Joint Accounts in the Fintech Era: How Does Labelling, Transparency and, Approval Rules Affect Spousal Financial Decision Making?

This study revisits the topic of joint accounts and assess how their terms, increasingly varied due to innovations in the fintech sector, affect intrahousehold allocation decisions. Through an experimental lab setting in Kolkata, India, we assess the impact of labeling, transparency and spousal approval on expenditure allocations of spouses. We also assess the impact of personally earning money on subsequent expenditure decisions of couples and specifically under different joint account terms.

This study uses random assignment to establish a causal relationship between different types of intervention and outcomes. There are two randomization levels: one, based on a work requirement of female partners, and second, based on various joint account terms such as privacy, labeling, and communication. The first level of randomization determines whether the female partner would work during the first half an hour. The second level of randomization determines differential levels of information, labeling, and communication among couples concerning their financial decisions.

Under the different types of treatment at the first and second randomization levels, couples make decisions on how they wish to allocate a certain amount of money into six different options such as: depositing in their own bank account, depositing in their partners’ bank account. depositing to a third persons’ bank account, a personal gift voucher to buy female clothing, footwear and other accessories, a personal gift voucher to buy male clothing, footwear and other accessories, and a shared gift voucher to buy household items. Participants are also asked how they plan to spend the amount towards different types of expenditure which ranges from savings to children’s expenditure to different types of daily expenditure goods such as food items and personal goods. We will also capture information on for whom participants would like to spend the amount on, for example, for self or partner or children or other family, non-family members.

2021-11-26

2021-12-31

We rely on participants’ decisions during the allocation game to measure the impact of labeling, transparency, spousal approval, and negotiation on couples’ expenditure decisions under different work requirements. We explore the following set of primary outcomes.

1. Share to Self.
2. Share to Spouse.
3. Share to Household Good.
4. Visibility.
5. Sharing with Spouse.
6. Share towards Expenditure - Food and necessities.
7. Share towards Expenditure - Personal goods and services.
8. Share towards Expenditure - Savings and investments.
9. Beneficiary of expenditure.

The primary outcomes and its definition are as below:

1. Share to Self - Share of money allocated towards own account and personal gift voucher.

2. Share to Spouse - Share of money allocated towards partner’s’ account and partner’s gift voucher.

3. Share to Household Good - Share of money allocated towards shared gift voucher.

4. Visibility (Binary (0/1)) - 1 if participant chooses deposit the amount in their partner’s bank account, a personal gift voucher to buy female clothing, footwear and other accessories, a personal gift voucher to buy male clothing, footwear and other accessories, or a shared gift voucher to buy household items, 0 otherwise.

5. Sharing with Spouse (Binary (0/1)) - 1 if participant chooses deposit the amount in their partner’s bank account, a personal gift voucher to buy female (male) clothing, footwear and other accessories, a shared gift voucher to buy household items, 0 otherwise.

6. Share towards Expenditure - Food and necessities (Binary (0/1) and percentage share)
Binary (0/1) - 1 if participant chooses items in ‘food and necessities’ for the follow up question on expenditure, 0 otherwise.
Percentage share - share of money participant spends on ‘food and necessities’.

7. Share towards Expenditure - Personal goods and services (Binary (0/1) and percentage share)
Binary (0/1) - 1 if participant chooses items in ‘personal goods and services’ for the follow up question on expenditure, 0 otherwise.
Percentage share - share of money participant spends on ‘personal goods and services’.

8. Share towards Expenditure - Savings and investments (Binary (0/1) and percentage share)
Binary (0/1) - 1 if participant chooses items in ‘savings and investments’ for the follow up question on expenditure, 0 otherwise.
Percentage share - share of money participant spends on ‘savings and investments’.

9. Beneficiary of expenditure -
Equal to "wife" if participant chooses items in the follow up question on expenditure for wife, 0 otherwise.
Equal to "husband" if participant chooses items in the follow up question on expenditure for husband, 0 otherwise.
Equal to "children" if participant chooses items in the follow up question on expenditure for children, 0 otherwise.
Equal to "other family members" if participant chooses items in the follow up question on expenditure for children, 0 otherwise.
Equal to "everyone in the household" if participant chooses items in the follow up question on expenditure for everyone in the household, 0 otherwise.

In addition, we also measure participant’s emotional well being after the allocation games using a set of variables

1. Emotional well being.
2. Mental Bandwidth.

The secondary outcomes and their definition are as below:

1. Emotional well being - Standardized index of variables measuring how often participant felt the following during the session (options range from ‘often’ to ‘never’)
Nervous, tense or uneasy,
Felt difficult to concentrate on what they were doing,
Sad,
Tired,
Could not cope with things asked to do, Felt confident about the future.

2. Mental Bandwidth - Measured by standardized index and individually of the below four tests.
(i) Psychomotor vigilance test (reactive function): Average and best reaction time taken to click on a target that appears on the screen.
(ii) Hearts and flower test (executive function): Number of times participant accurately answer tests based on congruent and incongruent blocks.
(iii) Memory test: Number of correct answers respondent gives to memory test.
(iv) Raven's test (abstract reasoning): Number of correct answers participant gives in a test to determine the missing element in a pattern.

Participants in this study are 1,000 couples who live in semi-urban Kolkata, India selected at random from the client pool maintained by several microfinance organizations and NGOs based in Kolkata. Participants are initially contacted via phone and asked if they are willing to participate in the study.

In order to be eligible for the study, couples have to meet the following criteria:
1. The household head is married and his/her spouse lives within the household.
2. Both the head and spouse are literate in local language.
3. Both the head and spouse have individual bank accounts.
4. Both the head and spouse are 60 or less years of age.

This study uses random assignment to establish a causal relationship between different types of intervention and outcomes. One thousand couples are selected based on the eligibility criteria determined by their age, literacy, relationship status and having an individual bank account each. There are two randomization levels: one, based on a work requirement of female partners, and second, based on various joint account terms such as privacy, labeling, and communication. The first level of randomization determines whether the female partner would work during the first half an hour. The second level of randomization determines differential levels of information, labeling, and communication among couples concerning their financial decisions. Two hundred couples will be in each treatment arm which has restrictions on information, labeling, and communication to complete an allocation game, with half coming from the ‘task’ stream and half from the ‘gift’ stream.

Once participants fulfill the eligibility criteria and consent to participate, couples are randomized at two levels. First, based on a work requirement of female partners, and second, based on various joint account terms such as privacy, labeling, and communication. We perform the randomization using STATA 17, a statistical software.

One thousand couples are selected based on the eligibility criteria determined by their age, literacy, relationship status and having an individual bank account each. There are two randomization levels: one, based on a work requirement of female partners, and second, based on various joint account terms such as privacy, labeling, and communication. The first level of randomization determines whether the female partner would work during the first half an hour. The second level of randomization determines differential levels of information, labeling, and communication among couples concerning their financial decisions. Two hundred couples will be in each treatment arm which has restrictions on information, labeling, and communication to complete an allocation game, with half coming from the ‘task’ stream and half from the ‘gift’ stream.

No

Not applicable

1,000 couples

1,000 couples will be randomized into a ‘task’ or ‘gift’ stream depending on the task allotted to them during the first hour. Hence, there will be 500 couples in 'task' stream and 500 in 'gift' stream. After the first level of working or waiting, all couples will be randomized further into five treatment groups with restrictions on information, labeling, and communication regarding their allocation decisions. Two hundred couples will be in each treatment arm at the second level of randomization, with half coming from the ‘task’ stream and half from the ‘gift’ stream.

A total of 1,000 couples, with 500 per treatment arm in the first level of randomization based on work requirements is sufficient to detect a minimum detectable effect size (MDES) of 0.085. For comparison between treatment arms with varying levels of restrictions for accounts, 200 couples in each of the five groups are sufficient to detect a MDES of 0.135. To find the effect of account types on allocation decisions, a sample size of 100 couples in each of the 10 groups can detect MDES of 0.187 for couples where female partner works for an hour and 0.192 for couples where female partner does not work during the first hour of the experiment. All power calculations for this study are performed at an 80% power and a 5% significance level.

Pre Analysis Plan Jose, King, Bedi