Behavior Revealed in Mobile Phone Usage Predicts Loan Repayment

Many households in developing countries lack formal financial histories, making it difficult for banks to extend loans, and for potential borrowers to receive them. However, many of these households have mobile phones, which generate rich data about behavior. This paper shows that behavioral signatures in mobile phone data predict loan default, using call records matched to loan outcomes. In a middle income South American country, individuals in the highest quintile of risk by our measure are 2.8 times more likely to default than those in the lowest quintile. On our sample of individuals with (thin) financial histories, our method outperforms models using credit bureau information, both within time and when tested on a different time period. The method forms the basis for new forms of lending that reach the unbanked.

[1]  Robert Kirkpatrick,et al.  Big Data for Development , 2013, Big Data.

[2]  Nuria Oliver,et al.  MobiScore: Towards Universal Credit Scoring from Mobile Phone Data , 2015, UMAP.

[3]  Xin Lu,et al.  Approaching the Limit of Predictability in Human Mobility , 2013, Scientific Reports.

[4]  Dean Karlan,et al.  Six Randomized Evaluations of Microcredit: Introduction and Further Steps † , 2015 .

[5]  A. Banerjee,et al.  Do Firms Want to Borrow More? Testing Credit Constraints Using a Directed Lending Program , 2004 .

[6]  Darrell Grissen,et al.  Behavior Revealed in Mobile Phone Usage Predicts Loan Repayment , 2017 .

[7]  Craig McIntosh,et al.  The supply- and demand-side impacts of credit market information , 2010 .

[8]  Allen N. Berger,et al.  A more complete conceptual framework for SME finance , 2006 .

[9]  A. Barabasi,et al.  Quantifying social group evolution , 2007, Nature.

[10]  Silvia Angela Osmetti,et al.  Modelling small and medium enterprise loan defaults as rare events: the generalized extreme value regression model , 2013 .

[11]  Lucy H. Jones,et al.  Scoring for Impact Evaluation Microcredit in Theory and Practice : Using Randomized Credit , 2011 .

[12]  Víctor Soto,et al.  Prediction of socioeconomic levels using cell phone records , 2011, UMAP'11.

[13]  D. Lazer,et al.  The Parable of Google Flu: Traps in Big Data Analysis , 2014, Science.

[14]  Bart Baesens,et al.  Credit scoring for microfinance: is it worth it? , 2012 .

[15]  William Jack,et al.  Documenting the birth of a financial economy , 2012, Proceedings of the National Academy of Sciences.

[16]  Albert-László Barabási,et al.  Understanding individual human mobility patterns , 2008, Nature.

[17]  A-L Barabási,et al.  Structure and tie strengths in mobile communication networks , 2006, Proceedings of the National Academy of Sciences.

[18]  David J. Hand,et al.  Measuring classifier performance: a coherent alternative to the area under the ROC curve , 2009, Machine Learning.

[19]  Christopher Woodruff,et al.  Experimental evidence on returns to capital and access to finance in Mexico , 2008 .

[20]  D. McKenzie,et al.  Returns to Capital in Microenterprises: Evidence from a Field Experiment , 2007, SSRN Electronic Journal.

[21]  Jill Luoto,et al.  Credit Information Systems in Less Developed Countries: A Test with Microfinance in Guatemala , 2007, Economic Development and Cultural Change.

[22]  Johan A. K. Suykens,et al.  Benchmarking state-of-the-art classification algorithms for credit scoring , 2003, J. Oper. Res. Soc..

[23]  Mark Schreiner,et al.  Scoring Arrears at a Microlender in Bolivia , 2004 .

[24]  Gabriel Cadamuro,et al.  Predicting poverty and wealth from mobile phone metadata , 2015, Science.