A reassessment of intermediation and size effects of financial systems


Several recent studies on the finance–growth nexus highlight that too much financial development, as it has been established in many advanced economies, harms growth. Beck et al. (J Financ Stab 10:50–64, 2014) criticize this literature for only focusing on intermediation activities of financial systems, even though financial sectors in advanced countries have extended their scope beyond traditional tasks. In line with this argument, Beck et al. find for a panel of high-income countries that financial sector size and non-intermediation activity stimulate growth, while intermediation activity has no effect. However, they focus only on OLS regressions with a very limited number of control variables. We test for the robustness of these results. Our findings show that they depend on outliers and are not robust against alternative specifications or estimation approaches. Further, a big financial sector and too many non-intermediation activities are found to reduce growth in some specifications. Our results suggest that Beck et al.’s criticism of the “too much finance” literature is grounded on thin empirical evidence.

This is a preview of subscription content, access via your institution.

Fig. 1


  1. 1.

    Arcand et al. (2012) show that this finding also holds for industry-level data.

  2. 2.

    Beck et al. (2014) drop all observations after 2007. They justify this as follows: “Although the potential instability associated with a large financial sector is central to our argument, we exclude the recent crisis from the sample period in order to be able to draw more general conclusions. Given the sudden large output declines which are reflected in the data as from 2008, results would be dominated by this event.” (Beck et al. 2014, p. 53) This is to say, they include the observations covering the years of the buildup of the financial bubble, where financial development went hand in hand with high but unsustainable growth, but exclude the period of the correction of the bubble. This seems problematic; such a decision potentially biases the results.

  3. 3.

    This is done to sweep out business cycle fluctuations from the data. Because proxies for financial development are highly pro-cyclical, it is important to address this issue. Sturn and Epstein (2014) show that 5-years averaging does not successfully sweep out business cycle effects, and therefore potentially biases results.

  4. 4.

    To see why, consider that Growth is defined as \(\varDelta y=y_{t}-y_{t-1}\), where y is GDP in logarithms. Thus, Eq. (1) can be rewritten as \(y_{t}-y_{t-1}={\upbeta }_{1}y_{t-1}+{\upbeta }_{2}W_{it}+{\upvarepsilon }_{it}\), which is identical to \(y_{t} = ({\upbeta }_{1}+1)y_{t-1}+{\upbeta }_{2}W_{it}+{\upvarepsilon }_{it}\) (see Bond et al. 2001).

  5. 5.

    We choose the Blundell and Bond (1998) estimator, designed for highly persistent series as in our case, to initialize the bias correction. However, our results are robust against the use of the Arellano and Bond (1991) estimator instead. Following standard practice when applying this estimator, we estimate standard errors using a parametric bootstrap method (see Bruno 2005b) with 400 resamples. To avoid loosing one observation per country when applying this dynamic estimator, we merge data on real GDP in local currency units from the World Development Indicators (WDI). Thus, Initial GDP in specifications applying the bias-corrected LSDV estimator slightly differs from Initial GDP in the OLS specifications, where it is defined as real GDP per capita in US dollars as in the original data set. All our central findings also hold when applying the uncorrected least-squares dummy variable estimator on the original data set. When applying the system GMM estimator, the coefficients are very imprecisely estimated and all variables measuring financial development are statistically insignificant in all specifications.

  6. 6.

    Beck et al. (2014) perform this analysis also for a slightly different and smaller sample of 27 developed countries (see their Table 4, Panel A). Our results also hold for this smaller sample, but do not provide additional information, and are thus not reported.

  7. 7.

    Luxembourg is also an outlier regarding Employment share, Hours share, and Compensation share.

  8. 8.

    This result is confirmed if estimated by the fixed effects estimator, whereas the thresholds are found to lie at 2.5 and 2.8, respectively.

  9. 9.

    The Hansen tests never reject the null, and thus provide support for the validity of the instruments. All regressions reject the null of no first-order autocorrelation, and do not reject the null of no second-order autocorrelation.

  10. 10.

    We treat Education, Openness, and Government Consumption as exogenous. We also experimented with various other specifications—e.g., treating all explaining variables as endogenous, allowing for more or less lags as instruments, collapsing the instrument matrix—with qualitatively similar results.

  11. 11.

    This result also holds when applying the OLS or fixed effects estimator.


  1. Aghion P, Howitt P, Mayer-Foulkes D (2005) The effect of financial development on convergence: theory and evidence. Q J Econ 120(1):173–222

    Google Scholar 

  2. Arcand J-L, Berkes E, Panizza U (2012) Too much finance? IMF Working Paper 12/161, IMF, Washington

  3. Arellano M, Bond S (1991) Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. Rev Econ Stud 58(2):277–297

    Article  Google Scholar 

  4. Arellano M, Bover O (1995) Another look at the instrumental variable estimation of error-components models. J Econom 68(1):29–51

    Article  Google Scholar 

  5. Beck T, Büyükkarabacak B, Rioja FK, Valev NT (2012) Who gets the credit? And does it matter? Household vs. firm lending across countries. B.E J Macroecon 12(1):1–46

    Article  Google Scholar 

  6. Beck T, Degryse H, Kneer C (2014) Is more finance better? Disentangling intermediation and size effects of financial systems. J Financ Stab 10:50–64

    Article  Google Scholar 

  7. Beck T, Levine R (2004) Stock markets, banks, and growth: panel evidence. J Bank Finance 28(3):423–442

    Article  Google Scholar 

  8. Bezemer D, Grydaki M, Zhang L (2014) Is financial development bad for growth? SOM Research Report 14016

  9. Blundell R, Bond S (1998) Initial conditions and moment restrictions in dynamic panel data models. J Econom 87(1):115–143

    Article  Google Scholar 

  10. Bond SR, Hoeffler A, Temple J (2001) GMM estimation of empirical growth models. CEPR Discussion Papers 3048

  11. Bruno GSF (2005a) Approximating the bias of the LSDV estimator for dynamic unbalanced panel data models. Econ Lett 87(3):361–366

    Article  Google Scholar 

  12. Bruno GSF (2005b) Estimation and inference in dynamic unbalanced panel-data models with a small number of individuals. Stata J 5(4):473–500

    Google Scholar 

  13. Cecchetti S, Kharroubi E (2012) Reassessing the impact of finance on growth. BIS Working Papers 381

  14. Favara G (2003) An empirical reassessment of the relationship between finance and growth. IMF Working Paper 03/123, IMF, Washington

  15. Hung F-S (2009) Explaining the nonlinear effects of financial development on economic growth. J Econ 97(1):41–65

    Article  Google Scholar 

  16. Kneer C (2013) Finance as a magnet for the best and brightest: implications for the real economy. De Nederlandsche Bank Working Papers 392

  17. Law SH, Singh N (2014) Does too much finance harm economic growth? J Bank Finance 41(C):36–44

    Article  Google Scholar 

  18. Levine R (2005) Finance and growth: theory and evidence. In: Aghion P, Durlauf NS (eds) Handbook of economic growth, vol 1, part A, pp 865–934. Elsevier, Amsterdam

  19. Levine R, Loayza N, Beck T (2000) Financial intermediation and growth: causality and causes. J Monet Econ 46(1):31–77

    Article  Google Scholar 

  20. Masten AB, Coricelli F, Masten I (2008) Non-linear growth effects of financial development: does financial integration matter? J Int Money Finance 27(2):295–313

    Article  Google Scholar 

  21. Nickell SJ (1981) Biases in dynamic models with fixed effects. Econometrica 49(6):1417–1426

    Article  Google Scholar 

  22. Philippon T, Reshef A (2013) An international look at the growth of modern finance. J Econ Perspect 27(2):73–96

    Article  Google Scholar 

  23. Roodman D (2009) A note on the theme of too many instruments. Oxf Bull Econ Stat 71(1):135–158

    Article  Google Scholar 

  24. Rousseau PL, Wachtel P (2011) What is happening to the impact of financial deepening on economic growth? Econ Inq 49(1):276–288

    Article  Google Scholar 

  25. Sturn S, Epstein GA (2014) Finance and growth: the neglected role of the business cycle. PERI Working Paper 339

  26. Windmeijer F (2005) A finite sample correction for the variance of linear efficient two-step GMM estimators. J Econom 126(1):25–51

    Article  Google Scholar 

Download references


We thank Jesus Crespo Cuaresma for very helpful econometric suggestions, and Thorsten Beck for generously sharing his data. Remaining errors are ours. Financial support from the Oesterreichische Nationalbank Anniversary Fund (Grant No. 15330) is gratefully acknowledged.

Author information



Corresponding author

Correspondence to Simon Sturn.



See Tables 4 and 5.

Table 4 Summary statistics for sample of developed economies, 1995–2007
Table 5 Summary statistics for sample of developed and developing economies, 1980–2007

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Sturn, S., Zwickl, K. A reassessment of intermediation and size effects of financial systems. Empir Econ 50, 1467–1480 (2016). https://doi.org/10.1007/s00181-015-0979-y

Download citation


  • Financial intermediation
  • Financial sector size
  • Finance–growth nexus
  • Too much finance
  • Robustness

JEL Classification

  • G10
  • G21
  • O16
  • O40