Experimental Design
We select all mortgage customers of the partnering bank with a letter opt-in as possible letter recipients unless they are subject to at least one of the following three restrictions: (i) mortgage activity (such as loan origination) over the past two years; (ii) expected mortgage activity still in 2024; and (iii) expected residual loan amount at reset of less than €10,000. Customers can have multiple loans, so if one loan is not subject to these restrictions, we keep the customer as a possible letter recipient. 75% of possible letter recipients then actually receive a letter. The remaining 25% serve as a control group. The bank invites via email all letter recipients as well as the control group to participate in the post-treatment survey if they have an email opt-in too. (The email invitation extends to all mortgagors who can be reached via email, not only those who are part of the letter experiment.) We additionally study the choices of treatment and control groups in the data provided by the partnering bank.
The letter comprises four sections. In the first section, we explain the collaboration between the bank and Goethe University Frankfurt, and that this letter provides information relevant to borrowers. We also show graphically that mortgage rates in Germany have increased substantially since early 2022. Second, we state that future rates are uncertain, citing a study that documents large disagreement among German households about future loan interest rates. Third, a simple example illustrates how higher interest rates can increase monthly payments once the rate-fixation period ends and mortgagors might have to refinance their mortgage at current market rates: a two-percentage-point rate increase raises monthly interest payments of a €100,000 mortgage by €2,000 initially. Fourth, we list options to cope with higher mortgage rates. Specifically, we mention mortgage prepayments, locking in existing rates through so-called forward loans or home-savings contracts, and increasing savings to finance higher future payments. We mention various factors determine which option is best for each mortgagor, and that the bank is happy to provide advice.
We construct three variants of the letter. The objective is to better understand which letter sections, if any, induce effects. The first letter variant includes all four sections. The second omits information on uncertain future rates (section 2). The third variant omits information on how higher rates might translate into higher payments (section 3).
We investigate effects of the letter in both survey and bank data. We run a survey around three weeks following letter provision. The survey has four objectives: measure (i) attention paid to the letter; (ii) effects on beliefs, in particular about past, current, and future rates, and about the payment impact of a rate change; (iii) past and planned mortgage choices; and (iv) mechanisms underlying treatment effects (e.g., lack of trust as a reason behind not seeking bank advice). In the bank data, we can test for observed choices induced by the letter, such as meetings with bank advisors, mortgage repayments, and, for a subset of borrowers, consumption and savings/investment decisions.
The survey also comprises a hypothetical vignette in which respondents make choices based on a €100,000 mortgage that needs refinancing in the future. We primarily ask respondents about whether they would interact with a bank advisor and would consider mortgage prepayments, as well as the reasoning behind their choices. The vignette section also includes questions on information acquisition, the rate-fixation period, propensity to lock in future rates, and saving and mortgage-search behavior. We vary the mortgage-rate change at and distance to reset as part of the vignette, allowing us to study the sensitivity of choices to changes in the size of the rate increase and distance to fixation-period end.