Working Paper: NBER ID: w28956
Authors: Paolina C. Medina; Michaela Pagel
Abstract: We analyze an experiment involving 3.1 million bank customers who were encouraged to save through SMS messages. We first theoretically show that by examining their spending, saving, and borrowing responses we can distinguish between the leading explanations for coholding liquid savings and credit card debt. Using a machine learning algorithm, we then predict individual-level treatment effects and find that the most responsive individuals reduce spending and increase their savings by 5.1% (225 USD PPP per month), while their credit card debt remains unchanged. We argue that these joint findings suggest people co-hold because they mentally separate savings and debt accounts
Keywords: No keywords provided
JEL Codes: D14; G5
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Decrease in spending (D12) | Mental accounting theory (G41) |
| Increase in savings (E21) | Mental accounting theory (G41) |
| Saving nudges (Y60) | Decrease in spending (D12) |
| Saving nudges (Y60) | Increase in savings (E21) |
| Saving nudges (Y60) | No change in credit card debt (G51) |
| Saving nudges (Y60) | Co-holding of low-interest savings and high-interest debt (G51) |
| Increase in savings (E21) | Negligible change in borrowing costs (G19) |