Following the World Health Organization’s (WHO) declaration of COVID-19 as a global pandemic on 11 March 2020, countries worldwide implemented lockdown measures to mitigate its spread (WHO, 2020; Singh and Tembo, 2022). In South Africa, these measures included the closure of schools and workplaces, restrictions on non-essential activities, bans on social gatherings, and limitations on movement. The food service industry in particular was subject to regulations such as social distancing, curfews, and restricted opening hours (South African Government, 2020). These restrictions significantly altered household food expenditure patterns, shifting spending from food-away-from-home (FAFH) to food-at-home (FAH)(1).
The decline in FAFH expenditure can be attributed to both macro- and micro-level impacts of lockdown measures. At the macro level, the closures of schools, workplaces, and event venues, coupled with the cancellation of large gatherings, reduced opportunities for FAFH consumption in these settings. At the micro level, social distancing, curfews, and restricted operating hours further limited FAFH consumption at restaurants and other eateries (Barrientos-Gutiérrez et al., 2020). As a result, households redirected their spending toward FAH, in line with global consumption shifts in consumption patterns observed during the pandemic.
Globally, the consumption of FAH rose due to restaurant closures, movement restrictions, and heightened health concerns. As a result, many consumers shifted toward purchasing foods with longer shelf lives. For example, in Italy, Denmark, Germany, and Slovenia, consumption of fresh produce and meat declined, while demand for frozen and shelf-stable foods surged as consumers prioritized longer-lasting items (Fanelli, 2021; Janssen et al., 2021). Similarly, in the United States, fast food consumption dropped, while home-prepared meals became more common (Chenarides et al., 2020). In contrast, Qatar exhibited a different trend, with an increase in the consumption of fresh produce and dairy products during the pandemic (Hassen et al., 2020).
Mirroring global trends, South Africa saw a decline in FAFH consumption, particularly in fast food, as restaurants and other FAFH establishments closed during the hard lockdown (Mkasi and Lefophane, 2023). In response, households increasingly turned to home-prepared meals, and the consumption of staple foods grew as consumers prioritized products with longer-lasting items to minimize shopping trips. This shift toward home cooking during the lockdown reflected similar patterns seen in countries like Italy, Denmark, and Germany (Fanelli, 2021; Janssen et al., 2021).
However, the intake of fresh produce declined as restrictions on informal trading curbed households’ access to fresh food (Mkasi and Lefophane, 2023). Additionally, dairy and meat consumption fell, largely due to price increases during the pandemic (Stats SA, 2021; Fin24, 2021). These shifts highlight the profound impact of the pandemic on food consumption patterns, with long-term consequences for both household diets and the food service industry.
Empirical studies have identified four key shifts in food purchasing, consumption, and expenditure patterns among households during the COVID-19 pandemic. First, to minimise shopping trips and avoid long queues at grocery stores, households stockpiled food due to concerns about contracting the virus (Hassen et al., 2020; Pappalardo et al., 2022; Truong and Truong, 2022). Second, while spending on FAFH decreased, online grocery shopping saw a notable increase (Ellison et al., 2021). Third, pandemic-induced price shocks reduced both the likelihood of spending on food and the tendency to purchase FAFH (Codjia and Saghaian, 2022). Finally, there was a clear shift from FAFH to FAH, reflected in a rise in FAH sales and demand, alongside a decline in FAFH sales and expenditure. This shift occurred as FAFH establishments were closed for being non-essential, while grocery stores remained open as essential services (Hobbs, 2020; Beckman and Countryman, 2021; Marchesi and McLaughlin, 2022; Codjia and Saghaian, 2022).
While these studies offer valuable insights into food expenditure behaviour during the pandemic, they primarily focus on periods of strict restrictions and high virus transmission. Consequently, household spending patterns likely shifted with the easing of lockdown measures, vaccination rollouts, and a decline in COVID-19 cases. For example, data from Statistics South Africa revealed increased spending on restaurants and hotels, along with a revenue rebound in takeaway and fast-food outlets after the lockdown (Pawson, 2023). This indicates that households may have shifted some of their grocery spending back to FAFH as eateries reopened at full capacity.
Nevertheless, these observations remain tentative, highlighting the need for empirical analysis to better understand FAFH behaviour post-COVID-19 and the factors that drive it. Previous studies (Labadarios et al., 2011; D’Haese et al., 2013; Thamaga-Chitja and Morojele, 2014) have largely overlooked post-lockdown trends, rural household behaviour, and socio-economic factors. Blick et al. (2018) focused primarily on pre-pandemic FAFH expenditure determinants at the national level using time-series data but did not address post-lockdown changes, rural dynamics, or a detailed socio-economic analysis. This reveals a gap in the literature that requires further investigation.
This study aims to address this gap by examining FAFH expenditure among rural households in South Africa after the COVID-19 lockdown restrictions. It contributes to the existing literature in three key ways: First, it focuses on the rural context, acknowledging that the impact of lockdown restrictions varies across geographic areas over time (Richards and Rickard, 2020; Sheth, 2020; Ko et al., 2023). Second, it provides a more comprehensive analysis by incorporating 16 socio-economic factors, overcoming the limitations of previous studies, such as Blick et al. (2018), which relied on national time-series data and only included seven socio-economic variables. Third, the study uses cross-sectional data to capture the rapid shifts in consumption behaviour brought on by the pandemic (Ellison et al., 2021), enabling a more nuanced understanding of FAFH expenditure changes among rural households post-lockdown.
In sum, this study contributes empirical evidence on whether FAFH expenditures among rural households in South Africa have rebounded post-lockdown and identifies the key factors driving these changes. By doing so, it enhances the understanding of rural household food economics in the post-pandemic landscape.
This paper is organized as follows: The next section reviews the lockdown restrictions imposed on South Africa’s food service industry and their effects on households’ food purchasing, consumption, and expenditure patterns. The following section details the study area, justifies the focus on rural households, and outlines the data collection and sampling methods, while also exploring the application of the double-hurdle model in the empirical analysis. The subsequent section presents and interprets the descriptive and empirical findings. Finally, the paper concludes with key insights, implications, and recommendations.
South Africa’s lockdown restrictions were implemented under the “National State of Disaster,” declared on 15 March 2020 (ENSafrica, 2020). A five-level “COVID-19 Alert System” was introduced, ranging from Alert Level 5 (starting 26 March 2020) to Alert Level 1 (ending 5 April 2022). Alert Level 5 (26 March–30 April 2020), the most stringent, imposed severe movement restrictions, including a ban on inter-provincial travel, non-essential activities, social gatherings, and street vending. Under this level, schools, non-essential workplaces, and FAFH establishments (such as restaurants and fast-food outlets) were closed, as they were deemed non-essential (South African Government, 2020; Mkasi and Lefophane, 2023). In contrast, grocery stores, classified as essential, were permitted to remain open but prohibited from selling prepared food, alcohol, and snacks (Mkasi and Lefophane, 2023). As a result, these measures dampened FAFH expenditure while driving an increase in FAH spending, as households were limited to purchasing groceries for home preparation.
During Alert Level 4 (1 May–31 May 2020), FAFH establishments were allowed to reopen for online ordering and delivery, although operating hours, curfews, and a ban on alcohol sales remained. Financial viability became a concern as delivery revenue failed to cover operational costs. For example, Nando’s initially hesitated to reopen due to anticipated financial losses (Lefophane, 2020). Despite these challenges, chains such as Nando’s, Chicken Licken, Debonairs, KFC, McDonald’s, and Steers reopened selected outlets (Reuters, 2020). Household spending on FAFH likely increased but remained constrained by ongoing restrictions.
Alert Level 3 (1 June–17 August 2020) saw further easing, with FAFH establishments allowed to offer sit-down services alongside delivery. The relaxation of movement restrictions and resumption of most economic activities likely boosted FAFH expenditure. By Alert Level 2 (18 August–20 September 2020), most economic activity restrictions were lifted, curfews were shortened, and movement restrictions eased, further stimulating FAFH spending. Finally, Alert Level 1 (21 September 2020–28 December 2020) removed curfews and most COVID-19 regulations, culminating in the termination of the National State of Disaster on 5 April 2022.
With the full reopening of FAFH establishments, fast-food and takeaway outlets saw a rapid recovery, with FAFH expenditure rising 41% above pre-COVID-19 levels from 2019 (Pawson, 2023). This sharp rebound signals a strong recovery in FAFH spending post-lockdown. As a result, many of the COVID-19-related factors that previously influenced FAFH expenditure may no longer be relevant. Therefore, it is crucial to examine the current factors influencing FAFH expenditure in South Africa during the post-lockdown period.
The study was conducted in Ga-Molepo village, located within the Polokwane Local Municipality, approximately 52.6 km from Polokwane, the capital of Limpopo Province. South African fast-food chains have increasingly targeted consumers in townships and rural areas, expanding beyond traditional urban centres. Consequently, several fast-food outlets are now accessible in nearby Mankweng Township (24.0 km from Ga-Molepo) and Paledi Mall (23.1 km from Ga-Molepo). Table 1 presents the top 12 fast-food chains in South Africa by the number of stores. While these fast-food outlets are not currently operating in Ga-Molepo, they are easily accessible to households in the village, particularly in Mankweng, Paledi Mall, and Polokwane.
Top fast-food chains in South Africa
| No. | Fast food chain (owner) | No. of stores |
|---|---|---|
| 1 | KFC (Yum! Brands) | 1,057 |
| 2 | Debonairs (Famous Brands) | 710 |
| 3 | Steers (Famous Brands) | 652 |
| 4 | Wimpy (Famous Brands) | 458 |
| 5 | McDonald’s (McDonald’s Corp.) | 377 |
| 6 | Nando’s (Nando’s Group) | 302 |
| 7 | Chicken Licken (Golden Fried Chicken Pty Ltd) | 286 |
| 8 | Roman’s Pizza (Roman’s Catering Enterprises) | 250 |
| 9 | Fishaways (Famous Brands) | 242 |
| 10 | Hungry Lion (Hungry Lion Fast Food Pty Ltd) | 204 |
| 11 | Kauai (Real Foods) | 200 |
| 12 | Burger King (Restaurant Brands International) | 130 |
Source: BusinessTech, 2024.
For this study, FAFH is specifically defined to include only fast-food outlets listed in Table 1, reflecting the unique motivations for fast-food consumption, which prioritize convenience, affordability, and speed. This definition excludes other types of eateries, such as sit-down restaurants, pubs, bars, and cafes, which are typically chosen for social dining or leisurely experiences that often involve higher spending and cater to different consumer demographics.
Primary data were collected through face-to-face interviews conducted between November 2022 and January 2023, after the lifting of COVID-19 lockdown restrictions. The study used a simple random sampling method to select respondents from a sampling frame of households in Ga-Molepo. A list of potential respondents was provided by the tribal authority of Ga-Molepo, which maintains household records for the area. A total of 360 households were selected from a population of 4,048 households, using Yamane’s (1967) sample size formula. Household numbers were written on slips of paper and randomly drawn for interviews. Respondents were individuals responsible for food consumption decisions within the household, including purchasing, cooking, preparing, and serving food. This approach ensured that the data accurately reflected the perspective and behaviours of the primary food decision-makers.
This study used a double-hurdle model to analyse rural households’ participation in and spending on FAFH. The double-hurdle model assumes that households make two separate decisions when it comes to purchasing FAFH: a participation decision (whether to purchase FAFH) and an expenditure decision (how much to spend on FAFH). Each decision is influenced by a distinct set of explanatory variables. To observe a positive level of expenditure, households must pass through both hurdles. Accordingly, two latent variables were used to model each decision process: a binary choice model (Probit model) for the participation decision and an uncensored model (OLS model) for the expenditure decision (Blundell and Meghir, 1987).
The first hurdle involves the decision to participate in FAFH. Respondents were asked: “Did you or any of your family members purchase FAFH post-lockdown restrictions?” The participation decision can be modelled as:
In line with previous studies (Aguiar and Hurst, 2005; Gao et al., 2021; Davis and You, 2009; Nguyen and Nguyen, 2023), respondents were asked the following question after their participation decision: “How much money did you or any of your family members spend on FAFH per month after the lockdown restrictions were lifted?” If a household decided to participate in FAFH (i.e. y1 = 1), the second decision hurdle model concerns the level of expenditure. The expenditure decision is modelled as follows:
Households that did not spend on FAFH were reported to have zero expenditures. Based on Angulo et al. (2002), zero expenditure can be attributed to three factors: (1) households were unable to afford FAFH at prevailing prices and income levels (i.e., corner solutions), (2) the survey period may have been too short to capture the expenditure amount (i.e., infrequent purchases), or (3) households may not have had an interest in purchasing FAFH. The double-hurdle model is appropriate for this study because it accounts for zero expenditures while modelling two separate decision processes for households.
An alternative to the double-hurdle model is the Heckman two-step model (Cragg 1971; Heckman, 1979). However, the Heckman model is not suitable for this study, as it assumes a quasi-absence of zero expenditure in the second hurdle once the first hurdle is passed (Codjia and Saghaian, 2022). Therefore, the double-hurdle model was employed in this study.
This section presents the descriptive statistics for rural households, providing an overview of the demographic and socio-economic characteristics of the sample population. These characteristics serve as explanatory variables that may influence both participation in and expenditure on food away from home (FAFH) in the double-hurdle model analysis. Table 2 shows these explanatory variables, along with their categories, frequencies, and percentages for categorical variables, and the minimum, maximum, mean, and standard deviation for continuous variables.
Descriptive statistics for explanatory variables
| Variable | Outcome | Frequency (N) | Percentage (%) |
|---|---|---|---|
| 1 | 2 | 3 | 4 |
| Age | Less than 18 | 0 | 0 |
| 18–35 | 62 | 17.2 | |
| 35–60 | 180 | 50.0 | |
| Above 60 | 118 | 32.8 | |
| Gender | Male | 168 | 46.7 |
| Female | 192 | 53.3 | |
| Marital status | Single | 171 | 43.0 |
| Married | 124 | 38.9 | |
| Divorced | 37 | 10.3 | |
| Widowed | 28 | 7.8 | |
| Level of education | No formal education | 97 | 26.9 |
| Primary education | 99 | 27.5 | |
| Secondary education | 106 | 29.4 | |
| Tertiary education | 58 | 16.1 | |
| Employment status | Unemployed | 215 | 59.7 |
| Full-time employed | 119 | 33.1 | |
| Part-time employed | 26 | 7.2 | |
| Source of income | Business | 22 | 6.1 |
| Grant | 131 | 36.4 | |
| Salary | 94 | 26.1 | |
| Pension | 35 | 9.7 | |
| Farm income | 11 | 3.1 | |
| Other income sources | 67 | 18.6 | |
| Household income level | < ZAR 5000 | 196 | 54.4 |
| ZAR 5000–8000 | 84 | 23.3 | |
| ZAR 8000–10 000 | 47 | 13.1 | |
| > ZAR 10 000 | 33 | 9.2 | |
| Household structure | Single parent | 120 | 33.3 |
| Extended family | 27 | 7.5 | |
| Grandparent family | 110 | 30.6 | |
| Both parents’ family | 103 | 28.6 | |
| Changes in household income | Decreased | 24 | 7.5 |
| Increased | 66 | 19.2 | |
| Constant | 270 | 73.3 | |
| Social grants | Yes | 113 | 31.4 |
| No | 247 | 68.6 | |
| Type of sector | Formal | 60 | 41.4 |
| Informal | 85 | 58.6 | |
| Employment sector | Agriculture | 25 | 17.2 |
| Government | 40 | 27.6 | |
| Education | 30 | 20.7 | |
| Health | 20 | 13.8 | |
| Retail | 15 | 10.3 | |
| Finance | 08 | 5.5 | |
| Others | 07 | 4.8 | |
| Continuous variables | ||||
|---|---|---|---|---|
| Variable | Minimum | Maximum | Mean | St. deviation |
| Household size | 1 | 14 | 6 | 3.25 |
| Number of employed people in the household | 0 | 8 | 1 | 2.00 |
| Households’ members receiving social grant | 0 | 6 | 1 | 1.53 |
| Households’ members receiving SRD grant | 0 | 3 | 1 | 1.50 |
Note: R18.46 = 1 USD (average exchange rate during the data collection period, November 2022 to January 2023)
Source: own elaboration.
The results show that the largest proportion of respondents (50%) were aged between 35 and 60 years, while a smaller proportion (17.2%) were in the 18–35 age group. This confirms that the study focused exclusively on the adult population, with no respondents under 18 years of age. In terms of gender, females represented a slightly higher proportion (53.3%) than males (46.7%).
Regarding marital status, the largest group of respondents were single (43%), followed by those who were married (38.9%). Smaller proportions were divorced (10.3%) or widowed (7.8%). Concerning educational attainment, 26.9% of respondents had no formal education, while 27.5% completed primary education, 29.4% had secondary education, and only 16.1% attained tertiary education.
Employment status revealed that the majority of respondents were unemployed (59.7%), while 33.1% were employed full-time and 7.2% were employed part-time. Among those with a source of income, social grants (36.4%) were the most common, followed by salaries (26.1%). Other income sources included pensions (9.7%), business activities (6.1%), and farm income (3.1%), while 18.6% reported earnings from unspecified sources.
Household income distribution showed that the majority (54.4%) earned less than ZAR 5,000 per month, with 23.3% earning between ZAR 5,000 and ZAR 8,000. Smaller proportions earned between ZAR 8,000 and ZAR 10,000 (13.1%) or more than ZAR 10,000 (9.2%) per month. The average income was approximately ZAR 5,748 per month, with a median income of ZAR 6,500. This suggests that half of the households earned less than ZAR 6,500 per month, while the other half earned more.
Regarding household structure, 33.3% of respondents lived in single-parent households, 30.6% in grandparent-headed households (where one or both grandparents act as primary caregivers for their grandchildren), 28.6% in two-parent households, and 7.5% in extended family households (including additional relatives, such as aunts, uncles, or cousins).
Income stability data revealed that 73.3% of households reported stable income, 19.2% experienced an increase and 7.5% reported a decrease. Additionally, 31.4% of respondents received social grants, while 68.6% did not. The number of household members receiving social grants varied from zero to six, with an average of one per household.
Employment sector data showed that 58.6% of employed respondents worked in the informal sector, while 41.4% were employed in the formal sector. Among these, 27.6% worked in public administration, 20.7% in education, 17.2% in agriculture, 13.8% in health, 10.3% in retail, 5.5% in finance, and 4.8% in other sectors.
The household size ranged from one to 14 members, with an average of six people per household. The number of employed individuals per household ranged from zero and eight, with an average of one. Similarly, the number of household members receiving social relief of distress (SRD) grants ranged from zero to three, with an average of one(2).
Table 3 presents the participation rates (first hurdle) and expenditure patterns (second hurdle) in food-away-from-home (FAFH) during and after the COVID-19 lockdown restrictions.
Participation and expenditure results
| During COVID-19 lockdown restrictions (First hurdle) | |||
|---|---|---|---|
| Variable | Outcome | Frequency (N) | Percentage (%) |
| Did you purchase FAFH during COVID-19 lockdown restrictions? | Yes | 107 | 29.7 |
| No | 253 | 70.3 | |
| Did you purchase FAFH post-COVID-19 lockdown restrictions? | Yes | 186 | 58.1 |
| No | 174 | 41.9 | |
| Post COVID-19 lockdown restrictions (Second hurdle) | ||||
|---|---|---|---|---|
| Variable | Minimum | Maximum | Mean | St. deviation |
| Amount spent on FAFH during lockdown restrictions | 0 | 250 | 136,07 | 53.95 |
| Amount spent on FAFH post lockdown restrictions | 0 | 300 | 149,25 | 59,625 |
Note: Average exchange rate during data collection (Nov 2022–Jan 2023): R18.46 = 1 USD
Source: own elaboration.
To address the first hurdle, respondents were asked whether they had purchased FAFH during and after the lockdown. The results show that 29.7% of households bought FAFH during the lockdown, while 70.3% did not. This suggests that most rural households in the sample did not participate in FAFH during the lockdown, likely due to the shutdown of FAFH establishments during Alert Level 5 (26 March to 30 April 2020), the partial reopening with restrictions during Alert Level 4 (1 May to 31 May 2020), and conditional operations during Alert Level 3 (1 June to 17 August 2020) (South African Government, 2020). These findings align with Codjia and Saghaian’s (2022) study, which observed a similar decline in FAFH purchases in the United States during the pandemic.
Post-lockdown, households’ participation in FAFH increased by 28.4%, rising from 29.7% during lockdown to 58.1%. Meanwhile, the percentage of non-participating households decreased by the same margin, from 70.3% to 41.9%. This suggests that the relaxation of restrictions and the full reopening of eateries contributed to higher participation. These findings are consistent with data from Statistics South Africa, which showed an increase in household spending on restaurants and hotels after restrictions were lifted (Pawson, 2023).
After addressing participation, respondents were asked about their FAFH expenditures during and after the lockdown. During the lockdown, the average expenditure was ZAR 136.07, with a maximum of ZAR 250 and a minimum of ZAR 0 for households that did not purchase FAFH. Post-lockdown, the average spending increased slightly to ZAR 149.25, with the maximum reaching ZAR 300, and ZAR 0 for those not purchasing FAFH. This suggests that the easing of restrictions and the full reopening of eateries led to an increase in household spending on FAFH. This trend is supported by the 41% revenue growth observed in takeaway and fast-food outlets compared to pre-COVID-19 levels.
Overall, the results highlight a significant shift in both participation and expenditure on FAFH following the lockdown. While most households refrained from FAFH during the lockdown, the post-lockdown period saw a notable increase in both participation and spending. The rise in household expenditure further suggests that lifting restrictions facilitated a return to pre-pandemic consumer behaviour, including increased spending on discretionary items like FAFH. Given these findings, it is important to further investigate the factors driving participation and expenditure in FAFH after the lockdown, particularly in rural areas. Understanding these factors will help inform strategies and policies to support the recovery and growth of the FAFH sector in the post-pandemic period.
Table 4 presents the results of the double-hurdle model, which examines the factors influencing household participation in food-away-from-home (FAFH) and subsequent expenditure post-lockdown. The probit model results identify variables affecting rural households’ decisions to purchase FAFH, while the OLS results highlight factors influencing the amount spent on FAFH.
Double-hurdle model results
| Variables | Decision to purchase FAFH (Probit model results) | Expenditure on FAFH (OLS results) | |||
|---|---|---|---|---|---|
| β | t-statistics | marginal effects | β | t-statistic | |
| 1 | 2 | 3 | 4 | 5 | 6 |
| Age | –0.764 (0.235) | –3.251 | –0.01 | –0.020 (0.018) | –1.111 |
| Gender | –0.667 (0.304) | –2.194 | –0.60 | –0.030 (0.012) | –2.500 |
| Marital status | 0.346 (0.169) | 2.047 | 0.15 | 0.025 (0.014) | 1.786 |
| Level of education | –0.055 (0.139) | –0.396 | –0.02 | –0.727 (0.162) | –4.488 |
| Employment status | 0.212 (0.098) | 2.163 | 0.40 | 0.049 (0.018) | 2.722 |
| Household size | –0.108 (0.054) | –2.000 | –0.05 | –0.932 (0.278) | –3.353 |
| Source of income | –0.137 (0.095) | –1.442 | 0.35 | 0.013 (0.012) | 1.083 (0.279) |
| Household income | 0.198 (0.069) | 2.869 | 0.25 | 0.095 (0.040) | 2.375 |
| Households’ members employed | 0.201 (0.075) | 2.680 | 0.20 | –0.020 (0.018) | 4.880 |
| Households’ members receiving social grant | 0.459 (0.200) | 2.295 | 0.075 | 0.371 (0.184) | 2.473 |
| Social grant received | 0.218 (0.085) | 2.565 | 0.08 | –0.025 (0.019) | –1.315 |
| Type of sector | –0.101 (0.085) | –0.532 | 0.45 | 0.008 (0.016) | 0.500 |
| Employment sector | 0.116 (0.115) | 1.009 | –0.10 | –0.011 (0.024) | –0.458 |
| SRD grant | 0.182 (0.078) | 1.051 | 0.18 | 0.093 (0.034) | 2.735 |
| Household structure | 0.096 (0.119) | 0.807 | –0.03 | 0.048 (0.022) | 2.181 |
| Changes in household’s income | –0.080 (0.146) | –0.548 | 0.12 | –0.004 (0.018) | –0.207 |
| Predicted probability | 0.65 | ||||
| Probit model summary | OLS model summary | ||||
| (–2) Log-likelihood | 54.665 | R-Square | 0.759 | ||
| Pseudo R-square | 0.78 | Adjusted R-Square | 0.744 | ||
| Prob > chi2 | 0.0000 | F-Statistic | 0.0000 | ||
Note 1. Values in parentheses represent standard errors. *, **, and *** denote p < 0.1, p < 0.05, and p < 0.01, respectively.
Note 2. For non-dichotomous variables, one category is used as the base category, with the other categories treated as non-base categories in comparison. The coefficients for these non-base categories indicate the difference in the probability of FAFH participation for each category relative to the base category.
Source: own elaboration.
The model’s pseudo-R-squared value of 0.78 suggests that the probit model explains a substantial portion of the variation in the decision to purchase FAFH. The significant chi-square value (p < 0.0001) confirms the model’s overall statistical significance, indicating that the independent variables collectively influence FAFH participation. The predicted probability of 0.65 suggests that the probit model correctly predicts household participation in FAFH approximately 65% of the time. The R-squared value of 0.759 indicates that the OLS model explains approximately 76% of the variation in FAFH expenditures. The adjusted R-squared value of 0.744 confirms a good model fit while accounting for the number of predictors. Finally, the F-statistic (p < 0.0001) confirms the overall significance of the OLS model, indicating that the independent variables have a statistically significant relationship with FAFH expenditures.
Age has a negative and significant effect on FAFH participation post-lockdown, with older households being 1 percentage point less likely to purchase FAFH than younger ones, ceteris paribus. However, among households who do participate, age does not significantly impact FAFH expenditure, suggesting that the household head’s age has little influence on spending.
Gender significantly affects both FAFH participation and expenditure, with female-headed households being 60 percentage points less likely to participate in FAFH than male-headed households. Among those that do participate, the negative and statistically significant coefficient for gender indicates that female-headed households spend less on FAFH than their male-headed counterparts, ceteris paribus.
Marital status has a positive and significant effect on FAFH participation, with married individuals being 15 percentage points more likely to participate compared to single, widowed, or divorced individuals, ceteris paribus. However, marital status does not significantly impact FAFH expenditure among rural households.
Education has no significant effect on FAFH participation, suggesting that education level does not strongly influence the decision to purchase FAFH. However, it has a positive and significant effect on FAFH expenditure, with higher-educated households spending more than those with lower educational levels.
Employment status positively and significantly influences both FAFH participation and expenditure. Employed individuals (either full-time or part-time) are 40 percentage points more likely to engage in FAFH than unemployed individuals, ceteris paribus. Furthermore, the positive and significant coefficients for employment status indicate that both full-time and part-time employed individuals spend more on FAFH than unemployed individuals, ceteris paribus. This effect on expenditure suggests that employed individuals allocate more to FAFH, likely due to greater financial stability and higher disposable income.
Household size has a negative and significant effect on both FAFH participation and expenditure. Each additional household member reduces the likelihood of FAFH participation by five percentage points, ceteris paribus. Among participating households, FAFH expenditure decreases by 0.932 units per additional household member.
Household income has a positive and significant effect on both participation and expenditure. Higher-income households are 25 percentage points more likely to purchase FAFH and tend to spend more on it, ceteris paribus.
The number of employed household members significantly increases the likelihood of FAFH participation, with each additional employed person raising the probability by 20 percentage points, ceteris paribus. However, expenditure decreases by 0.02 for each additional employed household member.
Social grants positively and significantly affect FAFH participation, increasing the likelihood by 8 percentage points, ceteris paribus. However, social grants do not significantly influence FAFH expenditure once a household participates.
The number of household members receiving social grants has a positive and significant effect on both FAFH participation and expenditure. Each additional grant recipient increases the likelihood of FAFH participation by 7.5 percentage points, reflecting the expanded consumption possibilities provided by higher aggregate social grant income (Aguilar-Lopez and Kuhar, 2022). In terms of expenditure, as the number of grant recipients rises, FAFH spending increases by 0.371.
SRD grant recipients, in contrast, have no significant effect on participation but have a positive and significant effect on FAFH expenditure. The absence of a significant effect on participation suggests that the SRD grant does not directly influence the decision to engage in FAFH, likely due to its temporary and limited nature, designed to address immediate financial distress rather than provide sustained disposable income. However, the positive and significant effect on expenditure shows that SRD grant recipients who participate in FAFH tend to spend more than those without the grant. Specifically, for each additional person receiving the SRD Grant, FAFH expenditure increases by 0.093 units, ceteris paribus.
Household structure does not significantly impact FAFH participation, suggesting that household composition does not affect the likelihood of engaging in FAFH. However, it has a positive and significant effect on FAFH expenditure, suggesting that different household compositions influence the amount spent on FAFH.
This section expands on the double-hurdle model results by exploring how socio-economic factors influence both participation in and expenditure on FAFH post-lockdown. It compares the findings with existing literature, contextualises observed trends, and offers insights into post-pandemic consumer behaviour. Since the study specifically focuses on fast-food outlets, its implications are more relevant to fast-food eateries than sit-down restaurants, which may be driven by different consumption factors.
The finding that age negatively affects FAFH participation is consistent with Liu et al. (2015) and Aguilar-Lopez and Kuhar (2022), who found that younger households in China and Mexico were more likely to consume FAFH due to socialising and lifestyle factors. However, the absence of a significant effect on expenditure contrasts with Aguilar-Lopez and Kuhar (2022), who reported that older households reduced both participation and expenditure during COVID-19 due to vulnerability and restrictions. This discrepancy may be explained by post-pandemic recovery, increased vaccine availability, and diminished health concerns (Gallo Marin et al., 2021).
The significant negative effect of gender on both FAFH participation and expenditure aligns with findings from Blick et al. (2018) in South Africa and Fernández and Agüero (2018) in Argentina, who found higher FAFH expenditures in male-headed households. Despite increased female labour force participation (Stats SA, 2021), persistent income disparities and gender pay gaps (McKenzie, 2023) may explain the lower FAFH spending among female-headed households. This aligns with earlier studies (Nayga and Capps, 1992; Ham et al., 2004) which found that male-headed households tend to spend more on FAFH, even as female labour force participation rises (Blick et al., 2018).
The positive effect of marital status on FAFH participation supports the work of Ham et al. (2004) and Binkley (2006), who observed higher FAFH participation in dual-income households in the U.S. and Korea. However, the lack of a significant effect on expenditure contrasts with Nayga and Capps (1992), who found that married households tend to spend more on FAFH due to greater financial stability. Post-pandemic economic constraints may have tempered FAFH expenditure despite continued participation, as noted by Venn et al. (2018) in Australia.
The positive relationship between education and FAFH expenditure aligns with Langellier (2015) and Abebe et al. (2022), who found that higher education levels are associated with increased FAFH spending in Mexico and Canada. However, since fast-food consumption is more common among lower-educated individuals, the higher expenditure among educated households may stem from time constraints and work demands, rather than education itself. This is consistent with Abebe et al. (2022) who found that educated Canadians were more likely to spend on FAFH due to work-related lifestyle pressures.
The positive and significant effect of employment status on both FAFH participation and expenditure aligns with Venn et al. (2018) in Australia and Aguilar-Lopez and Kuhar (2022) in Mexico. Both studies showed that full-time employment increases reliance on FAFH due to factors like time constraints and financial stability. Full-time employment typically involves longer working hours, limiting time for home meal preparation. As a result, employed individuals are more likely to purchase ready-made meals to save time and ensure convenience. These findings align with Goolsbee and Syverson (2021), who highlighted the impact of COVID-19 and shifts in labour dynamics.
The negative relationship between household size and FAFH participation and expenditure supports Aguilar-Lopez and Kuhar (2022) and Blick et al. (2018), who found that larger households tend to reduce FAFH spending due to economies of scale and preferences for home cooking. This suggests that the higher FAFH expenditure among educated households may be driven more by time constraints, demanding work schedules, and convenience than by education itself. Educated individuals, especially those employed full-time, may rely on fast food as a quick and accessible meal option, despite potential health concerns. This is consistent with earlier findings (Blick et al., 2018; Aguilar-Lopez and Kuhar, 2022).
The significant effect of income on both FAFH participation and expenditure supports Engel’s Law, which posits that as household income increases, spending on FAH decreases, while spending on FAFH rises (Lusk and McFadden, 2021). The findings align with pre-pandemic studies such as Stewart et al. (2004) and Blick et al. (2018), which found that higher-income households allocate more resources to FAFH. This is consistent with Saksena et al. (2018), who suggest that FAFH is considered a luxury good.
The mixed effects of social grants on FAFH participation and expenditure align with Waidler and Devereux (2019), who found that while grants improve access to food, they may not necessarily lead to increased spending on FAFH due to financial constraints. However, the significant effect of SRD grants on expenditure, as supported by Aguilar-Lopez and Kuhar (2022), suggests that temporary income boosts may facilitate increased spending on FAFH. This is further corroborated by Mkasi and Lefophane (2023), who observed that while SRD grant recipients reduced overall food consumption, they were still able to maintain food security during the pandemic.
Finally, the significant effect of household structure on FAFH expenditure, particularly for extended households, aligns with Lozada-Urbano et al. (2022), who found that extended households in Peru tend to consume more FAFH. This increase in consumption is primarily due to pooled incomes, where multiple earners contribute to a greater collective disposable income. As a result, extended households – comprising not only nuclear families but also extended family members – are better able to purchase meals away from home. The financial pooling within these households allows for greater flexibility in spending on discretionary items like FAFH. This is supported by Bezerra et al. (2013), who highlighted the impact of household composition on FAFH spending.
This study employs a double-hurdle model to analyse the factors influencing rural households’ participation in, and expenditure on, food away from home (FAFH) in South Africa post-lockdown. The descriptive analysis reveals a significant shift in FAFH behaviour post-lockdown, with both participation and expenditure increasing notably. This resurgence is attributed to the lifting of restrictions, widespread vaccination, and the return to normal operations of eateries. These findings provide empirical evidence that FAFH expenditures among rural households have rebounded, highlighting how eased restrictions and the reopening of eateries have played a key role in boosting both participation and spending.
The empirical analysis utilized a probit model in the first hurdle to identify the determinants of FAFH participation and an OLS model in the second hurdle to assess the factors influencing expenditure levels. Both models indicate that employment status, household size, household income, the number of employed household members, and the number of social grant recipients are significant predictors of both FAFH participation and expenditure. Additionally, gender and marital status influence participation decisions, while education level, household structure, and SRD grants have a significant impact on expenditure levels. By identifying these determinants, this study fills a critical gap in the literature on rural household dynamics in the post-pandemic era and provides valuable empirical insights into the factors shaping FAFH behaviour in a shifting socio-economic landscape.
Based on these findings, several policy strategies are recommended to improve food security, enhance access to nutritious food, and support vulnerable households, particularly those in low-income and informal sectors:
Income and Employment Support: Policies should prioritize job creation, skills development, and financial literacy training to enhance household income and financial stability. This would lead to higher employment rates, improved financial security, and greater access to diverse and nutritious food options.
Strengthening Informal Food Systems: Supporting small-scale food vendors, promoting affordable, healthy meal alternatives, and expanding nutrition education initiatives can improve the availability of nutritious foods while fostering local food economies.
Targeted Food Assistance Programs: Subsidies, meal programs, and direct cash transfers should focus on vulnerable households, particularly children and the elderly, to reduce food insecurity and ensure access to balanced meals.
While this study provides meaningful insights, several limitations highlight opportunities for further research:
Expanding the scope to include urban areas would allow for a broader understanding of post-pandemic FAFH trends and regional disparities.
Investigating the role of food prices in FAFH decisions, particularly through income elasticity estimates, could provide deeper economic insights.
A longitudinal approach could track how FAFH consumption and spending patterns evolve as economic conditions stabilise.
Exploring additional factors such as marketing strategies, cultural influences, health perceptions, digital food ordering trends, and the impact of social grants would offer a more nuanced understanding of FAFH behaviour in diverse rural contexts.
The lack of detailed data on SRD grant recipients in the study area limits direct comparisons between the sample and the broader beneficiary population. Future research should address this gap by collecting more granular data at the village level, possibly through collaboration with local authorities or community-based surveys.
FAFH establishments encompass sit-down restaurants, quick service restaurants, pubs, bars, cafes and other dining venues. In contrast, FAH refers to food purchased from retail outlets, or other food stores for preparation and consumption at home (U.S. Department of Agriculture, Economic Research Service, 2025).
The Social Relief of Distress (SRD) grant is a temporary financial assistance program provided by the government to support individuals facing extreme economic hardship, particularly those who are unemployed and do not receive other forms of social grants or Unemployment Insurance Fund (UIF) payments (SASSA, 2025).