ACCESS TO AND USE OF FINANCIAL SERVICES: EVIDENCE FROM PERU

Transcription

1 SBS Documentos de Trabajo 2015 Superintendencia de Banca, Seguros y Administradoras Privadas de Fondos de Pensiones. Este documento expresa el punto de vista del autor y no necesariamente la opinión de la Superintendencia Banca, Seguros y Administradoras Privadas de Fondos de Pensiones. DT/03/2015 SUPERINTENDENCIA DE BANCA, SEGUROS Y ADMINISTRADORAS PRIVADAS DE FONDOS DE PENSIONES (SBS) ACCESS TO AND USE OF FINANCIAL SERVICES: EVIDENCE FROM PERU Jacqueline Talledo 1 SBS Microfinance Analysis Department Aprobado por Manuel Luy December, 2015 Abstract This study examines the determinants of financial access and use, in particular, the effect of the proximity of formal service points, approximated by reported travel time, using cross-sectional data from a demand-side pilot survey conducted in Peru in A 30 minute reduction in travel time is found to predict an increase in the probability of formal loan application by 2.9 percentage points. Probabilities of active account use and loan usage for business benefit from proximity as well. There is also evidence for complementarity between formal and informal financial services. JEL classification codes: G28, O16, R51 Keywords: financial access, financial use. jtalledo@sbs.gob.pe * I am grateful to Sanchari Roy, Barry Reilly and Narda Sotomayor for their insights and comments as well as their valuable advice. 1 Jacqueline Talledo is Senior Analyst at the Microfinance Analysis Department of the SBS (Lima, Peru).

2 I. INTRODUCTION Lack of access to financial services remains as a major challenge, especially for the poor in developing countries. The Global Findex Database 2014 reports that 62 percent of adults in developing countries have an account at a financial institution, and in the poorest 40 percent of households this number drops to 54 percent for adults. Recent empirical studies have found that financial development and access to financial services are important channels to improve small and medium enterprise s investments as well as household consumption and risk coping. In this context, lack of access to financial services in developing countries may significantly restrain further economic development. Hence, there is a need to understand the factors underlying the lack of access to financial services in order to guide policy design and implementation. The literature recognizes factors affecting access to financial services from the supply side (such as availability and transaction costs) and the demand side (for instance, financial literacy and perceptions); demand side aspects have recently been at the core of studies since the existence of an underlying demand for financial services (and thus, only focusing on supply aspects) is no longer taken for granted, especially in rural areas where formal institutions may find it less profitable to operate, and informal sources come into play. Financial access is understood as the availability of quality and affordable formal financial service points that enable an individual the use of financial services, while financial use refers to the actual use of financial services, and is related to quality, regularity and duration of use over time. Availability of financial service points is a necessary condition for access from the supply side, since access would not be possible without close and affordable service points. In addition, convenience in accessing financial service points is a sufficient condition, implying ability and will to efficiently access and use financial services, both characteristics from the demand side. Both availability and convenience can be approximated by the proximity of financial service points to the population, which would determine first if it facilitates access to financial services and henceforth their use. In this context, the objective of this study is to examine the determinants of accessing and using financial services, employing proximity as a proxy of availability and convenience of financial service points, in addition to individual and household characteristics and perceptions towards financial products. Most studies have focused on the determinants of financial access employing financial use as a proxy; however, this approach does not capture the transition from access to use adequately. The contribution of this study to the financial access literature is the evaluation of financial access and use separately as well as the use of travel time to formal financial service points reported by individuals to measure proximity of service points (branches, ATMs and agents). Reported travel time is richer than other variables employed to measure proximity of financial service points, in the sense that it provides both information regarding the awareness of the location of financial institutions and pecuniary and non-pecuniary costs of accessing those services; in addition, it supersedes information of number of service points from the supply side which may not exist in a geographical location but still can be reached by the individual by traveling to a different 2

3 place. On the other hand, the relationship between proximity of financial service points and financial access and use is assessed for both formal and informal services allowing examination of their complementary role. Financial access information from the supply side is available from financial regulators and is based on data collected from supervised financial institutions. In contrast, demand side information is more limited, being the most known sources the Global Findex database at country level, and the FinScope survey for some African countries at the individual or enterprise level. In Latin America, Mexico collects information regarding credit from family surveys as other countries in the region although in a more limited way. In the case of Peru, while supply side information (such as indicators of financial outreach) is widely available, demand side data has been scarce. However, recent efforts by the financial regulator have led to the undertaking of a pilot survey to collect information aimed at measuring the level of financial access and use, as well as understanding the barriers associated with the lack of them from the demand side. This pilot survey was conducted in four regions in Peru in 2012, which comprises information at the individual level on access and use of financial services. The empirical strategy considers linear probability models to evaluate the determinants of the probability of applying for a loan (proxy for financial access), and the probabilities of having savings, using accounts for different transactions, and using loans for business purposes (as proxies for financial use). In the case of accounts, four outcome variables are modeled: the probability of making at least two withdrawals in one month, of having at least two deposits in the same time span, of using the account for payments and money transfers, and of using it to receive money. Formal and informal sectors are considered in the case of applying for a loan, having savings and using a loan for business purposes, whereas only the formal sector is analyzed for accounts, where the formal sector is comprised of supervised institutions by the national financial regulator. The estimation is carried out by ordinary least squares as well as instrumental variable procedures as there is a potential for simultaneity bias since travel time to financial service points and the probability of access and use may be jointly determined. Lower travel time to a service point, that is, higher proximity of formal branches, ATMs or agents is expected to inversely determine the probability of access and use of financial services; nonetheless, higher access and use may also attract financial institutions, especially in the case of credit, and may lead to the establishment of more service point in a geographical location, resulting in endogeneity of the travel time regressor and thus biased estimated parameters. There is also a possibility of measurement error in the travel time explanatory variable. Results are robust to alternative forms of the travel time regressors, specifically conversion into a distance measure, travel time by only public transport, and trimming the highest two percent of the variable distribution. The main findings are as follows. Firstly, proximity of formal financial service points, in terms of reduction in reported average travel time using public transport, is associated with an increase in the probability of formal access and a decrease in the probability of informal access: on average and ceteris paribus a 30 minute reduction in travel time predicts a 2.9 percentage point higher probability of applying for a formal loan, while a 0.6 of a percentage point lower probability of asking for an informal loan. Higher educational attainment, marriage, number of dependents and a higher level of welfare are positively associated with the probability of applying for a loan, especially in a formal 3

4 institution. In addition, those perceiving formal credit costs and process length as the most important factors in loan application are more likely to ask for a formal loan; although these last results should be taken with caution, they might still be indicative of barriers to access to credit from the demand side, especially for those lacking financial education (costs) or needing a more flexible funding source (process length). Secondly, probability of having formal savings is found to be positively correlated with travel time to service points; this may be explained by the fact that a service point located far away reduces convenience and might prevent individuals from taking money out of the account, since the travel costs may be too high; there is a positive correlation with the probability of informal savings as well, however the effect is not significant at conventional levels. On the other hand, lower travel time is correlated with an active account usage: a decrease in 30 minutes in travel time to a formal service point is associated with an increase of 2.8 percentage points in the probability of making at least two withdrawals in a month, of 3.9 percentage points in the probability of having two deposits or more during the same time span, as well as of 1.7 percentage points in the probability of using the account to pay and transfer money. Thirdly, on average and ceteris paribus, a decrease in 30 minutes in travel time to a financial service points predicts an increase in the probability of using a loan for business purposes in 8.6 percentage points, while reducing the probability of getting an informal loan for business in 0.2 of a percentage point. On the other hand, those that consider credit costs and process length as important factors when applying for a credit in the formal system are less likely to use the loan for business purposes compared to those who do not consider these factors important. This is a reflection of the barriers to access from the demand side, specifically of perceptions regarding the formal financial system. The document is organized as follows. In the second section the economic literature on financial access and use is reviewed, with a focus in the distinction between the concepts of access and use of financial services, and the identification of their determinants and barriers from the supply and demand sides. Subsequently, the third section presents some indicators on Peruvian financial outreach based on supply side data, showing that while there has been significant improvement in the last years in financial access, understood as availability of services, and usage of financial services, the levels vary across the country. The fourth section presents information from the Pilot Survey of Access and Use of Financial Services undertaken by the Peruvian financial regulator to examine the level of access and use from the demand side; it also includes a description of the variables employed in the current study. Section V shows the empirical strategy followed to address the research questions, remarking the advantages and disadvantages of the linear probability model and outlining the estimation by instrumental variables process. Section VI presents the estimation results for the models previously described, while Section VIII shows the robustness tests. The last section concludes with some points for policy discussion and future research agenda. 4

5 II. LITERATURE REVIEW The positive impact of financial development on growth and poverty alleviation has been part of the economic literature for over twenty years (Honohan, 2004a; Guillaumont & Kpodar, 2008), especially in developing countries (Jalilian & Kirkpatrick, 2002); and lack of finance is associated with inequality and slower growth (World Bank, 2008). Indeed, as financial development within a country is not evenly distributed among the population (Claessens, 2006), during the last decade, the focus has turned towards the study of access to finance. By raising consumption, promoting investment opportunities, improving risk management and absorbing financial shocks (especially among the most vulnerable), financial access has been found to be an important aspect of economic development (Claessens, 2006; Beck et al, 2009; Ellis et al, 2010). Furthermore, lifting financial constraints has been shown to positively impact economic behavior (consumption and investment decisions) and poverty alleviation. These effects can be mediated through households or enterprises (Beck et al, 2012). While the impact of financial access on savings has been found to have a positive effect on variables such as poverty alleviation or growth (Burgess & Pande, 2004; Ashraff, 2006; Honohan & King, 2009; Dupas & Robinson, 2013a,b), access to credit still has mixed results. Microfinance has been also evaluated, although results have not been conclusive (Honohan, 2004b; Banerjee et al, 2015); on the other hand, borrowing constraints have been found to deter human capital accumulation (Jacoby, 1994) as well as investment (Love & Sanchez, 2009). Given the importance of financial access, understanding its determinants and barriers become relevant in order to establish policies in the right direction, especially in developing countries. However, the definition of access and use should first be assessed. II.1 Access is different from usage Access is a wider concept than use. One way to understand access has been as the absence of price or non-price obstacles for use (World Bank 2008). In that sense, the concept is related to the supply side as the availability of quality financial service points at reasonable costs (Claessens, 2006), a view shared by the financial regulators including the Peruvian one (AFI, 2014). Thus, access is related to physical proximity and affordability concepts. The problem with this view is that it implies the existence of an underlying demand for financial services, and that once service points are available, individuals will use these services, which is not always the case (Kostov et al 2012). A more recent definition for financial access is related to the ability to use available financial services and products from formal financial institutions (AFI; 2010), or the ease to use financial services whether or not the individual chooses to do it (Ellis et al, 2010). This responds to the idea that having financial service points nearby does not automatically imply their use, a decision that may result from voluntary or involuntary exclusion from financial services. Individuals may self-exclude from using financial services (even if they are available to them) 5

6 because of lack of need, religious or cultural reasons, or access to other sources, while the involuntarily excluded may be so as a consequence of risk, discriminatory policies, contractual deficiencies, and price and product features (World Bank, 2006; Claessens, 2006; Beck et al, 2009). In addition, there are several dimensions to consider when measuring access, such as the institutional, functional, and product level (Kumar, 2005); or the related to availability, cost, range and quality, the latter based on reliability, convenience, continuity and flexibility (Claessens, 2006). These different dimensions make it even more difficult to define and measure access to financial services. Financial access has been also approximated with indicators of use of financial services (Ellis et al 2010), a favored approach given the wider availability of statistics from the supply side, and due to the difficulties of separating the voluntarily and involuntarily excluded (Kumar, 2005). However, these measures leave out the financially excluded, which is the target population when it comes to financial inclusion policies. Another disadvantage of this approach is that it leaves out information regarding informal financial services, which complement the role of the formal financial system (Guirkinger, 2008; Ellis et al, 2010). 2 On the other hand, usage refers to the actual consumption of financial services. In a standard supplydemand framework, usage would be given by the intersection of the supply and demand schedules (Claessens, 2006). Use of financial products is related not only to the quality, but also to the regularity and duration of use over time (AFI, 2014). Usage is also subject to voluntary and involuntary exclusion, with the latter perhaps more influenced by supply side factors related to information asymmetries (adverse selection and moral hazard), contractual issues, transaction costs, institutional factors, or even country characteristics (Claessens, 2006; Beck & De La Torre 2007; Ghosh, 2000). II.2 Determinants of financial access and use The determinants of access and use of financial services can be seen in the framework of supply and demand for financial services. The main factors in that context are financial services prices (in terms of interest rates and fees) and consumers income (for instance, not having enough money to save or steady income to apply for a loan). However, other non-price factors from the supply side such as distribution points, risk management (including eligibility requirements) and institutional environment (such as financial regulation or legal facilities for contracts) are also important in determining financial access and use. Regarding the demand side, financial illiteracy, cultural barriers (religion, ethnicity), lack of need, lack of awareness, and perceptions regarding the individual (creditworthiness related) and the financial system (regarding costs of financial products and trust in financial institutions) may restrain financial access and usage, especially voluntary (Claessens, 2006; Beck & De La Torre, 2007; Beck et al, 2009; Kostov et al, 2012; Campero & Kaiser, 2013; Karlan et al, 2014). 2 Furthermore, World Bank (2005) differentiates between the financially served, either formally (by banks or other regulated financial institutions) or informally, and the financially excluded, which is the population with no access to financial services at all. 6

7 Investigations on the determinants of financial access from the supply side have focused on indicators of outreach in terms of availability of financial service points as a measure for access (Beck et al, 2005; Beck et al, 2008). Beck et al (2008) explore the barriers to access banking services using cross-bank and cross-country data from a survey sent to the five largest banks from 62 countries of different financial and economic development backgrounds in 2004 and With this information, they construct indicators of financial outreach per capita (branch penetration, deposits and loans per capita) and barriers to financial access, in terms of physical access, affordability and eligibility. The authors find that bank size and availability of physical infrastructure are robust predictors of barriers, that financial outreach is negatively correlated with barriers, and that more economically and financially developed countries are likely to have less barriers to bank access, despite the issue of reverse causality between barriers to access and its determinants. Similarly, Honohan & King (2009) suggest that higher income countries may be better endowed for financial access; thus lower income countries may need to work on other aspects such as legal and regulatory institutions and organizational and technological efficiency of financial providers in order to promote financial access. Recent empirical work on the determinants of financial access has moved to the use of firm and household data, with a special focus on demand side factors. Using a multivariate probit model, Bendig et al (2009) find that household demand for savings, loans and insurance in rural Ghana is determined not only by income or other socioeconomic factors but also by risk assessment and past exposure to shocks from the household, together with trust in institutions and products. Campero & Kaiser (2013) investigate the role of awareness of financial sources and services in the decision of individuals engaging in formal or informal financial credit in Mexico, using a three-step multinomial logit to correct for selection bias due to awareness of the sources of credit. They find that household income and schooling increases the probability of awareness of formal and informal sources of lending, albeit reducing the likelihood of using an informal source. In addition, they find that informal sources of credit (in particular family and friends) are relevant when facing income shocks, arguing for a complementarity of formal and informal sources of credit, which attend different segments of the population. In contrast, Love & Sanchez (2009) characterize the demand for credit of individual entrepreneurs and enterprises in rural Mexico, as well as the determinants of the credit constrained. They use a probit model in a twostep procedure, where the first stage controls for selection bias in loan demand, and the second stage estimates the probability of access. Their results indicate that micro entrepreneurs are more likely to demand loans when they are middle-aged, have formal education, formal savings, buy inputs on credit and were affected by adverse events; similar characteristics are presented by credit-constrained micro entrepreneurs. In addition, they find that enterprises demanding more credit are the ones planning to make business improvements, buying on credit and reporting some problems; and those with more productive assets or formal savings and older are less likely to be credit constrained, with regional differences across the country. In terms of perceptions as financial demand determinants, Kostov et al (2012) study the role of financial literacy, attitudes and motivations in the context of the Mzansi intervention, a program designed to provide a basic 7

8 account to low-income individuals in South Africa. They use a generalized linear model with a logit link function for the choice of opening the account, and employ regularization techniques (Dantzig selector) in two steps for appropriate variable selection and model fitting. Their estimation results for 3900 households show that trust in financial institutions, financial aspirations and perceptions, and financial literacy increase the probability of holding an Mzansi account. Therefore, financial access and usage are closely interrelated, with access being a first stage that given appropriate conditions from the supply and demand sides, would translate into the use of financial services, which would be a second stage. Availability of financial service points is a necessary condition from the supply side for access, since access would not be possible without close and affordable service points. In addition, convenience in accessing financial service points is a sufficient condition, which implies ability and will to efficiently access and use financial services. Most studies have focused on the determinants of financial access employing financial use as a proxy; however, this approach does not capture the transition from access to use adequately. The objective of this study is to assess the determinants of financial access and use, employing as a proxy of availability and convenience of financial service points their proximity. This study adds to the financial access literature by using as a proxy for proximity of service points the travel time to formal financial service points reported by individuals. This information is richer than the number of financial service points, in the sense that it provides both information regarding the awareness of the location of financial institutions and the monetary and non-monetary costs of accessing those services, thus being more informative about the convenience for individuals to access financial service points. In addition, the relationship between proximity of financial service points and financial access and use is assessed for both formal and informal services (understanding formal services as those provided by supervised financial institutions) allowing examination of their complementary role. III. THE PERUVIAN FINANCIAL SYSTEM The formal Peruvian financial system is composed by institutions supervised by the Superintendency of Banking, Insurance and Private Pension Funds (SBS), the financial regulator and supervisor that authorizes their operation. Within this system, the financial intermediaries are the most important in size, group composed as of June 2015 by 19 banks (privately and state-owned), 12 financial companies, 12 municipal institutions of credit and savings (municipal cajas), 9 rural institutions of credit and savings (rural cajas), and 11 entities for the development of microenterprises (EDPYME), all of which are categorized as multiple operations institutions according to the Law of the Financial System (SBS, 2008). By June 2015, total assets of these institutions reached 381,614 million Nuevos Soles, approximately 120,118 million US Dollars (SBS, 2015b). Regarding ownership status, Banco de la Nacion is the most important state-owned bank, since it competes with private banks for the retail loan market. A mixed case is given by municipal cajas, whose stakeholders are the municipality and private investors. Regarding target markets, cajas and EDPYMEs have mainly focused on the 8

9 microenterprise segment ever since their creation, whereas banks and finance companies have addressed large commercial loans, although many of them have also entered the microfinance sector in recent years. In regards of operations, all of these institutions intermediate funds sourced on loans or deposits, except for EDPYMEs, which are not allowed by law to collect deposits from the public, a reason why many of these entities have requested their conversion into financial companies in recent years. The SBS regularly publishes information on financial outreach (access, use and depth) based on supply data gathered from supervised institutions (SBS, 2015c). Access to financial services is measured as the number of financial service points, namely branches 3, ATMs 4 and branchless service points known as agents. 5 On the other hand, use of financial services is measured by the number of borrowers and of accounts, while depth by the average loan or deposit balance as a percentage of GDP per capita. These indicators show that Peruvian financial access, use and depth have steadily improved in the last five years. The number of service points per 100,000 inhabitants increased from 109 by June 2010 to 443 by June 2015, and the number of service channels per 1,000 Km 2 rose from 13 to 61 during the same period 6. Similarly, the number of borrowers per 1,000 adults increased from 237 to 311. As for outreach depth, the average loan as a percentage of GDP per capita was 0.91 in June 2010, modestly increasing to 0.92 by June 2015; the opposite direction is observed in the case of the average deposit, which decreased from 0.26 to 0.20 during the same period. In this last case, limitations due to banking secrecy could be affecting statistics for deposits. Despite these trends, the levels of financial access and usage are not the same across the country. 7 By June 2015, Lima (the capital) had 600 service points per 100,000 inhabitants, while the rest of the departments in the country had on average 349. In addition, out of 1837 districts in the whole country, only 56% of them had access to a service point (although it increased from 30% in June 2010). Regarding the use of financial services, Lima registered 491 borrowers per 1,000 adults while the rest of the regions had on average 277. Nonetheless, some departments have outperformed Lima in recent years, especially in the case of access indicators. Arequipa registered 863 service points per 100,000 inhabitants; while 423 borrowers per 1,000 adults. On the other hand, Junin and La Libertad 3 Branches are defined by the Peruvian regulation as physical establishments in which financial institutions operate providing financial services to the public; as such, their range of services is the broadest (SBS, 2013). 4 ATMs are defined as electronic devices interconnected with financial institutions, allowing them to provide financial services by using credit or debit cards or other identifying mechanisms that require electronic signatures or similar. 5 Agents are defined as service points that operate in fixed or mobile establishments owned by individuals or legal entities different from supervised financial entities, and offer authorized operations on behalf of supervised financial institutions through point of sale (POS) devices (called cajeros corresponsales). Agents are allowed, among other things, to open basic accounts and to receive loan applications as well as applications to open other types of deposit accounts different from basic accounts. A basic account is a simplified account created under the anti-money laundering regulation, characterized by some prudential limits and fewer information requirements at opening, making it more accessible to the public (SBS, 2011). 6 When referring to agents, the SBS financial inclusion statistics differentiate between the number of agent establishments and the number of agent POS, where the first is equal or lower than the second as one establishment can have more than one POS. The number of establishments is classified as a service channel; while the number of POS falls under the service point category (SBS, 2015c). In this study, the focus will be on the number of agent establishments (service channels), even though they are referred to as service points. 7 Peru is comprised of 24 departments and a constitutional province. Departments are comprised by provinces, these by districts, and the latter by towns; the number of provinces, districts and towns vary across the country. At the national level, there are a total of 194 provinces and 1837 districts. 9

10 registered 411 and 400 service points per 100,000 inhabitants and 308 and 304 borrowers per 1,000 adults, respectively. Given the limitations of supply side data, on one hand, related to accounts given the restrictions in data collection due to banking secrecy (which may overestimate the number of depositors), and on the other hand, related to coverage and penetration of financial services, the SBS started in 2011 a project to collect information on access and use of financial services from the demand side. The results and database from that project are employed in the current study, and are discussed in the following section. IV. DATA IV.1 Pilot Survey of Access and Use of Financial Services Data employed in the current study corresponds to the Pilot Survey of Access and Use of Financial Services, carried out in Peru in 2012 by the SBS with funding from the Inter-American Development Bank (IDB). This was a first comprehensive survey designed with the objective to collect information on access and use of financial services as well as their barriers from the demand side. The pilot survey was conducted in four departments in Peru: Lima (the capital, located in the central coast), Arequipa (southern coast), Junín (central highland) and La Libertad (northern coast). The survey covered urban and rural areas of the last three departments, and only the urban areas of Lima. The target population was the group of adults between 18 and 65 years old, residing in the aforementioned departments. A random sample of individuals was chosen after a process of several stages, with stratification at the district, town, residence and household level. Some geographical areas were excluded from the sampling process, following a cost-benefit criterion; on average, 95% of the population in the selected departments was considered by the sampling procedure. The survey was undertaken by a total of 2,799 individuals selected under this process, 1,600 in urban areas and 1,200 in rural areas (SBS, 2014b). Additional details regarding the survey geographical outreach are shown in Table A1 in Appendix A. The survey collected information related to the sociodemographic characteristics of the individuals; access to formal financial service points: branches, ATMs and agents of private banks, financial companies, Banco de la Nación (BN), municipal and rural cajas (cajas), and EDPYMEs; savings and accounts, credit and loans; and other financial products and services, including credit cards, insurance, remittances, online banking and mobile banking. Together with objective questions regarding access and use, inquiries about attitudes and perceptions of the financial products and services were also included. Data is representative at the department level, for both urban and rural areas (SBS, 2014a). 10

11 IV.2 Survey results Survey results show that 89 percent of the individuals in the sample are aware of the existence of a formal financial service point, by reporting how far in time these are located relative to their home or job as well as the type of transportation employed. On average, 84 percent of the sample reports the travel time to a BN banking branch, 58 percent to a private banking branch (42 percent to a private banking agent) and 50 percent to a caja branch (19 percent to an agent); meanwhile, 80 percent of the sample reports using public transportation, 13 percent walking to the service point and 7 percent employing private transportation. In addition, it takes 97 minutes on average to reach a private banking branch using public transportation, and 61 minutes walking (for a BN branch, the average time is 110 minutes and 67 minutes, respectively); similarly, for cajas the average travel time is 112 minutes with public transportation and 24 minutes walking. In the case of agents, average travel time is shorter, with 71 minutes using public transportation and 11 minutes walking to get to a banking agent; while it takes 77 minutes and 16 minutes to reach a caja agent by public transportation and walking, respectively. Further information regarding travel time is presented in Table A2 in Appendix A. On the other hand, travel time varies depending on the residence area, with higher time reported in rural areas. On average, walking to a private banking branch takes 2,190 minutes (one and a half day) in rural areas while it only takes 11 minutes in urban areas; using public transportation reduces the average to 150 minutes in rural areas, although time increases to 66 minutes in urban areas. Without considering outlier time values, which are mainly encountered in rural areas, average walking travel time drops to 120 minutes and to 94 minutes using public transportation. 8 Further information regarding travel time is presented in Table A3 in Appendix A. There are noteworthy differences in reported travel time which may be a consequence of both supply-side as well as demand-side factors. Firstly, the sample is characterized by a wide range of travel time per type of channel especially for non-banking institutions (such as cajas) and for branches; this may respond to the fact that in Peru banks (private and public) are more widespread across the country than other types of institutions, and that retail agents are more cost-efficient than branches. Secondly, time is, in many cases, longer using public transportation than walking, which may sound counterintuitive as the use of a motorized vehicle would be expected to reduce travel time; however, concentration of financial service points in a certain location (usually in urban areas) would allow individuals to walk for a short time interval to get to them as they are close, forcing individuals located farther (typically in rural areas) to use motorized transport to reach them. Thirdly, long travel time intervals are predominant in rural areas, leading to outlier values; this may respond to different factors including misreported time (for example, the respondent might have considered not only travel time itself but also stopovers or waiting times), availability of transport infrastructure, geography, or measurement error. Inspection of towns with outliers using mapping services (such as Google Maps) shows that the nearest financial service point is located in another province, and yet other individuals from the same town either report lower travel time, lack of awareness or lack of 8 The means are influenced by high time values reported by 18 individuals in different districts within the Condesuyos province in the Arequipa department. For example, the maximum value reported to get to a private banking branch by any transport type is 4,320 minutes (more than three days), in the rural areas of the aforementioned zones. Without considering this group, the maximum travel time drops to 900 minutes. 11

12 existence of a financial service point. Since there is no certainty regarding the correctness of long reported travel time to a financial service point, model estimation will be done using the whole sample and it will also exclude outlier values as a robustness test. Regarding savings, results show that 38 percent of the sample has savings either in formal or informal financial institutions. Of those who save, 30 percent does so in formal financial institutions while 73 percent recur to informal mechanisms, mainly under the mattress (54 percent), in construction materials and animals (29 percent) or in rotating savings and credit associations ROSCAs (13 percent). Those who do not save report not having enough income to save as the most important reason (85 percent); the second important reason is not having the need to save (7 percent) and the third is that they prefer to spend today (7 percent). On the other hand, only 18 percent of the sample reports having or using a deposit account in a formal financial institution, and uses it primarily to receive salaries (60 percent); the main advantage associated with having an account is that money is safe (60 percent), while the main disadvantage is the fees charged. Those who do not have an account report as the most important reasons the elevated costs (23 percent), far distance of financial institutions (20 percent) and not finding it advantageous to have an account (15 percent); the first two reasons closely relate to the barriers to access from the supply side while the latter relates to the demand side, discussed in section II. Regarding credit, 40 percent of the sample reports having applied for a credit at some point either to formal or informal financial institutions. Of those who applied for a credit, 60 percent did so in a formal financial institution, while 50 percent asked for a loan in informal entities. Regarding credit in formal institutions, 50 percent of those who applied considers interest rates as the most important factor when applying for a loan; in addition, 87 percent of individuals who applied for a loan in a formal entity received it, using it mainly for productive activities such as working capital (28 percent) or starting or expanding a business (21 percent), followed by housing purposes such as buying, repairing or expanding a property (27 percent). Access barriers were also identified for credit. Those who did not receive a formal loan (23 percent of those that applied for one) consider low income (31 percent) or lack of steady income (16 percent), lack of collateral (16 percent) and lack of credit history (14 percent), as the primary reasons for not being granted the loan; whereas those who did not apply for a loan in a formal financial institution (76 percent of the sample), mention as the most important reasons for not applying the lack of need (22 percent), that they think they would not get it (14 percent), far distance from the institution (13 percent) and high interest rates (12 percent). As regards to those who asked for a loan the informal financial system, 89 percent recurred to family and friends, while 11 percent asked for a loan from a ROSCA; additionally, 25 percent informed a fast process as the most important reason for asking for a loan in these entities, followed by payment flexibility (12 percent). In summary, survey results show that while an important proportion of the sample is aware of the availability of financial service points, their location may not be convenient for all individuals, as seen by the differences in 12

13 reported travel time between types of service points and residence areas. On the other hand, financial use is important within the sample, especially credit; however, there are still several barriers from the supply and demand sides that limit use of financial services. IV.3 Variable construction Several questions from the survey as well as combinations of them were employed to construct the variables of interest. Table A4 in the Appendix shows the description of the variables used, along with the questions employed in their construction. Financial access. As seen in the section II, financial access is understood as the availability of formal financial service points which enable an individual to use financial services given appropriate conditions from the supply side (physical proximity and affordability, or availability) and demand side (ability and will to use, or convenience); these conditions, especially the latter, distance this process from being an automatic one. Availability of service points is a necessary but not a sufficient condition for access to formal financial services. Proximity of financial service points, in terms of availability and convenience, is measured by the travel time to a financial service point reported by survey respondents. In the sample, 89 percent of individuals report a travel time; therefore, they are aware of the service points and financial services are available to them. Whether they are able or not to actually use these services may respond to how far are these service points are located from them (distance), which may in turn increase transaction costs (decrease affordability), decrease convenience, and worsen other access barriers previously discussed. It is worth noting that the travel time variable provides valuable information regarding a supply side variable, but from the demand side perspective. 9 In contrast, the other 11 percent that did not report a travel time indicated either that the service point did not exist or that they did not know their location; in principle, this would mean that they do not have access to formal financial services (since necessary condition is not met); however, given that there could be many other reasons behind not reporting the travel time (including misreport and misunderstanding from the interviewee or interviewer), that the number of observations is small, and that there was no town with only non-respondents (all towns have both individuals that report time and that do not), the focus will remain with the 89 percent that reported a travel time to a service point (2488 individuals). These reasons also rule out the possibility of selectivity bias in the sample. A first approximation to the ability to use financial services once service points are available to individuals, henceforth financial access, is to go to the service points in order to ask for a financial service. This can be measured by loan application or asking to open an account, since both products are complementary, and having 9 Information of available service points from the supply side reveal that by March 2012 (when survey interviews were carried out), on average 40% of the districts of the survey sample had a service point; however, travel time to a service point was reported for all districts of the survey sample. 13

14 access to one can provide access to the other one. 10 While loan application may better suit the idea of access, as applying for a credit is the first step towards using the product, in the case of accounts, it may be more difficult to distinguish access from use, since it is unlikely to be denied account opening in Peru. 11 In addition, the survey does not contain information on applications to open accounts. Thus financial access is measured by loan applications only in this study. On the other hand, since informal services are undertaken by a large proportion of the sample, financial access is evaluated both for formal and informal services. On average, 24 percent of the sample has applied for a loan in the formal system and 20 percent in the informal financial system. It is worth mentioning that formal services and informal financial services are delimited by the survey questions. Specifically, formal institutions are comprised of banks (private banks and Banco de Nación), financial companies, cajas (municipal and rural) and EDPYME. Informal services are sourced on entities not supervised by the SBS, including ONGs, pawnshops, moneylenders, and friends and family. Financial usage. Financial use is focused on the probability of having savings, using an account actively as well as using a loan for business purposes. In the last two cases, the sample is restricted to users of accounts and business loans. The probability of having savings is explored in order to compare the use of passive products from formal and informal financial sources. In the case of accounts, active use of the account is approximated by the probability of making at least two or more withdrawals or deposits on average in one month and the probability of using the account to make payments and money transfers or to receive money (salaries, remittances or pensions). In the case of credit use, the variable employed is the probability of using the received loan conditional on applying for it for business purposes (to start it or expand it, or as working capital). On average, 12 percent has savings in a formal financial institution, while 28 percent has savings in informal sources. On the other hand, from those who have an account 41 and 32 percent made at least two withdrawals or deposits within one month, respectively. In the meantime, on average 14 percent used the account to make other transactions, while 80 percent used it to receive money. Regarding credit, on average 45 percent of those receiving a loan used it for business purposes. Travel time. Travel time from home or work of the interviewee is reported in the survey for different types of service points, types of formal financial institutions and of transportation; with most individuals reporting travel time for several or all types of categories. For the case of credit applying for a loan or using it travel time considers only the time to branches and agents, since they are authorized to manage loan application and granting credit. 12 On the other hand, in the case of savings and accounts all types of services points (branches, ATMs and agents) are considered, as all of them provide some platform for usage. 10 For instance, in the formal system having a savings account may improve the chances of getting a credit by acting as collateral (as is the case in Peru); this is also suggested in Honohan & King (2009). In ROSCAs, both credit and savings are usually part of the same scheme depending on a person s turn to receive the money. 11 With basic accounts, it could be even easier for an individual to get access to an account in a formal financial institution. However, denial to open accounts may occur in the case of some business accounts, although these are not under study here. 12 Refer to notes 3 and 5. 14

15 Average travel time is a proxy for the proximity of service points to the individual s location. In order to have a unique measure of time and to avoid reducing sample size by estimating separate models by transport types and losing observations 13, travel time using different transportation types are converted into equivalents of time using public transportation. This is done using ratios of time to get to a service point using public transportation with respect to time using private transportation and to walking. For instance, in Pataz province (in the La Libertad deparment) walking to a banking branch takes on average 0.6 of the time it takes using public transportation; thus, 5 minutes walking would be equivalent to 8.3 minutes with public transport. These proportions are constructed at the province level, due to the fact that the point of departure (home or work) and of arrival (branch, ATM or agent) may not be the same for all individuals within a certain location. However, it may be safe to assume that they would travel to a different district within a province since distances are relatively small between districts, especially when using motorized transportation. These converted times are then averaged through all the types of service points. The limitation of this approach is that it may overestimate the travel time for those having a closer service point (the bottom of the distribution). As seen in Section IV.2, those that walk reach a service point in less time than using public transport, implying that they have a service point closer than those using a motorized vehicle; this higher time interval would reflect the higher cost for him of using the second transport type. To account for this limitation, travel time irrespective of the transport type is also used in the estimations as a robustness test. Average travel time to any point of financial service has a mean of 121 minutes in the whole sample; however, the average time varies with different samples employed depending on the outcome variable under analysis, as observed in Table 1. Since the distribution of average travel time is skewed to the right, the log of this variable is used in order to expand the distribution. Other variables. Individual and household characteristics, along with perceptions regarding financial services are also considered from the survey database. Individual characteristics include gender, age, educational attainment, employment and marital status, number of dependent family members and whether a person receives or sends remittances. Household characteristics are comprised in a welfare index, constructed by the Peruvian financial regulator based on variables related to housing materials, access to basic utilities and to telecommunications, number of people living the residence and affiliation to public insurance, using a principal component analysis (SBS, 2014c); in addition, whether there is a home business and residence ownership are included. Perceptions about financial products are related to costs and process length. Pairwise correlations of the variables employed in the estimations are presented in Table A5 in the Appendix. 13 While 11% of the sample reports walking to a financial service point, transportation use in the survey is not mutually exclusive; thus, sample loss would be higher than 11% if those that walk would be excluded from the estimation sample. 15