Efficiency and Gender Bias in Business Investment Decisions by Households: A Lab Experiment in Ghana

This lab-in-the field experiment has been designed with the goal of evaluating two related hypotheses: (i) limited cooperation within household is associated with loss of income and (ii) this limited cooperation is associated with gender biases in resource allocation, and as a result affects disproportionately the business opportunities controlled by women.

This project is motivated by findings from earlier studies which showed that cash grants had a positive effect on men-owned but not on women-owned businesses, while in-kind transfers have a positive effect on both men- and female-owned businesses (Fafchamps et al. 2014). Decision-making processes within the household that divert resources away from investing in women-owned businesses could explain this. The money is for example spent on the business of the husband: Bernhardt et al. (2017) show that women's enterprises receive less investment in a cash transfer program when their partner also runs a business, compared to women from households where she is the only business owner in the household. This behavior is not necessarily driven by gender biases: Bernhardt et al. show that in many cases the return to investment in the man's business are higher than investing in the woman’s business, partially due to differences in sectoral choice.

In our experiments, we control for differential returns to investment by experimentally varying whether the man or the woman owns a business with a profitable investment opportunity. This way we can identify gender-driven differences in investment behavior of couples and the extent this results in inefficiencies.

The existing literature provides a range of insights into the incomplete overlap of household members' preferences, and in the cost of limited cooperation when these preferences diverge. It shows that households' choices regarding resource allocation vary depending on the gender of the household member receiving (and controlling) different income streams (Duflo 2003, Duflo and Udry 2004). It also suggests that household make costly choices when the preferences of their members are far apart (Schaner 2015). This experiment builds upon these insights, observing variations in husbands and wives' decisions to invest in either shared or common opportunities, depending on who controls the returns associated with these opportunities.

Specifically, to test for the prevalence of gender biases in households' decisions over resource allocation, we implement a number of decision-making games with female entrepreneurs and their partners. In these games, participants face a tradeoff between keeping the money for themselves, investing this in a business opportunity owned by one of them or investing this in a shared opportunity (see Munro 2017 for an overview of experiments using this particular game).
Additional variations on these voluntary contribution mechanism (VCM) games are implemented to assess the additional effect of different playing conditions, including:

1. switching from private to joint decision making;
2. introducing the possibility the owners of the business opportunity can reallocate their earnings to their partners;
3. introducing two "business opportunities", one owned by the male partner, the other by the female partner, with different level of returns;
4. having participants play with a stranger rather than their spouse;
5. introducing joint ownership of the "common pot", this time framed as a "shared opportunity", with the earnings being split equally between the two players;
6. reframing this "shared opportunity" as an investment or expenses that is commonly viewed in the local context as primarily a male or female responsibility;
7. combining for one of the players the opportunity to invest in a "shared opportunity" with that of investing in a "business opportunity" that the player controls.

These games provide an opportunity to assess the extent of gender biases in the allocation of resources under different playing rules.

The couples play variants of the VCM (voluntary contribution mechanism) game. In this game, both partners receive an endowment of 10 tokens (each worth 1 Ghana cedi), which they can either keep or invest in a business opportunity that is owned by one of the partners. The money in this opportunity gets multiplied by three and is subsequently given to the partner who owns the business opportunity. In most games, no re-allocation is possible. The earnings are kept private (except for in the treatment with open decision-making and communication, where the earnings can be inferred).

Participants play six or seven games. In these games, we introduce experimental variation in several dimensions. First, we vary who owns the business opportunity. In some games this is the man (denoted by an “M” suffix, for male), in other games this is the woman (denoted by an “F” suffix, for female) and in other games neither the man nor the woman owns a business opportunity (denoted by an “N” suffix, for neutral).

Second, we vary whether communication is allowed, whether in-game transfers are possible, the nature of the opportunity and whether participant plays with their spouse or a stranger. The following VCM games are played (see also Table 1 of the pre-analysis plan):

- The private game (PM and PF). This is the game described above. No communication is allowed and partners are seated away from each other.
- The open (joint) decision-making game (JM and JF). In this game, the couples are seated next to each other and are allowed to communicate. All choices are made openly.
- The re-allocation game (RM and RF). In this game, the owner of the business opportunity can make a transfer to their partner after receiving the tripled tokens.
- The multiple opportunities game (XM and XF). In this game, besides the business opportunity that leads to tripled tokens, there is a second business opportunity, belonging to the other partner, which only yields a doubling of tokens. Both partners can invest in both business opportunities.
- The stranger game (SM and SF). In this game, participants play with a stranger instead of their own partner.
- The common opportunity game (CN, CF and CM). In this game, there is a common opportunity. Both partners can invest their endowment in this common opportunity. The tokens invested in the common opportunity are multiplied by 1.5 and then split equally between the two partners. In game CN, there is no private opportunity. In game CF, the woman can also invest in a private opportunity. In game CM, the man has a private opportunity. In each of the last two games, the other partner cannot invest in the private opportunity.

For the common opportunity games (CN, CF and CM), we introduce one further degree of experimental variation, by varying the framing of the common opportunity. We introduce three frames:
- Neutral frame. In this frame, the common opportunity is phrased as a shared opportunity, without any example or explicit frame.
- Female frame. In this frame, the common opportunity is phrased as investing in kitchen utensils, which benefits the household as a whole. Kitchen utensils are traditionally considered as an expense incurred by women.
- Male frame. In this frame, the common opportunity is phrased as investing in housing, which benefits the household as a whole. Housing costs are traditionally considered as an expense incurred by men.

Furthermore, a subset of participants play the dictator game (DGF, DGM) in addition to the experiments described. In this game, only one of the parties receives an endowment and they can choose how to give of this to their partner.

More details on the experimental protocol can be found in the pre-analysis plan.

2017-07-19

2017-08-31

The individual and joint investment choices (i.e., the number of tokens) made in the experimental game.

See the pre-analysis plan for further detail.

The couples play variants of the VCM game. In this game, both partners receive an endowment of 10 tokens (each worth 1 Ghana cedi), which they can either keep or invest in a business opportunity that is owned by one of the partners. The money in this opportunity gets multiplied by three and is subsequently given to the partner who owns the business opportunity. In most games, no re-allocation is possible.

See the pre-analysis plan.

Randomization by a computer on the experimental session level.

Experimental sessions (containing 4-11 couples).

No

98 sessions.

590 participants. A full breakdown is found in Table 2 of the pre-analysis plan.

Depends on treatment, varying from 50 (for the stranger games) to more than 400 (for the private games).