Does Economic Policy Uncertainty Affect Foreign Remittances? Linear and Non-linear ARDL Approach in BRIC Economies

Figures & data

Table 1. Variables and Measures

Variables	Measure of variables	Source
Economic policy uncertainty	EPU index	www.policyuncertainty.com
Foreign remittances	Personal remittances received (current US$)	world development indicator (WDI)
Real effective exchange rate	Real effective exchange rate index (2010 = 100)	Federal Reserve bank ST. Louis (FRED)
Inflation	Consumer price index	International Financial Statistics (IFS)
Unemployment rate	Unemployment, total (% of the total labor force)	world development indicator (WDI)
Gross Domestic Product (GDP)	GDP per capita (current US$)	world development indicator (WDI)

Table 2. Descriptive Statistics

Variables	Mean	Std. Dev	Max	Min	skewness	kurtosis	Jarque-Bera	Obs
	Brazil
FR	21.64	0.31	22.02	20.94	−1.07	2.68	54.07(0.00)	276
EPU	4.85	0.52	6.52	3.10	0.05	3.17	0.47(0.79)	276
CPI	7.98	0.40	8.62	7.27	−0.17	1.89	15.29(0.00)	276
EX	78.12	15.57	109.92	42.03	−0.09	2.13	9.12(0.01)	276
GDP	8.82	0.49	9.49	7.95	−0.37	1.69	26.01(0.00)	276
UN	9.24	1.79	13.54	6.66	0.58	2.47	18.86(0.00)	276
	Russia
FR	21.83	1.12	23.06	19.24	−0.93	2.46	42.83(0.00)	276
EPU	4.71	0.76	6.67	2.52	−0.18	3.04	1.57(0.00)	276
CPI	4.35	0.69	5.25	2.42	−0.80	3.04	29.19(0.00)	276
EX	82.49	16.61	109.75	43.43	−0.39	2.34	11.85(0.00)	276
GDP	8.76	0.78	9.67	7.19	−0.65	1.89	33.47(00)	276
UN	7.35	2.40	13.26	4.50	1.16	3.42	64.25(0.00)	276
	India
FR	24.59	0.50	25.15	23.47	−0.84	2.21	21.06(0.00)	216
EPU	4.40	0.51	5.64	3.21	0.11	2.49	2.78(0.25)	216
CPI	4.51	0.37	5.16	3.91	−0.18	1.454	20.29(0.00)	216
EX	96.01	4.86	104.60	84.13	−0.15	2.04	9.08(0.01)	216
GDP	7.09	0.42	7.63	6.15	−0.62	2.02	19.36(0.00)	216
UN	5.54	0.32	7.76	5.27	4.78	28.68	67.62(0.00)	216
	China
FR	22.69	1.26	24.22	19.65	−0.75	2.52	28.87(0.00)	276
EPU	4.94	0.88	6.87	2.20	0.28	2.89	3.90(0.14)	276
CPI	4.34	0.16	4.62	4.12	0.11	1.52	25.73(0.00)	276
EX	104.18	14.74	130.94	81.83	0.30	1.66	24.81(0.00)	276
GDP	6.89	0.55	6.02	0.55	−0.24	1.57	26.18(0.00)	276
UN	4.29	0.48	4.96	3.23	−1.44	3.54	99.62(0.00)	276

The table includes mean, Standard deviation, minimum and maximum, skewness and kurtosis, and normality test (JB) values for all variables in the sample along with observations. FR denotes foreign remittances, EPU denotes economic policy uncertainty, CPI is a consumer price index, EX is the real exchange rate, GDP is per capita gross domestic product and UN is the unemployment rate.

Table 3. Correlation Analysis

Variables	EPU	CPI	EX	GDP	UN
Brazil
EPU	1.000
CPI	0.581	1.000
EX	−0.068	0.249	1.000
GDP	0.336	0.657	0.627	1.000
UN	0.246	0.239	−0.463	−0.261	1.000
Russia
EPU	1.000
CPI	0.696	1.000
EX	0.323	0.524	1.000
GDP	0.634	0.648	0.608	1.000
UN	−0.650	−0.728	−0.623	−0.758	1.000
India
EPU	1.000
CPI	0.174	1.000
EX	−0.143	0.443	1.000
GDP	0.246	0.609	0.507	1.000
UN	−0.040	0.003	0.128	−0.050	1.000
China
EPU	1.000
CPI	0.732	1.000
EX	0.687	0.689	1.000
GDP	0.674	0.678	0.624	1.000
UN	0.387	0.632	0.351	0.686	1.000

Table 4. Unit-root Tests

	Augmented Dickey-Fuller(ADF)				Phillips-Perron (PP)
	Brazil	Russia	India	China	Brazil	Russia	India	China
FR	−1.61	−2.02	−1.33	−2.00	−0.93	−4.62***	−2.51	−2.35
EPU	−4.05***	−3.89***	−2.68*	−2.17	−8.19***	−9.66***	−5.51***	−4.83***
CPI	−0.46	−4.83***	−0.24	−0.55	−0.42	−3.53***	−0.24	−0.35
EX	−2.34	−1.83	−0.91	−1.32	−2.15	−1.29	−1.01	−1.39
GDP	−1.15	−1.83	−2.32	−1.01	−0.86	−0.95	−3.84***	−0.03
UN	−2.29	−1.39	−3.07***	−2.36	−1.52	−1.38	−1.97	−1.96
	First differences
FR	−3.90***	−4.01***	−2.740*	−12.00***	−5.751***	−4.054***	−2.82*	−12.00***
EPU	−13.8***	−19.5***	−16.69***	−15.98***	−57.58***	−61.33***	−29.77***	−18.85***
CPI	−9.32***	−5.71***	−0.24***	−13.86***	−16.70***	−13.15***	−11.53***	−17.17***
EX	−12.8***	−11.10***	−11.49***	−9.15***	−12.78***	−10.72***	−11.38***	−9.17***
GDP	−2.61*	−2.98**	−2.47	−4.02***	−2.73*	−3.01**	−2.60*	−3.98***
UN	−3.49***	−14.3***	−3.61**	−7.05***	−14.26***	−14.37***	−3.74***	−7.14***

*, **, *** indicate the rejection of the null hypothesis at 10, 5, and 1 percentage levels, respectively.

Table 5. Linear ARDL (short-run) and diagnostic results

	Brazil	Russia	India	China
∆FR t-1	0.85(28.79)***	0.87(30.24)***	0.87(30.353)***	0.91(18.39)***
∆EPU t	6.04(0.11)	0.00(02.29)*	0.00(0.968)	−0.00(−1.54)
∆Ex t	−0.00(−1.13)		−7.31(−0.892)	0.00(0.51)
∆Ex t-1	−0.00(−3.47)***
∆UN t	−0.08(−11.65)***	0.01(1.73)*	0.00(0.528)	1.36(4.50)***
∆UN t-1	0.07(10.03)***			−0.95(−2.98)***
∆GDP t	−0.45(−8.97)***	0.39(5.48)***	0.01(3.833)***	8.79(23.51)***
∆GDP t-1	0.39(8.32)***	−0.27(−4.13)***		−7.67(−15.36)***
∆CPI t	0.04(0.54)		−0.01(−2.581)**	−0.15(−1.81)*
∆CPI t-1	0.14(1.97)*
F-stat.	2.76	4.23***	3.67**	2.85*
ECM t-1	−0.00(−4.44)***	−0.01(−5.50)***	−0.01(−5.14)***	−0.00(4.52)***
Adj. R^2	0.99	0.99	0.89	0.85
Ramsey RESET	[0.09]	[0.68]	[0.62]	[0.71]
LM test	[0.66]	[0.94]	[0.96]	[0.97]
Het test	[0.47]	[0.11]	[0.28]	[0.12]
critical values for F- test bounds
		10%	5%	1%
Lower		2.08	2.39	3.06
Upper		3	3.38	4.15

The F-statistic tests the null hypothesis of no cointegration $({\alpha }_4={\alpha }_5={\alpha }_6=0)$ against the alternative of cointegration $({\alpha }_4\ne {\alpha }_5\ne {\alpha }_6\ne 0)$. The ECMt-1 test is used to determine cointegration in the long run. The Adj.R² measures the fit of the model, while the LM test checks for Breusch-Godfrey serial correlation. The HET test examines for Breusch-Pagan-Godfrey heteroscedasticity, and the RESET test performs a Ramsey test for lack of functional specification. The ARDL model lags are selected using the Akaike Information Criterion (AIC). The figures in square brackets represent the probability, while the figures in parentheses are the absolute t-values. The symbols *, **, and *** indicate rejection of the null hypothesis at significant levels of 10%, 5%, and 1%, respectively.

Table 6. Linear ARDL (long-run) results

	Brazil	Russia	India	China
EPU	0.05(0.11)	0.17(1.85)*	0.05(1.05)	3.10(0.21)
EX	0.06(0.65)	0.003(0.39)	−0.007(−1.02)	−0.13(−0.16)
CPI	6.54(0.81)	0.72(3.02)***	−0.82(−2.24)**	−0.15(−1.81)*
GDP	−6.36(−0.72)	0.39(1.49)	1.79(5.76)***	0.06(2.51)**
UN	−0.34(−0.81)	−0.04(−0.63)	0.04(0.53)	−0.02(−1.92)
C	23.15(2.45)**	14.54(8.67)***	15.78(20.39)***	−20.36(−1.69)*

*, **, *** Specify a null hypothesis rejection at significant levels of 10, 5, and 1 percent, respectively.

Table 7. Nonlinear ARDL (short-run) and diagnostic results

	Brazil	Russia	India	China
∆FR t-1	0.85(28.53)***	0.88(30.27)***	0.87(30.47)***	0.87(17.57)***
∆EPU⁺	3.98(−0.07)	−0.00(−0.48)	0.04(0.45)**	−0.00(−1.25)
∆EPU⁻	0.00(0.186)	0.00(3.08)***	0.00(0.97)	−0.00(−0.28)
∆EX	−0.00(−1.14)		−9.79(−1.17)	0.00(0.87)
∆EX t-1	−0.00(−3.45)***
∆CPI	0.04(0.57)		−0.01(−1.51)	0.05(0.46)
∆CPI t-1	0.14(1.89)*
∆GDP t	−0.45(−8.95)***	0.39(5.51)***	0.03(3.91)***	8.79(23.92)***
∆GDP t-1	0.39(8.09)***	−0.29(−4.36)***		−7.19(−13.97)***
∆UN	−0.09(−11.63)***	0.01(1.65)	0.00(1.25)	1.37(4.59)***
∆UN t-1	−0.08(9.74)***			−0.93(−2.93)***
F-stat.	2.41	3.87**	3.50**	3.82**
Wsʀ	0.35[0.53]	11.37[0.00]***	342.24[0.00]***	0.73[0.48]
Wʟʀ	275.48[0.00]***	308.12[0.00]***	309.91[0.00]***	125.89[0.00]***
ECM t-1	−0.00(−4.45)***	−0.01(−5.64)***	−0.01(−5.38)***	−0.00(5.06)***
Adj. R^2	0.99	0.87	0.89	0.87
RESET	[0.04]	[0.69]	[0.96]	[0.85]
LM test	[0.65]	[0.94]	[0.87]	[0.75]
HET test	[0.55]	[0.11]	[0.29]	[0.19]
critical values for F-test bounds
		10%	5%	1%
Lower		1.99	2.27	2.88
Upper		2.94	3.28	3.99

The F-statistic tests the null hypothesis of no cointegration $({\rho }_0={\eta }^{+}={\eta }^{-}={\rho }_1=0)$.Wʟʀ and Wsʀ denote wald test of long-run symmetry (${\delta }^{+}={\delta }^{-}$) and short-run symmetry ($\sum _{j=0}^p{\gamma }_{2j}^{+}$=$\sum _{j=0}^n{\gamma }_{3j}^{-}$) respectively. ECM t-1 test is used to indicate cointegration in the long run. Adj. R² indicates whether the model fits well. LM denotes the Breusch-Godfrey serial correlation test. HET denotes the Breusch-Pagan-Godfrey heteroscedasticity test. RESET is a Ramsey test for lack of functional specification. The ARDL model lags are selected by AIC. The figures in square brackets denote the probability. The figures in parentheses are absolute t-values. *, **, *** Specify a null hypothesis rejection at significant levels of 10, 5, and 1 percent, respectively.

Figure 1. CUSUM Graphs.

Figure 2. CUSUMQ Graphs.

Figure 3. Dynamic Multipliers Graphs.

Table 8. Nonlinear ARDL (Long-run) results

	Brazil	Russia	India	China
EPU⁺	0.03(0.07)	0.11(0.93)	0.02(0.47)	1.35(0.39)
EPU⁻	0.07(0.18)	0.11(0.94)	0.05(1.04)	0.32(0.24)
EX	0.05(0.65)	0.00(0.13)	−0.01(−1.38)	−0.12(-(0.31)
CPI	7.31(0.88)	0.91(2.49)*	−0.46(−1.35)	1.65(1.97)*
GDP	−5.21(−0.68)	0.42(1.61)	1.98(6.62)***	−0.06(−0.44)
UN	−0.28(−0.72)	−0.04(−0.63)	0.09(1.32)	0.37(0.57)
C	8.56(0.15)	14.69(8.17)***	13.19(6.89)***	−18.97(−1.19)

*, **, *** Specify a null hypothesis rejection at significant levels of 10, 5, and 1 percent, respectively.