Revisiting the finance–growth nexus: the Turkish case, 1980–2010

ABSTRACT

In Turkey, the empirical results on the link between financial development and economic growth are mixed. The existing studies do not take into account the fact that Turkey has experienced endemic political and economic instabilities over extended periods. This study aims to analyse the role of macroeconomic instability and public borrowing on the finance–growth nexus in Turkey by using time series econometric techniques over the 1980–2010 period. In doing so, we attempt to extend the existing literature by taking into account the role of macroeconomic instability as well as public borrowing. Our results reveal that there are additional – albeit indirect – channels between finance and growth via the effects of macro instability and public borrowing on financial development and economic growth. After taking into account the effects of overall instability and public borrowing, we found that growth–financial development relationship is bidirectional and permanent. In other words, in Turkish case, economic growth and financial development are jointly determined. Thus, our results shed some light on the ambiguity of the evidence on the link between financial development and economic growth for Turkey.

KEYWORDS:

• financial development
• public debt
• macroeconomic instability
• growth
• Turkey

JEL CLASSIFICATION:

• E10
• E44
• E20
• O40

Acknowledgement

We are grateful for constructive and insightful comments from the anonymous referee that help improve the paper.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ However, the risk averse behaviour of the banking sector should not be mixed with the positive role of public debt in developing financial sectors by providing a relatively safe asset – called ‘safe asset’ view (Hauner 2009).

² See the data Appendix for the definition and the measurement of macroeconomic instability.

³ The share of public (state-owned) banks in the banking sector, according to assets, was around 45% during the 1980s and early 1990s and this ratio fell to about 35% during late 1990s and stabilized around 30% after 2002. (The Banks Association of Turkey 2016)

⁴ Cross-country studies also cover the studies that employ time series methodology on multiple countries.

⁵ In this study we will use an index, comprising of four instability variables such as inflation rate and budget deficit, to measure overall macro instability (see the data Appendix for more detail).

⁶ Bastı and Köksal (2011), by using banking data, found that public debt has unfavourable effects on the financial development in Turkey. It should be noted that, in their study, financial development is proxied by efficiency and profitability of banks.

⁷ Some other variables, such as M2 to GDP ratio and deposit to GDP ratio, have been used in the literature to represent financial development. Here, we use credit to private sector to GDP ratio as an indicator of financial development since it shows the extent of financial intermediation. According to the McKinnon/Shaw model and endogenous growth literature, financial intermediation, e.g. measured by credit to private sector, is the key determinant of economic growth.

⁸ We did not resort to Granger causality tests since results drawn from such tests are, generally, not reliable (could be spurious) when variables are non-stationary. See, for example, Rodriguez-Caballero and Ventosa-Santaularia (2014) for recent evidence.

⁹ For instance, six banks went bankrupt in 1983–1984 period (see Önder and Özyıldırım 2008).

¹⁰ The arbitrage opportunity generated by high real interest rates led the banks to lend to governments via creating open foreign exchange positions (see Alper and Öniş 2003 and the references cited therein).

¹¹ The share of private banks in the banking sector, according to assets, has risen from about 55% during the late 1980s to 65% during the late 1990s (The Banks Association of Turkey 2016).

¹² The attractiveness of the government debt instruments were not only limited to the factors like high interest rate, low risk and arbitrage opportunity but also related to other factors such as high liquidity, no-provision requirements and low tax rates.

¹³ However, this increased intermediary function of the banks did not lead to high growth since credits are mainly used to finance consumption rather than the investments. During this period the growth rate of consumer credits were significantly higher than the growth rate of corporate credits (CBRT 2010).

¹⁴ As noted before the banking sector is the primary buyer of domestic debt instruments in Turkey.

¹⁵ See the data Appendix for the definitions and the sources of the variables.

¹⁶ For PRIVR, the null hypothesis of a unit root is rejected at only 6.5% significance level. However, KPSS and ERS tests support the stationarity of ΔPRIVR.

¹⁷ Considering the non-presence of a deterministic trend in all variables, there is no trend term in cointegration relation but there exist a restricted constant term (these are in line with the suggestions of Hendry and Juselius 2001). Schwarz Bayesian Criterion selected the lag length of the VAR as 1. 2001 crisis is an outlying observation, following Hendry and Juselius (2001), we include an impulse dummy in our cointegration system. The residuals are homoscedastic, normal and not serially correlated.

¹⁸ See Altuğ and Öz (2012) and CBRT (2010).