Frequent and informal transfers with kin and social networks have been well documented in Africa (and other parts of the globe). Strong social norms often dictate that an individual who has cash on hand provides support to friends and relatives. This support can be important for households in times of crisis. But what are the potential costs of these redistributive pressures? Specifically, can such norms result in individuals shifting labor, savings, and investment behaviors at a cost in terms of increasing their own economic standing? In this way, the kin system can, at times, be a poverty trap when it results in a “status quo bias” and collective (if unstated) opposition to or costs of efforts for individuals to purse income opportunities (see, for example the discussion in Hoff and Sen 2006). Carranza, Donald, Grosset, and Kaur test this with an innovative study in Cote d’Ivoire focusing on labor supply and earnings. And they find big (big!) effects.
Before getting to their study, I want to commend these authors for an excellent and incredibly thorough collection of citations. For anyone wanting to know what papers are out there on the topic of “social tax” broadly, this paper is a great resource.
The setting is cashew-processing plant in Côte d’Ivoire , where they study a sample of 474 piece-rate workers (almost all women). The paper has a nice description of the social tax phenomena among these workers before jumping into their intervention. The workers, all women, transfer on average between 25-35% of their earnings to people not in their household, most (~70%) are extended family and mostly people not working in the factory. Requests for cash tend to come on or just after payday (twice a month), when workers are paid in cash. Workers report that if they earn more, they expect to get more requests from their network. And they report that there is a social cost (described as stigma and isolation) to not sharing their income. Workers partly avoid this social tax by deploying strategies to get rid of their cash: buying household items immediately after payday, storing money with others, and participating in ROSCAs.
The intervention entails offering these largely unbanked workers a blocked savings account for earnings increases above baseline earnings. Workers pick a threshold above their baseline income where additional earning beyond the threshold will be deposited. These funds are blocked for 3-9 months. These accounts, as the paper says, “are designed to make it more likely that any increases in productivity are retained by workers for their own future use,” hence reducing the social tax (and increasing labor supply).
But, of course, blocked accounts also might address self-control problems, thereby also resulting in more labor supply. To further isolate the extent of the existence of a social tax and its labor effects, they assigned workers to either private accounts or non-private accounts. The non-private account is made known to the social network of those workers, explained to be part of the bank’s efforts to increase take-up generally. See the paper for more details, but in short this entails two text messages to the worker’s social network, one of which is timed with the unblock date. A third group of workers are not offer a blocked savings account. The authors note that their design builds on lab-in-the-field experiments which tests if people are willing to pay to prevent their social network learning about cash windfalls.
The experiment is divided into two phases. The first phase, simply offering blocked accounts, was intended to build trust and to check the operationalization of their experiment in general. Workers in the second phase are either in phase 1 or additional workers add, and all are assigned one of the two account types (private and not private). There are no non-treated workers in phase 2. This was explained as a strategic decision to maximize (presumably thru greater sample size with a fixed research budget) ability to test key assumptions. Since comparing the non-private account intervention to workers with no account was not core to the set of predictions the paper lays out, it’s a smart decision.
First, a look at take-up. Take-up of private accounts goes up from Phase 1 (43%) to Phase 2 (60%); 90% of those who has a private account in phase 2 and received another offer in Phase 2. This is described as a signal of trust among the workers to this banking service. I would also add that it is an important lesson on considering the sensitivity of respondents to take-up when offering something unusual/new.
For non-private accounts: yikes, only 14% take-up the offer. On it’s own, this could be driven by many factors. But these authors get out in front of this concern with additional data that suggests that concerns about a social tax are the reason for this low take-up. Almost all (96%) who decline the offer report that they worry that if social networks are looped-in about their accounts, they would get more transfer requests.
Account thresholds were relatively high, and yet actively used, consistent with the labor and earnings effects they find. For those offered a private account, income goes up by 11% (that’s big), in part driven by a decline in absenteeism. For each two -week period, workers with the private account work more than one extra day. Reducing redistributive pressure has sizable labor impacts. Income from other activities (outside the factory) does not change. These effects are larger for workers who report greater social pressure to hand over some of their earnings. Is this mainly happening with couples (within household)? No. They find even larger impacts for women with no partner.
What happens with the private account workers have their cash unblocked? The evidence is suggestive that they increase their transfers. This could reflect some altruism as well as some social tax that occurs when the account gets unblocked (these accounts won’t always stay a secret). This paper also tackles an array of other explanations consistent with their main findings to that of a social tax: is this reflecting self-control, did those with accounts become better goal-setters, are privacy concerns explaining the unblocked account results, and are unselected workers disgruntled and working less (no, no, no, and no).
Bottom line: the social tax is high. They calculate that the average worker faces a social tax of 9-14%, with some estimates much larger depending on labor supply elasticity assumptions. As the authors say, social insurance can turn into social taxation and, then, lower labor effort and earnings. These distortions are economically meaningful. They can contribute to an understanding of labor recruitment, retention, and productivity challenges facing formal firms in Africa. Circling back to the start of this blog, the underpinnings of a social tax (bad!) are also those of social insurance and risk-sharing (good!). So the challenge, as the authors also conclude, is how to develop scalable tools to lower social taxation without undermining risk-sharing arrangements.
Join the Conversation