Inflation, financial intermediation and growth: the case of Egypt

Abstract

There is now a consensus in the theoretical and the empirical literature that a nonlinear relationship exists between the rate of inflation and the rate of economic growth. Using a threshold regression technique, this paper re-examines this relationship and the critical role that financial intermediation plays in it. Data to examine our hypothesis are from Egypt. We find that inflation contributes positively to economic growth until it reaches a threshold rate of about 12%, after which it becomes detrimental to growth. We then show that such a nonlinear relation is connected to whether or not financial deepening has crossed a certain threshold level. Given these thresholds, a coordination of policies (especially monetary) and financial reform is needed to achieve success in economic growth.

Keywords:

• inflation
• economic growth
• financial intermediation
• Egypt
• threshold regression

JEL Classifications:

• E31
• E44
• O16

Acknowledgements

We thank two anonymous referees and the editor for extremely helpful comments. The views expressed in this paper are those of the authors and do not necessarily represent those of the IMF or IMF policy.

Notes

1. A possible future area of research would be to use Hansen and Seo's (2002) technique to estimate a threshold regression in the context of a co-integration framework for Egypt. This belongs to a separate line of inquiry that we have not pursued in this paper.

2. We are aware of one study in a panel context that considers a different question, namely the role of inflation in the relation between finance and growth (Eggoh, 2012) but we are not aware of any threshold time series methods in this context regardless of which variable is the catalyst and which is causal.

3. The terms financial depth and financial development will be used interchangeably in this study, following Levine (1997, 2004).

4. This earlier version of Hansen's work does not develop distribution for the threshold parameter. As a result, it is more appropriate for point estimates than for hypothesis testing.

5. If we move from level of financial development to growth of finance with a focus on developed countries we find that researchers are more agnostic: a stimulating recent analysis by Cecchetti and Kharroubi (2012) finds that too much growth of finance retards economic growth by drawing resources and talent away from other sectors, a sort of Dutch disease. However, it is found that moderate growth of finance is good for economic growth.

6. For a detailed analysis of the dynamics of inflation and growth in Egypt see Abou-Ali and Kheir-El-Din (2009, pp. 63–66).

7. The empirical literature largely agrees on using domestic credit to the private sector as a share of GDP (CRPRV) as the indicator of financial development/financial depth (see Khan and Senhadji, 2000, p. 8; Levine, 2004, p. 56).

8. We tried the same regressions with real instead of nominal exchange rates and found a threshold rate of inflation very close to the thresholds found in this paper (11% using real exchange rate instead of 12% using nominal exchange rate, as we shall see below). Those results are available from the authors. However, here we chose to report the results with nominal exchange rates in order to allow for a closer comparison with other papers on Egypt, namely Abou Ali and Kheir El Din (2009) who also used nominal exchange rates.

9. If a variable has a unit root (is non-stationary), then the OLS estimates will be biased downward.

10. Abou-Ali and Kheir-El-Din (2009, p. 69) consider similar specifications to the models presented here, but they do not include the (INF)^∧2 variable. Therefore, results are not directly comparable.

11. We chose to report the adjusted R² rather than R² because the former penalizes the addition of more explanatory variables.

12. We report BIC since it penalizes larger models more heavily than AIC.

13. Threshold estimates of (inflation) have also been conducted using models 1 and 2. Results (not reported) show quite similar results. Coefficient signs, magnitudes and significance are mostly the same.

14. Threshold estimates of (CRPRV) have also been conducted using models 1 and 2. Results (not reported) show quite similar results. Coefficient signs, magnitudes and significance are mostly the same.

15. Low and high financial depth in this context refer to levels of financial depth that are below and above the estimated threshold level, respectively.