The impact of the Corona crisis on the environmental behaviors of different socioeconomic groups

The Corona pandemic led to changes in consumptions patterns, which are both positive and negative. Past research suggests that crises present opportunities for adopting sustainable consumption practices. Alternatively, they tend to increase frugality which can marginalize environmental considerations. This study extends research conducted in 2018 evaluating the environmental impacts of consumption patterns among Israel’s different socioeconomic deciles. The present research returned to the same respondents, during the first lockdown, assessing how consumptions patterns among different socioeconomic deciles, and support for different environmental policy options were influenced by the Corona crisis. The findings show that the poorest deciles increased their environmentally destructive behavior, while wealthier deciles showed modest improvements. All deciles displayed greater frugality in purchasing. Support was greater for policy interventions framed as environmental than for taxes on daily consumer products. The findings confirm the need for environmental programs which make sustainable consumption more readily accessible to poorer socioeconomic groups.


Introduction
This unprecedented economic downturn of the Corona pandemic brought with it changes in all aspects of life, including environmental behavior and consumption patterns. These were manifested in increased on-line shopping via the internet, which is considered to increase excess consumption and therefore has adverse environmental consequences (Aboujaouda et al 2015, Khyzer et al 2015, Aboujaouda et al 2017 alongside general reductions in retail purchases (Libscar 2020). A change in consumer preferences emerged, from products that promote environmental values to products that ostensibly contribute to improved health (Insights Website 2020). These findings are consistent with research suggesting that during crises, there is a marked drop in environmental concerns and an increase in skepticism about climate change (Brulle Robert et al 2012, Ratter et al 2012. Economic hardship has also been associated with negative effects on people's support for mitigation of greenhouse gas emissions (Matto and Anthony 2017)  The relationship between sustainability, consumption patterns and economic crises, therefore can appear inconsistent and even contradictory. On the one hand, saving patterns in households lead to reduce household resource utilization, such as energy and general purchases, producing a positive environmental impact. Conversely, a shift to cheaper products, such as old cars and fast fashion, often carry high environmental costs, leading to environmental damage. To this complex equation must also be added socio-economic dimensions and individual affluence, which influence consumption patterns and associated environmental impacts (Carsin 2009, Bass-Spector 2011, Peleg-Mizrachi and Tal 2019, Alcorn 2020).
In addition, despite prevailing understanding and emerging insights about post-crisis, consumption patterns, as well as research conducted prior to the pandemic about the relationship between green consumerism and socioeconomic status, there is still a pressing need for empirical assessments that integrate the two. Most post-crisis, consumer studies only focus on responses to the adverse economic conditions, much as many social scientists did following the 2008 financial crisis. But the Covid epidemic is unique because it creates a natural experiment, which for the first time allows for evaluation of a crisis that is linked to sustainability. The present research is the first study to consider how consumption patterns and environmental attitudes are affected by a global crisis that is generally perceived as having ecological-zoonotic origins. Associated findings can help in understanding how to make the climate crisis more accessible to new audiences and inform policies that seek to further environmental objectives when societies face acute health and economic challenges in the future.
By evaluating the change in sustainable consumption behavior among a representative sample of Israelis following the Covid crisis, this article offers sociological insights into consumer psychology and priorities. It also explores public policy's potential for upgrading sustainable consumption patterns and accessibility to environmentally friendly products among different socioeconomic clusters in times of economic downturn.

Methods
In 2018, we conducted a study that characterized consumption patterns among the various socioeconomic deciles in Israel in numerous consumption categories, from an environmental perspective. Our findings reveal that patterns are more nuanced than were previously recognized and that there are areas in which poorer populations manifest less sustainable consumption patterns. We also concluded that there is room for greater government interventions to facilitate more sustainable consumption patterns.
In addition, we found that sustainability patterns and attitudes towards environmental issues among different socio-economic deciles is largely influenced by economic ability and the way it is perceived by a given societal cohort (Carsin 2009).
In the midst of the Corona crisis, while the economic capabilities of many households in Israel changed dramatically for the worse, we returned to these study participants using an instrument containing many of the same questions from the 2018 survey. The primary research objective was to examine the extent to which respondents were financially affected by the epidemic and how individual consumption patterns and attitudes about the environment were affected as a result. Accordingly, the study was designed to assess what changes in Environmental Behaviors and consumption patterns occurred during the health crisis, what changes might be expected to occur in the future, and what kind of public policy could contribute to improved individual environmental consumerism. Figure 1 provides a flowchart of the progression of the study: We hypothesized that the most meaningful change in consumption patterns would be a tendency towards frugality, a pattern that is likely to become stronger as the economic impact of the pandemic became greater. In light of the findings from our previous research, in which we found that high socio-economic deciles tend to exhibit environmental behavior, with a modest contribution (e.g. recycling), at a higher rate than low socio-economic deciles (Peleg-Mizrachi and Tal 2019). We also hypothesized, that there would be a tendency for low to medium environmental involvement, reflected in recycling patterns, among the higher deciles. At the same time, we anticipated long-term, significant changes in consumption patterns to be manifested equally among all deciles. Finally, we hypothesized that among those deciles hit hardest financially from the Corona economic downturn, there would be a decline in environmental considerations when making consumer decisions, an increase in abusive behaviors towards the environment and a resistance to environmental taxation.
Given the new focus, we added several variables to those measured in the 2018 survey. These include: the degree of economic harm from the Covid pandemic, changes in consumer behavior during the Covid-closure period, projected future changes in consumption patterns and the degree of support for different environmental policy measures. The independent variables ultimately assessed included respondents' socioeconomic decile and the degree of economic harm reported due to the Covid pandemic. The dependent variables selected were changes in consumer behavior during the Covid closure period, projected future changes in consumption patterns and the degree of support for different environmental policy measures. Data collection was conducted through 'Panel Project Sampling' , a company specializing in web surveys for research purposes, during April 2020, 1 month after the start of the first closure, when the citizens of Israel were still in the midst of the closure. All subjects received the same fee for their participation. Of the six hundred respondents in the 2018 probability-based panel, a total of 385 were available for the 2020 survey. Given the well-known challenge of low recruitment participation rates in probabilitybased panels and natural attrition among participants (Hays et al 2015), the large number of respondents available for a follow-up survey was deemed fortuitous. The survey instrument itself is long and detailed, with well over a hundred questions about consumption and attitudes. Yet, 374 of the 2020 respondents completed the questionnaire, reflecting a high completion rate of 97.15%. The average time for completing the questionnaire was 32 min.
With a sample roughly two-thirds thirds the size of the original cohort, it was important to confirm that the participants in the Covid survey continued to constitute a representative sample of Israeli society. At first, we rebuilt the socioeconomic decile variable relying heavily on our previous research, and on the methodology used by Israel's Central Bureau of Statistics, which assigns every Israeli to one of ten socio-economic groups accordingly to the following sub-variables: education, geographic district, income, receipt of welfare payments, self-determination of economic status and seniority (years living) in Israel Each socio-economic, sub-variable was scored for every participant. The sub-variables were then recoded according to the individual response. Pursuant to these calculations, each participant received a composite score ranging from 18 to 52 (average 30.55, SD 5.25). All subjects' scores were then aggregated and summed. Finally, we divided the participants into ten groups based on these final figures (see table 1, in the appendix), producing the socio-economic deciles which serve as the basis for comparison. This process ensured the full representation of all socioeconomic sectors within Israel's society. All scores of the sub-variables were tagged according to the socioeconomic decile categories assigned by the Central Bureau of Statistics.
As illustrated in tables 2-8 in appendix, the representation of Israeli society is faithfully maintained, among the sampled population, despite the change in the composition of the participants In this way, for example, 27.9% of the participants were from central Israel, 11.8% from the south, 9.9% from the north, 7.2% from Jerusalem, 15.5% from Haifa, 4.8% from the West Bank and 22.8% from Tel Aviv-Yafo. These numbers closely reflect the geographical distribution of Israel's populations. Another example confirming the representativeness of the study sample involves the distribution of income: 24.9% reported that they earn far below average, 21.9% that they earn below average, 39.6% reported that they earn average wages, 10.2% reported that they earn above average and 3.5% earn far above average. These data also faithfully represent Israeli society, in which of income is not evenly distributed across socio-economic deciles. The rate of people earning a wage much lower than the mean wage in Israel is significantly higher than the rate of people earning wages that are far higher.
In order to assess the degree of financial damage caused by the Corona crisis, we examined the degree of economic harm on a scale of 1-5 as well as the nature of any changes in employment status: shifting to work from home, reduction in job status, unpaid leave, dismissal (but still eligible for unemployment benefits), dismissal (without entitlement to unemployment benefits), reduced income among independent business owners and stability among essential workers who continued their jobs throughout the closure period.
To examine changes in consumer behavior, subjects were asked about 17 behavioral changes involving the environment that might have occurred during the closure period. Among these are increased composting waste separation, setting aside bottles and batteries for recycling, reduced clothing purchases, avoiding the eating of meat, repairing broken objects, preventing food wastage,increased use of disposable plastic, etc (see table 9 in appendix).
To better characterize the general trends in consumer behavior, these 17 behavioral-environmental changes underwent factor analysis and were divided into five groups: frugality, recycling, proenvironmental behavior, fashion -related consumption and non-environmental changes (see tables 9 and 10 in appendix). In order to ensure that the variables are not independent and that factor analysis can be performed, Kaiser-Meyer-Olkin and Bartlett's Test were conducted (KMO 0.811, BT 1526.766, df 153, Sig 0).
The study also sought to assess behavioral changes likely to occur as a result of the epidemic. Subjects were asked about six consumer and environmental behaviors ranked on a scale of 1-5 in relation to the likelihood of their adoption. Participants who stated that they were merely continuing pro-environmental behavior patterns they had already adopted were recoded as zero, representing no change from baseline behavior .
Participants were also asked the extent to which they supported 17 different types of economicenvironmental measures, on a scale of 1-5. The policy types were grouped into four factors that explain 57% of the variance (see tables 11 and 12 in appendix). 'domestic levies' included increased taxation for eggs and beef products, health care fees and electricity prices; 'luxury taxes' included taxes on flights, new vehicles and fuel as well as customs for overseas packages; 'environmental fees' referred to measures such as pollution taxes, benefits for employers' that allow work from home and waste disposal fees; and 'ending pro-natal incentives' which included discontinuing child allowances or limiting the number of children receiving child allowances. In cases where we found cross-loadings for items in dependent variables, the factor with the higher loadings was selected.
The effect of socio-economic level on consumer preferences exhibited during the closure period (the five factors) was assessed, along with the planned consumer changes in the future and the degree of support for the various policy measures using analysis of variance and ANOVA testing. Duncan and Scheffe tests were conducted to examine the source of the differences between the groups. Similarly, the degree of economic damage from the Covid crisis and its impact on the five factors of consumer changes identified during closure, the anticipated consumer changes and the support for political measures was also assessed. In addition, we examined the distribution of economic harm reported across the various socio-economic deciles. The correlation between the socio-economic deciles and the degree of economic harm were examined by the Spearman test, which is designed to examine the relationship between ordinal variables (see table 13 in appendix).

Findings
In examining the distribution of economic harm across the various socio-economic groups in Israel we found that on average, people in all socio-economic deciles were affected somewhat by the Corona downturn. At the same time, the percentage of people significantly affected was higher among the low deciles (average 39.6% in deciles 1-3). Respondents, whose income was not affected at all were predominantly from the high socio-economic deciles (average 46% in deciles 7-10) (see figure 2).
The distribution of changes in consumer behavior during the Corona closure period was assessed across socio-economic deciles by ANOVA testing. We found that higher deciles tended to increase recycling and purchase food directly from farmers. Lower deciles tended to increase disposable plastic use patterns, while central deciles (3-6) tended to increase patterns relating to frugality, such as: avoiding clothing purchases and maintaining low power consumption. In addition, among all deciles, the patterns most frequently reported involved frugality (see tables 14 and 15 in appendix). These findings are consistent with our previous research which calculated the ecological footprint of lower deciles in the food sector as larger than that among higher deciles with respondents in higher deciles tending to recycle more than lower deciles (Peleg-Mizrachi and Tal 2019).
The most common manifestation of economic harm in consumer behavior was increased frugality, with the least common change being an increase in environmental considerations in consumption decisions. The increase in frugality was also confirmed in ANOVA testing as significantly higher when behavioral changes during quarantine were tested according to type of employment change (see table 16 in appendix).
Respondents among the full spectrum of economic harm reported behavioral changes, with significant differences found for two behaviors directly related to the environment: recycling and activities categorized as harmful to the environment (see table 17 in appendix). In all four patterns of behavior, the highest level of change was found among respondents who were only slightly affected financially by the Covid crisis (see figure 3 and table 18 in appendix).
We also considered the distribution of projected future changes in consumption patterns across socio-economic deciles. Here, we found a significant trend emerging in three of six possible types changes: Reduced purchase of luxury items and flights, was the most common found among higher deciles (see table 19 in appendix). The second and third change manifested in behavior during the crisis was a shift  from private car travel to public transportation and the adopting of a vegetarian/vegan diet. These two changes were significant, but without a clear trend in relation to socioeconomic status. Thus, the average probability of a change in diet was 1.3 in deciles 1, 4 and 8, compared with 1.9 in the second decile and 1.7 in the seventh decile (see table 20 in appendix).
In the ANOVA test, reduction of luxury purchases and increased utilization of public transportation were found to be significant (see table 21 in appendix).
A deeper look at the results reveals that adverse economic impacts tend to be associated with specific changes. Reduced luxury purchases and increased public transportation utilization were significantly more common among respondents who were highly affected economically. Expanded recycling was the least common change reported among respondents' most affected economically. By contrast, recycling and reducing flights emerged as the most common change among those who were only slightly harmed economically, while avoiding the use of disposable plastic and switching to a vegetarian or vegan diet were reported almost equally among respondents, regardless of degree of economic harm (see figure 4). These findings are consistent with our previous study, in which we found that although the lower deciles had a larger ecological footprint in the food sector, as a result of increased use of disposable plastics and fewer purchases of sustainably grown or organic food. In the distribution of vegetarians and vegans across the deciles, however, we found no trend. In other words, the rate of vegans and vegetarians was similar in both the lowest and highest deciles.
The study also sought to assess any trends in public backing for different environmental policy measures. We found different levels of support among different socioeconomic groups. Among all deciles, the type of measures receiving the most significant support were 'environmental fees' or generic interventions to protect the environment. The next level of support backed 'luxury taxes' . This proved to be particularly popular among lower socio-economic deciles. 'Ending pro-natal incentives' policy measures received greater support among the higher deciles than among poorer respondents. Finally, 'domestic levies' , involving increased taxes on basic commodities, was the least popular policy measure among all deciles (see figure 5, tables 22 and 23 in appendix).
When respondents are divided according to degree of economic harm caused by the Corona pandemic, the distribution of support for different environmental policy measures is similar. The type of policy receiving the greatest support is still environmental fees while domestic levies was the policy intervention that received the least support at all levels of vulnerability. Unlike the distribution of opinions among socio-economic deciles, luxury taxes were the second most popular policy supported, regardless of economic harm level, with no differential preferences found for measures relating to ending pro-natal incentives (see figure 6 and table 24 in appendix). It is important to note that in a Duncan, post-hoc test, designed to identify the source of these disparities, there was a significant difference between those whose income was not affected at all and those who were (see tables 25 and 26 in appendix).

Discussion
Similar to the climate crisis, the Covid crisis affects different socio-economic groups differently. Nonetheless, an overall impact of exacerbated existing socio-economic gaps can be identified worldwide (Alcorn 2020), as well as in Israel (Aviram Nitzan and Kidder 2020). The widening of these gaps also magnifies preexisting environmental inequalities in Israel, where there is a high correlation between low socioeconomic status and the likelihood of harm from environmental hazards (Carsin 2009, Bass-Spector 2011, Reza Khan et al 2016.
In this study, we found that the most common type of behavioral change during closure was frugality, whereas the least common type of change was enhanced environmental behavior. When we examined the degree of support for environmental policy measures, the measures receiving the highest  support were those directly presented as environmental taxes. Those that received the lowest support were domestic levies, or taxes on basic consumer goods-even for products with significant environmental impacts.
These findings suggest that as long as there is little recognition about the environmental consequences of consumption patterns among consumers. Support was conspicuously absent for 'domestic levies' or restriction on basic products that pose negative environmental impacts. The strong general concern expressed for environmental issues and the shared desire among respondents from all socio-economic groups to establish sound environmental policies, leads us to conclude that effective policies targeting environmental consumption require a parallel effort to promote awareness about the associated environmental implications of different consumer behaviors.
Moreover, in examining the relationship between actual behavioral changes, as reflected in consumption patterns, and changes in attitudes toward the environment, as expressed in support of policy measures, it can be seen that those most economically affected respondents not only supported domestic levies to a lesser extent, but also reported relatively lower support for luxury taxes (1.99, compared with 2.19 and 2.15). At the same time, support for general environmental fees was relatively high. These findings are counterintuitive because a relatively high rate of economically affected participants also stated that they intend to reduce purchases of luxuries (clothing, objects, etc). They are also the group most likely to replace travel in a private vehicle with public transportation. This suggests that the taxation of luxury goods would affect them less than general domestic levies. Nevertheless, tax policies targeting luxury items did not receive meaningful support from them.
On the other hand, respondents that reported fewer economic consequences or who were not affected at all by the Covid crisis economically, overall supported similar kinds environmental policies. Among the four levels of economic vulnerability, their general support, for policy changes, however, was the lowest (2.91 versus 3.41, 3.46 and 3.64). Similarly, regarding support for luxury taxes and termination of child benefits, theirs' was also the lowest among the four groups. The relatively low support for policy measures that may lead to environmental change is correlated with the relatively low rate of actual change in consumption patterns that this group demonstrated during the closure. As a general rule, of the four levels of economic harm, the most affluent group reported the lowest rate of changes in attitudes and behavior. We believe that these findings are consistent with the notion that all sectors of the general public should be made aware of the ecological significance of consumer choices, and the potential of public policy to affect environmental outcomes.
In light of these findings, there are two possible explanations for the enhanced environmental commitments expressed in the wake of the Corona pandemic in Israel. The first is that environmental quality (e.g. air and water quality) become an increasingly important element in people's general quality of life, even during times of economic recession (McCright and Dunlap 2011). The increasingly meaningful role of clean air can be understood in in light of the mobility restrictions imposed on citizens during the Covid closure, which limited allowable distance for walking or jogging.
The second explanation is that frequently, the positive external benefits of environmental policies are not considered sufficiently by the public and by decision makers when calculating the economic costs of environmental plans. As a result, environmental policy is unjustifiably perceived as particularly expensive (Matto and Anthony 2017). In any case, our findings confirm the notion that economic downturn need not undermine environmental progress. For example, the high support for environmental taxes among all respondents, regardless of the economic harm they endured, and among all socioeconomic levels, indicates that when emerging from the economic crises, there is a need to re-frame the environmental discourse, with a greater emphasis on environmental values in economic policy.
The importance of framing environmental discourse in positive environmental terms is supported by numerous other studies that found that anticipating negative emotions, by potential consumers constitutes a barrier preventing adoption of higher environmental involvement. In the field of sustainable consumerism (Claudy et al 2013, Onwezen et al 2013, Schuitema et al 2013. Indeed, positive anticipated emotions may actually serve as drivers of pro-environmental, consumer behavior. There is also a correlation between the degree of economic harm sustained, as well as socio-economic level and the degree of change in the environmental orientation of consumers during the closure. The lowest deciles, who tended to suffer the greatest economically harm, for example, increased the use of disposable plastic, which has a negative impact on the environment. In contrast, people in the higher deciles, only slightly affected by the economic downturn, actually increased their recycling activities.

Conclusions
Like previous studies around the world (Mayer and Smith 2017, Harrabin 2020), our findings suggest that there is considerable room for ambitious environmental policies, even under challenging economic conditions. This is especially true during the recovery phase. The potential for new consumption patterns to be environmentally beneficial underlies the importance of choice architecture (nudges): often for only a modest aggregate cost, an environmental option can become the more affordable one. As economic uncertainty pushes all of society towards greater frugality (with prices dominating consumer decisions) such interventions are more important than ever.
We also emphasize the importance of making environmentally friendly alternatives the least expensive option. For example, notwithstanding their lower carbon footprint, in Israel, plant-based food products, such as vegetables and soy milk are significantly more expensive than food products based on animals, such as eggs, cheese and chicken, which enjoy government subsidies. Moreover, during the first Corona-driven closure, for several weeks in Israel, there was a shortage of eggs. The government quickly responded by importing eggs at a cost of 12 million NIS (Moses 2020).
The increase in the use of disposable plastic, most prevalent among the lower socio-economic deciles also requires a policy response. This should begin with education about the negative implications of disposable plastic. Experience suggests, however, that it is unwise to rely exclusively on voluntary changes in consumer consumption patterns. Effective regulation (including bans and phase outs) is the most effective way to reduce the use of disposable plastic.
The climate crisis, much like the Corona crisis, also requires new approaches to encourage sustainable consumption. This can take the form of regulation, choice architecture and ensuring accessibility to environmental infrastructure. It is important that decision makers are cognizant of socioeconomic differences and constraints. The exit strategy from the Corona crisis, therefore should be nuanced and informed by the different consumer patterns and financial realities across society.
As countries around the world face increasingly complex environmental and economic challenges, it is important to understand what factors explain the socioeconomic class differences in environmental attitudes and behaviors. These insights should inform the requisite policy interventions. Eliciting public support, and promoting individual actions is crucial to the success of environmental policy . This is especially true in Israel, given its fragile political climate. For the Israeli government, efforts to move the country out of the Covid-induced economic crisis constitutes an important opportunity for fostering green values. As this study suggests, with properly framing of public policy measures for citizens at all socioeconomic levels (Accenture 2020a, 2020b) a convergence between economic stimulus exit strategies and environmental progress is possible.

Data availability statement
The data that support the findings of this study are available upon reasonable request from the authors.         Increasing the minimum wage .769 Raising the price of electricity .695 Raising university tuition .599 Applying a waste clearance fee, for unusual amounts of waste  Since the harm variable is constructed so that a higher score = lower harm, the significant positive relationship shows that the higher the deciles, the less harm there is to the employment situation.