Pepper & Nettle (P&N) make an ambitious and compelling attempt to isolate a common cause of a large set of related behaviors that seem to differ across rungs on the socioeconomic ladder. The focus of this paper is a set of behaviors that are related through their temporal aspect of “front-loading” consumption and gratification to the present despite potential negative consequences in the future. The common cause is hypothesized to be limited control associated with lower socioeconomic status (SES). We agree with the authors' view that limited control, as well as the perception of limited control, can indeed help explain part of the difference in observed present-oriented behavior between the poor and the rich. However, we disagree with the use of mortality risk as the primary illustrative example of a mechanism leading to this apparent impatience: In our view, several other mechanisms are more plausible.

Intrinsic mortality appears like a reasonable mechanism behind present-oriented behavior observed in low-SES individuals: If mortality risk makes the future uncertain, it makes sense to consume now. When we combine data on temporal discounting from 53 different countries recently reported by Wang et al. (Reference Wang, Reiger and Hens2016) with country-specific mortality rates from the World Health Organization (WHO), we do indeed find a negative correlation between mortality and discount factors (r = −0.36, p = 0.0096), with lower discount factors being associated with more discounting (see also Heimer et al. Reference Heimer, Myrseth and Schoenle2017 for further evidence). However, observed rates of temporal discounting are much too high to be accounted for by mortality risk, even when we generously ascribe all mortality risk to extrinsic causes. Specifically, people discount 46% over one year in Wang et al. (Reference Wang, Reiger and Hens2016) – that is, they are indifferent between receiving a payment of $x one year from today and $x * 0.46 today, which translates into a required interest rate of more than 116%. However, average mortality risk over one year in the countries in this dataset is only 0.148%;Footnote ¹ thus, if the risk of dying before a future payment were realized were the only factor influencing discounting, people would be indifferent between receiving $x in one year and $x * 1/(1 + 0.00148) = $x * 0.999 today. Mortality risk can therefore only account for 0.13% of the observed discounting. To produce discounting on the order of magnitude observed in the data, people would have to mis-estimate the prevailing mortality risk by a factor of 769. This mismatch would be even more egregious if we restricted the mortality risk to extrinsic (i.e., uncontrollable) causes, as argued by P&N, and would remain significant even when using the lower discount rates typically estimated with the convex time budget method (Andreoni & Sprenger Reference Andreoni and Sprenger2012). Thus, even if mortality rates partially explain the behavioral constellation of deprivation, it seems unlikely that it is the most important explanatory factor.

However, in our view, the authors' main hypothesis is correct; in the following, we illustrate three examples for the kind of uncertainty that could produce differences in observed discount rates at different rungs of the SES ladder.

First, poor individuals often face unpredictable income streams and liquidity constraints. The magnitude of these fluctuations can be substantial, and they mechanically create a preference for immediate payments over delayed payments. An example comes from Carvalho et al. (Reference Carvalho, Meier and Wang2016), who study time preferences of poor individuals before and after payday, finding that these people are more present-biased before payday for monetary but not effort outcomes. These findings suggest that liquidity constraints and income uncertainty in resource-poor contexts can lead to a focus on the present.

A completely different illustration of the environmental risks faced by individuals in low-income contexts comes from attrition rates in household surveys, such as those often undertaken by economists in developing countries. In our own work in Kenya, we typically expect 10% to 15% of attrition between survey rounds one year apart; i.e., we cannot find the same respondent one year later, even though the survey usually provides incentives on the order of half a daily wage. Now, imagine relying on others as business partners to deliver on promises in this context, relying on return visits from a health professional, or relying on public service delivery from government officials: It is likely that even higher rates of “attrition” are found in such situations, creating strong incentives to realize transactions now rather than later.

Finally, although the above risks are external, in our view there is a significant “internal” risk that creates incentives to act now in poor contexts: forgetting. In our own work, we have found that when individuals in Kenya have the opportunity to earn half a day's wage by simply sending a text message on a specified future date, they forget to perform this simple action at high rates, reaching about 50% for delays of a month (Haushofer Reference Haushofer2015). When this risk is combined with the inferior availability of automatic transactions or reminder technology in resource-poor contexts, it creates strong incentives to want to act today. In line with this argument, our respondents in Kenya actually prefer to send the text message sooner rather than later, because they are aware of their own likelihood of forgetting.

In sum, P&N have outlined a provocative and compelling hypothesis for the prevalence of short-sighted behaviors in resource-poor contexts. Their hypothesis makes several testable predictions. Most importantly, it predicts that individuals in resource-poor contexts should want to act now for gains and losses: They should not only wish to consume immediately but also wish to incur costs that lead to larger benefits immediately rather than later. Our text-message study is one such example; one might imagine similar studies that use health behaviors, such as vaccination, as outcome variables. Future work of this kind promises to provide important insights into the mechanisms that drive behavior at the low end of the income distribution and point to interventions that could improve health and other outcomes in these populations.