Current diets and food expenditure
Current diets were directly addressed in the survey through the question “Which of the following diets do you identify with?”, with the five diets mentioned earlier and an additional “other” option as possible answers. Moreover, using the answers regarding meals per week for each primary food, it was possible to identify the current diets with greater robustness. A comparison of results between both questions showed some dissonance, where the self-identified (perceived) diet was different from the actual diet (considering the meals consumed per week). Table 4 shows the differences.
Table 4 Comparison between perceived diet and actual diet Full size table
Some self-identified plant-based consumers reported eating animal-based foods at least once per week. Of the 40 self-identified vegans, in practice, 8 are ovo-lacto-vegetarians, 5 are flexitarians, and 1 is pescatarian, reducing the number of actual vegans to 26. The number of vegetarians also decreases, while flexitarians and pescatarians increase in number. It is not uncommon for participants to report being plant-based and also report sporadically consuming animal-based foods, as observed in the seminal work by Dietz et al. (1995). In fact, past studies have also identified this dissonance. Lusk and Norwood (2016) report that 5.2% are vegetarian, but only 2.2% are actual plant-based consumers. Self-identified plant-based consumers may tend to eat less animal-based foods. However, they may also be more likely to pay a higher price for these foods for their higher quality (e.g., organic, ethical). Jung et al. (2022) suggest that when the price of cultured meat was high, the purchase intention for novel food was also higher. Thus, it is vital to correct the data of actual vegetarians and vegans, as the present research does, so that this bias does not affect the results.
In the subsample assessed, 2.5% of the participants follow a vegan diet and 3.27% follow an ovo-lacto-vegetarian diet. In comparison with past studies, Centro Vegetariano and AC Nielsen (2017) reported that, in 2017, 0.6% of the population followed a vegan diet and 1.2% a vegetarian (ovo-lacto-vegetarians and vegans) diet. More recently, Jung et al. (2022) reported, for the USA, similar shares of vegans (2.2%). Thus, although the share of omnivorous is still large, 20% of the participants report eating less meat, concluding that in 4 years the numbers of alternative diets have increased considerably.
Figure 1 depicts food expenditure at-home for all diets. In general, self-reported food expenditures are lower among vegans. For example, more than 10% of the vegans reported spending less than 20€ per week, while only 2% of omnivorous reported the same amounts. Among vegans, 23% report spending between 20€ and 39€ per week, while for pescatarians, 24% report spending between 40€ and 59€, and 24% of vegetarians reported spending between 60€ and 79€. This might be a first indication that plant-based consumers spend less on food compared to consumers following other diets. Additionally, food expenditures are generally higher for omnivorous consumers, compared to the other consumers. On average, omnivorous consumers reported a food expenditure at-home of 75.96€/week. This is the highest expenditure level compared with the other diets, particularly in comparison with 62.35€/week for pescatarians, 68.6€/week for flexitarians, 59.39€/week for ovo-lacto-vegetarians, and 47.78€/week for vegans.
Fig. 1 Food expenditure at-home per week by actual diets (in % of consumers of each diet) Full size image
Moreover, Fig. 2 shows the responses for the food expenditure away-from-home, which are quite different than the ones reported for expenditure at-home. Nonetheless, more than half of the vegans report spending less than 5€/week outside their home, while only 22% of animal-based consumers report spending the same amount. Considering both food expenditures at-home and away-from-home, the descriptive statistics suggest that vegan consumers may spend less on food than their omnivorous counterparts, but the same may not be concluded for ovo-lacto-vegetarians.
Fig. 2 Food expenditure away-from-home among actual diets (in % of consumers of each diet) Full size image
Dietary preferences, however, may not be the only aspect influencing food expenditures. Having children, living with family, buying food for others or not, being responsible to shop and cook, having more disposable income, are all characteristics, among others, that may also affect expenditure. Thus, demographics may play a profound role in shaping how much consumers spend on food. These aspects are also considered in the models so that the true effect of diets is estimated, minimizing the omitted-variable bias, that is, vegans may spend less because they live alone, compared to omnivorous consumers who have children. Only with the results from the empirical analysis, it is possible to ascertain which consumers in reality spend less, since the models take into account several other variables as potential confounders.
Falsifiability of hypotheses
Following the hypotheses defined, Figs. 3, 4, and 5 show the coefficientsFootnote 5 of the modelsFootnote 6 described in the general specifications (Eqs. 2–5). Although the coefficients should not be interpreted in terms of magnitude since they do not reflect a marginal change, the overall effect (positive or negative) can be derived. It is thus possible to test each hypothesis. Additionally, the marginal effects (Tables 5, 6, 7, 8) have the advantage to quantify the impact of an increase in the predictors’ value on each category of food expenditures, i.e., the impact of a one unit-increase on the likelihood of spending less than 20€, or on the likelihood of spending 60€–79€. These are analysed in Sect. 3.3. The econometric software STATA 15 was used.
Fig. 3 Coefficients of actual diets on food expenditure (Eqs. 2, 3) Full size image
Testing H1: Plant-based diets, compared to omnivorous diets, positively affect the likelihood of spending more on food, Fig. 3 shows that a vegan consumer, compared to an omnivorous one, is associated with a lower food expenditure, for both at-home (Eq. 1) and away-from-home (Eq. 2). This means that vegan consumers are less likely to spend more on food (or more likely to spend lower amounts of money, according to the detailed marginal effects described in Table 5) compared to omnivorous consumers. The same can be said about pescatarians for food expenditures at-home. Therefore, H1a (Eq. 2, \(\beta_{115} > 0,\) p value = 0.014) and H1b (Eq. 3, \(\beta_{115} > 0,\) p value = 0.008) are rejected and it can be concluded that plant-based consumers do not spend more, but rather less than omnivorous consumers, for the sample assessed. However, additional insights can be derived concerning the cost associated with the number of meals consumed; these are presented further.
Considering H2: higher frequency of plant-based meals is associated with higher food expenditures, Fig. 4 shows that there is no relationship between the number of vegan meals and food expenditures, for both at-home (Eq. 3) and away-from-home (Eq. 4). However, an increase in ovo-lacto-vegetarian meals is associated with higher food expenditures away-from-home. Additionally, an increase in red meat meals and fish meals is also associated with higher food expenditures, for both FEAH and FEAFH. Therefore, H2a (Eq. 4, \(\beta_{5} > 0,\) p value = 0.55) and H2b (Eq. 5, \(\beta_{5} > 0,\) p value = 0.582) cannot be rejected for plant-based diets (vegan). Nonetheless, eating more red meat and fish meals increases both food expenditures.
Fig. 4 Coefficients of different meals on food expenditure (Eqs. 4, 5) Full size image
Following Fig. 5, the hypotheses concerning with food-related preferences can be testedFootnote 7. First, H3: informed consumers frequently spend less on food can be verified through the variable INFO (looks for information before buying). The hypothesis cannot be rejected since the coefficient shows no statistical significance level either regarding at-home or away-from-home. Second, consumers who favour biologic/organic foods tend to spend more on food (H4) is not rejected as indeed results show a positive relationship between favour biologic and spending more on food both at-home (H4a) and away-from-home (H4b). Third, for H5: consumers who cook for themselves and others, use leftovers, and own/receive local food production frequently spend less on food, results show that only one aspect (use of leftovers) of a pro-active consumer is generally associated with lower food expenditures, while the other aspects do not show any statistically significant relationship.
Moreover, other inferences can be derived from Fig. 5. For example, consumers with a higher education spend less on food at-home, while the wealthier the consumer, the higher is the food expenditure both at-home and away-from-home. Consumers who eat ready-to-eat meals more frequently generally end up spending more, although ready-to-eat meals are generally perceived as cheaper and more convenient than preparing meals with fresh food. Furthermore, the expected effects from the rest of the covariates corroborate the good consistency of the models. For example, a consumer who buys for the household spends more than a consumer who buys only for herself/himself. The same is shown for a consumer who lives with her/his family, a consumer who spends more than another consumer who lives alone or shares the house with non-family members. Additionally, the number of away-from-home meals is positively associated with spending more away-from-home, while lunchbox meals show a negative effect. These effects are in accordance with what is expected from reality and thus reinforce the robustness of the models.
Marginal effects and food policy implications
The marginal effects computed from the coefficients described above make it possible to assess the effect of the different diets on food expenditure in more detail, particularly for each category of spending. Tables 5, 6, 7 and 8 present the marginal effects associated with the different diets assessed for food expenditure at-home (FEAH). The * denotes statistical significance at the 5% level. As described above, both pescatarian and vegan consumers tend to generally spend less on food expenditure compared with omnivorous consumers. Table 5 suggests that, on average, following a vegan diet increases a consumer’s probability of spending 20–39€ and 40–59€ by 0.082 and 0.107, respectively, compared with an omnivorous diet. Additionally, the predicted probabilities of spending 120–139€, and 140€ or more is on average 0.034 and 0.048 lower for a vegan consumer than for an otherwise similar omnivorous consumer. Therefore, vegan consumers are more likely to spend lower amounts of money and less likely to spend higher amounts on food at-home compared with omnivorous consumers, which again rejects H1a, that plant-based consumers spend more than omnivorous consumers (at-home).
Table 5 Marginal effects of diets on food expenditure at-home Full size table
Table 6 Marginal effects of diets on food expenditure away-from-home Full size table
Table 7 Marginal effects of meals on food expenditure at-home Full size table
Table 8 Marginal effects of meals on food expenditure away-from-home Full size table
Regarding pescatarian consumers, similar marginal effects are found, although at lower predicted probabilities. The coefficients described in Fig. 3 only show the effect of the different diets analysed comparing them to an omnivorous diet as the base. Table 5, however, shows all the possible associations between the five diets assessed. These results show that a vegan consumer is more likely to spend lower amounts of money on food and less likely to spend higher amounts than a flexitarian. Hence, it is possible to conclude that vegans also tend to spend less on food at-home than otherwise similar flexitarians. Nevertheless, the marginal effects of vegans are higher when compared with omnivorous consumers, followed by ovo-lacto-vegetarians and finally flexitarians.
When analysing food expenditure away-from-home, the results described in Table 6 show similar patterns for plant-based and omnivorous diets. Vegan consumers are more likely to spend amounts less than 10€ and less likely to spend any other amount above 9€, compared with omnivorous consumers, for a statistical significance level of 1%. No statistically significant difference in food expenditure away-from-home was found between pescatarians and omnivorous consumers. Ovo-lacto-vegetarians, however, show higher probabilities of spending lesser amounts, compared with omnivorous consumers (although only at a 10% significance level). Additionally, plant-based consumers also tend to spend less on food away-from-home when compared with both pescatarians and flexitarians (which is also true for vegetarians compared with pescatarians). Therefore, to further corroborate previous findings, vegan consumers are more likely to spend lower amounts and less likely to spend higher amounts of money on food away-from-home compared with all other diets except ovo-lacto-vegetarians, rejecting H1b that plant-based consumers spend more than omnivorous consumers (away-from-home).
The marginal effects concerning food choices presented in Table 7 are interpreted differently, since the variables are ordinal instead of nominal. On average, a one unit-increase in the frequency of red meat meals is associated with a 0.021 increase in the probability of spending 140€ or more, and a 0.033 decrease in the probability of spending 40–59€ on food at-home. These marginal effects have a statistical significance level of 1%. Fish meals show similar marginal effects, although with lower magnitudes. Therefore, results show that an increase in red meat meals and fish meals per week increases the amount spent on food consumed at-home.
Since omnivorous diets include red meat and fish at considerable levels, at least with no intention of reduction (which are the flexitarian diets), results suggest that omnivorous consumers increase their food expenditure if the frequency of meals increases. The same goes for pescatarians and fish meals, while the frequency of vegetarian and vegan meals is not statistically significant. As consumers cannot eat infinitely, they incur trade-offs and choose between different meals. Choosing red meat meals, for example, incurs an opportunity cost of not choosing white meat meals or vegan meals, which could be cheaper than red meat. This explains why the positive association was found for red meals and fish meals.
Concerning food expenditure away-from-home, red meat meals show similar marginal effects as for food expenditure at-home (Table 8). The marginal effect of red meat meals is higher for the probability of spending less than 5€, i.e., a one unit-increase in the frequency of consuming a red meat meal is associated with a 0.023 decrease in the probability of spending less than 5€. The marginal effects found for fish meals are statistically significant only at the 10% level, while ovo-lacto-vegetarian meals show a similar marginal effect as red meat meals. A one unit-increase in the frequency of vegetarian meals is associated with a decrease in the probability of spending lower amounts on food away-from-home (less than 5€, and 5–9€), while spending higher amounts increases.
Globally, the results suggest that Portuguese consumers who follow a plant-based diet end up spending less on both food consumed at-home and away-from-home, compared with an otherwise similar Portuguese consumer who follows an omnivorous diet. The same can be said when compared with flexitarians. Evidence also shows that plant-based consumers spend less than pescatarians away-from-home. These insights, particularly that plant-based consumers spend less than omnivorous consumers on a weekly basis, follow the findings highlighted in the literature for the USA (Lusk and Norwood 2016), Sweden (Grabs 2015), and the UK (Berners-Lee et al. 2012; Hoolohan et al. 2013). To the authors’ knowledge, the present study is the first to assess these expenditure patterns in a Mediterranean country with strong cultural dietary roots, such as Portugal.
To promote healthy and sustainable diets, these also need to be affordable for all. Considering that plant-based diets are often highlighted in the literature as healthier and more sustainable, according to the present study, plant-based consumers are also shown to spend less than consumers following other diets. Thus, promoting plant-based diets as possibly cheaper can also be advantageous for achieving more acceptance and reaching a wider range of consumers, as price is an important factor when purchasing, particularly for poorer consumers. This away, policymakers can use plant-based diets as a mechanism to mitigate, not only climate change, but also food insecurity present in poorer households. Additionally, since plant-based consumers spend less on food, there will be a saving which will be spent elsewhere. Thus, a plant-based consumer will have a greater budgetary availability for meeting other needs. However, from an environmental point-of-view, in some cases, these other needs could have a negative impact which could offset the environmental benefits of choosing a vegan diet. In these cases, the rebound effect of the savings should be considered to estimate the net effect.