Asymmetric Effects on Risks of Virtual Financial Assets (VFAs) in different regimes: A Case of Bitcoin

Zhenghui Li; Hao Dong; Zhehao Huang; Pierre Failler; Zhenghui Li; Hao Dong; Zhehao Huang; Pierre Failler

doi:10.3934/QFE.2018.4.860

Quantitative Finance and Economics

2018, Volume 2, Issue 4: 860-883. doi: 10.3934/QFE.2018.4.860

Previous Article Next Article

Research article

Asymmetric Effects on Risks of Virtual Financial Assets (VFAs) in different regimes: A Case of Bitcoin

1.
Guangzhou International Institute of Finance, Guangzhou University, Guangzhou, P.R.China
2.
School of Economics and Statistics, Guangzhou University, Guangzhou, P.R.China
3.
Department of Economics and Finance, Portsmouth Business School, University of Portsmouth, Portsmouth, P01 3DE, UK

Received: 12 November 2018 Accepted: 20 November 2018 Published: 22 November 2018
JEL Codes: C32, C44, C51

The rapid development of VFAs allows investors to diversify their choices of investment products. In this paper, we measure the return risk of VFAs based on GARCH-type model. By establishing a Markov regime-switching Regression (MSR) Model, we explore the asymmetric effects of speculation, investor attention, and market interoperability on return risks in different risk regimes of VFAs. The results show that the influences of speculation and investor attention on the risks of VFAs are significantly positive at all regimes, while market interoperability only admits a positive impact on risk under high risk regime. All of the three factors exert asymmetric effects on risks in different regimes. Further study presents that the risk regime-switching also shows asymmetric characteristic but the medium risk regime is more stable than any others. Therefore, transactions of investors and arbitrageurs are monitored by certain policies, such as limiting the number of transactions or restricting the trading amount at high risk regime. However, when return risk is low, it will return to a medium level if we encourage investors to access.

Keywords:

Citation: Zhenghui Li, Hao Dong, Zhehao Huang, Pierre Failler. Asymmetric Effects on Risks of Virtual Financial Assets (VFAs) in different regimes: A Case of Bitcoin[J]. Quantitative Finance and Economics, 2018, 2(4): 860-883. doi: 10.3934/QFE.2018.4.860

Related Papers:

[1]	Thomas C. Chiang . Economic policy uncertainty and stock returns—evidence from the Japanese market. Quantitative Finance and Economics, 2020, 4(3): 430-458. doi: 10.3934/QFE.2020020
[2]	Saeed Sazzad Jeris, Ridoy Deb Nath . Covid-19, oil price and UK economic policy uncertainty: evidence from the ARDL approach. Quantitative Finance and Economics, 2020, 4(3): 503-514. doi: 10.3934/QFE.2020023
[3]	Sudeshna Ghosh . Asymmetric impact of COVID-19 induced uncertainty on inbound Chinese tourists in Australia: insights from nonlinear ARDL model. Quantitative Finance and Economics, 2020, 4(2): 343-364. doi: 10.3934/QFE.2020016
[4]	Md Akther Uddin, Mohammad Enamul Hoque, Md Hakim Ali . International economic policy uncertainty and stock market returns of Bangladesh: evidence from linear and nonlinear model. Quantitative Finance and Economics, 2020, 4(2): 236-251. doi: 10.3934/QFE.2020011
[5]	Gülserim Özcan . The amplification of the New Keynesian models and robust optimal monetary policy. Quantitative Finance and Economics, 2020, 4(1): 36-65. doi: 10.3934/QFE.2020003
[6]	Diogo Matos, Luís Pacheco, Júlio Lobão . Availability heuristic and reversals following large stock price changes: evidence from the FTSE 100. Quantitative Finance and Economics, 2022, 6(1): 54-82. doi: 10.3934/QFE.2022003
[7]	Simiso Msomi, Damien Kunjal . Industry-specific effects of economic policy uncertainty on stock market volatility: A GARCH-MIDAS approach. Quantitative Finance and Economics, 2024, 8(3): 532-545. doi: 10.3934/QFE.2024020
[8]	Chiao Yi Chang, Fu Shuen Shie, Shu Ling Yang . The relationship between herding behavior and firm size before and after the elimination of short-sale price restrictions. Quantitative Finance and Economics, 2019, 3(3): 526-549. doi: 10.3934/QFE.2019.3.526
[9]	Serçin ŞAHİN, Serkan ÇİÇEK . Interest rate pass-through in Turkey during the period of unconventional interest rate corridor. Quantitative Finance and Economics, 2018, 2(4): 837-859. doi: 10.3934/QFE.2018.4.837
[10]	Ongan Serdar, Gocer Ismet . Re-considering the Fisher equation for South Korea in the application of nonlinear and linear ARDL models. Quantitative Finance and Economics, 2019, 3(1): 75-87. doi: 10.3934/QFE.2019.1.75

Abstract

1. Introduction

Since introduction of the measure of policy uncertainty by Baker et al. (2016), the impact of policy uncertainty on macro variables have gained momentum. While Baker et al. (2016) demonstrated its adverse effects on economic activity, Wang et al. (2014) assessed the response of corporate investment to the new uncertainty measure. On the other hand, Pastor and Veronesi (2013), Ko and Lee (2015), and Brogaard and Detzel (2015) investigated response of risk premium and market returns to uncertainty. Other macro variables which are said to be affected by policy uncertainty include oil prices which was subject of an analysis by Kang and Ratti (2013) as well as Bahmani-Oskooee et al. (2018a). Additionally, while Bahmani-Oskooee et al. (2016) assessed the impact of policy uncertainty on the demand for money in the U.S., Bahmani-Oskooee and Ghodsi (2017) investigated the response of house prices in each state of the U.S.

The new policy uncertainty measure is now constructed on a monthly basis by the Policy Uncertainty Group for countries other than the U.S.¹ It is an index based on the volume of news associated with any uncertainty. Recently, Bahmani-Oskooee and Saha (2019) assessed the short- run and long-run effects of the policy uncertainty measure on stock returns in Canada, Japan, Korea, U.S., and U.K. These were the five countries for which a multivariate model could be estimated since monthly data were available for all the variables for these countries. Their main finding was that in all countries, increased uncertainty has adverse short-run effects but not long-run effects on stock prices. The main assumption included in Bahmani-Oskooee and Saha's (2019) study as well as other studies is that they have assumed that the effects of policy uncertainty on macro variables are symmetric. The symmetry assumption implies that if an x% increase in uncertainty hurts a variable by y%, an x% decrease in uncertainty will boost that variable by y%. However, this may not be the case since investors' reaction could be different to an increase in policy uncertainty as compared to a decline in uncertainty. If due to an x% increase in uncertainty investors shift their assets from stock to safer assets by y%, an x% decline in uncertainty may induce them to shift their portfolio into stock market by less than y%, if they expect decreased uncertainty to be short-lived, hence asymmetric effects. Similar arguments were recently made by Bahmani-Oskooee and Maki-Nayeri (2019) who assessed the asymmetric effects of policy uncertainty on domestic investment.

¹ For more details on constructing this index visit: www.policyuncertainty.com.

Therefore, the main purpose of this paper is to extend Bahmani-Oskooee and Saha's (2019) study from symmetric analysis to an asymmetric analysis of policy uncertainty on stock prices for the same five countries. To that end, we introduce the model and the methodology in Section Ⅱ that is followed by the results in Section Ⅲ. A summary is provided in Section Ⅳ and data definition and sources are cited in an Appendix.

2. The models and methodology

The model adopted here closely follows the model by Bahmani-Oskooee and Saha (2019), who followed the literature and used the following specification:²

² The same model is used by Boonyanam (2014) and Moore and Wang (2014) with no policy uncertainty measure. Other variants of the model without policy uncertainty measure is also used by Granger et al. (2000), Ansari and Kolari (2001), Nieh and Lee (2001), Phylaktis and Ravazzolo (2005), Yau and Nieh (2006), Pan et al. (2007), Richards et al. (2009), Liu and Tu (2011), Tsai (2012), Lin (2012), Tsagkanos et al. (2013), Groenewold et al. (2013), and Bahmani-Oskooee and Saha (2016).

$L n S P_{t} = a+b L n E X_{t}+c L n I P I_{t}+d L n C P I_{t}+e L n M_{t}+f L n P U_{t}+\varepsilon_{t}$

(1)

where SP denotes the stock prices, EX is the nominal effective exchange rate, IPI is a measure of output proxied by Index of Industrial Production (IPI), CPI is the Consumer Price Index as a measure of the price level, M is a measure of nominal money supply, and finally PU is the measure of policy uncertainty.

Specification (1) is a long run model and coefficient estimates are long-run estimates. Once the long run estimates are obtained, an estimate of b is expected to be negative or positive depending upon if firms associated with the specific stock are export or import oriented. Since an increase in economic activity measured by IPI is expected to increase the demand for stocks, an estimate of c is expected to be positive. An estimate of d is also expected to be negative or positive depending on whether inflation hurts firm costs and profit margin or if stocks are considered a hedge against inflation in the long run.³ An estimate of e is also expected to be positive or negative, depending whether increase in money supply leads to lower interest rates and economic growth or inflation. Finally, if increase in uncertainty is to hurt the markets, we expect an estimate of f to be negative.

³ See Fama (1981) for the negative effects and Anari and Kolari (2001) for the positive effects.

As mentioned above, specification (1) is a long-run model and in order to assess the short-run effects of right-hand side variables, we must incorporate the short-run dynamics. Bahmani-Oskooee and Saha (2019) followed Pesaran et al.'s (2001) ARDL (Auto Regressive Distributed Lag) bounds testing approach and considered the following error-correction model:

$\begin{array}{l} \Delta LnS{P_t} = \alpha + \sum\limits_{i = 1}^{n1} {{\beta _i}} \Delta LnS{P_{t - i}} + \sum\limits_{i = 0}^{n2} {{\delta _i}} \Delta LnE{X_{t - i}} + \sum\limits_{k = 0}^{n3} {{\varphi _i}} \Delta LnIP{I_{t - i}} + \sum\limits_{i = 0}^{n4} {{\theta _i}} \Delta LnCP{I_{t - i}}\\ + \sum\limits_{i = 0}^{n5} {{\pi _i}} \Delta Ln{M_{t - i}} + \sum\limits_{i = 0}^{n6} {{\omega _i}} \Delta LnP{U_{t - i}} + {\lambda _1}\ln S{P_{t - 1}} + {\lambda _2}LnE{X_{t - 1}} + \\ + {\lambda _3}LnIP{I_{t - 1}} + {\lambda _4}LnCP{I_{t - 1}} + {\lambda _5}Ln{M_{t - 1}} + {\lambda _6}LnP{U_{t - 1}} + {\mu _t} \end{array}$

(2)

In (2), the short-run effects of each variable is judged by the coefficient estimates attached to the first-differenced variables and the long-run effects by the estimates of λ₂–λ₆ normalized on λ_1.⁴ However, for the long-run estimates to be meaningful, cointegration must be established. Pesaran et al. (2001) propose two tests: the F test to establish joint significance of lagged level variables in (2) and the t-test to establish significance of λ₁ which must be negative.⁵ They demonstrate that the distribution of both tests are non-standard, hence they tabulate new critical values which also account for degree of integration of all the variables in a given model. Their upper bound critical values should be used to establish cointegration. These values are valid, even if some variables are I(0) and some I(1) and this is one of the main advantage of this approach since almost all macro variables are either I(0) or I(1).⁶

⁴ Note that once normalization is done we will have $\frac{{{{\hat \lambda }_2}}}{{ - {{\hat \lambda }_1}}} = \hat b, \frac{{{{\hat \lambda }_3}}}{{ - {{\hat \lambda }_1}}} = \hat c, \frac{{{{\hat \lambda }_4}}}{{ - {{\hat \lambda }_1}}} = \hat d, \frac{{{{\hat \lambda }_5}}}{{ - {{\hat \lambda }_1}}} = \hat e, {\rm{ }}and\; \; {\rm{ }}\frac{{{{\hat \lambda }_6}}}{{ - {{\hat \lambda }_1}}} = \hat f.$

⁵ Note that the estimate of λ₁ in (2) is the same as estimate of coefficient attached to lagged error-correction term in Engle and Granger (1987) error-correction specification. For empirical demonstration of this, see Bahmani-Oskooee (2019) and Bahmani-Oskooee and Ghodsi (2019).

⁶ Another advantage of this approach is that since short-run adjustment process is included in estimating the long-run effects, any feedback effects among variables are allowed to exert themselves which help reduce multicolinearity or endogeniety (Pesaran et al. 2001).

Shin et al. (2014) modified Pesaran et al.'s (2001) approach so that it could be used to assess asymmetric effects of any variable of concern. Since our concern is to assess asymmetric effects of policy uncertainty measure, PU, we take the following steps. First, we form ΔLnPU which includes positive as well as negative changes. Second, we separate increases in uncertainty from declines and form two new time-series variable as follows:

$PO{S_t} = \sum\limits_{j = 1}^t {\max } \left({\Delta LnP{U_j}, 0} \right), \quad NE{G_t} = \sum\limits_{j = 1}^t {\min } \left({\Delta LnP{U_j}, 0} \right)$

(3)

Where POS_t is the partial sum of the positive changes and reflects only increased uncertainty and NEG_t is the partial sum of negative changes in uncertainty and reflects only the decline in uncertainty. Next, we move back to the error-correction model (2) and replace the LnPU variable by the two partial sum variables to arrive at:

$\begin{array}{l} \Delta LnS{P_t} = \alpha + \sum\limits_{i = 1}^{n1} {{\beta _i}} \Delta LnS{P_{t - i}} + \sum\limits_{i = 0}^{n2} {{\delta _i}} \Delta LnE{X_{t - i}} + \sum\limits_{i = 0}^{n3} {{\varphi _i}} \Delta LnIP{I_{t - i}} + \sum\limits_{i = 0}^{n4} {{\theta _i}} \Delta LnCP{I_{t - i}}\\ + \sum\limits_{i = 0}^{n5} {{\pi _i}} \Delta Ln{M_{t - i}} + \sum\limits_{i = 0}^{n6} {\omega _i^ + } \Delta PO{S_{t - i}} + \sum\limits_{i = 0}^{n7} {\omega _i^ - } \Delta NE{G_{t - i}} + {\lambda _1}LnS{P_{t - 1}} + {\lambda _2}\ln E{X_{t - 1}} + \\ + {\lambda _3}LnIP{I_{t - 1}} + {\lambda _4}LnCP{I_{t - 1}} + {\lambda _5}Ln{M_{t - 1}} + \lambda _6^ + PO{S_{t - 1}} + \lambda _6^ - NE{G_{t - 1}} + {\mu _t} \end{array}$

(4)

Since the construction of the partial sum variables introduces nonlinear adjustment of policy uncertainty variable into our specification, Shin et al. (2014) refer models such as (4) as nonlinear ARDL model whereas, (2) is commonly referred to as the linear ARDL model. However, both models are subject to the same estimation, interpretation, and diagnostic tests.⁷ Once (4) is estimated, a few asymmetric assumptions could be tested. First, short-run effects of policy uncertainty will be asymmetric if at any given lag order, i, $\hat \omega _i^ + \ne \hat \omega _i^ -$ . However, if the Wald-test rejects the null hypothesis of $\sum {\hat \omega _i^ + } = \sum {\hat \omega _i^ - }$ , short-run cumulative or impact asymmetry will be established. Second, short-run adjustment process will be asymmetric if ΔPOS and ΔNEG take a different lag order, i.e., if n6 ≠ n7. Finally, long-run effects of policy uncertainty on stock prices will be asymmetric if we reject the null hypothesis of $\frac{{\hat \lambda _6^ + }}{{ - \hat \lambda _1^{}}} = \frac{{\hat \lambda _6^ - }}{{ - {{\hat \lambda }_1}}}$ . Again, the Wald test will be used for this purpose.⁸

⁷ Shin et al. (2014) even argue that the critical value of the F test should stay at the same high level when we move from (2) to (4) even if (4) has one more variable.

⁸ For some other application of these and other nonlinear models see Apergis and Miller (2006), Wimanda (2014), McFarlane et al. (2014), Gogas and Pragidis (2015), Durmaz (2015), Baghestani and Kherfi (2015), Al-Shayeb and Hatemi-J (2016), Lima et al. (2016), Nusair (2017), Aftab et al. (2017), Arize et al. (2017), Gregoriou (2017), Bahmani-Oskooee and Kanitpong (2018), Olaniyi (2019), and Istiak and Alam (2019).

3. The results

Since the linear model (2) has already been estimated by Bahmani-Oskooee and Saha (2019) for Canada, Japan, Korea, U.K., and the U.S. using monthly data, we restrict ourselves to estimating only the nonlinear model (4) and compare our findings to theirs. We have updated the data and used monthly data over the period January 1985–October 2018. Following their approach we impose a maximum of eight lags on each first-differenced variable and use Akaike Information Criterion to select the optimum number of lags. In any case, if there was evidence of serial correlation, following Bahmani-Oskooee and Fariditavana (2019) we added additional lags of dependent variable to reduce autocorrelation. Using required critical values reported in the notes to each table, we identify significant statistics by * at the 10% level and ** at the 5% level. The results are reported in five tables, one for each country.

From the short-run estimates in Panel A of each table, we gather that either ΔPOS or ΔNEG variable carry at least one significant coefficient in every optimum model, implying that changes in policy uncertainty has short-run effects on stock prices in all the five countries. This was also the case in the linear model. However, the new results reveal that except in the case of Japan, short-run effects are asymmetric since at the same lag order, the estimate attached to the ΔPOS variable is different than the one attached to the ΔNEG variable. Thus, in the four remaining countries, short-run effects of changes in policy uncertainty are asymmetric. However, sum of the coefficients attached to the ΔPOS variable is different than the sum attached to the ΔNEG variable only in the results for the U.S., since the Wald test reported as Wald-S in Panel C of each table is significant only in Table 5 for the U.S. In sum, while the short-run effects of policy uncertainty are asymmetric in four countries, evidence of cumulative or impact asymmetric effects is discovered only in one country (the U.S.).

Table 1. Full information estimate of the nonlinear ARDL model for Canada.

Panel A: short-run coefficient estimates
Lag order	0	1	2	3	4	5	6	7	8
ΔLnSP
ΔLnEX	0.50
ΔLnEX	(3.65)**
ΔLnIPI	0.21	0.47
ΔLnIPI	(0.93)	(2.17)**
ΔLnCPI	0.76	0.03	-0.67	-0.31	-1.56
ΔLnCPI	(1.30)	(0.06)	(1.15)	(0.53)	(2.72)**
ΔLnM	0.001
	(0.93)
ΔPOS	-0.06
ΔPOS	(5.30)**
ΔNEG	-0.007	-0.02	0.03
ΔNEG	(0.56)	(1.39)	(2.18)**
Panel B: long-run coefficient estimates
Constant	LnEX	LnIPI	LnCPI	LnM	POS	NEG
12.94	0.41	0.09	-1.84	0.01	-0.18	-0.25
(0.83)	(1.34)	(0.12)	(1.86)*	(0.03)	(2.27)**	(3.08)**
Panel C: Diagnostics
F	t-test	LM	RESET	${{\bar R}^2}$	CUSUM (CUSUMQ)	Wald-L	Wald-S
2.57	-0.08	0.02	12.15**	0.19	S (S)	3.77*	1.33
	(4.25)**
Notes: a. Numbers inside the parentheses are absolute value of t-ratios. , * indicate significance at the 10% and 5% levels respectively. b. The upper bound critical value of the F-test for cointegration when there are five exogenous variables is 3.35 (3.79) at the 10% (5%) level of significance. These come from Pesaran et al. (2001, Table CI, Case Ⅲ, p. 300). c. The critical value for significance of ECM_t-1 is -4.04 (-4.38) at the 10% (5%) level when k = 6. These come from Pesaran et al. (2001, Table CII, Case Ⅲ, p. 303). d. LM is the Lagrange Multiplier statistic to test for autocorrelation. It is distributed as χ² with 1 degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level. e. RESET is Ramsey's test for misspecification. It is distributed as χ² with one degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level. f. Both Wald tests are also distributed as χ² with one degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level.

| Show Table

DownLoad: CSV

Table 2. Full information estimate of the nonlinear ARDL model for Japan.

Panel A: short-run coefficient estimates
Lag order	0	1	2	3	4	5	6	7	8
ΔLnSP
ΔLnEX	-0.01
ΔLnEX	(0.39)
ΔLnIPI	0.04	0.36	-0.32
ΔLnIPI	(0.22)	(2.17)**	(1.88)*
ΔLnCPI	-0.65
ΔLnCPI	(2.62)**
ΔLnM	-2.71
	(1.73)*
ΔPOS	-0.002
ΔPOS	(2.01)**
ΔNEG	-0.002
ΔNEG	(2.47)**
Panel B: long-run coefficient estimates
Constant	LnEX	LnIPI	LnCPI	LnM	POS	NEG
66.48	-0.25	-1.00	-12.34	0.17	-0.03	-0.03
(0.92)	(0.41)	(0.63)	(1.98)**	(0.06)	(1.94)*	(2.79)**
Panel C: Diagnostics
F	t-test	LM	RESET	${{\bar R}^2}$	CUSUM (CUSUMQ)	Wald-L	Wald-S
2.98	-0.05	1.08	0.00	0.07	S (S)	0.34	1.93
	(4.48)**
Notes: a. Numbers inside the parentheses are absolute value of t-ratios. , * indicate significance at the 10% and 5% levels respectively. b. The upper bound critical value of the F-test for cointegration when there are five exogenous variables is 3.35 (3.79) at the 10% (5%) level of significance. These come from Pesaran et al. (2001, Table CI, Case Ⅲ, p. 300). c. The critical value for significance of ECM_t-1 is -4.04 (-4.38) at the 10% (5%) level when k = 6. These come from Pesaran et al. (2001, Table CII, Case Ⅲ, p. 303). d. LM is the Lagrange Multiplier statistic to test for autocorrelation. It is distributed as χ² with 1 degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level. e. RESET is Ramsey's test for misspecification. It is distributed as χ² with one degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level. f. Both Wald tests are also distributed as χ² with one degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level.

| Show Table

DownLoad: CSV

Table 3. Full information estimate of the nonlinear ARDL model for Korea.

Panel A: short-run coefficient estimates
Lag order	0	1	2	3	4	5	6	7	8
ΔLnSP	-	0.05	-0.10	0.15
ΔLnSP	(0.71)	(1.47)	(2.30)**
ΔLnEX	-0.42	-0.20	-0.06	-0.08	0.40
ΔLnEX	(1.94)*	(0.85)	(0.27)	(0.36)	(2.48)**
ΔLnIPI	0.42	0.41
ΔLnIPI	(1.95)**	(1.96)**
ΔLnCPI	-4.12	0.31	-1.07	-2.85	-0.88	-4.58	0.74	-3.39
ΔLnCPI	(3.24)**	(0.24)	(0.82)	(2.24)**	(0.70)	(3.87)**	(0.62)	(2.98)**
ΔLnM	-0.07
	(0.38)
ΔPOS	-0.04
ΔPOS	(2.05)**
ΔNEG	-0.0003
ΔNEG	(0.13)
Panel B: long-run coefficient estimates
Constant	LnEX	LnIPI	LnCPI	LnM	POS	NEG
25.32	0.36	-1.27	1.23	-0.61	0.06	-0.002
(0.54)	(0.73)	(1.36)	(0.34)	(0.38)	(2.47)**	(0.13)
Panel C: Diagnostics
F	t-test	LM	RESET	${{\bar R}^2}$	CUSUM (CUSUMQ)	Wald-L	Wald-S
6.84**	-0.13	2.36	0.24	0.31	S (U)	1.92	1.02
	(7.00)**
Notes: a. Numbers inside the parentheses are absolute value of t-ratios. , * indicate significance at the 10% and 5% levels respectively. b. The upper bound critical value of the F-test for cointegration when there are five exogenous variables is 3.35 (3.79) at the 10% (5%) level of significance. These come from Pesaran et al. (2001, Table CI, Case Ⅲ, p. 300). c. The critical value for significance of ECM_t-1 is -4.04 (-4.38) at the 10% (5%) level when k = 6. These come from Pesaran et al. (2001, Table CII, Case Ⅲ, p. 303). d. LM is the Lagrange Multiplier statistic to test for autocorrelation. It is distributed as χ² with 1 degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level. e. RESET is Ramsey's test for misspecification. It is distributed as χ² with one degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level. f. Both Wald tests are also distributed as χ² with one degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level.

| Show Table

DownLoad: CSV

Table 4. Full information estimate of the nonlinear ARDL model for the U.K.

Panel A: short-run coefficient estimates
Lag order	0	1	2	3	4	5	6	7	8
ΔLnSP
ΔLnEX	-1.19
ΔLnEX	(2.58)**
ΔLnIPI	1.99	1.67
ΔLnIPI	(2.91)**	(2.57)**
ΔLnCPI	0.23
ΔLnCPI	(0.35)
ΔLnM	-0.69	-0.67	-1.29	-1.50
	(1.10)	(1.00)	(1.94)*	(2.32)**
ΔPOS	-0.04
ΔPOS	(3.70)**
ΔNEG	-0.01
ΔNEG	(4.36)**
Panel B: long-run coefficient estimates
Constant	LnEX	LnIPI	LnCPI	LnM	POS	NEG
-83.52	-2.55	4.87	1.52	2.60	-0.25	-0.07
(4.10)**	(1.57)	(2.61)**	(0.36)	(3.82)**	(3.17)**	(4.45)**
Panel C: Diagnostics
F	t-test	LM	RESET	${{\bar R}^2}$	CUSUM (CUSUMQ)	Wald-L	Wald-S
3.69*	-0.14	0.31	1.73	0.18	S (S)	3.28*	0.13
	(5.05)**
Notes: a. Numbers inside the parentheses are absolute value of t-ratios. , * indicate significance at the 10% and 5% levels respectively. b. The upper bound critical value of the F-test for cointegration when there are five exogenous variables is 3.35 (3.79) at the 10% (5%) level of significance. These come from Pesaran et al. (2001, Table CI, Case Ⅲ, p. 300). c. The critical value for significance of ECM_t-1 is -4.04 (-4.38) at the 10% (5%) level when k = 6. These come from Pesaran et al. (2001, Table CII, Case Ⅲ, p. 303). d. LM is the Lagrange Multiplier statistic to test for autocorrelation. It is distributed as χ² with 1 degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level. e. RESET is Ramsey's test for misspecification. It is distributed as χ² with one degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level. f. Both Wald tests are also distributed as χ² with one degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level.

| Show Table

DownLoad: CSV

Table 5. Full Information Estimate of the Nonlinear ARDL Model for U.S.A.

Panel A: short-run coefficient estimates
Lag order	0	1	2	3	4	5	6	7	8
ΔLnSP	-	0.05	0.03	0.12
ΔLnSP	(1.00)	(0.59)	(2.32)**
ΔLnEX	-0.33
ΔLnEX	(2.28)**
ΔLnIPI	-0.93	0.75	0.61
ΔLnIPI	(2.49)**	(2.06)**	(1.61)
ΔLnCPI	0.25
ΔLnCPI	(1.80)*
ΔLnM	-0.79	-0.98
	(1.14)	(1.40)
ΔPOS	-0.04	-0.03	0.005	0.03
ΔPOS	(3.77)**	(2.21)**	(0.39)	(2.45)**
ΔNEG	-0.01	0.004	0.03
ΔNEG	(1.35)	(0.39)	(2.27)**
Panel B: long-run coefficient estimates
Constant	LnEX	LnIPI	LnCPI	LnM	POS	NEG
-23.17	-0.34	2.00	3.29	0.36	-0.05	-0.01
(1.18)	(0.65)	(2.92)**	(1.86)*	(0.63)**	(2.28)**	(0.94)
Panel C: Diagnostics
F	t-test	LM	RESET	${{\bar R}^2}$	CUSUM (CUSUMQ)	Wald-L	Wald-S
2.98	-0.06	0.29	12.03**	0.16	S (S)	0.67	4.83**
	(4.51)**
Notes: a. Numbers inside the parentheses are absolute value of t-ratios. , * indicate significance at the 10% and 5% levels respectively. b. The upper bound critical value of the F-test for cointegration when there are five exogenous variables is 3.35 (3.79) at the 10% (5%) level of significance. These come from Pesaran et al. (2001, Table CI, Case Ⅲ, p. 300). c. The critical value for significance of ECM_t-1 is -4.04 (-4.38) at the 10% (5%) level when k = 6. These come from Pesaran et al. (2001, Table CII, Case Ⅲ, p. 303). d. LM is the Lagrange Multiplier statistic to test for autocorrelation. It is distributed as χ² with 1 degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level. e. RESET is Ramsey's test for misspecification. It is distributed as χ² with one degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level. f. Both Wald tests are also distributed as χ² with one degree of freedom. The critical value is 2.70 (3.84) at the 10% (5%) significance level.

| Show Table

DownLoad: CSV

Short-run effects of policy uncertainty last into the long-run significant effects in all countries. In all the countries, either the POS variable or the NEG variable carry a significant coefficient that is reported in Panel B of each table. These long-run effects are meaningful since cointegration is supported in all five countries either by the F or by the t-test, reported in Panel C. This was not the case in Bahmani-Oskooee and Saha (2019), who failed to find significant long-run effects of policy uncertainty on stock returns in any of the five countries, except Canada. These new significant long-run effects must be attributed to the nonlinear adjustment of policy uncertainty. Furthermore, the long-run effects are asymmetric in the results for Canada and the U.K., since the Wald test reported as Wald-L in Panel C is significant in these two cases, rejecting the equality of normalized coefficients attached to the POS and NEG variables.

In each table we have also reported additional diagnostic statistics. To test for serial correlation, the Lagrange Multiplier (LM) statistic is reported in Panel C and since it is insignificant in all models, there is no evidence of serial correlation. Ramsey's RESET test is also reported to check for misspecification. It is insignificant in three models and significant in two models. We have also applied the CUSUM and CUSUMSQ tests to the residuals of each optimum model to establish stability of estimated short-run and long-run coefficients. The stable coefficients are indicated by "S" and the unstable ones by "U". As can be seen, all estimates are stable.

4. Summary and conclusion

In a recent study by the same authors they asked, "are these adverse effects of different types of uncertainty on stock prices transitory or do they have long run implication?". By using the policy uncertainty measure and a linear ARDL model, they concluded that indeed the effects are short run only.

In this paper, when we separated the increases in uncertainty from the declines and estimated a nonlinear multivariate model for the same five countries, we found that, short-run effects of policy uncertainty lasts into the long run which we attribute to the nonlinear adjustment of policy uncertainty. More precisely, we found that policy uncertainty has short-run asymmetric effects in Canada, Korea, U.K., and the U.S. but not in Japan. Short-run effects translated into significant and meaningful long-run effects in all five countries. However, the results are country-specific. While in the cases of Canada, Japan, and the U.K., we found that increased policy uncertainty hurts stock prices, decreased uncertainty boosts them, though at different rates. In the U.S. however, we found that while increased uncertainty hurts stock returns, decreased uncertainty does not have any long-run effects, a clear sign of long-run asymmetric effects. All in all, it appears that reducing uncertainty could benefit stock returns in the short run as well as in the long run.

Acknowledgements

Valuable comments of two anonymous referees are greatly appreciated. Remaining errors, however, are ours.

Conflict of interest

We, hereby, declare that the submitted paper is not associated with any kind of conflict of interest.

References

[1]	Andersen TG, Bollerslev T (1997) Intraday periodicity and volatility persistence in financial markets. J Empir Financ 4: 115–158. doi: 10.1016/S0927-5398(97)00004-2
[2]	Balcilar M, Elie B, Rangan G, et al. (2017). Can volume predict Bitcoin returns and volatility? A quantiles-based approach. Econ Model 64: 74–81.
[3]	Blau BM (2017) Price dynamics and speculative trading in bitcoin. Res in Int Business and Financ 41: 493–499. doi: 10.1016/j.ribaf.2017.05.010
[4]	Baur DG, Hong K, Lee AD (2017) Medium of exchange or speculative Assets? Working paper, SSRN.
[5]	Bollerslev T (1986) Generalized autoregressive conditional heteroscedasticity. J Econom 3:307–327.
[6]	Bouri E, Jalkh N, Molnar P, et al. (2017a) Bitcoin for energy commodities before and after the December 2013 crash: diversifier, hedge or safe haven? Applied Economic 49: 5063–5073.
[7]	Bouri E, Rangan G, Aviral KT, et al. (2017b) Does Bitcoin hedge global uncertainty? Evidence from wavelet-based quantile-in-quantile regressions. Financ Res Lett 23: 87–95.
[8]	Carrick J (2016) Bitcoin as a Complement to Emerging Market Currencies. Emerg Mark Financ Trade 52: 2321–2334. doi: 10.1080/1540496X.2016.1193002
[9]	Cabedoa JD, Moyab I (2003) Estimating oil price 'Value at Risk' using the historical simulation approach. Energy Econ 25: 239–253.
[10]	Chen Y (2018) Blockchain tokens and the potential democratization of entrepreneurship and innovation. Bus Horiz 61: 567–575. doi: 10.1016/j.bushor.2018.03.006
[11]	Cocco L, Marchesi M (2016) Modeling and Simulation of the Economics of Mining in the Bitcoin Market. PloS ONE 11: e0164603. doi: 10.1371/journal.pone.0164603
[12]	Davidson S, Filippi PD, Potts J (2018) Blockchains and the economic institutions of capitalism. J Inst Econ 14: 639–658.
[13]	De Bondt WFM, Thaler RH (1995) Financial decision-making in markets and firms: A behavioral perspective. Handbooks in operations res and management sci 9: 385–410. doi: 10.1016/S0927-0507(05)80057-X
[14]	Dyhrberg AH, Foley S, Svec J (2018) How investible is Bitcoin? Analyzing the liquidity and transaction costs of Bitcoin markets. Econ Lett 171: 140–143.
[15]	Feng WJ, Wang YM, Zhang ZJ (2018) Informed trading in the Bitcoin market. Financ Res Lett 26: 63–70. doi: 10.1016/j.frl.2017.11.009
[16]	Garcia D, Tessone C, Mavrodiev P, et al. (2014) The digital traces of bubbles: feedback cycles between socio-economic signals in the Bitcoin economy. J of the Royal Society Interface 11: 1–8.
[17]	Gkillas K, Katsiampa P (2018). An application of extreme value theory to cryptocurrencies. Econ Lett, 164: 109–111. doi: 10.1016/j.econlet.2018.01.020
[18]	Glaser F, Zimmarmann K, Haferhorn M, et al. (2014) Bitcoin-Asset or currency? Revealing users' hidden intentions. ECIS 2014, Tel Aviv.
[19]	Grinberg R (2012) Bitcoin: An Innovative Alternative Digital Currency. Hastings Sci & Technology Law J 4: 159–207.
[20]	Groshoff D (2014) Kickstarter My Heart: Extraordinary Popular Delusions and the Madness of Crowdfunding Constraints and Bitcoin Bubbles. William & Mary Business Law Rev 5: 489–557.
[21]	Gunter FR (2017) Corruption, costs, and family: Chinese capital flight, 1984–2014. China Econ Rev 43: 105–117. doi: 10.1016/j.chieco.2017.01.010
[22]	Hamilton JD (2008) Regime switching models. In: Durlauf, N., & Blume, L.E. (Eds.), The New Palgrave Dictionary of Economics, 2nd edition. London: Palgrave Macmillan.
[23]	Hayes AS (2016) Cryptocurrency value formation: an empirical study leading to a cost of production model for valuing bitcoin. Telematics and Informatics 34: 1308–1321.
[24]	Jacobs E (2011) Bitcoin: A Bit Too Far? J of IntBanking and Commerc 16: 1–4.
[25]	Kristoufek L (2015) What are the main drivers of the bitcoin price? evidence from wavelet coherence analysis. PLOS ONE 10: e0123923.
[26]	Kristoufek L (2013) Bitcoin meets Google trends and wikipedia: quantifying the relationship between phenomena of the Internet era. Scientific Reports 3:3415. doi: 10.1038/srep03415
[27]	Lan J, Timothy L, Tu Z (2016) Capital Flight and Bitcoin Regulation. Int Rev Financ 16: 445–455. doi: 10.1111/irfi.12072
[28]	Llorente G, Michaely R, Saar G, et al. (2002) Dynamic volume-Return relation of individual stocks. Rev Financ Stud 15: 1005–1047. doi: 10.1093/rfs/15.4.1005
[29]	Li Z, Chen S, Chen S (2017) Statistical Measure of Validity of Financial Resources Allocation. EURASIA J of Mathematics, Science and Technology Education 13: 7731–7741.
[30]	Lukáš P, Taisei K (2017) Volatility Analysis of Bitcoin Price Time Series. Quantitative Financ and Econ 1: 474–485. doi: 10.3934/QFE.2017.4.474
[31]	Luther WJ (2016) Bitcoin and the Future of Digital Payments. The Independent Rev 20: 397–404.
[32]	Mertzanis C (2018) Complexity, big data and financial stability. Quantitative Financ and Econ 2: 637–660. doi: 10.3934/QFE.2018.3.637
[33]	Murad Z, Sefton M, Starmer C (2016) How do risk attitudes affect measured confidence? J Risk Uncertain 52: 21–46. doi: 10.1007/s11166-016-9231-1
[34]	Nelson DB (1991) Conditional Heteroskedasticity in Asset Returns: A New Approach. Econometrica 59: 347–370. doi: 10.2307/2938260
[35]	Osterrieder J, Lorenz J (2017) A statistical risk assessment of Bitcoin and its extreme tail behavior. Annals of Financ Econ 12: UNSP 1750003.
[36]	Pilkington M (2013) Bitcoin and Complexity Theory: Some Methodological Implications. Working Paper, University of Burgundy, France.
[37]	Plassaras NA (2013) Regulating Digital Currencies: Bringing Bitcoin within the Reach of the IMF. Chicago J of Int Law 14: 377–407.
[38]	Rockafellar RT, Uryasev S (2002) Conditional value-at-risk for general loss distributions. J of Banking & Financ 26: 1443–1471.
[39]	Sapuric S, Kokkinaki A (2014) Bitcoin Is Volatile! Isn't that Right?. In: Abramowicz W., Kokkinaki A. (eds) Business Information Systems Workshops. BIS 2014. Lecture Notes in Business Information Processing, vol 183. Springer, Cham.
[40]	Sornette D, Cauwels P, Smilyanov G (2018) Can we use volatility to diagnose financial bubbles? lessons from 40 historical bubbles. Quantitative Financ and Econ 2: 1–105.
[41]	Yermack D (2013) Is Bitcoin A Real Currency? An Economic Appraisal. Working paper, National Bureau of Economic Research.
[42]	Yermack D (2015) Bitcoin, innovation, financial instruments, and big data. Handbook of Digital Currency 31–43.

This article has been cited by:

1.	Zhifeng Dai, Huiting Zhou, Fenghua Wen, Shaoyi He, Efficient predictability of stock return volatility: The role of stock market implied volatility, 2020, 52, 10629408, 101174, 10.1016/j.najef.2020.101174
2.	Md Akther Uddin, Mohammad Enamul Hoque, Md Hakim Ali, International economic policy uncertainty and stock market returns of Bangladesh: evidence from linear and nonlinear model, 2020, 4, 2573-0134, 236, 10.3934/QFE.2020011
3.	Afşin ŞAHİN, Mabruka MOHAMED, Cementing the Asymmetric Confluence of Exchange Rate and Stock Prices: NARDL Modeling for Kuwait, 2020, 1308-5549, 10.18074/ckuiibfd.636490
4.	Tinghui Li, Junhao Zhong, Hai Zhang, Pierre Failler, Chinese financial cycle spillovers to developed countries, 2019, 1, 2643-1092, 364, 10.3934/GF.2019.4.364
5.	Sudeshna Ghosh, Asymmetric impact of COVID-19 induced uncertainty on inbound Chinese tourists in Australia: insights from nonlinear ARDL model, 2020, 4, 2573-0134, 343, 10.3934/QFE.2020016
6.	Jieqiong Yang, Panzhu Luo, Yong Tan, Contingent Decision of Corporate Environmental Responsibility Based on Uncertain Economic Policy, 2020, 12, 2071-1050, 8839, 10.3390/su12218839
7.	Gizem Uzuner, Sudeshna Ghosh, Do pandemics have an asymmetric effect on tourism in Italy?, 2020, 0033-5177, 10.1007/s11135-020-01074-7
8.	Yue Liu, Pierre Failler, Jiaying Peng, Yuhang Zheng, Time-Varying Relationship between Crude Oil Price and Exchange Rate in the Context of Structural Breaks, 2020, 13, 1996-1073, 2395, 10.3390/en13092395
9.	Muhammad Asif Khan, Masood Ahmed, József Popp, Judit Oláh, US Policy Uncertainty and Stock Market Nexus Revisited through Dynamic ARDL Simulation and Threshold Modelling, 2020, 8, 2227-7390, 2073, 10.3390/math8112073
10.	Fenghua Wen, Cui Li, Han Sha, Liuguo Shao, How does economic policy uncertainty affect corporate risk-taking? Evidence from China, 2020, 15446123, 101840, 10.1016/j.frl.2020.101840
11.	Jiaying Peng, Zhenghui Li, Benjamin M. Drakeford, Dynamic Characteristics of Crude Oil Price Fluctuation—From the Perspective of Crude Oil Price Influence Mechanism, 2020, 13, 1996-1073, 4465, 10.3390/en13174465
12.	Prince Mensah Osei, Reginald Djimatey, Anokye M. Adam, Dehua Shen, Economic Policy Uncertainty Linkages among Asian Countries: Evidence from Threshold Cointegration Approach, 2021, 2021, 1607-887X, 1, 10.1155/2021/6656176
13.	Liming Chen, Ziqing Du, Yong Tan, Sustainable exchange rates in China: Is there the heterogeneous effect of economic policy uncertainty?, 2019, 1, 2643-1092, 346, 10.3934/GF.2019.4.346
14.	Rong Li, Sufang Li, Di Yuan, Keming Yu, Does economic policy uncertainty in the U.S. influence stock markets in China and India? Time-frequency evidence, 2020, 52, 0003-6846, 4300, 10.1080/00036846.2020.1734182
15.	Yanhong Feng, Dilong Xu, Pierre Failler, Tinghui Li, Research on the Time-Varying Impact of Economic Policy Uncertainty on Crude Oil Price Fluctuation, 2020, 12, 2071-1050, 6523, 10.3390/su12166523
16.	Liuguo Shao, Hua Zhang, Jinyu Chen, Xuehong Zhu, Effect of oil price uncertainty on clean energy metal stocks in China: Evidence from a nonparametric causality-in-quantiles approach, 2021, 73, 10590560, 407, 10.1016/j.iref.2021.01.009
17.	Emmanuel Asafo-Adjei, Daniel Agyapong, Samuel Kwaku Agyei, Siaw Frimpong, Reginald Djimatey, Anokye M. Adam, Junhai Ma, Economic Policy Uncertainty and Stock Returns of Africa: A Wavelet Coherence Analysis, 2020, 2020, 1607-887X, 1, 10.1155/2020/8846507
18.	Tangyong Liu, Xu Gong, Lizhi Tang, The uncertainty spillovers of China's economic policy: Evidence from time and frequency domains, 2020, 1076-9307, 10.1002/ijfe.2385
19.	Jian Liu, Ziting Zhang, Lizhao Yan, Fenghua Wen, Forecasting the volatility of EUA futures with economic policy uncertainty using the GARCH-MIDAS model, 2021, 7, 2199-4730, 10.1186/s40854-021-00292-8
20.	Djula Borozan, Bartol Borozan, The asymmetric effect of economic policy uncertainty on energy consumption, 2022, 15, 1570-646X, 10.1007/s12053-022-10037-w
21.	Sang Hoon Kang, Jose Arreola Hernandez, Mobeen Ur Rehman, Syed Jawad Hussain Shahzad, Seong-Min Yoon, Spillovers and hedging between US equity sectors and gold, oil, islamic stocks and implied volatilities, 2023, 81, 03014207, 103286, 10.1016/j.resourpol.2022.103286
22.	Abdel Razzaq Al Rababa’a, Mohammad Alomari, Mobeen Ur Rehman, David McMillan, Raed Hendawi, Multiscale relationship between economic policy uncertainty and sectoral returns: Implications for portfolio management, 2022, 61, 02755319, 101664, 10.1016/j.ribaf.2022.101664
23.	Meiyu Tian, Wanyang Li, Fenghua Wen, The dynamic impact of oil price shocks on the stock market and the USD/RMB exchange rate: Evidence from implied volatility indices, 2021, 55, 10629408, 101310, 10.1016/j.najef.2020.101310
24.	Huiming Zhu, Hao Wu, Yinghua Ren, Dongwei Yu, Time-frequency effect of investor sentiment, economic policy uncertainty, and crude oil on international stock markets: evidence from wavelet quantile analysis, 2022, 54, 0003-6846, 6116, 10.1080/00036846.2022.2057912
25.	Wen-Yi Chen, Mei-Ping Chen, Twitter’s daily happiness sentiment, economic policy uncertainty, and stock index fluctuations, 2022, 62, 10629408, 101784, 10.1016/j.najef.2022.101784
26.	Kuashuai Peng, Guofeng Yan, A survey on deep learning for financial risk prediction, 2021, 5, 2573-0134, 716, 10.3934/QFE.2021032
27.	Salah A. Nusair, Jamal A. Al-Khasawneh, Impact of economic policy uncertainty on the stock markets of the G7 Countries:A nonlinear ARDL approach, 2022, 26, 17034949, e00251, 10.1016/j.jeca.2022.e00251
28.	Salah A. Nusair, Jamal A. Al-Khasawneh, Changes in oil price and economic policy uncertainty and the G7 stock returns: evidence from asymmetric quantile regression analysis, 2023, 1573-9414, 10.1007/s10644-023-09494-9
29.	Mohammed Armah, Ahmed Bossman, Godfred Amewu, Information flow between global financial market stress and African equity markets: An EEMD-based transfer entropy analysis, 2023, 9, 24058440, e13899, 10.1016/j.heliyon.2023.e13899
30.	Asiye Tutuncu, Burcu Savas Celik, Sukran Kahveci, Do the Political Uncertainty and Geopolitical Risk Indexes in the G-7 Countries Relate to Stock Prices? Fourier Causality Test Evidence, 2024, 1016-8737, 1, 10.1080/10168737.2024.2408483
31.	Lydia N. Kotur, Goodness C. Aye, Peter I. Ater, 2024, 9780443157851, 10.1016/B978-0-44-313776-1.00283-X
32.	Masudul Hasan Adil, Amrita Roy, Asymmetric effects of uncertainty on investment: Empirical evidence from India, 2024, 29, 17034949, e00359, 10.1016/j.jeca.2024.e00359
33.	Jonathan E. Ogbuabor, Oliver E. Ogbonna, Onyinye I. Anthony-Orji, Davidmac O. Ekeocha, Obed I. Ojonta, Symmetric or Asymmetric: How is Economic Growth Responding to Global Economic Uncertainty in Africa's Oil Exporters?, 2023, 32, 12100455, 446, 10.18267/j.pep.836
34.	Lydia N. Kotur, Goodness C. Aye, Josephine B. Ayoola, Asymmetric Effects of Economic Policy Uncertainty on Food Security in Nigeria, 2024, 17, 1911-8074, 114, 10.3390/jrfm17030114
35.	RAHEEL GOHAR, MOHAMED OSMAN, EMMANUEL UCHE, P. A. MARY AUXILIA, BISHARAT HUSSAIN CHANG, THE ECONOMIC POLICY UNCERTAINTY EXTREME DYNAMICS AND ITS EFFECT ON THE EXCHANGE RATE, 2022, 22, 2194-5659, 10.1142/S2194565923500069
36.	Chien-Chiang Lee, Xueli Wen, How Does Exchange Rate Policy Uncertainty Affect Corporate Performance: Evidence from China, 2023, 59, 1540-496X, 3060, 10.1080/1540496X.2023.2205545
37.	Segun Kamoru FAKUNMOJU , ECONOMIC POLICY UNCERTAINTY AND STOCK MARKET DEVELOPMENT IN NIGERIA, 2024, 29, 2353-0758, 29, 10.7862/rz.2024.mmr.18
38.	Burak Gülmez, GA-Attention-Fuzzy-Stock-Net: An optimized neuro-fuzzy system for stock market price prediction with genetic algorithm and attention mechanism, 2025, 11, 24058440, e42393, 10.1016/j.heliyon.2025.e42393

Reader Comments

Your name:*

Email:*
© 2018 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)