Kydland and Prescott (1977) show that discretionary monetary policy leads to time inconsistency and inflationary bias. In the long term, it accentuates economic imbalances. These authors raise a number of points to explain the ineffectiveness of monetary policy:

In their work, Barro and Gordon (1983) propose that central banks commit to an inflation target in order to reduce the inefficiency of monetary policy. Their main idea is based on the notion of discretionary and reputational regimes in money management. In a discretionary regime, the central bank may be tempted to print more money and create inflation surprise. Monetary rules that fix the growth of money or target the rate of inflation would limit the discretionary capacity of monetary authorities. In this dynamic, Barro and Gordon have proposed a reputational equilibrium that leads monetary authorities to be cautious in their monetary policy so as not to weaken their scope.

Rogoff (1985) points out that in discretionary monetary regimes, the authorities are often tempted to use monetary policy for political ends, such as reducing unemployment; this encourages inflation. Discretionary policy leads to high and volatile inflation in the long term. To remove this ambiguity, he advocates the idea of a Central Bank totally independent of the government, which could encourage an inflationary monetary policy to reduce unemployment. This independence would keep inflation low, strengthen the Central Bank’s reputation and align the behavior of economic agents with economic policy objectives.

Aguir (2013) , on a study based on 54 developing countries classified by income level for the period 1991-2011, shows that inflation is a purely monetary phenomenon. He argues that central bank independence keeps inflation at an acceptable level, and that this independence does not affect output growth or unemployment. He considers money to be neutral in relation to real economic activity, and to reduce inflationary bias, it is necessary to strengthen the independence of the Central Bank and ensure better coordination between monetary policy and exchange rate policy.

In Bozkurt’s (2018) work carried out in Türkiye and the Shanghai Cooperation Organization zone on panel data, the results present a dichotomy between the real and monetary sectors, since monetary shocks do not impact real output, real interest rates and real wages. Instead, they affect nominal variables, prices and nominal wages.

In his work, Kasiwa (2022) distinguishes currency neutrality from super-neutrality. In the first case, currency neutrality refers to the fact that changes in the currency have no impact on output, employment or the real interest rate. In the second, super-neutrality, the growth rate of the money supply has no effect on real variables. The results of his work, estimated on an ARIMA model, show that money has no impact on short- and long-term production in the Congo.

De Grauwe (1999) considers that monetary policy suffers from problems of time inconsistency and credibility on the part of the monetary authorities. He believes that only a monetary policy that takes economic agents by surprise can have an impact on output growth and unemployment.

Using an impulse-response matching approach to estimate the parameters of their model, Christiano et al. (2001) find that monetary policy is not neutral in the economy, affecting both nominal and real variables in the short term. However, these results are highly sensitive to economic conditions and the expectations of economic agents.

With the introduction of unconventional monetary policy instruments, Nakamura et al. (2013) show in their work the absence of currency neutrality with the US Federal Reserve’s use of “forward guidance”. They stipulate that money varies real and nominal interest rates and thus impacts output growth in the short term.

Oloufade (2015) , using error-correction and panel VAR models, shows the positive effects of currency shocks on industrial output in the WAEMU, more specifically in Senegal in the short term and in Côte d’Ivoire in the long term.

Pastén et al. (2018) challenge the money neutrality thesis and emphasize the importance of monetary policy in economic stability and financial market regulation. For these authors, any increase in the money supply translates in the short term into higher financial asset prices during strong inflationary periods.

The work of Brandao-Marques et al. (2020) on a panel of 40 developing countries demonstrates the absence of monetary neutrality, and shows that positive monetary policy shocks have a negative influence on output growth and inflation. The extent to which monetary shocks are transmitted is strongly linked to the exchange rate in countries pegged to the dollar.

Using a staggered lag model and a Bayesian vector autoregression model, Gbenou (2021) estimates that upward variations in the money market interest rate lead to deflation and have a negative impact on effective growth in WAEMU countries.

Empirical literature remains mixed on the effects of monetary policy on short-term economic activity. Some studies have attempted to highlight the absence of monetary neutrality, and hence the strong dichotomy between the real and monetary sectors. Other works question this dichotomy and show the influence of money on short-term economic activity.

In order to highlight the dynamic impact of monetary shocks on short-term fluctuations in economic activity, we will use structural VAR (SVAR) methodology with short-term constraints. The use of VAR models was introduced into economic analysis by Sims in 1980. The idea is based on the observation that macro-econometric models impose a priori assumptions without any statistical justification, and the exogeneity of certain postulated economic policy variables is not tested. According to Clément and Germain (1993) , in a correctly specified general equilibrium model, each variable depends a priori on all the other variables in the system, notably through the formation of expectations. This general idea leads to the use of VAR models.

The monetary policy framework of the Banque Centrale des Etats de l’Afrique de l’Ouest (BCEAO) is the inflation targeting policy, which stipulates that the average annual inflation rate must be maintained at a maximum of 3% per year. This policy requires the central bank to use its main monetary policy instrument, the interest rate, to achieve its inflation target. When the inflation rate exceeds its target level, the central bank activates a restrictive monetary policy, which translates into a rise in the interest rate, and vice versa. However, for this targeting policy to be possible, the currency must be neutral. Since currency neutrality means the absence of effects of monetary shocks on real economic variables such as output, employment and real interest rates, the model will be built using four economic variables: the money supply growth rate (tmm), the real interest rate (tint), the inflation rate (tinf) and the output growth rate (tcp); the study period running from the first quarter of 1990 to the last quarter of 2022.

In order for the model to be able to provide information on the lags in the absorption of monetary shocks, then assess the lags and prevent shocks from being permanent, the data must be stationary. In addition, the stationarity procedure remedies three pitfalls:

The completion of the SVAR model presented for each WAEMU country should, however, lead us towards a methodological extension capable of identifying the difficulties linked to cyclical regulation by monetary policy in the States of the Union. Taking these dynamics into account is of paramount importance in defining the development strategies of the countries of the Union, as well as in the orientation of their economic policies.

After studying the stationarity of the series, it is possible to write the dynamic theoretical structural form of the model in the following form:

AY_t = B₁Y_t−1 + B₂Y_t−1 + B₃Y_t−1 + … + B_pY_t−p + U_t (1)

where

A is the (4 × 4) matrix of simultaneity relations between the variables of the economy; Y_t is the vector of endogenous² variables composed of the money supply growth rate (tmm), the real interest rate (tint), the inflation rate (tinf) and the output growth rate (tcp); B_i is the (4 × 4) matrix associated with the lag; U_t is the vector of structural residuals of type iid N(0,π) with π, a diagonal (nxn) matrix.

This assumption implies that the residuals of the model are independent of each other and can be considered as orthogonal structural shocks in pairs.

The reduced form of the above model is given by:

Y_t = A⁻¹B₁Y_t−1 + A⁻¹B₂Y_t−2 + A⁻¹B₃Y_t−3 + … + A⁻¹B_pY_t−p + A⁻¹U_t

Y_t = C₁Y_t−1 + C₂Y_t−2 + C₃Y_t−3 + … + C_pY_t−p + A⁻¹U_t (2)

This last writing of the structural form constitutes the structural VAR model in its general form. For its estimation, it is necessary to determine the optimal number of shifts, evaluate the parameters of the standard VAR, define the identifying constraints and then proceed with the Choleski decomposition of the variance-covariance matrix of the innovations of the estimated standard VAR.

a) Estimating delays using information criteria³

The number of variables in the economy is n = 4. To be able to estimate the SVAR, we therefore introduce n(n − 1)/2 constraints, i.e., 4(4 − 1)/2 = 6 short-term constraints which are linked in particular to slow adjustment phenomena on certain economic variables. These adjustments concern the diffusion effects of certain measures relating to the growth rate of the money supply and the existence of lags in the reaction of output, inflation and interest rates following a variation in the growth rate of money. With regard to the arrangement of the variables, the classification is made taking into account the objectives of the study and the instruments used in the monetary policy framework. The money supply and the interest rate are the instruments of monetary policy; inflation growth and output growth are the objective variables of the model. Taking into account the lags in the reaction of certain variables on others, we can ask:

We incorporate the constraints defined above in a triangular form, which justifies Choleski’s methodology in order to be able to trace them back to the structural form. The matrix is given by:

$\left[\begin{array}{l}tmm\\ t\mathrm{int}\\ t\inf \\ tcp\end{array}\right]=\left[\begin{array}{cccc}1& 0& 0& 0\\ C21& 1& 0& 0\\ C31& C32& 1& 0\\ C41& C42& C43& 1\end{array}\right]\times \left[\begin{array}{l}{\epsilon }_{tmm}\\ {\epsilon }_{t\mathrm{int}}\\ {\epsilon }_{t\inf }\\ {\epsilon }_{tcp}\end{array}\right]$

After ensuring that the structural shocks are normal⁵ and uncorrelated⁶, it is possible to calculate the matrix of simultaneity relationships associated with the various shocks. Estimation of the instantaneous coefficients (see Appendix 1) gives the following results:

From these results, we can write the matrix of simultaneity relations as follows:

BENIN

$\left[\begin{array}{l}tmm\\ t\mathrm{int}\\ t\inf \\ tcp\end{array}\right]=\left[\begin{array}{cccc}1& 0& 0& 0\\ -0.010& 1& 0& 0\\ +0.298& -0.376& 1& 0\\ -0.055& +0.133& -0.099& 1\end{array}\right]\times \left[\begin{array}{l}{\epsilon }_{tmm}\\ {\epsilon }_{t\mathrm{int}}\\ {\epsilon }_{t\inf }\\ {\epsilon }_{tcp}\end{array}\right]$

BURKINA FASO

$\left[\begin{array}{l}tmm\\ t\mathrm{int}\\ t\inf \\ tcp\end{array}\right]=\left[\begin{array}{cccc}1& 0& 0& 0\\ -0.012& 1& 0& 0\\ +0.097& -0.613& 1& 0\\ +0.055& -0.083& -0.117& 1\end{array}\right]\times \left[\begin{array}{l}{\epsilon }_{tmm}\\ {\epsilon }_{t\mathrm{int}}\\ {\epsilon }_{t\inf }\\ {\epsilon }_{tcp}\end{array}\right]$

COTE D’IVOIRE

$\left[\begin{array}{l}tmm\\ t\mathrm{int}\\ t\inf \\ tcp\end{array}\right]=\left[\begin{array}{cccc}1& 0& 0& 0\\ -0.023& 1& 0& 0\\ +0.096& -0.196& 1& 0\\ +0.027& +0.793& -0.182& 1\end{array}\right]\times \left[\begin{array}{l}{\epsilon }_{tmm}\\ {\epsilon }_{t\mathrm{int}}\\ {\epsilon }_{t\inf }\\ {\epsilon }_{tcp}\end{array}\right]$

MALI

$\left[\begin{array}{l}tmm\\ t\mathrm{int}\\ t\inf \\ tcp\end{array}\right]=\left[\begin{array}{cccc}1& 0& 0& 0\\ +0.011& 1& 0& 0\\ +0.101& -0.841& 1& 0\\ +0.091& -0.457& +0.148& 1\end{array}\right]\times \left[\begin{array}{l}{\epsilon }_{tmm}\\ {\epsilon }_{t\mathrm{int}}\\ {\epsilon }_{t\inf }\\ {\epsilon }_{tcp}\end{array}\right]$

NIGER

$\left[\begin{array}{l}tmm\\ t\mathrm{int}\\ t\inf \\ tcp\end{array}\right]=\left[\begin{array}{cccc}1& 0& 0& 0\\ -0.011& 1& 0& 0\\ -0.047& -0.587& 1& 0\\ +0.032& -0.746& +0.103& 1\end{array}\right]\times \left[\begin{array}{l}{\epsilon }_{tmm}\\ {\epsilon }_{t\mathrm{int}}\\ {\epsilon }_{t\inf }\\ {\epsilon }_{tcp}\end{array}\right]$

SENEGAL

$\left[\begin{array}{l}tmm\\ t\mathrm{int}\\ t\inf \\ tcp\end{array}\right]=\left[\begin{array}{cccc}1& 0& 0& 0\\ -0.019& 1& 0& 0\\ +0.475& -0.506& 1& 0\\ -0.052& -0.104& -0.201& 1\end{array}\right]\times \left[\begin{array}{l}{\epsilon }_{tmm}\\ {\epsilon }_{t\mathrm{int}}\\ {\epsilon }_{t\inf }\\ {\epsilon }_{tcp}\end{array}\right]$

The estimates show that monetary shocks have no instantaneous effect on real interest rates in WAEMU countries. We note that the instantaneous multipliers associated with interest rates are low and insignificant in the countries of the Union. This result is quite understandable, as when the Central Bank injects liquidity as part of its monetary policy, it takes time for real interest rates to be affected. When economic agents anticipate the effects of monetary shocks, their often rational behavior attenuates their impact.

The results also show that the rate of production growth reacts instantaneously and significantly to monetary shocks. However, the impact remains very weak, below 0.10 in all Union countries except Côte d’Ivoire, where the instantaneous effect is not significant. The other factor limiting the instantaneous impact of monetary shocks on production is price and wage rigidity. With fixed labor contracts and prices, it is difficult for production to adjust to changes in monetary policy.

The aim here is to analyze the reactions of output growth and inflation rates to monetary shocks. We thus present the cumulative multipliers (or total effects) following a shock to money supply growth (see Appendix 2 and Appendix 3). The results are summarised in Table 1 and Table 2 below:

Analysis of cumulative response functions following monetary shocks leads to the following comments:

However, the response of economic activity to monetary shocks highlights the heterogeneous nature of the countries in the Union. Indeed, the WAEMU zone is made up of low-income and middle-income countries whose economic convergence is not yet perfect. Given the contrasting trends in national variables, it would be surprising if the conduct of centralised monetary policy were to suit all the member countries, which have different economic structures (differences in living standards, labour productivity and, above all, excessive budget imbalances).

In this subsection, we calculate the pass-through coefficients of monetary shocks on the interest rate, inflation rate and output growth rate. Pass-through coefficients are obtained from the ratio between the cumulative impulse response of the interest rate, inflation rate and output growth rate, after h years, and the cumulative response of the money supply growth rate following the initial shock to the money supply growth rate, after h years. In other words, for each of the prices considered, at forecast horizon h, the pass-through coefficient (PT) is calculated as follows:

PT = |O_t,t+h|/M_t,t+h

where O_t,t+h is the cumulative change in the target variables and M_t,t+h is the cumulative change in the growth rate of the money supply. The results of the calculations of the estimated Pass-Through coefficients are given in the following Table 3 :

Source: Authors’ calculations using WINRATS software.

It follows from the estimates of the pass-through coefficients that the pass-through of monetary shocks to real interest rates, the inflation rate and the output growth rate is partial, insofar as all the estimated values are less than unity. However, it should be noted that the transmission of monetary shocks has a much greater impact on the inflation rate than on the output growth rate. These results are also confirmed by the analysis of the variance decomposition of the error on the inflation rate and output growth rate (see Appendix 4). Indeed, monetary shocks are transmitted:

To get a better idea of these effects, we can examine the reaction of key rates to monetary shocks. In most of the Union countries, interest rates vary negatively following monetary innovations, except in Mali, where positive effects are observed. Such negative interest rate behavior following monetary shocks in the Union countries favors increased demand for goods and services. If the supply of goods and services fails to keep pace with this demand, inflation ensues, weakening household purchasing power and thus reducing aggregate demand and economic growth. The real divergences following interest rate shocks can be explained by the fact that their effects occur at different stages of an economic cycle, often making policy pro-cyclical. With a single currency and a common monetary policy, there is only one interest rate, and consequently it cannot be used differently in different countries. For a monetary policy not to have asymmetrical effects, it is important that the countries have similar economic structures, which is not the case for the EU countries.