)
Explanatory
Real per capita income	GDP per capita (constant 2015 US$)	12,972.8 (419.8)
Food price index	Consumer food prices index (2015 = 100)	109.4 (33.19)
Population	Total population of all residents	42,700,000 (15,100,000)
Per capita wheat consumption	Consumption of wheat per person (kg/capita/year)	72 (48)
Per capita rice consumption	Consumption of rice per person (kg/capita/year)	43 (54)
Per capita sugar consumption	Consumption of sugar per person (kg/capita/year)	40 (25)
Per capita pulses consumption	Consumption of pulses per person (kg/capita/year)	6 (6)
Per capita eggs consumption	Consumption of eggs per person (kg/capita/year)	7 (5)
Per capita fish consumption	Consumption of fish per person (kg/capita/year)	20 (19)
Per capita fruits consumption	Consumption of fruits per person (kg/capita/year)	80 (51)
Per capita meat consumption	Consumption of meat per person (kg/capita/year)	49 (29)
Per capita milk consumption	Consumption of milk per person (kg/capita/year)	84 (80)
Biomass	Population with primary reliance on biomass (wood, crop residues, and dung) for cooking	12,900,000 (1,325,357)
Charcoal	Population with primary reliance on charcoal for cooking	1,239,228 (69,251.08)
Coal	Population with primary reliance on coal for cooking	822,562.4 (133,601.1)
Electricity	Population with primary reliance on electricity for cooking	5,943,890 (701,426.5)
Gas	Population with primary reliance on gas (LPG, natural gas, and biogas) for cooking	20,400,000 (1,747,930)
Kerosene	Population with primary reliance on kerosene for cooking	530,575.1 (79,581.62)
Observations	Number of observations	1,870

Source. Computed by authors based on panel data from 170 countries.

Note. Standard deviation in parentheses.

However, only one-third of countries adopted various GHG emissions mitigation measures in their national food systems (Ivanovich et al., 2023). We observe that the average per capita income in the selected countries is USD 12,973. The average food price index in the selected countries is 109. This shows that the residents of the selected countries pay comparatively higher prices for food. Recently, extreme weather events caused by climate change, rising energy costs, currency fluctuations, rapid urbanization, and population growth have raised the prices of various commodities on a global scale, contributing to further food price inflation (Hovhannisyan & Devadoss, 2020; Kunawotor et al., 2022). It is observed that the average population in the selected countries is 42.7 million. Furthermore, on average, the highest per capita consumption is observed for milk while the lowest per capita consumption is observed for fish. From the mean values of fuel types, it is observed that on average majority of the population (63%) in the selected countries relied primarily on clean fuels for cooking, such as electricity, LPG, natural gas, and biogas. The rest of the population (37%), mostly living in lower and middle-income countries, relied on polluting fuels such as kerosene, biomass, charcoal, and coal.

Results and Discussions

Multicollinearity

The heat plot matrix in Figure 2, presents the pairwise correlations among the dependent and explanatory variables used in this study. For all the explanatory variables the values of correlations are around 0.80 or lower than this value. Based on these values, we did not find any evidence of multicollinearity among the explanatory variables. The results of VIF and TOL are reported in Table 2. The values of VIF for all the explanatory variable did not exceeds 10 while values of TOL are closer to 1. This shows that explanatory variables are not collinear with each other. Therefore, we found no multicollinearity among the explanatory variables.

Figure 2.

Heat plot matrix.

Table 2.

Results of VIF and TOL for Checking Multicollinearity.

Variables	VIF	TOL = 1/VIF
Real per capita income	6.3	0.15
Food price index	1.0	0.96
Population	3.5	0.28
Per capita wheat consumption	2.7	0.36
Per capita rice consumption	2.0	0.49
Per capita sugar consumption	3.2	0.32
Per capita pulses consumption	1.3	0.77
Per capita eggs consumption	3.5	0.28
Per capita fish consumption	1.6	0.62
Per capita fruits consumption	1.2	0.83
Per capita meat consumption	3.6	0.27
Per capita milk consumption	3.3	0.31
Biomass	2.7	0.36
Charcoal	1.9	0.51
Coal	1.6	0.61
Electricity	1.8	0.53
Gas	2.8	0.35
Kerosene	1.7	0.59

Source. Computed by authors based on panel data from 170 countries.

Serial Correlation and Panel Endogeneity

The results of the Wooldridge test statistics are given in the last section of Tables 3 and 4. The Wooldridge test results for the nine models are statistically significant at the 1% and 5% levels, indicating that our models suffer from serial correlation. However, this issue is addressed by using Driscoll-Kraay standard errors and introducing country and time fixed effects into the models. The results of the Durbin-Wu-Hausman test are given in the last section of Tables 3 and 4. For model (6) and model (8) we found insignificant test statistics. Based on this insignificant test statistics values, we accept the null hypothesis of no endogeneity in the model. However, the significant test statistics for the seven models show endogeneity in the models. Therefore, for removing the endogeneity problem we included country and time fixed effect into our models.

Table 3.

Fixed Effect Regressions Results for 170 Countries: Effects of Economic Growth, Food Prices, Inflation, and Population Growth on Household Food-related GHG Emissions.

Model	(1)	(2)	(3)
Dependent variables	CO2 emissions (ln)	N2O emissions (ln)	CH4 emissions (ln)
Income:
Real per capita income (ln)	2.844*** (0.239)	4.108*** (0.475)	3.020*** (0.368)
Square of real per capita income (ln)	−0.140*** (0.0140)	−0.200*** (0.0273)	−0.153*** (0.0212)
Food price
Food price index (ln)	0.140*** (0.0259)	0.201*** (0.0345)	0.0879*** (0.0142)
Population
Population (ln)	1.920*** (0.0628)	1.808*** (0.110)	1.872*** (0.0734)
Per capita food consumption
Per capita wheat consumption (ln)	0.313*** (0.0634)	0.259*** (0.0661)	0.258*** (0.0630)
Per capita rice consumption (ln)	−0.0421 (0.0394)	−0.0934*** (0.0282)	−0.0523 (0.0296)
Per capita sugar consumption (ln)	0.140*** (0.0367)	0.0106 (0.0298)	0.146*** (0.0328)
Per capita pulses consumption (ln)	0.0347** (0.0122)	0.0196 (0.0176)	0.0626** (0.0236)
Per capita eggs consumption (ln)	−0.0110 (0.0307)	0.0236 (0.0307)	0.0196 (0.0351)
Per capita fish consumption (ln)	−0.117*** (0.0275)	−0.0916*** (0.0212)	−0.0828*** (0.0260)
Per capita fruits consumption (ln)	−0.0180 (0.0179)	0.0328* (0.0167)	0.0703*** (0.0147)
Per capita meat consumption (ln)	0.183** (0.0703)	0.163*** (0.0343)	0.0365 (0.0685)
Per capita milk consumption (ln)	0.0335** (0.0141)	−0.0422 (0.0337)	−0.0162 (0.0183)
Types of cooking fuels
Biomass (ln)	−0.0121* (0.00637)	−0.00468 (0.00908)	−0.0143** (0.00571)
Charcoal (ln)	0.00804** (0.00306)	0.0148** (0.00521)	0.00632** (0.00281)
Coal (ln)	0.0172*** (0.00362)	0.00870 (0.00524)	0.00680** (0.00258)
Electricity (ln)	−0.00990** (0.00400)	0.0170** (0.00551)	−0.0110*** (0.00256)
Gas (ln)	−0.0459** (0.0177)	−0.0174 (0.0119)	−0.0549** (0.0231)
Kerosene (ln)	−0.00331 (0.00351)	−0.0159** (0.00601)	0.00129 (0.00287)
Lages
First lag
Fixed effects
Country	Yes	Yes	Yes
Time	Yes	Yes	Yes
Constant	−27.29*** (1.694)	−44.02*** (3.057)	−35.52*** (1.853)
Observations	1,870	1,870	1,870
R2	.99	.99	.99
F-statistics	1,050.5***	3,690.3***	2,997***
Root MSE	0.19	0.24	0.23
Wooldridge test for serial correlation	127.2***	86.6***	74.2***
Durbin-Wu-Hausman	208.1***	73.4***	141.9***

Source. Computed by authors based on panel data from 170 countries. Standard errors in parentheses.

***

p < .01. **p < .05. *p < .1.

Table 4.

Fixed Effect Regressions Results for Low and High-Income Countries: Effects of Economic Growth, Food Prices, Inflation, and Population Growth on Household Food-Related GHG Emissions.

	Low-income countries			High-income countries
Model	(4)	(5)	(6)	(7)	(8)	(9)
Dependent variables	CO2 emissions (ln)	N2O emissions (ln)	CH4 emissions (ln)	CO2 emissions (ln)	N2O emissions (ln)	CH4 emissions (ln)
Income
Real per capita income (ln)	−6.436** (2.593)	−12.43*** (1.585)	−7.922** (2.858)	5.623*** (0.996)	5.361*** (1.009)	4.561*** (1.190)
Square of real per capita income (ln)	0.454** (0.170)	0.956*** (0.112)	0.607** (0.200)	−0.262*** (0.0448)	−0.245*** (0.0480)	−0.213*** (0.0541)
Food price
Food price index (ln)	−0.0207 (0.0890)	0.429*** (0.0498)	0.0649 (0.0715)	0.206 (0.123)	0.110 (0.135)	0.0301 (0.141)
Population
Population (ln)	3.344*** (0.657)	3.387*** (0.372)	2.860*** (0.483)	1.501*** (0.122)	1.199*** (0.224)	1.497*** (0.151)
Per capita food consumption
Per capita wheat consumption (ln)	0.282*** (0.0855)	0.116* (0.0607)	−0.135* (0.0673)	0.301*** (0.0454)	0.436*** (0.0684)	0.323*** (0.0948)
Per capita rice consumption (ln)	−0.114 (0.119)	0.193 (0.134)	0.0386 (0.0880)	−0.0320 (0.0337)	−0.0883 (0.0531)	−0.0671 (0.0621)
Per capita sugar consumption (ln)	0.386* (0.206)	0.118 (0.0650)	0.360** (0.148)	−0.107*** (0.0150)	−0.117* (0.0549)	−0.0422 (0.0539)
Per capita pulses consumption (ln)	0.156 (0.203)	−0.554** (0.200)	−0.360* (0.179)	−0.00156 (0.00558)	−0.0107 (0.0103)	−0.00614 (0.0243)
Per capita eggs consumption (ln)	0.190 (0.143)	−0.0543 (0.0583)	0.125 (0.0980)	0.0466* (0.0209)	0.169*** (0.0415)	−0.0346 (0.0590)
Per capita fish consumption (ln)	−0.0702 (0.212)	0.0375 (0.132)	−0.184 (0.153)	−0.164 (0.0950)	−0.0912 (0.161)	−0.215** (0.0863)
Per capita fruits consumption (ln)	0.188** (0.0743)	0.480** (0.162)	0.378*** (0.0925)	0.0801* (0.0362)	0.0918 (0.0532)	0.0304 (0.0560)
Per capita meat consumption (ln)	−0.0750 (0.0930)	−0.252 (0.222)	−0.173 (0.181)	0.379*** (0.111)	0.239 (0.145)	0.389*** (0.108)
Per capita milk consumption (ln)	0.0649 (0.103)	−0.0579 (0.124)	0.123 (0.115)	0.101*** (0.0236)	0.0313 (0.0322)	−0.00734 (0.0572)
Types of cooking fuels
Biomass (ln)	0.302*** (0.0944)	0.0744 (0.135)	0.0330 (0.219)	−0.0463 (0.106)	0.106 (0.0942)	0.0386 (0.117)
Charcoal (ln)	0.0130 (0.0130)	−0.0325** (0.0143)	−0.0147 (0.0125)	−0.00522 (0.00453)	−0.0130** (0.00489)	0.00445 (0.00346)
Coal (ln)	0.0257* (0.0138)	0.0101 (0.0151)	−0.0102 (0.0133)	−0.0142 (0.0109)	−0.00938 (0.00992)	−0.0265** (0.0110)
Electricity (ln)	−0.0391*** (0.00912)	0.0239** (0.00756)	−0.0402*** (0.00608)	0.00139 (0.0226)	0.0121 (0.0235)	−0.0107 (0.0255)
Gas (ln)	0.0239 (0.0283)	0.0185 (0.0170)	0.0258 (0.0228)	−0.0172 (0.0219)	−0.00971 (0.0239)	−0.0162 (0.0200)
Kerosene (ln)	0.0318*** (0.00879)	−0.00928 (0.0136)	0.0485*** (0.0117)	0.00206 (0.00463)	−0.00278 (0.00349)	−0.00414 (0.00526)
Lages
First lag
Fixed effects
Country	Yes	Yes	Yes	Yes	Yes	Yes
Time	Yes	Yes	Yes	Yes	Yes	Yes
Constant	−22.87 (15.65)	−15.43 (16.18)	−14.83 (15.46)	−33.13*** (6.159)	−40.53*** (6.250)	−35.94*** (7.809)
Observations	220	220	220	550	550	550
R-squared	.98	.96	.98	.99	.99	.99
F-statistics	11,429.5***	3,857.9***	237.3***	2,242.8***	377.22***	598.5***
Root MSE	0.26	0.27	0.26	0.13	0.17	0.15
Wooldridge test for serial correlation	60.3***	21.5**	75.4***	47.5***	71.5**	78.2***
Durbin-Wu-Hausman	234.14***	27.7*	15.1	30.7*	22.9	249.4***

Source. Computed by authors based on panel data from 170 countries. Standard errors in parentheses.

***

p < .01. **p < .05. *p < .1.

Results of the Fixed Effect Regressions

In this section, we report the findings from fixed-effect regressions that we conducted to examine the effects of economic growth, food prices inflation, and population growth on household food-related GHG emissions. The regression analysis is carried out for the whole panel (Table 3), for low-income countries and for higher-income countries (Table 4). We estimate the fixed effect models for each of the three alternative GHG emissions that we use: CO₂, N₂O, and CH₄ emissions. The regressions models are also estimated through system GMM given in Appendix Table A1; however, in term of significant coefficient the fixed effect regressions provide more reasonable results. Therefore, we interpret the results of the fixed effect regressions with time and country fixed effects given in Tables 3 and 4.

The last section of Tables 3 and 4 presents the findings of the diagnostic tests conducted on the nine fixed effect models. The R² of the nine models ranged from 0.96 to 0.99. We observed relatively higher R² for the nine models; however, it is common since we include country and time fixed effects to our models. The inclusion of time and country fixed effects reduces unexplained variance or residual sum of squares (RSS), leading to an increase in R². The F test’s findings shows that each of the nine regression models are statistically significant at 1% level indicates that the explanatory variables in the nine regression models contribute significantly to explaining variations in food-related GHG emissions. The range of the root mean square error (Root MSE) in the nine models is between 0.13 and 0.27, it means that the models provide better predictions, with fewer and smaller errors.

Income

We find that the impact of real per capita income on the CO₂, N₂O, and CH₄ emissions is quadratic for the whole sample (Table 3), that is, real per capita income has a significant positive effect on GHG emissions up to a certain level of income beyond which it has a significant negative effect on the GHG emissions. This result confirms the ECK hypothesis used by previous studies (Awan et al., 2022; Olale et al., 2018). This shows that in the early stages of economic growth, households may have limited resources and must rely on less efficient, resource-intensive agriculture and food production systems. Furthermore, low-income households may consume diets that are more reliant on resource-intensive food commodities like red meat and processed meals, which contributes to increased GHG emissions. As incomes rise beyond a certain threshold, households may start to adopt more sustainable food consumption habits. This could include adopting plant-based diets, reducing food waste, and increasing consumption of locally sourced and sustainably produced foods. These changes can result in a reduction in GHG emissions from household food consumption. Same result is found for high-income countries (Table 4). However, for low-income countries (Table 4), we observed the opposite results (U-shaped). This result challenges the EKC hypothesis in low-income countries, suggesting that they have not yet reached the income level where economic growth leads to reduced environmental degradation (Sirag et al., 2018; Twerefou et al., 2016). This shows that household in low-income countries still have limited resources and they may rely on agriculture food production systems. However, when the income of the household in low-income countries increases, they may start consuming diets that are more reliant on resource-intensive food commodities like red meat and processed meals, which contributes to increased GHG emissions. The EKC hypothesis results for the whole sample and for high-income countries found in this study confirms the results of Olale et al. (2018), who found an EKC relationship between GHG emissions and economic growth in Canada. Awan et al. (2022) tested the EKC hypothesis for the SAARC region. They found that at the earlier stages, of the economic growth CO₂ emissions increase in SAARC countries because most of the nations had a week financial and political system but with the passage of time emissions reduces as economy gets more stable.

Food Price

In Table 4, a one percent point increase in food prices increases the CO₂, N₂O, and CH₄ emissions from households’ food consumption by 0.14, 0.20, and 0.08 percent points, respectively. This shows that as food prices rise, households may choose cheaper food options, which frequently include products with a higher environmental footprint. Pang et al. (2021) found similar result for China in the long-run. They found that in the short-run, an increase in food prices will lead to a decrease in GHG emissions from food consumption; however, in the long-run, an increase in food prices will lead to an increase in GHG emissions. In the short run, as food prices rise, consumers may opt to temporarily limit food consumption, resulting in a decline in market demand, which influences supply and reduces GHG emissions. However, in the long run, food consumption will not decrease as living standards rise; rather, food consumption will rise further, promoting an increase in GHG emissions (Pang et al., 2021). However, the low-income countries exhibit different impact of food prices on GHG emissions. For low-income countries (Table 4), a one percent point increase in food prices increases the N₂O emissions from households’ food consumption by 0.43 percent points. N₂O is a potent greenhouse gas primarily emitted from agricultural activities, particularly due to the use of synthetic fertilizers, manure management, and soil cultivation. In low-income countries majority of the households are relied on agriculture food system; thus, any increase in the food prices in low-income countries can push back more household to rely on agriculture food system, which can further increase the N₂O emission from their consumption (World Bank, 2025).

Population

A one percent point rise in the population leads to a 1.9 percent points increase in CO₂ emissions and 1.8 percent points increase in both N₂O and CH₄ emissions from households’ food consumption (Table 3). Similarly, a one percent point rise in the population in low-income countries leads to a 3.3 percent points increase in CO₂ emissions, 3.4 percent points increase in N₂O emissions, and 2.8 percent point increase CH₄ emissions from households’ food consumption (Table 4). A one percent point rise in the population in high-income countries leads to a 1.2 percent points increase in N₂O emissions and 1.5 percent points increase in CO₂ and CH₄ emissions, respectively (Table 4). These results confirm the Ehrlich–Commoner model. The Ehrlich–Commoner model is a way to understand the environmental impact of human activities, which in our case is food consumption. It suggests that the adverse impact on the environment, which is the increase in GHG emissions from food consumption in our case is a result of three factors: population growth, food consumption per person, and technology (which refers to the environmental impact per unit of food consumption). In simple terms, the model explains that the more people there are, the more food they consume, and the more polluting the technology they use for food production, the greater the GHG emissions (Commoner, 2013; Ehrlich & Holdren, 1971). However, population growth in low-income countries is higher than in high-income countries, which is why the intensity of food-related GHG emissions is greater in low-income countries than in high-income countries (Pickson et al., 2024). Researchers like Mrówczyńska-Kamińska et al. (2021) argue that a significant increase in the global population leads to higher GHG emissions from household food consumption. As the population grows, the demand for food rises, leading to greater production, transportation, and consumption of food, all of which contribute to increased GHG emissions. Similarly, He et al. (2021) found that population expansion is a key driver of demographic transitions and the resulting increase in food consumption-based GHG emissions.

Per capita Food Consumption

A one percent point increase in per capita wheat consumption increases the CO₂ emissions from households’ food consumption by 0.31 percent points and N₂O and CH₄ emissions by 0.25 percent points, respectively (Table 3). This outcome is consistent with Kassem et al. (2022); they found that wheat consumption emits the highest amount of GHG among plant-based foods. However, this result is in contradiction with the findings of previous studies in which researchers found that per capita wheat consumption emits lower GHG (Chen & Zhong, 2022; Green et al., 2018). This discrepancy suggests differing assessments of the environmental impact of wheat, potentially due to variations in data, methodology, or regional differences in production practices and consumption patterns. Similar results are obtained for high-income countries, where a one percent point increase in per capita wheat consumption increases the CO₂ emissions from households’ food consumption by 0.30 percent points, N₂O emissions by 0.44 percent points, and CH₄ emissions by 0.32 percent points (Table 4). On the other hand, in low-income countries the wheat consumption increases the CO₂ and N₂O emissions by 0.28 and 0.12 percent points, respectively; while decreases the CH₄ emissions by 0.14 percent points (Table 4). These results indicates that high-income countries are primarily responsible for the increase in GHG emissions associated with wheat consumption, whereas low-income countries contribute comparatively less. It is reasonable because the average annual per capita wheat consumption in high-income countries is very high (87 kg) as compared to the average annual per capita wheat consumption in low-income countries (35 kg; FAO, 2024). Besides, a one percent point increase in per capita rice consumption decreases the N₂O emissions from households’ food consumption by 0.09 percent points (Table 3). Though, rice cultivation, still contributing to greenhouse gas emissions (CH₄); however, increased consumption of rice, which is often associated with a lower overall carbon footprint per calorie compared to more resource-intensive foods, might lead to a small reduction in the total N₂O emissions from household food consumption (Xiong et al., 2020).

Furthermore, a one percent point increase in per capita sugar consumption increases the CO₂ and CH₄ emissions from household food consumption by 0.14 and 0.15 percent points, respectively (Table 3). Similar trend is observed for low-income countries, where a one percent point increase in per capita sugar consumption increases the CO₂ and CH₄ emissions from household food consumption by 0.38 and 0.36 percent points, respectively (Table 4). However, in high-income countries, an opposite pattern emerges, where a one percent point increase in per capita sugar consumption leads to a 0.11 and 0.12 percent point reduction in CO₂ and N₂O emissions from household food consumption, respectively (Table 4). Agricultural and industrial processes involved in sugar production significantly contribute to CO₂ emissions (Smith et al., 2014). Additionally, the cultivation of sugarcane and sugar beets can contribute to CH₄ emissions indirectly through land-use changes and decomposition of organic matter in soils. As sugar consumption rises, it drives higher demand for these energy-intensive processes and agricultural practices, thereby increasing both CO₂ and CH₄ emissions (Xiong et al., 2020). Our results regarding the whole sample and low-income countries are against the findings of Kassem et al. (2022), who found that sugar consumption emitted the least amount of GHG. However, Kassem et al. (2022), analyzed emissions from specific consumption patterns, while the current findings highlight the GHG emissions from household food consumption at a larger scale. However, the results of the same study are in line with the findings for high-income countries. The higher GHG emissions from sugar consumption in low-income countries may not be directly related with the household consumption pattern but rather with the inefficient production methods used in low-income countries in sugar production. Low-income countries often rely on sugar beet farming practices, which involve higher energy use and excessive water consumption (Afrouzi et al., 2023).

A one percent point increase in per capita pulses consumption increases the CO₂ and CH₄ emissions from household food consumption by 0.03 and 0.06 percent points, respectively (Table 3). Although pulses are generally considered environmentally friendly due to their lower carbon footprint compared to animal-based proteins, their cultivation still contributes to GHG emissions. However, these increases are relatively small, indicating that shifts toward higher pulses consumption have a much lower environmental impact compared to many other food sources. This result is against the findings of Green et al. (2018) who found that per capita pulses consumption decreased GHG emissions. The discrepancy might stem from different research scopes, Green et al. (2018) focused on the impact of specific food consumption on emissions, whereas the current findings address emissions from household food consumption at a larger scale. However, in low-income countries, an opposite pattern emerges, where a one percent point increase in per capita pulses consumption leads to a 0.55 and 0.36 percent point reduction in N₂O and CH₄ emissions from household food consumption, respectively (Table 4). As compared to high-income countries, the annual average per capita pulses consumption in low-income countries are high (14 kg and 4 kg, respectively; FAO, 2024). As per the findings of Green et al. (2018) pulses are generally considered environmentally friendly due to their lower carbon footprint compared to animal-based proteins; therefore, shifts toward higher pulses consumption in low-income countries can further reduces the environmental impact associated with food consumption.

Eggs consumption significantly contributes to GHG emissions in higher-income countries. A one percent point increase in per capita eggs consumption increases the CO₂ and N₂O emissions from household food consumption by 0.05 and 0.16 percent points, respectively (Table 4). The greater affordability and dietary preferences in high-income countries leads to higher per capita consumption of animal-based proteins, including eggs (Poore & Nemecek, 2018). This increasing demand leads to higher production of eggs while eggs production is resource-intensive, which in turn raises GHG emissions (Lumsden et al., 2024). A one percent point increase in per capita fish consumption decreases the CO₂, N₂O, and CH₄ emissions from household food consumption by 0.12, 0.09, and 0.08 percent points, respectively (Table 3). This decrease is due to the relatively lower environmental impact of fish compared to other animal proteins. These findings are consistent with the findings of Silva et al. (2023). They found that, as compared to meat consumption, fish consumption has lower GHG emissions. Thus, switching from meat to low-carbon proteins like fish can be an alternative to lessening the effect of food consumption on GHG emissions (Silva et al., 2023). Almost similar findings are observed in high-income countries where a one percent point increase in per capita fish consumption decreases the CH₄ emissions from household food consumption by 0.22 percent points (Table 4). This result is important for policymakers in the high-income countries where the households consume higher amount of animal-based proteins, including meat, poultry, and eggs. The sifts of consumer preferences from meant and poultry to fish and seafood can reduce the food-related GHG emissions in high-income countries.

A one percent point increase in per capita fruit consumption increases the N₂O and CH₄ emissions from household food consumption by 0.03 and 0.07 percent points, respectively (Table 3). Fruits generally have a lower GHG footprint, their cultivation involves agricultural practices that can release N₂O and CH₄, particularly in soils with high organic content. Moreover, the production and transportation of fruits contribute to overall emissions, albeit less significantly compared to other foods. Thus, even though fruits are typically more environmentally friendly, their increased consumption can still lead to a slight rise in N₂O and CH₄ emissions due to these factors in the food supply chain (Xiong et al., 2020). Results from the low-income countries confirmed this explanation where a one percent point increase in per capita fruit consumption increases the CO₂, N₂O, and CH₄ emissions from household food consumption by 0.18, 0.48, and 0.37 percent points, respectively (Table 4). Although in most of the low-income countries’ fruits are consider luxury food items and the households with lower incomes barely able to consume fruits in their daily diets (Akram, 2020). However, due to fertile lands the low-income countries produced a significant number of fruits like banana, mangoes, and pineapples. Therefore, instead of fruit consumption, fruit production is the major cause of GHG emissions in low-income countries. Conversely, in high-income countries a one percent point increase in fruit consumption increases the CO₂ emission by 0.08 percent points. In high-income countries fruits are consider necessity food items and often households with higher incomes includes fruits in their daily diets which can further raise the carbon footprint of the fruits’ consumption.

Moreover, a one percent point increase in per capita meat consumption increases the CO₂ and N₂O emissions from household food consumption by 0.18 and 0.16 percent points, respectively. Almost similar emissions pattern is observed for high-income countries where a one percent point increase in per capita meat consumption increases the CO₂ and CH₄ emissions from household food consumption by 0.37 and 0.38 percent points, respectively. The substantial GHG emissions associated with meat production highlight the environmental cost of increased meat consumption and underscore the potential benefits of reducing meat intake for mitigating food-related GHG emissions. These results align with previous studies that identified meat consumption as a leading source of food-related GHG emissions (Chen & Zhong, 2022; Green et al., 2018; Kassem et al., 2022; Silva et al., 2023). It suggests that to reduce GHG emissions, especially in high-income countries, economic measures such as increasing the price of meat and decreasing the price of fish products could be effective. These adjustments in pricing could encourage consumers to shift toward more sustainable food options, thereby lowering overall emissions (Chen & Zhong, 2022).

A one percent point increase in per capita milk consumption increases the CO₂ emissions from household food consumption by 0.03 percent points. Similarly, a one percent point increase in per capita milk consumption in high-income countries increases the CO₂ emissions from household food consumption by 0.10 percent points. As compared to low-income countries, high-income countries typically have greater dairy consumption per capita. Therefore, increased demand for dairy products in high-income countries leads to higher CO₂ emissions from dairy farming, processing, transportation, refrigeration, and milk products preparations (Mazzetto et al., 2022). Dairy farming in the high-income countries is basically carried out at large-scale with energy-intensive technologies, such that mechanized milking, automated feeding, and temperature-controlled storage, all of which contribute to CO₂ emissions (Sorley et al., 2024).

Types of Cooking Fuels

A one percent point increase in population using biomass for cooking decreases the CO₂ and CH₄ emissions from households’ food consumption by 0.01 percent points, respectively. This minor reduction can be attributed to the fact that biomass fuels, such as wood and agricultural residues, often involve combustion processes that emit CO₂ and CH₄. However, when biomass is used in more efficient stoves or cooking practices, the overall corban and methane emissions can be lower compared to traditional open fires or inefficient stoves. Furthermore, biomass fuels can be less associated with enteric fermentation, which is a significant source of carbon and methane in livestock production. According to Bailis et al. (2003), improved biomass cookstoves can reduce emissions compared to traditional methods. The studies carried out by Berkouwer and Dean (2023) and Kaur-Sidhu et al. (2020) show that adopting clean cookstoves can lower CO₂-equivalent emissions by 3.9 tons per household annually. Therefore, they pointed out that transitioning to improved stoves or biogas systems helps minimize CO₂ and CH₄ emissions, promoting environmental sustainability, and better indoor air quality (Berkouwer & Dean, 2023; Kaur-Sidhu et al., 2020). However, in the low-income countries we observed opposite situations where a one percent point increase in population using biomass for cooking increases the CO₂ emissions from households’ food consumption by 0.3 percent points. Majority of the households in low-income countries use biomass in traditional open fires or in inefficient stoves for cooking, very few households have access to modern cooking stoves which leads to more CO₂ emissions (Baul et al., 2022).

A one percent point increase in population using charcoal for cooking increases the CO₂, N₂O and CH₄ emissions from households’ food consumption by 0.008, 0.01, and 0.006 percent points, respectively. Charcoal combustion releases CO₂, N₂O, and CH₄ as byproducts. When charcoal is used for cooking, the combustion process emits these greenhouse gases into the atmosphere, which further increases the GHG emissions from household food consumption. The results of Bailis et al. (2003) are in agreement with this outcome; they found that charcoal stoves have 6 to 13 times larger emissions than woodstoves. However, the results are in contradiction with this outcome Singh et al. (2014). They concluded that kerosene, biogas, and charcoal for rural areas in India have the lowest environmental footprint. This difference could be due to variations in study methodologies, fuel types, or regional conditions affecting the environmental impact of these fuels. As per the estimations of Bhattacharya et al. (2002) and Pennise et al. (2001), charcoal combustion emits 2,155 to 2,567 g of CO₂ per kg burned, this is much higher than combustion of wood per kg burned. However, N₂O emissions from combustion of charcoal are lower, ranging from 0.011 to 0.30 g per kg of charcoal (Bhattacharya et al., 2002; Pennise et al., 2001). For low and high-income countries, we observed different results. In low-income countries, a one percent point increase in population using charcoal for cooking decreases the N₂O emissions from households’ food consumption by 0.003 while it increases N₂0 emissions by 0.01 percent points in and high-income countries, respectively. These results are in line with the outcomes of Bhattacharya et al. (2002) and Pennise et al. (2001), who found that N₂O emissions from combustion of charcoal are lower (Bhattacharya et al., 2002; Pennise et al., 2001). However, in low-income countries majority of the population mainly use polluting fuels for cooking which can increase the CO₂ emissions but can reduce the N₂O emissions caused by using charcoal for cooking foods (Stoner et al., 2021). On the other hand, in high-income countries the households are shifting from traditional cooking fuels to modern cooking fuels like gas and electricity; therefore, the N₂O emissions from household food consumption gradually falls (Stoner et al., 2021).

Similarly, a one percent point increase in population using coal for cooking increases the CO₂ and CH₄ emissions from households’ food consumption by 0.02 and 0.007 percent points, respectively. Like charcoal, coal combustion releases CO₂ and CH₄ as byproducts. When coal is used for cooking, the combustion process emits these gases into the atmosphere, contributing to the observed increase in households’ food emissions. According to the World Health Organization (WHO), coal stoves not only emit GHG but it also emits other particulate matter (PM2.5), sulfur dioxide (SO₂), and carbon monoxide (CO), leading to respiratory diseases and premature deaths. The WHO warns against using polluting fuels due to their environmental and health impacts and suggested for household to use sustainable alternatives for cooking include electric cooking, LPG, biogas, improved biomass stoves, and solar cookers, which can reduce emissions and indoor air pollution (WHO, 2018). The outcomes of M. S. B. Khan and Lohano (2018) also indicating the severity of using polluting fuels. They found that children in households using polluting fuels are 1.5 times more likely to have symptoms of acute respiratory infection (ARI) than children in households using cleaner fuels (M. S. B. Khan & Lohano, 2018). Comparing the low and high-income countries, it is observed that a one percent point increase in population using coal for cooking increase the CO₂ emissions from households’ food consumption by 0.03 percent points in low-income countries while reduces the CH₄ emissions by 0.03 percent points in high-income countries. As compared to high-income countries, the households in low-income countries still cook their meals over open fires or on basic stoves, breathing in harmful smoke released from burning coal. Recently, a progress in Asia and Latin America regarding using clean energy fuels for cooking is observed; however, the number of people without access to clean cooking fuels has never stopped growing in low-income countries. Therefore, the CO₂ emissions from household food consumption still a problem in low-income countries (IEA, 2023).

A one percent point increase in population using electricity for cooking increases the N₂O emissions from households’ food consumption by 0.02 percent points while decreases the CO₂ and CH₄ emissions by 0.009 and 0.01 percent points, respectively. A similar pattern is emerged for low-income countries where a one percent point increase in population using electricity for cooking increases the N₂O emissions from households’ food consumption by 0.02 percent points while decreases the CO₂ and CH₄ emissions by 0.04 percent points, respectively. This pattern arises because while electricity use for cooking generally reduces CO₂ and CH₄ emissions compared to biomass or fossil fuels, due to fewer direct emissions from combustion, there is a slight increase in N₂O emissions. This increase can be attributed to the broader electricity generation mix, which may include fossil fuels and other sources that contribute to N₂O emissions during production. Therefore, while switching to electricity for cooking reduces direct GHG emissions from cooking itself, the overall impact depends on the energy mix used to generate the electricity (Im & Kim, 2020). This result is in line with the results of Aemro et al. (2021) who concluded that, compared with traditional wood-fired cookstoves, electric cookstoves can reduce energy consumption in Sub-Saharan Africa by 95.7% and CO₂ emissions by almost 100%. This result is consistent with the findings of Ramirez et al. (2017), who reported that reduction in carbon footprint takes place when households shift from traditional LPG to a predominantly hydroelectric-powered cooking system. However, the carbon footprint may increase when cooking relies on fossil fuel–derived electricity, which can occur during periods of peak electricity demand (Ramirez et al., 2017).

A one percent point increase in population using gas for cooking decreases the CO₂ and CH₄ emissions from households’ food consumption by 0.04 and 0.05 percent points, respectively. Stoves using cleaner fuels like natural gas or LPG are considered more efficient than those using solid fuels like charcoal or coal. Efficient combustion from gas stoves results in lower emissions of CO₂ and CH₄ per unit of energy produced, contributing to the observed decrease in emissions from household food consumption. Similarly, natural gas and LPG are cleaner-burning fuels compared to solid fuels such as charcoal and coal. They contain fewer impurities and produce fewer emissions of CO₂ and CH₄ when combusted, leading to a lower environmental impact associated with household cooking activities. This result is consistent with the findings of Permadi et al. (2017), who found that Indonesia's national policy of eliminating kerosene use in cooking had successfully replaced kerosene with LPG in 30 designated provinces. Consequently, the net GHG emissions from household cooking would reduce by 2% in 3 years. This result is also consistent with the results of Singh et al. (2014), who concluded that LPG and biogas for urban areas in India have the lowest environmental footprint.

Finally, a one percent point increase in population using kerosene for cooking decreases the N₂O emissions from households’ food consumption by 0.01 percent points. Kerosene emits less N₂O compared to solid fuels like charcoal or coal. Its cleaner combustion characteristics contribute to the observed decrease in households’ food emissions. Besides, kerosene stoves are typically designed to burn the fuel efficiently, minimizing waste and maximizing heat output. Efficient combustion reduces the amount of fuel needed for cooking tasks, thereby lowering household emissions of N₂O per unit of energy produced. This result is in line with previous study in which researchers concluded that kerosene, biogas, and charcoal for rural areas in India have the lowest environmental footprint (Singh et al., 2014). However, in low-income countries a one percent point increase in population using kerosene for cooking increases the CO₂ and CH₄ emissions from households’ food consumption by 0.03 and 0.05 percent points, respectively. According to Lam et al. (2012), kerosene remains a common cooking fuel in many low-income countries, particularly cities where fuelwood and other biomass has to be bought, with electricity and LPG being costly or irregular. A variety of kerosene stove designs exist for households in these regions, although they are generally divided into two main kinds contingent on how the fuel is combusted: wick stoves, which operate through the tube transmission of fuel, and the more effective, hotter-burning pressure stoves equipped with vapor-jet nozzles that aerosolize the fuel via physical pumping or heat. In economically vulnerable households, wick stoves are more common as these stoves are inexpensive, can readily deliver fester heat for essential foods, and lack nozzles that might become closed with soot. However, these stoves release significant quantities of fine particulates, CO₂, CH₄, nitrogen oxides (NO₂), and SO₂, thereby increasing household food-related emissions (Lam et al., 2012).

Conclusion and Recommendations

This study determines the effects of economic growth, food price inflation, and population explosion on food consumption related GHG emissions by employing panel data from 170 countries. Results from the two-way fixed effect regression supported the ECK hypothesis. However, the ECK hypothesis is hold for high-income countries but not for low-income countries. Moreover, an increase in food prices and population as well as an increase in per capita wheat, sugar, pulses, fruits, meat, and milk consumption increases the GHG emissions from households’ food consumption whereas the per capita rice and fish consumption decreases the GHG emissions. Furthermore, an increase in population using biomass, gas, and kerosene for cooking decreases the GHG emissions from households’ food consumption whereas an increase in population using charcoal, coal, electricity for cooking increases the GHG emissions.

The EKC hypothesis is hold in high-income countries, in low-income countries, the EKC hypothesis still does not hold. This placed duty on the national governments, rural support programs, and foreign donor organizations to step up efforts in low-income countries to combat poverty and provide employment and income-boosting strategies to improve household purchasing power and economic conditions in these countries. These national programs have the potential to increase household income levels and make it easier for people to purchase and consume environmentally friendly food products. Furthermore, an increase in food prices increases the household food consumption related GHG emissions. Therefore, efficient food prices control policies at national level especially in low-income countries are needed in this regard. Besides, the results that an increase in population increase GHG emissions confirms the Ehrlich–Commoner theory. The theory is hold for both low and high-income countries. The rapid population growth is considering a serious problem in low-income countries. Therefore, various population control measures may reduce the GHG emissions from household food consumption in these countries. By promoting family planning and increasing access to education, especially for women, low-income countries can manage population growth and subsequently decrease the demand for food, which in turn lowers emissions associated with food consumption.

Besides, an increase in per capita rice and fish consumption decrease the GHG emissions from food consumption whereas an increase in per capita wheat, sugar, pulses, fruits, meat, and milk consumption increase the GHG emissions. Therefore, switching out meat for low-carbon proteins like fish; and wheat, sugar, pulses, fruits, and meat for rice, vegetables, eggs, beans, lentils, peas, oats, barley, and plant-based milk will help reduce the impact of food consumption on GHG emissions. By increasing the price of wheat, sugar, pulses, fruits, and meat while lowering the price of fish, rice, vegetables, eggs, beans, lentils, peas, oats, barley, and plant-based milk, can further reduce GHG emissions. This can be done with the imposition of emission tax on consumption of wheat, sugar, pulses, fruits, and meat. Imposing an emission tax at the retail level can motivate consumers to move from high-emission food products to lower-emission replacements. However, the above-mentioned policies will be helpful only in both low and high-income countries by keeping the different food consumption and GHG emissions pattern persists in these countries. Another viable technique for influencing households’ food preferences is a non-monetary enticement, that might be helpful in high-income countries. Eco-labeling can tell consumers about the emissions produced by specific products. Consumers who care about the environment may choose to respond to such knowledge by limiting their demand for high-emissions products.

Finally, solid fuels for cooking like coal and charcoal increases food related GHG emissions whereas natural gas and LPG reduce food-related GHG emissions. Keeping in mind the varying GHG emissions intensity of different cooking fuels, policymakers especially in low-income countries must focus on promoting the adoption of cleaner and sustainable cooking fuels and technologies. In this regard, governments in low-income countries can offer subsidies or financial incentives to encourage households to shift from solid fuels like coal and charcoal to cleaner alternatives such as natural gas and LPG. International cooperation and partnerships aimed at promoting cleaner fuels and technologies for cooking will also be beneficial in this regard. A strong collaboration between governments, international organizations, non-governmental organizations (NGOs), and other stakeholders can facilitate the adoption of cleaner fuels and technologies for cooking at household level in the diverse communities.

While this study contributes valuable insights into the GHG emissions associated with household food consumption, it is important to acknowledge some limitations that offer avenues for future research. Although GHG emissions from household food consumption are included in this study, there are other sources of GHG emissions related to food production, processing, packaging, and transportation. Therefore, incorporating GHG emissions from food production, processing, packaging, and transportation would be a valuable direction for future research. Similarly, while the current study calculated GHG emissions from household food consumption at the global level, focusing on low and high-income countries, future research could explore these emissions at the country, regional, and sub-regional levels, as well as for lower-middle and upper-middle-income countries.