Abstract
Central African Economic and Monetary Community (CEMAC) Selected Issues
Cemac: The Impact of Oil Prices on Non-Oil Growth1
This paper aims to strengthen the analytical basis for the study of the economic impact of the recent oil-price shock on the Economic and Monetary Community of Central Africa (CEMAC) countries.2 The analysis, performed with the use of dynamic panel regressions, reveals a strong link between oil-price volatility and non-oil GDP growth in the oil-producing CEMAC countries. The empirical relationship is significant in statistical and in economic terms, and appears to be stronger in CEMAC than within other oil-producing countries. Subsequently, the regression coefficients are used to derive the impulse responses of non-oil GDP growth in CEMAC countries to the recent oil-price shock in the short to medium term. Finally, the analysis suggests that the impact of oil-price volatility on non-oil GDP growth is only partially intermediated by the concomitant change in government spending with other important channels being at play.
A. Introduction
1. Oil production is the key economic activity and a main source of foreign exchange and government revenues for five countries in CEMAC. Government oil revenues increased from 16 percent of all revenues in 1994 to a peak of 70 percent in 2008, before dropping to 58 percent in 2014. Oil dependency varies among CEMAC producers, which are at different stages of their oil sector development. In Cameroon, which is CEMAC’s smallest oil producer, production peaked more than 20 years ago, while in Chad, production started just over a decade ago and is currently on the rise. Congo showed a relatively steady output throughout the period under study. In Equatorial Guinea and Gabon, the CEMAC’s first and third largest oil producers, respectively, production is now entering a declining stage (Box 1).
2. CEMAC’s high dependency on oil has made its economies vulnerable to oil price shocks. This vulnerability manifested itself in the aftermath of the 1998 and the 2008 oil price shocks and is evident in the difficulty CEMAC oil producers experience in adjusting their policies to the current oil price slump, which started in the second half of 2014.
3. Part of this difficulty rests in that the link between oil price volatility and non-oil GDP growth is not well understood. This implies that the policy response (in most cases, fiscal tightening) may have side effects that are hard to foresee or quantify. This paper uses dynamic panel regression analysis to estimate the impact of oil price volatility on non-oil GDP growth, which could help to design appropriate policy responses to the current shock. The analysis also examines the role government spending plays as a transmission channel. The latter appears less effective, or at least less direct, than is commonly believed, when compared to other channels through which oil-price volatility affects non-oil economic activity.
CEMAC: Key Economic Facts 1994–2014
The history of oil production has been shaped by the increase in the number of oil producers from an initial three to the current five. In 1994, Cameroon, Congo, and Gabon were the three oil producers of the region, jointly producing about 650 thousand barrels of oil per day (bpd). The number of oil producers gradually increased with Equatorial Guinea and Chad joining in 1995 and 2003, respectively. Over time, the volume of oil production has expanded in Equatorial Guinea and Chad, and declined in Cameroon. The regional production peaked at 1.14 million bpd in 2005 and has been on a downward trend since then, amounting to 915 thousand bpd in 2014.
Over the last two decades, the increasing revenue from oil provided CEMAC with the resources needed to promote economic growth and build resilience. The characteristics of CEMAC economies have evolved significantly since 1994 with oil being the main factor in this change. CEMAC used to be a homogenous group of low-income countries. Two members (Equatorial Guinea and Gabon) became middle-income countries in the mid-2000s and the former acquired a high-income status toward the end of the last decade. With the CEMAC’s international reserves increasing from US$0.5 billion in 1994 to US$15 billion in 2013, the region has also gained the ability to cushion its path through an oil shock, but it has yet to achieve a greater degree of economic diversification and reduce its dependency on oil revenues.
CEMAC: Crude oil production
(Thousands of barrels per day)
Citation: IMF Staff Country Reports 2015, 308; 10.5089/9781513510583.002.A004
CEMAC: Oil revenue
(Billions of CFA francs)
Citation: IMF Staff Country Reports 2015, 308; 10.5089/9781513510583.002.A004
CEMAC: Oil exports
(Millions of US dollars)
Citation: IMF Staff Country Reports 2015, 308; 10.5089/9781513510583.002.A004
CEMAC: Official reserves
(Millions of US dollars)
Citation: IMF Staff Country Reports 2015, 308; 10.5089/9781513510583.002.A004
B. Stylized Facts of the Economic Impact of Oil-Price Volatility
4. Economic activity in CEMAC was highly affected by the two previous major oil price shocks.3 During the first shock (in 1996–98), oil prices dropped by 55 percent between November 1996 and December 1998. During the second shock (in 2008) they dropped by 67 percent between July and December (Figure 1). Oil exports and government revenues from oil contracted sharply as a result of the falling oil prices (Figure 2). In these episodes, real GDP growth declined from 10.8 percent in 1997 to 3.2 percent in 1999 and from 4.2 percent in 2008 to 1.0 percent in 2009 (Figure 3). In both episodes, there was a considerable impact on non-oil GDP, although oil production and investment were broadly unaffected. The impact on non-oil GDP was particularly sharp as a result of the 1998 shock, with the CEMAC non-oil GDP growth turning negative in 1999, reflecting a sharp non-oil GDP contraction in the Congo and Gabon.
CEMAC: Oil Prices and Real GDP, 1990–2014
(Percent and indices)
Citation: IMF Staff Country Reports 2015, 308; 10.5089/9781513510583.002.A004
Sources: World Economic Outlook (WEO) database and IMF Staff calculations.CEMAC: Oil Prices and Non-Oil Real GDP, 1990–2014
(Percent and indices)
Citation: IMF Staff Country Reports 2015, 308; 10.5089/9781513510583.002.A004
Sources: World Economic Outlook (WEO) database; and IMF Staff calculations.CEMAC: GDP Growth During Oil-Price Shock Episodes
Citation: IMF Staff Country Reports 2015, 308; 10.5089/9781513510583.002.A004
Sources: World Economic Outlook (WEO) database and IMF Staff calculations.Note: “t=0” refers to 2008.5. The economic impact of the oil price shocks has been broadly similar across CEMAC countries, although with some degree of heterogeneity. Following the 1998 shock, Congo and Gabon experienced a drop in overall GDP and an even larger drop in non-oil GDP. Cameroon, with the more diversified economy of the three, experienced only a minor deceleration in both overall and non-oil GDP growth. In 2008, only the economies of Cameroon and Congo were largely unaffected by the oil price shock. In Congo, this was because oil production increased in the year following the shock, offsetting the impact of the fall in oil prices. In Cameroon, which has to a large extent diversified away from oil and is less of an oil exporter than it used to be, the effect of the oil-price shock on government revenue and spending (e.g., fuel subsidies) largely canceled out and the non-oil activity was not significantly affected by the shock.
6. There is some evidence that changes in government expenditure has been an important transmission channel of oil-price volatility to the non-oil economy in recent years. With oil being a major source of government revenues and government being a major player in non-oil economic activity, it may be expected that changes in oil prices would lead to changes in the level of government revenues and, to the extent that these changes affect also government expenditure, they would also affect non-oil economic activity. Indeed, changes in oil prices have been positively correlated with CEMAC-wide real government expenditure; in turn, the latter has been correlated with non-oil real GDP growth although this correlation has been somewhat weaker, yet still significantly positive (Figure 4). This suggests that government spending may have been an important channel of transmission of oil-price changes to non-oil economic activity in CEMAC.
CEMAC: Government Expenditure, Oil Price, and Non-oil GDP Growth, 2000–14
(Percent)
Citation: IMF Staff Country Reports 2015, 308; 10.5089/9781513510583.002.A004
Sources: World Economic Outlook (WEO) database; and IMF Staff calculations.7. However, the government expenditure channel did not operate similarly during the two oil price shocks. During the 1998 episode, there was a sharp contraction in government expenditure in the countries that experienced a parallel decline in non-oil economic activity (Congo and Gabon), whereas, during the 2008 episode, some countries (Chad and Equatorial Guinea) did not reduce their spending levels (Figure 5). Chad and Equatorial Guinea had seen an unprecedented increase in their oil revenues immediately prior to the oil shock and kept increasing public spending by drawing on buffers accumulated in the previous years (Box 2). Yet, in both countries non-oil GDP growth declined markedly. A possible further factor leading to the differences in government expenditure contraction in 1998 and 2008 may have been the level of debt. While in 1998 most countries were highly indebted, with limited buffers and no access to additional borrowing, in 2008 most countries had already benefited from debt relief, which allowed some of them to implement countercyclical policies.
CEMAC: General Government Expenditure Growth in Real Terms, 1998–2008
(Percent)
Citation: IMF Staff Country Reports 2015, 308; 10.5089/9781513510583.002.A004
Sources: World Economic Outlook (WEO) database; and IMF Staff calculations.CEMAC: Evolution of Fiscal Buffers
Most CEMAC countries managed to build sizeable fiscal buffers starting in the mid-2000s, but had a tendency to deplete them rapidly during crisis episodes. During the 1998 crisis, all three oil producers (and the other non-oil producers) drew down their already low fiscal reserves, but these were not sufficient to support government expenditure beyond the very short term. All the countries in the region entered the millennium with usable government deposits amounting to less than two months of domestically financed expenditure. The commodity price boom in the early to mid-2000s allowed them to increase their buffers to at least three months of domestically financed expenditure by 2007. However, the 2008 crisis led most of them (Congo being the notable exception) to draw down their deposits again and left them with buffers largely depleted. Declining buffers since the 2008 oil-price shock have been caused in part by self-financed investment programs in most CEMAC countries. The resulting low buffers have amplified CEMAC countries’ vulnerability to the recent oil-price shock.
CEMAC: Government Deposits at the BEAC, 1995–2014
(In months of domestically-financed spending)
Citation: IMF Staff Country Reports 2015, 308; 10.5089/9781513510583.002.A004
Sources: World Economic Outlook (WEO) database; and IMF Staff calculations.8. The credit channel appears to have a played a very limited role in transmitting oil-price signals. Credit to the private sector appears to have been synchronized with oil price fluctuations, but it did not translate into higher private investment needed to spur non-oil growth (Figure 6). Although government investment in the CEMAC was highly correlated with oil price fluctuations, the extent of crowding out in downturns is likely to have been limited, because of the limited reliance of enterprises on bank financing.4 The increase in credit in boom periods went mainly to firms with activities linked to public investment projects, given that private investment remained subdued in 1994–2014. This implies that non-oil sector growth was mainly driven by activities linked to public investments projects, and not to autonomous private sector investments.
CEMAC: Credit to the Private Sector, and Public and Private Investment Growth, 1991–2014
(Percent and indices)
Citation: IMF Staff Country Reports 2015, 308; 10.5089/9781513510583.002.A004
Sources: CEMAC authorities and IMF staff estimates.9. The lack of an important government transmission channel in the 2008 episode suggests that other links between the oil and non-oil sector may be at play in CEMAC countries. This contradicts the common perception that the oil sector operates in an enclave. This issue will be revisited in the following sections.
C. Panel Regressions on the Economic Impact of Oil Price Changes
10. Panel regressions provide more refined evidence of the impact of oil price changes on non-oil GDP growth in CEMAC. Two sets of dynamic panel regressions were run to assess the impact of oil price fluctuations on economic activity and government expenditure. These results have been subsequently compared to the results of regressions performed on other oil-exporting countries across the world. The regressions have the following general specification:
11. An oil shock (OS) is the independent variable of primary interest. A shock is defined in line with IMF (2012), as the percentage change in international oil price multiplied by the share of oil exports in GDP, where the latter is the moving average of its values in the preceding three years. The specification allows for including up to two lags of this independent variable. The regressions were run on three sets of dependent variables (Y): real GDP growth, real non-oil GDP growth, and real government spending growth. The specification used also controls for the changes in the world demand (WD) that could affect domestic economic activity.
12. Data, drawn from the International Finance Statistics database, cover the period 1970–2014. The observation-years for the individual countries, which were not producing oil at the time of observation, were omitted. Regressions were run using standard dynamic panel techniques and specifications using random and fixed effects. The results proved robust to changes in the regression techniques. For brevity’s sake, only the results from the Arellano-Bond dynamic panel-data estimation, which encompasses both fixed and random effects, are documented below.
13. The results confirm a significant impact of oil-price fluctuations on all three dependent variables. The strength and the significance of the impact vary across the three variables and are generally consistent with the intuitions derived from casual data observation. The impact of the oil-price shock on real non-oil GDP growth is particularly strong and more robust than on the other two dependent variables.
The impact of an oil-price shock on real GDP growth is statistically significant contemporaneously and up to a one-year lag (Table 1). Based on the estimated coefficients, a “composite” CEMAC country with oil exports equivalent to 50 percent of its GDP (the average among the CEMAC oil-exporting countries), facing a 27 percent decline in the oil price (approximately the expected decline in the average international oil price in CFA terms between 2014 and 2015) would see its real GDP growth reduced by approximately one percentage point in the year of the shock and by two percentage points in the first year after the shock, after which the impact recedes and loses statistical significance (Figure 7).
The estimated impact of an oil shock on non-oil growth is twice as large, significant throughout three consecutive years, and robust across several regression techniques (Table 2). The estimated coefficients imply that a country with oil exports equivalent to 50 percent of GDP hit by a 27 percent decline in the oil price sees a reduction in non-oil GDP growth of 2.5 percentage points in the year of the shock and 4 percentage points in the following year. It fades to about 2 percentage points in the subsequent year, but remains statistically significant. An approximate magnitude of the shock at its peak (t+1), scaled by the oil-exports-to-GDP ratio, ranges from less than one percentage point in the case of Cameroon to over seven percentage points in the case of Equatorial Guinea (Figure 7). The impact of an oil price shock on non-oil GDP growth being stronger and more significant than the impact on overall GDP growth is consistent with the earlier observations suggesting that oil production levels in CEMAC countries have not been significantly affected by oil price changes.
Finally, the results confirm the existence of a significant impact of oil prices on real government expenditure. Nonetheless, in this case the estimated long-run elasticity is lower than the long-run elasticity of non-oil GDP growth with respect to oil prices and is not robust to a modification of the estimation technique. In a hypothetical (composite) CEMAC country with oil exports equivalent to 50 percent of GDP, real government expenditure growth was reduced by 3.6 percentage point of GDP in the year of the shock, and by a further 3.1 percentage point in the first year after the shock, while an increase of 3.2 percentage point could be expected in the second year after the shock. This impact is lower than the impact on non-oil growth and not statistically significant when using other panel regression methods (including System Generalized Method of Moments). The weaker link between the oil price and government expenditure may reflect the fact that real government expenditure remained relatively stable in the context of the 2008 oil price shock, as described in Section B.
CEMAC: Oil Shock Effect on Selected Aggregates1
Arellano-Bond dynamic panel-data estimation.
CEMAC: Oil Shock Effect on Selected Aggregates1
| Real GDP growth | Real non-oil GDP growth | Real government spending growth | |
|---|---|---|---|
| 1-year lagged dependent variable | 0.3812*** (0.0485) |
0.2552*** (0.06) |
0.0823*** (0.01) |
| World real GDP growth | 1.0022*** (0.38) |
0.9698*** (0.30) |
2.5982*** (0.63) |
| Oil shock | 0.0398** (0.02) |
0.1865* (0.11) |
0.2668*** (0.08) |
| 1-year lagged oil shock | 0.1317*** (0.05) |
0.2567*** (0.06) |
0.2320*** (0.07) |
| 2-year lagged oil shock | 0.0336 (0.04) |
0.1082* (0.06) |
−0.2422*** (0.05) |
| Constant | −0.8592 (0.86) |
−1.183 (1.10) |
−1.1555 (2.77) |
| Number of observations | 149 | 122 | 94 |
| LR-Elasticity of Oil Shock | 0.15 | 0.44 | 0.24 |
Arellano-Bond dynamic panel-data estimation.
CEMAC: Oil Shock Effect on Selected Aggregates1
| Real GDP growth | Real non-oil GDP growth | Real government spending growth | |
|---|---|---|---|
| 1-year lagged dependent variable | 0.3812*** (0.0485) |
0.2552*** (0.06) |
0.0823*** (0.01) |
| World real GDP growth | 1.0022*** (0.38) |
0.9698*** (0.30) |
2.5982*** (0.63) |
| Oil shock | 0.0398** (0.02) |
0.1865* (0.11) |
0.2668*** (0.08) |
| 1-year lagged oil shock | 0.1317*** (0.05) |
0.2567*** (0.06) |
0.2320*** (0.07) |
| 2-year lagged oil shock | 0.0336 (0.04) |
0.1082* (0.06) |
−0.2422*** (0.05) |
| Constant | −0.8592 (0.86) |
−1.183 (1.10) |
−1.1555 (2.77) |
| Number of observations | 149 | 122 | 94 |
| LR-Elasticity of Oil Shock | 0.15 | 0.44 | 0.24 |
Arellano-Bond dynamic panel-data estimation.
CEMAC: Real GDP and Non-Oil GDP Growth Response to Oil Price Shock1
(Percent)
Citation: IMF Staff Country Reports 2015, 308; 10.5089/9781513510583.002.A004
Sources: CEMAC authorities; and IMF staff estimates.1 The shock is a 27 percent decline in oil price (the projected magnitude of the 2014 to 2015 decline in domestic currency), assuming that exports-to-GDP ratio is 50 percent (CEMAC average).Oil Shock Effect on Non-Oil Real GDP Growth
Arellano-Bond dynamic panel-data.
System Generalized Method of Moments.
Oil Shock Effect on Non-Oil Real GDP Growth
| Random Effects | Fixed Effects | Dynamic Panel Data 1/ | System-GMM 2/ | |
|---|---|---|---|---|
| 1-year lagged dependent variable0.3430 | 0.3430*** (0.0800) |
0.2552*** (0.08) |
0.2552*** (0.06) |
0.2635*** (0.07) |
| World real GDP growth | 0.8441 (0.66) |
0.9698 (0.65) |
0.9698*** (0.30) |
0.8596 (0.61) |
| Oil shock | 0.2005*** (0.06) |
0.1865*** (0.06) |
0.1865* (0.11) |
0.2489*** (0.05) |
| 1-year lagged oil shock | 0.2483*** (0.06) |
0.2567*** (0.06) |
0.2567*** (0.06) |
0.2649*** (0.05) |
| 2-year lagged oil shock | 0.10 (0.06) |
0.1082* (0.06) |
0.1082* (0.06) |
0.1201** (0.05) |
| Constant | −1.2458 (2.39) |
−1.183 (2.36) |
−1.183 (1.10) |
−1.2017 (2.25) |
| Number of observations | 127 | 127 | 122 | 127 |
| LR-Elasticity of Oil Shock | 0.79 | 0.70 | 0.70 | 0.78 |
Arellano-Bond dynamic panel-data.
System Generalized Method of Moments.
Oil Shock Effect on Non-Oil Real GDP Growth
| Random Effects | Fixed Effects | Dynamic Panel Data 1/ | System-GMM 2/ | |
|---|---|---|---|---|
| 1-year lagged dependent variable0.3430 | 0.3430*** (0.0800) |
0.2552*** (0.08) |
0.2552*** (0.06) |
0.2635*** (0.07) |
| World real GDP growth | 0.8441 (0.66) |
0.9698 (0.65) |
0.9698*** (0.30) |
0.8596 (0.61) |
| Oil shock | 0.2005*** (0.06) |
0.1865*** (0.06) |
0.1865* (0.11) |
0.2489*** (0.05) |
| 1-year lagged oil shock | 0.2483*** (0.06) |
0.2567*** (0.06) |
0.2567*** (0.06) |
0.2649*** (0.05) |
| 2-year lagged oil shock | 0.10 (0.06) |
0.1082* (0.06) |
0.1082* (0.06) |
0.1201** (0.05) |
| Constant | −1.2458 (2.39) |
−1.183 (2.36) |
−1.183 (1.10) |
−1.2017 (2.25) |
| Number of observations | 127 | 127 | 122 | 127 |
| LR-Elasticity of Oil Shock | 0.79 | 0.70 | 0.70 | 0.78 |
Arellano-Bond dynamic panel-data.
System Generalized Method of Moments.
14. A broad cross-country comparison shows that non-oil economic activity in CEMAC seems to be more sensitive to oil-price changes than in other oil-exporting countries.5 This result holds even though the specification controls for the relative size of the oil sector (Table 4). The coefficients on real GDP growth and non-oil real GDP growth have the same sign but are lower and statistically less significant than in the CEMAC; the impact on non-oil GDP growth is insignificant, and the impact on the overall GDP growth becomes significant after two years, which probably reflects scaling up or down of oil-related investments in response to positive or negative oil price signals. Interestingly though, the impact of oil price fluctuations on government expenditure in non-CEMAC oil exporters is more significant, especially in high-income (mostly Middle-Eastern) countries.6
Non-CEMAC Oil-Exporting Countries1
Economies are categorized as oil-exporting countries when their oil-export earnings exceeded 50 percent of total exports on average between 2009 and 2014. Oil is defined by the Standard International Trade Classification (SITC) code 3, which includes mineral oils, lubricants, and related materials
Non-CEMAC Oil-Exporting Countries1
| Lower Middle Income | Upper Middle Income | High Income Non-OECD |
|---|---|---|
| Nigeria South Sudan Yemen |
Algeria Angola Azerbaijan Iran Iraq Kazakhstan Libya Turkmenistan Venezuela |
Bahrain Brunei Darussalam Kuwait Oman Qatar Saudi Arabia Trinidad and Tobago United Arab Emirates |
Economies are categorized as oil-exporting countries when their oil-export earnings exceeded 50 percent of total exports on average between 2009 and 2014. Oil is defined by the Standard International Trade Classification (SITC) code 3, which includes mineral oils, lubricants, and related materials
Non-CEMAC Oil-Exporting Countries1
| Lower Middle Income | Upper Middle Income | High Income Non-OECD |
|---|---|---|
| Nigeria South Sudan Yemen |
Algeria Angola Azerbaijan Iran Iraq Kazakhstan Libya Turkmenistan Venezuela |
Bahrain Brunei Darussalam Kuwait Oman Qatar Saudi Arabia Trinidad and Tobago United Arab Emirates |
Economies are categorized as oil-exporting countries when their oil-export earnings exceeded 50 percent of total exports on average between 2009 and 2014. Oil is defined by the Standard International Trade Classification (SITC) code 3, which includes mineral oils, lubricants, and related materials
Non-CEMAC: Oil Price Effect on Selected Variables1
Arellano-Bond dynamic panel-data estimation.
Non-CEMAC: Oil Price Effect on Selected Variables1
| Real GDP growth Real | non-oil GDP growth | Real government spending growth | |
|---|---|---|---|
| 1-year lagged dependent variable | −0.06530 (0.1585) |
−0.0703*** (0.00) |
−0.0305 (0.10) |
| World real GDP growth | 1.1047*** (0.21) |
−1.6584 (2.42) |
0.819 (0.94) |
| Oil shock | 0.0175 (0.03) |
−0.0091 (0.08) |
0.4042*** (0.12) |
| 1-year lagged oil shock | 0.05 (0.08) |
(0.08) (0.18) |
0.3704** (0.15) |
| 2-year lagged oil shock | 0.1007** (0.05) |
0.1022 (0.13) |
0.3313* (0.20) |
| Constant | −0.3827 (0.97) |
15.6734 (14.30) |
−0.2912 (3.63) |
| Number of observations | 548 | 303 | 330 |
| LR-Elasticity of Oil Shock | 0.07 | −0.09 | 0.80 |
Arellano-Bond dynamic panel-data estimation.
Non-CEMAC: Oil Price Effect on Selected Variables1
| Real GDP growth Real | non-oil GDP growth | Real government spending growth | |
|---|---|---|---|
| 1-year lagged dependent variable | −0.06530 (0.1585) |
−0.0703*** (0.00) |
−0.0305 (0.10) |
| World real GDP growth | 1.1047*** (0.21) |
−1.6584 (2.42) |
0.819 (0.94) |
| Oil shock | 0.0175 (0.03) |
−0.0091 (0.08) |
0.4042*** (0.12) |
| 1-year lagged oil shock | 0.05 (0.08) |
(0.08) (0.18) |
0.3704** (0.15) |
| 2-year lagged oil shock | 0.1007** (0.05) |
0.1022 (0.13) |
0.3313* (0.20) |
| Constant | −0.3827 (0.97) |
15.6734 (14.30) |
−0.2912 (3.63) |
| Number of observations | 548 | 303 | 330 |
| LR-Elasticity of Oil Shock | 0.07 | −0.09 | 0.80 |
Arellano-Bond dynamic panel-data estimation.
15. Although further analysis confirms the existence of the link between government expenditure and non-oil GDP, its economic significance appears to be weak. An oil-price shock is normally associated with the reduction in government expenditure in CEMAC countries, because of lower oil revenues. This effect is especially strong in the absence of sufficient fiscal buffers. The analysis shows that the change in government expenditure appears to have a statistically significant impact on non-oil GDP growth, for up to one year, but the related coefficient is small and therefore of low economic significance (Table 5). These results are corroborated by Granger causality tests (Table 6). Nevertheless, no strong inferences can be made out of these results bearing in mind the potential small sample bias and the bias that could result from the omission of potentially significant variables, given the use of very simplified regression specifications. It should be noted, however, that the evidence of a direct link between oil and non-oil activities unrelated to the government channel has already been established in other SSA countries, for example in Nigeria.7
CEMAC: Effect of Real Government Spending Growth on Real Non-Oil GDP Growth1
Arellano-Bond dynamic panel-data estimation.
CEMAC: Effect of Real Government Spending Growth on Real Non-Oil GDP Growth1
| Real non-oil GDP growth | |
|---|---|
| 1-year lagged dependent variable | 0.3537*** (0.0665) |
| Real government spending growth | 0.0484** (0.02) |
| 1-year lagged | 0.0773*** (0.02) |
| 2-year lagged | 0.0033 (0.01) |
| Constant | 3.4050*** (0.79) |
| Number of observations | 93 |
| LR-Elasticity of Oil Shock | 0.07 |
Arellano-Bond dynamic panel-data estimation.
CEMAC: Effect of Real Government Spending Growth on Real Non-Oil GDP Growth1
| Real non-oil GDP growth | |
|---|---|
| 1-year lagged dependent variable | 0.3537*** (0.0665) |
| Real government spending growth | 0.0484** (0.02) |
| 1-year lagged | 0.0773*** (0.02) |
| 2-year lagged | 0.0033 (0.01) |
| Constant | 3.4050*** (0.79) |
| Number of observations | 93 |
| LR-Elasticity of Oil Shock | 0.07 |
Arellano-Bond dynamic panel-data estimation.
CEMAC: Granger Causality Test: Government Spending and Non-Oil Growth
CEMAC: Granger Causality Test: Government Spending and Non-Oil Growth
| Panel A: Choice of the optimal lag for Granger causality test | ||||
| Information Criterion | No lag | One lag | Two lags | Optimal lag |
| AIC | 12.2 | 11.6 | 12.2 | 1 |
| HQIC | 12.1 | 11.4 | 11.9 | 1 |
| SBIC | 12.3 | 11.8 | 12.5 | 1 |
| Note: AIC stands for Akaike information criterion, HQIC for Hannan-Quinn information criterion, SBIC for Schartz Bayesian information criterion. | ||||
| Panel B: Causality test using the optimal lag | ||||
| Dependent variable | Expenditure growth | Non-oil GDP growth | ||
| Non-oil GDP Growth, lagged |
1.429*
(0.719) |
0.729*** (0.166) |
||
| Expenditure growth, lagged | 0.00817*** (0.00213) |
0.00169**
(0.0157) |
||
| Constant | 3.496 (4.253) |
0.927 (1.232) |
||
| Observations | 18 | 18 | ||
| R-squared | 0.140 | 0.404 | ||
| Dependent lags | 1 | 1 | ||
| Independent lags | 1 | 1 | ||
| F-test | F(1, 61) = 3.95 Prob > F = 0.0655 | F(1, 61) = 8.27 prob > F = 0.0116 | ||
| Causal inference | Causality | Causality | ||
CEMAC: Granger Causality Test: Government Spending and Non-Oil Growth
| Panel A: Choice of the optimal lag for Granger causality test | ||||
| Information Criterion | No lag | One lag | Two lags | Optimal lag |
| AIC | 12.2 | 11.6 | 12.2 | 1 |
| HQIC | 12.1 | 11.4 | 11.9 | 1 |
| SBIC | 12.3 | 11.8 | 12.5 | 1 |
| Note: AIC stands for Akaike information criterion, HQIC for Hannan-Quinn information criterion, SBIC for Schartz Bayesian information criterion. | ||||
| Panel B: Causality test using the optimal lag | ||||
| Dependent variable | Expenditure growth | Non-oil GDP growth | ||
| Non-oil GDP Growth, lagged |
1.429*
(0.719) |
0.729*** (0.166) |
||
| Expenditure growth, lagged | 0.00817*** (0.00213) |
0.00169**
(0.0157) |
||
| Constant | 3.496 (4.253) |
0.927 (1.232) |
||
| Observations | 18 | 18 | ||
| R-squared | 0.140 | 0.404 | ||
| Dependent lags | 1 | 1 | ||
| Independent lags | 1 | 1 | ||
| F-test | F(1, 61) = 3.95 Prob > F = 0.0655 | F(1, 61) = 8.27 prob > F = 0.0116 | ||
| Causal inference | Causality | Causality | ||
D. Conclusion
16. This paper establishes a strong link between oil price movements and non-oil GDP growth in CEMAC countries. Impulse response estimates, based on regression analyses of the link between oil price changes and non-oil activity in CEMAC countries should therefore be taken into account in policy formulation in the wake of an oil price shock, like the one currently affecting the region.
17. The strong and significant impact of oil price changes on CEMAC non-oil activity and the relatively weak results of the regressions related to the government transmission channel corroborate findings based on the stylized facts. The high sensitivity of non-oil GDP growth to changes in oil price, even when government expenditure is not much affected (as was the case in the context of the 2008 oil-price shock), is of particular interest. It implies that the direct channel through which oil-price fluctuations are transmitted to the non-oil economy is significant, contradicting the common perception of the oil sector operating in an economic enclave. The links that bypass the government expenditure channel could include the behavior of private contractors of oil companies, the consumption by oil sector employees, the liquidity supply to the banking sector as well as other, informal, liquidity channels. This is an area for further exploration.
References
International Monetary Fund, April 2012, “Commodity price swings and commodity exporters,” World Economic Outlook, Chapter 4.
International Monetary Fund, April 2015, “Uneven Growth: Short- and Long-Term Factors,” World Economic Outlook, Statistical Appendix.
International Monetary Fund, April 2015, “Navigating headwinds,” Regional Economic Outlook—Sub-Saharan Africa, Chapter 1.
Spatafora, Nicola and Irina Tytell, 2009, “Commodity terms of trade: The history of booms and busts,” IMF Working Paper No. WP/19/105.
Prepared by Samson Kwalingana, Azanaw Mengistu, Nérée Noumon, Gonzalo Salinas, and Jaroslaw Wieczorek.
For the purpose of this paper, unless otherwise indicated, CEMAC refers to its five oil-producing members, namely Cameroon, Chad, Congo (Republic of) Gabon, and Equatorial Guinea. The sixth CEMAC member, the Central African Republic, is not an oil producer.
Following the convention applied in the IMF’s African Regional Economic Outlook of April 2015, the term “oil-price shock” refers to oil price slump of more than 50 percent. Other noteworthy oil price drops, but below the 50 percent threshold, occurred from November 2000 to January 2002 (42 percent) and from July 2006 to may 2007 (18 percent).
Limited access to finance is well known in CEMAC and is due to many structural factors (see companion CEMAC country and Financial Sector Assessment reports).
There may be differences in public sector coverage between CEMAC and non-CEMAC countries, as some general government expenditure may be carried out by some large public companies in some CEMAC countries without this being recorded in the budget. We considered these potential differences to be relatively small.
The March 2015 Selected Issues Paper on Nigeria (15/85) establishes a strong link between oil price movements and non-oil GDP growth not only through government expenditure (a transmission channel that appears to have weakened in recent years), but also through private disposable income.
