FINANCIAL INCLUSION INDICATORS FOR DEVELOPING COUNTRIES*: The Peruvian Case
Giovanna Prialé Reyes and Edgar Salgado Chavez** Instituto de Finanzas Personales
Abstract This paper documents recent development of financial inclusion for the Peruvian case. We propose a set of financial inclusion indicators and explore their performance for the last decade. Despite the remaining spatial disparities in the deepening of the financial system, it has increased its coverage. Besides the indicators, we develop a simple econometric framework to investigate the microeconomic impacts of greater credit availability. We find that households, on their part are more likely to take credits, cope better with shocks, register their businesses and smooth their consumption and income in regions where credit availability has become more apparent.
This paper is based on a previous study done by Giovanna Prialé and Daniel Allaín. The authors work at the Instituto de Finanzas Personales, http://www.ifp-finanzas.com. Please send questions and comments to [email protected]
or [email protected]
This paper´s findings interpretations and conclusions are enterely those of the authors. All errors are responsibility of the authors.
The importance of sound and effective social inclusion policies has been widely accepted for more than a century. However, while the first hints of social inclusion policies were focused in the lack of participation of the population in the economic activities of the country and in governments’ efforts to fight against poverty; it was not until the last couple of decades that the term ‘inclusion’ became part of the sociopolitical vocabulary. This term was first related to education and diversity, and then broadened to cover issues as diverse as disability, health, race or gender discrimination, geographic location, religious or cultural backgrounds and globalization. Nowadays, social inclusion is generally perceived as a human right by itself, being widely conceived as a practice of ensuring that people in organizations and communities feel they belong, are engaged, and connected through their activities to the goals and objectives of the organization or community they are part of. However, one aspect of social inclusion has been left out until recent years, and that is financial inclusion, or inclusion of the population into the financial system. Issues relating to financial systems have been more focused in the soundness of the system itself, in providing the right incentives for financial institutions (FIs) to engage in their business but also taking into account and controlling their risk exposure. Credit risk, market risk and operational risk are terms no one in the financial business is unaware of. Regulators and supervisors have developed better techniques to understand the financial situation of any FI or financial conglomerate, and have been granted resources to take corrective action in case there is evidence of unmanaged or irresponsible risk behavior in any institution. The amount of data that both regulators and the market use to track the soundness of the financial system is enormous and it keeps increasing, as the complexity of financial markets also increases. Nevertheless, data related to the availability of financial services to everyone who could need and afford it under reasonable circumstances had been scarce until the first wave of efforts to promote microfinance raised awareness of the needs of the poor for financial services. Nonetheless, despite the effort toward the filling of this information gap, data collected on financial access and inclusion remains fragmented and incomplete, which makes it difficult to understand the size of the gap in the provision of financial services and the best policies that governments can put into operation to reduce it. The importance of financial inclusion, meaning broad access to financial products and services for a significant proportion of the population, is currently being debated. While its impact on economic development and poverty alleviation has not been clearly demonstrated (i.e. it has not been yet proved that increasing financial inclusion levels only in a population, keeping other factors controlled and constant, has caused a noticeable and significant, long-lasting impact in income and wealth1), there has been some evidence showing that the benefits drawn from the greater availability of financial products and services, helps stabilize poor people’s income cash flows and consumption. The availability of adequate credit sources and deposit accounts directed to people with low income allows them, to cope with unexpected expenses (i.e. a health emergency) without having to wait until their next payment. Well-developed, accessible financial systems allow businesses and households to finance investment projects outside their current budgets and promote a more competitive environment as they provide the resources needed for introduction of new projects and businesses into the market2. This is especially significant in environments where alternative channels (direct investment or capital market access) are not available or are prohibitive for low-income economic agents without collateral or a previous credit history. Availability of deposit accounts allow households to save resources for future use —which is a first approach to insurance of their household assets— and provides access to alternative payment channels. Furthermore, broad access to insurance coverage and pension funds allows the population to save for special needs (for uncertain events and retirement), which have a positive impact on households’ ability to accumulate wealth and conduct longer-term financial planning. The objective of this paper is to produce, estimate, and analyze a set of financial inclusion indicators aimed to assess is the profundity of current financial access; and to map out how financial inclusion has evolved during the past decade. Furthermore, we also analyze what measures can be taken in order to deepen financial access. In that sense, we have focused our work on developing financial inclusion indicators that can be measurable and meaningful for developing countries.
See Banerjee et. al (2010) See Pages (2010).
One of the first steps in producing effective indicators of financial inclusion levels is defining what we mean by financial inclusion in the first place. There has been some debate among specialists (Kendall et. Al 2010, ) and yet there is not a widely approved definition of ‘financial inclusion’, but every definition embodies certain core elements: broad access to a portfolio of financial products and services, financial education and a consumer protection framework. Also, there is a general definition of the minimum requirements for these financial products and services in terms of availability, quality, cost and sustainability. In this first approach, we have decided to use the following definition, as shown in box N° 1. Box N° 1 Definition of ‘Financial Inclusion’ Financial inclusion means that the majority of the population has broad access to a portfolio of quality financial products and services which include loans, deposit services, insurance, pensions and payment systems, as well as financial education and consumer protection mechanisms. Promoting financial inclusion requires creating or enhancing market incentives to develop and provide financial products and services focused in population with low levels of access or use of other types of financial products and services, as well as empowering financial users with the tools needed to better understand financial products and services offered and the channels required to enforce their consumer rights. Greater financial inclusion will promote economic development that contributes to a greater well-being of the society. This will be achieved through the establishment of mechanisms that allow for greater access to products and services of financial institutions; deeper knowledge about banks and microfinance institutions, insurance companies and private pension funds; and improved information disclosure regarding financial products and services’ features, benefits and costs for consumers.
We believe that developing a broad set of qualitative and quantitative financial inclusion indicators measuring access, use and geographical distribution of financial products and services will help to identify adequate measures to incentive the market; therefore financial institutions can provide better targeted products and services in terms of characteristics, distribution channels and prices. However, it is important to distinguish between two groups inside the population who are financially excluded. One of these groups is comprised of people who do not use the financial system due to the presence of barriers that prevent them from contracting with existing financial institutions, like geographical barriers, cultural barriers, trust issues or inadequate products and services for a specific environment but who would be able to access and use the financial products and services offered by financial institutions otherwise. The second group is described by people who do not use the financial system because they do not have the means or the resources to use the financial system; that is, their economic condition is so critical that they would not have any means of repaying a credit, nor generating enough resources to cover even their most basic needs. The first group of people, however financially excluded, is financially capable, which means that they could engage in a productive and benefiting use of the financial system if some conditions are met. The second group is financially incapable and there will be no financial inclusion policy that will help these people. Instead, they would need a more direct approach to combat their poverty issues: direct transfers or subsidies of resources, social programs to enhance labor and other actions that would have a direct impact in their income until their basic needs are met.. While we seek to improve financial inclusion, we are aware that only those people who are financially capable will be able to be included, whereas people who do not meet financial capability conditions will need other type of support before being able to become financially included.
We must point out that our analysis excludes new channels for accessing financial services, such as e-banking and mobile banking, or mobile-initiated transactions, because of a lack of separable user data. Instead, we focus on more traditional financial instruments like agencies and ATM coverage. We also make a first attempt to provide microeconomic evidence of the impact of financial deepening on Peruvian households. For that purpose we merge geographic data available from the Superintendence of Banking and Insurance (SBS in Spanish) and the National Household Survey (ENAHO in Spanish). Moreover, we construct a panel data set for the last three years covered in the analysis of this paper. The strategy consists on relating the greater credit access through Peruvian regions with greater take-up of credits at household level, better absorption of shocks and consumption and income smoothing. Preliminary results suggest that greater credit availability in Peruvian regions, after controlling for demographic and aggregate variables, holds a positive relation with greater credit take up. Moreover, households located in regions with greater financial access are more capable to cope with shocks, formalize their businesses and smooth their consumption. The rest of this document will be organized as follows. Section II presents the current economic environment in Peru and its recent evolution, in order to adequately interpret the proposed indicators. Section III discusses the methodology used to calculate our proposed set of financial inclusion indicators as well as the particular considerations about the data we have worked with, define and group those indicators into categories, and present our results both individually for each indicator and in aggregate. Section IV introduces a simple framework to test the microeconomic implications of financial deepening. Finally, Section V presents our conclusions and closing remarks.
CURRENT MACROECONOMIC AND FINANCIAL ENVIRONMENT IN PERU
The last decade has been one of consistent economic growth for Peru. With an average annual GDP growth of 5.1% between 2000 and 2009 despite the global financial crisis of 2008 – 2009, Peru has been able to adjust to the negative global economic environment without having to sacrifice its fiscal balance or resort to excessive expansionary monetary policy. As a result, consumer price growth has been modest in the last decade, with average annual inflation for that period being 2.6%. This result, together with the improving economic environment, allowed for sustained growth of the financial system while maintaining internal price stability during the last decade. These factors promoted an increase of the GDP per-capita from US$ 2035 in 2000 to US$ 4533 in 2009, which means a bit more than a 9% year-to-year increase in average. Figure 2: Gross domestic product and consumer prices growth, 2000 – 2009
GDP (var.% YoY)
Inflation (avg. var% YoY)
2.0 0.0 2000 2001
2004 2005 2006 2007 2008 2009
Figure 3: Gross domestic product per-capita, 2000 – 2009
Thousands of U.S. Dollars (US$)
4.0 3.0 2.0 1.0
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
These results are reflected in increased financial intermediation through regulated FIs, which include private and state-owned banks, finance companies and microfinance institutions (municipal savings and loans institutions, rural savings and loans institutions and entities for the development of micro and small enterprises), all of which are regulated and supervised by the SBS. Undoubtedly, the last decade has shown an explosive growth of loans and deposits, which have increased their size by more than 200% between 2000 and 2009, reaching US$ 37,371 and US$ 43,394 million by the end of 2009, respectively.
Figure 4: Direct Loans and Total Deposits in Financial Institutions, 2000 – 2009
Millions of U.S. Dollars (US$)
Direct Loans Total Deposits
15,000 10,000 5,000 0 2001 2002 2003 2004 2005 2006 2007 2008 2009
However, these economic and financial results have not translated into a deeper penetration of financial products and services in relation to the size of the economy, compared with other Latin American countries. Despite the ratio total deposits as a percentage of the GDP (a commonly used ratio for measuring financial penetration) having increased during the last five years, Peru is still below average (26.8% as of June 2009), far below Chile and Bolivia (who show ratios of 66.4% and 40.2%, respectively), considering deposits in banking institutions only. Figure 5: Financial penetration in Latin America, June 2009
75 60 45
Total Deposits / Annualized GDP (%)
In an environment of economic growth and increasing financial intermediation, the low financial penetration level of Peru compared to its peers means that there is still a significant portion of economic transactions taking place outside the financial system. This can only be explained by assuming that either: some individuals have no access to financial products and services (offered by regulated FIs); access is limited because of physical or economical barriers; or some economic agents are not willing to use existing financial products and services. All of these factors help explain why, despite strong economic growth, some individuals still resort to informal or alternative financial services’ providers.
Each of the above limits to overall use of formal financial services in Peru can be addressed through adequate policy actions such as fostering creation of low-cost, high-quality, simple products and services which could appeal to a broader part of the population; stimulating the creation and strengthening of alternative channels like banking agent, e-banking and mobile banking networks; investing in financial education initiatives; or strengthening recourse mechanisms and consumer protection regulation so economic agents can make informed choices and take proper action in case they have complaints about financial products or services provided by financial institutions. Even if there is still a significant portion of the population that is unbanked, financial intermediation levels of Peru have been growing since 2005 onwards, considering not only private and state-owned banks but also regulated FIs’ total deposits and direct loans as percentages of annualized GDP. Total deposits have increased from 26.2% of GDP in 2000 to 32.9% in 2009, while direct loans have risen from 22.3% to 28.3% in the same period, coinciding with the strengthening of the economy and the regulated financial system. Figure 6: Financial Intermediation in Peru, 2001 - 2009
As a percentage of Annualized GDP (%)
20.0 10.0 0.0
2001 2002 2003 2004 2005 2006 2007 2008 2009
It is in this context that we define and measure a first set of financial inclusion indicators, whose main objective is to provide us with information regarding access and use of financial products and services, so adequate measures can be taken to correct and improve any identified weaknesses, and also to allow us to measure the impact of any policy action taken on financial inclusion levels, so we can identify policy actions that are effective in boosting financial intermediation.
METHODOLOGY AND RESULTS
The financial inclusion indicators estimated in this document follow the work of Thorsten Beck et al. (2006)3, applying the indicators proposed in that paper to Peru, as well as expanding its set with five additional indicators, adding up to 13 financial inclusion indicators classified in three groups: indicators of access to financial products and services, indicators of use of financial products and services and indicators of geographical distribution inequality. a. Methodological considerations In order to understand the following indicators, there are some considerations that have to be taken into account: Population and territory size was obtained from the National Institute of Statistics and Information Technologies (INEI, in Spanish). Macroeconomic data was obtained from the Reserve Central Bank of Peru (BCRP, in Spanish) and financial data was obtained from the SBS, except for financial penetration data presented in Section II, which was obtained from the Banking Association (ASBANC). As mentioned before, the SBS supervises and regulates not only the banking system, but also other non-banking financial institutions, so we have credit, deposit and branch data for them. In order to show a more detailed picture, indicators for private banks (Multiple-operation banks, or BM), private banks and microfinance institutions (Multiple-operation institutions, or OM) and the whole financial system, including state-owned banks (Financial system, or SF) are presented. We have considered, following the recommendations made by Thorsten Beck et al., the number of debtors and depositors (users) instead of number of loans and deposits. These measures affect our calculations when estimating indicators of use of financial products and services. Additionally, data for number of depositors was added up from raw data of depositors by type of deposit for 2000 – 2004, instead of taking into account depositors with more than one type of deposit, which caused overestimation of the total number of depositors. This error was corrected from 2005 onwards, which has caused an artificial decrease in the number of depositors and an artificial increase in the average size of deposits per depositor, but it does not alter the overall tendency of these series. The SBS classifies loans and creditors by use, identifying four types: commercial loans, small business loans, consumer loans and home mortgage loans, distinguishing between commercial and small business loans by total loans granted to a debtor, so a commercial loan implies previous or available access to the financial system for a debtor (see box N° 2). For that reason, we have decided to exclude loans and debtor data from commercial loans and debtors, since them are already “financially included” in the system. In order to focus on financial inclusion of the population, we have considered deposits and number of deposit accounts from individuals only. In this third draft, we have developed two different groups of indicators regarding geographical financial inclusion, both of them considering total loans and total deposits (including commercial loans and both profit and non-profit institutions as well): o Indexes of loan and deposit growth by location (separating between the capital city, Lima, and other cities). Since we are also interested in showing financial institutions’ ability to foster financial inclusion by providing easier access to their products and services, for these geographical indicators we have only considered municipal and rural financial institutions which have their main offices in a city other than Lima but branches both in Lima and in other provinces, making them ideal to study this effect. o Gini indexes comparing loans, deposits and number of branches against population by region. For this calculation we have used loans, deposits and branch data at the municipalities’ level,
aggregating it by departments. Gini indexes give us an idea of distribution inequality, where the lower the index, the lower the inequality in the distribution.
Box N° 2 Classification of Credit Type by Use in Peru According to the current Regulation for Debtor Classification (approved by Resolution N° 8082003), financial institutions should follow these directives for loan classification: Commercial Loans: Direct or indirect loans granted to people or businesses for financing production and commercialization of goods and services, in their different phases. Small Business Loans: Direct or indirect loans granted to people or businesses for financing production and commercialization of goods and services, with a total level of debt with the financial system not exceeding US$ 30,000. In case of loans granted to people, these debtors should have business activities as their main source of income. In case of people or businesses which are part of a financial conglomerate or an economic group, the limit will be measured against the conglomerate or the group, and not individually. Consumer loans: Loans granted to people for use in products, services or expenditures activities not related to a business activity, including credits granted by credit cards, leasing arrangements and any other financial operation not related to business activities. Home mortgage loans: Loans granted to people for use in acquisition, building, reparation, remodeling, enlargement, improvement or subdivision of an owned housing, if such loan is guaranteed by a properly registered mortgage, including loans of such characteristics granted to directors and officials of the granting financial institution. If a debtor has more than one type of loan granted by the same financial institution, their classification should be based on the riskier type of loan, not taking into account consumer and small business loans lesser than twenty Nuevos Soles (S/. 20.00). Our indicators have been calculated for the period 2001 – 2009, except for financial institutions’ branch data, which was available from 2000 to 2009; financial institutions’ ATMs and banking agents, which was only available from 2008 to 2009; and loan and debtor-related data, which covers the period 2002 – 2009. b. List and classification of indicators of financial inclusion The indicators of financial inclusion we have constructed and calculated are the following: I) Indicators of Access 1. Number of branches per 1000 km² 2. Number of branches per 10,000 pop. 3. Number of ATMs per 1000 km² 4. Number of ATMs per 10,000 pop. 5. Number of agents per 1000 km² 6. Number of agents per 10,000 pop. II) Indicators of Use 7. Number of depositors per 1000 pop. 8. Number of debtors per 1000 pop. 9. Average size of total deposits per depositor to GDP per capita
10. Average size of total loans per debtor to GDP per capita III) Indicators of Geographical Inequality Distribution 11. Difference between participation of loans in provinces and participation of deposits in provinces (in percentage points) 12. Total loans in provinces to total deposits in provinces (index) 13. Gini indexes between population, loans, deposits and branch offices per region c. Results i. Indicators of Access
The following group of indicators provides a broad figure for estimating the existence and quantities of provision channels available to the population; that is, how many locations providing access to financial products and services are available. 1) Number of branches per 1000 km² The following indicator has been calculated considering the total number of physical branches by both banks and all regulated FIs, and measures geographical penetration of FIs’ branch networks, as a proxy of the average distance to a branch, if branch distribution was geographically uniform (which it is not). This indicator shows a constant increase during the past decade, reaching more than 2.3 branches per 1000 km2 in 2009, considering all FI branches (and almost 2 branches considering private FIs only). According to the Financial Access 2009 report by CGAP4, the World median is approximately 9.2 branches5 while the South American average is 1.99, so this figures show that Peru, despite being still under the 50th percentile, it is not far away from its peers. Figure 7: Financial institutions’ branches per 1000 km2, 2000 - 2009
BM OM SF
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
2) Number of branches per 10,000 pop. This indicator measures demographical penetration of the branch network, estimating how many people each branch would have to serve, so a higher number of branches would indicate than less people have to be served by each branch, which would imply easier access. In Peru, there are just over 1 FIs branch per 10,000 people
See CGAP (2009) While the reported average on the 139 countries surveyed is 40.02 branches per 1000 km2, this figure is strongly biased due to the presence of two outliers: Singapore (599.4 branches) and Hong Kong (1386.8 branches).
(0.87 branches belonging to private FIs), while the World median estimated by the Financial Access 2009 report is 1.68 and the South American average is 1.23. These figures mean that Peru is still behind the average branch network size for its population, although this indicator also shows a constant increase in recent years. Figure 8: Financial institutions branches per 10,000 pop., 2000 - 2009
1.00 0.80 0.60 0.40 0.20 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
BM OM SF
3) Number of ATMs per 1000 km² This indicator provides information about geographical penetration of the ATM network. In Peru, there are 3.5 ATMs per 1000 km2 (2.97 of them belonging to private FIs’ networks). Again, the World median according to the Financial Access 2009 report was 15.5 ATMs per 1000 km2 while the South American average is 6.6 ATMs per 1000 km2, which means that Peruvian’s ATM network development, while expanding, is still very immature related with other countries. Figure 9: Number of ATMs per 1000 km², 2008 - 2009
4) Number of ATMs per 10,000 pop. This indicator estimates how many people should be served by each ATM, so the bigger this indicator is, the lower the number of people using the same ATM will be, thus providing easier access to the ATM network. In
Peru, there are 1.53 ATMs per 10,000 people (1.3 ATMs belonging to private banks and microfinance institutions), which is also under the median estimated in the Financial Access 2009 report: 3.34. South America average, in comparison, is a bit higher: 3.75 ATMs per 10,000 people. Figure 10: Number of ATMs per 10,000 pop., 2008 - 2009
3.00 2.50 2.00 1.50 1.00 0.50 2008 2009
BM OM SF
5) Number of agents per 1000 km² Informally called ‘banking agents’, these independent businesses work with financial institutions by offering services connected to a financial institution’s network, thus being able to provide some of that financial institution’s products or services to their surroundings. The presence of these agents has greatly increased the geographic area and number of people FIs can serve. In Peru in 2009, there are 4.69 agents per 1000 km2, more than double the figure for FI branches. There are no banking agents established by state-owned banks. Figure 11: Number of agents per 1000 km², 2008 - 2009
5.00 4.00 3.00
6) Number of agents per 10,000 pop. The total population covered by correspondent tellers (the Peruvian Government’s official name for these ‘banking agents’) has also increased in the past years, and in 2009 there were 2.1 agents per 10,000 pop.
Figure 12: Number of agents per 10,000 pop., 2008 - 2009
3.00 2.50 2.00 1.50
1.00 0.50 2008
Indicators of Use
The following group of indicators provides broad estimates of the portion of the population who actually use financial products and services through any available channel; that is, who decide to use those products, services and access channels made available to them by FIs. 7) Number of depositors per 1000 pop. This pair of indicators (this and the next one) estimate the proportion of the population using the most demanded financial services: deposits and loans. For deposits, there has been a constant growth of the number of people using them, and there are now more than 425.2 people with a deposit account per 1000 people, or a little less than 3 in 7 people. Figure 13: Number of depositors per 1000 pop., 2000 - 2009
400 350 300 250 200 150
2001 2002 2003 2004 2005 2006 2007 2008 2009
BM OM SF
8) Number of debtors per 1000 pop.
This indicator, as the previous one, estimates the proportion of the population who has access, has decided to ask for and was granted a loan. In Peru, the figure has grown between 2002 and 2009 from a little less than 90 people to 247.5 people per 1000 people (or about 1 in 4 people). This means that in recent years FIs have extended their reach to those previously outside the financial institutions’ sphere of influence with microfinance and low-income, consumer-oriented products and services. The fall in the private bank (BM) series observed in the graph is not a change of tendency in 2009, but the consequence of a consumer-oriented bank turning into a finance company (which is included in the OM series).
Figure 14: Number of debtors per 1000 pop., 2002 - 2009
250 200 150 100 50 2002 2003 2004 2005 2006 2007 2008 2009
BM OM SF
9) Average size of total deposits per depositor to GDP per capita This indicator estimates a rough proxy of the savings/income ratio of the population, but it has to be interpreted with care. The graphic shows a constant decrease of the total deposits per depositor/GDP per capita ratio over the last decade, going from about 48% to 29% in 8 years, which could be seen as a tendency to save less. However, if we take into consideration that both GDP per capita and deposits in financial institutions have been growing over the last decade, as well as the total deposits/GDP ratio, and the number of depositors per 1000 people has been also growing, we will see that this indicator shows that the average size of total deposits per depositor have reduced because new depositors with smaller deposit accounts have entered the financial system during recent years. That means we now have a less-concentrated deposit base, with more depositors and smaller deposits in average, which probably come from regions or groups previously unattended by FIs.
Figure 15: Average size of total deposits per depositor to GDP per capita, 2000 - 2009
70% 60% 50% 40% 30% 20%
2001 2002 2003 2004 2005 2006 2007 2008 2009
BM OM SF
10) Average size of total loans per debtor to GDP per capita Similar to the previous indicator, this measure provides an estimation of the average debt size per debtor (that is, their loan portfolio as a whole) against income. This indicator has also fallen from 58.4% to 50.3% in the last decade, although it shows a growing tendency since 2007. In a context of growing GDP and GDP per capita, growing financial institutions’ total loan portfolio size and growing number of debtors per 1000 people, this means that, on average, the average debt size of the population has been increasing in the last two or three years. More people are having access to credit and those with previous access are using more of it, something that is reasonable in a context of economic growth like the one observed in Peru in the last years. It is also probably a sign that financial institutions (specially private banks) are feeling more comfortable giving loans now than at the beginning of the decade, in order not to repeat the explosive consumer-oriented loan growth that was observed during 1993 – 1997 and which ended in over-indebtedness for a significant portion of the population –many of them people recently included in the financial system due to a consumer loan or a credit card– and a credit ‘crunch’ for them at the end of century.
Figure 16: Average size of total loans per debtor to GDP per capita, 2000 - 2009
BM OM SF
2002 2003 2004 2005 2006 2007 2008 2009
Indicators of Geographical Distribution Inequality
The following section presents two groups of indicators which provide a first approach to geographical distribution and impact of FIs’ branch networks in different regions. In the first group, we try to focus on the effect of having regional FIs participating outside their base cities. Specifically, we are interested in analyzing the behavior of municipal and rural savings and loans institutions, whose main offices are located outside of Lima, to study if they are using their branches in Lima to siphon resources out of its population and finance projects in lower-income cities with them; that is, if the presence of a bigger branch network allows for resource transfers between high-income and low-income regions. The hypothesis here is that municipal and rural savings and loans institutions are financing the creation of loans in Peruvian provinces with deposits raised in Lima, effectively contributing to a redistribution of resources. As a first approach, we will follow the evolution of credits and deposits generated in Lima and in other regions considering that Lima average levels of well-being are superior to those of other cities. The second group of indicators has been calculated from the distribution of loans, deposits and FIs branches, comparing them against the population distribution in Peru, at the departmental level. In this case we have estimate Gini indexes, which compare the cumulative distribution of resources against the cumulative distribution of population (the Lorenz curve), thus measuring deviations in resource distribution from the equality distribution. These indexes does not take into account other factors like regional production levels or initial wealth of the population in each department, thus allowing us to see the pure statistical distribution, which then can be used as a starting point to find some answers about financial inclusion in each department. 11) Difference between participation of loans in provinces and participation of deposits in provinces (in percentage points)
This indicator (shown in the graphic below as a blue bar) provides a broad picture of the evolution of the dynamics of loan and deposit creation in Peru, between Lima and other cities, considering that if the gap between the proportion of loans not in Lima and the deposits not in Lima grows (or the negative gap falls), those new loans not in Lima are probably being financed by deposits or funds from Lima or abroad. Normalizing the indicator to zero in 20016 (so we don’t have to deal with a negative indicator), we observe an increase in the participation gap of 15.9 percentage points, which means that the participation of loans in provinces has grown
The actual difference of between participation of loans in provinces (94,6%) and participation of deposits in provinces (97,9%) was 3,3 percentage points in 2001.
faster (or decreased slower) than participation of deposits in provinces, thus suggesting a transference of resources from Lima to other cities via the these specialized, province-based financial institutions. 12) Total loans in provinces to total deposits in provinces (index) This indicator (the red line in the graphic below) is the normalized ratio of total loans outside Lima over total deposits outside Lima, considering municipal and rural savings and loans institutions. We can see that this normalized ratio has also grown during the past decade, reaching a value of 1.25 in 2009, which means that loans in provinces outgrew deposits in provinces by near 3% per year. This also suggests that funding for those loans should have come from other sources of financing not located in these provinces, like deposits from Lima or credit lines from banking institutions from Lima or abroad. Figure 17: Indexes of Geographical Distribution of loans and deposits, 2001 - 2009
%CredProv - %DepProv
% Part. Loans in Prov - % Part. Deposits in Prov. (2001=0)
Loans in Provinces / Deposits in Provinces (2001=1)
1.25 1.20 1.15 1.10 1.05 1.00
15 12 9 6 3 0
2001 2002 2003 2004 2005 2006 2007 2008 2009
13) Gini indexes As it was mentioned at the beginning of sub-section III.c.iii, we have estimated three series of Gini indexes, comparing the cumulative distribution of Peruvian population per department against the total value of the loans, the total value of the deposits and the number of financial institutions’ branches opened in each department. Each series allows us to study the evolution of the inequality in the distribution of loans, deposits and FIs branches, while the three of them combined allow us to investigate the relationship between these three distributions. During the past decade, the Gini index for the distribution of total loans decreased from 0.60 to 0.46, due to the reduction in the participation of loans created in Lima from 85% to a little over 70% of the total portfolio of loans in the financial system. This process has been steady during the decade, with each year registering a constant increase of the participation of loans in provinces over the total loan portfolio and a constant decrease of the Gini index of about 0.02 points.
Figure 18: Lorenz curves of the distribution of loans against population, 2001 and 2009
Regarding deposit distribution, however, we don’t observe the same process. The Gini index of deposits against population decreased from 0.58 to 0.53 between 2001 and 2009, but most of that decrease was achieved between 2008 and 2009 (0.04 points). In fact, this Gini index increased during 2007, due to a significant raise of deposits in Lima. Total deposits in Lima have accounted for 82% - 83% of total deposits during most of this period, which could mean that people in Lima have a greater capability for create savings, but could also be the result of a strategy in the financial system for raising resources in Lima in order to finance their lending in other regions. Figure 19: Lorenz curves of the distribution of deposits against population, 2001 and 2009
Analyzing the Gini index of branches against population, we come to an unexpected result. One of the main suppositions regarding financial exclusion is lack of access to branch offices, but in the Peruvian case this has not been the case. The Gini index has been very low during the decade, ranging from 0.23 in 2001 to 0.18 in 2009, which means that inequality in the distribution of branches against population of each department has been low. Even more surprising, Lima is not located at the top the branch distribution, but Moquegua, and there are two other departments before Lima in 2009 (Lima has 1.41 branches per 10,000 people while Moquegua has 1.77 branches per 10,000 people in 2009). Lima branches have decreased their participation from 46% to 42% of the total branch network.
This result indicates that the observed inequalities in the geographical distribution of loans and deposits could be less related to lack of access and more related to other factors, like lack of knowledge about the products and services offered by the financial system, low levels of trust in the financial system or a perception that existing products and services are not adequate for part of the population (especially in the case of deposit accounts). Figure 20: Lorenz curves of the distribution of branches against population, 2001 and 2009
The following figure shows the evolution of the Gini indexes discussed in this section, and it’s clear from it that the branch distribution has never been as unequal as the other two, and it’s also visible that the loan distribution has been reducing its inequality levels in a slow but constant rate, while the deposit distribution is still very concentrated in Lima. We are seeing a financial inclusion process from the credit side (retail or small-business loans). Figure 21: Evolution of Gini indexes of loans, deposits and branches, 2001 – 2009
0.40 Loans Deposits
Summarizing all the results, we can see that financial institutions have been doing an effort to increase their network reach, especially in later years, with agents being one of their most potent tools in doing so. While it is true than their network is still very small compared with other countries, which could explain in part the low financial penetration levels observed in Peru, these efforts have shown some results, both in number of clients and in the source of those clients, being probably added from sectors with previous low or nil access to financial
services. We have also shown that presence of financial institutions not based in Lima can foster credit opportunities and financing for families and small businesses in provinces, which have been happening over the last years since municipal and rural savings and loans institutions were granted permission to operate in Lima in 2003; however, it has been also shown that while more funding is available for people living outside Lima, there is not a deposit-creation process of the same level going on, despite the existence of financial institution branches (often named as one of the most important drivers of deposit creation), which could mean that people outside Lima face other barriers – besides lack of physical access – to become financially included, at least from the savings side. IV. MICROECONOMETRIC EVIDENCE
In this section we provide econometric evidence on the relationship between financial access and a set of variables related to credit take up, shock resilience, business initiative and consumption and income smoothing at household level. In doing so, we rely on two data sources. First, we use the financial inclusion indicators developed in the above sections with SBS data. Our second data source is the ENAHO. Since the national survey is a very comprehensive data set covering all departments of Peru, data of financial access provided by SBS at departmental level can be easily matched to the data covered in ENAHO. In a similar vein, the work of Aguilar (2011) assesses the relation between the availability of microcredit and economic growth at the regional level. Using regional-level production data to estimate the rate of economic growth and the provision of loans, also at the regional level, the author finds evidence that suggests a positive relationship between economic growth and the expansion of microcredit availability. Interestingly, using an alternative measure of financial deepening such as bank intermediation, the author does not consider any effect of this variable on economic growth at the regional level. As a simulation exercise, it is shown that if the provision of loans from rural banks, municipal banks and banks specialized in microcredit reaches 10 percent of GDP, that would imply a 4 percentage point increase in the GDP per capita growth rate. An important drawback of this study is the assumption of homogeneity of the relationship between credit expansion and the development of regions. While it seems likely that certain activities with different credit requirements are concentrated in different regions, this possibility is not taken into account in the analysis. The strategy adopted in this paper consists on exploiting regional and time variability in financial access to gauge the effect of greater credit availability on household’s indicators. In the same token, the rich data provided by ENAHO allows us to control for demographic and geographic variables when modeling the impact. We construct a panel of households spanning from 2007 to 2009. The sample is constituted by an unbalanced sample of households. There are four groups of variables we are interested in: a. Take up In this first group we aim to measure whether households are more likely to take credit within regions with greater credit availability. The variables we look at are related to taking a credit for household repairs, in general, and differentiated by the type of institution: commercial bank or Banco de Materiales (a major source of funding for construction, managed by the government). We also look at the amount of credit taken. b. Shock reactions In this section we group variables indicating the capability to cope with unexpected events. More precisely, we assess the relationship of greater credit availability with the absorption of shocks underwent by the households. c. Business Here we have variables indicating whether the independent business of the household took certain actions under the context of greater credit availability. We have variables indicating business registration as a formal entity, the starting up of new businesses and the profits earned.
d. Welfare Smoothing In this last group of variables we explore the capability of the household to minimize the variability of the transitory component of both consumption and income. For groups a, b and c we follow a model as in equation (1) (1) Where “i” represents the household, “c” is the geographic location of the household, and “t” is the time dimension. is the fixed effect for the household whereas is a time dummy. The vector has control variables as household size and the consumption decile7. FI is the financial inclusion indicator (in logs) per region “c”. In the estimation we have relied in the access indicators: number of agencies per 10,000 individuals (lagged one year8). In order to provide more results, we have considered two types of institutions, banks and a second group comprised by Cajas Municipales and Cajas Rurales (termed “cajas”). is the error term. Finally, we present results for the national sample, and disaggregated by urban or rural status. The estimation procedure for group c required some assumptions. Since we are interested in evaluating the capability of smoothing transitory outcomes, we have to assume a structure that allows for permanent and transitory outcomes9. Equation (2) presents the model; (2) is the output variable we are interested in: consumption or income (per capita and logs). The first three elements in the right hand side of the equation represent the permanent component of the output. As in equation (1) captures the household fixed effect. With the element we are assuming each household follows an idiosyncratic trend10; and with the vector we are controlling for the variables that also capture permanent components of variable of interest (consumption or income): the household size and the number of adults in the household. Finally, the error term represents the transitory component of the variable of interest. For the purpose of the estimation of model in equation (2) we constraint the sample to a balanced panel of three years. In that sense, we end up with three transitory outputs for each variable of interest, per household. Since we are interested in consumption or income smoothing, we need to come up with a measure of transitory variability. The smaller the variability, the more stable the income or consumption of the household over the period. Thus, we estimate the standard deviation per household through the 2007-2009 period. That standard deviation is then assessed in the face of greater credit availability. Equation (3) shows the relationship: (3) is the standard deviation of the transitory output, per household. is the average variability of the transitory component of the sample, whereas is the growth rate of the financial inclusion indicator for the 2007-2009 period. We are interested in assessing whether households located in a region with greater growth in financial inclusion experienced a smaller variation in their transitory components of consumption or income. Results are shown in the appendix. Tables 1 to 3 present the results for model (1), whereas table 4 does so for model (3). As stated above, we have considered two types of variables representing credit access per department: number of bank agencies per 10,000 individuals and number of “caja” agencies per 10,000 individuals. Both variables in logs. Finally we have considered three types of estimations to investigate whether
That is, we divide the complete consumption per capita distribution between ten groups, from the poorest to the richest. This is a wealth control variable. 8 Data from ENAHO is collected through all the year, thus we have to allow for some time between the arrival of new agencies and their impact in the cities. 9 See Deaton (1998). 10 As stated in Wooldridge (2002), this is sometimes called a random trend model, as each individual is allowed to have its own trend. The additional individual-specific trend is another source of heterogeneity.
there is some heterogeneity driven the results: besides the national estimation, we present results for urban and rural samples. Looking at the results, In table 1 we show some credit take-up variables. First, there is no correlation in the overall credit up-taking for housing repairs for any of the samples, although the coefficients are suggestively positive. When looking at the credit up-taking from Banco de Materiales, we don’t find any significant result, although the coefficients turn to negative. It is in the third variable, credit from a commercial bank that we see a positive and significant correlation. Interestingly, the correlation seems to be driven by credit up-taking in rural households. In general, a 10% increase in the number of bank agencies per 10,000 inhabitants is related to a 1% increase in the likelihood of taking a credit for housing repairs. Data from ENAHO also allows us to see the amount the household requested in the loan for housing repairs. When assessed, we see a positive relation here too. Suggesting the major credit availability at regional level is related to an increase in the amount household requests. Although the national result seems to be driven by the coefficient in urban households. Thus, if there is a suggestive relation pointing out not only to an increase in the likelihood of credit take up, but also an increase in the amount the household requests. Finally within this group we assess a more general take up variable. Within the “perceptions” module in the ENAHO, the head of the household is asked for any type of financial service he/she took during the last year. This variable, then, could embody any type of financial service 11, not only credits. The result suggests there is a positive correlation, specially driven by urban households, and the effect is stronger for the banks variable. Once stated that there is a significant correlation between the greater credit availability and credit take up, it is worth asking about the implications this credit availability has on other variables. Table 2 relates credit availability to a set of variables that explain the capability of the household to cope with shocks12. The first variable indicates that the household suffered a shock13 in the last 12 months and had to use its own saving to cope with it. We see that in locations where credit availability has increased, the likelihood of depleting own savings is smaller. Thus, suggesting households have more choices when faced with a shock. The second variable tries to confirm this, but we have an unexpected result. The likelihood of using a loan to cope with a shock is smaller in locations with greater credit access. We suspect some error measurement in the variable, since the head of the household is not asked to specify what kind of loan he/she requested. So, it might be the case of informal loans, and the greater credit availability is operating here through a substitution effect. However, this cannot be confirmed with ENAHO data. The final column of table 2 shows a more extreme measure to cope with the shock: reducing consumption. Here we find a strong significant effect for “cajas”. Households that suffered a shock are less likely to reduce consumption in those regions where “cajas” have increased more. It is worth noticing the extent of the coefficient is higher for rural households. Another interesting dimension to look at is the business behavior of the households. Table 3 presents some results in this regard. Based on data from the “independent business” module in ENAHO, we construct three variables to investigate their relation to greater credit availability. The first column shows the likelihood the business is registered (opens up a tax payer registration number, RUC in Spanish), and we see that there is a positive overall correlation for “cajas”, which in turn is confirmed in the urban and rural subsamples. Also, greater credit availability measured by bank agencies has an impact only in the rural sample. This results is aligned with the within firms channel leading to formalization of firms, as explained in Moron, Salgado and Seminario (2012): “The within (or intensive) channel operates by encouraging formalization within the firm's size category. The idea is that access to credit requires compliance with tax and employment legislation. Thus, firms are more likely to incur such costs of formalization once bank credit is more widely available at a lower cost” As in their work, we find here a positive relation between greater credit availability in regions, and the likelihood of registering the business. More interestingly, we find this effect to be particularly strong in the rural subsample. When we look, however at the likelihood of starting new businesses, there is no statistically significant
Unfortunatelly, there is no a detailed explanation about the definition of “financial service”. We are interested in shocks with big impacts. Thus, the shock variable used in this estimation is defined as an income shock that had big consequences on the household. 13 Income shock.
association. This result, suggestive enough, points out to the fact that starting a business in Peru is not too difficult, thus credit access does not affect this behavior. Nevertheless, when it comes to improving the business, through the registration, credit becomes important. Finally in table 3 we see that businesses profits are higher for households located in regions where credit has expanded more intensively. This result is strong for all the samples. Finally, table 4 presents the results for the model explained in equations (2) and (3). We find that the growth in credit availability measured as the number of agencies per 10,000 inhabitants has a negative impact in the volatility of transitory consumption and income. Households located in regions that experienced a higher growth rate have less volatile consumption and income. This result confirms one of the major advantages of credit: the smoothing of consumption and income. The recent period of financial development in Peru is not the exception in this regard. V. CONCLUSIONS
Financial inclusion is not only important because it can potentially enhance economic growth and reduce external shocks on households and business alike (by allowing people to stabilize their cash flows), but because – like other types of inclusion – it makes individuals feel part of a group, thus raising their self-esteem and their wellbeing, finally having an impact in overall welfare. It is a human right; in the sense that anyone who needs a service provided in a region and is able to afford it under normal market circumstances should be able to access such a service. In this case financial inclusion guarantees the ability of people to transform wealth over time, accessing future wealth before time or storing wealth for the future (and future generations), which makes development planning over time feasible, reduces wealth fluctuation and contributes to population overall wellbeing. We believe that this paper, by providing a first set of tools with which to study and evaluate financial inclusion, contributes in a small way to this ultimate goal. In this paper we have defined and developed a set of financial inclusion indicators to provide effective information about the access and use of financial services by the population and complement other financial indicators commonly used in the literature. We have also estimated those indicators and provided a first explanation of their evolution over the last decade, and what it could suggest for the financial system and the regulator in order to increase both access and use of financial products and services. As a first step, we have provided broad indicators of access and use, and we have estimated a first group of indicators of geographical inequality in financial inclusion. Finally, we have merged some of these indicators with household data in order to evaluate its microeconomic relevance. Results suggest at households are reacting to the greater credit availability by taking up more credits. The greater credit availability has also different effects over household population. It helps them to better cope with shocks, fosters business registration and increases business profits. Finally we showed financial deepening is also related to a more stable consumption and income stream for the households. We believe that this data, aggregated over time, will help financial institutions, policymakers and researchers alike to understand the problem of access to financial products and services, propose adequate measures to enhance financial inclusion and evaluate their impact, both in the short and in the long term. Financial institutions have been doing an effort to increase their network reach, especially in later years, with agents being one of their most potent tools in doing so. While it is true than their network is still very small compared with other countries, these efforts have shown some results, both in number of clients and in the source of those clients. We have also shown that presence of financial institutions not based in Lima can foster credit opportunities and financing for families and small businesses in provinces, which have been happening over the last years since municipal and rural savings and loans institutions were granted permission to operate in Lima in 2003; however, it has been also shown that while more funding is available for people living outside Lima, there is not a deposit-creation process of the same level going on, despite the existence of financial institution branches (often named as one of the most important drivers of deposit creation), which could mean that people outside Lima face other barriers – besides lack of physical access – to become financially included, at least from the savings side.
Despite this spatial inequality, financial deepening experienced during the last years has some positive outcomes. Matching household survey data with aggregate data on financial deepening provides a useful way to investigate the microeconomic implications of greater credit availability. In this paper we have shown that credit take up increases for households located in regions where credit has become more available in the form of more agencies. But not only the likelihood of taking a loan has increased, but the amount the household takes also is greater for households located in regions with more financial deepening. Rural households are most responsive in terms of taking credits from commercial banks. We also evaluated household’s capability to absorb shocks and we found households located in regions with greater financial deepening are better prepared to endure shocks: they are less likely to use their own savings and reduce consumption in the event of an income shock. However, when asked whether the household took a loan to cope with the shock, our results show a decrease in the likelihood of doing so. However counterintuitive this result, it might be a consequence of error measurement in the “took a loan” variable to cope with the shock. There is no specification of the type of loan, thus it might also be an informal loan. Thus the effect of financial deepening is less clear in this respect. Business behavior was also assessed, and we found independent small businesses being more likely to formally register when credit becomes more available. This might be a consequence of formal requirements for accessing formal credit. This result is stronger for rural businesses. Also, there is a significant increase in profits reported by independent businesses operating in regions with greater credit availability. This result is strong for all samples. Finally, since one of the most important features of financial development is that it allows people to smooth their consumption and income, we developed a simple framework to test this. After estimating temporary components of consumption and income per capita, we find that higher financial deepening is related to a significantly smaller volatility of temporary income and consumption at household level.
References Aguilar, G. (2011): “Microcrédito y Crecimiento Regional en el Perú," PUCP Documento de Trabajo, Departamento de Economía, 317. Banerjee, A., E. Duflo, R. Glennerster, y C. KinnanThe (2010). “The Miracle of Microfinance?: Evidence from a randomized evaluation” MIT, Departamento de Economía. http://econ-www.mit.edu/files/5993 CGAP (2009), “Financial Access 2009: Measuring Access to Financial Services around the World” The World Bank. Deaton, A. (1998) “Economics and Consumer Behaior”. Cambridge University Press. Kendall, J., N. Mylenko, and A. Ponce (2010) “Measuring Financial Access around the World”. The World Bank, Policy Research Working Paper Series, 5253. Morón, E., E. Salgado y C. Seminario (2012) “Financial Dependence, Formal Credit and Firm Informality: Evidence from Peruvian Household Data” Inter American Development Bank Working Paper Series, 288. Pages (2010) “The Age of Productivity”. Inter American Development Bank. Prialé, G., L. Allaín y R. Mazer (2012) “Financial Inclusion Indicators for Developing Countries: The Peruvian Case” Thorsten Beck, Asli Demirguc-Kunt and Maria Soledad Martinez Peria (2006) “Reaching out: Access to and use of banking services across countries”. World Bank. Wooldridge (2002) “Econometric Analysis of Cross Section and Panel Data” MIT Press.
Table 1. Credit take up
Requested a loan for housing repairs All Urban Rural 0.006 0.007 0.004 (0.006) (0.010) (0.007) 0.003 (0.004) 0.003 (0.006) 0.003 (0.004)
Requested a loan for housing repairs - Banco de Materiales All Urban Rural -0.001 -0.000 -0.001 (0.002) (0.003) (0.001) -0.001 (0.001) 21455 -0.001 (0.002) 12669 -0.001 (0.001) 8786
Requested a loan for housing repairs - Commercial Bank All Urban Rural 0.009** 0.008 0.012*** (0.004) (0.007) (0.004) 0.006** (0.003) 21455 0.006 (0.005) 12669 0.008*** (0.002) 8786
Requested a loan for housing repairs - Amount (S/.) All Urban Rural 25.613 38.026 18.150 (69.647) (107.173) (69.735) 27.698 (45.420) 21455 23.404 (71.805) 21455 42.942 (43.822) 21455
Household has bought any financial service All Urban Rural 0.014** 0.026** -0.003 (0.007) (0.011) (0.004) 0.005 (0.004) 21455 0.011 (0.007) 21455 -0.003 (0.002) 21455
Obs 21455 12669 8786 Own estimations using data from ENAHO and SBS.
Table 2. Shock Reactions.
Suffered a shock and had to use savings All Urban Rural 0.002 -0.005 0.012 (0.008) (0.011) (0.013) -0.001 (0.005) -0.001 (0.007) -0.001 (0.008) Suffered a shock and had a loan All -0.012 (0.009) 0.000 (0.006) 21455 Urban -0.022* (0.012) -0.006 (0.008) 12669 Rural 0.004 (0.011) 0.010 (0.007) 8786 Suffered a shock and had to reduce food consumption All -0.000 (0.009) -0.005 (0.006) 21455 Urban -0.009 (0.009) -0.005 (0.006) 12669 Rural 0.012 (0.016) -0.005 (0.010) 8786
Obs 21455 12669 8786 Own estimations using data from ENAHO and SBS.
Table 3. Business
Registered household business All 0.009 (0.007) 0.015*** (0.005) Urban 0.009 (0.012) 0.025*** (0.008) Rural 0.009* (0.005) 0.003 (0.003) Started a new business in the household All 0.003 (0.005) 0.003 (0.003) 21455 Urban 0.006 (0.007) 0.001 (0.004) 12669 Rural -0.002 (0.006) 0.006 (0.004) 8786 Business profits (S/.) All 80.384*** (24.451) 56.345*** (15.944) 21455 Urban 107.116*** (38.328) 79.770*** (25.676) 12669 Rural 46.748** (21.953) 30.935** (13.796) 8786
Obs 21455 12669 8786 Own estimations using data from ENAHO and SBS.
Table 4. Smoothing
Standard deviaton of temporary consumption All Bank -0.035*** (0.011) Cajas 0.030*** (0.009) Urban -0.018 (0.014) 0.027** (0.011) Rural -0.031* (0.018) 0.017 (0.015) Standard deviaton of temporary income All -0.017 (0.017) 0.014 (0.013) 4429 Urban 0.009 (0.023) 0.014 (0.017) 2507 Rural -0.018 (0.024) -0.007 (0.020) 1922
Obs 4429 2507 1922 Own estimations using data from ENAHO and SBS.