Data and coding behind Figure 1(b)

Panel (b) of Figure 1 is constructed from the 2017 Macroprudential Policies Evaluation Database (MaPPED) from Budnik and Kleibl (2018). MaPPED documents the life cycle of macroprudential and closely related policy instruments in all EU member states and the United Kingdom, recording for each policy action its instrument category and subcategory, the market segment targeted, the type of action (activation of a new tool, change in level, change in scope, deactivation), whether it tightened or loosened the stance, its main character (macroprudential, microprudential, or other), and a multi-select field, “Objective of the measure,” in which the authority’s stated motivation is recorded. The objectives field is populated from national regulatory documents by the country experts who compiled the database.

We keep policy actions whose main character includes “Macroprudential” and whose category is “Lending standards restrictions,” which is MaPPED’s label for borrower-based instruments, and then restrict to actions whose market segment includes residential real estate. This yields 113 actions in 18 countries. The sample comprises 49 loan-to-value limits, 28 debt-service-to-income limits, 12 maturity and amortization restrictions, 5 other income requirements for loan eligibility, 2 debt-to-income limits, 1 loan-to-income limit, and 16 other restrictions on lending standards. Of these, 89 are tightenings and 58 are activations of a new tool. Ninety-seven are classified as purely macroprudential, with the remainder carrying a joint macroprudential and microprudential (or other) character. Announcement years, where recorded, range from 1982 to 2017, with 71 of the 91 dated actions falling in 2010 or later. The unit of observation is the policy action rather than the country-instrument, so countries that adjusted a given limit repeatedly contribute several observations.

MaPPED’s objectives field is multi-select and unranked. A measure may list several objectives, but the database does not identify the primary one. We therefore report the share of measures citing each objective. The mapping in Figure 1(b) into four broad categories plus a residual is as follows. Credit and household leverage collects “General credit growth,” “Sectoral credit growth,” and “Household leverage.” House prices corresponds to “Developments in asset prices.” Financial stability / resilience collects “Other addressing the resilience of the financial system,” “Currency risks,” “Interconnectedness / contagion risks,” “Market liquidity,” and “Exposure concentration.” Bank sector collects “Bank leverage,” “Maturity mismatches on banks’ balance sheets,” and “Bank-level oversight (microprudential).” Other collects “Level playing field / competition,” “Harmonisation,” and “Other.” We count a measure in a category if it cites any objective belonging to it, so categories overlap and shares do not sum to one.

We place “Currency risks” under financial stability rather than credit, on the grounds that the foreign-currency mortgage measures in Central and Eastern Europe that account for most of these cases were motivated by household exposure to exchange-rate shocks rather than by the volume of borrowing as such. Moreover, we place “Bank-level oversight (microprudential)” under the bank sector. Neither affects the ranking in Figure 1(b).