^{1}

^{2}

^{1}

^{1}

This study investigated the impact of crude oil price and exchange rate on economic growth in Nigeria using an autoregressive distributed lag model covering the period from 1982-2018. The results indicated that crude oil price and exchange rate have significant positive impact on economic growth in both the long-run and the short-run periods. The findings suggested that crude oil price and exchange rate which are the focal points of the study, could affect economic growth in both the long-run and the short-run. Therefore government should diversify its earnings in agriculture, industrialization and investment in order to reduce the heavy reliance on crude oil and income fluctuation resulting from the fluctuation in crude oil prices.

One of the most important driving forces of the global economy is the crude oil and changes in the price of this oil will have significant effects on economic growth and the well being of the population around the world [

The OPEC Reference Basket rebounded in September from low levels registered last month, mainly supported by supply disruptions and heightened geopolitical tensions in the Middle East, which helped to push all crude oil benchmarks higher. The ORB value rose by $2.74, or 4.6% in September to settle at $62.36/b. All basket component values increased following the attacks on Saudi Arabia’s oil processing facilities at Abqaiq and Khurais, which caused a supply disruption of about 5.7 mb/d, raising concerns on the tightening physical crude market. However, oil prices came down and the risk premium faded after Saudi Arabia restored production and fulfilled all scheduled shipments to customers [

Crude oil is a major source of foreign exchange earnings and the dominant source of revenue for the Nigerian economy. The Nigerian economy has been completely reliant on oil and the basis upon which government budgeting, revenue distribution and capital allocations are determined. Volatility in oil price is an upward and downward movement of oil prices globally. This assertion thus translates that these oil prices are exogenous because it’s determined by external influences that somewhat stagnate the Naira and Nigeria cannot moderate the causes of these oil price slides. Nigeria’s exports of oil at a time of peak prices have enabled the country to post merchandise trade and current account surpluses in recent years. Reportedly, 80% of Nigeria’s energy revenues flow to the government; 16% cover operational costs, and the remaining 4% go to investors Atukeren, (2003) as cited in [

According to [

The recent shock in crude oil prices has adversely affected Nigeria, especially in the areas of foreign exchange earnings, foreign reserves, decline in government revenue and threat in terms of ability to meet financial obligations as at when due. The average crude oil price further dropped in a row by USD 8.26 or 13.10% month to month to USD 54.77/b in December 2018; the lowest since October 2017. This decrease is due to concerns over unforeseen rise in global oil supply with decreased demand amidst ambiguity about worldwide economic growth [

The objective of this study was to look at the impact of crude oil price and exchange rate on the economic growth of Nigeria. Many studies conducted in Nigeria only looked at the impact of crude oil price alone on economic growth or impact of foreign exchange rate alone on economic growth. Therefore, this is the gap that the study intended to fill in the literature. In accordance with the literature, crude oil price is supposed to have a direct relationship with economic growth because an increase in the price of crude oil means an increase in the level of economic growth. Whereas foreign exchange rate can have direct and also indirect relationship with the level of economic growth, but only empirical tests could validate this. Accordingly, this research work sought to empirically test the effects of crude oil price and exchange rate on economic growth of Nigeria by employing an autoregressive distributed lag (ARDL) model advanced by Pesaran, Shin and Smith (2001).

This paper is organized in sections. The first section contains introduction, research problem, significance of the study, objective of the study and the organization of the paper. The second section consists of review of related literature, which provides some snapshots of similar empirical literature. The third section presents methodology and data. The fourth section presents results and discussion of the paper. Finally, the fifth section discusses the conclusions and policy recommendations.

Many research studies have been conducted in areas related to this. However, a major part of the researches conducted were on crude oil price and economic growth nexus or on exchange rate economic growth nexus. Review of literature shows that there is no many existing studies that specifically examine the impact of crude oil price and exchange rate on economic growth, especially in Nigeria. Though there is a bunch of literature that investigated the relationship between crude oil price and economic growth nexus and exchange rate and economic growth nexus. The literature here will give an overview of the existing literature on the related topic from general perspective and later narrow it down to focus on Nigeria.

[

[

[

Using generalized autoregressive conditional heteroscedasticity (GARCH), component generalized autoregressive conditional heteroscedasticity (CGARCH) and exponential generalized autoregressive conditional heteroscedasticity (EGARCH) to examines the macroeconomic effects of exogenous oil price shock in Nigeria, [

Using the simple ordinary least square (OLS) method and Granger causality test, [

[

[

[

[

[

[

[

The Nigeria’s annual data employed in this study spans from 1982 to 2018. The data were sourced from [

To derive the model, it is known that based on the Quantity theory of money equation (particularly income approach equation) where demand for money is equated to the supply of money. Therefore, the equation can be denoted as follow: M V = P Y

Y = M V P = f ( M V P − 1 )

where: Y is the economic growth, M will be dropped, V will be replaced with EXC and P − 1 will be replaced with COP. Therefore, the model can be written as: G D P C = f ( E X C , C O P ) .

Following the above model of the study, the econometric form of the model can be written in a simple log-linear form and augmented it with the crude oil price and exchange rate variables as follows:

ln G D P C t + β 0 + β 1 ln E X C t + β 2 ln C O P t + η t (1)

where: ln G D P P C t stand in for the natural log of GDP per capita, ln E X C t is the natural log of Exchange rate, ln C O P t is the natural log of Crude oil price and η t is a disturbance term.

The co-integration approach of the ARDL model has been employed to test for co-integration relationships between the variables of interest. Despite that there are other methods for achieving the same purpose, this approach has several advantages that include; its applicability regardless of the order of the variables in the model (i.e., whether they are all I (0), I (1) or mixture of the two); with the ARDL, both the short-run and long-run coefficients can be simultaneously obtained; it is also a good model for small sample (i.e. 30 to 80 observations); it has an indirect co-integration test within the model; and lastly it has the diagnostic tests within the model (e.g. using Microfit statistical software).

On the basis of these advantages, this study chose this approach and formulated the conditional error correction model as;

Δ ln G D P C t = δ 0 + ∑ i = 1 k χ i Δ ln G D P C t − i + ∑ i = o k λ i Δ ln E X C t − i + ∑ i = 0 k ϑ i Δ ln C O P t − i ψ 1 ln G D P C t − i + ψ 2 ln E X C t − 1 + ψ 3 ln C O P t − 1 + μ t (2)

Equation (2) is estimated using the OLS method to test for co-integration relationship among crude oil price and exchange rate on economic growth by conducting a Wald test/F-test to ascertain the joint significance of the lagged coefficients of the variables. To accomplish this task, the null hypothesis if no co-integration in Equation (2) is defined as

H_{0}: ψ 1 = ψ 2 = ψ 3 = 0 as against the alternative hypothesis, which states that co-integration exists (Ha: ψ 1 ≠ ψ 2 ≠ ψ 3 ≠ 0 ). To decide on the result, [

ln G D P C t = β 0 + ∑ i = 1 m χ 1 i ln G D P C t − i + ∑ i = 0 m λ 1 i ln E X C t − i + ∑ i = 0 m ϑ 1 i ln C O P t − i + ε 1 t (3)

Δ ln G D P C t = β 1 + ∑ i = 1 m χ 2 i Δ ln G D P C t − i + ∑ i = 0 m λ 2 i Δ ln E X C t − i + ∑ i = 0 m ϑ 2 i Δ ln C O P t − i + Φ E C T − 1 + ε 2 t (4)

where the coefficient of the error correction term (ECT) is denoted by Φ that shows the speed of adjustment of the variables toward long-run convergence.

Lastly, this study diagnosed the model by conducting tests for serial correlation (using Breusch-Pagan LM test), heteroscedasticity (using ARCH test for heteroscedasticity), normality (using Jarque-bera test), functional form (using Ramsey RESET test) and stability test using CUSUM and CUSUMSQ to be able to assess how stable the model is along the sampled periods.

Robustness Check Using Dynamic Ordinary Least Squares (DOLS) and Fully Modified Ordinary Least Squares (FMOLS)

To gauge the long-run estimate, we apply dynamic ordinary least square (DOLS) and fully modified ordinary least square (FMOLS). DOLS and FMOLS have the power to deal with endogeneity problem, simultaneity bias and small sample bias. These estimators are good for robustness check of ARDL estimates. DOLS and FMOLS have been advanced by Stock and Watson (1993) and Philip and Moon (1999), respectively to address the problem of serial correlation and small sample bias attributed to Ordinary Least Squares (OLS) estimator. The estimators can also be applied to mix order of integrated variables in co-integration framework. Considering the strengths of these estimators, their results will serve as robustness checks to ARDL long-run test results.

To begin the estimation, the time series properties of the data were first tested using augmented [

However, sometimes, ADF and PP tests may not produce reliable estimates if there is a presence of structural break in the series and as such they could produce a biased result. To avoid such doubt, we have equally employed Zivot-Andrews (1992) structural break trended unit root test.

Before testing the co-integration relationship among the variables using Equation (2), it was important to identify the optimum lag length to be used. From the results of the unrestricted vector auto regressive optimum lag selection criteria in

Having identified the optimum lag length, the next step was to estimate the long-run relationship among the variables by using ordinary least square (OLS). The null hypothesis of no co-integration (H_{0}: ψ 1 = ψ 2 = ψ 3 = 0 ) was tested against the alternative hypothesis of the existence of a co-integration relationship (Ha: ψ 1 ≠ ψ 2 ≠ ψ 3 ≠ 0 ). The result of this test presented in

The Johansen Juselius test for co-integration using model with Trace statistics and model with Max-Eigen value confirmed the existence of one co-integration equation in the trace statistics model. Therefore, we conclude that there is a long-run relationship among the dependent and independent variables and that the variables moved together in the long-run. This test result supported the result of ARDL bound test for co-integration (

Level | First Difference | ||||||||
---|---|---|---|---|---|---|---|---|---|

ADF | PP | ADF | PP | ||||||

Variables | Constant | Constant & trend | Constant | Constant & trend | Constant | Constant & trend | Constant | Constant & trend | I (d) |

ln G D P C t | −0.44 (0.889) | −1.54 (0.793) | −0.08 (0.943) | −2.76 (0.218) | −5.20*** (0.000) | −4.92*** (0.001) | −5.17*** (0.000) | −4.92*** (0.001) | I (1) |

ln C O P t | −1.00 (0.741) | −2.36 (0.391) | −1.00 (0.741) | −2.36 (0.391) | −5.58*** (0.000) | −5.54*** (0.000) | −5.57*** (0.000) | −5.53*** (0.000) | I (1) |

ln E X C t | −2.93* (0.051) | -1.81 (0.675) | -2.91* (0.053) | −1.23 (0.887) | −3.50** (0.013) | −4.31*** (0.008) | −3.53** (0.012) | −4.30*** (0.008) | I (0) |

Source: Eviews 9; Note: ***, ** & * stand for 1%, 5% & 10% levels of significance and values in parenthesis are the P-values, while I (d) stands for the interpretation of the results.

Level | First difference | ||||||||
---|---|---|---|---|---|---|---|---|---|

Variables | Constant | Break Point | Constant & trend | Break point | Constant | Break point | Constant & trend | Break point | I (d) |

ln G D P C t | −2.988 (2) | 2001 | −3.029 (2) | 1994 | −4.839 (1)* | 2000 | −3.805 (1) | 2010 | I (1) |

ln C O P t | −3.670 (0) | 2003 | −2.602 (0) | 2011 | −6.257 (0)*** | 2008 | −6.038 (0)*** | 2005 | I (1) |

ln E X C t | −1.949 (0) | 1991 | −5.562 (0)*** | 1995 | −4.266 (4) | 1996 | −3.042 (4) | 2007 | I (0) |

Source: Eviews 9; Note: ***, ** & * stand for 1%, 5% & 10% levels of significance and values in brackets are the lag lengths, while I (d) stands for the interpretation of the results.

Lag | LogL | LR | FPE | AIC | SC | HQ |
---|---|---|---|---|---|---|

0 | −53.05599 | NA | 0.005999 | 3.397333 | 3.533379 | 3.443108 |

1 | 87.67932 | 247.3530^{b} | 2.05e−06^{b} | −4.586625^{b} | −4.042441^{b} | −4.403524^{b} |

2 | 95.55101 | 12.40387 | 2.23e−06 | −4.518243 | −3.565920 | −4.197815 |

3 | 102.9613 | 10.32948 | 2.56e−06 | −4.421896 | −3.061435 | −3.964142 |

4 | 108.5848 | 6.816327 | 3.39e−06 | −4.217258 | −2.448658 | −3.622178 |

Source: Authors’ computation 2019 using Eviews 9. Note; ^{b}indicates lag order selected by the criterion.

Bound test critical values | |||||
---|---|---|---|---|---|

[Unrestricted intercept &no trend] | |||||

Model | F-stat. | Lag | Level of significance | I (0) | I (1) |

1982 to 2018 | 5.667 | 1 | 1% | 5.15 | 6.36 |

F(lnGDPC_{t}/lnCOP_{t}, lnEXC_{t}) | 5% | 3.79 | 4.85 | ||

K = 2 & n = 36 | 10% | 3.17 | 4.14 |

Source: Authors’ computation 2019 using Eviews 9.

Hypothesized | Eigenvalue | Trace Statistic | 0.05 | Max-Eigen | 0.05 |
---|---|---|---|---|---|

No. of CE(s) | Critical Value | Statistic | Critical Value | ||

r = 0 | 0.430 | 32.952** (0.021) | 29.797 | 19.130 (0.093) | 21.131 |

r ≤ 1 | 0.264 | 13.821 (0.088) | 15.494 | 10.422 (0.185) | 14.264 |

r ≤ 2 | 0.095 | 3.399 (0.065) | 3.841 | 3.399 (0.065) | 3.841 |

Sources: EViews 9; Note: Values in parentheses are the P-values and **represent statistically significant at 5% level.

After establishing a co-integration relationship among the variables, the long-run model in Equation (3) was estimated to obtain the long-run coefficients as presented in

The official exchange rate is also positive and significant at 1% level of significance which is more stringent. Meaning that, 1% increase in exchange rate will lead to 0.069% increase in economic growth of Nigeria in the long run. This result contradicts the findings of researchers such as [

The short run results from the estimation of Equation (4) are reported in

To further ensure the reliability of the estimates, diagnostic tests of serial correlation, functional form, normality and heteroscedasticity were conducted and reported in

As such, the model could produce reliable result. As suggested by Pesaran, Shin and Smith (2001), cumulative sum (CUSUM) and cumulative sum of square (CUSUMSQ) tests for stability of the model along the studied period were conducted. The results are shown in

Dependent variable, ln G D P C t Regressors | Coefficients | T-ratio (P value) |
---|---|---|

ln C O P t | 0.292 | 5.319*** (0.000) |

ln E X C t | 0.069 | 2.915*** (0.006) |

Constant | 6.194 | 34.841*** (0.000) |

Short-run estimation result | ||

Dependent variable, Δ ln G D P C t | ||

ln C O P t | 0.292 | 5.319*** (0.000) |

ln E X C t | 0.069 | 2.915*** (0.006) |

Constant | 6.194 | 34.841*** (0.000) |

ECM (-1) | −0.196 | −3.544*** (0.001) |

e c m = ln G D P C t − 0.2924 × ln C O P t + 0.0693 × ln E X C t + Constant R^{2}: 0.979, DW-statistic: 1.256, F-stat: 503.523*** (0.000) |

Note. ECM = error correction model. *** represent statistically significant at 1% level.

D V = ln G D P C t : | Dynamic OLS | Fully modified OLS | ||
---|---|---|---|---|

Regressors | Coefficients | SE | Coefficients | SE |

Long-run coefficients | ||||

Crude oil price | 0.321*** (8.986) | 0.035 | 0.298*** (7.943) | 0.037 |

Exchange rate | 0.040** (2.719) | 0.014 | 0.039** (2.734) | 0.014 |

Constant | 6.123*** (43.878) | 0.139 | 6.240484*** (53.770) | 0.116 |

Note. Numbers in brackets are the t-statistics. DV = Dependent variable, DOLS = dynamic ordinary least squares; FMOLS = fully modify ordinary least square; OLS = ordinary least square; SE = standard error. *** and ** indicate significant at 1% and 5% levels respectively.

model was stable along the studied periods as the residuals were within the critical bounds at 5% significance level except that there is a slight deviation in the cumulative sum of square (CUSUMSQ).

As a robustness check to the ARDL results, we have employed dynamic DOLS and FMOLS, and their estimated results are reported in

Next is the causal relationship between the variables and was examined by employing Granger causality test to test the direction of causality among the variables in the model. The existence of co-integration necessitates the existence of a causal relation in at least one direction. The Granger causality test results are presented in

This study employed an ARDL approach to co-integration to ascertain the impact of crude oil price and exchange rate on economic growth in Nigeria. The

Null Hypothesis: | Obs | F-Stat (P value) | Direction of causality |
---|---|---|---|

ln C O P t does not Granger Cause ln G D P C t | 34 | 3.008 (0.047)** | Unidirectional causality |

ln G D P C t does not Granger Cause ln C O P t | 1.488 (0.240) | No causality | |

ln E X C t does not Granger Cause ln G D P C t | 34 | 1.244 (0.312) | No causality |

ln G D P C t does not Granger Cause ln E X C t | 1.462 (0.247) | No causality | |

ln E X C t does not Granger Cause ln C O P t | 34 | 1.814 (0.168) | No causality |

ln C O P t does not Granger Cause ln E X C t | 2.576 (0.074)* | Unidirectional causality |

Note. Values in parentheses are the P-values and ** & * represent statistically significant at 5% and 10% significance levels.

study further employed Granger causality to test for the direction of causality among the variables. At first, this study tested for co-integration among the variables in the model after selecting optimum lags and found that all the variables in the model were co-integrated. The long-run model was estimated and the result revealed both the crude oil price and exchange rate impacted positively and significantly on economic growth within the study period. Besides, the long-run model estimations, a short-run model were also estimated for the model. The results also indicated that all the explanatory variables, that is, crude oil price and exchange rate were positive and significant in influencing economic growth in the short run. This suggested that both crude oil price and exchange rate could affect economic growth in both the long-run and the short-run periods.

The robustness check was conducted using dynamic OLS and fully modified OLS, and their results confirmed the result of the long-run ARDL model. The direction of causality was equally tested using Granger causality test, which revealed significant unidirectional causality running from crude oil price to economic growth and from crude oil price to exchange rate in the model.

The main policy recommendation from this study is that since crude oil price and exchange rate have significant positive impact on economic growth in both the short-run and long-run periods which means that increase in earnings from crude oil and appreciation of Naira increases the economic growth of the country and decrease in crude oil earnings and depreciation of Naira decreases the economic growth of the country. Therefore government should diversify its earnings in agriculture, industrialization and investment in order to reduce the heavy reliance on crude oil and income fluctuation resulting from the fluctuation in crude oil prices in order to protect the country’s economy.

The author(s) received no financial support for the research, author-ship, and/or publication of this article.

The author(s) declared no potential conflicts of interest with respect to the re-search, authorship, and/or publication of this article.

Musa, K.S., Maijama’a, R., Shaibu, H.U. and Muhammad, A. (2019) Crude Oil Price and Exchange Rate on Economic Growth: ARDL Approach. Open Access Library Journal, 6: e5930. https://doi.org/10.4236/oalib.1105930