Exchange Rate Volatility and Macroeconomic Performance in Nigeria

Abstract

The study examined the asymmetric relationship between exchange rate volatility and macroeconomic performance in Nigeria covering the period between 1986Q1 and 2019Q4. The Non-linear Generalised Autoregressive Distributive Conditional Heteroscedasticity (GARCH) model was employed. The study was motivated as a result of periodic increase in exchange rate of naira to a dollar and instability of macroeconomic variables in the economy. The presence of Autoregressive Distributive Conditional Heteroscedasticity (ARCH) effect established the use of non-linear GARCH models which showed that volatility was persistent over the period of study. Consequently, the result revealed that exchange rate volatility exhibited a positive relationship with trade balance, industrial output and inflation in the study period. Thus, good news prevailed more over bad news in the foreign exchange market. The study therefore recommended that monetary authorities in Nigeria should regulate exchange rate and macroeconomic variables in order to control the general price level in the economy.

1. Introduction

The obligation of every responsible government is to ensure a balance among the different macroeconomic indicators which reveals the health of the economy. However, the major macroeconomic goal of a country is to achieve rapid growth. But persistent volatile exchange rate is a current impediment for successful macroeconomic policies [1] in any country. Consequently, the monetary authority (Central Bank of Nigeria) has engaged in different exchange rate adjustment policies in attaining macroeconomic objective of price stability. Still, all have proved abortive as greater flexibility in exchange rate is more important to attain equilibrium level and curtail shocks associated with transition from one regime to another. Though, the major objective of exchange rate policy in accordance to [2] is to regulate the domestic currency towards maintaining favourable financial balances and overall macroeconomic stability in order to attain sustainable growth. Subsequently, effort has been made over the years to achieve this objective via adoption of several policy options in foreign exchange market, specifically adoption of Structural Adjustment Programmes (SAP) by developing countries which was initiated by World Bank and International Monetary Fund in 1986. This eventually resulted to instabilities in exchange rates and aggravated inflationary problems.

The [3] clarified the assertion that instabilities in macroeconomic variables cannot increase output, it could harm the economy wherein firms cannot perform effectively in incidence of high inflation. This indicates that the economy cannot grow unless the macroeconomic environment is stabilised. However, since the economy cannot exist in isolation, balance of trade with other countries is another important factor for macroeconomic performance. International trade is said to be sensitive to macroeconomic changes, in that if openness is positively related to growth, inflation which distorts the price of goods ought to be seriously monitored. However, it has been acknowledged that the breakdown of Bretton Woods system in 1973 rendered exchange rate of many countries unstable overtime. This has increased motivation in predicting exchange rate basically because it is an important price that links the world and domestic market for goods and assets. It as well designates competitiveness of a country’s exchange rate with the global market. Its uncontrollable cases have brought about currency crisis in the financial market in terms of output and investment which disrupt macroeconomic performance in Nigeria.

Nevertheless, it is appropriate to examine the interaction that exist among volatility of exchange rate and macroeconomic variables (trade balance, industrial output and inflation) due to the role they play in the overall development of the economy. The three macroeconomic variables are selected because of their sensitivity to exchange rate and their determinant of the general price level in the economy. In relation to trade, it could result to trade surplus or trade deficit therefore distorting trade balance [4]. When trade deficit arises, government might find it difficult to finance the deficit which pose a challenge to the economy. In terms of surplus, it inhibits consequences for government revenue and foreign reserves by upsetting the flow of imports and exports [5]. However, inflation has been acknowledged as the major cause of macroeconomic instability [6]. It was observed that whenever there is an increase in inflation, volatility in exchange rate rises but reverses in periods of relative stability. This implies that increase in exchange rate leads to higher inflation, therefore, variations in exchange rate influences inflation which might prevail for a long time in the economy [7]. Lastly, in relation to output, exchange rate volatility can initiate uncertainty among profit maximisation traders, and thus deteriorates trade balance and alter economic growth [8].

In respect to various empirical studies: [9, 10, 11], it is inappropriate to liberalise trade in the periods of macroeconomic instabilities especially inflation and exchange rate volatility. In this regards, macroeconomic policies will destruct and bring about loss of consumers’ and investors’ confidence and in turn impairs trade in such country. In Nigeria, financial authorities have played an essential role in implementing trade policies and different strategies to regulate important macroeconomic variable that relates to exchange rate. Unfortunately, it has witnessed several economic adversities which have all proved to no avail as the major problem is that Nigerian economy is still an import dependent country. Exchange rate which is the most important variable is still on the increase periodically and this is detrimental to macroeconomic performance.

The main objective of this study is to examine the asymmetric relationship between exchange rate volatility and macroeconomic variables in Nigeria. To accomplish this, the article is structured as follows: the next section reviews empirical literature, followed by theoretical review and methodology. Afterwards, the article dwells on data interpretation and then conclusion.

2. Literature review

Macroeconomist in developed and developing countries have emphasised the impact of exchange rate volatility on macroeconomic performance. In developed countries, using ARCH and GARCH models, [12] examine the link between exchange rate volatility and trade. The result disclosed that bilateral exchange rate volatility lowered trade between countries but displayed positive contribution to trade. Also, asymmetric effect was found between trade-depressing effect and trade promoting effect which are larger for external volatility and swings in volatility. [13] estimated the effect of exchange rate volatility on trade between 1999 and 2009. It was found that upsurge in real exchange rate volatility exhibited negative impact on import and export in the long-run. [14] examined the relationship between exchange rate and inflation in UK and Turkey from 2005 to 2014. The result revealed that purchasing power parity does not occur in Turkey and this might be due to some related factors.

In developing countries, employing the ARCH and GARCH models, [15] examined the effect of exchange rate volatility on macroeconomic performance in Sudan. GARCH technique and two stage Least Square Methods with data series from 1979 to 2009 showed that Real Effective Exchange Rate (REER) volatility exhibited harmful effects on flow of Foreign Direct Investment (FDI) and economic growth. [16] estimated the direct effect of real exchange rate volatility on trade balance in Iran between 1993 and 2011. Result showed that REER had no significant effect on trade balance and trade balance is not affected by export but import. [17] used annual data from periods between 1980 and 2013 to examine the effects and cause of volatility of exchange rate on growth in Ghana. It was established that extreme volatility is dangerous for economic growth. However, decomposition of shocks to exchange rate indicated that three quarter of exchange rate volatility is self-driven. Considering sub-Saharan countries (SSA), [18] investigated the impact of exchange rate volatility on trade applying data spanning from 1993 to 2014. The result found no effect on import when pooled mean group estimators and GARCH model was used. However, negative effect of exchange rate volatility on export transpired in short-run while positive effect occurred in long- run. Considering seven (7) developing countries, [19] investigated effect of exchange rate volatility on FDI and trade along “One Belt and One Road”. Panel data series from 1995 to 2016 in addition to techniques of Threshold Autoregressive Conditional Heteroskedasticity (TGARCH) was used. Result showed that exchange rate volatility affected FDI and trade in OBOR related countries.

In Nigeria, [20] reported the presence of overshooting volatility shocks while investigating consistency, severity and persistency of exchange rate volatility from 1986 to 2008. [21] focused on monthly data from 2000 to 2015 in examining impact of exchange rate volatility on trade balance in Nigeria, a long run relationship was found between exchange rate volatility and trade. [22] investigated the impact of exchange rate volatility and the role exchange rate policy plays in Nigerian economy. Data spanning from 1996 to 2017 showed causal relationship between GDP growth and exchange rate volatility are inversely related but a bidirectional relationship occurred between RGDP and exchange rate. [23] investigated relationship between exchange rate volatility and inflation. Quarterly data ranging from 1970Q1–2014Q4 was employed. Result showed there was no causal relationship between real exchange rate and inflation.

[24] examined the interaction among exchange rate volatility, interest rate and exchange rate pass-through in Nigeria. Monthly time series spanning from 1970 to 2008 was used. The result showed there exist positive relationship among the variables in long run but negative nexus existed between inflation and exchange rate volatility in short-run. [25] examined exchange rate volatility and sectorial export oil and non-export sectors in Nigeria using annual data from 1980 to 2011. GARCH techniques was employed to measure volatility of exchange rate while Seemingly Unrelated Regression was used to estimate coefficient of two-system equation. The result suggested exchange rate was unpredictable whereas the SUR model showed existence of negative relationship between export performance and exchange rate volatility of non-oil and oil sector. [26] used ARCH model and extension (GARCH, Exponential Generalised ARCH (EGARGH) and Threshold (TGARCH)) to examine effect of exchange rate volatility on export of non-oil in Nigeria. Quarterly data spanning from 1986Q1–2014Q4 was used in conjunction with ECM technique. The result confirmed presence of exchange rate volatility and existence of negative impact on non-oil export.

In view of the empirical literature above, the gap identified is the issue of measurement of exchange rate volatility. Several methods have been employed in measuring volatility: moving average, ARCH and GARCH models. [21] are of the opinion that GARCH is the right model for modelling volatility in Nigeria. But because the measurement of volatility is of great importance to macroeconomic performance, there is need to employ non-linear GARCH model due to its advantages of positive variance irrespective of estimated parameters and its asymmetric effects on innovations [27]. Furthermore, according to [28], asymmetric GARCH have revealed better results than simple GARCH models.

3. Theoretical review and methodology

The study is based on Mundell-Fleming model (MFM) developed by [29, 30]. The model is based on the extension of IS-LM model. The traditional IS-LM model describes an economy under autarky (i.e., closed economy) but MFM describes an open economy. It designates the relationship between output (short run), nominal exchange rate and interest rate in an economy. Thus, Mundell-Fleming model is adopted in this study since Nigeria is attributed to an open economy. In that, Nigeria is small to influence the world market in terms of world prices and interest rate and then assumed to have a perfect capital mobility.

3.1 Model specification

The overall model to capture the broad objective of the study which is to examine the asymmetric relationship between exchange rate volatility and macroeconomic variables in Nigeria is stated in the equation below:

$$Y_t=f\left({\mathit{ERV}}_t^{\ast }\right)$$E1

where $‘Y’$ is output and ‘ERV’ is exchange rate volatility.

The macroeconomic theory in relation to Mundell-Fleming framework implies that income, interest rate, exchange rate, price, net export and similar variables can be implicitly related.

$$Y_t=f\left({\mathit{ERV}}_t,Z_t\right)$$E2

where ‘$Y_t$’ is output, ‘${\mathit{ERV}}_t$’ is exchange rate volatility; $Z_t$ is vector of macroeconomic variables.

The explicit equation of the above is presented below as:

$$Y=f\left(i,e,P,\mathit{NX}\right)$$E3

where ‘i’ is interest rate, ‘e’ is exchange rate, ‘P’ is price and ‘NX’ is Net Export.

Also, macroeconomic theory behind the Mundell-Fleming framework makes us believe that income, interest rate, exchange rate, price, net export variables can be implicitly related as follows:

The Mundell-Fleming Model

$$Y=C\left(Y-t\left(Y\right)\right),r-E\left(\pi \right)+I\left(r-E\left(\pi \right),Y_{t-1}\right)+G+\mathit{NX}\left(e,Y,Y^{\ast }\right)$$E4

where $Y_t$ = Output; $C_t$ = Consumption; $I_t$ = Investment; $R_t$ = Interest rate; ${\mathit{\text{in}}}_t$ = Inflation rate; $G_t$ = Government spending; ${\mathit{TB}}_t$ = Trade balance; $E_t$ = Exchange rate.

Therefore, the model formation of this study is illustrated below as:

$$Y=f\left(i,R,E,\mathit{NX}\right)$$E5

where ‘i’ is inflation rate; ‘R’ is interest rate; ‘E’ is Exchange Rate and NX is ‘Net Export’.

3.2 Estimation of heteroskedasticity using non-linear GARCH model

The confirmation of the presence of ARCH in the model as presented in Table 1 enabled the study to proceed to non-linear GARCH. The ARCH model comprises of two parts: the mean equation and the variance equation as proposed by [31]. The mean equation can be mathematically specified as:

F-Statistics	9.9019	Prob. F (1,133)	0.0020
Obs*R-squared	9.3544	Prob. Chi-Square (1)	0.0022
Scaled explained SS	276.1406	Prob. Chi-Square (1)	0.0000

Table 1.

ARCH test.

Source: Authors computation 2021.

$$Y_t=\alpha +{\beta }^{!}{X}_t+{\mu }_t$$E6

where $Y_t$ is the vector of variables; $\alpha$ represents the constant term. ${\beta }^{!}$ is the vector of unknown parameter; $X_t$ represents the vector of unknown variable; then ${\mu }_t$ is the random error term.

The variance equation is presented as:

$$h_t={\gamma }_0+\sum _{i=1}^q{\gamma }_1{\mu }_{t-i}^2$$E7

ARCH model is majorly a moving average (MA) and the variance is mainly responding to errors. It does not capture the autoregressive (AR) part of the model. This prompted the use of more comprehensive models like GARCH propounded by [32].

Consequently, the GARCH model also lack some important features. It cannot explain the effect of events and news which exhibit asymmetric effect on exchange rates. However, investors react in different ways to incidence of good and bad news in the financial market, wherein bad news leads to higher volatility. But the non-linear GARCH (EGARCH) model is proficient in capturing events, news and incidence that lead to asymmetric impact and this gives preference to use EGARCH which is believed to be superior in accounting for asymmetric and non-linear effects [33].

The aim of achieving a robust result resulted in the consideration of EGARCH model for this study. To begin with, exchange rate returns was generated through the log of difference of exchange rate which is mathematically specified as:

$$dlog\left({\mathit{EXR}}_t-{\mathit{EXR}}_{t-1}\right)$$E8

The EGARCH model as propounded by Nelson (1991) is specified in a general form as:

$$log\left(h_t\right)={\gamma }_0+\sum _{i=1}^p{\beta }_i\mid \frac{{\mu }_{t-i}}{\sqrt{h_{t-i}}}\mid +\sum _{i=1}^q{\gamma }_i\frac{{\mu }_{t-i}}{\sqrt{h_{t-i}}}+\sum _{i=1}^m{\alpha }_{i}log\left(h_{t-1}\right)$$E9

where ${\gamma }_0$ is constant term, ${\beta }_i$ is a measure of ARCH effect, ${\gamma }_1$ is the leverage effect, ${\alpha }_i$ is the GARCH effect, positive value of ${\mu }_{t-i}$ with total effect as $\left(1+{\gamma }_i\right)\mid {\mu }_{t-i}$ denotes good news, negative value of ${\mu }_{t-i}$ with total effect as $\left(1-{\gamma }_i\right)\mid {\mu }_{t-i}$ denotes bad news. In this case, good news is assumed to have a greater effect on volatility compared to bad news. There is asymmetry if ${\gamma }_i\ne 0$ and symmetry when ${\gamma }_i=0$.

3.3 Source of data and variable selection

This research employed secondary (quarterly) data spanning from 1986Q1–2019Q4. The data was sourced from CBN Statistical Bulletin (2018; 2019; 2020) and International Financial Statistics (2019) editions. The macroeconomic variables used in the study are trade balance, industrial output and inflation while interest rate was used as control variable in the model.

4. Data interpretation

The descriptive statistics is made up of statistical distribution of a total of 136 observation presented in Table 2. The result showed that almost all the variables demonstrated significant level of consistency within the minimum and maximum values. The closeness of the mean and median indicates the variables have a normal distribution. It was discovered that the mean and the median values are positive Also, the mean and the median falls within the maximum value. This suggest that the individual values are normally distributed. The large difference between the maximum and minimum exchange rate indicates the high volatile nature of exchange rate. The standard deviation which measures the amount of dispersion of a variable revealed that the standard deviation of the variables reported were very low except for trade balance and interest rate which are greater than two. This revealed that the values of the variables clustered around their average with little or no variability. For a normal distribution, the skewness is zero while positive and negative skewness indicates distribution with right and left tail respectively. The standard deviation, skewness and kurtosis greater than zero indicates that the distribution is not normally distributed. The positive skewness of exchange rate returns, industrial output and interest rate implies their distributions are skewed to the right. Then, the negative skewness of exchange rate, trade balance and inflation indicate their distribution are skewed to the left. Negative skewness denotes a tail towards the left side of the distribution while positive skewness signifies a tail towards the right side of the distribution. Lastly, the Jarque-Bera statistics which is used to test the normality in the distribution of the series. The result indicated that the probability of J-B is or less than 0.05. Therefore, the null hypothesis of normal distribution is rejected and the alternative hypothesis that the variables have a non-normal distribution is accepted. The implication of the non-normality of trade balance and inflation verified their responses to the volatile nature of exchange rate movement in Nigeria.

The result of the unit root test presented in Table 3 was to verify the stationarity properties of the variables. The Augmented Dickey Fuller (ADF) and Phillips Peron (PP) test was used to determine the order of integration of the variables. The result revealed that exchange rate returns (RNER) and trade balance (TDB) are stationary at levels, i.e., I (0) while NER (exchange rate), industrial output (IDP), inflation (INF) and interest rate (INR) are stationary at first difference, i.e., I (1), hence the variables are integrated at levels and order one i.e., mixture of I (0) and I (1).

4.1 Preliminary test

In order to verify the presence of heteroscedasticity in the model, ARCH test was carried out using Breush-Pagan-Godfrey method. Therein, the evidence of ARCH effect enables the study to proceed to estimating non-linear GARCH model. The ARCH effect result is presented in Table 1 while Figure 1 shows the evidence of volatility clustering. Therein, the null hypothesis states that there is no presence of heteroscedasticity in the return series. But since the probability is 1%, the null hypothesis is rejected while the alternative hypothesis of presence of ARCH effect is accepted.

Figure 1.

Graphical representation of returns of exchange rate in Nigeria. Source: Authors computation (2021).

4.2 Non-linear GARCH results

The result of EGARCH (1,1) model for dependent variable trade balance is presented in Table 4. The normal distribution analysis reveals that the effect of exchange rate volatility on trade balance is negative at −0.013%. The ARCH result is 1.32% at a significance level of 1%, this showed a positive relationship between exchange rate volatility and trade balance. The leverage term is −0.045% which is significant at 1%. The negative leverage term implied that bad news prevails over good news in the foreign exchange market. Likewise, the GARCH result at 0.81% indicate that present volatility positively predicts future volatility at 1% level of significance. The result of student-t distribution shows the marginal effect of exchange rate volatility on trade balance is −0.02%. The result of ARCH is 1.32% at 1% significance which indicates a positive relationship between exchange rate volatility and trade balance. The leverage term at 0.07% was significant at 1%. The positive leverage effect implies that good news prevails over bad news in the foreign exchange market. Also, the GARCH term was 0.81% which indicates that present volatility positively predicts the future volatility at 1% level of significance. Lastly, the generalised error distribution result show that the marginal effect of exchange rate volatility on trade balance is −0.04%. The outcome of the ARCH result is 1.29% and this implies a positive relationship between exchange rate volatility and trade balance. The leverage effect is at 0.183% and it show that goods news prevails over bad news in the foreign exchange market. The GARCH term is 0.80% and this implies that present volatility predicts the future volatility at 1% significance level. Nevertheless, the best model for the distribution is the generalised error distribution with minimum variance of Schwarz Criterion value at 3.5145 and log likelihood value of −212.7013.

Variables	NER	RNER	TDB	IDP	INF	INR
Mean	3.6888	0.0064	10.3603	11.4508	3.3763	11.9218
Median	4.4165	0.0012	10.9364	11.1563	3.7163	11.5400
Maximum	5.1945	3.6477	13.4853	15.0223	5.6339	25.9300
Minimum	−0. 3567	−1.4628	1.7574	9.8521	−0.0619	4.6000
Std. Dev.	1.4173	0.3781	2.4548	1.3104	1.6539	3.9657
Skewness	−1.1526	5.9230	−0.6673	1.0456	−0.6659	0.8533
Kurtosis	3.2353	68.1764	2.7179	3.6343	2.2620	3.9657
Jarque-Bera	30.4238	2468.11	10.5427	27.0588	13.1386	21.5339
Probability	0.0000	0.0000	0.0051	0.0000	0.0014	0.0000
Observations	136	136	136	136	136	136

Table 2.

Descriptive statistics.

Note: NER is exchange rate; RNER is the returns of exchange rate, TDB is trade balance; IDP is industrial output; INF is inflation, INR is interest rate.

Source: Authors computation 2021.

Variables	Levels			First difference
	Intercept + trend			Intercept + trend
	ADF	PP	STATUS	ADF	PP	STATUS
NER	−3.6811*	−3.2571*	—	−11.5112***	−11.6093***	I (1)
RNER	−11.5112***	−11.6093***	I(0)	—	—
TDB	−5.7023***	−5.6849***	I(0)	—	—
IDP	−1.6265	−1.6846	—	−12.7396***	−12.7625***	I (1)
INF	−2.3329	−1.3505	—	−3.8607**	−7.4348***	I (1)
INR	−4.5889***	−3.4472**	—	−8.1564***	−7.6674***	I (1)

Table 3.

Unit root test.

Note: ***, **, * at 1%, 5% and 10% respectively.

Source: Authors computation 2021.

EGARCH (1,1)
Normal Distribution
Mean Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
RNER	−0.0133	0,3492	−0.0380	0.9697
Constant	14.7341	1.0363	14.2186	0.0000***
Variance Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
Constant	−0.9879	0.2023	−4.8832	0.0000****
Residual Square	1.3196	0.2863	4.6084	0.0000***
Leverage Term	−0.0453	0.1661	0.2809	0.7787
GARCH (−1)	0.8133	0.1125	7.2273	0.0000***
Log Likelihood	−210.7896
Schwarz Criterion	3.4498
Student.t Distribution
Mean Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
RNER	−0.0203	0.6953	−0.0292	0.9767
Constant	14.3965	1.0252	14.0427	0.0000***
Variance Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
Constant	−0.9671	0.2068	−4.6753	0.0000***
Residual Square	1.3157	0.3035	4.3346	0.0000***
Leverage Term	0.0725	0.1764	0.4107	0.6813
GARCH (−1)	0.8131	0.1201	6.7663	0.0000***
Log Likelihood	−211.0528
Schwarz Criterion	3.3623
Generalized Error Distribution
Mean Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
RNER	−0.0442	0.4357	−0.1016	0.9191
Constant	12.1887	1.0669	11.4235	0.0000***
Variance Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
Constant	−0.8419	0.2173	−3.8739	0.0001***
Residual Square	1.2879	0.4043	3.1854	0.0014***
Leverage Term	0.1832	0.2348	0.7800	0.4354
GARCH (−1)	0.7955	0.1201	6.6217	0.0000***
Log Likelihood	−212.7013
Schwarz Criterion	3.5145

Table 4.

EGARCH (1,1) result for dependent variable trade balance.

***, ** and * represent 1%, 5% and 10% respectively.

Source: Authors computation 2021.

The result of EGARCH (1,1) for dependent variable industrial output is presented in Table 5. The outcome of the normal distribution show that the effect of exchange rate volatility on industrial output is positive at 0.04%. The leverage term is 0.95% which indicates that good news prevails over bad news in the foreign exchange rate market. The positive outcome of ARCH is at 0.93% with a significant level of 1%, this shows the existence of a positive relationship between exchange rate volatility and industrial output. The GARCH result is 0.94% at 1% significance level and this indicates that present volatility positively predicts the future volatility at 1% level of significance. The student-t distribution result reveal that the marginal effect of exchange rate volatility on industrial output is 0.04% at 10% significance level. This indicate that exchange rate volatility exhibits a positive relationship with industrial output in Nigeria. The leverage effect is 0.08% which indicate goods news prevails over bed news in the foreign exchange market. The ARCH result of 0.08% infer that present volatility positively predicts the future volatility at 1% significance level. Moreover, the GARCH term at 0.94% infer that present volatility positively predicts the future volatility at 1% significance. The result of generalised error distribution showed that the effect of exchange rate volatility on industrial output is 0.04%. This designates a positive relationship between exchange rate volatility and industrial output in Nigeria. The leverage term at 0.09% indicate that good news prevails over bad news in foreign exchange market. In reporting the GARCH term at 0.93%, this infers that present volatility predicts future volatility at 1% significance. In relating the distributions in terms of goodness of fit, normal distribution is the best model to be considered with a minimum variance value of SC at 0.5102 and the log likelihood maximum value at −12.3675.

EGARCH (1,1)
Normal Distribution
Mean Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
RNER	0.0441	0.0319	1.3777	0.1683
Constant	10.0803	0.0522	193.0716	0.0000***
Variance Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
Constant	−0.8958	0.3099	−2.8902	0.0038***
Residual Square	0.9260	0.3001	3.0862	0.0020***
Leverage Term	0.0955	0.0996	0.9586	0.3378***
GARCH (−1)	0.9371	0.0546	17.1738	0.0000***
Log Likelihood	−12.3675
Schwarz Criterion	0.5502
Student.t Distribution
Mean Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
RNER	0.0439	0.0307	1.4310	0.1524*
Constant	10.0771	0.0508	198.4041	0.0000***
Variance Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
Constant	−0.8767	0.3143	−2.7895	0.0053***
Residual Square	0.9198	0.3079	2.9866	0.0028***
Leverage Term	0.0843	0.1404	0.6007	0.5480
GARCH (−1)	0.9406	0.0549	17.1186	0.0000***
Log Likelihood	−12.2295
Schwarz Criterion	0.5445
Generalized Error Distribution
Mean Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
RNER	0.0437	0.0368	1.1870	0.2352
Constant	10.0819	0.0559	180.3712	0.0000***
Variance Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
Constant	−0.9242	0.2920	−3.1645	0.0016***
Residual Square	0.9360	0.2763	3.3878	0.0007***
Leverage Term	0.0913	0.1163	0.7851	0.4324
GARCH (−1)	0.9332	0.0496	18.8167	0.0000***
Log Likelihood	−11.6548
Schwarz Criterion	0.5360

Table 5.

EGARCH (1,1) result for industrial dependent variable output.

***, ** and * represent 1%, 5% and 10% respectively.

Source: Authors computation 2021.

The EGARCH (1,1) result for dependent variable inflation is presented in Table 6. The normal distribution show that the effect of exchange rate volatility on inflation is negative at −0.03%. The ARCH result at 1.0% with a significant level at 1% infer that exchange rate volatility and inflation are positively related. The leverage term is −0.45% and indicates that bad news prevails over good news in the foreign exchange rate market. However, the GARCH effect is 0.79% which indicates that the present volatility predicts the future volatility at 1% significance level. The student-t result revealed that the marginal effect of exchange rate volatility on inflation is negative at −0.03%. The ARCH result is 1.01% and significant at 1% level indicate the existence of a positive relationship between exchange rate volatility and inflation. The leverage effect at −0.44% infer that bad news prevails over good news in the foreign exchange market. The GARCH term at 0.81% revealing that present volatility predicts the future volatility at 1% level of significance. Lastly, the generalised error distribution result showed that the marginal effect of exchange rate volatility on inflation is negative at −0.03%. The ARCH effect at 0.99% deduce there is a positive relationship between exchange rate volatility and inflation at 1% significant level. The leverage term at 0–0.48% indicate that bad news prevails over good news in the foreign exchange market. Then the GARCH result at 0.79% indicate that present volatility predicts future volatility at 1% level of significance. In choosing the best model for goodness of fit with minimum variance, the normal distribution with log likelihood value of 83.9817 and minimum variance of 1.7712 is selected.

EGARCH (1,1)
Normal Distribution
Mean Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
RNER	−0.03225	0.0961	−0.3385	0.7350
Constant	−6.3262	0.2901	−21.8100	0.0000***
Variance Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
Constant	−1.3526	0.2710	−4.9908	0.0000***
Residual Square	1.0013	0.2123	4.7158	0.0000***
Leverage Term	−0.4518	0.1698	−2.6609	0.0078***
GARCH (−1)	0.7986	0.0862	9.2690	0.0000***
Log Likelihood	−83.9817
Schwarz Criterion	1.7712
Student.t Distribution
Mean Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
RNER	−0.0326	0.0880	00.3705	0.7110
Constant	−6.5065	0.3386	−19.2155	0.0000***
Variance Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
Constant	−1.3407	0.3640	−3.6829	0.0002***
Residual Square	1.0066	0.2210	4.5536	0.0000***
Leverage Term	−0.4374	0.2089	−2.0936	0.0363***
GARCH (−1)	0.8074	0.0936	8.6283	0.0000***
Log Likelihood	−83.9067
Schwarz Criterion	1.6064
Generalized Error Distribution
Mean Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
RNER	−0.0335	0.1022	−0.3277	0.7431
Constant	−6.3774	0.3362	−18.9700	0.0000***
Variance Equation
Variables	Coefficient	Std. error	z-Statistics	Prob.
Constant	−1.3899	0.3633	−3.8261	0.0001***
Residual Square	0.9907	0.2116	4.6808	0.0000***
Leverage Term	−0.4797	0.1884	−2.5469	0.0109***
GARCH (−1)	0.7888	0.0865	9.1151	0.0000***
Log Likelihood	−83.8706
Schwarz Criterion	1.6059

Table 6.

EGARCH (1,1) result for dependent variable inflation.

***, ** and * represent 1%, 5% and 10% respectively.

Source: Authors computation 2021.

5. Conclusion and policy implication

The study examined the asymmetric relationship between exchange rate volatility and macroeconomic performance in Nigeria through the period between 1986Q1 and 2019Q4 using the non-linear GARCH model. In order to employ the non-linear GARCH model (EGARCH), ARCH effect was verified to confirm the presence of heteroscedasticity. The preliminary test such as unit root test and ARCH test to verify the presence of heteroscedasticity revealed the evidence of volatility clustering. The outcome of the analysis revealed that volatility movement was very high and persistent over the period of study. Exchange rate volatility exhibited a positive relationship with trade balance and industrial output but a negative relationship with inflation. The leverage effect of exchange rate volatility on trade balance and industrial output indicated the prevalence of good news over bad news. But contrary to the aforementioned, the leverage effect of exchange rate volatility on inflation revealed the prevalence of bad news over good news in the foreign exchange market. On the other hand, the present volatility predicts the future volatility for all the variables. Also, the non-linear GARCH model confirmed that the normal distribution is the best model for traders in the financial market. The implication of the findings imply that exchange rate volatility is an important factor in the macroeconomic performance of the Nigerian economy. It is therefore recommended that government should procure a stable economy through the monetary authorities (CBN) by implementing policies to control and regulate exchange rate and macroeconomic variables in order to control the general price level. Also, investors and financial analyst should consider exchange rate volatility in predicting macroeconomic performance in the future by ensuring stability in the financial market. Furthermore, even though investors in financial markets are risk averse, it is recommended that they should be more sensitive to good news rather than bad news in the events of movement of exchange rate volatility on macroeconomic performance in Nigeria.

• The probability density function of normal distribution is represented as:

  $$f\left(x|\mu ,\sigma \right)=\frac{1}{\sigma \sqrt{2\pi }}{e}^{\frac{-{\left(x-\mu \right)}^2}{2{\sigma }^2}}$$E10

• Its log likelihood function in GARCH term is:

  $$-\frac{n}{2}ln\left(2\pi \right)-\frac{n}{2}ln\left(h\right)-\frac{1}{2h}\sum _{j=1}^n{\left(x_j-\mu \right)}^2$$E11

• The probability density function for Student-t distribution is illustrated as:

  $$f\left(y,v\right)=\frac{\mathrm{\Gamma }\frac{v+1}{2}}{\sqrt{\pi \left(v-2\right)\mathrm{\Gamma }\frac{v}{2}{\left(1+\frac{v^2}{v-2}\right)}^{\frac{v+1}{2}}}}$$E12

• The log likelihood function in GARCH term is depicted as:

  $$\begin{array}{l}log\left[\mathrm{\Gamma }\left(\frac{v+1}{2}\right)\right]-log\left[\mathrm{\Gamma }\left(\frac{v}{2}\right)\right]-\frac{1}{2}log\left(\pi \left(v-2\right)\right)\\ -\frac{1}{2}\sum _{j=1}^n\left[log\left(h_t\right)+\left(v+1\right)log\left(1+\frac{{\epsilon }_t^2}{h_t\left(v-2\right)}\right)\right]\end{array}$$E13

• The probability density function for generalised error distribution is mathematically represented as:

  $$f\left(x|\mu ,\sigma ,k\right)=\frac{e^{-\frac{1}{2}\mid x-\mu \mid \frac{1}{n}}}{2^{k+1}\sigma \mathrm{\Gamma }\left(k+1\right)}$$E14

• Its log likelihood function in GARCH term is:

  $$-\frac{1}{2}{\left|x-\mu \right|}^{\frac{1}{n}}-\left(k+1\right)log\left(2\right)-log\left(h\right)-log\left(\mathrm{\Gamma }\right)-log\left(k+1\right)$$E15

JEL Classification

B22, C22, E31, E58, F14