We remind you that the whole of this document is accessible on the Amazon site:
Presentation of study data
As part of this project of memory project, we are particularly interested in economic growth in Africa. Variables dictated by literary and empirical journals will be retrieved from the World Bank database. In this case, GDP per capita, fossil energy consumption, physical capital ... etc. Other unobservable variables will be able to be calculated. In this sense, it is the savings rate, the net capital inflow, the rates of commercial opening. Our values are in current dollars. The study period or the interval by which we analyze the theme is between 1960 and 2016. In truth, more data remains unavailable for the year 2017. We will then solicit methods of clearing and correction of the missing values before to be able to carry out studies whose prospects seem already purified in the general introduction.
2. Analysis of econometric results
The present table shows the extent of the Granger causality test, which highlights the simultaneous result of 9 causality tests. Recall that the prayer vocation of this study concerns the classification of variables from the most exogenous to the most endogenous.
The table opposite shows us the optimal delay of the annual VAR model. Apart from the problem of selecting the significant variables, the approach for selecting the number of the optimal delay in the process is the same. It is also easy to see that the number of parameters to estimate increases with the number of delays. Focusing on an approach taking into account the parametric dimension. These include the FEC (Final prediction Error, 1969) criterion based on predictive power, AIC (Akka Information Criterion, 1973), Hannan and Quinn (1979), Schwarz (1978). By principle of parsimony, we will consider the criterion that minimizes the number of optimal delay to integrate into the process. Obviously, the criteria of AIC, FPE and LR indicate an optimal delay of order 7 on the delay retained for the annual VAR.
The table above presents the estimates of the Gaussian (left) and Bayesian (VAR) annual VAR (6) model. The relationships highlighted are between economic growth (represented by GDP per capita), co2 emissions, and income inequality. Indeed, we detect a positive relationship between per capita GDP and per capita CO2 emissions lagged by a similar period for income inequality. Regarding CO2 emissions, it is positively related to GDP. Notwithstanding, income inequalities are both negatively linked to IP but also to CO2 emissions. We will further deepen the notion of relationship by Granger causality between economic growth CO2 emissions and income inequalities. The table below tells us an advantage. Each variable explained by its historical values.
The Portemanteau test is based on the null hypothesis of no autocorrelation between residues. The Q-Stat statistic of Box Pierce modified and that of Chi-Two significantly stipulate the validity of the null hypothesis at the risk threshold of 1%. There is no residual autocorrelation in the simultaneous equation model.
The graph below shows the dynamic behavior of CO2 emissions followed by a shock on GDP, which is nothing other than the standard deviation of its errors in the UMA Common Area. CO2 emissions respond instantaneously and negatively as a result of the shock on economic growth per capita until the fifth year before returning to equilibrium starting at 13th, since the variables are stationary. In this sense, it is clear that a shock on GDP per capita is manifested by a negative effect on CO2 emissions in the AMU. In other words, it is to say that a shock on the GDP negatively impacts the environment up to 5 years before signaling an adaptation on the part of the environment.
The decomposition of the variance indicates that the variance of the GDP forecast error is due to 93.44% to its own innovations, 3.51% to those of CO2 emissions and 3.04% to those of income inequalities. This is to say that a shock to GDP per capita is more important to income inequalities than to CO2 emissions in the WAMU common area.
This graph of the inverse roots of the characteristic AR polynomial; see Lütkepohl (1991) indicates that the estimated VAR is stable (stationary) if all roots have a modulus less than one and are within the unit circle. If the VAR is not stable, some results (such as standard impulse response errors) are not valid.
Analysis of Quarterly Variables (Continued UMA Analysis)
We go in the section below to transform the annual variables of GDP, CO2 emissions and the GINI coefficient into quarterly frequency of grace with the DENTON algorithm. It calculates the proportional interpolation method of a low-frequency time series (eg, quarterly) using a higher-frequency index associated with it and also imposes the constraints that the interpolated series obeys to the original series totals. at low frequency. Benchmarking (2001) defines it as "relatively simple, robust, and well suited for large-scale applications." It requires, however, that the low frequency variable be an annual or quarterly time, while the indicator variable may be quarterly or monthly frequency. Although the procedure is generally applied to flow series (such as GDP).
Abdi-Basid ADAN, 2018.
ADAN, A.-B. (2018). Croissance Economique et Développement Durable en Afrique: Analyses Statistiques et Econométriques [Kindle]. Amazon Media EU S.à r.l.
© 2026 The Abdi-Basid Courses Institute (tABCi). All rights reserved.
🌍 The EcoClimate Hub | Focus: Climate Science & Environment.
Contact : Abdi-Basid ADAN [abdi-basid@outlook.com]
© 2017-2026 The Abdi-Basid Courses Institute (tABCi). All rights reserved.
