The appropriate response of Spanish Gitanos: short-run orientation beyond current socio-economic status

Abstract

Humans differ greatly in their tendency to discount future events, but the reasons underlying such inter-individual differences remain poorly understood. Based on the evolutionary framework of Life History Theory, influential models predict that the extent to which individuals discount the future should be influenced by socio-ecological factors such as mortality risk, environmental predictability and resource scarcity. However, little empirical work has been conducted to compare the discounting behavior of human groups facing different socio-ecological conditions. In a lab-in-the-field economic experiment, we compared the delay discounting of a sample of Romani people from Southern Spain (Gitanos) with that of their non-Romani neighbors (i.e., the majority Spanish population). The Romani-Gitano population constitutes the main ethnic minority in all of Europe today and is characterized by lower socio-economic status (SES), lower life expectancy and poorer health than the majority, along with a historical experience of discrimination and persecution. According to those Life History Theory models, Gitanos will tend to adopt “faster” life history strategies (e.g., earlier marriage and reproduction) as an adaptation to such ecological conditions and, therefore, should discount the future more heavily than the majority. Our results support this prediction, even after controlling for the individuals' current SES (income and education). Moreover, group-level differences explain a large share of the individual-level differences. Our data suggest that human inter-group discrimination might shape group members' time preferences through its impact on the environmental harshness and unpredictability conditions they face.

Introduction

In nature, individuals of different species often have to choose between outcomes realized at different times (Ainslie, 1975; Rachlin & Green, 1972). These inter-temporal choices are also ubiquitous in the lives of humans, for instance, in the spheres of marriage and reproduction, education and work, as well as during social and market interactions (Espín, Brañas-Garza, Herrmann, & Gamella, 2012; Espín, Exadaktylos, Herrmann, & Brañas-Garza, 2015; Frederick, Loewenstein, & O'Donoghue, 2002; Nettle, Coall, & Dickins, 2011; Woodburn, 1980). When faced with such decisions, individuals tend to discount the value of delayed rewards. The preference for sooner-smaller rewards over later-larger rewards has been referred to as delay discounting (DD) (Frederick et al., 2002; Kirby et al., 2002). DD is considered to be a measure of one of the multiple domains of impulsivity, namely “impulsive choice” (Bevilacqua & Goldman, 2013; Reynolds, Ortengren, Richards, & de Wit, 2006).

DD tends to be a stable individual characteristic (Kirby, 2009; Ohmura, Takahashi, Kitamura, & Wehr, 2006) – although it may be momentarily influenced by short-term state manipulations (e.g., Kidd, Palmeri, & Aslin, 2013; Read & van Leeuwen, 1998) – and people differ greatly in the extent to which they discount the future (Frederick et al., 2002). However, the factors underlying such inter-individual differences remain poorly understood. On the one hand, there is evidence suggesting that DD rates are heritable to some extent (Anokhin, Golosheykin, Grant, & Heath, 2011; Anokhin, Grant, Mulligan, & Heath, 2015; Aycinena & Rentschler, 2017; Bevilacqua & Goldman, 2013). On the other hand, people's current socio-economic conditions, as proxied by variables such as education and income, also seem to be related to DD: poorer and less educated individuals have been found to discount the future more heavily (Harrison, Lau, & Williams, 2002; Kirby et al., 2002; Tanaka, Camerer, & Nguyen, 2010), although the causal direction is unclear (Becker & Mulligan, 1997). In addition, a number of behavioral disorders (e.g., attention-deficit/hyperactivity disorder, aggression, suicide, and substance abuse) have been associated with high DD (Barkley, Edwards, Laneri, Fletcher, & Metevia, 2001; Bickel & Marsch, 2001; Dombrovski et al., 2011).

The latter evidence has been taken to support the notion of high DD as a maladaptive trait. However, under certain socio-ecological conditions, discounting the future can be a contextually appropriate response. To be more specific, developing a preference for the short-run may be fitness-maximizing in harsh and unpredictable environments (Becker & Mulligan, 1997; Daly & Wilson, 2005; Frankenhuis, Panchanathan, & Nettle, 2016; Hill, Jenkins, & Farmer, 2008; Pepper & Nettle, 2017).

According to Life History Theory (del Giudice, Gangestad, & Kaplan, 2015; Kaplan & Gangestad, 2005; Roff, 1993), variation in life history traits (e.g., size and number of offspring, parental investment, longevity, time to first reproduction, sociability) can be understood in terms of trade-offs in the allocation of resources to competing life functions such as maintenance, growth, and reproduction. The accumulated set of resource allocation decisions during life constitutes the individual's life history strategy, which leads to the development of an integrated collection of life history traits. The most common approach to life history strategies poses a continuum from slow to fast (Promislow & Harvey, 1990). Leading models based on Life History Theory rely on this slow-to-fast approach to predict variation of traits and strategies both between and within species. In this vein, unpredictable and harsh environments are particularly related to the development of fast life history strategies that divert resources from long-term outcomes in favor of short-term outcomes, while predictable and secure settings lead to strategies in the opposite, slow end of the continuum (Kaplan & Gangestad, 2005).

Life history strategies are often seen as species-distinctive characteristics but humans (as well as other organisms) have evolved mechanisms of phenotypic plasticity that allow them to adjust life history strategies to match local conditions during their lifetime (Belsky, Steinberg, & Draper, 1991; Brumbach, Figueredo, & Ellis, 2009; Ellis, 2004; McCullough, Pedersen, Schroder, Tabak, & Carver, 2013; Nettle et al., 2011; Nettle, Coall, & Dickins, 2010; Worthman, 2003). These strategies would lead to the maximization of individuals' average lifetime inclusive fitness (del Giudice et al., 2015; Kaplan & Gangestad, 2005; Roff, 1993). However, it is worth noting that existing theories are almost entirely based on verbal models, or on formal models designed to study the evolution of variation in life history strategies between species/populations, which tend to ignore the plasticity in life history strategies within species/populations.¹ Few formal models to date have examined how suits of traits (e.g., onset of puberty, number of offspring, investment per offspring) should covary as a result of phenotypic plasticity (e.g., along a fast-slow continuum), depending on environmental conditions (Mathot & Frankenhuis, 2018). For pioneering verbal models of within-species/populations variation in life history traits along the fast-slow continuum, see Belsky et al. (1991), Ellis, Figueredo, Brumbach, and Schlomer (2009) and Réale et al. (2010).

Yet, although the theoretical literature on life history clearly stresses the role of environmental conditions in shaping individuals' DD, more research needs to be conducted to assess this link empirically. While a number of individual-level studies provide support for the hypothesized relationships by showing, for example, a connection between high DD and variables such as low income and education, which can be considered as proxies for unfavorable (potentially harsh and/or unpredictable) conditions (Becker & Mulligan, 1997; Chipman & Morrison, 2015; Green, Myerson, Lichtman, Rosen, & Fry, 1996; Griskevicius, Tybur, Delton, & Robertson, 2011; Kirby et al., 2002; Pender, 1996), few studies to date have examined to what extent the conditions of harshness and future-unpredictability faced by different groups can predict individual differences in DD. Given that those individual-level studies do not use grouping variables that are exogenous to the individual, they cannot put the focus on an environment-based analysis, as prevails in the theoretical literature. Ramos, Victor, Seidl-De-Moura, and Daly (2013) is, to the best of our knowledge, the first study directly approaching the question. Consistent with the theoretical predictions, the authors show that slum-dwelling youth in Brazil (highly exposed to violence) discount future hypothetical rewards more heavily than university students from more affluent neighborhoods. A more recent study enlarged the number of comparison groups (46 countries were used as observation units) and found that country-level life expectancy can predict the average DD in a sample of university students from the country (Bulley & Pepper, 2017), also as hypothesized by a life history approach. Finally, Lee, Debruine, and Jones (2018) extended the latter analysis to a larger sample from 54 countries using individual-specific life expectancy (i.e., age-, sex-, year-, and country-specific life expectancy), and show that the theoretical relationship holds especially when waiting is arguably more beneficial, that is, when the later-larger reward is relatively high and delay is short. While these pieces of work have significant value, in the three cases, the authors compare groups of young individuals who, moreover, differ in a large number of uncontrolled characteristics that are not necessarily related to harsh and unpredictable living conditions but may influence DD. Among those are neighborhood facilities (which may translate into different access to services, for instance), geography, political regime, climate, and so forth. Furthermore, the tasks used to elicit the participants' time preferences (i) did not allow for a parametrization of the individuals' discount functions and (ii) were survey-based, without real incentives. In this paper we further contribute to fill this empirical gap.

To test the predictions of the evolutionary framework of Life History Theory, this study explores the differences in DD between two populations which often face different socio-ecological pressures even if they live in the same geographic areas, even in the same villages and neighborhoods. More specifically, using data from an economic experiment involving real monetary rewards, we compare the discounting behavior of a sample of Romani people from Southern Spain (Gitanos, or Calé, as they typically refer to themselves) with that of their non-Romani neighbors (i.e., the majority Spanish population).² Technically speaking, we set up a quasi-experimental design where ethnicity is the only variable that a priori differs between treatment and control groups. Note that a pure experimental design cannot be applied here because the socio-ecological conditions under which the individuals develop, as proxied by their ethnicity, cannot be exogenously manipulated for obvious reasons. We consider that this design is as close to a controlled experiment as a study of these characteristics can be since ethnicity and its associated socio-ecological conditions are eminently exogenous, thus leaving little room for endogenous determination.

The localities where we conducted our experiments are characterized by a particularly high concentration of Gitano people, amounting to over 25% of the total population, compared to 1–1.5% in the whole of Spain (Gamella, 1996; Gamella, Martín Carrasco-Muñoz, & Quesada Garrido, 2014). However, the Gitano population is clearly differentiated in their demographic and cultural profile, and faces a markedly different socio-economic “ecology” than the majority. Hence it constitutes a paradigmatic ethnic group for the goal of this study. Notwithstanding this, it is worth noting that theoretically identical opportunities (Rawls, 1971; Sen, 1992) in terms of access to public education, social benefits and healthcare, are provided (at least legally, on paper) to the Gitano population, at least since the advent of democracy in Spain, four decades ago.

Compared to the dominant majority, the Gitano population of Spain is characterized by a lower socio-economic status (SES), including lower income and education, and also by poorer health, lower life expectancy and higher fertility rates (Casals et al., 2011; Cook, Wayne, Valentine, Lessios, & Yeh, 2013; Gamella, 2011; Gamella et al., 2014; la Parra Casado, Gil González, & de la Torre Esteve, 2016; MSC-FSG, 2005). These processes have generated a differentiated demographic profile. For instance, in the study area, the Gitano population had a mean age of 27 years compared to the near 42 of the overall local population (Gamella, 2011). Life expectancy at birth has increased in this population almost continuously since the mid 1940s. But still today it seems to be from 5 to 10 years below that of their non-Gitano neighbors (Gamella, 2011; MSC-FSG, 2005). This may have resulted in a different set of aggregated needs and outlooks.

However, despite a long-lasting coexistence in many local areas, social exclusion, forced assimilation and discrimination by the majority have considerably influenced the lives of Gitanos as well as of other Romani groups in almost every part of the world (Matras, 2015). In Spain this still affects the lives of Gitano people, particularly in the most segregated areas.

Several distinctive features of Gitano social life seem to reflect adaptations to these negative environmental and historical conditions. For instance, Gitanos maintain a strong and oppositional sense of identity and high levels of ethnic and familial endogamy. In the study area the Gitano minority presented rates of inbreeding five to eight times larger than those of the general population, and have maintained these until the present (Gamella & Martín, 2007; Martín & Gamella, 2005; Núñez Negrillo, 2016). These endogamous strategies tend to increase their social and perhaps their genetic homogeneity (Bittles, 2012). However they also might work as a protection against external threats associated with the discriminatory environments that Gitano confronted as a group (Fraser, 1995; Gamella, Martín Carrasco-Muñoz, & Núñez Negrillo, 2013; Matras, 2015). Interestingly, the rate of incarceration of Gitanos is still nowadays much higher than that of the majority population. As an example, in a number of recent studies, Gitano women accounted for over 25% of the female prison population, a huge over-representation (Cerezo, 2016; Feintuch, 2013).

In addition, most Gitano groups, as other Romani groups through Europe, also maintain patterns of early and pronatalist marriage. Gitano women in the study area were found to have a mean age of first childbirth of 18–19 years (over a decade earlier than the Spanish average), and total fertility rates that doubled and even tripled those of the Spanish population at large (Gamella, 2011; Martín & Gamella, 2005). Infant mortality rates displayed by Gitanos have declined sharply during the last 60 years but are still nowadays considerably larger (about 40%) than those observed in the non-Gitano population (Gamella & Martín, 2017; Martín & Gamella, 2005; MSC-FSG, 2005).

These patterns of Gitanos can be understood as life history strategies situated at the (relatively) fast end of the fast-slow continuum, which are typically adopted by people who grow up in unpredictable and harsh environments (Brumbach et al., 2009; Dickins, Johns, & Chipman, 2012; Frankenhuis et al., 2016; Johns, Dickins, & Clegg, 2011). When the future is uncertain or predictably harsh (Bulley, Pepper, & Suddendorf, 2017; McGuire & Kable, 2013), therefore, the appropriate response might be to develop a short time horizon (Becker & Mulligan, 1997; Daly & Wilson, 2005; Pepper & Nettle, 2017) and to adopt strategies such as giving birth, as soon as possible, to the maximum number of offspring (Dickins et al., 2012; Johns et al., 2011; Nettle et al., 2011; Worthman, 2003). In these ecologies, long-term resource allocation may not pay off because there is uncertainty (i.e. the distribution of outcomes has unknown properties such as the mean or SD due to random variation) that the organism will live until late adulthood. Although environmental harshness (risk of mortality-morbidity) and unpredictability (stochastic variation in salient environmental conditions which prevents ex-ante accurate predictions due to factors such as a lack of information or excessive complexity) are theoretically and empirically dissociable and may have differential effects on several life history traits (Belsky, Schlomer, & Ellis, 2012; Brumbach et al., 2009; Ellis et al., 2009), an orientation to the short-run is by definition predicted by both factors (Ellis et al., 2009; Frankenhuis et al., 2016; Roff, 1993). In addition, the evidence suggests that the effects of environmental harshness and unpredictability on life history strategies are additive, not interactive (Brumbach et al., 2009). We further expand on this point in the Discussion section.

In this vein, the aforementioned models based on Life History Theory predict that Gitanos will tend to display higher DD rates than their non-Romani neighbors, due to the different ecologies faced. Moreover, it is expected that individual socio-economic factors such as current (at adult age) education and income account for some but not all of the difference because life-history-related behavioral traits such as temporal orientation are thought to be mainly shaped in earlier stages of development (e.g., Belsky et al., 1991; Griskevicius et al., 2011; McCullough et al., 2013; Mell, Safra, Algan, Baumard, & Chevallier, 2018). Group differences should indeed explain a large share of the individual differences, given the shared environmental influences during development within each group.