Page 43 - Azerbaijan State University of Economics

Fatih Chellai: Regime-Dependent Effects of Public Spending in Algeria: A Structural VAR and
Markov-Switching Approach

Identification of structural shocks
The choice of this matrix of contemporary restrictions for the SVAR model is based on a
theoretical logic consistent with macroeconomic transmission mechanisms and the
institutional specificities of the Algerian economy. In this matrix, real GDP (LPIB) is placed
in first position and assumed not to react to any shocks from other variables at the moment,
which is economically justified by the fact that aggregate output takes time to adjust to
economic shocks, especially in a country like Algeria where delays in implementing public
policies and structural rigidities are frequent. Real public expenditure (EXP), on the other
hand, can react immediately to the level of GDP, reflecting potentially counter-cyclical
behavior on the part of the government, which adjusts its spending in line with current
economic performance (e.g., higher spending in the event of slower growth). However,
EXP is assumed not to react instantly to inflation or unemployment, as budgetary decisions
often require a slower adjustment process, particularly in rigid institutional contexts.

Table 4: Contemporary restraint matrix
LPIB EXP INFLATION UNEMP
LPIB 1 0 0 0
EXP C(1) 1 0 0
INFLATION C(2) C(4) 1 0
UNEMP C(3) C(5) C(6) 1
Source: By author

Inflation (INFLATION) is modeled as responding immediately to shocks to GDP and
government spending, reflecting the fact that any stimulation of aggregate demand (via
production or government spending) can generate immediate pressure on prices, especially
in an import-dependent economy like Algeria. On the other hand, it does not respond
instantly to unemployment, which follows the logic of a Phillips curve that is more
structural than cyclical. Finally, the unemployment rate (UNEMP) is assumed to react
contemporaneously to all other variables, reflecting the labor market's direct sensitivity to
changes in output, public spending (especially in an economy where the state is a major
employer), and price levels. This configuration reflects the reality of the Algerian labor
market, which is often influenced in the short term by public policies and the general
economic situation , with little dynamic autonomy.

Estimation results for the SVAR model with A-B type identification (where Ae=Bu,
and E(uu′)=I) show that several contemporaneous relationships between endogenous
variables are not statistically significant. The coefficients in matrix A, notably C(1),
C(2), C(4), C(5) and C(6), have p-values greater than 0.05, indicating that shocks in
one variable do not produce a significant instantaneous effect on the others in the
model. Only the coefficient C(3), which relates unemployment to GDP, is significant
(p = 0.049), suggesting that a shock to real GDP has an immediate and positive effect

43