Data and sample
In this research study, we specify the variables exchange rate, inflation rate, RE, BOP, GDP, money supply, population size and trade of the emerging economies for the period 1990 to 2019. The data employed in this study is secondary data and is retrieved from the official website of the Organization for Economic Cooperation and Development (OECD), www.oecd.org. In this study, we use a study sample of seven countries known as the emerging seven (E7 economies): China, India, Brazil, Mexico, Russia, Indonesia and Turkey.
The exchange rate is the price of one country’s currency expressed as a ratio of that of another country. Thus, the exchange rate shows the value of one country’s currency related to that of other countries. The exchange rate values used in this study are expressed as the home currency per one United States dollar. Thus, an increase in the exchange rate causes a depreciation of the home country’s currency and vice versa. Inflation is a continuous increase in a country’s goods and services and is not suitable for the country’s economy since it erodes the currency’s buying power. RE is referred to as the contribution of renewables to the total primary energy supply, and it comes in the form of hydro, tidal, geothermal, wind, solar and wave sources. Money supply is the total amount of money available in circulation in the money market and in this study, the number of notes and coins (M1) in circulation is taken to represent money supply. GDP refers to the total value of all products produced within the boundaries of a country and is measured in million dollars. Population size is the number of people currently residing in a country. Balance of payment is the current account record of one nation’s international transactions with those of other countries in the world and trade in goods and services is the transactions in goods and services between residents of a country and non-residents.
Method
Based on the findings of Deka and Dube (2021), and Deka, Cavusoglu and Dube (2021) who found a significant link between currency appreciation and the use of RE, in this study, we seek to further examine whether the association really holds. Cointegration Regression analysis, including Fully Modified Ordinary Least Square (FMOLS) by Phillips and Hansen (1990) and Dynamic Ordinary Least Square (DOLS) by Stock and Watson (1993), is used as the study methodology to ascertain the relationship between exchange rate as the dependent variable and the explanatory variables (RE, GDP, BOP, money supply, population size, trade and inflation rate) of the E7 economies. The FMOLS and DOLS models provide robust long-run results on panel data that is cointegrated, not stationary at level and stationary at first difference. Hence, in this study as there is cointegration among the variables and they are integrated of order one, we employ these models. FMOLS and DOLS are cointegration regressions that require two essential prerequisites to be met: the variables should be non-stationary at level and stationary at first difference (integrated of the same order 1) and they must have at least one cointegration equation.
In this study, we employ the Augmented Dicker Fuller (ADF) test and Phillips Peron (PP) test to check the stationarity of the variables (Box & Jenkins, 1976; Granger, 1986; Engle & Granger, 1987). The ADF test was pioneered by Dickey and Fuller (1979), and it is an extension of the Dickey test developed by Dickey (1976). The Dickey test is only capable of checking the stochastic trend of a variable but not the deterministic trend and the ADF test can identify both stochastic and deterministic trends in a time series (see Gujarati, 2004). The other test of unit root in a variable is the PP test by Phillips and Peron (1988). The PP test is used together with the ADF test for robustness (Gujrat, 2004; Granger, 1986). Granger (1986) and Engle and Granger (1987), suggested that if the variables are non-stationary at level and stationary at first difference, then a relationship such as Zt exists that is stationary, and this is known as the cointegration relationship. Granger (1986) stated that if variables Yt and Xt, are connected through a relationship that contains a constant term, then the following cointegration relationship that is stationary will be obtained.
Zt = Yt – aXt ~I (0) (1)
Where Zt is the cointegration relationship between Yt and Xt that is stationary at level, and a is a unique constant term connecting Yt and Xt, in a cointegration relationship, see Granger (1986). The relationship expressed in Equation 1 above may contain some seasonal problems (Granger, 1986). However, another relationship such as the one below may exist that has no seasonal issues.
Yt = aXt (2)
If variables are cointegrated, this means that a long-run relationship among them exists. They have a long-run association, and this association can be ascertained by employing cointegration regressions such as FMOLS and DOLS models, among others. For this study, the following equation (Equation 3) represents the statistical representation of the cointegration regression (FMOLS and DOLS) used in this study. Equation 3 below shows the statistical relationship between the exchange rate and the explanatory variables employed. In this study, we used the exchange rate as the explained variable.
ER = β0 + β1RE + β2INFL + β3BOP + β4MS + β5GDP + β6PP + β7TRD + et (3)
In Equation 3 above, ER represents the exchange rate, RE is the renewable energy, INFL represents inflation rate, BOP is the balance of payment, MS is the money supply, GDP is the Gross Domestic Product, PP is the population size, TRD is the trade of goods and services and et is the error term, while β0 to β7 are the model’s coefficient values.
Data analysis
Descriptive statistics
The results of this study’s descriptive statistics are provided in Table 1. The total number of observations per each variable, exchange rate, balance of payment, inflation, RE, money supply, trade, GDP, and population size is 210. In Table 1, we also provide the mean, median, maximum, standard deviation and total value of every indicator under study. It is shown that in the E7 economies for the period under study, the average values of BOP, exchange rate, GDP, inflation, money supply, population size, RE and trade are -0.1191, 1190, 9164.9, 58.77, 39.97, 454.16 21.38 and 20.6. respectively (see Table 1 below). Moreover, their Sum values are -25.02, 249903, 1924649, 12341.7, 8393.4, 95373.75, 4490.34, and 4426.14, respectively. For the median, maximum, minimum and standard deviation values of the indicators under study, please see Table 1 below.
Table 1
Descriptive statistics results
|
BOP
|
ER
|
GDP
|
IFL
|
MS
|
PP
|
RE
|
TRD
|
Mean
|
-0.1191
|
1190.0
|
9164.9
|
58.77
|
39.97
|
454.16
|
21.38
|
20.60
|
Median
|
0.00
|
8.49
|
8624.5
|
6.95
|
22.5
|
182.14
|
18.14
|
22.57
|
Maximum
|
10.35
|
14236.9
|
29189
|
2947.7
|
170.6
|
1433.8
|
46.71
|
48.19
|
Minimum
|
-8.82
|
0.00
|
0.00
|
-1.40
|
0.00
|
55.12
|
0.00
|
0.00
|
Std. Dev.
|
2.87
|
3259.3
|
7489.34
|
294.15
|
41.99
|
498.51
|
14.80
|
9.72
|
Sum
|
-25.02
|
249903
|
1924649
|
12341.7
|
8393.4
|
95373.75
|
4490.34
|
4326.14
|
Obs.
|
210
|
210
|
210
|
210
|
210
|
210
|
210
|
210
|
Correlation analysis
Table 2 below provides the correlation results of exchange rate, balance of payment, inflation, RE, money supply, trade, GDP, and population size for the E7 economies. The results of the correlation analysis show that population size is negatively correlated with inflation rate and GDP and positively correlated with BOP. Furthermore, the results provided in Table 2 indicate that money supply is negatively correlated with BOP and inflation rate, while it is positively correlated with GDP and population size. RE use is negatively correlated with BOP, inflation rate, GDP, and money supply, but positively correlated with exchange rate, and population size. Trade is also negatively correlated with inflation rate, population size and RE use, but positively correlated with BOP, exchange rate, GDP and money supply. A negative significant correlation shows that the variable has an inverse association such that a rise in one variable will cause a decline in the other, whereas a positive significant correlation implies that the variables move in the same direction such that a rise in one causes the other to rise as well.
Table2: Correlation results
Correlation
t-Statistic
|
BOP |
ER |
IFL |
LGDP |
LPP |
MS |
RE |
TRD |
BOP
|
1.000
|
|
|
|
|
|
|
|
|
-----
|
|
|
|
|
|
|
|
ER
|
-0.0536
|
1.0000
|
|
|
|
|
|
|
|
-0.6954
|
-----
|
|
|
|
|
|
|
IFL
|
-0.0202
|
-0.0829
|
1.0000
|
|
|
|
|
|
|
-0.2623
|
-1.0792
|
-----
|
|
|
|
|
|
LGDP
|
-0.0702
|
-0.0913
|
-0.0577
|
1.0000
|
|
|
|
|
|
-0.9119
|
-1.1886
|
-0.7495
|
-----
|
|
|
|
|
LPP
|
0.2596
|
0.0171
|
-0.3555
|
-0.6268
|
1.0000
|
|
|
|
|
3.4837***
|
0.2212
|
-4.9294***
|
-10.4269***
|
-----
|
|
|
|
MS
|
-0.1480
|
0.0735
|
-0.3796
|
0.5199
|
0.1409
|
1.0000
|
|
|
|
-1.9398*
|
0.9549
|
-5.3189***
|
7.8902***
|
1.8447*
|
-----
|
|
|
RE
|
-0.3086
|
0.4089
|
-0.1280
|
-0.3297
|
0.203000
|
-0.0104
|
1.0000
|
|
|
-4.2049***
|
5.8082***
|
-1.6729*
|
-4.5269***
|
2.6871***
|
-0.1348
|
-----
|
|
TRD
|
0.3109
|
0.2158
|
-0.0453
|
0.2010
|
-0.0520
|
0.0518
|
-0.4263
|
1.000
|
|
4.2402***
|
2.8646***
|
-0.5888
|
2.6597***
|
-0.6761
|
0.6724
|
-6.1082***
|
-----
|
***significant at 1% level |
Unit root test results
In this section of the study, we provide the findings of the ADF and PP unit root tests. In Table 3 below, the results of the unit root test are given for all the variables that have been employed in this research, including exchange rate, balance of payment, inflation, RE, money supply, trade, GDP, and population size. The findings of both ADF and PP tests show that all variables are stationary at first difference, and hence, they are integrated of order one. They are non-stationary at level and become stationary after first difference. It is only inflation rate that is stationary at level at 1% level of significance according to the results of both tests. However, it is stationary again at first difference. Cointegration regressions such as FMOLS, DOLS, VECM require that variables should be stationary at first difference (Granger, 1986); Engle and Granger, 1987).
Table 3
|
ADF
|
PP
|
|
level
|
First Difference
|
level
|
First Difference
|
ER
|
11.1195
|
41.9224***
|
9.96939
|
65.6024***
|
RE
|
6.02590
|
27.9352**
|
2.65510
|
37.5590***
|
lnGDP
|
11.2178
|
27.0733**
|
5.24613
|
99.4086***
|
lnPP
|
10.1000
|
29.2318**
|
269.945***
|
43.6814***
|
IFL
|
31.5105***
|
93.5171***
|
43.6471***
|
93.5171***
|
BOP
|
7.54456
|
41.9177***
|
8.14009
|
88.8733***
|
MS
|
26.6199*
|
33.5923***
|
12.2573
|
46.7127***
|
TRD
|
16.3276
|
61.5499***
|
14.0176
|
126.053***
|
***significant at 1% level |
Cointegration analysis
In this study, we employ the Johansen and Kao cointegration tests to analyze the cointegration relationship among the variables specified in the research model of this study. Table 4 below shows that for both the trace test and max-eigen test, the variables of exchange rate, balance of payment, inflation, RE, money supply, trade, GDP, and population size are significantly cointegrated at 1% level of significance. This is because the p-values of both the Trace test and max-eigen test are less than 0.01, indicating that the variables are cointegrated at 1% level. The results of the Kao cointegration test also show that we should reject the null hypothesis of no cointegration and accept that the variables are cointegrated. This shows that exchange rate, balance of payment, inflation, RE, money supply, trade, GDP, and population size have a long-run relationship (Granger, 1986); Engle and Granger, 1987); Granger and Weiss, 1983). Therefore, the variables move together in the long run.
Table 4
Cointegration test results
Johansen Fisher Panel Cointegration Test
|
Kao Cointegration Test
|
Hypothesized No. of CE(s)
|
Trace test)
|
Max-eigen test
|
t-Statistic
|
None
|
325.9***
|
73.52***
|
-2.901470***
|
At most 1
|
189.9***
|
106.7***
|
|
At most 2
|
228.6***
|
107.2***
|
|
At most 3
|
141.5***
|
68.58***
|
|
At most 4
|
86.98***
|
40.74***
|
|
At most 5
|
55.29***
|
32.74***
|
|
At most 6
|
33.34***
|
24.63***
|
|
At most 7
|
27.14***
|
27.14***
|
|
***significant at 1% level |
*significant at 10% level |
FMOLS and DOLS results
At this juncture, we have ascertained that the variables under study are integrated of the same order I(1), and they are cointegrated, indicating that they have a long-run relationship (Granger, 1986); Engle and Granger, 1987). In Table 5 below, we provide the FMOLS and DOLS results. In this model, for both FMOLS and DOLS, exchange rate is the explained variable, while balance of payment, inflation, RE, money supply, trade, GDP, and population size are the explanatory variables. The results of the FMOLS model in Table 5 below show that BOP, inflation rate, population size and money supply have a significant positive impact on the foreign exchange rate of the E7 nations. Therefore, when BOP, inflation rate, money supply and population size increase in these nations, then this will have the effect of causing a depreciation in the nation’s currencies. When the inflation rate of a nation rises, its currency will tend to depreciate; thus, our results concur with the Relative Purchasing Power Parity theory of Fisher (1930). The results also support the findings of Deka and Dube (2021) and Deka, Cavusoglu and Dube (2021), who alluded that a high inflation rate has the effect of depreciating the nation’s currency and vice versa. Additionally, an increase in money supply is seen to cause currency depreciation in the E7 nations; thus, it should be kept a minimum level as printing too much money is not healthy for the economy since it triggers demand pull inflation. Population size should also be kept at normal levels to ensure stability in the currency exchange, since overpopulation will result in a struggle for limited and scarce resources, leading to a crisis.
The results in Table 5 below also show that GDP, RE use and trade have a significant negative association. It follows that an increase in the Gross Domestic Product of the E7 economies will result in a decrease in their foreign exchange rate, indicating that economic growth in the E7 nations tends to encourage appreciation of the currency value. This is a good sign for the E7 nations as there is no dilemma faced in trying to achieve a high growth rate and stabilize the currency at the same time. RE use is also observed to have a significant negative effect on foreign exchange rate in the emerging economies, showing that increased use of RE has the effect of encouraging currency appreciation. These findings support the results of Deka and Dube (2021), and Deka, Cavusoglu and Dube (2021), who alluded that RE use enhances currency appreciation. Therefore, the use of RE should be encouraged in the emerging economies since it contributes towards the stabilization of the currency value. Trade is also seen to encourage appreciation of the currency exchange value in the E7 nations.
Table 5
|
FMOLS
|
DOLS
|
|
Coefficient
|
t-Statistic
|
Coefficient
|
t-Statistic
|
BOP
|
91.0124
|
6.9028***
|
-0.0458
|
-0.9033
|
IFL
|
18.9255
|
10.2581***
|
0.0705
|
6.0982***
|
LGDP
|
-1499.777
|
-12.7779***
|
-4.0602
|
-7.0136***
|
LPP
|
8242.744
|
13.5037***
|
27.8472
|
2.5143*
|
MS
|
7.3445
|
5.6770***
|
-1.4605
|
-0.0007
|
RE
|
-158.6387
|
-14.5778***
|
-0.0475
|
-2.3985*
|
TRD
|
-51.2979
|
-8.9636***
|
0.1056
|
3.8863**
|
R-squared
|
0.9089
|
|
0.998953
|
|
Adjusted R-squared
|
0.9010
|
|
0.994068
|
|
Long-run variance
|
85739.60
|
|
0.000193
|
|
Dependent variable: Exchange rate |
***significant at 1% level |
**significant at 5% level |
To verify the results of FMOLS in this study, we also employ the DOLS model. The results of the DOLS model are given in Table 5 above. The results of DOLS concur with the FMOLS results that inflation rate and population size in the emerging economies positively raises the exchange rate value. Therefore, low levels of inflation rate in these nations tend to encourage stability in the currency exchange value and low population size also brings stable foreign exchange values as opposed to overpopulated nations where people scramble for resources. The findings of the DOLS model in Table 5 above also concur with those of FMOLS that GDP and RE use has a strong negative association with currency exchange value. An increase in the use of RE will help to lower the currency exchange value in the E7 nations, hence causing the currency to appreciate (Deka and Dube, 2021; Deka, Cavusoglu and Dube, 2021). However, the findings of the DOLS model indicate that exchange rate does not have any significant association with BOP and money supply in the E7 economies and that trade rather has a significant positive association with foreign exchange rate.
Residual Diagnostic testing
To test for the validity, robustness and reliability of the findings provided above by the FMOLS and DOLS models, residual diagnostic testing is performed. We employ the Correlogram of residual test and Correlogram of residual squared test to check if the model’s residuals are free from Autocorrelation and Partial Autocorrelation problems. We also employ the Jarque-Bera Normality Test to check if the model’s residuals are normally distributed. In Table 6 below, the results shows that the FMOLS model suffers from Autocorrelation and Partial Autocorrelation problems. The statistics values are greater than the critical values and the p-values are less than 1% significant level showing that the model has Autocorrelation and Partial Autocorrelation problems (Gujarati, 2004; Deka & Resatoglu, 2019). Moreover, the residuals of the FMOLS model are not normally distributed. Thus, the results of the FMOLS model are not reliable and valid.
However, the t-Statistic residual diagnostic results in Table 6 below for Autocorrelation and Partial Autocorrelation are less than the critical value and the p-value is greater than 5% significant level showing that we can accept the null hypothesis that there are no Autocorrelation and Partial Autocorrelation problems in the DOLS model. Furthermore, the Jarque-Bera Normality Test results shows that we should accept the null hypothesis that the residuals are normally distributed. Thus, the DOLS model provides valid, reliable and robust results since it has no Autocorrelation problems and the residuals are normally distributed. Hence, when drawing conclusions and giving policy recommendations in this research, we use the results given by the DOLS model.
Table 6
Residual Diagnostics results
|
FMOLS
|
DOLS
|
Lag
|
Autocorrelation
|
Partial Correlation
|
Autocorrelation
|
Partial Correlation
|
Correlogram of Residuals
|
1
|
0.645***
|
0.645***
|
-0.356
|
-0.356
|
2
|
0.386***
|
-0.050***
|
-0.075
|
-0.231
|
3
|
0.151***
|
-0.134***
|
-0.142
|
-0.311
|
4
|
-0.115***
|
-0.254***
|
-0.244
|
-0.608
|
5
|
-0.125***
|
0.169***
|
0.531**
|
0.028**
|
6
|
-0.129***
|
-0.024***
|
-0.219**
|
-0.250**
|
Correlogram of Residuals Squared
|
1
|
0.816***
|
0.816***
|
-0.209
|
-0.209
|
2
|
0.539***
|
-0.383***
|
0.168
|
0.131
|
3
|
0.296***
|
0.001***
|
0.249
|
0.326
|
4
|
0.154***
|
0.089***
|
-0.292
|
-0.238
|
5
|
0.107***
|
0.063***
|
0.262
|
0.083
|
6
|
0.108***
|
0.010***
|
-0.321
|
-0.297
|
Normality Test
|
Jarque Bera
|
813.21***
|
|
0.1389
|
|
Dependent variable: Exchange rate |
***significant at 1% level |
*significant at 10% level |