**3. Methodology**

Even though government efficiency has been studied to some extent, we reveal a significantly wider knowledge gap. First, conceptual specification, based on which empirical examinations of government efficiency is analyzed, is not prevailing combining theory and empirical analysis. Secondly, we identify six structural VAR models. To our knowledge, it has not been applied to WB 6 data. VARs turn out to be one of the key empirical tools in modern macroeconomics, and they allow one to model macroeconomic data informatively [39].

The range of the data is from January 2006 to December 2018. In order to control for time trends in our analysis, we include dummy variables. The expression referring to an SVAR model is used as follows:

$$Y\_t = \mathfrak{a}\mathfrak{o} + \beta \mathbf{X}\_t + \mathfrak{u}\_t \tag{1}$$

Here, we present parameter estimates and the main characteristics of the models. The identified recursive SVAR model is as follows:

$$y\_t = a\_t + \beta\_1 GovE\sharp f\_t + \beta\_2 \pi\_t + u\_t \tag{2}$$

where *yt* is the gross domestic product for each of the WB 6, *GovEfft* is the government efficiency indicator, and *π<sup>t</sup>* represents inflation (a proxy for macroeconomic stability). This specification contains independently identically distributed stochastic disturbance term *ut IID* 0; *σ*<sup>2</sup> *u :* The above model will allow us to observe how economic governance shocks and macroeconomic stability impact GDP growth and vice versa*.* Of particular interest for this paper is to examine the role of economic integrations and macroeconomic stabilization in determining the growth of GDP in Albania, Bosnia and Hercegovina, Kosovo, Montenegro, North Macedonia, and Serbia. Thus, government efficiency and inflation are considered as important explanatory factors. For North Macedonia Model, we added purposely the corruption indicator variable, in order to observe the potential shocks to growth. As we will see, the indicator to this specific case shows no impact.

How well the models describe the dynamic behavior of economic variables? We will proceed with our VAR models for structural inference and policy analysis. One of the main objectives of our VAR model is forecasting, and it has common

*Governance and Growth in the Western Balkans: A SVAR Approach DOI: http://dx.doi.org/10.5772/intechopen.91731*

characteristics as a univariate AR model. Zivot and Wang emphasize that forecasting future values of a matrix *Yt*, when the parameters Π of the Var(p) process are assumed to be known and there are no deterministic terms of exogenous variables, the best linear predictor, in terms of minimum mean squared error (MSE) of *Yt*þ<sup>1</sup> or one-step forecast, is [40]:

$$Y\_{(T+1|T)} = c + \Pi\_1 Y\_T + \dots + \Pi\_p Y\_{T-p+1} \tag{3}$$

and forecasts for longer horizons h (h-step forecasts) may be obtained using the chain rule of forecasting as:

$$Y\_{(T+h|T)} = c + \Pi\_1 Y\_{(T+h-1|T)} + \dots + \Pi\_p Y\_{(T+h-p|T)} \tag{4}$$

and h-step forecast error may be expressed as:

$$\Psi\_{T+h} - Y\_{(T+h|T)} = \sum\_{\varepsilon=0}^{h-1} \Psi\_{\varepsilon \mathcal{E}\_{T+h-\varepsilon}} \tag{5}$$

where the matrices Ψ*<sup>s</sup>* are determined by recursive substitution:

$$\Psi\_s = \sum\_{j=1}^{p-1} \Psi\_{s-j} \Pi\_j \tag{6}$$

with Ψ<sup>0</sup> ¼ *In* and Π *<sup>j</sup>* ¼ 0 for j>p.

As already emphasized in the literature review, the logic behind employing these variables is clear: in an economically free societal environment, people and companies are free to work, manufacture, utilize their disposable income, and make investments in any way they please, with that liberty both ensured and protected by the state and unconstrained by the state [41]. Besides, low and stabilized inflation significantly indicates faster and mounting economic growth.
