A micro dataset on delinquency risk, loan motivations and credit access in Chile

This article provides data that allows to estimate how delinquency risk and consumer loan motivations changed in Chile before and after the new Interest Rate Ceiling Law of 2013. The data is of particular interest for those interested in the heterogeneity of household borrower profiles and their reactions to new loan regulation. The codes are in Stata format and the datasets are in Excel.


Data
Two data files are included: CS_debts_EFHall2.xlsx and EFH_TMC_CS.xlsx. These datasets can be easily imported to Stata for analysis. These are the raw micro datasets.
The data file CS_debts_EFHall2.xlsx contains 21,319 cross-sectional household observations from the Chilean Household Finance Survey (in Spanish, Encuesta Financiera de Hogares, hence on EFH) for the survey waves 2007, 2008, 2009, 2010, 2011, 2014 and 2017. It provides information on the households' demographic characteristics (years of education, age, number of household members), income (monthly income, income strata at the national level, with strata given by the percentiles 1e50, 51e80, 81e100), and consumer loan amount (in a real monetary index labelled as UF, which is widely used in Chile e one UF corresponds roughly to 40 USD during the period of analysis). It also gives information on whether the household has a loan of 0e50 UF or 50e200 UF with banks or with non-E-mail address: cmadeira@bcentral.cl. The data file EFH_TMC_CS.xlsx contains data for all the households in the EFH survey between 2007 and 2014 and it includes information on all the individual family members. It has observations on 16,770 households and 56,986 household members, with information on the role of each household member (household head, spouse/partner, child, sibling, parent or in-law, other relative, or domestic worker). It has information on the household members' age, gender, civil status, monthly income, number of household members, monthly debt service at the household level, delinquency status (1 month or more in arrears, 3 months or more in arrears) for mortgage loans, overall consumer loans, banking consumer loans, banking and retail credit cards. Finally, it contains risk-adjusted interest rate variables for consumer loans between 0-50 UF and 50e200 UF, plus dummy variables with a value of Specifications Table   Subject Economics, Econometrics and Finance. Specific subject area Household Finance. Impact evaluation of programs and legislation. Type of data Value of the Data The data can be used to study how delinquency risk changed before and after the 2013 Chilean Maximum Interest Rate law (see C ordova and Toro [2], Cuesta and Sepúlveda [3], Hurtado [4], and references therein). This is an important regulation topic for several countries (Maimbo and Henriquez [5]), especially as household credit keeps growing. The data can be used by researchers interested in delinquency risk of consumer loans and how these relate to demographics (age, civil status, college education, number of household members), household income (currently monthly income measure at the time of the survey, annual permanent household income), indebtedness (the ratio of consumer debt to annual permanent household income, the ratio of total monthly debt service to household income), and unemployment risk. The data includes dummy variables for mortgage and consumer loan delinquency contracts after 30 days or more and after 90 days or more. It also contains measures of payment difficulties for retail and banking credit cards. Researchers can use the insights of this data to study how interest rate ceilings are likely to affect developing countries, accounting for heterogeneity of families and how the impact may depend on characteristics such as age and income. Since most underdeveloped countries have scarce microeconomic data on this topic (Maimbo and Henriquez [5]), the dataset is likely to be influential with policy makers, international institutions, besides academic researchers. Finally, the data is useful for researchers interested in studying how borrowers' behaviour and motivations differs across different lenders. In particular, the data includes information on the debtor status of households: Both Banks and Retail Stores, Banks only, Retail Stores only, Labour and Credit Unions, Other Loans, whether households have no wish for consumer debt (No Wish for Debt), or whether households applied for loans but were refused (No Access to Debt). The data also includes measures of the share of consumer loans motivated by "Current consumption purchases", "Durable goods and investments", "Pay previous debts", and "Health".

Data in brief
one when the risk-adjusted interest rate is above the Legal Interest Rate Ceiling in Chile for 0e50 UF and 50e200 UF, respectively. The article also provide two Stata codes with an example of how to use these datasets: Debt_-motivations_TMC.do and EFH_2011regs2.do. The code Debt_motivations_TMC.do uses the data CS_debts_EFHall2.xlsx to replicate the analysis in section 2.2 of Madeira [1]. The code EFH_2011-regs2.do uses the data EFH_TMC_CS.xlsx to replicate the analysis shown in sections 2.2 and 3.2 of Madeira [1]. These 2 codes create the analysed final tables in sections 2.2 and 3.2 of Madeira [1]. The Stata codes use the raw datasets to replicate the 12 analysed data Excel files in the folder "analy-sed_data" described succinctly in Table 1 below.

Experimental design, materials, and methods
The data consists of a formatting of the EFH dataset, which creates delinquency variables at the household level for mortgage and consumer loans, loan motivations, and non-payment of banking or retail credit card debt. Household measures of permanent income and unemployment risk were included based on workers of similar characteristics from the Chilean Employment and Income Survey The sequence for the final data creation can be described as follows:  1) The EFH dataset (waves 2007 to 2014) are used, which includes a total of 16,770 urban households interviewed at the cross-sectional level. Socioeconomic information on age, years of education, three income strata (national household income percentiles 1 to 50, 51 to 80, and 81 to 100), number of household members, total monthly household income, a dummy variable for the Santiago Metropolitan Region and the type of county (low income level, high income level, and upper income level) is obtained directly from the survey. Variables of delinquency in arrears and credit card non-payment are included. 2) The variable "deudacon" classifies all the EFH households into 7 mutually exclusive categories of borrowers according to their largest consumer loan amount held: 1) borrowers in Banks (but not in Retail Stores), 2) households with consumer loans both in Banks and Retail Stores, 3) households with loans in Retail Stores (but not in Banks), 4) households with consumer loans in Labour and Credit Unions, 5) Other Loans (car sellers, pawnshops, informal loans), 6) households with No Desire for Debt, and 7) households with No Access to Debt (because their loan applications were rejected or expected to be rejected).
3) The share of consumer debt represented by each of 4 types of loan motivations ("Current consumption purchases", "Durable goods and investments", "Pay previous debts or loan consolidation", and "Health") is created by grouping more detailed loan categories. In particular, the category "Current consumption purchases" is an aggregation of the motivations: "Purchase of articles for the home and living expenses", "To purchase clothes", "Other". The category "Durable goods and investments" also aggregates more detailed classifications of motives: "To buy a vehicle or other means of transport, maintenance and repair expenses of vehicles", "Vacations", "To finance a business or professional activity", "For investment in financial assets", "To refurbish or renovate the residence", "For education purposes", "To purchase real estate assets", "To provide funds or make a loan to another person or relative". 4) Using data from workers of similar characteristics from the Chilean Employment Survey (ENE/ESI), conditional on their education, age, industry, income quintile and region, a measure of permanent labour income was built for each household member, P k(i),t , accounting for their earnings while employed, their probability of unemployment spells and their replacement ratio of earnings while in unemployment. The households' permanent income is obtained as the sum of their non-labour income plus the permanent income of their members: P i,t ¼ a i þ P k P k(i),t . The household unemployment risk is estimated as a permanent-income weighted average of the unemployment risk of its labour force members: u i,t ¼ P k P k(i),t /(P i,t -a i )u k(i),t .