Finance for all? Banking structure reform may not be the way

Thorsten Beck, Martin Brown 06 October 2010



Access to financial services is viewed as a key determinant of economic wellbeing, especially for households in low-income countries. Savings, credit, and insurance products make it easier for households to align income and expenditure patterns across time, to insure themselves against income and expenditure shocks, as well as to undertake investments in human or physical capital. Given the importance attributed to financial service access it is striking that there is little cross-country evidence which documents how financial access differs across households and, in particular how the structure of the financial sector affects which households are banked, and which are not (see World Bank 2007 for an overview).

In recent research we examine how bank-ownership, deposit insurance, payment systems development and creditor protection affect the composition of the banked population in 29 transition economies (Beck and Brown 2010). We find that foreign bank ownership, generous deposit insurance, better payment systems and creditor protection encourage high-income, well-educated, professional and urban households to use banking services. We find no evidence that state-bank ownership fosters the use of banking services by low-income, rural, or minority households.

The data

Our data are taken from the EBRD-World Bank Life in Transition Survey implemented in 2006. The survey covers 29 countries including 28 transition countries in which the EBRD operates and Turkey. In each country, 1,000 interviews were conducted with randomly selected households.

We employ two indicators of household use of banking services. The dummy variable Account measures whether any member of the household has a bank account. The dummy variable Card measures whether any member of the household has a bank (debit or credit) card. Figure 1 shows that there is substantial variation in the use of banking services across countries, with banked households much more common in Central Europe than in the CIS countries. More than 75% of households in Croatia, the Czech Republic, Estonia, Slovakia and Slovenia have a bank account, while less than 5% of households in Armenia, Azerbaijan, Kyrgyzstan, Tajikistan and Uzbekistan do so.

Figure 1. The share of households which have a bank account by country.

Source: Beck and Brown (2010)

Access to finance and household characteristics

In the first step of our empirical analysis we relate our indicators of banking service use to characteristics of the household. We conduct a pooled regression controlling for country level determinants with country-fixed effects, as well as country-by-country regressions. Our results largely confirm recent findings on household access to finance in Africa (Honohan and King 2009):

  • There is a strong income elasticity of bank services, with richer households significantly more likely to have a bank account. Raising household expenses by one standard deviation ($2,331) from the sample mean ($2,570) increases the probability of having a bank account by roughly 12% and that of having a bank card by 10%.
  • Wealth and geography matter as well. Homeowners are 3% more likely to have a bank account, although they are not more likely to have a bank card. This result, however, is driven only by a few countries. Urban households are 5% more likely to have a bank account and 8% more likely to have a bank card than rural households.
  • Households with formal employment are more likely to use bank services than those that rely on self-employment or transfer income. For example, households that rely on transfer income are 11% (15%) less likely to have a bank account (card) than households which have wage-income.
  • Households in which the responding adult has professional education are 9% more likely to have a bank account and 5% more likely to have a bank card, suggesting that literacy (and thus maybe also financial literacy) does affect bank use.
  • There is a significant impact of social status and religion on the use of banking services. Not speaking the official language reduces the likelihood of having a bank account by 8%. Being a Muslim reduces the probability of having a bank account / card by 8% and 5%, respectively. These findings are again, though, driven by a few countries.

Banking sector structure and the composition of the banked population

Country by country regressions show that the relation between household characteristics and use of banking services differs strongly across countries. The composition of the banked population may be primarily driven by general economic development of a country (e.g. GDP per capita). However, important for policy makers, it may also be related to the underlying structure of the banking sector.

In the second step of our empirical analysis we therefore assess how the ownership structure of banks (foreign and state-owned banks) as well as the development of the financial infrastructure (deposit insurance coverage, payment system development, creditor protection, credit information sharing) affects which households are banked. We focus on the interaction of household and country characteristics, i.e. on the effect of banking market structure and financial infrastructure. Our findings can be summarised as follows:

  • Foreign bank presence is positively associated with the use of banking products among high-income and well-educated households and households which have wage-income. These interactions are mostly driven by greenfield foreign banks rather than domestic bank taken over by international banks.
  • We find no evidence that state-bank ownership leads to a broader use of banking products among low-income or rural households.
  • Improvements in the financial infrastructure, i.e. higher deposit insurance coverage, better electronic payment system development, stronger creditor rights and more effective credit registries, are associated mostly with higher use of banking services among high-income and high-wealth households.

Policy conclusions

Our results shed doubt on the ability of policy levers to broaden the financial system to disadvantaged groups. Specifically, attempts to broaden the use of financial services through state-owned banks and deposit insurance do not disproportionally increase the likelihood that poorer, less wealthy, and socially less included segments of the population use formal financial services. Similarly, a better contractual and information framework seems to benefit mostly the higher-income and wealthy segments of the population, most likely by allowing the banks to differentiate more carefully among potential clients. Our results do not imply that these policies do not help improve financial sector development rather that it is difficult to target them to certain groups.


Beck, T and M Brown (2010). “Which Households use Banks: Evidence from the Transition Economies”, Tilburg University - European Banking Center Discussion Paper 2010-25.

Honohan, P and M King (2009), “Cause and effect of financial access: cross-country evidence from the Finscope surveys”, World Bank mimeo.

World Bank (2007), Finance for All? Policies and Pitfalls in Expanding Access. Washington, DC.



Topics:  Development Financial markets

Tags:  development, financial regulation, access to finance

Professor of Banking and Finance, The Business School (formerly Cass); Director, Florence School of Banking and Finance

Professor of Banking, University of St. Gallen


CEPR Policy Research