Abstract
Morocco: Selected Issues
Fiscal Multipliers In Morocco1
A. Introduction
1. Interest in estimating the size of the fiscal multiplier in Morocco has surged in recent years, as the country has experienced several exogenous shocks, including from the fiscal sector. The overall fiscal deficit in Morocco declined from a peak of 7.3 percent in 2012 to 4.9 percent of GDP in 2014, due in part to exogenous factors. This has revived interest in understanding the impact of fiscal shocks, particularly from government expenditure, on aggregate demand as the adjustment has been tilted towards government consumption. Estimating the fiscal multiplier, defined as the change in the real GDP caused by one unit increase in government consumption arising from exogenous shocks, helps address some of these questions.
2. The magnitude of the fiscal multiplier may depend on the transmission and the dynamics of the shocks to the economy. To better understand the size of the fiscal multiplier, it may be necessary to better comprehend the relationship between key macroeconomic variables in both the short and long run. Changes in government expenditure could induce long-run and/or short-run effects on the economy; long-run effects tend to impact the steady equilibrium of the economy, while short-term effects only affect the cyclical components of macroeconomic variables, which implies that these effects are short lived.
3. This note follows a three-step process toward assessing the impact of government spending shocks on output. First, it provides stylized facts on the long-run correlations between key macroeconomic variables. Second, it assesses the short-run effects of government consumption using VAR techniques (specifically, the Blanchard and Perotti (2002) structural identification methodology). Finally, it assesses the size of the fiscal multiplier, using impulse response functions. The note is organized as follows: section B briefly reviews long-run stylized facts; section C presents the methodology used to analyze the short-run effects of government consumption; section D presents the short-term stylized facts; section E describes the impulse response analysis; section F discusses the size of fiscal multipliers in Morocco; and, section G concludes.
4. The analysis suggests that the government consumption multiplier in Morocco is relatively small. However, its size falls within the range expected for emerging economies. The transmission channels of transitory shocks seem weak as the co-movement among macroeconomic variables is weak in the short run. In contrast, permanent adjustments of government consumption could have long-lasting effects on economic activity.
B. Long-Term Stylized Facts
5. The long-run stylized facts focus on correlations between key macroeconomic variations, abstracting from cyclical movements. We filtered the data using a Hodrick-Prescott filter to eliminate short-run volatility, and then calculated the correlations among the trend components of the variables.2 Table 1 (below) shows some basic, long-run correlations among key macroeconomic variables.
Morocco: Long-Run Correlations
Morocco: Long-Run Correlations
| Real GDP | Agriculture GDP | Non-Agriculture GDP | Government consumption | Reel effective Exchange rate | Trade balance | Central government debt | |
|---|---|---|---|---|---|---|---|
| Real GDP | 1.0 | ||||||
| Agriculture GDP | 0.86 (12.25) | 1.00 ----- | |||||
| Non-Agriculture GDP | 0.98 (32.65) | 0.72 (7.70) | 1.00 ----- | ||||
| Government consumption | 0.99 (58.83) | 0.79 (9.43) | 0.99 (67.76) | 1.00 ----- | |||
| Reel effective Exchange rate | −0.98 (-35.39) | −0.84 (-11.59) | −0.96 (-24.31) | −0.97 (-28.26) | 1.00 ----- | ||
| Trade balance | −0.90 (-15.11) | −0.98 (-40.51) | −0.78 (-9.26) | −0.84 (-11.55) | 0.86 (12.34) | 1.00 ----- | |
| Central government debt | 0.93 (18.87) | 0.61 (5.68) | 0.99 (46.02) | 0.97 (28.97) | −0.90 (-15.54) | −0.69 (-7.01) | 1.00 ----- |
Morocco: Long-Run Correlations
| Real GDP | Agriculture GDP | Non-Agriculture GDP | Government consumption | Reel effective Exchange rate | Trade balance | Central government debt | |
|---|---|---|---|---|---|---|---|
| Real GDP | 1.0 | ||||||
| Agriculture GDP | 0.86 (12.25) | 1.00 ----- | |||||
| Non-Agriculture GDP | 0.98 (32.65) | 0.72 (7.70) | 1.00 ----- | ||||
| Government consumption | 0.99 (58.83) | 0.79 (9.43) | 0.99 (67.76) | 1.00 ----- | |||
| Reel effective Exchange rate | −0.98 (-35.39) | −0.84 (-11.59) | −0.96 (-24.31) | −0.97 (-28.26) | 1.00 ----- | ||
| Trade balance | −0.90 (-15.11) | −0.98 (-40.51) | −0.78 (-9.26) | −0.84 (-11.55) | 0.86 (12.34) | 1.00 ----- | |
| Central government debt | 0.93 (18.87) | 0.61 (5.68) | 0.99 (46.02) | 0.97 (28.97) | −0.90 (-15.54) | −0.69 (-7.01) | 1.00 ----- |
6. The transmission channel of permanent shocks appears strong as macroeconomic variables co-move along the growth path. Correlations among real variables are high in the long run. The GDP is particularly correlated with most variables in the long run, with the coefficients of correlation generally close to 0.9, while the relationship is almost one-to-one between real GDP and non-agricultural GDP. This suggests a strong co-movement along the growth path, which could imply more persistent effects from permanent shocks, specifically supply shocks.
7. Moreover, permanent adjustments of government consumption could have long-lasting effects on economic activity. In the long run, government consumption is positively correlated with all the macroeconomic variables, with the exception of trade balance and the real exchange rate. This strongly positive relationship could be a result of the framework of government consumption (such as wages and social benefits) and the transfer system to local collectivities or public entities. However, the strongly negative correlation between the government spending, the trade balance and the real exchange rate indicates possible dampening effects to permanent government consumption shocks. The dampening effects from the trade account are driven by leakages. The long-run correlation between government spending and real exchange rates suggest that permanent government consumption shocks could play an important role in long real exchange rate movements.
C. Short-Term Stylized Facts
8. The short-run stylized facts focus on co-movements of cyclical movements of key macroeconomic variables. Empirical methodologies in the literature on business cycles require the removal of trend from macroeconomic variables through the first difference technique.3 To obtain a comprehensive picture on short-term fluctuations, we estimate a vector autoregression (VAR) model including GDP, non-agricultural GDP, government consumption, trade balance, and the real effective exchange rate. We then augment the VAR with nominal variables, specifically inflation and the money market interest rate, in order to capture the relationship between nominal and real variables. We compute the VAR-based autocorrelation functions (ACFs),4 over a horizon period of 20 quarters, to assess the persistence and co-movements among macroeconomic variables in the short run (Figure 1).5
Morocco: VAR-Based Cross-Correlations Between Morocco Key Macroeconomic Variables
Citation: IMF Staff Country Reports 2016, 036; 10.5089/9781484393215.002.A001
Morocco: VAR-Based Cross-Correlations Between Morocco Key Macroeconomic Variables
Citation: IMF Staff Country Reports 2016, 036; 10.5089/9781484393215.002.A001
Morocco: VAR-Based Cross-Correlations Between Morocco Key Macroeconomic Variables
Citation: IMF Staff Country Reports 2016, 036; 10.5089/9781484393215.002.A001
9. In contrast to the findings on permanent shocks, the transmission channels of transitory shocks seem weak, despite the large fluctuations. Fluctuations in real variables are particularly large at the end of the sample period—with substantially large amplitudes between peaks and troughs—characterizing a period during which Morocco experienced severe exogenous shocks. Similarly, there are high frequency fluctuations in the government consumption, which exhibit large swings from strong policy action to contain wage bill and to reform subsidies particularly toward the end of the sample period. Although the “cyclical components” of macroeconomic variables display large fluctuations, their correlations seems to be low, suggesting that the transmission mechanism is weak.
10. Moreover, the growth rate of government consumption displays weak correlation, including in the leads and lags. The correlation between (the growth rates of) real GDP and government consumption is about 0.3 at its peak, but becomes close to zero in the very short-run horizon. The correlation between government consumption and the trade balance is negative and appears to be important in the first quarter. The government consumption is also negatively correlated with inflation. However, the correlation between the interest rate and the real exchange rate appears positive.
D. Structural VAR: Methodology
11. The empirical methodology used to estimate the impact of government spending shock on key macroeconomic variables is the following:
A structural VAR is needed as it enables us to identify fiscal shocks. This is important because fiscal policy requires some time to respond to news about the economy. Structural identification rests on the assumption that fiscal policy shocks affect output through wealth effects, intertemporal substitution, and distortions.
Using the empirical approach based on quarterly data utilized by Ilzetzki and others (2013) and the Blanchard and Perotti (2002) structural identification methodology, we estimate the SVAR on the Log difference of the government consumption Δ(LnGCt), the Log difference of the real effective exchange rate Δ(LnREERt), the trade balance to GDP ratio (TBt/GDPt), and the Log difference of the real GDP Δ(LnGDPt).6
We then use the SVAR impulse functions to estimate the size of fiscal multipliers for Morocco and to explore the relevant factors that could explain their magnitude.
We complement the analysis of SVAR with Bayesian impulse responses of the observable macroeconomic variables following a one-off temporary change in government consumption.
E. Impulse Response Analysis
12. The impulse response analysis focuses on the response of the economy to an exogenous shock to government consumption, as most of the fiscal adjustment has fallen on consumption. It also helps to examine the dynamic properties of our model, check its stability, and identify the variables that might display complex and interesting dynamics; for example, by undershooting or overshooting their steady state values. The magnitude of the government consumption shock is set to a one (positive) unit standard deviation of the exogenous shock. The figure below shows the response of the trade balance, real effective exchange rate, and the real GDP from a SVAR estimated on the full sample.
13. The findings suggest that a government consumption shock has a weak impact on most macroeconomic variables, particularly real GDP. A positive government consumption shock increases real GDP by less than 0.005 percent. Moreover, the effect is short lived since it dies out completely after eight quarters. However, the government spending shock would worsen the trade balance and lead to real appreciation. The trade balance worsens initially by about 0.004 percent, while the real effective exchange rate appreciates significantly by the same magnitude. To some extent, the weak cross-correlation between real GDP and government consumption (growth rates) in the short run could echo the similarity that emerges from their impulse responses to exogenous shocks.
Morocco: Estimated Impulse Responses to a One (Positive) Unit Standard Deviation of Government Consumption Shock
Citation: IMF Staff Country Reports 2016, 036; 10.5089/9781484393215.002.A001
Morocco: Estimated Impulse Responses to a One (Positive) Unit Standard Deviation of Government Consumption Shock
Citation: IMF Staff Country Reports 2016, 036; 10.5089/9781484393215.002.A001
Morocco: Estimated Impulse Responses to a One (Positive) Unit Standard Deviation of Government Consumption Shock
Citation: IMF Staff Country Reports 2016, 036; 10.5089/9781484393215.002.A001
Morocco: Estimated Impulse Responses to a One (Positive) Unit Standard Deviation of Government Consumption Shock
Citation: IMF Staff Country Reports 2016, 036; 10.5089/9781484393215.002.A001
Morocco: Estimated Impulse Responses to a One (Positive) Unit Standard Deviation of Government Consumption Shock
Citation: IMF Staff Country Reports 2016, 036; 10.5089/9781484393215.002.A001
14. The propagation of the government spending shock using a Bayesian VAR unveils identical patterns. However, with Bayesian behaviour, unanticipated government consumption shocks seem to have relatively larger effects.
F. Size of Fiscal Multipliers
15. In this section, we focus on the impact multiplier—that is, the change in the GDP at the moment the impulse to the government consumption occurs—and the cumulative multiplier—the change in the GDP over the period the impulse to government consumption vanishes.
16. On average, fiscal multipliers for Morocco are relatively small: the average impact multiplier varies between 0.095 and 0.3, while the cumulative multiplier is estimated around 0.6.7 By their size, fiscal multipliers in Morocco are not significantly different from estimates for emerging economies (see table below). The magnitude of the fiscal multipliers could be affected by several factors, including the degree of exchange rate flexibility, the openness of the economy, the monetary policy stance, and the public debt level. Household behaviors are critical in determining the size of fiscal multipliers. Specifically, the high propensity to consume and import foreign goods dampens the impact of a government spending shock on the real GDP, as significant leakages exist. Overall, the analysis tends to suggest that a well-designed fiscal consolidation (for instance, one tilted toward public consumption) would reduce public debt with a limited contractionary effect on growth.
G. Conclusion
17. There are strong correlations among real variables in the long run, but Morocco’s macroeconomic variables display low persistence and co-movements in a very short-run period. The transmission channels of transitory shocks seem weak as the co-movement among macroeconomic variables is weak in the short run. In contrast, permanent adjustments of government consumption could have long-lasting effects on economic activity.
18. Government consumption shock has a weak impact on most macroeconomic variables, particularly real GDP. Government consumption shocks would worsen the trade balance, lead to real appreciation, and raise output. While the impacts on the trade balance and real effective exchange rate are statistically significant, the impact on real GDP appears limited. However, with Bayesian behaviour, unanticipated government consumption shocks seem to have relatively larger effects
19. The government consumption multipliers in Morocco are relatively small. However, their size falls within the range expected for emerging economies.
Morocco: Comparison of the Size of Multipliers with the Literature
cumulative over three years.
Morocco: Comparison of the Size of Multipliers with the Literature
| Countries | Model | Variables | S-T Multipliers | L-T multipliers 1/ | |
|---|---|---|---|---|---|
| Gulf Council Countries | Panel models | total expenditure | 0.2 - 0.3 | 0.4 - 0.7 | |
| Espinoza and Senhadji, 2011 | capital expenditure | 0.2 - 0.3 | 0.6 - 1.1 | ||
| current expenditure | 0.2 - 0.3 | 0.3 - 0.7 | |||
| 22 developing countries | Panel SVAR | Developing countries dummy | −0.2 | 0.2 | |
| Ilzetzki, Mendoza, and Vegh, 2010 | Exchange rate | 0.1 | 1.5 | ||
| openness (exports + imports) | −0.3 | −0.7 | |||
| Emerging economies | VARs | total expenditures | 0.1 - 0.2 | 0 | |
| IMF, WEO, 2008 | |||||
| G-20 Countries | capital expenditure | 0.5 - 1.8 | |||
| IMF, G-20 note, 2008 | current expenditure | 0.3 - 1.0 | |||
| Emerging Asia | Structural | capital expenditure | 0.7 | 0.8 | |
| Freedman et al, 2008 | model | transfers | 0.4 | 0.2 | |
cumulative over three years.
Morocco: Comparison of the Size of Multipliers with the Literature
| Countries | Model | Variables | S-T Multipliers | L-T multipliers 1/ | |
|---|---|---|---|---|---|
| Gulf Council Countries | Panel models | total expenditure | 0.2 - 0.3 | 0.4 - 0.7 | |
| Espinoza and Senhadji, 2011 | capital expenditure | 0.2 - 0.3 | 0.6 - 1.1 | ||
| current expenditure | 0.2 - 0.3 | 0.3 - 0.7 | |||
| 22 developing countries | Panel SVAR | Developing countries dummy | −0.2 | 0.2 | |
| Ilzetzki, Mendoza, and Vegh, 2010 | Exchange rate | 0.1 | 1.5 | ||
| openness (exports + imports) | −0.3 | −0.7 | |||
| Emerging economies | VARs | total expenditures | 0.1 - 0.2 | 0 | |
| IMF, WEO, 2008 | |||||
| G-20 Countries | capital expenditure | 0.5 - 1.8 | |||
| IMF, G-20 note, 2008 | current expenditure | 0.3 - 1.0 | |||
| Emerging Asia | Structural | capital expenditure | 0.7 | 0.8 | |
| Freedman et al, 2008 | model | transfers | 0.4 | 0.2 | |
cumulative over three years.
References
Aguiar, M. and Gopinath, G., 2007. Emerging Market Business Cycles: The Cycle Is the Trend. Journal of Political Economy, University of Chicago Press, vol. 115, pp. 69–102.
Blanchard, O. and Perotti, R., 2002. An Empirical Characterization of the Dynamic Effects of Changes in Government Spending and Taxes on Output. Quarterly Journal of Economics 117, pp. 1329–1368.
Canova, F., 2007. Methods for Applied Macroeconomic Research. Princeton, NJ: Princeton University Press.
Dejong, D. and Dave, C., 2007, Structural Macroeconometrics, Princeton, NJ: Princeton University Press.
Hamilton, J. D., 1994. Time Series Analysis. Princeton. N.J.: Princeton University Press.
Ilzetzki, E., Mendoza, E., Vegh, C.A., 2013. “How Big (Small?) are Fiscal Multipliers?” Journal of Monetary Economics 60, pp. 239–254.
Prepared by Jean Frédéric Noah Ndela Ntsama.
In the Hodrick-Prescott filter, the parameter λ equals to the ratio of the volatility of the cycle component over the volatility of the trend. For emerging economies, the trend could generally be more volatile than in developed economies (see Aguiar and Gopinath, 2007), which might require a value of λ lower than the 1600 (calibrated for US quarterly data). However, given the smooth path of Morocco variable, the trend seems stable with a low standard deviation, thus we use λ = 1600.
There are three leading approaches for removing trend from macroeconomic time series: de-trending using a linear trend, differencing, and filtering (the most common used filters are the Hodrick-Prescott (H-P) filter and the Band Pass (B-P) filter) (see Canova, 2007, or Dejong and Dave, 2007). In this analysis, we obtain the cyclical component through differencing technique primarily because we are interested on the growth impact, which could easily be connected to the macro outlook analysis.
Note that a vector autoregressive model VAR (p) specified for an m x 1 vector has a companion form zt = Azt–1 + et. The sth -order covariance matrix is given by Γ(s) = AsΓ(0), where
On Figure 1, ΔLnGDPt stands for the Log difference of the real GDP, ΔLnNGDPt for the Log difference of the non-agriculture real GDP, ΔLnGCt the Log difference of the government consumption, TBt/GDPt the trade balance to GDP ratio, πt the inflation rate, it the nominal interest rate, and ΔLnREERt the Log difference of real effective exchange rate.
We conduct a series of test on the VAR representation. The VAR satisfies the stability condition since all its roots lie inside the unit circle. The maximum number of lags, three, introduced in the VAR model was determined using the sequential modified likelihood ratio test, and the Akaike and Schwarz based lower maximum likelihood criteria. The Portmanteau autocorrelation test, the normality test, and the White heteroskedasticity test performed on the residuals also confirm this VAR representation.
We use a rolling window SVAR (40 observations) and obtain to multiplier for each sub-period, then calculate the average. The idea is that Moroccan economy undergoes significant structural changes, which affect the transmission of shocks including the fiscal multipliers.
References
Agénor, P. R., and El Aynaoui, K., (2015). Morocco: Growth Strategy for 2025 in an Evolving International Environment. OCP.
AfDB (2014), Diagnostic de Croissance du Maroc.
Conseil Supérieur de l’Éducation, de la Formation et de la Recherche Scientifique, La Mise en Œuvre de la Charte Nationale d’Education et de Formation 2000-2013, Acquis, Déficits et Défit, December 2014.
Griogli 2014, A Hybrid Approach to Estimating the Efficiency of Public Spending on Education in Emerging Economies, IMF Working Paper No. 14/19.
Grigoli, F., and Kapsoli, J., Waste Not, Want Not: The Efficiency of Health Expenditures in Developing Economies, IMF Working Paper No. 13/187.
Gupta, Sanjeev, and Marijn Verhoeven, 2001, “The Efficiency of Government Expenditure. Experiences from Africa,” Journal of Policy Modelling, Vol. 23, pp. 433–67.
Gupta, Sanjeev, Gerd Schwartz, Shamsuddin Tareq, Richard Allen, Isabell Adenauer, Kevin Fletcher, and Duncan Last, 2007, “Fiscal Management of Scale-Up Aid,” IMF Working Paper No. 07/222 (Washington: International Monetary Fund).
Herrera, Santiago, and Gaobo Pang, 2005, “Efficiency of Public Spending in Developing Countries: An Efficiency Frontier Approach,” World Bank Policy Research Working Paper No. 3645 (Washington: World Bank).
IBRD (2014), Revue des Dépenses Publiques, Tome II: Secteur de l’Education.
Jayasuriya, Ruwan, and Quentin Wodon, 2003, “Efficiency in Reaching the Millennium Development Goals,” World Bank Working Paper No. 9 (Washington: World Bank).
Kumbhakar, S. C., and Knox Lovell, C. A., (2000). Stochastic Frontier Analysis. Cambridge University Press.
Prepared by Gregory Auclair and Dominique Fayad.
Abdeslam Seddiki, Minister of Labor and Social Affairs in Morocco, Interview” 'In Morocco, youth unemployment is driving up inequality," The Guardian, August 20, 2014.
The gender gap index can be interpreted as the percentage of the inequality between women and men that has been closed. The index is built on four groups of variables: economic participation and opportunity, educational attainment, political empowerment, and health and survival, among 130 countries.
Conseil Supérieur de l’Éducation, de la Formation et de la Recherche Scientifique, « La Mise en Œuvre de la Charte Nationale d’Education et de Formation 2000-2013, Acquis, Déficits et Défit », December 2014.
AfDB, Diagnostic de Croissance au Maroc, Agénor OCP.
Gupta and Verhoeven (2001), Herrera and Pang (2005), Gupta and other (2007), Jayasuriya and Wodon (2003).
Grigoli, F., and Kapsoli, J.
We apply the SFA on a panel of 44 countries. We assume a normal/half normal distribution of errors. The dependant variable is the 8th grade TIMSS scores averaged across math and sciences available for 2003, 2006, 2011. The explanatory variable is the logarithm of educational expenditure in primary and secondary education per student (USD PPP basis), average for the 4 preceding years. Above 8000 USD PPP of public spending per student, countries face decreasing marginal returns, therefore, we drop these countries from the sample in order to estimate the maximum efficiency gains Morocco can achieve by closing the gap with the frontier. We apply a SFA on the 8th grade TIMSS scores averaged across math and sciences and the public expenditures in primary and secondary education per student (USD PPP basis), average for the 4 preceding years in order to take the lagged effects of public spending on education into account.
Diversion of public funds is an indicator from the World Economic Forum that assesses how common is the reallocation of public funds to companies, individuals, or groups due to corruption.
IBRD (2014), Revue des Dépenses Publiques, Tome II: Secteur de l’Education. AfDB (2014), Diagnostic de Croissance du Maroc. Sutherland, D., Price, R., and Gonand, F. (2009), Improving Public Spending Efficiency in Primary and Secondary Education, OECD.
