The Basic Analytics of Access to Financial Services

BY THORSTEN BECK AND AUGUSTO DE LA TORRE

Access to financial services, or rather the lack thereof, is often indiscriminately decried
as problem in many developing countries. This paper argues that the “problem of access”
should rather be analyzed by identifying different demand and supply constraints. We use
the concept of an access possibilities frontier, drawn for a given set of state variables, to
distinguish between cases where a financial system settles below the constrained optimum,
cases where this constrained optimum is too low, and—in credit services—cases where
the observed outcome is excessively high. We distinguish between payment and savings
services and fixed intermediation costs, on the one hand, and lending services and different
sources of credit risk, on the other hand. We include both supply and demand side frictions
that can lead to lower access. The analysis helps identify bankable and banked population,
the binding constraint to close the gap between the two, and policies to prudently expand the
bankable population. This new conceptual framework can inform the debate on adequate
policies to expand access to financial services and can serve as basis for an informed
measurement of access.

I. INTRODUCTION
Access to financial services or outreach of the financial system has become a major
concern for many policymakers in developing countries. While the use of financial
services—measured as having deposit accounts with banks—reaches over 90% in
most high-income countries, in many low- and even middle-income countries the
use of formal financial services is still restricted to a small number of firms and
households (Peachey and Roe, 2004; Beck, Demirguc-Kunt, and Martinez Peria,
2007; Honohan, 2006). Moreover, the intense financial sector reforms undertaken
by many emerging economies over the past decade—doing away with interest
rate controls and directed credit, liberalizing entry and privatizing state-owned
banks—have not led to the type of broadening of access to financial services that
was initially expected, particularly for lower-income households and small and
medium-size enterprises (SMEs).
Broad access to financial services is related to the economic and social devel-
opment agenda for at least two reasons. First, a large theoretical and empirical
literature has shown the importance of a well developed financial system for eco-
nomic development and poverty alleviation (Beck, Levine, and Loayza, 2000;
Beck, Demirguc-Kunt, and Levine, 2007; Honohan, 2004a). To be sure, while a
causal link running from financial depth to growth has been rather convincingly
established by empirical research, the search for causality between the breadth of
access and growth is still on. However, as noted by De la Torre and Schmukler
(2006a), the discussion of the plausible channels through which financial depth
C 2007 The Authors. Journal compilation
c 2007 New York University Salomon Center, Financial Markets, Institu-
tions & Instruments, V. 16, No. 2, May. Published by Blackwell Publishing, Inc., 350 Main St., Malden, MA 02148,
USA, and 9600 Garsington Road, Oxford OX4 2DQ, UK.
80 Thorsten Beck and Augusto de la Torre

could cause economic growth often resorts to access-related stories. 1 Prominent

in this regard is the Schumpeterian view that finance leads to growth because it
fuels “creative destruction” by allocating resources to efficient newcomers. That
is, through broader access to external funds, talented newcomers are empow-
ered and freed from the disadvantages that would otherwise arise from their lack
of inherited wealth and absence of connections to the network of well-off in-
cumbents (Rajan and Zingales, 2003). Second, access to financial services can
be seen as a public good that is essential to enable participation in the bene-
fits of a modern, market-based economy, in an analogous way as is the access
to safe water, basic health services, and primary education (Peachey and Roe,
2004).
A low level of observed use of financial services, however, has to be care-
fully distinguished from a problem of access. In a purely theoretical—and rather
uninteresting—world characterized by the absence of transaction costs, uncer-
tainty, and asymmetric information there is no “problem” of access. Decisions to
accumulate savings, take out loans, and make payments would be equally open
to all and the implementation costless. Banks would not be needed to mobilize
savings, facilitate payments, and allocate loans, as savers would assign their sav-
ings directly to borrowers based on perfect knowledge of investment possibilities.
Hedging or insurance products would not be required given the absence of un-
certainty. Access to external finance would be frictionless, limited only by the
inter-temporal wealth constraint of the borrower, which would be known equally
well and with certainty by both the lender (saver) and the borrower (investor). In-
vestment decisions would be independent of financing and consumption decisions.
The choice between borrowing and lending (saving) would be determined purely
by inter-temporal preferences and investment opportunities, and changes in bor-
rowing and lending would reflect changes in demand and investment opportunities
rather than changes in the possibility of access. In this ideal world, the lack of use
of finance by some agents would certainly not be a “problem” in the commonly
used sense of the word. Agents that do not borrow for consumption would be those
that do not need to smooth their consumption over time subject to their life-time
wealth. And projects that do not borrow for investment would be those that do not
meet the condition of a positive real net present value.
Problems of access do arise in some well-defined sense, however, in the real
world and are essentially linked to such crucial real-world facts as transaction costs,
uncertainty about project outcomes, and information asymmetries. These introduce
frictions that can limit access to financial services and that can make it difficult
to de-couple investment from financing decisions in most cases. In a world with
frictions, investment decisions may reflect credit supply constraints, and not just
preferences and business opportunities. It is precisely these frictions which give

1
For micro-evidence, see Gine and Klonner (2005) on the degree to which access to finance determined
the switch to more efficient and thus profitable fisher boats in South India.
 The Basic Analytics of Access to Financial Services 81

rise to organized financial markets, financial institutions, and broader contractual

entities. 2 However, the efficiency with which financial markets and institutions
overcome these frictions depends on the macroeconomic environment, market
structure, and overall contractual and informational environment (Beck, 2006).
Across countries and over time, we can observe a large variation in the efficiency
with which financial markets and institutions are able to overcome market frictions
and provide financial services.
To say that problems of access to financial services arise due to transactions
costs, information asymmetries, and uncertainty does not entail, however, that
an access “problem” is always easy to identify. On the demand side, economic
agents, households and enterprises alike, might have no impediment to access
financial services, but may simply not want to use them. It would be wrong to
argue that voluntary self-exclusion constitutes a “problem of access,” except in
the cases where self-exclusion reflects unduly low levels of financial literacy or
is a psychological response to past systematic discrimination. On the supply side,
creditors that face large macroeconomic risks and/or major difficulties in mitigating
problems of adverse selection, moral hazard, and contract enforcement may decide
to deny loans to certain borrowers. Doing so would be a matter of prudence in the
use of depositors’ funds. Again, whether this situation constitutes a “problem of
access” is debatable and the opposite is easier to argue—that it would be a serious
“problem” if creditors made loans to certain borrowers under such circumstances,
as many a banking crisis illustrates. The key point is that, once the existing market
frictions in an economy are taken into account, the observed lack of use and
outreach of financial services might be the rational and prudent outcome. But,
should such an outcome deserve the label of a “problem”? And if so, in what
sense?
Traditionally, access problems have been defined by reference to some form
of observable limitation that leads to a contrast between the active use of a given
financial service (say, a loan) by a certain group, on the one hand, and the low use (or
lack of use) of that service by another group, on the other hand. Thus, we talk about
geographic limitations—reflected, for instance, in the absence of bank branches
or delivery points in remote and sparsely populated rural areas that are costlier to
service. 3 We also talk about socio-economic limitations—when financial services
appear inaccessible to specific income, social or ethnic groups either because
2
Transactions costs, agency problems, and uncertainty are key reasons why institutions and organiza-
tions exist (North 1990). In a world with financial market frictions, basic financial services are typically
categorized into savings, loan, insurance, and payment services. By offering payment services, financial
institutions and markets reduce transaction costs in the exchange of goods and services between people
and over time. By offering savings and loan services financial institutions and markets allow firms
and households to overcome frictions that prevent them from de-linking consumption from invest-
ment decisions as discussed above. By offering insurance mechanisms, financial intermediaries allow
households and firms to hedge and diversify risks and smooth consumption. Compare the overview in
Levine (1997, 2005).
3
Beck, Demirguc-Kunt, and Martinez Peria (2007) find a positive cross-country association of geo-
graphic branch and ATM penetration with population density and physical infrastructure.
82 Thorsten Beck and Augusto de la Torre

of high costs, rationing, financial illiteracy, or discrimination. And we also talk

about limitations of opportunity—where, for instance, talented newcomers with
profitable projects are denied finance because they lack fixed collateral or are not
well connected. 4
While limitations to access along these three dimensions are due to market
frictions, the observed outcome could be a constrained optimum—the result of
rational agents maximizing their utility and profit functions given the constraints
imposed by the existing market frictions. In what cases and in what sense, then, can
we say that a low level and unevenly distributed access constitute a “problem”? Our
approach in this paper is to define the “problem of access” in terms of an “access
possibilities frontier,” which is drawn for a given set of “state variables.” In our
framework, thus, an access problem is defined in three different ways: (i) when the
economy settles at a point below the access possibilities frontier, given the state
variables; (ii) when the possibilities frontier is too low relative to countries with
comparable levels of economic development; and (iii) when imprudent lending
practices lead to an excessive credit expansion beyond the constrained optimum.
To simplify the discussion we will bundle the market frictions that are relevant
to the supply of financial services into two groups: (i) transaction costs and the re-
sulting scale economies of financial services at the level of the user, the institution,
and the market, and (ii) systemic and idiosyncratic risks. On the demand side, we
will differentiate between economic and non-economic factors that may lead to
self-exclusion. While this is clearly a major simplification of the access problem,
it will help us derive an analytical tool to better discuss access issues and relevant
policies. In section II, we analyze access to simple payment and savings services.
In section III, we examine access to lending services. Section IV concludes.

II. THE ANALYTICS OF ACCESS TO PAYMENT

AND SAVINGS SERVICES
This section discusses supply and demand factors for access to payments and
savings services. While there is a considerable diversity of payments and savings
services, even in relatively underdeveloped financial systems, our analysis focuses
on the demand and supply of the most plain-vanilla version: a payment service
based on a simple checking account and instrumented through either a check or a
debit card, and a saving service consisting of a passbook savings account that pays
a zero real interest rate and is redeemable at par and on demand. 5 We, therefore,
emphasize in this section the transactional and custodial functions of these services,
4
In an alternative classification Honohan (2004b) distinguishes between price barrier (a financial
service is available but too expensive), information barrier (a firm’s or household’s credit worthiness
cannot be established) and product and service barrier (services most needed by certain groups are not
offered). Beck, Demirguc-Kunt, and Martinez Peria (2006) document the large cross-country variation
in barriers to banking such as minimum balances, documentation required to open an account and fees
for different financial services
5
We assume, therefore, that checking and savings accounts pay a nominal interest rate that is equal to
the local inflation rate.
 The Basic Analytics of Access to Financial Services 83

respectively, rather than their interest-earning dimension. We further assume that

deposits in checking and passbook savings accounts are invested by banks in risk-
free securities (a narrow bank scenario). Hence, for the purposes of our analysis,
the price of payments and savings services is given by a fee, the intermediation
margin earned by banks on the corresponding accounts is negligible, and risk
considerations are of no relevance. 6 These simplifying assumptions—which imply
no loss of generality for the analysis of issues in access to this type of services—
allow us to focus on costs as the driver behind the supply of payments and savings
services. Although we consider only plain-vanilla fee-based services, we do make
a distinction that matters for access, as discussed below—the distinction between
the production of high-value payments and savings services, on the one hand, and
low-value/high-volume services, on the other.

FIXED TRANSACTION COSTS

Fixed transaction costs in financial service provision result in decreasing unit
costs as the number or size of transactions increase. These fixed costs exist at
the transaction, client, institution, and even financial system level. Processing
an individual payment or savings transaction entails costs that are, at least in
part, independent of the value of the transaction. Maintaining an account for an
individual client also implies costs that are largely independent of the number
and size of transactions the client makes. At the level of a financial institution,
fixed costs are crucial and span across a wide range—from the brick-and-mortar
branch network, to computer systems, to legal services, to accounting systems,
and to security arrangements—and are rather independent of the number of clients
served or the number of transactions processed. Fixed costs also arise at the level
of the financial system, including in terms of regulatory costs and the costs of
payment clearing and settlement infrastructure, which are again, and up to a point,
independent of the number of institutions regulated or participating in the payment
system. The resulting economies of scale at all levels make it unprofitable to stay
in the business of payment and savings service provision unless the associated
scale economies are captured in some form. 7
The effect of fixed costs on financial service provision can be reinforced by
network externalities, where the marginal benefit to an additional customer is
determined by the number of customers already using the service (Claessens et al.,
2003). This is especially relevant for payment systems, where benefits and thus
6
To be sure, even in the narrow-bank scenario assumed here, there are some forms of risk, notably
operational risk. Clearly, risks (credit, liquidity, price, etc.) come into the center of the stage as we
depart from the narrow-bank assumption and banks invest the sight deposits into loans and other risky
assets. We introduce credit risk as a core element of the analysis of access to lending services in the
next section.
7
While the literature has failed to find evidence for scale economies after a certain threshold, there
seems to be evidence for scale economies for small banks (Berger and Humphrey, 1994). Also, in cross-
country, cross-bank comparisons, smaller banks show higher operating cost to assets ratios (Demirguc-
Kunt, Laeven, and Levine, 2004).
84 Thorsten Beck and Augusto de la Torre

demand increases as the pool of users expands. High fixed costs can trap a small
financial system at a low level equilibrium because of the system’s inability to reap
the necessary scale economies and network externalities.
In sum, fixed costs can constitute an important limitation to outreach in the pro-
vision of payment and savings services and, hence, a key barrier to the broadening
of access to these services. Profitable and sustainable financial intermediaries have
to exploit scale economies either through sufficiently high-volume or high-value
transactions, but not necessarily through both. In particular, the competitive en-
vironment could be such that deposit-taking institutions could stay in business
without much outreach efforts, by specializing in large-value payments and sav-
ings services.
Because of scale economies and network externalities, problems of access to
payments and savings services in many developing countries are related to the oft-
found triple problem of smallness—small transactions, small financial institutions,
and small market size. Unless a way is found to raise transaction volumes to seize
scale economies, low-income clients with the need for small and few payment
and savings transactions would not constitute a profitable clientele for financial
service providers. A small, community-based financial institution might not be
able to spread the fixed costs of its physical installations, technological platform,
legal and accounting infrastructure, etc. over a sufficiently large business volume
to be profitable. Similarly, a bank might not be willing to open a branch in smaller
towns if there is not sufficient market potential to cover its fixed costs of setting up
shop. Finally, a small market might be able to support only a few viable financial
institutions (i.e., institutions that reach an efficient scale), thereby having to forego
the type of competition that—as we argue below—appears crucial to foster the
broadening of access.
It is useful to think of the process of supply-originated broadening of access
to payments and savings services as driven either by changes in “state variables”
or by idiosyncratic cost management for a given level of state variables. 8 While
idiosyncratic cost management refers to actions within the realm of individual
financial institutions, we treat as state variables those that are largely outside the
control of the managers of financial intermediaries and that change slowly over
relatively long time periods, including the following: market size, macroeconomic
fundamentals, available technology, the average level and distribution of per capita
income, and system-wide costs of doing business related, for instance, to the quality
of transport and communication infrastructure, the effectiveness of the contractual
and informational frameworks, and the degree of general insecurity associated
with crime, violence, terrorism etc. We argue that the intensity of competition can
vary for a given level of state variables, but that competition is also a key driver
behind changes in state variables over time. An oligopolistic market structure, for
example, can result in dominant banks successfully opposing any improvement

8
It is of course somewhat arbitrary to define the set of state variables, not least because the definition
is not independent of the relevant time horizon.
 The Basic Analytics of Access to Financial Services 85

Typical
value

Number of clients

Figure 1: Scale economies in payment and savings services

in the contractual and informational frameworks as this would reduce their

rents.
The interplay between state variables and idiosyncratic cost management in
determining supply outreach for payments and savings services is illustrated in
Figure 1. It shows, via iso-profit curves, the combinations of transactions value
and transactions number of payment and savings services that yield the same
profit for a given financial institution. The horizontal axis measures the number of
transactions which we assume, for ease of presentation, to increase only with the
number of clients—i.e., we hold the number of transactions per client constant.
The vertical access measures the “typical” value (or the mode) of the payment or
savings transaction handled by the institution. 9 The iso-profit curves are drawn for
a given level of state variables and are downward sloping because an increase in the
number of transactions (clients) is needed for a financial entity to make the same
profit while moving towards a lower typical transaction value. The curvature of
the iso-profit curves depends on the characteristics of the production function for
payments and savings services. The iso-profit curves are parallel, with curves more
distant to the origin representing higher profits. A movement along an iso-profit
curve towards a larger number of clients represents, by definition, a supply-induced
broadening of access.
It might be not that easy, however, for financial institutions to trade-off smoothly
(beyond a certain range) a higher transaction value for a higher number of clients
while still making the same profit. Discontinuities along this trade-off could arise
if the supply of large-value payments and savings services is not generated by
the same production function as the supply of low-value/high-volume services,
and changes from one production function to another imply switching costs.
9
This is equivalent to assuming that the vertical axis measures the average value of the transactions
of a client and that such value increases only with the value of each individual transaction, so that for
ease of argument the number of transactions per client is held constant.
86 Thorsten Beck and Augusto de la Torre

Technology, marketing and costumer service approach, location, size, configu-

ration of bank offices, and even the profile of staff are all a function of the type
of service produced and switching to a different technology, marketing, etc. to
reach a new clientele is costly. The case of two production functions is depicted in
Figure 1, where financial institutions cannot move from point A to point B without
incurring switching costs, even if both yield in the end the same profit. This, in
turn, might be a barrier to move down-market to expand access.
Discontinuities in the iso-profit line explain not only why financial institutions
cannot move freely along a given curve beyond a certain range of values. They
also can also help explain why financial systems in many emerging economies
show a clustering of financial institutions at the North-West corner of the space
in Figure 1, with few if any institutions at the South-East corner. That type of
clustering reflects a situation where banks cater mainly to large value clients and
are unmotivated to incur the costs to change their production function to move
down-market to a point like B. It also reflects a situation where newcomer banks
do not enter, for one reason or another, business niches located in the South-East
corner of Figure 1.
Why would that clustering equilibrium arise even if the production technol-
ogy is available to make point B feasible and equally profitable as point A (as
is suggested in Figure 1)? One obvious explanation would be that profits in that
type of banking system are not under strong competitive pressures. Due to insuf-
ficient market contestability, payment and savings service providers could make
money comfortably by targeting larger firms and wealthier households and would
have little incentive to manage their costs efficiently to reach out to new clients
by, say, re-engineering their internal processes and services and optimizing on
the use of available information technology. 10 Insufficient competitive pressures
may also be associated with barriers that deter newcomer service providers—
who by definition do not incur switching costs—from entering the business of
payments and savings service provision to target the smaller firms and poorer
households.
The clustering of payments and savings service provision in the North-West
corner of Figure 1 could also occur—even if competition was strong—because of
low levels in certain state variables. For instance, a small market size and/or lack of
technology to reduce costs sufficiently could hinder supply outreach. In this case,
the iso-profit curves would not extend into the South East corner of the graph.
Even in the absence of discontinuities, the supply-driven broadening of access
to payments and savings services could conceivably encounter difficulties if iso-
profit curves were “too flat.” Relatively flat curves would mean that to remain at
least equally profitable a financial institution would have to increase the number
of transactions (clients) very substantially for each small decrease in the typical

10
Given the fixed cost character of the switching costs, bigger banks are in a better position to incur
these costs than small banks.
 The Basic Analytics of Access to Financial Services 87

value of the transactions it handles. 11 If iso-profit curves were flat, increased mar-
ket contestability would be insufficient to promote supply-induced broadening
of access, as that flatness would be a reflection of state variables. A significant
growth in market size, a technological breakthrough in information and commu-
nications technology, 12 a noticeable improvement in road or telecommunication
infrastructure, or a palpable reduction in general insecurity would be required
to create greater curvature in iso-profit lines, thereby facilitating supply-induced
broadening of access.
While Figure 1 illustrates different supply constraints to access, we have to take
into account demand side constraints. Further, the analysis up to now has focused
on individual institutions, while ultimately we are only interested in institutions that
work on the margin, i.e., with the most marginal customers. Therefore, we now turn
to a discussion of demand-side constraints to outreach before combing aggregate
demand and supply side analysis to derive the access possibilities frontier.

ECONOMIC AND NON-ECONOMIC DEMAND FACTORS

Price and income level are the salient economic determinants of the demand for
payments and savings services. 13 Economic development and the associated rise
in per capita income increases the need for more and more sophisticated versions
of these services. However, demand is not only driven by economic but also by
socio-cultural factors. Further, we have to isolate pure demand factors from demand
reductions that are due to the expectation of supply constraints. 14
In the following, we will distinguish between two demand curves—a potential
demand curve, driven purely by economic factors, and an actual demand curve,
that might be below the potential one due to non-economic factors. We can write
the potential (individual or aggregate) demand as D∗ = f{income, price}, with
demand increasing in the first argument and decreasing in the second. The actual
demand can be lower than potential demand for a given price and income level, due
to self-exclusion arising from such non-economic reasons as financial illiteracy
and ethnic or religious factors. 15 Taking into account these non-economic demand
11
Whether flat iso-profit curves are not just a theoretical possibility but also a phenomenon of empirical
relevance is a question beyond the scope of this paper. To our knowledge there is no study that
investigates this empirical issue.
12
The IT revolution has allowed new and cheaper forms of delivering and processing financial services,
including via the phone and internet. See Claessens, Glaessner, and Klingebiel (2002) for a discussion
of the potential role of e-finance in emerging markets.
13
Peachey and Roe (2004) show that the increase in the share of population with bank accounts across
OECD countries over the past 50 years can be attributed to the entry of women into the work force and
the associated higher disposable household incomes.
14
This would include lack of demand due to expected discrimination-based rejection.
15
Financial illiteracy might lead to lower demand for services than predicted by economic capacity
to pay as people might not know of the existence or affordability of specific payments or saving
services, partly due to a generally low level of education. See Bertrand et al. (2005) for an example
of how marketing can increase the take-up of financial products without changing the price or product
features.
88 Thorsten Beck and Augusto de la Torre

Price

D*
S

D IV
I
III
II
S*

D C B A Population

Figure 2: Access Possibilities Frontier for payment and savings services

factors, we can write actual demand (individual or aggregate) as: D= f{income,

price, financial illiteracy, cultural barriers}, with demand decreasing in the third
and fourth arguments. Abstracting from the theoretical possibility that illiteracy
of cultural biases might lead to over-demand of financial services, actual demand
will be bounded by potential demand.

THE ACCESS POSSIBILITIES FRONTIER FOR PAYMENT AND SAVING SERVICES

Aggregate supply and demand determine the bankable population, i.e., the share
of population a financial system can reach with payment and savings services. We
define the Access Possibilities Frontier for payment and saving services as the
maximum share of population that could be served by financial institutions, for
a given set of state variables. Figure 2 combines the demand and supply analysis
of the previous two sections into the traditional price-quantity graph. It assumes
away aggregation problems; specifically while there may be different production
functions in the supply of high-value services versus low-value/high-volume ser-
vices at the level of individual financial institutions, Figure 2 assumes that we can
speak cogently of only one type of payments (or savings) service at the aggregate
level, i.e., that the mentioned differences in production do not prevent a smooth
blending of individual supply curves as they are aggregated into the overall supply
schedule. 16
In line with the assumptions discussed at the beginning of this section, the price
measured along the vertical axis is a fee, defined on a per-transaction basis. We
assume further that the fee is flat, i.e., independent of the size of the transaction.
On the horizontal axis, rather than the quantity of service transactions, we plot the
16
Alternatively, we could have assumed discontinuities, which could result in multiple equilibria. For
the ease of discussion, we exclude this possibility in our analysis.
 The Basic Analytics of Access to Financial Services 89

share of the population (households and firms) engaging in payments and savings
transactions. We order the population share along the axis starting with agents that
engage in transactions that are large in value and number and moving towards
agents with transactions of increasingly lower value and number. To simplify
this ordering, we make two further assumptions. First, that clients with a high
transactions volume are also characterized by transactions of high value. Second,
that while the volume of transactions per period and value of each transaction vary
across agents, each agent engages in a fixed number of transactions of equal value
per period, independently of the fee. 17 These assumptions imply that movements
of the equilibrium outcome towards the right of the graph unequivocally represent
a broadening of access to payments and savings services, as marginal customers
that demand low-value/low-volume transactions enter the financial system, raising
the share of the banked population.
Let us first discuss the potential and actual demand curves for payments and sav-
ings services as function of its price and for a given set of state variables (which, as
noted earlier, are beyond the influence of individual service providers and include
market size, macroeconomic fundamentals, the average level and distribution of
per capita income, available technology, physical infrastructure, the contractual
environment, and the degree of general insecurity). The demand reflects willing-
ness to pay and we expect customers with demand for large-value/high-volume
transactions to have a higher marginal willingness to pay than customers with few
and small transactions. A higher price thus reduces the share of the population that
will demand payment and savings services, thus resulting in a downward sloping
demand curve. As the price falls, marginal users (i.e., smaller firms and poorer
households) add their demands for typically lower-value transactions. Curve D∗
denotes the potential demand as function of economic factors only, while D denotes
the actual demand, as predicted by both economic and non-economic factors, in-
cluding self-exclusion. The distance between the two curves denotes self-exclusion
and is not necessarily constant across the population spectrum. However, if one
conjectures that self-exclusion and illiteracy are inversely correlated with income,
the gap between potential and actual demand opens up as marginal households
and firms are added to the demand, as depicted in Figure 2.
We now turn to supply. As the fee increases, it becomes profitable for service
suppliers to reach a larger share of the population, by targeting agents (households
and firms) characterized by increasingly lower-value transactions. This results in an
upward sloping supply curve. The higher steepness of the supply curve in the outer
regions could be explained by either switching costs or flat iso-profit curves, as
discussed above. The curve S∗ denotes the potential supply, representing the share
of population that can be serviced for a given price by efficient financial service
providers, given the state variables. Note that the S∗ curve assumes efficiency in
supply, i.e., that service providers maximize profits for a given price and are, thus,
17
This implies that when the fee falls, individuals do not react by demanding more transactions, but
rather more individuals will demand. Or, in the terminology of the trade literature, changes in the fees
have effects only on the extensive not on the intensive margin.
90 Thorsten Beck and Augusto de la Torre

located on the highest feasible iso-profit curve. The potential supply, S∗ , differs
from the actually observed supply curve, S, which denotes the share of population
that currently existing financial service providers are actually willing to cater to
for a given price. As illustrated in Figure 2, S is to the left of S∗ , implying that
the former is a higher cost supply due to some form of inefficiency, including
distortionary regulatory policies, an excessive number of banks for the size of the
market (leading to a failure to achieve economies of scale), or lack of adequate
market contestability.
We are now ready to analyze various key points in Figure 2. Point I, the in-
tersection of potential demand D∗ and potential supply S∗ schedules, denotes the
access possibilities frontier, i.e., the maximum outreach point for payments and
savings services that can be reached in a country’s financial system, given the state
variables. Point I is, thus, the constrained optimum, associated with a bankable
population at point A on the horizontal axis. Hence, 1-A is the non-bankable pop-
ulation, that is, the share of population that cannot be provided with market-based
payments and savings services, given the state variables.
The access possibilities frontier can be moved by shifts (or changes in curvature)
in the potential supply and the potential demand curve as a result of changes in
state variables. An outward shift of the potential demand curve due, for instance,
to a higher level of per capita income increases access, as measured by the equi-
librium share of bankable population. Similarly, a downward shift in the potential
supply curve due to improvements in a country’s infrastructure or institutional
environment will result in an expansion of the bankable population.
Starting from the Access Possibilities Frontier and the bankable population
(Points I and A), we can identify three different types of access problems. First
is the access problem implied in point II, which denotes a lower than potential
possibilities frontier as a result of non-economic factors that lead to self-exclusion
of agents in the demand schedule. In this case, the equilibrium level of banked
population (B) is lower than the bankable potential, given state variables. The A-B
distance can be interpreted as a measure of a demand-driven access problem.
A second type of access problem is illustrated by points III or IV. Both are
points of constrained sub-optimality, i.e., an inefficient (high cost) supply leads to
an equilibrium where the banked population is lower than the bankable population,
given the state variables. The access problem is worse at point IV which, in addition
to supply inefficiency, reflects also self-exclusion. The distances A − C and B −
D are measures of access problems driven purely by supply inefficiency, as they
hold the type of demand (potential or actual, respectively) constant. The distance
A − D is a measure of an access problem caused by a combination of demand
self-exclusion (A- B) and supply inefficiency (B − D).
The third type of access problem would obtain if the bankable population as-
sociated with point I is “too low” relative to countries with comparable levels of
economic development. This situation could arise, for example, if the country in
question lags behind its comparators in certain state variables (say, higher level of
general insecurity or weaker informational and contractual environments).
 The Basic Analytics of Access to Financial Services 91

POLICIES TO FOSTER ACCESS TO SAVINGS AND PAYMENTS SERVICES

In order to design effective policies to expand access to payments and savings
services, policymakers have to be aware which of the three access problems is
the most binding one and/or can be addressed in the most effective way. More
specifically, policy makers should try to understand if the access problem is mainly
because their financial system is inside the possibilities frontier, and why, or mainly
because the achievable frontier is too low compared to similar countries and why.
The latter problem will generally call for market-developing policies that expand
the possibilities of outreach through structural reforms that improve institutions
and other state variables, thus moving the frontier outwards over time. The former
problem rather calls for market-enabling policies that, depending on the underlying
reason, would put emphasis on raising financial literacy to mitigate self-exclusion
or providing incentives to financial service providers to operate more efficiently. 18
Consider first market-enabling policies. While the IT and communication rev-
olution allows financial institutions to improve cost management, the motivation
to seize such opportunities typically comes from the pressures of competition. In
effect, competitive pressure and the search for profits are key factors behind such
examples of outreach as offering of services that are tailor-made for low-income
clients (e.g., simple debit accounts at lower costs than regular checking accounts)
or the use of mobile branches or cell phone banking to reach populations in remote
areas at low costs. 19 Policy makers have an important role to play in providing
financial institutions with the necessary incentives to expand outreach through
different regulatory policies and actions. 20
If the lack of innovation and the corresponding increase in outreach is due to
imperfect competition, policy responses should aim at enhancing market contesta-
bility, so that service providers—existing or new—are under competitive pressures
to achieve cost minimization through efficiency gains, holding the state variables
constant. Allowing foreign bank entry or avoiding overly high licensing and mini-
mum initial capital requirements can foster contestability and more specifically the
emergence of financial institutions catering to the needs of low-value customers. 21
Preventing the establishment of closed clubs (such as payment networks limited
to large banks) can keep a system contestable and thus competitive. 22 Looking be-
yond the commercial banking system and allowing competition from the non-bank
financial sector can be important. For instance, preventing payments infrastructures
from becoming inaccessible to financial institutions outside a restricted “club” and
18
Compare a similar classification in Porteous (2004).
19
For a discussion of examples from Sub-Saharan Africa, see World Bank (2007).
20
As example, see Porteous (2006) for a discussion on the regulatory framework for cell phone banking.
21
The deregulation wave in OECD over the past 30 years and the consequent increased competition led
many financial service providers to look for new markets, effectively expanding the banked population
(Peachey and Roe, 2004).
22
For cross-country evidence on the link between regulatory policies and firms’ access to finance,
see Beck, Demirguc-Kunt, and Maksimovic (2004). For cross-country evidence on the link between
market structure, regulatory policies and the competitiveness of a financial system, see Claessens and
Laeven (2004).
92 Thorsten Beck and Augusto de la Torre

ensuring that retail payments infrastructures are open and interoperable can help
the financial system cater to marginal customers at low cost. If inefficiency re-
flects instead the inability to capture economies of scale because, say, the number
of banks is too large relative to the size of the market, policy responses should
first aim at the consolidation of the banking system, through the exit of unviable
institutions via orderly closures and mergers.
Regulatory policies can have a profound impact on financial institutions’ costs
and efforts to increase outreach, even if they do not target directly the competi-
tiveness of a financial system. On the negative side, high compliance costs with
unduly complicated “Know Your Customer” (KYC) and anti-money laundering
(AML) regulations may prevent financial institutions from moving to the frontier
(or prevent the establishment of institutions at the frontier) to service marginal
customers (Claessens, 2006). On the positive side, relaxing branching restrictions
by allowing financial institutions to offer limited services through non-financial
correspondents can significantly reduce the fixed cost element of financial service
provision and thus problems of diseconomies of scale in remote and small market
places. 23 A further step would be to allow the use of public post office networks
as platform for service provision by different financial institutions, as in India
and South Africa (see World Bank, 2004a,b). This cannot only help overcome
problems of scale economies, but address concerns of competition compared to
the situation where only one institution is allowed to use post offices as corre-
spondents as in Brazil or a situation where one financial institution dominates
service provision in remote areas, as is the case in large parts of Sub-Saharan
Africa.
Sometimes, the role of government in fostering access might have to go beyond
competition policies and take the form of “affirmative regulatory policy.” Examples
include the moral suasion exercised by authorities to make South African banks in-
troduce the Mzansi (basic transaction) account or make British banks introduce the
Basic Bank Account (BBA). 24 Inducing banks to share or ensure interoperability
of payments infrastructures (including ATM networks) can help avoid undesirable
competition on access to infrastructure while enhancing desirable competition on
price and quality of service, thereby facilitating the achievement of cost-reducing
scale economies and lowering entry barriers to new financial service providers.
Market enabling policies can also be important on the demand side. Financial
literacy programs or financial products tailored to specific groups can increase
access to and use of financial services by bringing actual demand closer to potential

23
In Brazil, for instance, some of the largest banks have expanded their network through correspondent
agreements with the Post Office, lottery shops and supermarkets (Kumar, 2005).
24
See Napier (2005) for a discussion of the Financial Access Charter in South Africa. This is very
different from Universal Service Obligations, as legislated in several European countries (Kempson,
Atkinson, and Pilley, 2004), which require banks to open a bank account for anyone. Such obligations
can force banks to move access beyond the constrained optimum and raises questions on sustainability
and efficiency.
 The Basic Analytics of Access to Financial Services 93

demand and move a financial system closer to the frontier. 25 The government,
in its role as financial market participant, can also have an important role. For
instance, in 1999, the U.S. Treasury started a program to make all federal benefit
payments though electronic transfer accounts, using subsidies to encourage banks
and beneficiaries to open such accounts (Caskey, Ruiz Duran, and Solo, 2006;
Claessens, 2006).
Market-enabling policies, however, find their limits in the state variables. Defi-
ciencies in state variables require more profound structural reforms, mostly outside
the reach of even financial sector policy makers. We denote such reforms as market-
developing policies. One often neglected area are distortions in the input markets
for financial service provision, including labor and communication markets. A
deficient phone infrastructure and lack of competition in the telephone market can
drive up costs for financial institutions. 26 A deficient transportation infrastructure
can increase the cost of outreach into more remote areas, even through innovative
techniques such as mobile banking. Generally high costs of doing business, due
to widespread corruption, insecurity, or even deficient electricity provision, can
drive up the fixed cost element of financial service provision with negative reper-
cussions for outreach to low-value/high-volume customers. Finally, economic size
can be a restricting factor as discussed above. Small economies should therefore
put a premium on encouraging entry of foreign banks that are able to reap benefits
of scale economies across subsidiaries in different countries, on integrating their
local markets with international financial markets, and on allowing their citizens
access to financial services across borders. 27

III. THE ANALYTICS OF ACCESS TO CREDIT

This section expands the concept of the Access Possibilities Frontier to lending
services and introduces risk as the key feature that distinguishes credit services
from savings and payment services. While we will be referring to lending services
in general, most of our analysis focuses on a given loan product rather than on
aggregate lending services, as different types of loans entail different technologies
and, hence, different production functions, implying switching costs in moving
from the provision of one type of lending to another.

25
Compare Financial Services Authority (2000) and the literature quoted therein. See Ashraf, Karlan,
and Yin (2006a,b) for an example of how the introduction of commitment savings product can impact
savings rates.
26
Compare the detailed discussion for the case of South Africa in World Bank (2004b).
27
For a more detailed discussion on policy options for small financial systems, see Bossone, Honohan,
and Long (2002). Ultimately, the last recommendation leaves open the question to which extent domestic
financial systems are viable at all in small developing economies and agents should not rather rely on
providers abroad for certain services. Even in the extreme, some local markets will be needed, especially
for local-currency denominated financial services, which are important not just for financial access but
also to avoid financial vulnerabilities stemming from currency mismatches.
94 Thorsten Beck and Augusto de la Torre

RISK AS BARRIER TO OUTREACH IN CREDIT SERVICES SUPPLY

In addition to costs, the outreach in the supply of credit services is constrained
by risks, especially default risk. These can be either borrower-specific or sys-
temic. For the purposes of this paper, we define as systemically originated credit
risk that which is non-diversifiable within a given domestic jurisdiction and as a
consequence affects all credit contracts therein. 28
Systemic risk typically stems from high macroeconomic uncertainty (e.g., sig-
nificant volatility in the rate of inflation, the terms of trade, the real interest, and the
real exchange rate), weaknesses in the contractual and informational environment
(e.g., poorly defined and difficult to enforce creditor rights, deficient accounting
and disclosure practices, and lack of a well functioning credit bureau), or geo-
graphical limitations (e.g., a small country prone to flooding or hurricanes). 29
Regardless of its origin, systemic risk hinders the supply of credit because it raises
the default probability and the loss given default for all credit contracts written in a
given jurisdiction. This leads to a higher than otherwise cost of funds and, hence, a
higher floor for the interest rate required by a creditor to make a loan. As systemic
risk increases, it enlarges the set of borrowers/projects that find the cost of credit
unaffordable and are thus priced out of access to credit.
Idiosyncratic credit risks are specific to individual borrowers/projects and are,
hence, not correlated with systemic risk. As a result, the cost of finance and/or
availability of credit services will differ across debtors/projects depending on their
differences in idiosyncratic riskiness, and is priced as a spread over the interest
rate floor set by systemic risk. Importantly, however, the ability of the lender to
manage idiosyncratic risk is influenced by the systemic risk environment as we will
discuss below. Two factors are particularly important in explaining the differences
in interest spreads across debtors (for a given type of loan) that are induced by
idiosyncratic risk: agency problems and limits to the diversification of risks that
are not related to agency problems. Let us explain these factors in some detail.
Consider first the constraints on the ability to reduce non-agency related risk
(for a given return) through diversification. Idiosyncratic risk would in principle
be diversifiable or insurable (and thus not priced in the interest rate spread) in
an idealized world where markets are complete. 30 The limits to idiosyncratic risk
diversification observed in the real world are thus a reflection of some form of mar-
ket incompleteness, including the lack of sufficient markets for hedges and other
insurance products. If unable to diversify non-agency related risks in a competitive
market, risk adverse creditors will include a risk premium in the lending interest
rate. Such risk premium increases the lending interest rate beyond the level nec-
essary to cover the creditor’s marginal cost of funds plus the cost she has to incur
28
In our definition, therefore, systemic risk might be diversifiable through international financial
markets.
29
De la Torre and Schmukler (2005) analyze the role of systemic risk in credit contracts and show that
the terms of loan contracts—particularly duration, currency of denomination, and jurisdiction—are
adapted so as to cope with systemic risk.
30
Compare discussion in any standard finance textbook, such as Copeland, Weston, and Shastri (2005).
 The Basic Analytics of Access to Financial Services 95

in order to provide the credit service. In all, high risk premiums undermine credit
supply outreach to the extent that they render lending interest rates unaffordable for
certain borrowers. The lack of agricultural lending in many developing countries
has often been explained by the inability of financial institutions to diversify the
high risk stemming from agricultural activity and therefore agricultural lending.
Consider now agency problems. They give rise to a misalignment of debtor
incentives with the interests of the creditor. The most studied source of agency
problems is information asymmetry, whereby the debtor is privy to relevant infor-
mation about herself and her project that the creditor may not be able to secure
or only at a prohibitively high cost, and which can lead to two conceptually dis-
tinct sources of credit risk: adverse selection and moral hazard. 31 Although higher
agency-related idiosyncratic credit risk can be compensated by increasing the risk
premium as discussed above, such an increase can lead to adverse selection and
exacerbate the moral hazard problem.
Regarding adverse selection, a higher risk premium would tend to attract riskier
borrowers, i.e., borrowers more likely to default in bad states of the world, to the
pool of loan applicants while the creditor may lack sufficient information to sort
them ex-ante. As a result, instead of raising the risk premium, creditors try to
mitigate adverse selection through non-price screening devices (such as collateral
or character assessment) or simply deny credit to loan applicants that are not
favorably screened through such devices. Moral hazard, for its part, arises from
the borrower’s incentive to use the proceeds of the loan in endeavors that are riskier
than those specified in the credit contract, while being able to conceal such behavior
from the creditor. By taking on greater risks, the debtor can capture the windfall
gains in the good states of the world while limiting her loss to her investment in the
project (assuming limited liability) in the bad states of the world. An increase in
the non-agency related risk premium, furthermore, can intensify moral hazard by
further exacerbating the debtor’s incentives to divert the loan resources to riskier
uses. Thus, rather than increasing the interest rate, creditors may again resort to
non-price devices to mitigate moral hazard or simply curtail the quantity of credit
supplied.
Figure 3 allows us to illustrate more formally the way in which systemic and
idiosyncratic risks influence the cost and availability of credit supply. The figure
is drawn for a given type of loan (say a consumer, mortgage, working capital,
or investment loan) supplied by financial institutions in a given jurisdiction. The
lending interest rate, i, is measured along the horizontal axis, and the expected
return to the lender, r, is measured along the vertical axis. In an ideal world devoid
of transactions costs and risks, i and r would coincide and their relation would take
the form of a straight 45 degree line from the origin. In the real world of market
frictions and risks, however, there is a wedge between the interest rate charged by
the creditor and his expected return. We will now show how the three different risk
31
The classic article on the effects of information asymmetry on credit supply is Stiglitz and Weiss
(1981). Though conceptually distinct, adverse selection and moral hazard are difficult to distinguish
empirically, as discussed by Karlan and Zinman (2004).
96 Thorsten Beck and Augusto de la Torre

r
Expected
return for bank

wedge

r* Risk premium
I

Systemic risk
II

i
loan interest rate
imc i*

Figure 3: Systemic risk, agency problems, risk premiums and credit rationing

categories lead to a deviation of the lender’s return from the lending interest rate
and potentially to credit rationing.
Take first country-level systemic risk, illustrated by a parallel line to the 45 de-
gree line, and with this difference denoting the country risk premium, c. Denoting
with i mc the marginal cost of funds for the creditor bank—which, due to arbitrage,
is assumed to be equal to the interest rate paid on short-term government debt
securities—we can establish a floor for lending interest rates and a first wedge
as r mc = i mc − c. The country risk premium, c, can be explained by the different
elements of systemic risk explained above—macroeconomic uncertainty, deficien-
cies in the contractual and informational frameworks and geographic limitations.
This systemic risk is common to all credit contracts in a country and cannot be
diversified away within that jurisdiction.
Take next the non-linearities due to scale and agency problems, illustrated by
curve I. 32 The nonlinear wedge between the 45 degree line and curve not only
implies that the default probability increases with the lending interest rate, causing
r to rise less than i. It also implies that, as the lending rate increases beyond a
given threshold, denoted in Figure 3 by i∗ , the expected return begins to decrease.
Thus, at (i∗ , r∗ ), the marginal revenue to the creditor due to a contractual increase
in the lending interest rate is fully offset by the marginal expected loss due to
a higher probability of default. This nonlinear relationship can be explained by
agency problems, for the reasons discussed above. But it could also be rationalized
without resort to agency problems. For example, as shown by Williamson (1987), a
similar nonlinear wedge can arise from the existence of transaction costs in lending
combined with uncertainty in debtors’ revenue streams. 33
32
Note that this curve starts at i mc , as no loan would be made by a creditor at an interest rate below his
marginal cost of funds or opportunity cost of lending.
33
In such a situation, an increase in average transaction costs increases the loan interest rate the lender
has to charge in order to recover her costs and thus increases the probability that the borrower may not
be able to repay due to a negative output shock, independently of her willingness to repay.
 The Basic Analytics of Access to Financial Services 97

Consider finally the incapacity to diversify non-agency related risk. Curve I is

drawn after subtracting from the interest rate any non-agency related idiosyncratic
risk premium. Hence, along curve I, a higher interest rate incorporates only higher
lending costs due to smaller scale and/or higher agency-related idiosyncratic risk.
Curve II, on the other hand, takes into account the risk premium and, hence, is
always to the right of curve I, with the vertical distance between the two curves
measuring the premium charged by creditors for non-diversifiable risk. To the
extent that the risk premium increases with the level of the lending rate (reflecting
the increase in the ex-ante probability of default), curve II would be flatter than
curve I and would have a lower flexion point, as drawn in Figure 3. Note that
the widening of the wedge between i and r as i increases is common to both
curves. This is because the probability of default rises with the lending interest
rate, independently of the reasons (costs, risk-adjusted profits, or risk premium)
that push that rate up. 34
Figure 3 allows us to focus on three types of distinct but interrelated state
variables that constrain the supply of credit services. The first state variable is
macroeconomic risk which, as noted, is assumed to be fully summarized in the
interest rate paid by the sovereign on its debt. Higher macroeconomic risk raises
the opportunity cost of lending, i.e., i mc , for all creditors in a given jurisdiction,
thereby increasing the floor for the minimum interest rate that can feasibly be
charged on loans to the private sector. The higher this state variable, the larger the
set of households and firms that will be priced out of the credit market.
The second state variable is the quality of the contractual and informational en-
vironment. The lower such quality, the more difficult it will be for creditors to miti-
gate information asymmetry and contract enforcement problems in “arms-length”
credit transactions and, hence, the more they will resort to non-price screening and
monitoring mechanisms, such as highly personalized “relationship lending” and/or
heavy reliance on fixed (mainly real estate) collateral from debtors. The lower the
quality of the contractual and informational environment the flatter Curves I and
II in Figure 3 and the wider the wedge between i and r that cannot be reduced
through idiosyncratic effort, i.e., effort by the individual lender. As a result, the
lower such quality, the larger the set of debtors that, despite having viable projects,
i.e., projects that debtors would be willing to finance out of their own resources if
available, will have no access to external finance. 35
The third state variable is given by systemic limits to the diversification of non-
agency related risks. As noted, such limits reflect undiversified productive sectors

34
The supply function for loans to the private sector can be easily derived from Curve II in Figure 3, by
noting that loan supply, L s , is a positive function of expected return, r, which is in turn a function of the
lending interest rate. Formally, L s = f{r(i,. . .)} = g{i,. . .}, where δg/δi > 0 up to a certain threshold.
Moreover, the non-linear relationship between nominal interest rate and expected return to lender can
result in backward-bending supply curve-δg/δi < 0-and credit rationing, as shown by Stiglitz and Weiss
(1981).
35
De la Torre and Schmukler (2006a) use this “agency wedge” as their working definition of an “access
to finance problem.” They note that, according to this definition, an access problem is reduced to the
extent that the share of viable projects that are able to obtain external finance increases.
98 Thorsten Beck and Augusto de la Torre

and financial underdevelopment, manifested in missing markets for certain loan

products as well as for hedges and insurance products. 36 The narrower the scope
to diversify away idiosyncratic risk, the higher the risk premium that will be
incorporated in the interest rate for any given type of loan. High risk premiums
further enlarge the set of potential debtors that are priced out of the credit market.
High risk premiums, moreover, exacerbate agency problems, further enlarging the
set of debtors that is rationed out of the loan market.
As state variables are out of the control of individual lenders, their main job
consists of managing idiosyncratic risks and costs so as to make a profit. Lenders
must develop loan technologies that enable them to choose such debtors that, for
any given lending interest rate, will give the highest risk-adjusted expected net
return. To select such debtors for a given type of loan, the creditor must compare:
(i) the all-in costs of lending to different debtors/projects; (ii) the differences in
expected returns (capacity to pay) across such debtors/projects that the lender
considers to be equally risky; and (iii) the differences in risks and willingness to
pay across such debtors/projects that the lender considers to be of equal expected
returns (or capacity to pay). These comparisons require approaches that differ
across different types of loans, ranging from scoring methods based on the law
of large numbers for the more homogeneous loan products (such as credit card
loans, micro loans, or even mortgage loans), to the low-cost low-risk lending to
the most reputable, creditworthy, and typically large debtor firms, and to the more
intense and personalized approaches in lending to the opaque and clearly more
heterogeneous SMEs. After having selected the debtor/project that will be offered
a loan, the lender has to monitor the debtor’s behavior and has to decide what to do
in case of default, i.e., to write-off the loan or engage in post-default value recovery
efforts. Finally, the scope for optimization that the lender will have in managing
lending costs and risks will be constrained by the state variables. The weaker the
state variables, the less the maneuvering room for credit supply optimization. Given
the constraints, a prudent lender will rather not offer loans at a higher interest rate
if she is not reasonably sure of her ability to appropriately measure costs, sort out
risks for a given expected return, and identify expected returns for a given risk.
There is a correlation, but far from perfect, between cost and risk management.
Risk management activities aim at better measuring and controlling default risk
exposure in the production of credit services, while cost management activities
aim at minimizing costs for a given unit of output. A given management activ-
ity may accomplish both aims. For instance, the introduction of computer-based
statistically-driven scoring models to measure and control default risk in consumer

36
The missing market problem, moreover, can be exacerbated by deficiencies in the other two state
variables. For example, high macroeconomic volatility can itself be a root cause of missing derivatives
markets or missing markets for long-duration local-currency debt. The problem of missing markets not
only limits the ability of creditors to diversify risks but also makes it more likely for debtors to incur
currency and duration mismatches in their balance sheets, which, in equilibrium, can be an optimal
outcome in high systemic risk environment (De la Torre and Schmukler, 2005).
 The Basic Analytics of Access to Financial Services 99

or micro credit may also be seen as a cost management activity to the extent that
it enables a reduction in the number of loan officers and branches (given that
scoring methods could be applied via e-banking). 37 By contrast, some activities
may only affect directly cost management (e.g., upgrading the IT hardware) or
only risk management (e.g., defining the types of assets that would be accepted as
collateral).
Lenders must confront complex interactions between costs, returns, and risks.
Clients that are equally risky may be associated with different expected returns (or
capacity to pay). Clients with similar payment capacity may have different risks
and willingness to pay. Clients with equal risk and capacity to pay may represent
different costs for the bank. Costlier clients are often, but not always, also the
riskier and smaller clients. For instance, small-scale enterprises demand smaller
loans, are less transparent, and can often offer fewer risk-alleviating remedies
such as fixed collateral and guarantees. In contrast, the marginal borrower in long-
term investment loans is not necessarily small, as the inability of lenders to offer
longer-term investment loans is mostly due to a missing market problem arising
from systemic risk considerations and independent of loan size.
Be it as it may, holding the state variables constant, a supply-driven broadening of
access to credit services can be generally defined as the incorporation of additional,
marginal borrowers into the credit circuit for any given lending interest rate and
loan type. The marginal borrower would generally be the one that, in the view
of the lender, will give a risk-adjusted return that is just sufficient to cover costs.
Compared to our previous analysis of savings and payment services, however,
where the marginal client was identified as the smaller firm and poorer household
(characterized by low value transactions), the marginal client in the supply of credit
services is not necessarily defined along one dimension. It may be the costlier client
to serve (due to opacity, riskiness, or smallness); it may be the client with a lower
expected return among equally risky debtors; or it may be the client with the higher
risk among debtors with equal expected returns.
We can now take the analysis one step further by considering the supply and
demand for credit services, which allows us to develop the Access Possibilities
Frontier for loan services in a similar fashion as was done in Section II for payment
and savings services.

THE ACCESS POSSIBILITIES FRONTIER FOR CREDIT SERVICES

As we combine demand and supply for credit services, our focus on access means
that we are interested in the marginal borrower/project. However, as noted, identi-
fying the marginal borrower/project is not a simple matter, given the complex inter-
play of costs, returns, and risks in credit markets. Because the multi-dimensionality

37
For a discussion on the effect of credit scoring on bank efficiency and outreach in the U.S., see
Frame, Padhi, and Woosley (2001) and Berger, Frame, and Miller (2005).
100 Thorsten Beck and Augusto de la Torre

of lending constraints is captured in one price—the interest rate—there is not a

unique combination of cost, size, risk, and return of projects that maps one-to-one
to interest rates. Furthermore, given information asymmetries, the ranking of bor-
rowers according to, say, expected returns or riskiness, will be different depending
on whether it is done by the lenders or the debtors. Thus, demand and supply can-
not be drawn in the same space that has the interest rate on the vertical and a given
ordering of borrowers on the horizontal axis. This is different from payment and
savings services, where transaction costs and size of transaction can be mapped
directly into the price and thus into demand and supply. Further, changes in the
price, i.e., the interest rate, cause changes in borrowers’ behavior beyond ability to
pay (unlike in the case of payments and savings services) as discussed above. Due
to asymmetric information, however, these changes are not obvious to the lender;
while she might know the overall riskiness of the portfolio, she does not know
the riskiness of individual borrowers and their precise change in behavior as the
interest rate changes. In all, the multidimensionality of lending constraints and the
reaction of borrowers to changes in interest rates make it impossible to combine,
in the traditional way, demand and supply into one graph. We will therefore take
an alternative, two-step approach, first defining demand for a given loan product
and given a certain i mc . Taking this defined universe of potential loan applicants,
we will then illustrate supply as a function of the marginal interest rate, introduce
demand in an indirect and non-traditional way, and derive an Access Possibilities
Frontier that will help us determine the share of borrowers in a pool of loan appli-
cants that are potentially bankable and actually banked, as function of the interest
rate.
Consider first demand, illustrated in Figure 4. The vertical axis denotes the
nominal marginal interest rate charged by lenders, while the horizontal axis has
the number of loan applicants, who are ordered according to the inverse of the

iIII

i
I

II
III

A
imc
imc’
Loan applicants,
ordered by inverse
BII BI BIII BIII’ of expected return

Figure 4: Loan demand as function of expected return, non-economic factors

and agency costs
 The Basic Analytics of Access to Financial Services 101

expected return of the project that they advertise to lenders. 38 Hence, as we move
along the horizontal axis and to the right we encounter applicants with lower and
lower expected returns. For simplicity, we assume that each applicant demands
only one loan, although we allow loan sizes to vary across applicants. We can thus
be sure that the downward slope in the demand curve implies that a decrease in the
lending interest rate adds a new applicant with a marginally lower expected return
to the pool of demanders (rather than adding a new loan by a borrower already
in the demand pool). 39 We draw three different demand curves in Figure 4, cor-
responding to (i) demand in a world without market frictions, defined as a world
where borrowed resources are treated by the debtor as if they were his own re-
sources; (ii) demand taking into account self-exclusion due to cultural barriers and
financial illiteracy; and (iii) demand taking into account limited liability, asymmet-
ric information, and the resulting agency problems. All three demand curves are
contingent on the investment opportunities available in the economy—the lesser
the investment opportunities the more the demand curve would be to the left of the
graph.
Take first demand corresponding to a world without market frictions, represented
by Curve I, which serves as benchmark for the further analysis. This illustrates
demand for loans that borrowers would treat like their own resources or, alterna-
tively, demand that involves only those projects that the debtor would undertake
with her own resources if she had them. Imposing (for ease of argument) a zero-
profit condition on borrowers, this would imply that the marginal borrower with
expected return x is willing to pay an interest rate of up to x% on her loan. Demand
is cut off at i mc , i.e., no borrower will demand loans for a project with expected
return lower than i mc , the return that could be earned by investing in debt securities
of the local sovereign.
Not all potential borrowers with profitable investment projects will demand
loans, however. Cultural barriers to incurring debt or financial illiteracy resulting
in ignorance about borrowing possibilities might result in self-exclusion and, thus,
a loan demand as represented by Curve II, to the left of Curve I. 40
While cultural barriers and financial illiteracy might reduce demand, incen-
tives arising from limited liability cum asymmetric information are demand-
increasing—they result in a demand for loan resources that exceeds what the
borrower would invest in the project if she owned such resources. Limited liability
leads to an asymmetric participation of the borrower in the returns of a project—
the gains from undertaking the project with the lender’s money are potentially
unlimited but the losses are limited to the borrower’s own resources invested in

38
To simplify things, we assume that the expected return that borrowers advertise to lenders is equal
to the one they expect to get. As will become clear later, however, this advertised expected return may
be higher than the return that borrowers expect the project itself to deliver.
39
For ease of graphic illustration, we are using a straight line for the demand curve. In reality, this
does not have to be the case.
40
For ease of argument, we draw Curve II parallel to Curve I, although the distance between the two
could vary across different interest rates.
102 Thorsten Beck and Augusto de la Torre

the project. Asymmetric information might result in adverse selection and moral
hazard problems as discussed above. Thus, limited liability and asymmetric infor-
mation combine to create incentives for borrowers to undertake riskier projects.
Given that limited liability truncates the downside losses to the borrower, the ex-
pected return to the borrower can be higher than the expected return of the project
itself. Therefore, in a world of agency problems, a borrower has incentives to
finance her project at interest rates above the project’s expected return.
Curve III in Figure 4 is the demand curve that results from subtracting
the demand-reducing effects of self-exclusion from Curve I and then adding
the demand-increasing effects of limited liability and agency problems. We as-
sume that the latter effects outweigh the former and, hence, we draw Curve III to
the right of Curve I. Compared to Curve I, Curve III implies more loan applicants
for any level of the lending interest rate (or the lending spread over i mc ). The
distance between Curves II and III represents the severity of agency problems, as
applicants in Curve III demand more external resources because they do not have
to treat them as their own and are therefore willing to take greater risks and pay an
interest rate higher than the expected return of their project. As in Figure 3, where
the gap between interest rate and return to lender increases with increasing interest
rates, it is likely that the distance between Curves I and II, on the one hand, and
Curve III, on the other hand, will increase with higher interest rates, as agency
problems intensify.
Figure 4 allows us to define the universe of loan applicants or potential borrow-
ers. For a given level of the lending interest rate, the corresponding point in Curve
III gives the maximum number of loan applicants that would be observed, given
the state variables, in a world characterized by self-exclusion, limited liability, and
agency problems. The set of loan applicants grows bigger as the lending interest
rate declines and it reaches a maximum at point A, the intersection of Curve III and
the parallel to the horizontal axis at i mc . The maximum number of applicants will-
ing to take loans at i = i mc is denoted by B III . The marginal applicant in B III has the
lowest expected return which, due to agency problems, is higher than the expected
return of the project itself. Further, if i mc decreased to i mc ’ (as shown in Figure 4)
due to an improvement in the macroeconomic state variables, the maximum set of
applicants would increase to B III ’. Thus, a reduction in systemic risk has a direct
impact on total loan demand. Note, however, that along Curve III there will be an
imprudently high demand for loans, an excessive number of potential borrowers
compared to the number of applicants that what would obtain under Curve I, i.e.,
in the absence of the incentives distortions introduced by agency problems.
Let us now turn to the supply schedule illustrated in Figure 5, which is again
drawn for a given type of loan and where the marginal lending interest rate is
measured on the vertical axis. However, as discussed earlier, there is not a unique
combination of costs, expected return, and risk that maps one-to-one to the interest
rate. Rather, it is a mix of different borrower characteristics as perceived by the
lender that determines the marginal borrower at each interest rate along the supply
curve. To capture supply-driven changes in access, while keeping some indirect
 The Basic Analytics of Access to Financial Services 103

iIII
D
II
I
III
i*
S S’
S*
imc

Share of loan
B A C 100% applicants

Figure 5: Access Possibilities Frontier for credit services

reference to demand, we follow a strategy proposed in De la Torre and Schmukler

(2006a), namely, to focus on the share of loan applicants at a given interest rate that
actually receive a loan. To determine such share, we use Curve III in Figure 4. 41
Thus, the horizontal axis in Figure 5 denotes the share of loan applicants receiving
loans, which obviously reaches a maximum at 100%, denoted by the vertical line
D. Each point along the this vertical line corresponds to a point in Curve III of
Figure 4; as the interest rate increases along the vertical line D, the number of
borrowers demanding loans falls (as per Curve III), while the share remains of
course constant at 100%. Line D is thus a “normalized” demand curve, where the
underlying level of demand, i.e., number of loan applicants, falls as the marginal
interest rate increases, but the remaining loan applicants are re-normalized to one.
Line D is truncated at i III , the lending interest rate at which Curve III intersects the
vertical axis in Figure 4 and, hence, where the pool of loan applicants is empty.
It is the lender who determines which borrowers get a loan. At each lending
interest rate, the pool of loan applicants that request loans from the lender is
determined by Curve III in Figure 4. It is a pool that includes “good” borrowers, who
would treat the lender’s money as if it was their own, as well as “bad” borrowers,
who would not invest their own resources in projects for which they now demand a
loan but would be willing to gamble with the loan resources if and once obtained.
Faced with this problem, the lender uses a mixture of price and non-price devices
to try to select the “right” debtors out of this pool. Her ultimate objective is to
identify the subset of loan applicants that corresponds to Curve I in Figure 4,
that is, to sort out those loan applicants that would treat the loan resources as if
41
De la Torre and Schmukler (2006a), by contrast, discuss changes in access with reference to the
demand for loans for “viable projects,” which coincides with our definition of demand Curve I in
Figure 4.
104 Thorsten Beck and Augusto de la Torre

they were their own. The supply curve in Figure 5 is upward sloping but becomes
steeper and steeper and eventually can bend backward. At i = i mc the share of
loans applicants that get a loan from lenders is zero, as lenders would rather just
invest in debt securities of the local government. The shape of the supply curve
reflects a complex interaction between lenders’ costs and their ability to identify
appropriately expected returns and risks. Let us explain this better.
While the number of potential borrowers in the pool of loan applicants falls
with increases in the lending interest rate (along demand Curve III in Figure 4),
the riskiness of such pool increases. Nonetheless, in the lower region of the supply
curve, lenders provide loans to a rising share of loan applicants as the interest rate
increases. This is not just because a higher lending spread enables lenders to cover
more costs. It is also because lenders feel confident in their ability to assess risks and
thus to sort out appropriately such debtors/projects that have higher expected risk-
adjusted returns—and that would be thus able to repay the loan under reasonable
states of the world. However, as the lending interest rate increases further, the
task for the lender becomes increasingly more difficult because the pool of loan
applicants becomes riskier (and thus the distance between Curve III and Curve
I in Figure 4 widens). The lender becomes more and more reluctant providing
loans to a larger share of loan applicants at higher interest rates (the supply curve
steeps up) and, at a threshold interest rate, the supply curve reaches a flexion point
and may begin to bend backward. Prudent loan suppliers thus try to offset the
negative effects of higher interest rates on default probabilities via rationing—
denying loans to debtors perceived to be too risky. While rationing intensifies as
the lending interest rate rises, some rationing can happen at any point in the supply
curve, in the sense that the share of borrowers that actually get a loan might not
include some or many of the viable borrowers in Curve I of Figure 4. This is an
inevitable cost of prudence in a world of uncertainty and agency problems—some
“good” debtors have to be left out in order to minimize the set of “bad” debtors
that are let in. 42
Curve S∗ in Figure 5 denotes the potential supply curve for loans, i.e., the share
of borrowers/projects that can be efficiently and prudently serviced by lenders as
a function of the marginal interest rate and given the state variables, including i mc .
Prudent lenders that maximize risk-adjusted profits will supply loans up to the
flexion point of S∗ , at i∗ , where the contractual income gains of a higher interest
rate are just offset by the expected default losses. Therefore, i∗ is the “rationed
equilibrium” marginal interest rate in the market for a given type of lending product,
and the flexion of S∗ at point I denotes the Access Possibilities Frontier for credit
services, i.e., the maximum equilibrium outreach in terms of access to credit that
is prudently achievable given the state variables, including i mc . This corresponds
42
This is the classic tradeoff between “type I” and “type II” errors. The fuzziness in screening introduced
by uncertainty and agency problems confronts the lender with overlapping probability distributions of
good and bad debtors. Faced with this problem, lenders will in general try to minimize the error of
letting too many bad debtors in, which necessarily implies that some good debtors will be left out.
 The Basic Analytics of Access to Financial Services 105

to the bankable share of loan applicants A. Note that this bankable share depends
on the level of i∗ , as the pool of loan applicants is different for different levels
of the lending interest rate (as per Figure 4). 1-A is the unbankable share of loan
applicants at i∗ , given the state variables. 43
Again, we allow for the possibility that the financial system may not achieve this
potential so that actual supply, S, falls short of S∗ . Thus, a sub-optimal credit supply
can obtain, for instance, if creditors’ profits are not threatened by the pressures
of competition and, as a result, they have little incentive to undertake greater but
feasible cost and risk management efforts to reach out to new (costlier and riskier)
clients.
We can use Figures 4 and 5 to identify different types of access to finance
problems. To this end, we focus on deserving debtors/projects that do not get loans
at the “rationed equilibrium” marginal interest rate, rather than on the number of
deserving projects. 44 We start by identifying access problems that arise holding the
state variables constant. A first type of access problem in this context would consist
of too low a number of loan applicants simply because of self-exclusion resulting
from cultural barriers or financial illiteracy. This would be reflected explicitly in
a lower demand curve (Curve II) in Figure 4 but would only be implicit in the
vertical line D in Figure 5.
A second type of access problem can arise due to supply sub-optimization,
represented in Figure 5 by the supply curve S, which is below the curve S∗ . In this
case, the problem is that of too low a supply outcome where, due to, for instance,
regulatory distortions or insufficient contestability, lenders do not fully exploit
all the outreach opportunities, given the market interest rate, the available loan
technologies, and the state variables. As drawn in Figure 5, this not only results
in a higher maximum marginal interest rate, but also a lower banked share of loan
applicants at interest rate i∗ ; the distance A-B at i∗ is a measure of access problems
driven by supply inefficiencies. 45
A third and very different access problem is associated with “excess access,”
denoted by supply to the right of the prudent possibilities frontier given at point
I in S∗ . Curve S’, to the right of S∗ , would in effect imply that loans are granted
to a larger share of loan applicants than is prudently warranted, given the lending
interest rate, and given the state variables, particularly the quality of the informa-
tional and contractual environment. For example, at i∗ , the problem of “excess
43
To get a measure of the overall unbankable share of loan applicants we would have to add loan
applicants that are rejected at lower interest rates but decided not to demand a loan at this higher
interest rate i∗ .
44
Note that a key problem in emerging markets could be a lack of investment projects that deserve
financing based on their expected return. This would be reflected in a demand curve in Figure 4 that is
too far to the left or too steep. While this is a relevant problem, it is not an “access to finance problem.”
45
There are two effects explaining the difference between interest rates at the flexion points of S and
S∗ . On the one hand, S reflects higher costs resulting in higher interest rate for every borrower; on
the other hand, the flexion point obtains earlier, so that the “equilibrium” marginal interest rate for S
could also be lower than for S∗ . In the context of Figure 5, however, we conjecture that the first effect
is stronger than the second.
106 Thorsten Beck and Augusto de la Torre

iIII
D

I II III
i*

S*’’
S*

imc
S*’
imc’
Share of loan
A B C 100% applicants

Figure 6: Access Possibilities Frontier for credit services – changes in state

variables

access” can be represented by the distance C − A in Figure 5, which denotes the

increase in the share of loan applicants receiving a loan at i∗ that is not warranted
by prudent risk management. Point C in Figure 5 would be, given i∗ , “too close”
to the vertical line D. “Too close” in the sense that lenders would be allowing in
too many borrowers that are in Curve III but are not in the ideal Curve I (both in
Figure 4).
While the points of demand-driven self-exclusion, supply-driven constrained
sub-optimality and imprudent excessive access are defined holding constant state
variables, there are two other access problems that refer directly to the location of
the prudent access possibilities frontier and thus to deficiencies in state variables
compared to countries with similar levels of economic development. The first such
access problem would be due to an i mc, the opportunity costs of lending, which
is too high. The effect of lower opportunity costs is illustrated in Figures 4 and 6.
A lower i mc will increase the universe of potential loan demanders (from B III to
B III ’ in Figure 4). It will also give rise to supply curve S∗ ’, which is to the right of
S∗ , with a higher share of loan applicants served at any interest rate. For example,
at i = i∗ , the share of loan applicants receiving finance would increase from A to
B in Figure 6.
A second access problem related to state variables would arise due to a contrac-
tual and informational environment that is too weak compared to that in countries
with similar levels of per capital income and which would affect the curvature
of S∗ . An improvement in the contractual and informational environment would
result in a supply curve that is flatter than S∗ , allowing for a larger share of loan
applicants to be prudently and efficiently served by lenders at all relevant levels of
the interest rate. The effect of an improvement in the state variables is illustrated
 The Basic Analytics of Access to Financial Services 107

in Figure 6 by curve S∗ ; the effect on the bankable share of loan applicants at i∗
would be the horizontal distance C-A. 46

POLICIES TO FOSTER ACCESS TO CREDIT SERVICES

In order to design effective policies to expand access, policymakers have to be
aware of whether their financial system is inside the frontier, at the frontier, or
unsustainably beyond it, and/or whether the possibilities frontier is too low rela-
tive to countries of comparable levels of economic development. As in the case
of payment and savings services, we can distinguish between market-developing
policies—which raise the sustainable prudent possibilities frontier by changing the
state variables—and market-enabling policies—which provide incentives to, or re-
move obstacles for, private financial institutions to move closer to the frontier, given
the state variables. Additionally, however, we have to consider market-harnessing
policies that prevent the financial system from moving to an unsustainable impru-
dent equilibrium beyond the frontier. The weight given to each type of policies will
depend on the diagnosis of the access to finance problems and an identification of
where the more binding constraints lie. 47 It will also depend on the policy time
horizon.
Consider first market-developing policies. As they aim at improving the state
variables, these policies involve fundamental reforms that often confront polit-
ical resistance and are thus difficult to implement (Rajan and Zingales, 2003).
Market-developing reforms include, for instance, legal and even constitutional
changes, major modernization processes in public sector agencies (including civil
service reform), and substantial upgrades in macroeconomic, particularly fiscal,
performance. To be sure, the results of market-developing reforms can at times be
elusive—not least because of path dependence and problems associated with par-
tial reform of complex institutional arrangements. 48 However, their pursuit must
be given high priority where the binding constraints consist of major deficiencies
in the state variables. Market-developing policies can be categorized according to
the state variable they refer to: macro volatility, contractual and informational envi-
ronment, or the scope for idiosyncratic risk diversification. Let us briefly consider
each of these cases.
A high opportunity cost of lending may in some cases be the most important
hindrance to the broadening of access to credit to the private sector. It may re-
flect doubts about fiscal solvency and a history of asset confiscation or inflation
46
As in the case of the flexion points of S and S∗ in Figure 5, there are two offsetting effects on the
flexion points as we move from S∗ to S∗ ’. On the one hand, a lower i mc implies a lower interest rate
for every borrower; on the other hand, this allows lenders to increase the marginal interest rate more.
For ease of discussion, we assume that both effects cancel out in Figure 6, so that the flexion points of
both supply curves are at the same interest rate i∗ .
47
This very much resembles the “Growth Diagnostics” analysis by Hausmann, Rodrik, and Velasco
(2005).
48
See De la Torre and Schmukler (2006b) for a discussion on the elusiveness of results of capital
markets reforms in Latin America.
108 Thorsten Beck and Augusto de la Torre

volatility, and will be often associated with high financial crowding out—i.e., the
absorption by the government of a large share of society’s financial savings. Under
such circumstances, the appropriate policies to foster access to credit would be
those that aim at enhancing the resiliency of fiscal solvency and at establishing a
credible record of low and stable inflation.
A shallow credit market with low access may also result from major shortfalls in
the contractual and informational frameworks. 49 The appropriate policies in this
case would span a wide range: from titling of land property to the upgrading of
laws affecting collateral repossession or execution of guarantees; from the modern-
ization of corporate reorganization and bankruptcy proceedings to improvements
in the functioning of the judiciary; from raising accounting and disclosure stan-
dards to establishing the appropriate legal framework and right incentives for the
development of debtor information systems. Improvements in the contractual and
informational environment facilitate the expansion of arm’s-length financing: eq-
uity and debt contracts that are more impersonal in nature and that, therefore, rely
more on transparency and enforcement rules of general application.
A final class of market-developing policies aims at raising the possibilities fron-
tier by widening the scope for idiosyncratic risk diversification and thus, essentially,
completing markets. To a significant extent, policies in this category coincide with
those that aim at fostering macroeconomic stability. A minimum of macro stability,
combined with sound management of the public debt and actions to the develop the
local currency debt market are key in the formation of a reliable yield curve, which
is in turn crucial to the development of interest rate derivatives. Similarly, macro
stability and a flexible exchange rate regime are necessary in the development of
the market for currency derivatives. In small and undiversified economies, how-
ever, these policies might not suffice to give rise to markets for hedges and certain
insurance products. In such cases, a premium should be placed on accessing the
vast, yet vastly unexploited, risk pooling and diversification opportunities offered
by international capital markets—which range potentially from catastrophic insur-
ance to commodity price hedges, from weather and crop insurance to securitization
of export receivables, and from currency swaps to GDP-indexed securities. 50

49
There is considerable empirical evidence showing the adverse effects of a weak contractual and
informational environment on credit depth and access. See Beck and Levine (2005) for an overview and
IDB (2005) for an extensive survey of issues regarding creditor rights and debtor information systems
in Latin America, providing ample evidence of the adverse role of poor creditor rights systems on
credit depth, volatility, and access. There is also considerable empirical evidence showing that smaller,
younger and more opaque firms are more financially constrained in economies with less developed
financial and legal institutions and with less effective systems of credit information sharing (Beck,
Demirguc-Kunt, and Maksimovic, 2005; Beck, Demirguc-Kunt, Laeven, and Maksimovic, 2006; Love
and Mylenko, 2003).
50
For examples of weather insurance, see Hess (2003) and Gine, Townsend, and Vickery (2007). See
OECD (2005) for a collection of papers on catastrophe insurance and Auffret (2003) for a discussion
of the catastrophic insurance market in the Caribbean. Schiller (2003) provides a grand vision of
the potential for financial markets to aid in long-term and international risk management, including
reducing “gratuitous inequality”, the feasibility of which is within reach given recent development in
financial theory, information technology, and the science of psychology.
 The Basic Analytics of Access to Financial Services 109

Changes in the state variables involve changes in fundamental institutions and,

thus, take a long time to materialize. That does not necessarily mean, however,
that that there is no scope for policy in the shorter-run. To the extent that a fi-
nancial system is operating below the possibilities frontier, there is room for
market-enabling policies that may foster the broadening of access even in the
absence of perceptible changes in state variables. Where the main reason for be-
ing below the possibilities frontier is the demand problem of self-exclusion, the
appropriate policies would emphasize raising financial literacy. If—as is more
likely—the main problems reside with sub-optimization in credit supply, by con-
trast, a wider range of policy options can be considered, starting with competition
policy.
Competition is one obvious area for market-enabling policies aimed at enhanc-
ing supply outreach. Greater competition can result in efficiency gains that would
move the system closer to the possibilities frontier. The salutary effects of greater
competition are illustrated, for instance, by the recent vigorous expansion of sus-
tainable and profitable micro- and consumer lending in emerging markets. It was
spurred in many cases by the removal of unnecessary regulatory distortions to
competition—including, for instance, allowing the use of scoring methods in credit
risk management, calibrating entry capital requirements for microfinance institu-
tions, abolishing of interest rate caps and providing regulatory incentives to the use
of credit bureaus in screening debtors. 51 As competition thrived in the microcredit
market, it stimulated innovation in loan products and the maximum use of available
technologies, with consequent improvements in cost and risk management. The
result has been a sustained broadening of access to micro credit, even where state
variables have changed little. 52
Beyond targeting competition per se, market-enabling policies can also try pro-
duce a movement towards the possibilities frontier by addressing hindrances such
as coordination failures, first mover disincentives, and obstacles to risk distribu-
tion and sharing. While not easy to define in general terms, given their variety,
these government interventions tend to share a common feature in creating incen-
tives for private lenders and investors to step in, without unduly shifting risks and

51
In its early stages of the development, microfinance tended to mitigate agency problems through
group monitoring, i.e., lending to groups of borrowers rather than individuals with each member liable
for the group loan. With the maturation of the microfinance industry, individual lending has become
dominant. As regards collateral, rather than taking traditional collateral, microfinance institutions often
take security on crucial consumer durables such as TVs and refrigerators, since these have a higher
personal value for borrowers and thus reduce ex-post moral hazard problems. See Armendariz de
Aghion and Morduch (2005) for an overview.
52
Christen (2000) discusses microfinance’s evolution from an NGO-based grant-funded activity to
a vigorous and self-sustaining industry and associated “mission drift.” Otero and Marulanda (2005)
examine mayor trends and features in microfinance in Latin America, noting the simultaneous phe-
nomenon of “upscaling” (the transformation of NGOs into regulated entities) and “downscaling” (the
entering of well-established commercial banks into the microfinance business). The current status
of microfinance and the opportunities and challenges going forward are discussed in Helms (2006).
Microfinance case studies include Navajas, Schreiner, Meyer, Gonzalez-Vega, and Rodriguez-Meza
(2002) for Bolivia; Benavente (2006) for Chile, and Zaman (2004) for Bangladesh.
110 Thorsten Beck and Augusto de la Torre

costs to the government. 53 For all the questions these market promoting interven-
tions raise—and their systematic study is at a too early stage to provide definitive
answers—they clearly stand in sharp contrast with the traditional and typically
ill-fated market-substituting policies consisting of government-directed lending at
subsidized interest rates or through government-owned institutions.
Given the central role of risk in credit services, there is a third category of
government policies related to access, which we define as market-harnessing and
which try to prevent the financial system from moving to an unsustainable equilib-
rium beyond the frontier due to imprudent lending. Such imprudent lending binges
can arise from the same competition that market-enabling policies try to foster if
not accompanied by a proper defined regulatory and supervisory safety net. Indis-
criminate free entry for new deposit-taking credit institutions or the intensification
of competition among existing institutions can be a recipe for a banking crisis
in waiting, in the context of implicit or explicit government guarantees, poor ac-
counting and disclosure practices, deficient early warning system and prompt cor-
rective action regimes and dysfunctional failure resolution frameworks. 54 Market-
harnessing policies therefore aim at keeping banks’ incentives to take aggressive
risks in check through a mix of measures aimed at strengthening market and su-
pervisory discipline. The absence of too generous deposit insurance—implicit or
explicit—and disclosure and transparency requirements give large creditors and
depositors incentives and possibilities to monitor and discipline banks. In addition,
market signals in the form of deposit interest rates, yields on subordinated debt or
equity prices of publicly listed banks moving in response to risk taking and per-
formance of banks provide additional information to bank supervisors and should
be coupled with effective official intervention into institutions that the market has
identified as weak. Market-harnessing policies are also important on the demand
side to avoid predatory lending, which results in unsustainable overborrowing by
individual borrowers. Predatory lending practices flourish in environments where
borrowers face information gaps, imperfect competition and inadequate legal

53
An analysis of this type of “market-friendly roles for the visible hand” is found in De la Torre
and Schmukler (2006a), which presents case studies of such intriguing examples as: (i) the creation
by NAFIN (a Mexican development bank) of an internet-based market for the discounting of post-
delivery receivables by SMEs; (ii) a Chilean program (FOGAPE) to promote lending to SMEs via
the auctioning of partial government guarantees; and (iii) a variety of structured finance packages
orchestrated by FIRA (a Mexican development fund) to finance agricultural production (e.g., shrimp,
corn); these packages succeed in involving, through risk sharing arrangements and partial guarantees,
key stakeholders (multinational commercialization firm, suppliers, producers, banks, etc.) so to generate
finance that would have otherwise not materialized.
54
See Carletti and Hartmann (2003) for an overview over the theoretical and empirical literature
on stability and competition. De Juan (2002) has a classical discussion of how, under conditions of
unsound competition, good bankers turn into bad bankers. See also De la Torre et al. (2001) and De
Krivoy (2000) for a discussion of how the weak application of fit-and-proper criteria at entry as well as
deficient official monitoring led to unhealthy competition and banking crises in the cases of Ecuador
and Venezuela, respectably. See Gavin and Hausmann (1996) for a discussion that even in the absence
of moral hazard and deficient supervision competition can lead to crisis, as “good times are bad times
for learning,” in the sense that information asymmetries intensify during credit booms, which often
result in crises.
 The Basic Analytics of Access to Financial Services 111

consumer protection. Disclosure requirements and education programs can help

reign in loan sharks, while usury laws are much more difficult to calibrate. 55 In
all, market-enabling policies focused on enhancing competition must ensure that
gains in efficiency and access have to be coupled with market-harnessing policies
that keep banks’ incentives to take aggressive risks in check.

IV. CONCLUDING REMARKS

This paper introduced a new tool to evaluate the outreach of a country’s financial
system and help design policies to increase prudent access and outreach. On the
supply side we have focused on the fixed component of transaction costs and on
lending risk as barriers for the financial system to reach out to clients with demand
for low-value payment and savings transactions and to riskier borrowers. We dis-
tinguished between potential supply denoting the maximum outreach a financial
system can provide given state variables such as the contractual and informational
frameworks, the macroeconomic environment, technology and other country char-
acteristics, and actual supply, which takes into account market structure and con-
testability resulting in no financial institution working at the frontier and catering
to marginal customers. On the demand side, we distinguished between potential
demand as predicted by economic factors and actual demand, which takes into ac-
count voluntary self-exclusion resulting in limited or no use of financial services
by some customers. In the case of payment and savings services we defined the
access possibilities frontier as the intersection of potential demand and supply and
thus the bankable population as the share of population that could be served by
the financial system given constraints imposed by the state variables and demand
as defined by economic factors. Starting from this frontier, we defined three dif-
ferent access problems: first the lack of demand due to voluntary self-exclusion;
second, supply of financial services below the potential due to lack of competition
or other supply side constraints and third a frontier that is too low in international
comparisons and explained by the state variables. Since credit risk is central to
access to lending services, there is an additional possibility of excessive access
due to imprudent lending practices.
The Access Possibilities Frontier has important implications for the debate on
how to expand access to financial services. It shows the potential for private solu-
tions to expand access to financial services, but also points to an important role of
the government in fostering such private solutions through competition and regu-
latory policies and through interventions characterized as private-public solutions.
However, it also shows the limit to these efforts due to systemic deficiencies in
the contractual, macroeconomic and informational environment that prevent the
financial system from expanding access beyond a very low frontier. Finally, in the
case of lending services, we discussed the role of government in harnessing market
forces to avoid excessive imprudent access to credit.
55
See Honohan (2004c) for a more detailed discussion.
112 Thorsten Beck and Augusto de la Torre

The concept of the Access Possibilities Frontier is also useful for empirical
attempts to measure outreach and access, bankable and banked population, for
specific financial products. 56 A combination of household and supplier surveys
can help compute demand and supply side factors and constraints, as we want to
illustrate with the following example for the case of savings and payment services.
First, household surveys can be used to determine the share of the population that
uses a specific service, such as checking account, savings account, or ATM ser-
vices (Point III in Graph 2). Second, using survey questions on the reasons why
certain people do not use a specific service, one can identify the (i) unbankable
(no sufficient income or economic activity), (ii) unbanked due to voluntary exclu-
sion (by choice) and (iii) unbanked due to involuntary exclusion (access related
obstacles such as costs, lack of necessary documentation, distance etc.). Adding
the banked population, the unbanked by choice and the unbanked due to invol-
untary exclusion, one can get to a first approximation to the bankable population
(point I in Graph 2) or the frontier. The relative shares of unbanked due to volun-
tary and involuntary exclusion also allow quantifying the effect to be expected by
market-enabling policies aimed at the demand and supply side. 57
The tool of access possibilities frontier can also be used to assess the impact of
cost- or risk-reducing technological change and of changes in state variables such as
the legal system on the bankable population. The asset holdings of enterprises and
households can be compared to collateral requirements of financial institutions
and as impacted by the contractual framework. If legal system reforms allow a
“new” asset (e.g., movable assets) to be used as collateral, enterprise and household
surveys can be used to calculate the share of potential borrowers that would become
bankable as they are now able to use their assets as collateral, thus advising policy
makers on the binding constraint and most beneficial reform in terms of higher
outreach.
While we have developed and discussed the access possibilities frontier on a
conceptual level, we hope to have convinced the reader that it can very well be
operationalized for both empirical and policy work. Disentangling different access
problems and the impact of policy reforms is much more difficult for lending
services than for payment and savings services, as the assessment and management
of risk adds a crucial additional element. Nevertheless, we are confident that this
tool can be refined further to serve both the debate on the measurement of access
to finance as well as the policy debate on how to improve outreach of the financial
system and access to financial services.

56
Porteous (2005) provides an example for such a calculation for the Mzansi account in South Africa,
while Melzer (2006) offers an example for the housing loan market in South Africa.
57
Alternatively, the bankable population could be computed through estimates of the costs of delivering
certain payment services, affordability assumptions (x% if monthly income) and information on the
income distribution in the country. These estimates can then be compared to a supplier survey on the
actual prices for the same products to calculate the share of the unbanked population that is excluded
due to supply constraints. The notorious unreliability of such cost estimates, due to the important role
of overhead costs in financial institutions, is an important caveat for such an exercise.
 The Basic Analytics of Access to Financial Services 113

V. REFERENCES
Armendariz de Aghion, B. and J. Morduch. 2005. The Economics of Microfinance.
Cambridge, MA: MIT Press.
Ashraf, N., D. Karlan, and W. Yin. 2006a. “Tying Odysseus to the Mast: Evidence
from a Commitment Savings Product in the Philippines.” Quarterly Journal of
Economics 121:636–72.
Ashraf, N., D. Karlan, and W. Yin. 2006b. “Household Decision Making and
Savings Impact: Further Evidence from a Commitment Savings Product in the
Philippines.” Unpublished paper, Yale University.
Auffret, P. 2003. “Catastrophe Insurance Market in the Caribbean Region. Market
Failures and Recommendations for Public Sector Interventions.” Unpublished
paper, The World Bank.
Beck, T. 2006. “Creating an Efficient Financial System: Challenges in a Global
Economy.” Pp. 193–224 in Economic Growth, eds. South African Reserve Bank,
Banco de Mexico, and The People’s Bank of China.
Beck, T., A. Demirguc-Kunt, and R. Levine. 2007. “Finance, Inequality and the
Poor: Cross-Country Evidence.” Journal of Economic Growth, forthcoming.
Beck, T., A. Demirguc-Kunt, and V. Maksimovic. 2004. “Bank Competition, Fi-
nancing Constraints, and Access to Credit.” Journal of Money, Banking and
Credit 36:627–648.
Beck, T., A. Demirguc-Kunt, and V. Maksimovic. 2005. “Financial and Legal
Constraints to Firm Growth: Does Firm Size Matter?” Journal of Finance 60,
137–177.
Beck, T., A. Demirgüç-Kunt, L. Laeven, and V. Maksimovic. 2006. “The Deter-
minants of Financing Obstacles.” Journal of International Money and Finance
25:6:932–952.
Beck, T., A. Demirguc-Kunt, and M. Martinez Peria. 2006. “Banking Services for
Everyone? Barriers to Bank Access around the World.” Unpublished paper, The
World Bank.
Beck, T., A. Demirguc-Kunt, and M. Martinez Peria. 2007. “Reaching Out: Ac-
cess to and Use of Banking Services across Countries.” Journal of Financial
Economics, forthcoming.
Beck, T., R. Levine, and Loayza. 2000. “Finance and the Sources of Growth.”
Journal of Financial Economics 58:261–300.
Beck, T. and R. Levine. 2005. “Legal Institutions and Financial Development.”
Pp. 251–278 in Handbook of New Institutional Economics, eds. C. Menard and
M. Shirley. Dordrecht: Springer
Benavente. 2006. Programa de Crédito para la Microempresa. Background paper
for De la Torre and Schmukler. (2006a). Innovative Experiences in Access to
Finance: Market Friendly Roles for the Visible Hand, Latin America Regional
Study. The World Bank.
Berger, A., S. Frame, and N. Miller. 2005. “Credit Scoring and the Availability,
Price, and Risk of Small Business Credit.” Journal of Money, Credit and Banking
37:191–222.
114 Thorsten Beck and Augusto de la Torre

Berger, A. and D. Humphrey. 1994. “Bank Scale Economies, Mergers, Concen-

tration, and Efficiency: The U.S. Experience.” Unpublished paper, Board of
Governors of the Federal Reserve System.
Bertrand, M., D. Karlan, S. Mullainathan, E. Shafir, and J. Zinman. 2005. “What’s
Psychology Worth? A Field Experiment in the Consumer Credit Market.” Un-
published paper, University of Chicago.
Bossone, B., P. Honohan, and M. Long. 2002. “Policy for Small Financial Sys-
tems.” Pp. 95–128 in Financial Sector Policy for Developing Countries: A
Reader, eds. G. Caprio, P. Honohan, and D.Vittas. Oxford University Press.
Carletti, E. and P. Hartmann. 2003. “Competition and Stability: What’s Special
about Banking?” Pp. 202–229 in Monetary History, Exchange Rates and Finan-
cial Markets: Essays in Honor of Charles Goodhart, ed P. Mizen. Cheltenham,
UK: Edward Elgar.
Caskey, J., C. Ruiz Duran, and T. Solo. 2006. “The Urban Unbanked in Mexico
and the United States.” Unpublished paper, The World Bank.
Christen, R. P. 2000. “Commercialization and Market Drift: The Transformation
of Microfinance in Latin America.” CGAP Occasional Paper 5.
Claessens, S. 2006. “Access to Financial Services: A review of the Issues and
Public Policy Objectives.” World Bank Research Observer 21:207–240.
Claessens, S., G. Dobos, D Klingebiel, and L. Laeven. 2003. “The Growing Impor-
tance of Networks in Finance and its Effects of Competition.” Pp. 110–135 in
Innovations in Financial and Economic Networks, ed. A. Nagurney. Northamp-
ton, MA: Edward Elgar Publishers.
Claessens, S., T. Glaesner, and D. Klingebiel. 2001. “Electronic Finance: Reshap-
ing the Financial Landscape around the World.” Journal of Financial Services
Research 22:29–61.
Claessens, S. and L. Laeven. 2004. “What Drives Banking Competition? Some
International Evidence.” Journal of Money, Banking and Credit 36:563–583.
Copeland, T. E., J. F. Weston, and K. Shastri. 2005. Financial Theory and Corporate
Policy. Addison Wesley.
De Juan, A. 2002. “From Good Bankers to Bad Bankers.” Pp. 19–30 in Financial
Sector Policy for Developing Countries: A Reader, eds. G. Caprio, P. Honohan,
and D.Vittas. Oxford University Press.
De Krivoy, R. 2000. Collapse: The Venezuelan Banking Crisis of 1994. Group of
Thirty
De la Torre, A., R. Garcia-Saltos, and Y. Mascaro. 2002. “Banking, Currency, and
Debt Meltdown: Ecuador Crisis in the Late 1990s.” Unpublished paper, The
World Bank.
De la Torre, A. and S. Schmukler. 2005. “Coping with Risks Through Mismatches.”
International Finance 7:1–42.
De la Torre, A. and S. Schmukler. 2006a. “Innovative Experiences in Access to
Finance: Market Friendly Roles for the Visible Hand.” Latin America Regional
Studies Series. The World Bank.
De la Torre, A., J. C. Gozzi, and S. Schmukler. 2006b. “Capital Market Develop-
ment. Whither Latin America?” Unpublished paper, The World Bank.
 The Basic Analytics of Access to Financial Services 115

Demirguc-Kunt, A., L. Laeven, and R. Levine. 2004. “Regulations, Market Struc-

ture, Institutions, and the Cost of Financial Intermediation.” Journal of Money,
Credit, and Banking 36:593–622.
Financial Services Authority. 2000. “ In or Out? Financial Exclusion: A Literature
and Research Review.” Consumer Research 3.
Frame, S., M. Padhi, and L. Woosley. 2001. “The Effect of Credit Scoring on Small
Business Lending.” Financial Review 39:34–54.
Gavin, M. and R. Hausmann. 1996. “The Roots of Banking Crises: The Macroe-
conomic Context.” Unpublished paper, Inter-American Development Bank.
Gine, X. and S. Klonner. 2005. “Credit Constraints as Barrier to Technology Adop-
tion by the Poor: Lessons from South-Indian Small-Scale Fishery.” Unpublished
paper, The World Bank.
Gine, X., R. Townsend, and J. Vickery. 2007. “Rainfall Insurance Participation in
India.” Unpublished paper, The World Bank.
Hausmann, R., D. Rodrik, and A. Velasco. 2005. “Growth Diagnostics.” Unpub-
lished paper, Harvard University.
Helms, B. 2006. “Access for All: Building Inclusive Financial Systems.” Unpub-
lished paper, The World Bank.
Hess, U. 2003. “Innovative Financial Services for Rural India: Monsoon-Indexed
Lending and Insurance for Smallholders.” Unpublished paper, The World Bank.
Honohan, P. 2004a. “Financial Development, Growth and Poverty: How Close
Are the Links?” Pp. 1–37 in Financial Development and Economic Growth:
Explaining the Links, ed. C. Goodhart. London: Palgrave.
Honohan, P. 2004b. “Measuring Microfinance Access: Building on Existing Cross-
country Data.” Unpublished paper, The World Bank.
Honohan, P. 2004c. “Financial Sector Policy and the Poor. Selected Findings and
Issues.” Unpublished paper, The World Bank.
Honohan, P. 2006. “Household Financial Assets in the Process of Development.”
Unpublished paper, The World Bank.
Inter-American Development Bank. 2004. Unlocking Credit. The Quest for Deep
and Stable Bank Lending. Maryland: The Johns Hopkins University Press.
Karlan, D. and J. Zinman. 2004. “Observing Unobservables: Identifying Infor-
mation Asymmetries with a Consumer Credit Field Experiment.” Unpublished
paper, Yale University.
Kempson, E., A. Atkinson, and O. Pilley. 2004. “Policy Level Response to Finan-
cial Exclusion in Developed Economies: Lessons for Developing Countries.”
Unpublished paper, University of Bristol.
Kumar, A. 2005. “Access to Financial Services in Brazil.” Unpublished paper, The
World Bank.
Levine, R. 1997. “Financial Development and Economic Growth: Views and
Agenda.” Journal of Economic Literature 35-688-726.
Levine, R. 2005. “Finance and Growth: Theory and Evidence.” Pp. 865–934 in
Handbook of Economic Growth, eds. P. Aghion and S. Durlauf. The Netherlands:
Elsevier Science.
116 Thorsten Beck and Augusto de la Torre

Love I. and N. Mylenko. 2003. “Credit Reporting and Financing Constraints.”

Unpublished paper, The World Bank.
Melzer, I. 2006. “How Low Can You Go? Charting the Housing Finance Access
Frontier: A Review of Recent Demand and Supply Data.” Unpublished paper,
Finmark Trust.
Napier, M. 2005. “Engaging the Private Sector: The Case for Financial Access
Charters in Sub-Saharan Africa.” Unpublished paper, Finmark Trust.
Navajas, F., M. Schreiner, R. L. Meyer, C. Gonzalez-Vega, and J. Rodriguez-
Mesa. 2000. “Microcredit and the Poorest of the Poor: Theory and Evidence
from Bolivia.” World Development 28:333–346.
North, D. 1990. Institutions, Institutional Change, and Economic Performance.
Cambridge, UK: Cambridge University Press.
Organization for Economic Cooperation and Development. 2005. OECD Confer-
ence on Catastrophic Risks and Insurance, Paris, France.
Otero, M. and B. Marulanda. 2005. “The Profile of Microfinance in Latin America
in 10 Years: Vision and Characteristics.” Unpublished paper, Acción Interna-
tional.
Peachey, S. and A. Roe. 2004. “Access to Finance: A Study for the World Savings
Banks Institute.” Unpublished paper, WSBI.
Porteous, D. 2004. “Making Financial Markets Work for the Poor.” Unpublished
paper, Finmark Trust.
Porteous, D. 2005. “The Access Frontier as a Tool in Making Markets Work for
the Poor.” Unpublished paper, Finmark Trust.
Rajan, R. and L. Zingales. 2003. Savings Capitalism from the Capitalists. New
York: Crown Business.
Shiller, R. J. 2003. The New Financial Order: Risk in the 21st Century. Princeton,
NJ: Princeton University Press.
Stiglitz, J. and A. Weiss. 1981. “Credit Rationing in Markets with Imperfect In-
formation.” American Economic Review 71:393–410.
Williamson, S. D. 1987. “Costly Monitoring, Loan Contracts and Equilibrium
Credit Rationing.” Quarterly Journal of Economics 102:135–46.
World Bank. 2004a. “India—Scaling-Up Access to Finance for India’s Rural Poor.”
Unpublished paper, The World Bank.
World Bank. 2004b. “South Africa: Technology and Access to Financial Services.”
Unpublished paper, The World Bank.
World Bank. 2007. “Making Finance Work for Africa.” Unpublished paper, The
World Bank.
Zaman, H. 2004. “The Scaling-Up of Microfinance in Bangladesh: Determinants,
Impact and Lessons.” Unpublished paper, The World Bank.

VI. NOTES ON CONTRIBUTORS/ACKNOWLEDGMENTS

Thorsten Beck is a senior economist in the Finance and Private Sector Develop-
ment Team of the Development Research Group of the World Bank. His research,
 The Basic Analytics of Access to Financial Services 117

academic publications and operational work have focused on two major questions:
What is the relationship between finance and economic development? What poli-
cies are needed to build a sound and effective financial system? Recently, he has
concentrated on access to financial services, including SME finance, as well as the
incentive-compatible design of financial safety nets. He has published numerous
academic papers and is co-author of the Making Finance Work for Africa Flag-
ship report and the forthcoming Policy Research Report on Access to Finance:
Measurement, Impact and Policy. His country experience, both in operational and
research work, includes Bangladesh, Bolivia, Brazil, China, Mexico, Russia and
several countries in Sub-Saharan Africa. He holds a Ph.D. from the University of
Virginia and an M.A. from the University of Tübingen in Germany.

Augusto de la Torre has worked for the World Bank since October 1997. He has
been running the Financial Systems Unit of FPD since the fall of 2006. Prior to his
current position, he worked as Senior Regional Financial Sector Advisor for Latin
America and the Caribbean and provided technical and conceptual leadership to
World Bank financial sector operations and research in the Region. He is actively
involved in the joint IMF-World Bank Financial Sector Assessment Program. Be-
fore joining the Bank he headed the Central Bank of Ecuador, from 1993 to 1996.
In November 1996, he was chosen by Euromoney Magazine as the year’s “Best
Latin Central Banker.” He is a member of the Carnegie Network of Economic
Reformers. From 1986 to 1992 he was an Economist with the International Mon-
etary Fund and during 1991 and 1992 was the IMF’s Resident Representative in
Venezuela. Augusto de la Torre earned his M.A. and Ph.D. degrees in Economics
at the University of Notre Dame and holds a Licenciatura in Philosophy from the
Catholic University of Ecuador.

This paper has been prepared as part of the Latin American Regional Study on
Access to Finance. It has benefited from discussions and comments by Abayomi
Alawode, Ole Andreassen, Jerry Caprio, Stijn Claessens, Asli Demirguc-Kunt,
Patrick Honohan, Martin Jung, Soledad Martinez Peria, Sergio Schmukler, Tova
Solo, Robert Stone, as well as from comments and suggestions from participants
at various seminars in the World Bank, the Asobancaria seminar on Bancarización
in Cartagena 2004, the World Savings Banks Institute conference on Access to
Finance in Brussels 2004 and the Dia de la Competencia in Mexico City 2005.
This paper’s findings, interpretations, and conclusions are entirely those of the
authors and do not necessarily represent the views of the World Bank, its Executive
Directors, or the countries they represent.