|
Eduvest – Journal of
Universal Studies Volume 4 Number 10, October,
2024 p- ISSN 2775-3735- e-ISSN 2775-3727 |
||
|
|
|
|
|
WORLDWIDE RECESSIONS
AND HERDING BEHAVIOUR: A COMPARATIVE ANALYSIS OF THREE COUNTRIES |
|
|
|
Nurtantyo Pratomo Suyadi1,
Zaäfri Ananto Husodo2 1,2Faculty
of Economics and Business, Universitas Indonesia, Indonesia Email:
[email protected]1,
[email protected]2 |
|
|
|
ABSTRACT |
||
|
The
general suggestion that behavioral science plays a part in creating
abnormalities within the financial sector has been studied and proposed many
times in the past. This study aims to prove the existence of behavioral
sciences, specifically herding behavior, in three countries with different
market conditions: Indonesia (Emerging), Vietnam (Frontier), and the United
States (Developed). We developed our methodology using quantile regression to
study the existence of herding behavior, and our findings were as follows:
(1) As expected, the US didn’t have any indication of a statistically
significant herding presence; they do, however, indicate an insignificant
presence of herding behavior in the post-covid period under bearish
conditions (2) Vietnam does not indicate significant herding tendencies, (3)
Surprisingly, Indonesia did not exhibit statistically significant herding
presence, but both Indonesia and Vietnam exhibited the slight presence of
herding behavior but still relatively insignificant. |
||
|
KEYWORDS |
COVID-19, Herding Behaviour, Emerging Markets, Frontier Markets, Developed
Markets. |
|
|
|
This work is licensed under a Creative Commons
Attribution-ShareAlike 4.0 International |
|
INTRODUCTION
Behavioral science is crucial for understanding
financial market abnormalities, especially during crisis periods such as
pandemics. Concepts like prospect theory illuminate how individuals make
decisions under risk and uncertainty, revealing that investor behavior is often
swayed by emotional factors and cognitive biases rather than pure rationality
A notable example of herding behavior can be
observed in emerging markets like Indonesia, particularly during times of
economic uncertainty. For instance, the 2008 subprime mortgage crisis, while
primarily rooted in the U.S. housing market, highlights how herding can emerge
in different contexts
Before the pandemic, Indonesia's economy was
thriving, with its GDP rising to 1.119 trillion USD in 2019. However, the
pandemic caused a drop in GDP to 1.059 trillion USD, although inflation
remained low
According to the International Monetary Fund
(IMF), herd behavior in financial markets has several potential causes:
imperfect information, concern for reputation, and compensation structures. IMF
also mentioned that there are two types of herd behavior: true (intentional)
and spurious (unintentional). There are differences in effectivity between
those two, where intentional herding tends to be inefficient and is usually
characterized by fragility and idiosyncrasy. Influencers would usually try to
influence unassuming, inexperienced traders to induce “herd buying” to try and
increase the value of a fundamentally invaluable stock. The act of herd buying
or selling has historically started large, unfounded market rallies
According to FTSE Russel’s research, frontier
markets are defined as markets that represent developing countries with high
rates of economic growth but relatively illiquid stock markets. MSCI identified
29 different nations as frontier markets, one of which is Vietnam. An article
by VN Express International mentioned that in 2022, Vietnam’s retail stock
investors reached 4.93 million and will only grow larger in the future.
Vietnam’s stock market started to boom in 2020 after the plunge due to
COVID-19. An article by Nguyen et al.
Investing is an ever-growing trend in Vietnam;
Vietnam has two stock exchanges: the Hanoi Stock Exchange (HNX) AND the Ho Chi
Minh Stock Exchange (HOSE). At the end of 2020, the number of investor accounts
was just over 4.5 million accounts. This was an astonishing number, considering
the population of Vietnam is just over 98 million. In April 2023, the number
increased marginally to over 7 million, approximately 7% of the total
population. The market capitalization has also increased marginally from 5,416
trillion VND at the end of April 2023 to 6 quadrillion VND at the end of 2023,
which is equivalent to about 62% of the total GDP. The increase in investors over
the years and the increase in market capitalization indicate that Vietnam has
what it takes to be considered a frontier market.
Vietnam’s investors consist mostly of individual
household investors, with few institutional investors as a minority. Based on
the research provided by CFA Community Vietnam (2020), the securities assets
held by domestic investors, primarily individual investors, make up some
portions of the overall stock market, which indicates that retail investors
rather than institutional investors largely drive the market. This leads to
unstable transactions, and since many of the individual investors are still
unaware of legal conditions and have very limited understanding and knowledge,
they often act according to hearsay. Also, since the stock market in Vietnam is
still in its early stages of development, the psychological aspects are still a
contributing element to stock movements. False rumors and inflationary pressure
affect investors' decision-making in these markets tenfold compared to
developed markets. Buying and selling shares according to rumors and
misinformation is what ultimately leads to the formation of herd instinct, as
with low individual knowledge and confidence comes imitation
With over 158 million investors and a market
capitalization of $51.47 trillion, the U.S. stock market remains highly
productive, even after participation fell following the 2008 recession. Despite
being a developed economy, herding behavior still occurs, particularly during
volatile periods like the financial crisis. This is driven by macroeconomic
data and uncertainty, as observed in both institutional and individual
investors. In contrast, Indonesia (emerging) and Vietnam (frontier) show
similar tendencies, with Indonesia's strong GDP and Vietnam's rapid growth
making them appealing to investors.
This study compares herding behavior in
Indonesia, Vietnam, and the U.S. during the COVID-19 pandemic, examining
whether market efficiency minimizes herding in developed markets, as proposed
by Eugene Fama's efficient market hypothesis (EMH). Data will focus on each
country's top 100 performing stocks, selected for their high liquidity and
large market capitalization. These factors are associated with stability,
making the stocks attractive for investors while allowing us to observe herding
tendencies across different market classifications.
RESEARCH METHOD
Research
Design
This research employs a
quantitative exploratory design to compare herding behavior across frontier,
emerging, and developed markets. It aims to challenge traditional models like
CAPM and EMH. Variables and methodology are based on Nguyen et al.
Methodological
Framework
Data
Analysis Method
The study by Nguyen et al. (2023) uses two main
steps: a stationarity test, descriptive statistics, and quantile regression,
with descriptive statistics as a preliminary step. Descriptive statistics
summarize the relationship between variables across three time periods:
pre-COVID, post-COVID, and the full COVID period, each divided into bearish and
bullish markets with varying trading volumes. Quantile regression, a robust
alternative to OLS, is used to address distributional tails, offering a more
comprehensive analysis of the impact of variables on CSAD. It is particularly
effective in analyzing herding behavior under different market conditions, with
results indicating herding if certain parameters are significantly negative.
RESULT AND DISCUSSION
The Effects of the COVID-19 Pandemic Towards Herd
Presence in Different Market Conditions
During
COVID-19 Under Bullish Conditions
The results for the DUR-COV period under bullish conditions are similar to those from the period in default conditions. The
coefficients are highly positive and, therefore, do not exhibit the existence
of herding behavior within this period during bullish market conditions. These
results align with our assumption that herding will not be present in positive
and stable markets. However, this contradicts the research by Vidya et al.
|
Index |
Quantile |
Absolute MR |
(t-stat) |
Squared MR |
(t-stat) |
Pseudo R |
||
|
|
|
Coeff |
Std. Error |
|
Coeff |
Std. Error |
|
|
|
Kompas100 |
Q10 |
0.111 |
0.0530 |
(2.096) |
6.6766 |
0.6902 |
(9.802) |
0.244 |
|
|
Q25 |
0.206 |
0.0653 |
(3.152) |
5.744 |
0.8492 |
(6.764) |
0.261 |
|
|
Q50 |
0.250 |
0.0758 |
(3.302) |
5.127 |
0.9864 |
(5.198) |
0.308 |
|
|
Q75 |
0.567 |
0.0945 |
(6.002) |
1.838 |
1.2303 |
(1.494) |
0.397 |
|
|
Q90 |
0.526 |
0.1786 |
(2.946) |
1.918 |
2.3242 |
(0.825) |
0.507 |
|
VN-100 |
Q10 |
0.075 |
0.0907 |
(0.825) |
11.746 |
2.4586 |
(4.778) |
0.183 |
|
|
Q25 |
0.056 |
0.0965 |
(0.577) |
11.409 |
2.6148 |
(4.363) |
0.175 |
|
|
Q50 |
-0.267 |
0.1761 |
(-1.519) |
22.707 |
4.7717 |
(4.759) |
0.168 |
|
|
Q75 |
-0.047 |
0.4496 |
(-0.104) |
14.954 |
12.1828 |
(1.227) |
0.154 |
|
|
Q90 |
-1.117 |
0.6566 |
(-1.702) |
66.986 |
17.7917 |
(3.765) |
0.151 |
|
DUR-COV |
Q10 |
0.006 |
0.0593 |
(0.108) |
4.111 |
0.9210 |
(4.464) |
0.127 |
|
|
Q25 |
0.020 |
0.0604 |
(0.339) |
3.645 |
0.9389 |
(3.882) |
0.130 |
|
|
Q50 |
0.027 |
0.0702 |
(0.389) |
4.736 |
1.0907 |
(4.343) |
0.185 |
|
|
Q75 |
0.66 |
0.0846 |
(0.783) |
5.711 |
1.3143 |
(4.345) |
0.266 |
|
|
Q90 |
0.181 |
0.2127 |
(0.853) |
3.521 |
3.3038 |
(1.066) |
0.362 |
During
COVID-19 Under Bearish Conditions
The negative coefficients in the DUR-COV periods started appearing under bearish conditions with negative market returns. The negative coefficients could especially be seen in Vietnam’s VN-100 index at lower quantiles (10%; 25%) and the US Nasdaq100 index at higher quantiles (75%; 90%). Indonesia’s Kompas100 index, however, stayed relatively stable with no exhibits of negative coefficients across all quantiles under bearish conditions. The negative numbers in VN-100, considering it is located at lower quantiles, mean that the dispersions between its CSAD and MR are greater with lower CSAD values than its middle to higher CSAD values. It’s a different case with the US’ Nasdaq100 index as this index indicates negative coefficients at higher quantiles, this means that the dispersions are much more pronounced than those in lower quantiles.
Table
2. Regression Output of Kompas100, VN-100, and
NASDAQ100 Index during the Covid-19 Pandemic Under Bearish Condition
|
Index |
Quantile |
Absolute MR |
(t-stat) |
Squared MR |
(t-stat) |
Pseudo R |
||
|
|
|
Coeff |
Std. Error |
|
Coeff |
Std. Error |
|
|
|
Kompas100 |
Q10 |
0.126 |
0.1036 |
(1.213) |
10.162 |
2.2742 |
(4.468) |
0.266 |
|
|
Q25 |
0.093 |
0.0917 |
(1.015) |
10.162 |
2.0131 |
(5.075) |
0.260 |
|
|
Q50 |
0.374 |
0.1579 |
(2.367) |
5.416 |
3.4666 |
(1.562) |
0.280 |
|
|
Q75 |
0.137 |
0.2474 |
(0.554) |
13.671 |
5.4325 |
(2.517) |
0.294 |
|
|
Q90 |
0.214 |
0.4401 |
(0.487) |
14.529 |
9.6645 |
(1.503) |
0.378 |
|
VN-100 |
Q10 |
0.532 |
0.1426 |
(3.733) |
-0.375 |
2.7608 |
(-0.136) |
0.258 |
|
|
Q25 |
0.727 |
0.1617 |
(4.496) |
-3.398 |
3.1308 |
(-1.085) |
0.329 |
|
|
Q50 |
0.281 |
0.1690 |
(1.664) |
8.670 |
3.2715 |
(2.650) |
0.404 |
|
|
Q75 |
0.230 |
0.2532 |
(0.909) |
12.026 |
4.9018 |
(2.453) |
0.473 |
|
|
Q90 |
0.224 |
0.5177 |
(0.433) |
10.866 |
10.0242 |
(1.084) |
0.520 |
|
DUR-COV |
Q10 |
-0.050 |
0.0619 |
(-0.813) |
1.745 |
0.6562 |
(2.660) |
0.140 |
|
|
Q25 |
-0.080 |
0.0487 |
(-1.646) |
1.435 |
0.5162 |
(2.779) |
0.147 |
|
|
Q50 |
-0.191 |
0.0626 |
(-3.055) |
0.448 |
0.6636 |
(0.675) |
0.181 |
|
|
Q75 |
-0.250 |
0.1265 |
(-1.977) |
-0.230 |
1.3415 |
(-0.172) |
0.188 |
|
|
Q90 |
-0.478 |
0.3120 |
(-1.532) |
-2.298 |
3.3078 |
(-0.695) |
0.168 |
Late
COVID-19 Under Bullish Conditions
Higher negative γ2 coefficients start to appear in these results, specifically the regression results of the LATE-COV period under Bullish conditions. It is especially apparent in Indonesia’s Kompas100 and Vietnam’s VN-100 indexes. Indonesia’s Kompas100 index indicated negative γ2 coefficients at most quantiles (10%, 50%, 75, 90%), while Vietnam’s VN-100 index exhibited negative γ2 coefficients across all quantiles (10%, 25%, 50%, 75%; 90%). In contrast, The US Nasdaq100 index indicated different results with highly positive γ_2 coefficients across the board.
The results indicate that positive market returns within recovery periods actually manifest a lot more dispersion between CSAD and MR in developing and frontier markets. The higher γ2 coefficients observed in Indonesia’s Kompas100 and Vietnam’s VN-100 indices may imply that the investors investing in these indices are more prone to herding tendencies under bullish conditions in recovery periods. Conversely, the US Nasdaq100 index exhibits no herding tendencies at all, maintaining a high degree of market efficiency during recovery periods. This could be attributed to the more liquid nature of the US Nasdaq100 index, allowing less information asymmetry and independent decision-making among investors.
Table
3. Regression Output of Kompas100, VN-100, and
NASDAQ100 Index during late Covid-19 Pandemic Under Bullish Condition
|
Index |
Quantile |
Absolute MR |
(t-stat) |
Squared MR |
(t-stat) |
Pseudo R |
||
|
|
|
Coeff |
Std. Error |
|
Coeff |
Std. Error |
|
|
|
Kompas100 |
Q10 |
0.381 |
0.2558 |
(1.489) |
-13.480 |
13.3937 |
(-1.006) |
0.015 |
|
|
Q25 |
0.047 |
0.2900 |
(0.162) |
9.333 |
15.1822 |
(0.615) |
0.013 |
|
|
Q50 |
0.501 |
0.3721 |
(1.346) |
-14.710 |
19.4803 |
(-0.755) |
0.025 |
|
|
Q75 |
0.551 |
0.6014 |
(0.917) |
-20.069 |
31.4858 |
(-0.637) |
0.018 |
|
|
Q90 |
1.356 |
1.0170 |
(1.333) |
-58.808 |
53.2457 |
(-1.104) |
0.042 |
|
VN-100 |
Q10 |
0.389 |
0.1298 |
(2.997) |
-1.204 |
3.4339 |
(-0.351) |
0.062 |
|
|
Q25 |
0.260 |
0.1628 |
(1.595) |
-0.052 |
4.3048 |
(-0.012) |
0.054 |
|
|
Q50 |
1.070 |
0.4008 |
(2.670) |
-15.845 |
10.6011 |
(-1.495) |
0.063 |
|
VN-100 |
Q75 |
0.880 |
0.7767 |
(1.133) |
-16.549 |
20.5429 |
(-0.806) |
0.031 |
|
|
Q90 |
1.572 |
1.2822 |
(1.226) |
-25.452 |
33.9120 |
(-0.751) |
0.026 |
|
NASDAQ100 |
Q10 |
0.45 |
0.0988 |
(0.456) |
7.256 |
2.9379 |
(2.470) |
0.194 |
|
|
Q25 |
-0.31 |
0.1015 |
(-0.304) |
9.845 |
3.0179 |
(3.262) |
0.203 |
|
|
Q50 |
-0.37 |
0.1253 |
(-0.297) |
10.661 |
3.7262 |
(2.861) |
0.222 |
|
|
Q75 |
-0.237 |
0.1414 |
(-1.674) |
16.782 |
4.2051 |
(3.991) |
0.277 |
|
|
Q90 |
-0.303 |
0.2649 |
(-1.146) |
18.441 |
7.8783 |
(2.337) |
0.347 |
Late COVID-19 Under Bearish Conditions
The LATE-COV period under bearish conditions yielded opposite results where Indonesia’s Kompas100 and Vietnam’s VN-100 indices do not exhibit any negative coefficients across all quantiles while the US’ Nasdaq100 exhibited negative coefficients at all of its quantiles (10%; 25%; 50%; 75%; 90%). The results suggest a significant difference in the reaction of investors within developing, frontier, and developed markets. Considering the negative market return condition, it is speculated that the negative numbers in the US Nasdaq100 are caused by higher market anxiety and risk aversion among investors.
Table
4. Regression Output of Kompas100,
VN-100, and NASDAQ100 Index during late Covid-19 Pandemic Under Bullish
Condition
|
Index |
Quantile |
Absolute MR |
(t-stat) |
Squared MR |
(t-stat) |
Pseudo R |
|||
|
|
|
Coeff |
Std. Error |
|
Coeff |
Std. Error |
|
|
|
|
Kompas100 |
Q10 |
-0.023 |
0.1865 |
(-0.122) |
14.049 |
5.7700 |
(2.435) |
0.064 |
|
|
|
Q25 |
0.359 |
0.1650 |
(2.175) |
5.272 |
5.1040 |
(1.033) |
0.125 |
|
|
Kompas100 |
Q50 |
0.355 |
0.2206 |
(-1.608) |
3.752 |
6.8242 |
(0.550) |
0.112 |
|
|
|
Q75 |
0.311 |
0.3491 |
(0.890) |
5.056 |
10.8018 |
(0.468) |
0.126 |
|
|
|
Q90 |
0.152 |
0.5024 |
(0.302) |
3.778 |
15.5447 |
(0.243) |
0.106 |
|
|
VN-100 |
Q10 |
-0.324 |
0.1187 |
(-2.727) |
23.042 |
2.6287 |
(8.765) |
0.156 |
|
|
|
Q25 |
-0.269 |
0.2457 |
(-1.093) |
20.231 |
5.4440 |
(3.716) |
0.159 |
|
|
|
Q50 |
-0.601 |
0.3761 |
(-1.599) |
28.986 |
8.3325 |
(3.479) |
0.150 |
|
|
|
Q75 |
-0.89 |
0.4092 |
(-0.217) |
24.241 |
9.0647 |
(2.674) |
0.235 |
|
|
|
Q90 |
0.432 |
0.5361 |
(0.806) |
13.058 |
11.8761 |
(1.100) |
0.393 |
|
|
NASDAQ100 |
Q10 |
-0.219 |
0.0885 |
(-2.468) |
-0.499 |
2.2325 |
(-0.224) |
0.209 |
|
|
|
Q25 |
-0.355 |
0.919 |
(-3.856) |
-2.920 |
2.3184 |
(-1.259) |
0.276 |
|
|
|
Q50 |
-0.326 |
0.0895 |
(-3.648) |
-1.539 |
2.2558 |
(-0.682) |
0.282 |
|
|
|
Q75 |
-0.460 |
0.1367 |
(-3.365) |
-2.077 |
3.4477 |
(-0.602) |
0.316 |
|
|
|
Q90 |
-0.399 |
0.2377 |
(-1.680) |
-0.648 |
5.9929 |
(-0.108) |
0.286 |
|
Overall, in the case of Indonesia’s Kompas100, Vietnam’s VN-100, and the US Nasdaq100, the existence of herding behavior within said indices is still unproven. However, we can see hints of negative (t-stat) coefficients in different periods and market conditions. The LATE-COV period is the most notable period, which indicated negative γ2 coefficients in bullish and bearish conditions. Granted, within bullish conditions, the indices indicating negative γ2 coefficients are Indonesia’s Kompas100 and Vietnam’s VN-100, while the US’ Nasdaq100 only indicated negative γ2 coefficients in bearish conditions.
The presence of negative γ2 coefficients in specific conditions and periods aligns with the notion that herding tendencies might appear to be more pronounced in certain market phases, particularly when investors are faced with optimism and/or uncertainty. However, considering the lack of statistical significance within the regression results, the findings are closer to Eugene Fama's Efficient Market Hypothesis (1970), which assumes all market actors behave rationally and independently.
CONCLUSION
This study examines herding behavior in three markets during the COVID-19 pandemic: Indonesia, Vietnam, and the United States, representing developing, frontier, and developed markets, respectively. Analyzing daily stock closing prices from 2019 to 2023, the research found no statistically significant evidence of herding behavior in any of the markets during the specified periods, despite some negative coefficients that did not meet the threshold for significance. While these findings align with previous studies in Indonesia, they contradict reports of herding in Vietnam, potentially due to differences in sample size and observation periods. The limitations of this study include a sample restricted to the 100 largest stocks, which may not capture broader market dynamics, and a relatively short time frame that could affect the robustness of the conclusions. Consequently, while the study accepts the null hypothesis regarding the absence of significant herding behavior during COVID-19, further research with larger and more diverse samples is warranted to fully understand herding dynamics in various market conditions.
REFERENCES
Adnan, M. (2023). Modeling Herding Behavior in the
Indonesian Capital Market. International Journal of Economics, Business and
Management Research, 07(04), 167–179.
https://doi.org/10.51505/IJEBMR.2023.7413
Bennett, D., Mekelburg, E., & Williams, T. H. (2023). BeFi meets DeFi: A behavioral finance approach to decentralized finance asset pricing. Research in International Business and Finance, 65, 101939. https://doi.org/10.1016/j.ribaf.2023.101939
Chang, C.-L., McAleer, M., & Wang, Y.-A. (2020). Herding behaviour in energy stock markets during the Global Financial Crisis, SARS, and ongoing COVID-19*. Renewable and Sustainable Energy Reviews, 134, 110349. https://doi.org/10.1016/j.rser.2020.110349
Chen, Z., & Zheng, H. (2022). Herding in the Chinese and US stock markets: Evidence from a micro-founded approach. International Review of Economics & Finance, 78, 597–604. https://doi.org/10.1016/j.iref.2021.11.015
Choijil, E., Méndez, C. E., Wong, W.-K., Vieito, J. P., & Batmunkh, M.-U. (2022). Thirty years of herd behavior in financial markets: A bibliometric analysis. Research in International Business and Finance, 59, 101506. https://doi.org/10.1016/j.ribaf.2021.101506
Fama, E. F. (1970). Efficient Capital Markets: A Review of Theory and Empirical Work. The Journal of Finance, 25(2), 383. https://doi.org/10.2307/2325486
Kuramoto, Y., Khan, M. S. R., & Kadoya, Y. (2024). Behavioral Biases in Panic Selling: Exploring the Role of Framing during the COVID-19 Market Crisis. Risks, 12(10), 162.
Mihajlovic, S., Nikolic, D., Santric-Milicevic, M., Milicic, B., Rovcanin, M., Acimovic, A., & Lackovic, M. (2022). Four Waves of the COVID-19 Pandemic: Comparison of Clinical and Pregnancy Outcomes. Viruses, 14(12), 2648. https://doi.org/10.3390/v14122648
Mishra, P. K., & Mishra, S. K. (2023). Do Banking and Financial Services Sectors Show Herding Behaviour in Indian Stock Market Amid COVID-19 Pandemic? Insights from Quantile Regression Approach. Millennial Asia, 14(1), 54–84. https://doi.org/10.1177/09763996211032356
Mubarok, F., & Fadhli, M. M. (2020). Efficient Market Hypothesis and Forecasting of the Industrial Sector on the Indonesia Stock Exchange. Journal of Economics, Business, & Accountancy Ventura, 23(2), 160–168. https://doi.org/10.14414/jebav.v23i2.2240
Muth, J. F. (1961). Rational Expectations and the Theory of Price Movements. Econometrica, 29(3), 315. https://doi.org/10.2307/1909635
Nguyen, H. M., Bakry, W., & Vuong, T. H. G. (2023). COVID-19 pandemic and herd behavior: Evidence from a frontier market. Journal of Behavioral and Experimental Finance, 38, 100807. https://doi.org/10.1016/j.jbef.2023.100807
Sánchez-Granero, M. A., Balladares, K. A., Ramos-Requena, J. P., & Trinidad-Segovia, J. E. (2020). Testing the efficient market hypothesis in Latin American stock markets. Physica A: Statistical Mechanics and Its Applications, 540, 123082. https://doi.org/10.1016/j.physa.2019.123082
Swandari Budiarso, N., & Pontoh, W. (2022). Market efficiency and global issues: A case of Indonesia. Investment Management and Financial Innovations, 19(4), 1–13. https://doi.org/10.21511/imfi.19(4).2022.01
Vidya, C. T., Ravichandran, R., & Deorukhkar, A. (2023). Exploring the effect of Covid-19 on herding in Asian financial markets. MethodsX, 10, 101961. https://doi.org/10.1016/j.mex.2022.101961