How do Households Learn to Use a New Financial Product? Savings Behavior in the Plan, Crowd-Out, and the Role of Social Interactions

Project Start:04/2016
Researchers:Satyajit Dutt, David Love, Henriette Prast, Nathanael Vellekoop
Category: Household Finance
Funded by:LOEWE

In 2006, a new tax-favored employee savings plan was introduced in the Netherlands. Savings in the plan could only be used to self-finance periods of leisure during the working life. Examples are early retirement, parental leave, and sabbaticals. We plan to study the take-up and monthly savings behavior, the crowd-out of other household savings, and the role of social interactions in this new savings product. We link administrative data with monthly wages of all Dutch employees to household wealth data for the years 2006-2013. In this unique panel we have the universe of savers, and the complete history of monthly contributions into the plan. We focus on three, related questions.

(1) What is the savings behavior in a new financial product? For example: who are the early adopters, and which firms do they work for? How much entry and exit is there from year to year? How many people are there that try the plan for a year? How do savers choose the monthly savings level and the annual profile? Since we have information on the entire population, we will also look at whether the target population was reached that policy makers had in mind.

(2) A longstanding question in public finance is how households change other savings in response to the presence of subsidized savings. The theoretical prediction is that households should shift savings one-for-one (full crowd-out). We will estimate the extent of crowd-out, as well as changes in the composition of asset holdings of households (stocks, savings accounts). In order to identify crowd-out we use a regression-discontinuity design, where we exploit the differences in maximum contribution limits of certain cohorts. Since this is a new savings product, we will also document dynamics after opening an account. This will be informative about whether people “learn” to crowd-out.

(3) The third question is what the role of the environment is in take-up, level of savings, and crowd-out. We investigate two possible channels of social interactions: coworkers and family (siblings). For the coworkers we use three instruments: firm mergers, the place of the worker in the within-firm wage distribution (Duflo and Saez, 2002), and the co-workers of the spouse. With our unique data we can contrast which of the two channels is more effective in learning how to use a new financial product.

This project mainly fits in the household finance area, but also will provide useful insights for firms, policy makers and financial institutions.