We introduce three types of variables in our model. First of all, the growth variable, a variable dependent on the model. We then orient financial development indicators. Finally, we introduce a conditional information matrix to control for variables that affect long-term economic growth.
3-1- Sample and period
a. Sample
Our region is a sample which is made up of 16 countries namely Bahrain, United Arab Emirates, Jordan, Kuwait, Qatar, Saudi Arabia, Indonesia, Malaysia, Tunisia, Turkey, Morocco, Egypt, Sudan, Iran, Algeria, Yemen. Our sample is made up of 16 countries shared over the MENA region and East Asia and the Pacific, we have compiled a database of international macroeconomic data available in “World Bank CD’’.
b. Period :
The sample of countries selected is made up of 16 MENA and East Asia and Pacific countries: 5 African countries, 8 Gulf countries, 2 East Asian and Pacific countries and 1 Mediterranean. Depending on the availability of data, our study period extends from 1990 to 2018 over a period of 29 years. The great diversity in terms of geography and in terms of country performance increases the robustness of our analyses.
3-2- Definitions and measurements of variables
a. Growth Indicator:
We chose the noted GDP Per Capita Growth Rate (GDP) (Levine et al., 2000, Beck et al., 2000, and Beck and Levine, 2004).
b. Indicators of financial development:
We propose the following indicators.
• Islamic financial development: In their 1998 study, Levine and Zervos add the measuring the development of the banking sector to cross-sectional studies of growth. According to these authors, this measure is equal to private sector bank credit divided by GDP denoted FI (Finis/GDP): Qard Hasan, Mourabahah, Ijarah, Moudarabah, Moucharakah, Salam, Istisna‘. ).
• Investment: Gross fixed capital formation, is the aggregate which measures, in national accounts, the investment (acquisition of production goods) in fixed capital of the various resident economic agents. (FDI/GDP).
• Control variables: We retained as control variables, for this work, the ratio of government expenditure to GDP (GC) as an indicator of macroeconomic stability (Easterly and Robelo (1993) and Fisher (1993)), the value ratio of trade (export + import) / GDP to capture the degree of openness (Sachs and Warver (1995)) noted (TRAD) andThe tertiary enrollment rate to control the accumulation of human capital noted (HK).
• Dummy variables: We use this nature of the variables (variable dummy: DV) because our study region is formed by countries that apply Islamic finance and others that do not. So, we note 1 for the countries that practice Islamic finance and 0 for the others.
c- Quality of governance indicator
• Governance quality index: “IQG indicator of quality of governance:The mean value of ICRG variables”. After calculating the quality of governance index, we will present descriptive statistics of this synthetic indicator..
-Political stability noted (PS): IMGs are not used by the World Bank Group to allocate resources. The impact of institutional factors namely political stability noted (PS) and realized by Kaufman D. Kraay A. and Mastruzzi M. (2003).
3-3- Financial development, FDI and economic growth in the MENA region and Asia Pacific
a- Simultaneous Equations Model
We will estimate the simultaneous equation model that we will specify later. The model to be estimated responds, in a mathematical way, to the following three equations:
b- Simultaneous equation model in panel data
-
There are several steps to follow, namely:
-
The design, i.e. the writing or specification of the model.
-
Estimation of the model equations, using appropriate techniques.
-
Endogeneity problem
-
Method REG3 (Three-Stage least squares)
c- Preliminary tests
The main results obtained, their interpretations and their debates compared to previous studies.
-
Stationarity tests
-
Collinearity study between the independent variables
-
Multi-collinearity problem and model selection
-
Model equations identification problem modèle
-
Exclusion Restrictions
This restriction consists of assigning a zero coefficient for each endogenous or exogenous variable that does not appear in a structural equation. In our model, the variable "FDI" appears at the level of the last equation is endogenous whose exogenous variables "PS", "TRAD" and "HK" appear only at the level of the last equation and do not appear at the level of the first or second equation. There are variables that appear at the level of the first and third equations and do not appear at the level of the second (IQG).
Some model specifications require that variables be assigned an identical coefficient. This type of restriction is not present in our model. Once the restrictions on the coefficients have been made, it is essential to proceed with the identification of the system of equations. There are two identification conditions: order conditions (necessary conditions) and rank conditions (sufficient conditions).
After having selected the variables to be integrated into the model, a step prior to the step of processing simultaneous equation models is to perform model identification tests to choose the most appropriate estimation method..
In our case, we note for the model to be studied, that all the equations are over-identified. Indeed, we have three endogenous variables in the model (i.e. W = 3) "GDP", "IFD" and "FDI" and five exogenous variables: "TRAD", "INV", "GC", "PS", " VD”, “IQG” and “HK” (i.e. K = 7)
-
The first equation has five exclusion restrictions and no constraint restrictions. By applying the identification conditions, the variables appearing in the human capital equation give: W' = 1, K' = 4 and r = 0 with W' being the number of endogenous variables appearing in an equation and K ' is the number of exogenous variables appearing in an equation. So let: W – W'+ K – K' = 3–1 + 6– 4 = 4 > W – 1 = 3–1 = 2, the first equation is therefore over-identified.
-
The second equation has five exclusion restrictions but no constraint restrictions. We therefore have: W = 3, K = 6, W' = 1, K' = 4 and r = 0, which gives us: W – W' + K – K' = 3–1 + 6–4 = 4 > W – 1 = 2, so this equation is over-identified.
-
The third equation has six exclusion restrictions but no constraint restrictions. So we have W = 3, K = 6, W'=1, K'=4 and r = 0, which implies W-W'+ K- K'= 3 − 1 + 6– 5 = 3 > W – 1 = 2, the third equation is therefore over-identified. Since in our model all the equations are over-identified, the model will therefore be over-identified.
-
Sufficient conditions: Rank conditions
If the order conditions are verified, it is also necessary to verify the rank conditions (sufficient conditions)