Context

Financial technology (fintech), including mobile money services, peer-to-peer lending, and mobile insurance is promised as a means to financial or economic inclusion for the un(der)banked. However these online financial services require Internet access, adoption, and digital literacy. Given the important interdependencies between banking and Internet adoption in the United States, this paper merges datasets from the FDIC’s June 2015 Unbanked and Underbanked supplement and NTIA’s July 2015 Computer and Internet Use supplement to study the issue more closely. While the process of merging Current Population Survey (CPS) supplements at the household level is fairly straightforward, treating sampling weights and variance raises some complications. The paper proposes a methodology for merging CPS supplements and addressing these issues.

The study

Given the increasing interdependencies between online banking and Internet adoption in the United States, this paper merges datasets from two supplements to the US Census Bureau’s (Census’s) monthly Current Population Survey (CPS) on banking and Internet use to enable more sophisticated study of the issue. Combining these consecutive supplements permits longitudinal examination of households participating in the CPS. The paper offers a methodology for merging CPS supplement data and addressing weighting issues. Substantively, the results suggest that there is a relationship between households’ Internet use and their banking status. The study is based on data from c.35,000 households who completed both original surveys.

Key findings

• With the growing importance of fintech and increasing reliance on the Internet for banking, the issues of financial and digital inclusion appear to be converging. M-money services, peer-to-peer lending, and mobile insurance promise to provide un(der)banked consumers with greater access to the benefits of traditional banking institutions and lower cost alternative financial services. To be successful, however, these online financial services may require Internet access, adoption, and digital literacy. To ensure fintech services can reach these populations, policymakers need to understand the relationship between the un(der)banked and digital adoption.
• The merged dataset shows a strong relationship between the un(der)banked and the level of Internet activity, as measured both through the type of Internet service(s) used and the types of devices used by the household.
• For example, nearly half (47%) of unbanked households do not have a fixed or mobile Internet connection. Likewise, while less than 40% of the unbanked households rely on multiple types of devices, around 75% of the underbanked and banked households use multiple types of devices. Results suggest that to successfully expand financial inclusion in the US fintech could require Internet access, adoption, and digital literacy training.
• Analysis of the merged dataset shows that lack of connectivity is an issue that can have implications for banking status. The banking industry has suggested that mobile banking could motivate unbanked households to open bank accounts. The industry has also looked at mobile banking as a way to discourage the usage of alternative financial services. However, results suggest that lack of Internet adoption and digital literacy may be barriers to expanding access to the un(der)banked. The rate of online or text banking drops dramatically with the number of device types a household uses; there is a qualitative difference in the level of activity for those with fewer devices. Nearly half (46%) of unbanked households did not have internet access either through a fixed or mobile connection. Just over a quarter (26%) of unbanked households had internet access only through a mobile data plan. Access to internet services by underbanked households is similar to that of fully banked households.
• The authors suggest a simple model of banking status dependent on Internet activity and a host of other demographic attributes. Use of Internet access services remained a significant indicator of having a bank account after accounting for demographic differences.

Points to consider:

• This study is based on two unrelated data sets from rom the FDIC’s June 2015 Unbanked and Underbanked supplement and NTIA’s July 2015 Computer and Internet Use supplement. It creates a model to merge this data and accounts for weighting, then reweights the merged data. The validity of findings are therefore dependent on the strength of the original data and data merging methodology.
• The data is based on surveys from c.35,000 households in the US and is weighted to be nationally representative.

• The study is relevant and its findings are valuable for insight into the area of digital inclusion among the un(der)banked.