Abstract
Contents include: basic data and social demographic indicators, 1971-95. Liberalizing the economy -- structural reform issues: reform of the fiscal system; public enterprise reform and increased scope for private sector activity; price liberalization and reform of consumer subsidies; environmental issues; financial sector; factor markets; and exchange system liberalization. Labor market developments; recent trends; structure of unemployment; reasons for unemployment; reforming labor markets. Determinants of nongovernment investment, 1972-95: nongovernment investment data; estimates of investment functions; explaining the performance of nongovernment investment. The associaiton agreement and the new fisheries agreement between Morocco and the EU. Statistical tables.
III. Determinants of Nongovernment Investment, 1972-9537
1. Summary and conclusions
Under Morocco’s adjustment program, pursued for more than a decade through 1992, the share of overall investment in GDP declined through 1988 and recovered only modestly during 1989-90; subsequently, investment more or less stagnated at this level through 1995 (Chart 3). Government investment explains most of the downturn through 1988. After a surge following the phosphate boom in the mid-1970s, it was sharply cut back in the late 1970s, and was reduced further during the 1981-92 adjustment period before it broadly stabilized at about 3 percent of GDP during 1992-95. In fact, fiscal adjustment during 1981-92 took place mainly through cutbacks in public investment and increased taxation. Nongovernment investment also surged following the phosphate boom, and fell sharply during the late 1970s. Subsequently, however, nongovernment investment broadly stabilized at around 18 percent of GDP during 1980-88, and after a modest upturn during 1989-91 declined again during the first half of the 1990s. Weaknesses in the data on public enterprise investment make it difficult to disentangle the share of public enterprise and of “truly” private investment within nongovernment investment (see below). Nevertheless, available data indicate that the above pattern broadly held for both components of nongovernment investment through 1988; however, public enterprise investment did not increase during 1989-91 and rose substantially during 1993-94, mainly due to large investments in the energy sector.
Morocco Investment
Citation: IMF Staff Country Reports 1997, 006; 10.5089/9781451824650.002.A003
Sources: Moroccan authorities; and Fund staff estimates.1/ The sharp increase in public enterprise investment in 1993-94 reflects mainly major projects in the energy sector, undertaken by the electricity company ONE.Morocco Investment
Citation: IMF Staff Country Reports 1997, 006; 10.5089/9781451824650.002.A003
Sources: Moroccan authorities; and Fund staff estimates.1/ The sharp increase in public enterprise investment in 1993-94 reflects mainly major projects in the energy sector, undertaken by the electricity company ONE.Morocco Investment
Citation: IMF Staff Country Reports 1997, 006; 10.5089/9781451824650.002.A003
Sources: Moroccan authorities; and Fund staff estimates.1/ The sharp increase in public enterprise investment in 1993-94 reflects mainly major projects in the energy sector, undertaken by the electricity company ONE.A better understanding of the modest performance in nongovernment investment, despite broadly successful macroeconomic stabilization and progress in structural reform in a number of areas, may help to formulate policies that address remaining obstacles to higher investment and, thus, growth. Accordingly, after describing the sources of data, an attempt is made to assess the main determinants of nongovernment investment, including the role of economic policies and external shocks, on the basis of empirical estimates of various nongovernment investment functions.
The econometric results indicate that nongovernment investment in Morocco during 1972-95 was primarily influenced by domestic policies. In contrast to previous studies,38 only limited evidence was found regarding demand and debt overhang effects on nongovernment investment; the latter is largely determined by the real effective exchange (REER) rate, the real interest rate, the share of government investment in GDP, and the growth of credit to the nongovernment sector.
In all likelihood, investment was also influenced by risk and uncertainty as perceived by investors—given the forward-looking but irreversible nature of private investment. Major sources of uncertainties were the volatility of weather conditions and world market prices, which heavily affected economic activity, but potentially also the credibility of government policies. The role of uncertainty, however, proved difficult to quantify. The variance of inflation and real interest rates, the foreign exchange premium in parallel markets, and debt and debt service ratios were tried as proxies for policy-induced uncertainty, including the uncertainty associated with any perceived debt-overhang. However, only external debt was found to have possibly adversely affected investment; at the same time, the fact that the estimated equation, which overall explained actual investment reasonably well but underpredicted the decline during 1984-88 and the increase thereafter, may reflect growing uncertainty in the first period and decreasing uncertainty in the subsequent period, as reform became more credible and was increasingly perceived as irreversible.
As regards the more recent years, the estimates suggest that the stagnation of nongovernment investment since 1990 reflects policies operating through the REER, real interest rate, and government investment levels, and that these factors had broadly offsetting effects. Specifically, investment was depressed by the steady appreciation of the REER, while increases in government investment and credit availability were insufficient to fully overcome these effects.
2. Nongovernment investment data, 1970-95
Separate private and public enterprise investment data are not published in Morocco; accordingly, the subsequent analysis is based on nongovernment investment as reported in the national accounts statistics.39 Nongovernment investment at constant prices is estimated by using a weighted average of the price index for imported capital goods and deflators for the domestic value added of the construction and agriculture sectors:
The inclusion of public enterprises, however, complicates the analysis, as the behavior of most public enterprises may have differed from that of the “truly” private sector, owing to different budget constraints and different access to domestic or foreign credit. Moreover, partial estimates of public enterprise investment40 indicate that its share in nongovernment investment has been sizeable and varied considerably (Chart 3). In view of these considerations, an attempt has been made to also estimate separate investment functions for public enterprises and the private sector, respectively. However, these estimates have to be treated with caution, given the weaknesses in the available data.
3. Estimates of investment functions
Model specification
The initial specification of the function for nongovernment investment draws on the recent literature on determinants of investment in developing countries,41 and includes the following explanatory variables: (1) real growth rate of GDP, lagged one period, to capture demand effects42; (2) real interest rate, measured by the nominal interest rate on the 6-month treasury bills adjusted by the actual rate of CPI inflation, as a proxy for the real cost of capital; (3) real growth rate of banking sector credit to the nongovernment sector,43 to capture the influence of domestic borrowing constraints on investment (both the real interest rate and the volume of bank credit are included in the investment function because administrative restrictions affecting both these variables were in force during part of the sample period),44 (4) levels of foreign exchange reserves (expressed in months of imports) and of workers’ remittances, to reflect the access to foreign exchange by the private sector (the latter also attempts to capture income effects on household investment (e.g. housing)); (5) ratio of government investment to GDP (measured in constant prices), to examine possible complementarity or substitutability between nongovernment and government investments; (6) macroeconomic uncertainty, measured by the variance of the monthly CPI inflation rate as well as the foreign exchange premium (defined as the ratio of the parallel market rate to the official rate); (7) external debt overhang effects on nongovernment investment, measured by the ratio of the external debt to GDP and the ratio of debt service to exports of goods and services; (8) real effective exchange rate (REER) (the expected sign of the short-run impact of the REER on investment is ambiguous: a depreciation encourages investment in the traded goods sector, but also results in a higher relative price for imported capital goods); and (9) terms of trade (measured by the income effect of changes in export and in import prices).45 Finally, the one-period lagged investment rate (to capture adjustment lags and other inertial factors); and a dummy variable for 1980 (when due to a revision of the methodology in Morocco’s national accounts statistics, a statistical break in the GDP and related series occurred, affecting some of the explanatory variables) were also included.
In the empirical model, all variables are expressed either in scaled form or in rates of change to reduce the risk of spurious correlation among trended variables. All regressors were checked for nonstationarity by examining the behavior of the sample autocorrelation function. The equations were estimated with annual data for the period 1972-95, using ordinary least squares procedures; the standard errors shown are heteroskedastic-consistent estimates.46
Empirical results
The regression results are reported in Table 2. Estimates of the nongovernment investment function including all potential variables yielded implausible (wrong-signed) and insignificant parameter estimates for several variables (the variance of inflation rates, changes in the terms of trade, foreign exchange reserves, and import price shocks). An initial specification of the investment function, excluding these variables (equation 1 in Table 2), was then estimated; from this latter specification, statistically insignificant explanatory variables—the exchange rate premium, workers remittances, external debt service, external debt, real GDP growth, and export price shocks—were successively removed, to arrive at the final specification of the investment function (equation 7 in Table 2).47 The final equation indicates that despite the considerable accumulation of external debt and subsequent debt restructuring during this period, debt overhang effects had no statistically significant influence on nongovernment investment.48 Equally remarkable is the lack of significance in the demand variable. The growth rate of economic activity (lagged real GDP growth) has a positive, but not statistically significant effect which may reflect the fact that variations in GDP were dominated by weather-related supply shocks which had no large effect on overall nongovernment investment.
Morocco: Investment Function – Specification Search 1/
(Dependent variable: gross fixed nongovernment investment/GDP, constant prices, in logs) 2/
(OLS estimates, 1972 to 1995)
Standard errors and t-statistics (**=1%; *=5%) were computed using White’s heteroscedastic–consistent–variance–covariance estimator.
Government sector includes central government, local governments, public nonprofit organizations, and the national social security fund. Thus, nongovernment investment includes public enterprise investment.
Morocco: Investment Function – Specification Search 1/
(Dependent variable: gross fixed nongovernment investment/GDP, constant prices, in logs) 2/
(OLS estimates, 1972 to 1995)
| Variable: | EQ1 | EQ2 | EQ3 | EQ4 | EQ5 | EQ6 | EQ7 |
|---|---|---|---|---|---|---|---|
| Constant | 5.280** | 4.850** | 5.530** | 5.110** | 4.970** | 5.140** | 5.360** |
| (5.24) | (4.98) | (8.98) | (7.55) | (7.38) | (7.19) | (7.20) | |
| Real interest rate (level, deflated with CPI) | –0.003 | –0.006 | –0.004 | –0.006 | –0.007 | –0.008 | –0.013** |
| (–0.67) | (–1.26) | (–1.41) | (–1.18) | (–1.73) | (–1.66) | (–3.43) | |
| Real growth of credit stock to the nongovernment sector 2/ | 0.004* | 0.005* | 0.005** | 0.005** | 0.005** | 0.005** | 0.005* |
| (2.33) | (3.01) | (3.08) | (3.14) | (3.50) | (3.44) | (2.58) | |
| Government investment/GDP (constant prices, in logs) | 0.367* | 0.329* | 0.353** | 0.327** | 0.323** | 0.329** | 0.290** |
| (5.78) | (5.23) | (5.49) | (5.34) | (5.38) | (5.14) | (5.85) | |
| REER (level, in logs) | –1.020** | –0.937** | –1.090** | –1.030** | –0.993** | –1.020** | –1.010** |
| (–5.09) | (–4.58) | (–7.17) | (–6.16) | (–6.07) | (–6.12) | (–6.28) | |
| Lagged dependent variable 3/ | 0.476** | 0.513** | 0.554** | 0.621** | 0.578** | 0.560** | 0.495** |
| (3.76) | (4.25) | (4.20) | (5.99) | (5.78) | (5.59) | (5.29) | |
| Dummy for statistical break in GDP series in 1980 | 0.203** | 0.206** | 0.227** | 0.241** | 0.234** | 0.238** | 0.242** |
| (4.76) | (4.60) | (5.50) | (7.70) | (7.47) | (7.69) | (7.95) | |
| Export price shock (% of GDP) | 0.016* | 0.013 | 0.014 | 0.014 | 0.013 | 0.011 | |
| (2.62) | (1.88) | (1.82) | (1.83) | (1.78) | (1.37) | ||
| Real GDP growth (lagged one period) | 0.005* | 0.004 | 0.004 | 0.004 | 0.004 | ||
| (2.34) | (1.76) | (1.74) | (1.54) | (1.55) | |||
| External debt/GDP (in logs) | –0.089* | –0.073 | –0.049 | –0.024 | |||
| (–2.34) | (–2.13) | (–1.98) | (–0.90) | ||||
| External debt service/exports of goods and services G&S (in logs) | 0.045 | 0.036 | 0.031 | ||||
| (1.36) | (1.09) | (1.03) | |||||
| Workers’ remittances/GDP (current prices, in logs) | 0.129 | 0.119 | |||||
| (1.27) | (1.21) | ||||||
| Exchange rate premium (in logs) | –0.155 | ||||||
| (–1.13) | |||||||
| Number of observations | 24 | 24 | 24 | 24 | 24 | 24 | 24 |
| Adj. R-sq | 0.81 | 0.81 | 0.81 | 0.82 | 0.83 | 0.83 | 0.82 |
| S.E.E. | 0.07 | 0.07 | 0.07 | 0.06 | 0.06 | 0.06 | 0.06 |
| F-statistic | 8.9** | 10.0** | 10.9** | 12.8** | 15.1** | 16.6** | 18.4* |
| Durbin’s h for autocorrelation | –0.33 | –3.26** | –3.85** | –1.16 | –0.75 | –1.47 | –0.59 |
| Breusch-Pagan test for heteroscedasticity | 0.17 | 0.05 | 1.95 | 1.70 | 1.81 | 1.62 | 2.62 |
| Jarque-Bera test for normality of residuals | 0.17 | 0.88 | 1.97 | 1.46 | 1.48 | 1.08 | 0.90 |
Standard errors and t-statistics (**=1%; *=5%) were computed using White’s heteroscedastic–consistent–variance–covariance estimator.
Government sector includes central government, local governments, public nonprofit organizations, and the national social security fund. Thus, nongovernment investment includes public enterprise investment.
Morocco: Investment Function – Specification Search 1/
(Dependent variable: gross fixed nongovernment investment/GDP, constant prices, in logs) 2/
(OLS estimates, 1972 to 1995)
| Variable: | EQ1 | EQ2 | EQ3 | EQ4 | EQ5 | EQ6 | EQ7 |
|---|---|---|---|---|---|---|---|
| Constant | 5.280** | 4.850** | 5.530** | 5.110** | 4.970** | 5.140** | 5.360** |
| (5.24) | (4.98) | (8.98) | (7.55) | (7.38) | (7.19) | (7.20) | |
| Real interest rate (level, deflated with CPI) | –0.003 | –0.006 | –0.004 | –0.006 | –0.007 | –0.008 | –0.013** |
| (–0.67) | (–1.26) | (–1.41) | (–1.18) | (–1.73) | (–1.66) | (–3.43) | |
| Real growth of credit stock to the nongovernment sector 2/ | 0.004* | 0.005* | 0.005** | 0.005** | 0.005** | 0.005** | 0.005* |
| (2.33) | (3.01) | (3.08) | (3.14) | (3.50) | (3.44) | (2.58) | |
| Government investment/GDP (constant prices, in logs) | 0.367* | 0.329* | 0.353** | 0.327** | 0.323** | 0.329** | 0.290** |
| (5.78) | (5.23) | (5.49) | (5.34) | (5.38) | (5.14) | (5.85) | |
| REER (level, in logs) | –1.020** | –0.937** | –1.090** | –1.030** | –0.993** | –1.020** | –1.010** |
| (–5.09) | (–4.58) | (–7.17) | (–6.16) | (–6.07) | (–6.12) | (–6.28) | |
| Lagged dependent variable 3/ | 0.476** | 0.513** | 0.554** | 0.621** | 0.578** | 0.560** | 0.495** |
| (3.76) | (4.25) | (4.20) | (5.99) | (5.78) | (5.59) | (5.29) | |
| Dummy for statistical break in GDP series in 1980 | 0.203** | 0.206** | 0.227** | 0.241** | 0.234** | 0.238** | 0.242** |
| (4.76) | (4.60) | (5.50) | (7.70) | (7.47) | (7.69) | (7.95) | |
| Export price shock (% of GDP) | 0.016* | 0.013 | 0.014 | 0.014 | 0.013 | 0.011 | |
| (2.62) | (1.88) | (1.82) | (1.83) | (1.78) | (1.37) | ||
| Real GDP growth (lagged one period) | 0.005* | 0.004 | 0.004 | 0.004 | 0.004 | ||
| (2.34) | (1.76) | (1.74) | (1.54) | (1.55) | |||
| External debt/GDP (in logs) | –0.089* | –0.073 | –0.049 | –0.024 | |||
| (–2.34) | (–2.13) | (–1.98) | (–0.90) | ||||
| External debt service/exports of goods and services G&S (in logs) | 0.045 | 0.036 | 0.031 | ||||
| (1.36) | (1.09) | (1.03) | |||||
| Workers’ remittances/GDP (current prices, in logs) | 0.129 | 0.119 | |||||
| (1.27) | (1.21) | ||||||
| Exchange rate premium (in logs) | –0.155 | ||||||
| (–1.13) | |||||||
| Number of observations | 24 | 24 | 24 | 24 | 24 | 24 | 24 |
| Adj. R-sq | 0.81 | 0.81 | 0.81 | 0.82 | 0.83 | 0.83 | 0.82 |
| S.E.E. | 0.07 | 0.07 | 0.07 | 0.06 | 0.06 | 0.06 | 0.06 |
| F-statistic | 8.9** | 10.0** | 10.9** | 12.8** | 15.1** | 16.6** | 18.4* |
| Durbin’s h for autocorrelation | –0.33 | –3.26** | –3.85** | –1.16 | –0.75 | –1.47 | –0.59 |
| Breusch-Pagan test for heteroscedasticity | 0.17 | 0.05 | 1.95 | 1.70 | 1.81 | 1.62 | 2.62 |
| Jarque-Bera test for normality of residuals | 0.17 | 0.88 | 1.97 | 1.46 | 1.48 | 1.08 | 0.90 |
Standard errors and t-statistics (**=1%; *=5%) were computed using White’s heteroscedastic–consistent–variance–covariance estimator.
Government sector includes central government, local governments, public nonprofit organizations, and the national social security fund. Thus, nongovernment investment includes public enterprise investment.
These regression results point to the fact that the following domestic policy variables were the main determinants of nongovernment investment in Morocco:
The real interest rate has a negative and statistically significant effect on the rate of nongovernment investment—an increase in the real interest rate of 1 percentage point of GDP reduces investment by 0.08 percentage point of GDP (evaluated at the sample mean). The real rate of growth of domestic bank credit to the nongovernment sector also has a statistically significant positive effect on nongovernment investment, consistent with the presence of credit rationing during the sample period. This effect is smaller than that of the interest rate: a one percentage point increase in real credit raises the investment relative to GDP by 0.03 percentage point (evaluated at the sample mean).
The share of government investment (i.e. excluding public enterprises) in GDP is found to have a statistically significant positive effect on the nongovernment investment rate. This suggests a strong complementarity between government and nongovernment investment.49
The real effective exchange rate is found to have a sizeable negative and statistically significant effect on the nongovernment investment rate. An exchange rate depreciation of 10 percent in real effective terms would increase the investment rate by 6 percent; this would suggest that any negative “cost” effect of a real depreciation would be more than offset by the positive “competitiveness” effects for the economy as a whole.
While the preferred equation tracks the observed path of investment reasonably well, the model misses some turning points, for example in 1991 and in 1992, underpredicts the investment rate in the early 1980s, and overpredicts it during the mid-1980s (Chart 4). This may reflect the role of uncertainty—not captured in equation 7—on investment. The period 1979-83 was marked by a major shift in stabilization policies and the onset of serious structural reform in the mid-1980s was likely associated with considerable uncertainty. As a result, the incentive for investors to wait before embarking on irreversible investments may have been strong.
Morocco Actual and Estimated Investment, 1972–95 1/
(In percent of GDP)
Citation: IMF Staff Country Reports 1997, 006; 10.5089/9781451824650.002.A003
Sources: Moroccan authorities; and Fund staff estimates.Morocco Actual and Estimated Investment, 1972–95 1/
(In percent of GDP)
Citation: IMF Staff Country Reports 1997, 006; 10.5089/9781451824650.002.A003
Sources: Moroccan authorities; and Fund staff estimates.Morocco Actual and Estimated Investment, 1972–95 1/
(In percent of GDP)
Citation: IMF Staff Country Reports 1997, 006; 10.5089/9781451824650.002.A003
Sources: Moroccan authorities; and Fund staff estimates.The specification of the function for nongovernment investment (equation 7) was also estimated for public enterprise investment and private investment separately (Table 3), although it may not be the preferred specification if the specification search exercise were to be repeated for private investment and public enterprise investment separately. With this caveat in mind, the estimated equations point to some interesting differences between private and public enterprise investment behavior. While the real interest remains highly significant for private investment (with a larger coefficient), public enterprise investment seems less affected by the real interest rate—consistent with the hypothesis that public enterprise investment is less concerned with profit maximization. By contrast, credit to the nongovernment sector is, not surprisingly, significantly important for public enterprise investment, while the sensitivity of private sector investment to credit growth is lower, and insignificant. Thus, public enterprise investment appears to have been credit-constrained while the cost of capital (proxied by the real interest rate) and profitability in the tradables sector (real effective exchange rate) were found to be more important constraints for private sector investment. As indicated in Chart 4, overall these equations track actual public enterprise and private investment, respectively, less well than the model for overall nongovernment investment, particularly in the early 1990s.
Morocco: Private Sector and Public Enterprise Investment Functions 1/
(Dependent variables: gross fixed private investment/GDP and gross fixed public enterprise investment, constant prices, in logs)
(OLS estimates, 1972 to 1995)
Standard errors and t–statistics (in parentheses: **=1% *=5%) were computed using White’s heteroscedastic–consistent–variance–covariance estimator.
Morocco: Private Sector and Public Enterprise Investment Functions 1/
(Dependent variables: gross fixed private investment/GDP and gross fixed public enterprise investment, constant prices, in logs)
(OLS estimates, 1972 to 1995)
| Variable: | Private sector Investment Function | Public Enterprise Investment Function |
|---|---|---|
| Constant | 5.800** | 3.430* |
| (5.88) | (2.24) | |
| Real interest rate (level, deflated with CPI) | –0.016** | –0.007 |
| (–3.40) | (–1.11) | |
| Real growth of credit stock to the nongovernment sector | 0.002 | 0.009* |
| (1.26) | (2.78) | |
| Government investment/GDP (constant prices, in logs) | 0.265** | 0.351* |
| (4.67) | (2.89) | |
| REER (level, in logs) | –1.140** | –0.702 |
| (–5.26) | (–2.04) | |
| Lagged dependent variable | 0.491** | 0.414* |
| (4.75) | (2.42) | |
| Dummy for GDP break in 1980 | 0.344** | –0.009 |
| (7.93) | (–0.18) | |
| Number of observations | 24 | 24 |
| Adj. R–sq | 0.67 | 0.50 |
| S.E.E. | 0.10 | 0.16 |
| F–statistic | 8.73** | 4.87** |
| Durbin’s h for autocorrelation | –1.66 | 2.17* |
| Breusch–Pagan test for heteroscedasticity | 5.02 | 4.30 |
| Jarque–Bera test for normality of residuals | 0.47 | 5.54 |
Standard errors and t–statistics (in parentheses: **=1% *=5%) were computed using White’s heteroscedastic–consistent–variance–covariance estimator.
Morocco: Private Sector and Public Enterprise Investment Functions 1/
(Dependent variables: gross fixed private investment/GDP and gross fixed public enterprise investment, constant prices, in logs)
(OLS estimates, 1972 to 1995)
| Variable: | Private sector Investment Function | Public Enterprise Investment Function |
|---|---|---|
| Constant | 5.800** | 3.430* |
| (5.88) | (2.24) | |
| Real interest rate (level, deflated with CPI) | –0.016** | –0.007 |
| (–3.40) | (–1.11) | |
| Real growth of credit stock to the nongovernment sector | 0.002 | 0.009* |
| (1.26) | (2.78) | |
| Government investment/GDP (constant prices, in logs) | 0.265** | 0.351* |
| (4.67) | (2.89) | |
| REER (level, in logs) | –1.140** | –0.702 |
| (–5.26) | (–2.04) | |
| Lagged dependent variable | 0.491** | 0.414* |
| (4.75) | (2.42) | |
| Dummy for GDP break in 1980 | 0.344** | –0.009 |
| (7.93) | (–0.18) | |
| Number of observations | 24 | 24 |
| Adj. R–sq | 0.67 | 0.50 |
| S.E.E. | 0.10 | 0.16 |
| F–statistic | 8.73** | 4.87** |
| Durbin’s h for autocorrelation | –1.66 | 2.17* |
| Breusch–Pagan test for heteroscedasticity | 5.02 | 4.30 |
| Jarque–Bera test for normality of residuals | 0.47 | 5.54 |
Standard errors and t–statistics (in parentheses: **=1% *=5%) were computed using White’s heteroscedastic–consistent–variance–covariance estimator.
Stability of the estimated coefficients
Morocco experienced a marked shift in policy regime when the unsustainable expansionary policies of the 1970s were replaced by stabilization policies in the 1980s. Moreover, major structural reforms, such as the liberalization of interest rates since 1985, may well have altered the way in which investment responds to policies changes. For these reasons, the assumption of time-invariant parameters in the estimated nongovernment investment function may not be valid. Accordingly, the stability of the estimated investment function has been examined in order to investigate how sensitive the estimated coefficients are to changes in the sample period. The forward recursions are based on an initial sample of 1972 to 1987, to which annual observations are added until reaching the full sample (Chart 5). The backward recursions are based on the full sample of 1972-95, from which annual observations are subtracted one-by-one (Chart 6). The main results that emerge from the recursive estimates are the following:
Morocco: Recursive Least Square Estimates of Nongovernment Investment Equation
(Slope Parameters in Forward Recursions) 1/
Citation: IMF Staff Country Reports 1997, 006; 10.5089/9781451824650.002.A003
1/ Recursive least squares estimates were based on samples starting in 1972 and ending in the year displayed on the time axis. Forward recursions added annual observations one-by-one to an initial sample covering 1972-87. Standard errors were based on White’s heteroscedasticity-consistent variance-covariance estimator.Morocco: Recursive Least Square Estimates of Nongovernment Investment Equation
(Slope Parameters in Forward Recursions) 1/
Citation: IMF Staff Country Reports 1997, 006; 10.5089/9781451824650.002.A003
1/ Recursive least squares estimates were based on samples starting in 1972 and ending in the year displayed on the time axis. Forward recursions added annual observations one-by-one to an initial sample covering 1972-87. Standard errors were based on White’s heteroscedasticity-consistent variance-covariance estimator.Morocco: Recursive Least Square Estimates of Nongovernment Investment Equation
(Slope Parameters in Forward Recursions) 1/
Citation: IMF Staff Country Reports 1997, 006; 10.5089/9781451824650.002.A003
1/ Recursive least squares estimates were based on samples starting in 1972 and ending in the year displayed on the time axis. Forward recursions added annual observations one-by-one to an initial sample covering 1972-87. Standard errors were based on White’s heteroscedasticity-consistent variance-covariance estimator.
Morocco: Recursive Least Square Estimates of Nongovernment Investment Equation
(Slope Parameters in Forward Recursions) 1/
Citation: IMF Staff Country Reports 1997, 006; 10.5089/9781451824650.002.A003
1/ Recursive least squares estimates were based on samples starting in 1972 and ending in the year displayed on the time axis. Forward recursions added annual observations one-by-one to an initial sample covering 1972-87. Standard errors were based on White’s heteroscedasticity-consistent variance-covariance estimator.Morocco: Recursive Least Square Estimates of Nongovernment Investment Equation
(Slope Parameters in Forward Recursions) 1/
Citation: IMF Staff Country Reports 1997, 006; 10.5089/9781451824650.002.A003
1/ Recursive least squares estimates were based on samples starting in 1972 and ending in the year displayed on the time axis. Forward recursions added annual observations one-by-one to an initial sample covering 1972-87. Standard errors were based on White’s heteroscedasticity-consistent variance-covariance estimator.Morocco: Recursive Least Square Estimates of Nongovernment Investment Equation
(Slope Parameters in Forward Recursions) 1/
Citation: IMF Staff Country Reports 1997, 006; 10.5089/9781451824650.002.A003
1/ Recursive least squares estimates were based on samples starting in 1972 and ending in the year displayed on the time axis. Forward recursions added annual observations one-by-one to an initial sample covering 1972-87. Standard errors were based on White’s heteroscedasticity-consistent variance-covariance estimator.There is considerable stability in coefficient estimates for all the explanatory variables except the real interest rate, which exhibits weak stability,5051 in that one cannot, at a two-standard error confidence level, reject the possibility of a switch in the sign when the full sample is not included. This may be attributable to the lack of interest rate flexibility until the mid-1980s.
The standard errors of the coefficient estimates for all explanatory variables decreases as observations for the late-1980s and early 1990s are included in the sample. Only private sector credit standard errors diverge somewhat when observations from the early 1970s are included. Overall, however, the clear reduction in the standard errors substantiates the robustness of the estimated coefficients.
The absolute value of the negative coefficients of the real interest rate effect on investment declines when observations after 1990 are included, and the significance of the coefficient increases somewhat relative to the late 1980s.
The influence of bank credit on nongovernment investment equally falls when observations after 1990 are included. This may reflect the impact of financial liberalization and greater reliance on indirect instruments of monetary policy.
The complementarity of government and nongovernment investments (crowding-in) is observed throughout the sample recursions.
An appreciation of the REER has a significant negative effect on investment, and the standard error of the estimate is reduced when observations from the 1970s are included in the sample. This impact of the REER on nongovernment investment suggests that while the steady real depreciation through 1990 raised the relative price of imported capital goods, any adverse effect on investment was more than offset by positive effects from the depreciation, such as increased profitability of the tradables sector, as well as benefits from trade liberalization, such as lower tariff rates.
4. Explaining the performance of nongovernment investment
To assess how policy and exogenous factors influenced the rate of nongovernment investment, the preferred investment equation has been used to decompose the estimated change in nongovernment investment into the contribution of each explanatory variable. Table 4 presents this decomposition for four different subsamples based on different phases of the nongovernment investment cycle: strong decline (1977-79), pause or minor decline (1981-88), recovery (1988-90), and stagnation (1990-95).
Morocco: Contribution of Explanatory Variables to the Changes in Nongovernment Investment (NGI)
(Based on equation 7, Table 2)
Morocco: Contribution of Explanatory Variables to the Changes in Nongovernment Investment (NGI)
(Based on equation 7, Table 2)
| 1977–79 | 1981–88 | 1988–90 | 1990–95 | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Explanatory Variable | Estimated Coefficient 1972–95 | Estimated Change in NGI/GDP | Share in % explained by: | Estimated Change in NGI/GDP | Share in % explained by: | Estimated Change in NGI/GDP | Share in % explained by: | Estimated Change in NGI/GDP | Share in % explained by: |
| Real interest rate | –0.13 | –1.34 | 27 | –2.40 | 132 | 1.14 | 42 | 0.02 | –1 |
| Real growth of credit to the nongovernment sector | 0.005 | 0.38 | –7 | 0.04 | –2 | 0.19 | 7 | 0.95 | –58 |
| Government investment/GDP (constant prices) | 0.29 | –2.47 | 49 | –3.68 | 202 | –0.70 | –26 | 0.51 | –31 |
| REER | –1.01 | –0.59 | 11.5 | 5.30 | –291 | 1.01 | 37 | –2.55 | 155 |
| Lagged dependent variable | 0.495 | –1.03 | 19.5 | –1.08 | 59 | 1.08 | 40 | –0.58 | 35 |
| Total change in NGI/GDP | –5.05 | 100 | –1.82 | 100 | 2.72 | 100 | –1.65 | 100 | |
| Actual change in NGI/GDP | –6.05 | –3.15 | 3.25 | –0.46 | |||||
| Prediction Error | 1.00 | 1.33 | 0.53 | –1.19 | |||||
Morocco: Contribution of Explanatory Variables to the Changes in Nongovernment Investment (NGI)
(Based on equation 7, Table 2)
| 1977–79 | 1981–88 | 1988–90 | 1990–95 | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Explanatory Variable | Estimated Coefficient 1972–95 | Estimated Change in NGI/GDP | Share in % explained by: | Estimated Change in NGI/GDP | Share in % explained by: | Estimated Change in NGI/GDP | Share in % explained by: | Estimated Change in NGI/GDP | Share in % explained by: |
| Real interest rate | –0.13 | –1.34 | 27 | –2.40 | 132 | 1.14 | 42 | 0.02 | –1 |
| Real growth of credit to the nongovernment sector | 0.005 | 0.38 | –7 | 0.04 | –2 | 0.19 | 7 | 0.95 | –58 |
| Government investment/GDP (constant prices) | 0.29 | –2.47 | 49 | –3.68 | 202 | –0.70 | –26 | 0.51 | –31 |
| REER | –1.01 | –0.59 | 11.5 | 5.30 | –291 | 1.01 | 37 | –2.55 | 155 |
| Lagged dependent variable | 0.495 | –1.03 | 19.5 | –1.08 | 59 | 1.08 | 40 | –0.58 | 35 |
| Total change in NGI/GDP | –5.05 | 100 | –1.82 | 100 | 2.72 | 100 | –1.65 | 100 | |
| Actual change in NGI/GDP | –6.05 | –3.15 | 3.25 | –0.46 | |||||
| Prediction Error | 1.00 | 1.33 | 0.53 | –1.19 | |||||
A common feature of the first three investment phases is that the growth of real bank credit contributed little to the changes in investment. The late 1970s, marked by unsustainable expansionary policies coupled with macroeconomic instability, led to adverse effects including a sharp decline in nongovernment investment. A sharp rise in the real interest rate, a decline in government investment, and an appreciation of the real effective exchange rate all contributed to a decline of over 6 percentage points of GDP in nongovernment investment during 1977-79. Domestic adjustment policies were the major influences on nongovernment investment between 1981 and 1988, when it fell by 3.2 percentage points of GDP. The large depreciation of the exchange rate in real effective terms during this period exerted a strong positive impact on nongovernment investment, which was more than offset by a severe cut in government investment by 4 percentage points of GDP and an increase in real interest rates. However, only 58 percent of the change in investment between 1981 and 1988 is explained by the equation. Moreover, the path of actual investment during this period is underpredicted during 1981-83, then overpredrcted during 1984-88 by the equation; as discussed above, this may be indicative of the influence of uncertainty on investment decisions.
During 1988-90, nongovernment investment recovered by more then 3 percentage points of GDP to the highest level since the early eighties. A modest decline in real interest rates, a small further depreciation of the real effective exchange rate, and increased growth of credit to the private sector all had small positive effects on nongovernment investment. Government investment continued to be cut back in relation to GDP, but only by a small amount, and this had only a minor negative impact on nongovernment investment. The subsequent fluctuation of nongovernment investment around a broadly flat trend between 1990 and 1995 reflected primarily the offset of an adverse effect of the real effective appreciation of the dirham by positive effects from increased government investment and credit growth to the private sector.
Prepared by K. Enders, J. Zhou, and P. Kunzel.
See Schmidt-Hebbel and Müller (1991), and Goldsbrough et al. (1996).
National accounts data on gross fixed capital formation by the government and the nongovernment sectors, measured at current prices, are available through 1994. For 1995, staff estimates for government investment are based on data for capital expenditure in the central government budget. The government sector includes the central government, local administrations, public non profit organizations, and the public social security system. Nongovernment investment includes both investment undertaken by public enterprises and investment by the “true” private sector.
Based on data available for a group of major enterprises monitored by the Ministry of Finance and which cover a shorter period (1975-95), separate series were derived for public enterprise and “true” private sector investments, respectively. However, since coverage of the services may have varied overtime, this breakdown between public enterprise investment and private investment should be considered with caution.
See Goldsbrough et al (1996).
In order to avoid simultaneity bias, lagged rather than current real GDP growth has been used. The ratio of current to trend GDP was also tried in the initial estimations, to capture the effect of the output gap, but the results proved to be statistically insignificant.
This measure includes credit from the development banks. The real growth of credit is obtained using the investment deflator.
For example, ceilings were imposed on several categories of credit; also, even when firms are credit-constrained, a change in interest rates can still affect investment through its impact on the discount rate and thus on the choice of the use of retained earning.
Ordinary least squares procedures were used because in small samples this procedure is more robust to misspecification errors.
This reduced model has been derived by performing F-tests for the exclusion of explanatory variables.
The estimates of the same investment functions for private and public enterprise investments taken separately (Table 3) indicate, however, that debt overhang effects might have affected public enterprise investment, in line with Schmidt-Hebbel and Müller (1991) which had found evidence of debt overhang effects for nongovernment investment during 1970-88. Even then, though it is possible that indicators of external debt are measuring the availability of foreign financing for public enterprise investment rather than overhang effects.
Similar evidence of complementarity was also found by Schmidt-Hebbel and Müller (1991).
Estimates based on a sample for the years 1972-87 are given by the far left observation in the forward recursion chart, and on a sample for the late 1980s and early 1990s by the far right observation in the backward recursion chart.
A formal Chow test for stability was not possible because of the presence of heteroskedasticity in the residuals.
