An extensive empirical literature has shown a positive relationship between financial development and economic growth, where the former is often captured by total credit outstanding to the private sector divided by GDP (private credit to GDP). More recently, however, several studies have focused on the non-linearity in the relationship between both variables and even a possible negative relationship between them at very high levels of private credit to GDP (Arcand et al. 2011, Ceccetti and Kharroubi 2015). While not arguing against these findings, this column urges caution in their interpretation.
From theory to measurement
Financial development has been traditionally measured by the ratio of private credit to GDP, the total outstanding claims of regulated financial intermediaries in a country on non-financial domestic enterprises and households, normalised by economic activity. It is important to understand, however, that far from catching all the subtle complexities of financial sector development, this is a crude proxy measure for financial development, as I will detail in the following.
To clarify this point, one has to go back to theory to understand how the financial sector can support economic growth. Economic theory has pointed to an array of different channels and mechanisms through which a well-developed and efficient financial system can support economic development:
- By providing payment services and reducing transaction costs, thus enabling the efficient exchange of goods and services as well as specialisation of labour;
- By pooling savings from many individual savers, thus helping overcome investment indivisibilities and allowing scale economies to be exploited;
- By economising on screening and monitoring costs, thus increasing overall investment and improving resource allocation;
- By helping reduce liquidity risk, thus enabling long-term investment;
- By helping diversify cross-sectional and inter-temporal risk.
It is important to stress, however, that theory is not unambiguous in its prediction of a positive impact of finance on growth. By offering more efficient deposit and lending services, a more developed financial system might actually depress aggregate savings, with negative repercussions for investment and growth (Bencivenga and Smith 1991). A more developed financial system might also draw talent away from the real sector (Philippon 2010, Bolton et al. 2011).
The empirical literature has not been able to map these different functions into very specific empirical measures. Critically, several of these functions are complementary to each other. For example, the pooling and intermediating of savings goes hand in hand with the screening and monitoring of borrowers. Similarly, pooling and intermediating savings implies liquidity transformation. By offering payment services, banks learn about potential borrowers and thus get better at screening and monitoring them.
In the absence of variables capturing the individual functions of the financial sector, the literature has used crude proxy variables focusing on the size and activity of financial intermediaries and markets, most prominently the ratio of private credit to GDP. By capturing the total amount of financial intermediation by regulated intermediaries in the economy, this variable is a proxy for the total volume of intermediation. The main advantage of this indicator is that it is available for a large cross-section of countries and over longer time periods, as the underlying data (total credit outstanding and GDP) go back until 1960 for many countries. Confidence in the measure is further increased as it is significantly correlated with other indicators of the efficiency of the financial system, available for fewer countries and over shorter time periods, such as interest rate spreads and margins.
Given its crude nature, there are clear shortcomings in the private credit-to-GDP measure. It indicates quantity not quality, and focuses only on regulated financial institutions. It does not capture the maturity structure. It does not capture how widespread the use of credit services is among enterprises and households and the ease with which enterprises and households can access credit.
Importantly, it is not clear that there is a linear mapping from higher levels of private credit to GDP into more efficient and developed financial markets. Credit provision in an economy fluctuates substantially with the business cycle (Bernanke and Gertler 1989). Short-term variations in private credit to GDP for a given country are thus unlikely to reflect changes in the efficiency and development of financial markets and institutions. More importantly, credit cycles are often related to asset price cycles, so that rapid increases in private credit to GDP might reflect credit bubbles rather than rapid improvements in the efficiency and development of financial systems.
Numerous cross-country studies using private credit-to-GDP have shown a positive, though non-linear relationship between the measure and economic growth, using aggregate, industry-level, and firm-level data.1 It is important to note that many if not most of these studies directly account for a non-linear relationship between finance and growth by including private credit-to-GDP in logs rather than level, under the assumption that an increase from 20% to 40% has the same effect on credit as an increase from 40% to 80% or from 80% to 160%.
A parallel literature, however, has also shown that rapid growth in credit to the private sector is a significant predictor of banking crises. Systemic bank fragility, especially in middle- and high-income countries, is often preceded by credit booms, often associated with asset price bubbles. As the asset price cycle turns, credit booms turn to credit busts and ultimately large-scale non-performing assets and bank failures. Banking crises pose large economic costs and come more often than not with deep recessions, as documented by Laeven and Valencia (2010) and Claessens et al. (2013). Rapidly increasing private credit can turn into public credit in case of bailouts, as seen in several European countries, which ultimately is growth reducing.
While the finance-growth and banking crisis literatures have moved mostly in parallel, there have been some efforts to integrate these two effects and also to consider trade-offs. Just to give one example, Loayza and Ranciere (2006) show long-term positive effects of finance but negative short-term effects of rapid credit growth. Critical in this context is the question of what is ‘short term’ and what is ‘long term’. Growth regressions focus typically on five-year or longer periods to abstract from business cycles. Financial cycles, however, are typically longer, so that even averaging over five-year periods might only imperfectly control for such cycles.
Looking behind private credit-to-GDP
Banks’ balance sheets look very different across countries at different income levels. In low- and lower-middle income countries, banks focus often heavily on government bond holdings and corporate lending, often short term, with limited lending to small and medium-sized enterprises. Often very liquid, banks undertake little long-term financing and little consumer lending. In upper-middle income countries, consumer lending becomes more prominent, as does SME lending. As we move into high-income countries, banks’ balance sheets are dominated by anything but private sector lending (Langfield and Pagano 2015, Demirguc-Kunt and Huizinga 2010). And behind private credit-to-GDP is mostly household rather than enterprise lending in high-income countries.
In summary, as we move up the income level of countries, private credit-to-GDP – and more specifically enterprise credit – becomes first a more and then a less accurate indicator of banks’ business focus. For high-income countries, the traditional measure of financial development is thus less relevant than for developing and emerging markets.
Finance and growth in high-income countries – caveat emptor!
As I have written in a previous column (Beck 2013), there are many reasons why the growth in the financial sector in high-income countries has not provided any growth benefits, or has even damaged growth. The increasing importance of household, especially mortgage credit, in private sector lending is one important reason. The focus on financial centres rather than intermediation is another, which together with the focus on ‘my country, my banks’ has led to oversized banking systems in many European countries (Langfield and Pagano 2015). Finally, politically fuelled lending booms – again mostly in the area of mortgage lending – have led to an over-expansion of the financial system. None of these, however, has anything to do with the efficiency of the financial system, even if they are reflected in higher levels of private credit to GDP.
Private credit as a target variable
Identifying a variable such as private credit-to-GDP as entering positively and significantly in a growth regression results in the temptation to turn it into a policy target. We know from Goodhart’s Law, however, that "when a measure becomes a target, it ceases to be a good measure" (Goodhart 1975). Specifically, an extensive literature has shown that cross-country variation in private credit-to-GDP can be explained by differences in macroeconomic stability and in the efficiency of the contractual and informational environment. Private credit-to-GDP – as a proxy for the development and efficiency of financial systems – has thus to be seen as an outcome variable rather than a target variable in itself. The question then becomes not about the level of private credit to GDP per se, but whether this level is consistent with the macroeconomic environment and institutional framework in a country. Attempts to benchmark a country’s indicators of financial development to an array of structural and policy variables take a first step towards answering this question, as done, for example, by Barajas et al. (2013) and De la Torre et al. (2013).
To paraphrase Winston Churchill, private credit-to-GDP is the worst indicator to measure financial development, except for all others that are available. When interpreting regression results using this indicator, utmost caution has to be applied to distinguish the efficiency of the intermediation process from other phenomena represented by this indicator. Yes, there can be too much finance, as many countries have found out the hard way in recent crises, but this is not the same as saying that financial systems can become too developed or too efficient. While the latter can be true, using aggregate indicators, such as private credit-to-GDP, will not serve to test this hypothesis.
Arcand, J L, E Berkes, and U Panizza (2011), "Too Much Finance?” VoxEU.org, 7 April.
Barajas, A, T Beck, E Dabla-Norris, and S Reza Yousefi (2013), “Too Cold, Too Hot, Or Just Right? Assessing Financial Sector Development Across the Globe,” IMF Working Paper 13/81
Beck, T (2009), “The Econometrics of Finance and Growth.” In Palgrave Handbook of Econometrics, vol. 2, ed. T Mills and K Patterson, 1180–1211, Houndsmill: Palgrave Macmillan.
Beck, T (2013), “Finance and Growth”, VoxEU.org, 27 October.
Bencivenga, V R, and B D Smith (1991), “Financial Intermediation and Endogenous Growth,” Review of Economics Studies 58, 195–209.
Bernanke, B S, M Gertler (1989), “Agency costs, net worth, and business fluctuations”, American Economic Review 79 (1), 14-31.
Bolton, P, T Santos and J Scheinkman (2011), “Cream Skimming in Financial Markets,” NBER Working Paper 16804.
Boyd, J H, R Levine, and B D Smith (2001), “The Impact of Inflation on Financial Sector Performance,” Journal of Monetary Economics 47, 221-48.
Cecchetti, St and E Kharroubi (2015), “Why growth in finance is a drag on the real economy”, VoxEU.org, 7 July.
Claessens, S and M Ayhan Kose (2013), “Financial Crises: Explanations, Types, and Implications”, in S Claessens, M A Kose, L Laeven, and F Valencia (eds.), Financial Crises, Consequences, and Policy Responses, IMF Publications.
De la Torre, A, E Feyen, and A Ize (2013), “Financial Development: Structure and Dynamics”, World Bank Economic Review 27, 514-541.
Demirgüç-Kunt, A and H Huizinga (2010), "Bank activity and funding strategies: The impact on risk and returns," Journal of Financial Economics 98, 626-650.
Goodhart, C A E (1975a) ‘Monetary Relationships: A View from Threadneedle Street’ in Papers in Monetary Economics, Volume I, Reserve Bank of Australia, 1975.
Langfield, S and M Pagano (2015), “Bank bias in Europe: Effects on systemic risk and growth”, Economic Policy 85.
Laeven, L and F Valencia (2011), “Resolution of Banking Crises: The Good, the Bad and the Ugly”, IMF Working Paper 10/146.
Levine, R (2005), “Finance and Growth: Theory and Evidence.” In Handbook of Economic Growth, ed. P Aghion and S N Durlauf, 865–934. Amsterdam: Elsevier.
Phillipon, T (2010), “Financiers vs. Engineers: Should the Financial Sector be Taxed or Subsidized?” American Economic Journal: Macroeconomics, 158–82.
1 See Levine (2005) for an overview of the theoretical and empirical literature and Beck (2009) for a survey of the different methods applied.