Replication package of the study "What drives profits in savings groups?
Bayesian data mining evidence from the SAVIX database"
(Forthcoming: Review of Development Finance)
Data mining is the process of exploring a large dataset with the aim of discovering unknown patterns and identifying associations through statistical and mathematical techniques. The Cross Industry Standard Process for Data Mining (CRISP-DM) is a process to extract knowledge of a database using data mining. The CRISP-DM process was adapted to analyze the SAVIX database. The Savings Groups Information Exchange (SAVIX) database has information of more than 200 000 savings groups worlwide. Bayesian data mining of the SAVIX database was performed with Bayesian Model Averaging (BMA).
- R_savix_BDM_nov2021: R script to replicate the results of the study (BMA and Bayes Factors). Update 2021: includes an additional analysis for ROA, besides ROS.
- SAVIXpanel_for_R: database (in Stata .dta format). Note: this is not the original SAVIX database, it is a cleaned version of the SQL SAVIX database prepared to calculate between-effects (as group-level fixed effects and random effects are not feasible nor appropiate for the SAVIX).