Material Scarcity and Unethical Economic Behavior: A Systematic Review and Meta-Analysis


 Individuals around the globe experience different forms of material resource scarcity in terms of aspects such as hunger, thirst, or financial strains. As experiences of material scarcity have been found to make individuals more risk-taking, impulsive, and focused on regaining resources in the short-term, a growing body of research has investigated how such scarcity affects moral economic behavior. Yet, findings remain mixed and at times contradictory, thus calling for a systematic meta-analytical review on this overarching topic. In this pre-registered systematic review and meta-analysis, we evaluate qualitatively and quantitatively how material resource scarcity affects moral economic behavior. We analyze a comprehensive dataset including 44 published and unpublished studies comprising a total of 6,921 respondents across four distinct types of material scarcity: financial scarcity, physiological scarcity, scarcity reminders, and lower social class. Our findings show that acute scarcity significantly increases the propensity to engage in unethical economic behavior (gfinancial = .24, gphysiological = .39, greminders = .32). Importantly, we find no evidence that chronic experiences of scarcity in the form of low social class affect unethical economic behavior (gsocial class = .02). These results appear robust to the influence of publication bias and contextual sensitivity. We discuss how these findings advance our understanding of the psychological and moral consequences of scarcity and elaborate on implications for public policy.

Here, we present the first pre-registered systematic review and meta-analysis on the 11 relationship between experiences of material scarcity and unethical economic behavior. Based 12 on the systematic review, we argue that it is important to distinguish between different types 13 of acute and chronic experiences of material scarcity to understand their impact on financial 14 outcomes. Specifically, we identify financial and physiological scarcity as well as reminders 15 of scarcity as three fundamental forms of acute scarcity, whereas social class constitutes a key 16 indicator of chronic scarcity. 17 Using this typology, we then provide the first meta-analysis on how material resource 18 scarcity might affect individuals' propensity to engage in unethical economic behavior. 19 Analyzing 44 published and unpublished studies including 6921 respondents, we find that 20 acute financial and physiological scarcity as well as reminders of scarcity significantly increase 21 individuals' propensity to engage in unethical economic behavior. Importantly, we find no 22 evidence that more chronic material scarcity, in the form of lower social class, impacts 23 unethical economic behavior. Thus, based on the extant body of published research, the 24 propensity to engage in dishonest behavior for direct monetary gains appears to operate 25 irrespective of social class and is equally likely to occur among individuals from lower and 1 higher social classes. 2 Overall, the findings from the systematic review and the meta-analysis do not support 3 the general claim that scarcity increases the propensity to engage in unethical economic 4 behavior. Instead, our results highlight the importance of distinguishing between different 5 types of material scarcity, where only temporary or acute experiences of scarcity seem to 6 impact people's moral decision-making in the financial realm. Sensitivity analyses conducted 7 as part of our meta-analysis indicate that the results are not strongly impacted by contextual 8 sensitivity (cf. Van Bavel et al., 2016) and robust to the existence of publication bias (cf. 9 Mathur & VanderWeele, 2020; Simonsohn et al., 2015). 10 Together, the results from the qualitative systematic review and the quantitative meta-11 analysis highlight that the seemingly mixed findings on the relationship between material 12 resource scarcity and unethical economic behavior can be disentangled by distinguishing 13 between different types of temporary and acute versus more chronic forms of scarcity. 14 Consequently, rather than focusing on consequences of "the scarcity mindset," scholars may 15 benefit from theorizing on the psychological architecture and consequences of different forms 16 of material scarcity. As acute experiences of material resource scarcity can in effect apply to 17 individuals around the globe, we argue that the current review has important implications for 18 the theoretical understanding of how material scarcity can distort human moral judgment and Importantly, some research on how material scarcity affects decision-making argues 20 that the "scarcity-mindset" is induced independent of the resource in question. That is, whether 21 the current scarcity is experienced as a lack of food, water, or financial resources, the effect of 22 this lack of necessary resources will lead to similar cognitive and behavioral outcomes across 23 different types of material resource scarcity (Mullainathan & Shafir, 2014). It should be noted 24 that this conceptualization of scarcity does not entail that poor individuals are worse decision-25 makers (Mullainathan & Shafir, 2014). Instead, it prescribes to the idea that cognizant 1 experiences of relative scarcity make the individual focus on regaining the lack of resources in 2 the short-term, regardless of the individual's social class or demographic profile. In what follows, we present a systematic review 10 of extant scientific studies on this topic, which aligned with the inclusion criteria for the 11 subsequent meta-analysis. The aim of this section is threefold: (1) to provide an in-depth 12 overview of the different types of material scarcity that have been studied, (2)  Gino and Pierce (2010), investigated how economic inequality, as manipulated by 24 resource allocation (allocated randomly or subjectively), affected moral behavior in the form 25 of either helping or hurting others. Results showed that inequality significantly predicted 1 increased levels of dishonest behavior for individuals with less resources and a follow-up 2 experiment indicated that people behaved unethically to restore the perceived inequality. Also, 3 Sharma et al. (2014) showed that manipulating resource scarcity in the form of financial 4 deprivation lead individuals to cheat more for economic gains, and to judge such behavior as 5 being less immoral. This effect was mediated by whether individuals considered the 6 experienced deprivation as an acceptable reason for engaging in unethical behavior and if they 7 were made aware of that the unethical behavior could not alleviate the experienced scarcity, in 8 which case the effect no longer emerged (Sharma et al., 2014). 9 More recently, Birkelund and Cherry (2020) found that individuals experiencing 10 financial inequality (vs. equality) cheated significantly more for monetary resources in an 11 experimental dishonesty task and justified such behavior as result of the experienced scarcity. 12 Relatedly, Gino and Pierce (2009b) investigated the influence of relative resource scarcity, 13 indexed by inequity in endowments between individuals, on unethical economic behavior in a 14 laboratory setting. Albeit the incentives for dishonest behavior varied, results consistently 15 showed that inequity between partners in the experiment increased cheating and, importantly, 16 that pure self-interest was not the prime mechanism in the causal chain; rather, individuals 17 engaged in cheating due to emotional reactions (i.e., envy) elicited by the perceived resource 18 scarcity. Results also indicated that dishonest behavior triggered by inequity made individuals 19 more inclined to engage in dishonest helping due to empathy with the less fortunate partner 20 creating an environment where one's current relative financial scarcity was visually 1 emphasized, the authors documented that participants cheated more to alleviate this state, with 2 the cheating also provoking emotions of envy towards wealthy others. In a similar vein, an 3 experiment by John et al. (2014) used performance-based pay-rates, and found that dishonesty 4 emerged in individuals with lower pay-rates, but only when it was salient to them that there 5 was an opportunity of gaining a higher pay-rate (relative to their own rate), again emphasizing 6 the effects that relative scarcity has on activating a competitive orientation and a maximizing 7 mindset (Goldsmith et al., 2018;Roux et al., 2015). 8 In relation to specific cognitions and behaviors stemming from scarcity, scholars have 9 argued that such cognitive and behavioral responses could be the result of an evolutionary 10 response to harsh environments. Notably, the use of fast life history strategies (i.e., short-term 11 mating, low group altruism, higher criminal record, and higher risk taking) have been shown income rice farmers in Thailand would be more inclined to cheat for monetary gains when 24 experiencing resource scarcity. Unethical behavior was not found to increase by experiences 25 of scarcity; however, reminders of social-norms of morality cut cheating behaviors for richer 1 individuals, while it had no effect for individuals experiencing resource scarcity (Boonmanunt 2 et al., 2020). 3 Aksoy and Palma (2019) investigated the effects of scarcity on cheating and in-group 4 favoritism, using a lab-in-the-field experiment in Guatemala. While the authors found no effect 5 of scarcity on cheating for economic gains in a well-validated cheating task (Fischbacher &  6 Föllmi-Heusi, 2013), they did find that affluency (vs. scarcity) increased cheating behaviors 7 directed towards one's in-group. The mixed evidence of these findings taps into the ongoing 8 academic debate on how psychological constructs developed primarily in WEIRD (Western, 9 Educated, Industrialized, Rich and Democratic) contexts might not generalize to non-WEIRD  Grundmann and Lambsdorff (2017) found no effect of relative resource scarcity on 19 unethical behavior in a task where participants could earn different levels of income based on 20 skill and luck, and then rolled a die (either privately or in public) that would determine their 21 tax-rate of the earned income. Resource scarcity was manipulated by how much money 22 individuals earned in the task, while participants had the opportunity to cheat in the experiment 23 by reporting a lower tax rate (i.e., a lower die roll) in the private die-roll. In contrast to the 24 thesis that resource scarcity increases unethical behavior, Grundmann and Lambsdorff (2017) 25 found that individuals with higher incomes cheated marginally more to gain a lower tax rate.   The studies identified as eligible for inclusion were coded with respect to the dependent and 3 the independent variable. The coding procedure was carried out by two trained student coders, 4 with backgrounds in economics and cognitive science, and all coded information was verified 5 subsequently in correspondence with the first author. Specifically, we extracted information on 6 the dependent variable, independent variable(s), mediators, moderators, population type (e.g., 7 students, children, etc.), location, site of the study (i.e., onsite vs. online), research design (e.g., 8 lab experiment, lab-in-the-field experiment, etc.), and participant compensation from all the 9 included articles. Independent variables were divided into four sub-groups: Financial scarcity 10 (i.e., poverty and economic inequality), Social class (socioeconomic status), Physiological 11 scarcity (i.e., hunger and thirst), and Reminders of scarcity (i.e., the activation of cognitions 12 related to scarcity). 13 The coders also extracted the appropriate information from the reported test statistics Hedges & Olkin, 2014). 22 All effect sizes were computed and coded individually in the dataset using the R-1 package esc for meta-analytic effect size computation (Lüdecke, 2018). Wherever possible, 2 effect sizes were calculated based on means, standard deviations, F-statistics, t-statistics, r-3 statistics, and p-values. All reported tests of a given article concerning the relationship between 4 resource scarcity and moral behavior were coded in the dataset and the main test of each study 5 was dummy coded (1 = main study result on unethical behavior, 2 = additional test of unethical 6 behavior) to clarify the main findings of each article, while still preserving individual tests of 7 the relationship based on different sub-group analyses (e.g., gender differences, physiological 8 differences, etc.). Studies reporting significant as well as insignificant statistical tests were 9 included in the dataset. were randomly assigned between the three coders. As our sample of articles was smaller (N = 1 28), all three coders evaluated each of the included articles, but in a randomized order. Vitally, 2 this process still allowed us to calculate interrater reliability between the coders through 3 intraclass correlation coefficients (Bartko, 1966;Fleiss & Cohen, 1973). 4 In the rating of the articles, the three coders assessed how likely the findings of the 5 individual articles were to vary by context, which could be in terms of location (i.e., WEIRD 6 vs. non-WEIRD countries/regions, cf. Heinrich et al., 2010), sample type (i.e., student vs. non-7 student populations, differences in racial characteristics, etc.), time (i.e., before vs. after large 8 political changes, before vs. after recessions, etc.). Thus, the coding scheme aimed to capture 9 a selection of macrolevel contextual effects that could be expected to influence the 10 generalizability of the included research on scarcity and unethical economic behavior    The results of the P-curve analysis revealed that the curve was significantly right-1 skewed (z = -6.049, p < 0.001), while the test of curve flatness was insignificant (z = 2.612, p 2 > 0.999). The mean post-hoc statistical power for the significant studies was 69%, with a 95% 3 confidence interval ranging from 47.3% to 84.2%. These results suggest that a "true" effect is 4 present in the included studies, meaning that the findings cannot easily be explained through The results of our analysis of contextual sensitivity of the included studies disproved the 5 prediction that individual study characteristics would be correlated with the rated level of 6 contextual sensitivity in a (Bayesian) Pearson's correlation. Results are reported in Table 2. 7  In sum, contrary to what we hypothesized in our pre-registration, the results of our 16 contextual sensitivity analysis indicate no statistically significant relationship between how 17 contextually sensitive the included studies were rated and the magnitude of the extracted effect 18 sizes, standard errors, sample sizes, and p-values. Hence, our analysis of contextual sensitivity conveys two main messages: (1) that findings on the relationship between resource scarcity 1 and unethical economic behavior can be considered fairly generalizable across contexts, and 2 (2) that the influence of contextual sensitivity in general might not be as predictive of study 3 outcomes and future replication success, supporting Inbar (2016)  been utilized to investigate this relationship empirically. Importantly, general conclusions have 20 been lacking due to mixed and, at times, contradictory findings. This pre-registered systematic 21 review and meta-analysis sought to synthesize existing studies and establish the current known 22 aggregate effect of whether resource scarcity influences individuals' propensity to engage in 23 unethical economic behavior. Overall, our results show that acute experiences of financial 24 scarcity and physiological scarcity can shift people's propensity to engage in unethical 1 economic behavior. We also show that scarcity reminders exert the same impact. Importantly, 2 our results show that more chronic experiences of scarcity, in the form of lower social class, 3 do not have this effect. Thus, when studying the relationship between scarcity and direct 4 cheating for monetary gains, our findings indicate that it may not be sufficient to rely on "the 5 scarcity mindset" account. Rather, the present results imply that scholars may gain a deeper 6 and more nuanced understanding into this relationship by distinguishing between different 7 types of acute and chronic experiences of material scarcity. 8 A crucial distinction to make based on the present research is that social class is not a 9 predictor of unethical economic behavior: individuals from lower social classes are not more 10 inclined to engage in unethical economic behavior compared to their wealthier counterparts. 11 Instead, our results suggest that more acute experiences of relative material scarcity can 12 increase individuals' tendency to engage in unethical economic behavior to counteract and 13 alleviate the experienced lack of resources. Certain limitations of our analysis should be noted. First, the pooled estimate from our meta-5 analysis needs to be interpreted against the assumption of potential biases. While our meta-6 analysis was found to be robust to the impact of publication bias, at least based on conventional 7 tests to detect this bias source, the possibility of publication bias cannot be excluded. Our 8 analysis indicated that significant (vs. non-significant) studies would need to be close to 7 times 9 more likely to be published, which is not unlikely to be true considering previous research on 10 publication bias (Sterling, 1959;Sterling et al., 1995). Nevertheless, considering the number 11 of studies that reported non-significant results and were still published, we consider it unlikely 12 that publication bias should have a material impact on our results. this discrepancy largely stems from the fact that we used expert raters from three distinct 5 disciplines, which should be a more robust measure of contextual sensitivity, future research 6 could aim to assess this construct with different methods, such as crowd-sourced academic 7 rating schemes (Tierney et al., 2021). 8 Finally, it should be noted that this article has focused specifically on unethical 9 economic behavior. When discussing the larger topic of the relationship between scarcity and 10 morality, it is important to be aware of the sensitivity of the issue and the multiple dimensions 11 of ethical behavior. Hence, while it may be unethical economic behavior if hungry individuals 12 steal money to feed their starving children, a different philosophical perspective could 13 emphasize that such behavior could also reflect a humanitarian and parental action made to 14 protect innocent children. Thus, while a given behavior might be considered immoral from one 15 perspective, it may be perceived as defensible and even altruistic from other perspectives. 16 17

Directions for Future Research 18
The present work is the first of its kind to provide a generalizable quantification of the 19 relationship between material resource scarcity and unethical economic behavior. Furthermore, 20 our results suggest that more chronic forms of scarcity (i.e., lower social class) do not make 21 individuals more inclined to engage in such financially fraudulent actions. However, it remains 22 unknown whether individuals with lower (vs. higher) social class may experience acute 23 influences of relative material scarcity differently with respect to moral judgment and decision-24 making. Given that such an investigation could have important implications for structuring 25 policy initiatives aimed at helping individuals experiencing chronic resource scarcity, future 1 empirical work should examine this possibility. 2 We acknowledge that the meta-analysis is based on a relatively small sample (k = 44), 3 highlighting the rather narrow set of existing studies in this domain. Although the results of the 4 sensitivity analysis indicate that the findings from the meta-analysis are robust by conventional 5 standards, future research should aim to further test how different forms of material scarcity, 6 both acute and chronic, affect moral economic behavior across populations, contexts, and study 7 paradigms. 8 While our pre-registered work focused on the impact of material scarcity on 9 individuals' inclination to engage in unethical economic behavior, our review protocol also 10 resulted in a series of articles investigating how this form of scarcity may affect both unethical 11 and prosocial behaviors or prosocial behaviors alone. As a supplement, we included the articles 12 that focused on prosocial behavior but not unethical behavior in a separate meta-analysis 13 reported in Appendix B. It is likely that we did not identify all studies on the relationship 14 between material scarcity and prosocial economic behavior, given that our review protocol was 15 not designed to do so. Nevertheless, the findings from this supplementary analysis showed a 16 small positive and significant effect between material resource scarcity and prosocial economic 17 behaviors. Again, this adds to the inconsistencies in the literature on how resource scarcity 18 affects moral behaviors, and future research is needed to investigate these relationships further. 19 20

21
The results of this systematic review and meta-analysis show that acute experiences of relative 22 resource scarcity in the form of financial scarcity, physiological scarcity, and scarcity 23 reminders increase individuals' inclination to engage in unethical economic behavior. In 24 contrast, more chronic experiences of scarcity in the form of lower socioeconomic status do 25 not increase the propensity to engage in such behavior. Thus, our investigation highlights that 1 individuals from lower social classes are not more immoral economic decision-makers than 2 their wealthier counterparts. Instead, our findings suggest that acute experiences of relative 3 resource scarcity make individuals more inclined to engage in unethical economic behavior, 4 regardless of their social class. Thus, this meta-analysis emphasizes the benefits of 5 complementing "the scarcity mindset" by distinguishing between different types of acute and 6 chronic experiences of material scarcity to enrich our understanding of scarcity effects in the 7 moral psychology domain.

Inclusion Criteria and Integrated Scientific Works
This section provides more detail on the article selection process. First, we only included articles which concerned material resource scarcity. Some articles from our database search concerned cognitive depletion and cognitive scarcity, but such studies were not included based on our pre-registered research rationale. Second, our database search resulted in several studies that only dealt with material resource scarcity but not unethical behavior or vice versa. Because such studies did not investigate the relationship between these two factors, they were also excluded. Third, we excluded review articles as they did not abide to empirically test the relationship between resource scarcity and unethical behavior. If the criteria specified in our pre-registration were fulfilled, we extracted the effect size(s) from the available test statistics in the articles. In cases where such information was not available, we contacted the authors to retrieve the necessary data used in the chosen effect size calculation(s) (Hedges' g). Two articles had to be excluded, because the authors did not respond to requests about data sharing  Table A1 represents all studies and conditions on the relationship between resource scarcity and unethical behavior, while Table A2 represents all studies and conditions on the relationship between resource scarcity and prosocial behavior. Each row represents an extracted effect size.

Appendix B Supplementary Review and Meta-Analysis on Prosocial Behavior
As noted in our methods section (see sub-section, Search), our pre-registered review protocol (https://bit.ly/3t1zd8z) restricted our search terms and inclusion criteria to research that had investigated the relationship between resource scarcity and unethical behavior. However, as previously noted, our database search resulted in a series of articles that either investigated how scarcity might affect both unethical and prosocial behavior or which purely sought to investigate how scarcity might affect prosocial behavior. While considered irrelevant to the main body of the systematic review and meta-analysis, we decided that it would still be relevant to outline the results of such studies in connection with our main focus on unethical behavior.
Consequently, the below section outlines the meta-analytical results on how resource scarcity might affect prosocial behavior.

Meta-analysis
As with the main analysis presented in the paper, the meta-analysis of the relationship between resource scarcity and prosocial behavior followed a hierarchical approach, in which we firstly analyzed a global model on all the included studies, before breaking the analysis down to the subgroup level. Here, it should be noted that this supplementary analysis only included three subgroups: Financial Scarcity, Reminders of Scarcity, and Physiological Scarcity. The results are presented in Table B1. Again, the key variables of interest in this analysis is the Standardized Mean Difference (SMD) in the form of Hedges' g, the percentage of variability in effect sizes # , and the between study variance in the form of # .  analysis is presented in Figure B1. Our analysis showed no difference in effect size estimates between the two methods.
The results of the sensitivity analysis are presented in Figure C1. This analysis hence supports our initial prediction that the main results of our meta-analysis is robust to the use of different estimation algorithms. Figure C1. Sensitivity analysis of the difference between the Hartung-Knapp-Sidik-Jonkman estimation algorithm and the DerSimonian-Laird estimation algorithm. The algorithm used in the main meta-analysis is robust due to the extremely small difference in pooled estimates.

Meta-analysis of all extracted effect sizes
To further assess the robustness of our main findings on the relationship between resource scarcity and unethical behavior, we carried out a random-effect model meta-analysis by the use of every single extracted effect size from the included articles. While this entailed a redundancy in sample sizes (i.e., multiple tests for the same sample size) and, consequently, high levels of heterogeneity, we argue that such an analysis provides further validity and robustness of our main findings, given that this analysis was conducted on a much larger data set than that of the original. Table C1 reports the results this analysis. Note. k = number of studies, N = sample size, SMD = Standardized Mean Difference by Hedges' g, 95% CI = 95% confidence interval, p = p-value, # = percentage of variability in effect sizes, # = between-study variance.
The robustness analysis of all included effect sizes revealed a significant overall effect of 0.1973, close to, but lower than that of the main findings. As expected, the global model also revealed substantial heterogeneity # = 84.7%. An identical pattern emerged in the subgroup analysis. Here, all estimates for the subgroups were highly significant and yielded positive effect sizes for all groups except that of social class.
Following previous procedures, we conducted a GOSH analysis ( propensity to engage in unethical behavior, while social class does not.