Date of submission: August 17, 2021; date of acceptance: September 3, 2021. * Contact information: nagerisuccess2000@yahoo.co.uk, Department of Banking and Finance, Al-Hikmah University, Ilorin, Kwara State, Nigeria, phone: +2348056172296; ORCID ID: https://orcid.org/0000-0002-5569-2169. Copernican Journal of Finance & Accounting e-ISSN 2300-3065 p-ISSN 2300-12402021, volume 10, issue 4 Nageri, K.I. (2021). Risk-Return Relationship in the Nigerian Stock Market During Pandemic COVID-19: Sectoral Panel GARCH Approach. Copernican Journal of Finance & Accounting, 10(4), 97–116. http://dx.doi.org/10.12775/CJFA.2021.017 kaMaldeen ibraheeM nageri* Al-Hikmah University risk-return relationship in the nigerian stock Market during pandeMic covid-19: sectoral panel garch approach Keywords: COVID-19, risk-return, news, GARCH. J E L Classification: D81. Abstract: This study examines how the Nigerian Stock Exchange (NSE) is responding to the COVID-19 pandemic in the form of risk-return relationship and volatility. Panel data analyses of GARCH-in-mean and Threshold GARCH were estimated on three error distributional assumptions. All Share Index (ASI) from January 2020 to December 2020 for ten stock market indices on the NSE. Findings indicate that the cross-section return of the ten stock market indices returns exhibit a positive risk-return relationship du- ring COVID-19 and the impact of bad news was found to have no significant impact on return volatility on the NSE. This indicates that the policy response during the pande- mic is adequate to cushion the negative impact of COVID-19, which should be sustained.  Introduction Risk-return has a positive correlation whereas investors are risk-averse is the postulation of the fundamental studies of finance irrespective of analysis con- Kamaldeen Ibraheem Nageri98 ducted at industry, firm and national level (Mahmood & Shah, 2015). On the contrary, behavioural finance explained that the risk-return relationship is sensitive to the investor’s target or situation. According to prospect theory, in- vestors exhibit a risk-averse attitude in the gain domain and exhibit a risk-seek- ing attitude in the loss domain, calculated relative to a reference point. This im- plies that the risk-return relationship is negatively correlated. Ghysels, Plazzi and Valkanov (2016) find fundamental change in the traditional risk-return re- lationship during financial crises and separation between the traditional risk- return relationship and financial crises due to f light to safety (when investors sell assets perceived to be highly risky to invest in safer assets like gold). Various studies (Fisher & Hall, 1969; Neumann, Bobel & Haid, 1979; Glos- ten, Jagannathan & Runkle, 1993; Brandt & Kang, 2004; Guo & Whitelaw, 2006; Ludvigson & Ng, 2007) had been conducted at firm, industry and national lev- els and the controversy between risk-return relationship lingers on in the lit- erature. Corona Virus Disease (COVID) first identified in December 2019 in Wuhan, China, generally known as COVID-19, has spread to over 200 countries world- wide and on every continent. It is an acute respiratory disease, caused by a novel coronavirus (previously known as SARS-CoV-2, and 2019-nCoV) named COVID-19. COVID-19 is the third highly pathogenic and large-scale epidemic coronavirus after the Severe Acute Respiratory Syndrome Coronavirus (SARS-CoV) of 2002 and the Middle East Respiratory Syndrome Coronavirus (MERS-CoV) of 2012 in the human popula- tion in the twenty-first century (Hu, Guo, Zhou & Shi, 2020; Guo, Cao, Hong, Tan, Chen, Jin, Tan, Wang & Yan, 2020). World Health Organisation (WHO) declared COVID-19 as a pandemic in March 2020 and became the first coronavirus to be characterised as such. At the time of the declaration, there were 118,000 cases in 114 countries and 4,291 deaths but at the end of June 2020, there were over 10.5 million cases reported in 210 countries, over 514,000 deaths and more than 5.9 million infected people have recovered. When the virus was declared a global pandemic, stock markets globally ex- perienced a cumulative loss of 12.35% and over $9 trillion loss between Janu- ary and May 2020 and the global market exhibit increased volatility. For ex- ample, the stock prices in the United States declined by 32%, in the United Kingdom declined by 27.9%, and in some emerging stock markets like Brazil reduced by 40.5%, Russia 24.2% and China by 10.1% (Salisu, Ebuh & Usman, 2020). There exist different sentiments and analysts have divergence opinions risk-return relAtionshiP in the nigeriAn stock mArket… 99 on the impact of COVID-19, the fall in stock prices was attributed to investors’ panic, as many investors sold out of fear. Another view is that COVID-19 could cause the re-emergence of another global financial crisis and analysts also un- derstand that COVID-19 impact could be worse than the combination of Severe Acute Respiratory Syndrome (SARS) outbreak of 2003, the global financial cri- sis and World War II (International Monetary Fund (IMF), 2020a; International Monetary Fund (IMF), 2020b; Khan, Zhao, Zhang, Yang, Shah & Jahanger, 2020). Therefore, as a result of the foregoing problem, it becomes imperative to examine how the NSE is responding to the COVID-19 pandemic in the form of risk-return relationship and volatility (responding to good and bad news). The findings of the study will add to existing literature especially on the study of COVID-19 and stock market performance in the presence of policy response and implications. The subsequent parts of this paper contain a literature re- view in section two. Section three presents the methodology. Results and dis- cussion are presented in section four and section five provides the conclusion and recommendations of the study. Literature Review This section appraises relevant pieces of literature including the theoretical background and the empirical reviews. Theoretical Background The relationship between risk and return is one fundamental concept and prac- tice in the field of financial economics. The risk-return relationship indicates that an investment with higher risk should attract a higher expected return proportional to the risk-free return. Merton (1973) posits that the expected ex- cess equity market return is positively related to its conditional variance: markets like Brazil reduced by 40.5%, Russia 24.2% and China by 10.1% (Salisu, Ebuh & Usman, 2020). There exist different sentiments and analysts have divergence opinions on the impact of COVID-19, the fall in stock prices was attributed to investors’ panic, as many investors sold out of fear. Another view is that COVID-19 could cause the re-emergence of another global financial crisis and analysts also understand that COVID-19 impact could be worse than the combination of Severe Acute Respiratory Syndrome (SARS) outbreak of 2003, the global financial crisis and World War II (International Monetary Fund (IMF), 2020a; International Monetary Fund (IMF), 2020b; Khan, Zhao, Zhang, Yang, Shah & Jahanger, 2020). Therefore, as a result of the foregoing problem, it becomes imperative to examine how the NSE is responding to the COVID-19 pandemic in the form of risk-return relationship and volatility (responding to good and bad news). The findings of the study will add to existing literature especially on the study of COVID-19 and stock market performance in the presence of policy response and implications. The subsequent parts of this paper contain a literature review in section two. Section three presents the methodology. Results and discussion are presented in section four and section five provides the conclusion and recommendations of the study. Literature Review This section appraises relevant pieces of literature including the theoretical background and the empirical reviews. Theoretical Background The relationship between risk and return is one fundamental concept and practice in the field of financial economics. The risk-return relationship indicates that an investment with higher risk should attract a higher expected return proportional to the risk-free return. Merton (1973) posits that the expected excess equity market return is positively related to its conditional variance: E(R�) = R� + β�(E(R�) − R�) (1) Where E(R�) represents the expected return on the asset; R� represents the risk-free rate such as coupons from government bonds; β� represents the sensitivity of the expected excess return on the asset to the expected excess market returns, can be estimated through, β� = (1) Where E(Ri) represents the expected return on the asset; Rf represents the risk-free rate such as coupons from government bonds; βi represents the sen- sitivity of the expected excess return on the asset to the expected excess mar- ket returns, can be estimated through, β� = ��� ( ��� ��)��� ( �� ) ; E(R�) represents the expected return of the market; E(R�) − R� represents the market premium. When the rate of returns is independent and equally dispersed, the expected risk-return relationship is positive given the risk aversion of investors. When returns are not independent and equally dispersed, the risk-return relationship will include additional terms to recognise the hedging behaviour of investors (Merton, 1973). Empirical viewpoint has been found in both positive and negative relationships between return and risk (Guo & Whitelaw, 2006; Leon, Nave & Rubio, 2007; Lettau & Ludvigson, 2003; Nelson, 1991; Raputsoane, 2009). Empirical Review The risk-return relationship on stock returns has been a major area of research and studies had been conducted such as Nageri (2019a), Coffie (2015), Alade, Adeusi and Alade (2020), Ashraf (2020), Salisu and Vo (2020), Azimli (2020), among others. Ashraf (2020) examined the stock markets’ response to the COVID-19 pandemic in 64 countries between the period January 22, 2020, to April 17, 2020, employed panel data regression analysis technique and the finding suggests a strong negative market reaction throughout the early days of confirmed cases of COVID-19 and later between 40 to 60 days after the initial confirmed cases of COVID-19. Salisu and Vo (2020) evaluate the relevance of health news on the predictability of stock returns, using a panel data regression model with data of 20 top-worst-hit countries in terms of reported COVID-19 cases and deaths. The result shows that health news is significant when predicting stock return during COVID-19 pandemic. Azimli (2020) examines the impact of COVID-19 on risk-return dependence in the United States of America. The quantile regression result indicates that sectoral returns have an asymmetric dependence structure with the market portfolio. Alade et al. (2020) investigates the connection between COVID-19 confirmed cases and Nigerian stock market capitalization used the Vector regression model and finds that confirmed cases of COVID-19 have a mixed association, which is a negative but statistically insignificant relationship to the Nigerian stock market equity capitalization. Nageri (2019a) evaluates positive and negative news on the Nigerian stock market before and after the 2018/19 financial meltdown, employed GARCH variant models of TGARCH, EGARCH and ; E(Rm) represents the ex- Kamaldeen Ibraheem Nageri100 pected return of the market; E(Rm) – Rf represents the market premium. When the rate of returns is independent and equally dispersed, the expected risk- return relationship is positive given the risk aversion of investors. When re- turns are not independent and equally dispersed, the risk-return relationship will include additional terms to recognise the hedging behaviour of investors (Merton, 1973). Empirical viewpoint has been found in both positive and nega- tive relationships between return and risk (Guo & Whitelaw, 2006; Leon, Nave & Rubio, 2007; Lettau & Ludvigson, 2003; Nelson, 1991; Raputsoane, 2009). Empirical Review The risk-return relationship on stock returns has been a major area of research and studies had been conducted such as Nageri (2019a), Coffie (2015), Alade, Adeusi and Alade (2020), Ashraf (2020), Salisu and Vo (2020), Azimli (2020), among others. Ashraf (2020) examined the stock markets’ response to the COVID-19 pandemic in 64 countries between the period January 22, 2020, to April 17, 2020, employed panel data regression analysis technique and the find- ing suggests a strong negative market reaction throughout the early days of confirmed cases of COVID-19 and later between 40 to 60 days after the initial confirmed cases of COVID-19. Salisu and Vo (2020) evaluate the relevance of health news on the predicta- bility of stock returns, using a panel data regression model with data of 20 top- worst-hit countries in terms of reported COVID-19 cases and deaths. The re- sult shows that health news is significant when predicting stock return during COVID-19 pandemic. Azimli (2020) examines the impact of COVID-19 on risk- return dependence in the United States of America. The quantile regression re- sult indicates that sectoral returns have an asymmetric dependence structure with the market portfolio. Alade et al. (2020) investigates the connection between COVID-19 con- firmed cases and Nigerian stock market capitalization used the Vector re- gression model and finds that confirmed cases of COVID-19 have a mixed as- sociation, which is a negative but statistically insignificant relationship to the Nigerian stock market equity capitalization. Nageri (2019a) evaluates positive and negative news on the Nigerian stock market before and after the 2018/19 financial meltdown, employed GARCH variant models of TGARCH, EGARCH and PGARCH. The study reveals that good news inf luence stock return more signifi- risk-return relAtionshiP in the nigeriAn stock mArket… 101 cantly than bad news of similar extent before the meltdown, but bad news have a more insignificant impact on stock return than the good of similar extent af- ter the meltdown. Zhang, Hu and Ji (2020) studied the financial markets of the top 10 coun- tries on the list of confirmed COVID-19 pandemic cases as of 27 march, 2020 (the United States, Italy, China, Spain, Germany, France, the United Kingdom, Switzerland, South Korea and the Netherlands). The study used the quantile regression methodology for the daily S&P500 (SP) returns and findings reveal global financial market risks significantly increased during the COVID-19 pan- demic period of the study. Therefore, from the pieces of literature reviewed, it is obvious that a result of the uncertainties due to COVID-19 in the stock market globally, making the stock market as a major point of call indicating the effect of COVID-19 on econo- mies. As such, it is an important contribution to the literature and the economic rebound of the Nigerian economy to embark on a study of the risk-return rela- tionship of stock on the NSE to provide an initial estimate for policy direction and guide concerning past policies and future decisions. Data and Methodology The discussion in this section includes the method that was used for the analy- sis, mainly the theoretical framework, model specifications, sources and esti- mation procedure. Data The data are secondary, sourced from the Nigerian Stock Exchange using the daily selected sectoral All Share Index return of different sectors (see table 1) from January 1st till June 30th, 2020. The return series was generated by the fol- lowing for 123 observations for each sector: PGARCH. The study reveals that good news influence stock return more significantly than bad news of similar extent before the meltdown, but bad news have a more insignificant impact on stock return than the good of similar extent after the meltdown. Zhang, Hu and Ji (2020) studied the financial markets of the top 10 countries on the list of confirmed COVID-19 pandemic cases as of 27 march, 2020 (the United States, Italy, China, Spain, Germany, France, the United Kingdom, Switzerland, South Korea and the Netherlands). The study used the quantile regression methodology for the daily S&P500 (SP) returns and findings reveal global financial market risks significantly increased during the COVID-19 pandemic period of the study. Therefore, from the pieces of literature reviewed, it is obvious that a result of the uncertainties due to COVID-19 in the stock market globally, making the stock market as a major point of call indicating the effect of COVID-19 on economies. As such, it is an important contribution to the literature and the economic rebound of the Nigerian economy to embark on a study of the risk-return relationship of stock on the NSE to provide an initial estimate for policy direction and guide concerning past policies and future decisions. Data and Methodology The discussion in this section includes the method that was used for the analysis, mainly the theoretical framework, model specifications, sources and estimation procedure. Data The data are secondary, sourced from the Nigerian Stock Exchange using the daily selected sectoral All Share Index return of different sectors (see table 1) from January 1st till June 30th, 2020. The return series was generated by the following for 123 observations for each sector: 𝐴𝐴𝐴𝐴𝐴𝐴�� = (����� ������) ������ (2) Where 𝐴𝐴𝐴𝐴𝐴𝐴�� is All Share Index return at a particular day, 𝐴𝐴𝐴𝐴𝐴𝐴� is the All-share Index of a particular day, 𝐴𝐴𝐴𝐴𝐴𝐴��� is the previous day All-Share Index. Table 1 consists of the sectoral index used in the study, the description and the source of the data used as a proxy for each sector index. Table 1. Description of Selected Stock Market Indices (2) Where ASIrt is All Share Index return at a particular day, ASIt is the All-share In- dex of a particular day, ASIt–1 is the previous day All-Share Index. Table 1 con- Kamaldeen Ibraheem Nageri102 sists of the sectoral index used in the study, the description and the source of the data used as a proxy for each sector index. Table 1. Description of Selected Stock Market Indices Variables Description Source NSE30 It is the price index that tracks the top thirty listed companies with fully paid- -up common shares in terms of market capitalisation and liquidity on the Nigerian Stock Exchange. NSE NSE5O It is the price index of the top fifty listed companies with fully paid-up common shares. It is weighted by adjusted market capitalisation multiplied by closing prices of the companies, multiply by a capping factor. NSE NSEBAN It is the price index that tracks the performance of listed banks with the most capitalisation and liquidity performance. NSE NSECON It is the price index that tracks the performance of listed companies in the consumer goods sector (food, beverages and tobacco) based on the market capitalisation method. NSE NSEIND It is the price index based on market capitalisation and method that captures the performance of listed companies in the industrial sector. NSE NSEINS It is the price index that captures the performance of listed insurance compa- nies with the most capitalisation and liquidity. NSE NSEISL It is the price index based on the requirements of the Shari’ah Advisory Board that captures the performance of fifteen Shari’ah-compliant listed companies. NSE NSEOIL It is the price index based on market capitalisation methodology that captures the performance of listed companies in the oil and gas sector NSE NSEPEN It is the price index that captures the top forty listed companies in terms of adjusted (free float factor) market capitalisation and liquidity NSE NSEPRE It is the price index that tracks the performance of fully paid up ordinary shares of companies listed on the premium board NSE NSEALL It is the cross-section of the ten price indexes listed above Computed S o u r c e : Nigerian Stock Exchange (NSE). risk-return relAtionshiP in the nigeriAn stock mArket… 103 V ar ia bl es D es cr ip tio n So ur ce N SE 30 It i s th e pr ic e in de x th at t ra ck s th e to p th ir ty l is te d co m pa ni es w ith f ul ly p ai d- up co m m on s ha re s in te rm s of m ar ke t c ap ita lis at io n an d liq ui di ty o n th e N ig er ia n St oc k E xc ha n g e. N SE N SE 5O It i s th e pr ic e in de x of t he t op f if ty l is te d co m pa ni es w ith f ul ly p ai d- up c om m on sh ar es . I t i s w ei gh te d by a dj us te d m ar ke t c ap ita lis at io n m ul tip lie d by c lo si ng p ri ce s of th e co m pa ni es , m ul tip l y b y a ca pp in g fa ct or . N SE N SE B A N It i s th e pr ic e in de x th at t ra ck s th e pe rf or m an ce o f lis te d ba nk s w ith t he m os t ca pi ta lis at io n an d liq ui di t y p er fo rm an ce . N SE N SE C O N It is th e pr ic e in de x th at tr ac ks th e pe rf or m an ce o f lis te d co m pa ni es in th e co ns um er go od s se ct or ( fo od , be ve ra ge s an d to ba cc o) b as ed o n th e m ar ke t ca pi ta lis at io n m et ho d. N SE N SE IN D It i s th e pr ic e in de x ba se d on m ar ke t ca pi ta lis at io n an d m et ho d th at c ap tu re s th e pe rf or m an ce o f l is te d co m pa ni es in th e in du st ri al s ec to r. N SE N SE IN S It is th e pr ic e in de x th at c ap tu re s th e pe rf or m an ce o f lis te d in su ra nc e co m pa ni es w ith th e m os t c ap ita lis at io n an d liq ui di t y . N SE N SE IS L It is th e pr ic e in de x ba se d on th e re qu ir em en ts o f t he S ha ri ’a h A dv is or y B oa rd th at ca pt ur es th e pe rf or m an ce o f f if te en S ha ri 'a h- co m pl ia nt li st ed c om pa ni es . N SE N SE O IL It i s th e pr ic e in de x ba se d on m ar ke t ca pi ta lis at io n m et ho do lo gy t ha t ca pt ur es t he pe rf or m an ce o f l is te d co m pa ni es in th e oi l a nd g as s ec to r N SE N SE PE N It is th e pr ic e in de x th at c ap tu re s th e to p fo rt y lis te d co m pa ni es in te rm s of a dj us te d (f re e fl oa t f ac to r) m ar ke t c ap ita lis at io n an d liq ui di t y N SE N SE PR E It i s th e pr ic e in de x th at t ra ck s th e pe rf or m an ce o f fu lly p ai d up o rd in ar y sh ar es o f co m pa ni es li st ed o n th e pr em iu m b oa rd N SE N SE A L L It is th e cr os s- se ct io n of th e te n pr ic e in de xe s lis te d ab ov e C om pu te d So ur ce : N ig er ia n St oc k Ex ch an ge (N SE ). F ig ur e 1. G ra ph ic al P re se nt at io n of th e Se le ct ed S ec to ra l S ha re In de x on th e N ig er ia n St oc k E xc ha ng e So ur ce : N SE , 2 02 0. 1, 00 0 1, 50 0 2, 00 0 2, 50 0 3, 00 0 3, 50 0 M 1 M 2 M 3 M 4 M 5 M 6 M 7 M 8 M 9 M 10 M 11 M 12 20 20 N S E P R E 60 0 80 0 1, 00 0 1, 20 0 1, 40 0 1, 60 0 M 1 M 2 M 3 M 4 M 5 M 6 M 7 M 8 M 9 M 10 M 11 M 12 20 20 N S E P E N 16 0 18 0 20 0 22 0 24 0 26 0 M 1 M 2 M 3 M 4 M 5 M 6 M 7 M 8 M 9 M 10 M 11 M 12 20 20 N S E O IL 1, 20 0 1, 60 0 2, 00 0 2, 40 0 2, 80 0 3, 20 0 M 1 M 2 M 3 M 4 M 5 M 6 M 7 M 8 M 9 M 10 M 11 M 12 20 20 N S E IS L 10 0 12 0 14 0 16 0 18 0 20 0 M 1 M 2 M 3 M 4 M 5 M 6 M 7 M 8 M 9 M 10 M 11 M 12 20 20 N S E IN S 80 0 1, 00 0 1, 20 0 1, 40 0 1, 60 0 1, 80 0 2, 00 0 2, 20 0 M 1 M 2 M 3 M 4 M 5 M 6 M 7 M 8 M 9 M 10 M 11 M 12 20 20 N S E IN D 30 0 40 0 50 0 60 0 70 0 M 1 M 2 M 3 M 4 M 5 M 6 M 7 M 8 M 9 M 10 M 11 M 12 20 20 N S E C O N 20 0 25 0 30 0 35 0 40 0 45 0 50 0 M 1 M 2 M 3 M 4 M 5 M 6 M 7 M 8 M 9 M 10 M 11 M 12 20 20 N S E B A N -4 00040 0 80 0 1, 20 0 1, 60 0 2, 00 0 M 1 M 2 M 3 M 4 M 5 M 6 M 7 M 8 M 9 M 10 M 11 M 12 20 20 N S E 50 80 0 1, 00 0 1, 20 0 1, 40 0 1, 60 0 1, 80 0 M 1 M 2 M 3 M 4 M 5 M 6 M 7 M 8 M 9 M 10 M 11 M 12 20 20 N S E 30 Fi gu re 1 . G ra ph ic al P re se nt at io n of th e Se le ct ed S ec to ra l S ha re In de x on th e N ig er ia n St oc k E xc ha ng e S o u r c e : N SE , 2 02 0. Kamaldeen Ibraheem Nageri104 The sectoral index, as shown in figure 1, indicates that the sectors respond sim- ilarly during the early months of the year and at the inception of COVID-19 cas- es in Nigeria between January (M1) and April (M4). The sectorial indexes were increasing in January (M1) but experienced a decline afterwards till around March (M3) and April (M4). This was the period when COVID-19 was first re- ported and various forms of lockdown were also announced in Nigeria. During this period of lockdowns and decline, various policy responses by different eco- nomic agencies were announced and implemented to cushion the negative ef- fect of the pandemic in the country. NSEINS started to recover as early as March (M3) while other sectors started recovering toward April (M4), afterwards the policy responses, except for NSEOIL that indicates a second decline around May (M5) towards June (M6). The indexes indicate different magnitude because the index value of the sectors varies, the NSEPRE has the highest index followed by NSEISL while the lowest index is shown by NSEINS. Theoretical Framework Stock returns exhibit stochastic volatility and jumps (Nageri, 2019b) indicating the risk associated with deviation in returns (Campbell, Lettau, Malkiel & Xu, 2001; Pastor & Veronesi, 2006). Volatility models should capture error term’s heteroscedasticity and stylised fact in stock returns (Engle, 1982). The gener- al form of the conditional variance equation incorporates the ARCH processes with (p) lagged as follows: The sectoral index, as shown in figure 1, indicates that the sectors respond similarly during the early months of the year and at the inception of COVID-19 cases in Nigeria between January (M1) and April (M4). The sectorial indexes were increasing in January (M1) but experienced a decline afterwards till around March (M3) and April (M4). This was the period when COVID- 19 was first reported and various forms of lockdown were also announced in Nigeria. During this period of lockdowns and decline, various policy responses by different economic agencies were announced and implemented to cushion the negative effect of the pandemic in the country. NSEINS started to recover as early as March (M3) while other sectors started recovering toward April (M4), afterwards the policy responses, except for NSEOIL that indicates a second decline around May (M5) towards June (M6). The indexes indicate different magnitude because the index value of the sectors varies, the NSEPRE has the highest index followed by NSEISL while the lowest index is shown by NSEINS. Theoretical Framework Stock returns exhibit stochastic volatility and jumps (Nageri, 2019b) indicating the risk associated with deviation in returns (Campbell, Lettau, Malkiel & Xu, 2001; Pastor & Veronesi, 2006). Volatility models should capture error term’s heteroscedasticity and stylised fact in stock returns (Engle, 1982). The general form of the conditional variance equation incorporates the ARCH processes with (p) lagged as follows: 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔𝜔𝜔�𝜀𝜀���� 𝜔𝜔…………𝜔𝜔𝜔𝜔�𝜀𝜀���� (3) Equation 3, ARCH (1) model shows that next period’s variance of return which was the residuals of the mean equation (𝜎𝜎��), depends on squared residuals of the past period from period 1 till p period (𝜀𝜀���� ) known as the ARCH term, 𝜔𝜔� is the parameter of the ARCH term and 𝜔𝜔𝜔is the constant. An extension is the Generalised ARCH or GARCH model which adds the lags of the variance, 𝜎𝜎���� to the standard ARCH. GARCH model has one lag of the regression model’s squared residual (𝜀𝜀���� ) known as the ARCH term and one lag of the variance itself (𝜎𝜎���� ) known as the GARCH term (Bollerslev, 1986). 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔∑ 𝜔𝜔𝜀𝜀�������� 𝜔 ∑ 𝛽𝛽𝜎𝜎�������� (4) Equation 4 is the GARCH model developed by Bollerslev (1986). 𝜔𝜔, 𝛽𝛽 > 0 and (𝜔𝜔 + 𝛽𝛽) < 1 to avoid negative conditional variance and it indicates that present period’s return variance (3) Equation 3, ARCH (1) model shows that next period’s variance of return which was the residuals of the mean equation The sectoral index, as shown in figure 1, indicates that the sectors respond similarly during the early months of the year and at the inception of COVID-19 cases in Nigeria between January (M1) and April (M4). The sectorial indexes were increasing in January (M1) but experienced a decline afterwards till around March (M3) and April (M4). This was the period when COVID- 19 was first reported and various forms of lockdown were also announced in Nigeria. During this period of lockdowns and decline, various policy responses by different economic agencies were announced and implemented to cushion the negative effect of the pandemic in the country. NSEINS started to recover as early as March (M3) while other sectors started recovering toward April (M4), afterwards the policy responses, except for NSEOIL that indicates a second decline around May (M5) towards June (M6). The indexes indicate different magnitude because the index value of the sectors varies, the NSEPRE has the highest index followed by NSEISL while the lowest index is shown by NSEINS. Theoretical Framework Stock returns exhibit stochastic volatility and jumps (Nageri, 2019b) indicating the risk associated with deviation in returns (Campbell, Lettau, Malkiel & Xu, 2001; Pastor & Veronesi, 2006). Volatility models should capture error term’s heteroscedasticity and stylised fact in stock returns (Engle, 1982). The general form of the conditional variance equation incorporates the ARCH processes with (p) lagged as follows: 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔𝜔𝜔�𝜀𝜀���� 𝜔𝜔…………𝜔𝜔𝜔𝜔�𝜀𝜀���� (3) Equation 3, ARCH (1) model shows that next period’s variance of return which was the residuals of the mean equation (𝜎𝜎��), depends on squared residuals of the past period from period 1 till p period (𝜀𝜀���� ) known as the ARCH term, 𝜔𝜔� is the parameter of the ARCH term and 𝜔𝜔𝜔is the constant. An extension is the Generalised ARCH or GARCH model which adds the lags of the variance, 𝜎𝜎���� to the standard ARCH. GARCH model has one lag of the regression model’s squared residual (𝜀𝜀���� ) known as the ARCH term and one lag of the variance itself (𝜎𝜎���� ) known as the GARCH term (Bollerslev, 1986). 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔∑ 𝜔𝜔𝜀𝜀�������� 𝜔 ∑ 𝛽𝛽𝜎𝜎�������� (4) Equation 4 is the GARCH model developed by Bollerslev (1986). 𝜔𝜔, 𝛽𝛽 > 0 and (𝜔𝜔 + 𝛽𝛽) < 1 to avoid negative conditional variance and it indicates that present period’s return variance , depends on squared residuals of the past period from period 1 till p period The sectoral index, as shown in figure 1, indicates that the sectors respond similarly during the early months of the year and at the inception of COVID-19 cases in Nigeria between January (M1) and April (M4). The sectorial indexes were increasing in January (M1) but experienced a decline afterwards till around March (M3) and April (M4). This was the period when COVID- 19 was first reported and various forms of lockdown were also announced in Nigeria. During this period of lockdowns and decline, various policy responses by different economic agencies were announced and implemented to cushion the negative effect of the pandemic in the country. NSEINS started to recover as early as March (M3) while other sectors started recovering toward April (M4), afterwards the policy responses, except for NSEOIL that indicates a second decline around May (M5) towards June (M6). The indexes indicate different magnitude because the index value of the sectors varies, the NSEPRE has the highest index followed by NSEISL while the lowest index is shown by NSEINS. Theoretical Framework Stock returns exhibit stochastic volatility and jumps (Nageri, 2019b) indicating the risk associated with deviation in returns (Campbell, Lettau, Malkiel & Xu, 2001; Pastor & Veronesi, 2006). Volatility models should capture error term’s heteroscedasticity and stylised fact in stock returns (Engle, 1982). The general form of the conditional variance equation incorporates the ARCH processes with (p) lagged as follows: 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔𝜔𝜔�𝜀𝜀���� 𝜔𝜔…………𝜔𝜔𝜔𝜔�𝜀𝜀���� (3) Equation 3, ARCH (1) model shows that next period’s variance of return which was the residuals of the mean equation (𝜎𝜎��), depends on squared residuals of the past period from period 1 till p period (𝜀𝜀���� ) known as the ARCH term, 𝜔𝜔� is the parameter of the ARCH term and 𝜔𝜔𝜔is the constant. An extension is the Generalised ARCH or GARCH model which adds the lags of the variance, 𝜎𝜎���� to the standard ARCH. GARCH model has one lag of the regression model’s squared residual (𝜀𝜀���� ) known as the ARCH term and one lag of the variance itself (𝜎𝜎���� ) known as the GARCH term (Bollerslev, 1986). 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔∑ 𝜔𝜔𝜀𝜀�������� 𝜔 ∑ 𝛽𝛽𝜎𝜎�������� (4) Equation 4 is the GARCH model developed by Bollerslev (1986). 𝜔𝜔, 𝛽𝛽 > 0 and (𝜔𝜔 + 𝛽𝛽) < 1 to avoid negative conditional variance and it indicates that present period’s return variance known as the ARCH term, α1 is the parameter of the ARCH term and ω is the constant. An extension is the Gen- eralised ARCH or GARCH model which adds the lags of the variance, The sectoral index, as shown in figure 1, indicates that the sectors respond similarly during the early months of the year and at the inception of COVID-19 cases in Nigeria between January (M1) and April (M4). The sectorial indexes were increasing in January (M1) but experienced a decline afterwards till around March (M3) and April (M4). This was the period when COVID- 19 was first reported and various forms of lockdown were also announced in Nigeria. During this period of lockdowns and decline, various policy responses by different economic agencies were announced and implemented to cushion the negative effect of the pandemic in the country. NSEINS started to recover as early as March (M3) while other sectors started recovering toward April (M4), afterwards the policy responses, except for NSEOIL that indicates a second decline around May (M5) towards June (M6). The indexes indicate different magnitude because the index value of the sectors varies, the NSEPRE has the highest index followed by NSEISL while the lowest index is shown by NSEINS. Theoretical Framework Stock returns exhibit stochastic volatility and jumps (Nageri, 2019b) indicating the risk associated with deviation in returns (Campbell, Lettau, Malkiel & Xu, 2001; Pastor & Veronesi, 2006). Volatility models should capture error term’s heteroscedasticity and stylised fact in stock returns (Engle, 1982). The general form of the conditional variance equation incorporates the ARCH processes with (p) lagged as follows: 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔𝜔𝜔�𝜀𝜀���� 𝜔𝜔…………𝜔𝜔𝜔𝜔�𝜀𝜀���� (3) Equation 3, ARCH (1) model shows that next period’s variance of return which was the residuals of the mean equation (𝜎𝜎��), depends on squared residuals of the past period from period 1 till p period (𝜀𝜀���� ) known as the ARCH term, 𝜔𝜔� is the parameter of the ARCH term and 𝜔𝜔𝜔is the constant. An extension is the Generalised ARCH or GARCH model which adds the lags of the variance, 𝜎𝜎���� to the standard ARCH. GARCH model has one lag of the regression model’s squared residual (𝜀𝜀���� ) known as the ARCH term and one lag of the variance itself (𝜎𝜎���� ) known as the GARCH term (Bollerslev, 1986). 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔∑ 𝜔𝜔𝜀𝜀�������� 𝜔 ∑ 𝛽𝛽𝜎𝜎�������� (4) Equation 4 is the GARCH model developed by Bollerslev (1986). 𝜔𝜔, 𝛽𝛽 > 0 and (𝜔𝜔 + 𝛽𝛽) < 1 to avoid negative conditional variance and it indicates that present period’s return variance to the standard ARCH. GARCH model has one lag of the regression model’s squared residual The sectoral index, as shown in figure 1, indicates that the sectors respond similarly during the early months of the year and at the inception of COVID-19 cases in Nigeria between January (M1) and April (M4). The sectorial indexes were increasing in January (M1) but experienced a decline afterwards till around March (M3) and April (M4). This was the period when COVID- 19 was first reported and various forms of lockdown were also announced in Nigeria. During this period of lockdowns and decline, various policy responses by different economic agencies were announced and implemented to cushion the negative effect of the pandemic in the country. NSEINS started to recover as early as March (M3) while other sectors started recovering toward April (M4), afterwards the policy responses, except for NSEOIL that indicates a second decline around May (M5) towards June (M6). The indexes indicate different magnitude because the index value of the sectors varies, the NSEPRE has the highest index followed by NSEISL while the lowest index is shown by NSEINS. Theoretical Framework Stock returns exhibit stochastic volatility and jumps (Nageri, 2019b) indicating the risk associated with deviation in returns (Campbell, Lettau, Malkiel & Xu, 2001; Pastor & Veronesi, 2006). Volatility models should capture error term’s heteroscedasticity and stylised fact in stock returns (Engle, 1982). The general form of the conditional variance equation incorporates the ARCH processes with (p) lagged as follows: 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔𝜔𝜔�𝜀𝜀���� 𝜔𝜔…………𝜔𝜔𝜔𝜔�𝜀𝜀���� (3) Equation 3, ARCH (1) model shows that next period’s variance of return which was the residuals of the mean equation (𝜎𝜎��), depends on squared residuals of the past period from period 1 till p period (𝜀𝜀���� ) known as the ARCH term, 𝜔𝜔� is the parameter of the ARCH term and 𝜔𝜔𝜔is the constant. An extension is the Generalised ARCH or GARCH model which adds the lags of the variance, 𝜎𝜎���� to the standard ARCH. GARCH model has one lag of the regression model’s squared residual (𝜀𝜀���� ) known as the ARCH term and one lag of the variance itself (𝜎𝜎���� ) known as the GARCH term (Bollerslev, 1986). 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔∑ 𝜔𝜔𝜀𝜀�������� 𝜔 ∑ 𝛽𝛽𝜎𝜎�������� (4) Equation 4 is the GARCH model developed by Bollerslev (1986). 𝜔𝜔, 𝛽𝛽 > 0 and (𝜔𝜔 + 𝛽𝛽) < 1 to avoid negative conditional variance and it indicates that present period’s return variance known as the ARCH term and one lag of the variance itself The sectoral index, as shown in figure 1, indicates that the sectors respond similarly during the early months of the year and at the inception of COVID-19 cases in Nigeria between January (M1) and April (M4). The sectorial indexes were increasing in January (M1) but experienced a decline afterwards till around March (M3) and April (M4). This was the period when COVID- 19 was first reported and various forms of lockdown were also announced in Nigeria. During this period of lockdowns and decline, various policy responses by different economic agencies were announced and implemented to cushion the negative effect of the pandemic in the country. NSEINS started to recover as early as March (M3) while other sectors started recovering toward April (M4), afterwards the policy responses, except for NSEOIL that indicates a second decline around May (M5) towards June (M6). The indexes indicate different magnitude because the index value of the sectors varies, the NSEPRE has the highest index followed by NSEISL while the lowest index is shown by NSEINS. Theoretical Framework Stock returns exhibit stochastic volatility and jumps (Nageri, 2019b) indicating the risk associated with deviation in returns (Campbell, Lettau, Malkiel & Xu, 2001; Pastor & Veronesi, 2006). Volatility models should capture error term’s heteroscedasticity and stylised fact in stock returns (Engle, 1982). The general form of the conditional variance equation incorporates the ARCH processes with (p) lagged as follows: 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔𝜔𝜔�𝜀𝜀���� 𝜔𝜔…………𝜔𝜔𝜔𝜔�𝜀𝜀���� (3) Equation 3, ARCH (1) model shows that next period’s variance of return which was the residuals of the mean equation (𝜎𝜎��), depends on squared residuals of the past period from period 1 till p period (𝜀𝜀���� ) known as the ARCH term, 𝜔𝜔� is the parameter of the ARCH term and 𝜔𝜔𝜔is the constant. An extension is the Generalised ARCH or GARCH model which adds the lags of the variance, 𝜎𝜎���� to the standard ARCH. GARCH model has one lag of the regression model’s squared residual (𝜀𝜀���� ) known as the ARCH term and one lag of the variance itself (𝜎𝜎���� ) known as the GARCH term (Bollerslev, 1986). 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔∑ 𝜔𝜔𝜀𝜀�������� 𝜔 ∑ 𝛽𝛽𝜎𝜎�������� (4) Equation 4 is the GARCH model developed by Bollerslev (1986). 𝜔𝜔, 𝛽𝛽 > 0 and (𝜔𝜔 + 𝛽𝛽) < 1 to avoid negative conditional variance and it indicates that present period’s return variance known as the GARCH term (Bollerslev, 1986). risk-return relAtionshiP in the nigeriAn stock mArket… 105 The sectoral index, as shown in figure 1, indicates that the sectors respond similarly during the early months of the year and at the inception of COVID-19 cases in Nigeria between January (M1) and April (M4). The sectorial indexes were increasing in January (M1) but experienced a decline afterwards till around March (M3) and April (M4). This was the period when COVID- 19 was first reported and various forms of lockdown were also announced in Nigeria. During this period of lockdowns and decline, various policy responses by different economic agencies were announced and implemented to cushion the negative effect of the pandemic in the country. NSEINS started to recover as early as March (M3) while other sectors started recovering toward April (M4), afterwards the policy responses, except for NSEOIL that indicates a second decline around May (M5) towards June (M6). The indexes indicate different magnitude because the index value of the sectors varies, the NSEPRE has the highest index followed by NSEISL while the lowest index is shown by NSEINS. Theoretical Framework Stock returns exhibit stochastic volatility and jumps (Nageri, 2019b) indicating the risk associated with deviation in returns (Campbell, Lettau, Malkiel & Xu, 2001; Pastor & Veronesi, 2006). Volatility models should capture error term’s heteroscedasticity and stylised fact in stock returns (Engle, 1982). The general form of the conditional variance equation incorporates the ARCH processes with (p) lagged as follows: 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔𝜔𝜔�𝜀𝜀���� 𝜔𝜔…………𝜔𝜔𝜔𝜔�𝜀𝜀���� (3) Equation 3, ARCH (1) model shows that next period’s variance of return which was the residuals of the mean equation (𝜎𝜎��), depends on squared residuals of the past period from period 1 till p period (𝜀𝜀���� ) known as the ARCH term, 𝜔𝜔� is the parameter of the ARCH term and 𝜔𝜔𝜔is the constant. An extension is the Generalised ARCH or GARCH model which adds the lags of the variance, 𝜎𝜎���� to the standard ARCH. GARCH model has one lag of the regression model’s squared residual (𝜀𝜀���� ) known as the ARCH term and one lag of the variance itself (𝜎𝜎���� ) known as the GARCH term (Bollerslev, 1986). 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔∑ 𝜔𝜔𝜀𝜀�������� 𝜔 ∑ 𝛽𝛽𝜎𝜎�������� (4) Equation 4 is the GARCH model developed by Bollerslev (1986). 𝜔𝜔, 𝛽𝛽 > 0 and (𝜔𝜔 + 𝛽𝛽) < 1 to avoid negative conditional variance and it indicates that present period’s return variance (4) Equation 4 is the GARCH model developed by Bollerslev (1986). α, β > 0 and (α + β) < 1 to avoid negative conditional variance and it indicates that present period’s return variance The sectoral index, as shown in figure 1, indicates that the sectors respond similarly during the early months of the year and at the inception of COVID-19 cases in Nigeria between January (M1) and April (M4). The sectorial indexes were increasing in January (M1) but experienced a decline afterwards till around March (M3) and April (M4). This was the period when COVID- 19 was first reported and various forms of lockdown were also announced in Nigeria. During this period of lockdowns and decline, various policy responses by different economic agencies were announced and implemented to cushion the negative effect of the pandemic in the country. NSEINS started to recover as early as March (M3) while other sectors started recovering toward April (M4), afterwards the policy responses, except for NSEOIL that indicates a second decline around May (M5) towards June (M6). The indexes indicate different magnitude because the index value of the sectors varies, the NSEPRE has the highest index followed by NSEISL while the lowest index is shown by NSEINS. Theoretical Framework Stock returns exhibit stochastic volatility and jumps (Nageri, 2019b) indicating the risk associated with deviation in returns (Campbell, Lettau, Malkiel & Xu, 2001; Pastor & Veronesi, 2006). Volatility models should capture error term’s heteroscedasticity and stylised fact in stock returns (Engle, 1982). The general form of the conditional variance equation incorporates the ARCH processes with (p) lagged as follows: 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔𝜔𝜔�𝜀𝜀���� 𝜔𝜔…………𝜔𝜔𝜔𝜔�𝜀𝜀���� (3) Equation 3, ARCH (1) model shows that next period’s variance of return which was the residuals of the mean equation (𝜎𝜎��), depends on squared residuals of the past period from period 1 till p period (𝜀𝜀���� ) known as the ARCH term, 𝜔𝜔� is the parameter of the ARCH term and 𝜔𝜔𝜔is the constant. An extension is the Generalised ARCH or GARCH model which adds the lags of the variance, 𝜎𝜎���� to the standard ARCH. GARCH model has one lag of the regression model’s squared residual (𝜀𝜀���� ) known as the ARCH term and one lag of the variance itself (𝜎𝜎���� ) known as the GARCH term (Bollerslev, 1986). 𝜎𝜎�� = 𝜔𝜔 𝜔𝜔∑ 𝜔𝜔𝜀𝜀�������� 𝜔 ∑ 𝛽𝛽𝜎𝜎�������� (4) Equation 4 is the GARCH model developed by Bollerslev (1986). 𝜔𝜔, 𝛽𝛽 > 0 and (𝜔𝜔 + 𝛽𝛽) < 1 to avoid negative conditional variance and it indicates that present period’s return variance is determined by a constant (ω), estimates of the previous period’s squared residual of the mean return equation (𝜎𝜎��) is determined by a constant (𝜔𝜔), estimates of the previous period’s squared residual of the mean return equation (∑ 𝛼𝛼𝛼𝛼�������� ) and estimates of the previous period’s return variance (∑ 𝛽𝛽𝜎𝜎�������� ). Model Specification This study examined the risk-return relationship of returns and evaluates good and bad news during COVID-19. Variant Auto-Regressive Conditional Heteroscedasticity (ARCH) models were used to achieve the purpose of the study under three (3) distributional assumptions. The following is the general form of the panel GARCH model used in this research: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶𝐶𝛼𝛼��𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶𝐶𝛼𝛼��� Cross-section mean return equation for 𝐴𝐴𝐴𝐴𝐴𝐴��� (7) 𝜎𝜎��� = 𝜔𝜔 𝐶 𝛼𝛼𝛼𝛼����� 𝐶𝐶 𝐶𝛽𝛽𝜎𝜎����� Cross-section variance of return equation (8) Where 𝜎𝜎��� represents the cross-section return variance (error term from the mean return equation) is determined by 𝜔𝜔 representing the constant, 𝛼𝛼����� representing the ARCH term showing the past period’s cross-section squared error term derived from the mean return equation and 𝜎𝜎����� representing the GARCH term showing the past period’s cross-section variance of the return. Risk-Return during COVID-19 The panel GARCH-in-Mean (GARCH-M) model used to measure the risk-return relationship was as follows: 𝜎𝜎��� = 𝜔𝜔 𝐶 𝜔𝜔��𝑔𝑔𝑔𝜎𝜎��) 𝐶𝐶𝛼𝛼��� (9) Where 𝑔𝑔𝑔.) is the arbitrary volatility function of 𝜎𝜎��, the cross-section GARCH-M was specified with cross-section GARCH conditional variance specification and the function 𝑔𝑔𝑔𝜎𝜎��) is the standard deviation in mean (𝜔𝜔��). Positive 𝜔𝜔�� indicates higher risk leading to higher average return and vice versa. The a priori expectation of GARCH-in-Mean is 0 > 𝜔𝜔�� > 0, indicating that the risk-return relationship can be positive or negative. Good and Bad News during COVID-19 Threshold GARCH (TGARCH/TARCH) permits asymmetric effect between positive and negative news on cross-section stock returns with general description stated below:𝐶𝜎𝜎��� = 𝜔𝜔 + ∑ 𝛼𝛼������ 𝛼𝛼����� + ∑ 𝛾𝛾������ 𝛼𝛼����� 𝑑𝑑���� + ∑ 𝛽𝛽������ 𝜎𝜎����� (10) and estimates of the previous period’s return variance (𝜎𝜎��) is determined by a constant (𝜔𝜔), estimates of the previous period’s squared residual of the mean return equation (∑ 𝛼𝛼𝛼𝛼�������� ) and estimates of the previous period’s return variance (∑ 𝛽𝛽𝜎𝜎�������� ). Model Specification This study examined the risk-return relationship of returns and evaluates good and bad news during COVID-19. Variant Auto-Regressive Conditional Heteroscedasticity (ARCH) models were used to achieve the purpose of the study under three (3) distributional assumptions. The following is the general form of the panel GARCH model used in this research: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶𝐶𝛼𝛼��𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶𝐶𝛼𝛼��� Cross-section mean return equation for 𝐴𝐴𝐴𝐴𝐴𝐴��� (7) 𝜎𝜎��� = 𝜔𝜔 𝐶 𝛼𝛼𝛼𝛼����� 𝐶𝐶 𝐶𝛽𝛽𝜎𝜎����� Cross-section variance of return equation (8) Where 𝜎𝜎��� represents the cross-section return variance (error term from the mean return equation) is determined by 𝜔𝜔 representing the constant, 𝛼𝛼����� representing the ARCH term showing the past period’s cross-section squared error term derived from the mean return equation and 𝜎𝜎����� representing the GARCH term showing the past period’s cross-section variance of the return. Risk-Return during COVID-19 The panel GARCH-in-Mean (GARCH-M) model used to measure the risk-return relationship was as follows: 𝜎𝜎��� = 𝜔𝜔 𝐶 𝜔𝜔��𝑔𝑔𝑔𝜎𝜎��) 𝐶𝐶𝛼𝛼��� (9) Where 𝑔𝑔𝑔.) is the arbitrary volatility function of 𝜎𝜎��, the cross-section GARCH-M was specified with cross-section GARCH conditional variance specification and the function 𝑔𝑔𝑔𝜎𝜎��) is the standard deviation in mean (𝜔𝜔��). Positive 𝜔𝜔�� indicates higher risk leading to higher average return and vice versa. The a priori expectation of GARCH-in-Mean is 0 > 𝜔𝜔�� > 0, indicating that the risk-return relationship can be positive or negative. Good and Bad News during COVID-19 Threshold GARCH (TGARCH/TARCH) permits asymmetric effect between positive and negative news on cross-section stock returns with general description stated below:𝐶𝜎𝜎��� = 𝜔𝜔 + ∑ 𝛼𝛼������ 𝛼𝛼����� + ∑ 𝛾𝛾������ 𝛼𝛼����� 𝑑𝑑���� + ∑ 𝛽𝛽������ 𝜎𝜎����� (10) Model Specification This study examined the risk-return relationship of returns and evaluates good and bad news during COVID-19. Variant Auto-Regressive Conditional Het- eroscedasticity (ARCH) models were used to achieve the purpose of the study under three (3) distributional assumptions. The following is the general form of the panel GARCH model used in this research: (𝜎𝜎��) is determined by a constant (𝜔𝜔), estimates of the previous period’s squared residual of the mean return equation (∑ 𝛼𝛼𝛼𝛼�������� ) and estimates of the previous period’s return variance (∑ 𝛽𝛽𝜎𝜎�������� ). Model Specification This study examined the risk-return relationship of returns and evaluates good and bad news during COVID-19. Variant Auto-Regressive Conditional Heteroscedasticity (ARCH) models were used to achieve the purpose of the study under three (3) distributional assumptions. The following is the general form of the panel GARCH model used in this research: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶𝐶𝛼𝛼��𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶𝐶𝛼𝛼��� Cross-section mean return equation for 𝐴𝐴𝐴𝐴𝐴𝐴��� (7) 𝜎𝜎��� = 𝜔𝜔 𝐶 𝛼𝛼𝛼𝛼����� 𝐶𝐶 𝐶𝛽𝛽𝜎𝜎����� Cross-section variance of return equation (8) Where 𝜎𝜎��� represents the cross-section return variance (error term from the mean return equation) is determined by 𝜔𝜔 representing the constant, 𝛼𝛼����� representing the ARCH term showing the past period’s cross-section squared error term derived from the mean return equation and 𝜎𝜎����� representing the GARCH term showing the past period’s cross-section variance of the return. Risk-Return during COVID-19 The panel GARCH-in-Mean (GARCH-M) model used to measure the risk-return relationship was as follows: 𝜎𝜎��� = 𝜔𝜔 𝐶 𝜔𝜔��𝑔𝑔𝑔𝜎𝜎��) 𝐶𝐶𝛼𝛼��� (9) Where 𝑔𝑔𝑔.) is the arbitrary volatility function of 𝜎𝜎��, the cross-section GARCH-M was specified with cross-section GARCH conditional variance specification and the function 𝑔𝑔𝑔𝜎𝜎��) is the standard deviation in mean (𝜔𝜔��). Positive 𝜔𝜔�� indicates higher risk leading to higher average return and vice versa. The a priori expectation of GARCH-in-Mean is 0 > 𝜔𝜔�� > 0, indicating that the risk-return relationship can be positive or negative. Good and Bad News during COVID-19 Threshold GARCH (TGARCH/TARCH) permits asymmetric effect between positive and negative news on cross-section stock returns with general description stated below:𝐶𝜎𝜎��� = 𝜔𝜔 + ∑ 𝛼𝛼������ 𝛼𝛼����� + ∑ 𝛾𝛾������ 𝛼𝛼����� 𝑑𝑑���� + ∑ 𝛽𝛽������ 𝜎𝜎����� (10)  Cross-section mean return equation for ASIrit (7) (𝜎𝜎��) is determined by a constant (𝜔𝜔), estimates of the previous period’s squared residual of the mean return equation (∑ 𝛼𝛼𝛼𝛼�������� ) and estimates of the previous period’s return variance (∑ 𝛽𝛽𝜎𝜎�������� ). Model Specification This study examined the risk-return relationship of returns and evaluates good and bad news during COVID-19. Variant Auto-Regressive Conditional Heteroscedasticity (ARCH) models were used to achieve the purpose of the study under three (3) distributional assumptions. The following is the general form of the panel GARCH model used in this research: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶𝐶𝛼𝛼��𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶𝐶𝛼𝛼��� Cross-section mean return equation for 𝐴𝐴𝐴𝐴𝐴𝐴��� (7) 𝜎𝜎��� = 𝜔𝜔 𝐶 𝛼𝛼𝛼𝛼����� 𝐶𝐶 𝐶𝛽𝛽𝜎𝜎����� Cross-section variance of return equation (8) Where 𝜎𝜎��� represents the cross-section return variance (error term from the mean return equation) is determined by 𝜔𝜔 representing the constant, 𝛼𝛼����� representing the ARCH term showing the past period’s cross-section squared error term derived from the mean return equation and 𝜎𝜎����� representing the GARCH term showing the past period’s cross-section variance of the return. Risk-Return during COVID-19 The panel GARCH-in-Mean (GARCH-M) model used to measure the risk-return relationship was as follows: 𝜎𝜎��� = 𝜔𝜔 𝐶 𝜔𝜔��𝑔𝑔𝑔𝜎𝜎��) 𝐶𝐶𝛼𝛼��� (9) Where 𝑔𝑔𝑔.) is the arbitrary volatility function of 𝜎𝜎��, the cross-section GARCH-M was specified with cross-section GARCH conditional variance specification and the function 𝑔𝑔𝑔𝜎𝜎��) is the standard deviation in mean (𝜔𝜔��). Positive 𝜔𝜔�� indicates higher risk leading to higher average return and vice versa. The a priori expectation of GARCH-in-Mean is 0 > 𝜔𝜔�� > 0, indicating that the risk-return relationship can be positive or negative. Good and Bad News during COVID-19 Threshold GARCH (TGARCH/TARCH) permits asymmetric effect between positive and negative news on cross-section stock returns with general description stated below:𝐶𝜎𝜎��� = 𝜔𝜔 + ∑ 𝛼𝛼������ 𝛼𝛼����� + ∑ 𝛾𝛾������ 𝛼𝛼����� 𝑑𝑑���� + ∑ 𝛽𝛽������ 𝜎𝜎����� (10)  Cross-section variance of return equation (8) Where (𝜎𝜎��) is determined by a constant (𝜔𝜔), estimates of the previous period’s squared residual of the mean return equation (∑ 𝛼𝛼𝛼𝛼�������� ) and estimates of the previous period’s return variance (∑ 𝛽𝛽𝜎𝜎�������� ). Model Specification This study examined the risk-return relationship of returns and evaluates good and bad news during COVID-19. Variant Auto-Regressive Conditional Heteroscedasticity (ARCH) models were used to achieve the purpose of the study under three (3) distributional assumptions. The following is the general form of the panel GARCH model used in this research: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶𝐶𝛼𝛼��𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶𝐶𝛼𝛼��� Cross-section mean return equation for 𝐴𝐴𝐴𝐴𝐴𝐴��� (7) 𝜎𝜎��� = 𝜔𝜔 𝐶 𝛼𝛼𝛼𝛼����� 𝐶𝐶 𝐶𝛽𝛽𝜎𝜎����� Cross-section variance of return equation (8) Where 𝜎𝜎��� represents the cross-section return variance (error term from the mean return equation) is determined by 𝜔𝜔 representing the constant, 𝛼𝛼����� representing the ARCH term showing the past period’s cross-section squared error term derived from the mean return equation and 𝜎𝜎����� representing the GARCH term showing the past period’s cross-section variance of the return. Risk-Return during COVID-19 The panel GARCH-in-Mean (GARCH-M) model used to measure the risk-return relationship was as follows: 𝜎𝜎��� = 𝜔𝜔 𝐶 𝜔𝜔��𝑔𝑔𝑔𝜎𝜎��) 𝐶𝐶𝛼𝛼��� (9) Where 𝑔𝑔𝑔.) is the arbitrary volatility function of 𝜎𝜎��, the cross-section GARCH-M was specified with cross-section GARCH conditional variance specification and the function 𝑔𝑔𝑔𝜎𝜎��) is the standard deviation in mean (𝜔𝜔��). Positive 𝜔𝜔�� indicates higher risk leading to higher average return and vice versa. The a priori expectation of GARCH-in-Mean is 0 > 𝜔𝜔�� > 0, indicating that the risk-return relationship can be positive or negative. Good and Bad News during COVID-19 Threshold GARCH (TGARCH/TARCH) permits asymmetric effect between positive and negative news on cross-section stock returns with general description stated below:𝐶𝜎𝜎��� = 𝜔𝜔 + ∑ 𝛼𝛼������ 𝛼𝛼����� + ∑ 𝛾𝛾������ 𝛼𝛼����� 𝑑𝑑���� + ∑ 𝛽𝛽������ 𝜎𝜎����� (10) represents the cross-section return variance (error term from the mean return equation) is determined by ω representing the constant, (𝜎𝜎��) is determined by a constant (𝜔𝜔), estimates of the previous period’s squared residual of the mean return equation (∑ 𝛼𝛼𝛼𝛼�������� ) and estimates of the previous period’s return variance (∑ 𝛽𝛽𝜎𝜎�������� ). Model Specification This study examined the risk-return relationship of returns and evaluates good and bad news during COVID-19. Variant Auto-Regressive Conditional Heteroscedasticity (ARCH) models were used to achieve the purpose of the study under three (3) distributional assumptions. The following is the general form of the panel GARCH model used in this research: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶𝐶𝛼𝛼��𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶𝐶𝛼𝛼��� Cross-section mean return equation for 𝐴𝐴𝐴𝐴𝐴𝐴��� (7) 𝜎𝜎��� = 𝜔𝜔 𝐶 𝛼𝛼𝛼𝛼����� 𝐶𝐶 𝐶𝛽𝛽𝜎𝜎����� Cross-section variance of return equation (8) Where 𝜎𝜎��� represents the cross-section return variance (error term from the mean return equation) is determined by 𝜔𝜔 representing the constant, 𝛼𝛼����� representing the ARCH term showing the past period’s cross-section squared error term derived from the mean return equation and 𝜎𝜎����� representing the GARCH term showing the past period’s cross-section variance of the return. Risk-Return during COVID-19 The panel GARCH-in-Mean (GARCH-M) model used to measure the risk-return relationship was as follows: 𝜎𝜎��� = 𝜔𝜔 𝐶 𝜔𝜔��𝑔𝑔𝑔𝜎𝜎��) 𝐶𝐶𝛼𝛼��� (9) Where 𝑔𝑔𝑔.) is the arbitrary volatility function of 𝜎𝜎��, the cross-section GARCH-M was specified with cross-section GARCH conditional variance specification and the function 𝑔𝑔𝑔𝜎𝜎��) is the standard deviation in mean (𝜔𝜔��). Positive 𝜔𝜔�� indicates higher risk leading to higher average return and vice versa. The a priori expectation of GARCH-in-Mean is 0 > 𝜔𝜔�� > 0, indicating that the risk-return relationship can be positive or negative. Good and Bad News during COVID-19 Threshold GARCH (TGARCH/TARCH) permits asymmetric effect between positive and negative news on cross-section stock returns with general description stated below:𝐶𝜎𝜎��� = 𝜔𝜔 + ∑ 𝛼𝛼������ 𝛼𝛼����� + ∑ 𝛾𝛾������ 𝛼𝛼����� 𝑑𝑑���� + ∑ 𝛽𝛽������ 𝜎𝜎����� (10) rep- resenting the ARCH term showing the past period’s cross-section squared er- ror term derived from the mean return equation and (𝜎𝜎��) is determined by a constant (𝜔𝜔), estimates of the previous period’s squared residual of the mean return equation (∑ 𝛼𝛼𝛼𝛼�������� ) and estimates of the previous period’s return variance (∑ 𝛽𝛽𝜎𝜎�������� ). Model Specification This study examined the risk-return relationship of returns and evaluates good and bad news during COVID-19. Variant Auto-Regressive Conditional Heteroscedasticity (ARCH) models were used to achieve the purpose of the study under three (3) distributional assumptions. The following is the general form of the panel GARCH model used in this research: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶𝐶𝛼𝛼��𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶𝐶𝛼𝛼��� Cross-section mean return equation for 𝐴𝐴𝐴𝐴𝐴𝐴��� (7) 𝜎𝜎��� = 𝜔𝜔 𝐶 𝛼𝛼𝛼𝛼����� 𝐶𝐶 𝐶𝛽𝛽𝜎𝜎����� Cross-section variance of return equation (8) Where 𝜎𝜎��� represents the cross-section return variance (error term from the mean return equation) is determined by 𝜔𝜔 representing the constant, 𝛼𝛼����� representing the ARCH term showing the past period’s cross-section squared error term derived from the mean return equation and 𝜎𝜎����� representing the GARCH term showing the past period’s cross-section variance of the return. Risk-Return during COVID-19 The panel GARCH-in-Mean (GARCH-M) model used to measure the risk-return relationship was as follows: 𝜎𝜎��� = 𝜔𝜔 𝐶 𝜔𝜔��𝑔𝑔𝑔𝜎𝜎��) 𝐶𝐶𝛼𝛼��� (9) Where 𝑔𝑔𝑔.) is the arbitrary volatility function of 𝜎𝜎��, the cross-section GARCH-M was specified with cross-section GARCH conditional variance specification and the function 𝑔𝑔𝑔𝜎𝜎��) is the standard deviation in mean (𝜔𝜔��). Positive 𝜔𝜔�� indicates higher risk leading to higher average return and vice versa. The a priori expectation of GARCH-in-Mean is 0 > 𝜔𝜔�� > 0, indicating that the risk-return relationship can be positive or negative. Good and Bad News during COVID-19 Threshold GARCH (TGARCH/TARCH) permits asymmetric effect between positive and negative news on cross-section stock returns with general description stated below:𝐶𝜎𝜎��� = 𝜔𝜔 + ∑ 𝛼𝛼������ 𝛼𝛼����� + ∑ 𝛾𝛾������ 𝛼𝛼����� 𝑑𝑑���� + ∑ 𝛽𝛽������ 𝜎𝜎����� (10) representing the GARCH term showing the past period’s cross-section variance of the return. Risk-Return during COVID-19 The panel GARCH-in-Mean (GARCH-M) model used to measure the risk-return relationship was as follows: (𝜎𝜎��) is determined by a constant (𝜔𝜔), estimates of the previous period’s squared residual of the mean return equation (∑ 𝛼𝛼𝛼𝛼�������� ) and estimates of the previous period’s return variance (∑ 𝛽𝛽𝜎𝜎�������� ). Model Specification This study examined the risk-return relationship of returns and evaluates good and bad news during COVID-19. Variant Auto-Regressive Conditional Heteroscedasticity (ARCH) models were used to achieve the purpose of the study under three (3) distributional assumptions. The following is the general form of the panel GARCH model used in this research: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶𝐶𝛼𝛼��𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶𝐶𝛼𝛼��� Cross-section mean return equation for 𝐴𝐴𝐴𝐴𝐴𝐴��� (7) 𝜎𝜎��� = 𝜔𝜔 𝐶 𝛼𝛼𝛼𝛼����� 𝐶𝐶 𝐶𝛽𝛽𝜎𝜎����� Cross-section variance of return equation (8) Where 𝜎𝜎��� represents the cross-section return variance (error term from the mean return equation) is determined by 𝜔𝜔 representing the constant, 𝛼𝛼����� representing the ARCH term showing the past period’s cross-section squared error term derived from the mean return equation and 𝜎𝜎����� representing the GARCH term showing the past period’s cross-section variance of the return. Risk-Return during COVID-19 The panel GARCH-in-Mean (GARCH-M) model used to measure the risk-return relationship was as follows: 𝜎𝜎��� = 𝜔𝜔 𝐶 𝜔𝜔��𝑔𝑔𝑔𝜎𝜎��) 𝐶𝐶𝛼𝛼��� (9) Where 𝑔𝑔𝑔.) is the arbitrary volatility function of 𝜎𝜎��, the cross-section GARCH-M was specified with cross-section GARCH conditional variance specification and the function 𝑔𝑔𝑔𝜎𝜎��) is the standard deviation in mean (𝜔𝜔��). Positive 𝜔𝜔�� indicates higher risk leading to higher average return and vice versa. The a priori expectation of GARCH-in-Mean is 0 > 𝜔𝜔�� > 0, indicating that the risk-return relationship can be positive or negative. Good and Bad News during COVID-19 Threshold GARCH (TGARCH/TARCH) permits asymmetric effect between positive and negative news on cross-section stock returns with general description stated below:𝐶𝜎𝜎��� = 𝜔𝜔 + ∑ 𝛼𝛼������ 𝛼𝛼����� + ∑ 𝛾𝛾������ 𝛼𝛼����� 𝑑𝑑���� + ∑ 𝛽𝛽������ 𝜎𝜎����� (10) (9) Kamaldeen Ibraheem Nageri106 Where g( . ) is the arbitrary volatility function of σit, the cross-section GARCH-M was specified with cross-section GARCH conditional variance specification and the function g(σit) is the standard deviation in mean (θit). Positive θit indicates higher risk leading to higher average return and vice versa. The a priori expec- tation of GARCH-in-Mean is 0 > θit > 0, indicating that the risk-return relation- ship can be positive or negative. Good and Bad News during COVID-19 Threshold GARCH (TGARCH/TARCH) permits asymmetric effect between pos- itive and negative news on cross-section stock returns with general descrip- tion stated below: (𝜎𝜎��) is determined by a constant (𝜔𝜔), estimates of the previous period’s squared residual of the mean return equation (∑ 𝛼𝛼𝛼𝛼�������� ) and estimates of the previous period’s return variance (∑ 𝛽𝛽𝜎𝜎�������� ). Model Specification This study examined the risk-return relationship of returns and evaluates good and bad news during COVID-19. Variant Auto-Regressive Conditional Heteroscedasticity (ARCH) models were used to achieve the purpose of the study under three (3) distributional assumptions. The following is the general form of the panel GARCH model used in this research: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶𝐶𝛼𝛼��𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶𝐶𝛼𝛼��� Cross-section mean return equation for 𝐴𝐴𝐴𝐴𝐴𝐴��� (7) 𝜎𝜎��� = 𝜔𝜔 𝐶 𝛼𝛼𝛼𝛼����� 𝐶𝐶 𝐶𝛽𝛽𝜎𝜎����� Cross-section variance of return equation (8) Where 𝜎𝜎��� represents the cross-section return variance (error term from the mean return equation) is determined by 𝜔𝜔 representing the constant, 𝛼𝛼����� representing the ARCH term showing the past period’s cross-section squared error term derived from the mean return equation and 𝜎𝜎����� representing the GARCH term showing the past period’s cross-section variance of the return. Risk-Return during COVID-19 The panel GARCH-in-Mean (GARCH-M) model used to measure the risk-return relationship was as follows: 𝜎𝜎��� = 𝜔𝜔 𝐶 𝜔𝜔��𝑔𝑔𝑔𝜎𝜎��) 𝐶𝐶𝛼𝛼��� (9) Where 𝑔𝑔𝑔.) is the arbitrary volatility function of 𝜎𝜎��, the cross-section GARCH-M was specified with cross-section GARCH conditional variance specification and the function 𝑔𝑔𝑔𝜎𝜎��) is the standard deviation in mean (𝜔𝜔��). Positive 𝜔𝜔�� indicates higher risk leading to higher average return and vice versa. The a priori expectation of GARCH-in-Mean is 0 > 𝜔𝜔�� > 0, indicating that the risk-return relationship can be positive or negative. Good and Bad News during COVID-19 Threshold GARCH (TGARCH/TARCH) permits asymmetric effect between positive and negative news on cross-section stock returns with general description stated below:𝐶𝜎𝜎��� = 𝜔𝜔 + ∑ 𝛼𝛼������ 𝛼𝛼����� + ∑ 𝛾𝛾������ 𝛼𝛼����� 𝑑𝑑���� + ∑ 𝛽𝛽������ 𝜎𝜎����� (10) (𝜎𝜎��) is determined by a constant (𝜔𝜔), estimates of the previous period’s squared residual of the mean return equation (∑ 𝛼𝛼𝛼𝛼�������� ) and estimates of the previous period’s return variance (∑ 𝛽𝛽𝜎𝜎�������� ). Model Specification This study examined the risk-return relationship of returns and evaluates good and bad news during COVID-19. Variant Auto-Regressive Conditional Heteroscedasticity (ARCH) models were used to achieve the purpose of the study under three (3) distributional assumptions. The following is the general form of the panel GARCH model used in this research: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶𝐶𝛼𝛼��𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶𝐶𝛼𝛼��� Cross-section mean return equation for 𝐴𝐴𝐴𝐴𝐴𝐴��� (7) 𝜎𝜎��� = 𝜔𝜔 𝐶 𝛼𝛼𝛼𝛼����� 𝐶𝐶 𝐶𝛽𝛽𝜎𝜎����� Cross-section variance of return equation (8) Where 𝜎𝜎��� represents the cross-section return variance (error term from the mean return equation) is determined by 𝜔𝜔 representing the constant, 𝛼𝛼����� representing the ARCH term showing the past period’s cross-section squared error term derived from the mean return equation and 𝜎𝜎����� representing the GARCH term showing the past period’s cross-section variance of the return. Risk-Return during COVID-19 The panel GARCH-in-Mean (GARCH-M) model used to measure the risk-return relationship was as follows: 𝜎𝜎��� = 𝜔𝜔 𝐶 𝜔𝜔��𝑔𝑔𝑔𝜎𝜎��) 𝐶𝐶𝛼𝛼��� (9) Where 𝑔𝑔𝑔.) is the arbitrary volatility function of 𝜎𝜎��, the cross-section GARCH-M was specified with cross-section GARCH conditional variance specification and the function 𝑔𝑔𝑔𝜎𝜎��) is the standard deviation in mean (𝜔𝜔��). Positive 𝜔𝜔�� indicates higher risk leading to higher average return and vice versa. The a priori expectation of GARCH-in-Mean is 0 > 𝜔𝜔�� > 0, indicating that the risk-return relationship can be positive or negative. Good and Bad News during COVID-19 Threshold GARCH (TGARCH/TARCH) permits asymmetric effect between positive and negative news on cross-section stock returns with general description stated below:𝐶𝜎𝜎��� = 𝜔𝜔 + ∑ 𝛼𝛼������ 𝛼𝛼����� + ∑ 𝛾𝛾������ 𝛼𝛼����� 𝑑𝑑���� + ∑ 𝛽𝛽������ 𝜎𝜎����� (10) (10) Where Where 𝑑𝑑���� = � 1 𝑖𝑖𝑖𝑖 𝑖𝑖���� < 0 0, 𝑖𝑖𝑖𝑖 𝑖𝑖���� ≥ 0 When 𝑖𝑖���� indicates positive news, the effect would be given as 𝛾𝛾�� 𝑖𝑖����� , when 𝑖𝑖���� indicates negative news, the effects would be given as (𝜎𝜎+𝛾𝛾�� )𝑖𝑖���� . 𝛾𝛾�� is expected to be positive so that bad news would impact volatility. TGARCH/TARCH cross-section means return equation and cross-section return variance is stated as: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶 𝐶𝐶�� 𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶 𝑖𝑖��� Cross-section mean return equation of 𝐴𝐴𝐴𝐴𝐴𝐴��� (11) 𝜎𝜎��� = 𝜔𝜔 + 𝐶𝐶�� 𝑖𝑖����� + 𝛾𝛾�� 𝑖𝑖����� 𝑑𝑑���� + 𝛽𝛽�� 𝜎𝜎����� Cross-section return variance equation (12) Where 𝑑𝑑���� = 1 if 𝑖𝑖����� < 0 and 𝑑𝑑���� = 0 if 𝑖𝑖����� > 0. The a priori expectation of TGARCH specifies 𝛾𝛾�� < 0 as a measure of the impact of negative news on return volatility persistence. Three (3) conditional error distributions that are: Gaussian distribution, student’s-t distribution and the Generalised Error Distribution (GED) were used to estimate the parameters of the consistent residuals of the models. Estimation Procedure The estimation procedure involves the estimation of the descriptive statistics of each sectoral return series, the univariate and panel data unit-root pre-testing of the series to establish the absence of unit root were conducted. The effect of sectoral-specific was tested with the use of least squares dummy variable estimator for heteroscedasticity and autocorrelation and the Wald test statistic was employed to test the null hypothesis of the data pool ability These are done to satisfy the requirements for the use of the GARCH model in a panel data environment according to Cermeño and Grier (2001). Results and Discussion In this section, the analysis is carried out and results are presented and interpreted. Table 2. Descriptive Statistics of Selected Sectoral Share Index Return on the Nigerian Stock Exchange Stat Variable Mean Median Maximum Minimum Std. Dev. Skewness Kurtosis J.Bera Prob NSE30 0.001417 0.000421 0.060937 -0.055389 0.013190 0.407018 7.958619 261.9745 0.0000 When εit–i indicates positive news, the effect would be given as Where 𝑑𝑑���� = � 1 𝑖𝑖𝑖𝑖 𝑖𝑖���� < 0 0, 𝑖𝑖𝑖𝑖 𝑖𝑖���� ≥ 0 When 𝑖𝑖���� indicates positive news, the effect would be given as 𝛾𝛾�� 𝑖𝑖����� , when 𝑖𝑖���� indicates negative news, the effects would be given as (𝜎𝜎+𝛾𝛾�� )𝑖𝑖���� . 𝛾𝛾�� is expected to be positive so that bad news would impact volatility. TGARCH/TARCH cross-section means return equation and cross-section return variance is stated as: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶 𝐶𝐶�� 𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶 𝑖𝑖��� Cross-section mean return equation of 𝐴𝐴𝐴𝐴𝐴𝐴��� (11) 𝜎𝜎��� = 𝜔𝜔 + 𝐶𝐶�� 𝑖𝑖����� + 𝛾𝛾�� 𝑖𝑖����� 𝑑𝑑���� + 𝛽𝛽�� 𝜎𝜎����� Cross-section return variance equation (12) Where 𝑑𝑑���� = 1 if 𝑖𝑖����� < 0 and 𝑑𝑑���� = 0 if 𝑖𝑖����� > 0. The a priori expectation of TGARCH specifies 𝛾𝛾�� < 0 as a measure of the impact of negative news on return volatility persistence. Three (3) conditional error distributions that are: Gaussian distribution, student’s-t distribution and the Generalised Error Distribution (GED) were used to estimate the parameters of the consistent residuals of the models. Estimation Procedure The estimation procedure involves the estimation of the descriptive statistics of each sectoral return series, the univariate and panel data unit-root pre-testing of the series to establish the absence of unit root were conducted. The effect of sectoral-specific was tested with the use of least squares dummy variable estimator for heteroscedasticity and autocorrelation and the Wald test statistic was employed to test the null hypothesis of the data pool ability These are done to satisfy the requirements for the use of the GARCH model in a panel data environment according to Cermeño and Grier (2001). Results and Discussion In this section, the analysis is carried out and results are presented and interpreted. Table 2. Descriptive Statistics of Selected Sectoral Share Index Return on the Nigerian Stock Exchange Stat Variable Mean Median Maximum Minimum Std. Dev. Skewness Kurtosis J.Bera Prob NSE30 0.001417 0.000421 0.060937 -0.055389 0.013190 0.407018 7.958619 261.9745 0.0000 when εit–i indicates negative news, the effects would be given as Where 𝑑𝑑���� = � 1 𝑖𝑖𝑖𝑖 𝑖𝑖���� < 0 0, 𝑖𝑖𝑖𝑖 𝑖𝑖���� ≥ 0 When 𝑖𝑖���� indicates positive news, the effect would be given as 𝛾𝛾�� 𝑖𝑖����� , when 𝑖𝑖���� indicates negative news, the effects would be given as (𝜎𝜎+𝛾𝛾�� )𝑖𝑖���� . 𝛾𝛾�� is expected to be positive so that bad news would impact volatility. TGARCH/TARCH cross-section means return equation and cross-section return variance is stated as: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶 𝐶𝐶�� 𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶 𝑖𝑖��� Cross-section mean return equation of 𝐴𝐴𝐴𝐴𝐴𝐴��� (11) 𝜎𝜎��� = 𝜔𝜔 + 𝐶𝐶�� 𝑖𝑖����� + 𝛾𝛾�� 𝑖𝑖����� 𝑑𝑑���� + 𝛽𝛽�� 𝜎𝜎����� Cross-section return variance equation (12) Where 𝑑𝑑���� = 1 if 𝑖𝑖����� < 0 and 𝑑𝑑���� = 0 if 𝑖𝑖����� > 0. The a priori expectation of TGARCH specifies 𝛾𝛾�� < 0 as a measure of the impact of negative news on return volatility persistence. Three (3) conditional error distributions that are: Gaussian distribution, student’s-t distribution and the Generalised Error Distribution (GED) were used to estimate the parameters of the consistent residuals of the models. Estimation Procedure The estimation procedure involves the estimation of the descriptive statistics of each sectoral return series, the univariate and panel data unit-root pre-testing of the series to establish the absence of unit root were conducted. The effect of sectoral-specific was tested with the use of least squares dummy variable estimator for heteroscedasticity and autocorrelation and the Wald test statistic was employed to test the null hypothesis of the data pool ability These are done to satisfy the requirements for the use of the GARCH model in a panel data environment according to Cermeño and Grier (2001). Results and Discussion In this section, the analysis is carried out and results are presented and interpreted. Table 2. Descriptive Statistics of Selected Sectoral Share Index Return on the Nigerian Stock Exchange Stat Variable Mean Median Maximum Minimum Std. Dev. Skewness Kurtosis J.Bera Prob NSE30 0.001417 0.000421 0.060937 -0.055389 0.013190 0.407018 7.958619 261.9745 0.0000 is ex- pected to be positive so that bad news would impact volatility. TGARCH/TARCH cross-section means return equation and cross-section return variance is stat- ed as: Where 𝑑𝑑���� = � 1 𝑖𝑖𝑖𝑖 𝑖𝑖���� < 0 0, 𝑖𝑖𝑖𝑖 𝑖𝑖���� ≥ 0 When 𝑖𝑖���� indicates positive news, the effect would be given as 𝛾𝛾�� 𝑖𝑖����� , when 𝑖𝑖���� indicates negative news, the effects would be given as (𝜎𝜎+𝛾𝛾�� )𝑖𝑖���� . 𝛾𝛾�� is expected to be positive so that bad news would impact volatility. TGARCH/TARCH cross-section means return equation and cross-section return variance is stated as: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶 𝐶𝐶�� 𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶 𝑖𝑖��� Cross-section mean return equation of 𝐴𝐴𝐴𝐴𝐴𝐴��� (11) 𝜎𝜎��� = 𝜔𝜔 + 𝐶𝐶�� 𝑖𝑖����� + 𝛾𝛾�� 𝑖𝑖����� 𝑑𝑑���� + 𝛽𝛽�� 𝜎𝜎����� Cross-section return variance equation (12) Where 𝑑𝑑���� = 1 if 𝑖𝑖����� < 0 and 𝑑𝑑���� = 0 if 𝑖𝑖����� > 0. The a priori expectation of TGARCH specifies 𝛾𝛾�� < 0 as a measure of the impact of negative news on return volatility persistence. Three (3) conditional error distributions that are: Gaussian distribution, student’s-t distribution and the Generalised Error Distribution (GED) were used to estimate the parameters of the consistent residuals of the models. Estimation Procedure The estimation procedure involves the estimation of the descriptive statistics of each sectoral return series, the univariate and panel data unit-root pre-testing of the series to establish the absence of unit root were conducted. The effect of sectoral-specific was tested with the use of least squares dummy variable estimator for heteroscedasticity and autocorrelation and the Wald test statistic was employed to test the null hypothesis of the data pool ability These are done to satisfy the requirements for the use of the GARCH model in a panel data environment according to Cermeño and Grier (2001). Results and Discussion In this section, the analysis is carried out and results are presented and interpreted. Table 2. Descriptive Statistics of Selected Sectoral Share Index Return on the Nigerian Stock Exchange Stat Variable Mean Median Maximum Minimum Std. Dev. Skewness Kurtosis J.Bera Prob NSE30 0.001417 0.000421 0.060937 -0.055389 0.013190 0.407018 7.958619 261.9745 0.0000  Cross-section mean return equation of ASIrit (11) Where 𝑑𝑑���� = � 1 𝑖𝑖𝑖𝑖 𝑖𝑖���� < 0 0, 𝑖𝑖𝑖𝑖 𝑖𝑖���� ≥ 0 When 𝑖𝑖���� indicates positive news, the effect would be given as 𝛾𝛾�� 𝑖𝑖����� , when 𝑖𝑖���� indicates negative news, the effects would be given as (𝜎𝜎+𝛾𝛾�� )𝑖𝑖���� . 𝛾𝛾�� is expected to be positive so that bad news would impact volatility. TGARCH/TARCH cross-section means return equation and cross-section return variance is stated as: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶 𝐶𝐶�� 𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶 𝑖𝑖��� Cross-section mean return equation of 𝐴𝐴𝐴𝐴𝐴𝐴��� (11) 𝜎𝜎��� = 𝜔𝜔 + 𝐶𝐶�� 𝑖𝑖����� + 𝛾𝛾�� 𝑖𝑖����� 𝑑𝑑���� + 𝛽𝛽�� 𝜎𝜎����� Cross-section return variance equation (12) Where 𝑑𝑑���� = 1 if 𝑖𝑖����� < 0 and 𝑑𝑑���� = 0 if 𝑖𝑖����� > 0. The a priori expectation of TGARCH specifies 𝛾𝛾�� < 0 as a measure of the impact of negative news on return volatility persistence. Three (3) conditional error distributions that are: Gaussian distribution, student’s-t distribution and the Generalised Error Distribution (GED) were used to estimate the parameters of the consistent residuals of the models. Estimation Procedure The estimation procedure involves the estimation of the descriptive statistics of each sectoral return series, the univariate and panel data unit-root pre-testing of the series to establish the absence of unit root were conducted. The effect of sectoral-specific was tested with the use of least squares dummy variable estimator for heteroscedasticity and autocorrelation and the Wald test statistic was employed to test the null hypothesis of the data pool ability These are done to satisfy the requirements for the use of the GARCH model in a panel data environment according to Cermeño and Grier (2001). Results and Discussion In this section, the analysis is carried out and results are presented and interpreted. Table 2. Descriptive Statistics of Selected Sectoral Share Index Return on the Nigerian Stock Exchange Stat Variable Mean Median Maximum Minimum Std. Dev. Skewness Kurtosis J.Bera Prob NSE30 0.001417 0.000421 0.060937 -0.055389 0.013190 0.407018 7.958619 261.9745 0.0000   Cross-section return variance equation (12) Where Where 𝑑𝑑���� = � 1 𝑖𝑖𝑖𝑖 𝑖𝑖���� < 0 0, 𝑖𝑖𝑖𝑖 𝑖𝑖���� ≥ 0 When 𝑖𝑖���� indicates positive news, the effect would be given as 𝛾𝛾�� 𝑖𝑖����� , when 𝑖𝑖���� indicates negative news, the effects would be given as (𝜎𝜎+𝛾𝛾�� )𝑖𝑖���� . 𝛾𝛾�� is expected to be positive so that bad news would impact volatility. TGARCH/TARCH cross-section means return equation and cross-section return variance is stated as: 𝐴𝐴𝐴𝐴𝐴𝐴��� = 𝐶𝐶 𝐶 𝐶𝐶�� 𝐴𝐴𝐴𝐴𝐴𝐴����� 𝐶 𝑖𝑖��� Cross-section mean return equation of 𝐴𝐴𝐴𝐴𝐴𝐴��� (11) 𝜎𝜎��� = 𝜔𝜔 + 𝐶𝐶�� 𝑖𝑖����� + 𝛾𝛾�� 𝑖𝑖����� 𝑑𝑑���� + 𝛽𝛽�� 𝜎𝜎����� Cross-section return variance equation (12) Where 𝑑𝑑���� = 1 if 𝑖𝑖����� < 0 and 𝑑𝑑���� = 0 if 𝑖𝑖����� > 0. The a priori expectation of TGARCH specifies 𝛾𝛾�� < 0 as a measure of the impact of negative news on return volatility persistence. Three (3) conditional error distributions that are: Gaussian distribution, student’s-t distribution and the Generalised Error Distribution (GED) were used to estimate the parameters of the consistent residuals of the models. Estimation Procedure The estimation procedure involves the estimation of the descriptive statistics of each sectoral return series, the univariate and panel data unit-root pre-testing of the series to establish the absence of unit root were conducted. The effect of sectoral-specific was tested with the use of least squares dummy variable estimator for heteroscedasticity and autocorrelation and the Wald test statistic was employed to test the null hypothesis of the data pool ability These are done to satisfy the requirements for the use of the GARCH model in a panel data environment according to Cermeño and Grier (2001). Results and Discussion In this section, the analysis is carried out and results are presented and interpreted. Table 2. Descriptive Statistics of Selected Sectoral Share Index Return on the Nigerian Stock Exchange Stat Variable Mean Median Maximum Minimum Std. Dev. Skewness Kurtosis J.Bera Prob NSE30 0.001417 0.000421 0.060937 -0.055389 0.013190 0.407018 7.958619 261.9745 0.0000 The a priori expectation of TGARCH specifies γit < 0 as a measure of the impact of negative news on return volatility persistence. risk-return relAtionshiP in the nigeriAn stock mArket… 107 Three (3) conditional error distributions that are: Gaussian distribution, student’s-t distribution and the Generalised Error Distribution (GED) were used to estimate the parameters of the consistent residuals of the models. Estimation Procedure The estimation procedure involves the estimation of the descriptive statistics of each sectoral return series, the univariate and panel data unit-root pre-test- ing of the series to establish the absence of unit root were conducted. The effect of sectoral-specific was tested with the use of least squares dummy variable es- timator for heteroscedasticity and autocorrelation and the Wald test statistic was employed to test the null hypothesis of the data pool ability These are done to satisfy the requirements for the use of the GARCH model in a panel data en- vironment according to Cermeño and Grier (2001). Results and Discussion In this section, the analysis is carried out and results are presented and inter- preted. Table 2. Descriptive Statistics of Selected Sectoral Share Index Return on the Nigerian Stock Exchange Stat Variable Mean Median Maximum Minimum Std. Dev. Skewness Kurtosis J.Bera Prob NSE30 0.001417 0.000421 0.060937 -0.055389 0.013190 0.407018 7.958619 261.9745 0.0000 NSE50 0.001051 0.000251 0.199894 -0.116283 0.021080 2.683139 37.44343 12759.04 0.0000 NSEBAN 0.000743 0.000000 0.079884 -0.127868 0.026560 -0.595542 6.472291 139.8081 0.0000 NSECON 0.000288 -0.000017 0.225520 -0.174956 0.029467 1.863450 34.07173 10160.68 0.0000 NSEIND 0.002787 0.000036 0.086508 -0.075006 0.019561 0.630845 7.282481 206.7894 0.0000 NSEINS 0.001771 0.002363 0.061560 -0.056696 0.015838 -0.006462 4.693598 29.76007 0.0000 NSEISL 0.001849 0.000381 0.054492 -0.044845 0.013018 0.507936 6.914825 169.7127 0.0000 NSEOIL -0.000512 0.000000 0.052420 -0.056970 0.013119 -0.827499 9.643415 486.3175 0.0000 NSEPEN 0.001234 0.000000 0.096402 -0.089618 0.015971 0.453358 13.09519 1065.876 0.0000 Kamaldeen Ibraheem Nageri108 Stat Variable Mean Median Maximum Minimum Std. Dev. Skewness Kurtosis J.Bera Prob NSEPRE 0.020182 0.000161 1.036042 -0.502876 0.212936 3.041523 17.78441 2651.665 0.0000 NSEALL 0.003081 0.000028 1.023604 -0.502876 0.069947 9.227151 160.4829 2608424 0.0000 S o u r c e : authors computation, 2020. Table 2 indicates that the mean return of the sectoral indices is positive except for the mean return of NSEOIL while the median return of NSECON is nega- tive and all other sectoral median returns are positive. All the series exhibit positive maximum returns with all showing negative minimum returns during the COVID-19 year. The standard deviations of all the return series are low at a maximum of 3% except for NSEPRE with the Jarque-Bera P-values are all sta- tistically significant at 5% indicating a normal distribution of the return series. The pooled series (NSEALL) has a positive mean and median return, positive maximum return and negative minimum return with 6% standard deviation and a Jarque-Bera P-value of less than 5% significant level indicating a normal distribution of the NSEALL series. Table 3. Unit Root Test Result of Selected NSE Indices Series Variable Method Stat Prob. Method Stat Prob. NSE30R ADF Test Stat -7.403898 0.0000 PP Test Stat -11.34105 0.0000 NSE50R ADF Test Stat -16.47200 0.0000 PP Test Stat -16.46769 0.0000 NSEBANR ADF Test Stat -13.07916 0.0000 PP Test Stat -12.99907 0.0000 NSECONR ADF Test Stat -19.04778 0.0000 PP Test Stat -19.08406 0.0000 NSEINDR ADF Test Stat -13.22865 0.0000 PP Test Stat -13.53596 0.0000 NSEINSR ADF Test Stat -18.09120 0.0000 PP Test Stat -17.97058 0.0000 NSEISLR ADF Test Stat -7.747417 0.0000 PP Test Stat -12.10292 0.0000 NSEOILR ADF Test Stat -13.96485 0.0000 PP Test Stat -13.98646 0.0000 NSEPENR ADF Test Stat -13.71659 0.0000 PP Test Stat -13.80775 0.0000 NSEPRER ADF Test Stat -16.46982 0.0001 PP Test Stat -21.72942 0.0000 S o u r c e : authors computation, 2020. Table 2. Descriptive… risk-return relAtionshiP in the nigeriAn stock mArket… 109 Table 4. Panel Unit Root Test Result of NSEALL Variable Methods Stat Prob. NSEALL Levin, Lin and Chu Im, Pesaran and Shin ADF PP -9.76184 -19.2477 388.829 1122.30 0.0000 0.0000 0.0000 0.0000 S o u r c e : authors computation, 2020. Table 3 and 4 are the unit root test results for the series, the ADF and PP unit root test result for the individual ASI sectorial returns indicates no unit root while the cross-section NSEALL unit root result also indicates stationarity. Therefore, the data is suitable for econometrics analysis. Cross-section regression requires the establishment of the pool-ability of the data to know if there exists no sectorial specific effect in the data and ap- plicability of single intercept. To test for homogeneity (common intercept), the Least Square Dummy Variable (LSDV) according to Cermeño and Grier (2001) was used, the Auto-Regressive Conditional Heteroscedasticity (ARCH) effect test was used to test for ARCH effect and serial correlation was tested using the Ljung–Box Q-statistics and partial correlations tests for the residuals and squared residuals of the mean and variance equations. Table 5 presents the Wald test F-statistics with Chi-Square values of the LSDV cross-section mean equation were 0.289782 and 1.876113 respectively and they are not statistically significant, indicating that there exists homoge- neity (common intercept) within the sectorial return series. The F-statistics and the observed R-square values of the ARCH test from the pooled returns in- dicate that the null hypothesis of no ARCH effect is rejected indicating the pres- ence of ARCH effect in the residuals of the pooled mean equation. Table 5. Wald Test (Mean Equation) and ARCH Test (Pooled Regression) Wald Test (Mean Equation) Value Prob Heteroscedasticity Test (Pooled Regression) Value Prob F-statistics 0.289782 0.8914 F-statistics 132.8361 0.0000 Obs*R-square 1.876113 0.8914 Obs*R-square 126.1997 0.0000 S o u r c e : authors computation, 2020. Kamaldeen Ibraheem Nageri110 Table 6. Autocorrelation Result Residual PAC Q-Stat Prob Squared Residual PAC Q-Stat Prob -0.008 0.0216 0.002 0.225 126.40 0.0000 0.108 14.334 0.001 0.158 188.77 0.0000 0.035 17.747 0.002 0.112 219.80 0.0000 -0.083 16.680 0.002 0.081 236.30 0.0000 S o u r c e : authors computation, 2020. Table 6 shows the autocorrelation result of the pooled regression for the residu- al and the squared residuals from the mean equation and it indicates that there is no serial correlation in the residuals. The result of no serial correlations and the presence of the ARCH effect indicates the application of the GARCH model. Lastly, the cross-section variance equation was tested for individual effect in the NSE sectoral series using the Maximum log-likelihood estimate (MLE) as suggested by Cermeño and Grier (2001) and the MLE is statistically significant at 5% which shows that the sectorial return variance is not consistent between the market. Therefore, the GARCH model applies to the panel data for analysis. Table 7. Cross-Section GARCH-in-Mean Result for Sectorial Return in COVID-19 Parameters Gaussian Distribution Estimates P-Value Student’s-t Distribution Estimates P-Value Generalised Error Distribution Estimates P-Value C -0.007504 0.0000 -0.006850 0.4835 -0.056134 0.1859 θit 0.423549 0.0000 0.113079 0.4835 0.964359 0.1860 ω 0.000079 0.0000 0.005454 0.5914 0.001733 0.0000 αit 0.164938 0.0000 -0.004056 0.6061 0.000017 0.0015 βit 0.739167 0.0000 -0.263666 0.0382 0.488494 0.0000 AIC -4.667309 -5.528884 -5.658158 SC -4.653280 -5.512517 -5.641791 HQ -4.662214 -5.522941 -5.652215 S o u r c e : authors computation, 2020. risk-return relAtionshiP in the nigeriAn stock mArket… 111 Table 7 shows the GARCH-in-Mean cross-section return indicating positive θit (standard deviation) of 0.423549, 0.113079 and 0.964359 under the three error distributional assumptions respectively with P-values of 0.0000, 0.4835 and 0.1860 respectively indicating statistical significance at 5% under Gauss- ian distribution estimates but statistically insignificant under the Student’s-t distribution and the Generalised Error Distributions (GED). This implies a positive risk-return relationship in the cross-section return on the Nigerian Stock Exchange sectors during COVID-19. Higher risk leads to higher return and vice versa. The Akaike Information Criterion (AIC), Schwarz Criterion (SC) and the Hannan-Quinn Criterion show that the GED value is the lowest indicating its superior predictive ability of GARCH-in-Mean model esti- mate of cross-section risk-return relationship during COVID-19. This implies that the risk-return premium of the stock on the selected sectoral stocks is not risky to hold during COVID-19. The variance equation indicates αit and βit representing the ARCH term (past day’s return squared residual) and the GARCH term (past day’s variance of return) respectively with ω as the constant is positive and significant at 5% under Gaussian and GED distributions. This indicates that past day return and risk (variance) are significant and positive under the Gaussian and GED dis- tributional assumptions. The past day’s return squared residual is negative and statistically insignificant under the Student’s-t distribution while the past day’s variance of return and the constant are negative and statistically signifi- cant under the Student’s-t distribution. Table 8. ARCH Effect and Autocorrelation Result of GED GARCH-in-Mean Model Heteroscedasticity Test (Pooled Regression) Value Prob Squared Residual PAC Q-Stat Prob F-statistics 0.100427 0.8178 -0.148 54.105 0.918 Obs*R-square 0.100444 0.8177 -0.103 80.888 0.499 -0.100 105.17 0.547 -0.052 112.24 0.691 S o u r c e : authors computation, 2020. The ARCH effect and autocorrelation result in table 8 is the diagnostic result of the GED estimate suggested by the criterion as the best estimate. The result Kamaldeen Ibraheem Nageri112 shows that the null hypothesis of no ARCH effect and no serial correlation can- not be rejected, which confirms that the model is good and desirable for policy consideration and implementation. Table 9. Cross-Section TGARCH/TARCH Result for Sectorial Return in COVID-19 Parameters Gaussian Distribution Estimates P-Value Student-t Distribution Estimates P-Value Generalised Error Distribution Estimates P-Value ω 0.000063 0.0000 0.104493 0.0070 0.000085 0.0000 αit 0.072421 0.0000 1455.982 0.0094 1.058086 0.0000 γit 0.147225 0.0000 -383.8023 0.1403 0.140300 0.6700 βit 0.785788 0.0000 0.131148 0.0000 0.514905 0.0000 AIC -4.680251 -5.662720 -5.767163 SC -4.666223 -5.646353 -5.750796 HQ -4.675157 -5.656777 -5.761220 S o u r c e : authors computation, 2020. The TGARCH results of the cross-section return on the Nigerian Stock Ex- change during the COVID-19 half-year is shown in table 9. The value of γit un- der the Gaussian Distribution is positive and significant while γit is negative and insignificant under Student’s-t while γit is positive and insignificant under GED distributional assumptions. The TGARCH model specifies the estimates of γit < 0 and significant to show that bad news has more impact on return vol- atility than the good news of the same extent. Therefore, the cross-section vol- atility of return reacted more to good news than to bad news of equal extent on the Nigerian Stock Exchange during COVID-19 as shown by all the distribu- tional assumptions. The variance equation indicates that αit and βit representing the ARCH term (past day’s squared residual return) and the GARCH term (past day’s variance of the return) respectively and ω as the constant, are positive and significant at 1% as shown by all the distributional assumptions. The best-fitted estimates according to the estimates of the AIC, SIC and HQ selection criteria indicate that the GED value is the lowest indicating its supe- risk-return relAtionshiP in the nigeriAn stock mArket… 113 rior predictive ability of TGARCH estimate of news impact on cross-section re- turns during COVID-19. This is in agreement with the finding of Nageri (2019b). Table 10. ARCH Effect and Autocorrelation Result of GED TGARCH/TARCH Model Heteroscedasticity Test (Pooled Regression) Value Prob Squared Residual PAC Q-Stat Prob F-statistics 0.005834 0.9391 -0.002 0.0058 0.939 Obs*R-square 0.005839 0.9391 0.004 2.0487 0.976 -0.002 3.0607 0.996 -0.002 0.0727 0.999 S o u r c e : authors computation, 2020. Table 10 is the diagnostic result of the ARCH effect and autocorrelation result of the GED suggested by the criterion as the best estimate. The result specifies that the null hypothesis of no ARCH effect and no serial correlation cannot be rejected, which shows that the model is good and desirable for policy consid- eration and implementation.  Conclusion and Recommendations The study examined the risk-return relationship during COVID-19 on the Ni- gerian Stock Exchange (NSE) using panel data of ten (10) sector index returns. Panel GARCH methodology of GARCH-in-Mean and the TGARCH models were used for the analysis. The mean and variance equations were developed in a panel form as suggested by Cermeño and Grier (2001). Findings indicate that the return of selected ten sectorial returns exhib- its a positive risk-return relationship during the period under consideration, showing that the assets are not too risky to hold. This is in agreement with the conventional view that the higher the return, the higher the risk. On the other hand, stock returns respond to good news more than they do to the bad news of a similar extent on the Nigerian Stock Exchange during COVID-19 and current- day returns respond to past returns during the period. This indicates that the Nigerian stock market is resilient and was able to resist the impact of COVID-19 contrary to findings in the advanced stock market. The reason for this may be connected to the various policy responses of the financial sector regulatory Kamaldeen Ibraheem Nageri114 authorities such as the Central Bank of Nigeria (CBN) and Securities and Ex- change Commission (SEC) to cushion the effects of the pandemic. The top fifty listed companies, listed banks, listed insurance companies and the top forty listed have remained more attractive to investors. The performance of the oth- er sectors (consumer goods, industrial sector, shari’ah compliant companies, companies listed on the premium board and the top thirty listed companies) can be attributed to high inf lation, loss of income and growing unemployment. The oil and gas sector can be attributed to the high volatility in the price of oil throughout the world. Finally, the study suggests that policymakers should be more sincere, in- genuine and pay more attention to the Nigerian stock market for opportunities abound in the market for post-COVID-19 economic recovery and to leverage on the ability of the market to resist the pandemic to encourage more listing on the stock market.  References Alade, E.M., Adeusi, A.S., & Alade, O.F. (2020). Covid-19 pandemic and Nigerian stock market capitalization. Ilorin Journal of Economic Policy, 7(3), 12-23. Ashraf, B.N. (2020). Stock markets’ reaction to COVID-19: Cases or fatalities? Research in International Business and Finance, 54, 101249. http://dx.doi.org/10.1016/j.rib- af.2020.101249. Azimli, A. (2020). The impact of COVID-19 on the degree of dependence and structure of risk - return relationship: A quantile regression approach. Finance Research Let- ters, 36, 101648. http://dx.doi.org/10.1016/j.frl.2020.101648. Bollerslev, T. (1986). Generalised autoregressive conditional heteroscedasticity. Journal of Econometrics, 31(3), 307-327. http://dx.doi.org/10.1016/0304-4076(86)90063-1. Brandt, M.W., & Kang, Q. (2004). On the relationship between the conditional mean and volatility of stock returns: A latent VAR approach. Journal of Financial Economics, 72(2), 217-257. Campbell, J.Y., Lettau, M., Malkiel, B.G., & Xu, Y. (2001). Have individual stocks become more volatile? An empirical exploration of idiosyncratic risk. Journal of Finance, 56(1), 1-43. http://dx.doi.org/10.1111/0022-1082.00318. Cermeño, R., & Grier, K.B. (2001). Modelling GARCH processes in Panel Data: Theory, Simulations and Examples. LACEA 2001: Annual Conference. Montevideo, Uruguay. Coffie, W. (2015). Modelling and forecasting the conditional heteroskedasticity of stock returns using asymmetric models: Empirical evidence from Ghana and Nigeria. Journal of Accounting and Finance, 15(5), 109-123 Engle, R.F. (1982). Autoregressive conditional heteroskedasticity with estimates of the variance of United Kingdom inf lation. Econometrica, 50(4), 987-1007. risk-return relAtionshiP in the nigeriAn stock mArket… 115 Fisher, I.N., & Hall, G.R. (1969). Risk and corporate rate of return. Quarterly Journal of Economics, 83(1), 79-92. http://dx.doi.org/10.2307/1883994. Ghysels, E., Plazzi, A., & Valkanov, R. (2016). The risk-return relationship and financial crises. SSRN Working Paper, 1-19. http://dx.doi.org/10.2139/ssrn.2776702. Glosten, L.R., Jagannathan, R., & Runkle, D.E. (1993). On the relation between the ex- pected value and the volatility of the nominal excess return on stocks. Journal of Finance, 48(5), 1779-1801. http://dx.doi.org/10.1111/j.1540-6261.1993.tb05128.x. Guo, Y.R., Cao, Q.D., Hong, Z.S., Tan, Y.Y., Chen, S.D., Jin, H.J., Tan, K.S., Wang, D.E., & Yan, Y. (2020) The origin, transmission and clinical therapies on coronavirus disease 2019 (COVID-19) outbreak – an update on the status. Military Medical Research, 7(11), 1-10. http://dx.doi.org/10.1186/s40779-020-00240-0. Guo, H., & Whitelaw, R.F. (2006). Uncovering the risk-return relation in the stock market. Journal of Finance, 61(3), 1433-1463. http://dx.doi.org/10.1111/j.1540- 6261.2006.00877.x. Hu, B., Guo, H., Zhou, P., & Shi, Z. L. (2020). Characteristics of SARS-CoV-2 and COVID-19. Nature Reviews Microbiology, 19, 141-154. http://dx.doi.org/10.1038/s41579-020- 00459-7. International Monetary Fund (IMF) (2020a). World economic outlook 2020: The great lockdown, https://www.imf.org/en/Publications/WEO/Issues/2020/04/14/weo- april-2020 (accessed: 12.06.2020). International Monetary Fund (IMF) (2020b). Global Financial Stability Report, https:// www.imf.org/en/Publications/GFSR (accessed: 12.06.2020). Leon, A., Nave, J., & Rubio, G. (2005). The relationship between risk and expected re- turn in Europe. Journal of Banking and Finance, 31(2), 495-512. http://dx.doi. org/10.1016/j.jbankfin.2006.07.011. Lettau, M., & Ludvigson, S. (2003). Measuring and modeling variation in the risk-return trade-off. In Y. Aït-Sahalia, L.P. Hansen (Eds.). Handbook of Financial Econometrics. North Holland: Elsevier Science B.V. Ludvigson, S.C., & Ng, S. (2007). The empirical risk-return relation: A factor analysis ap- proach. Journal of Financial Economics, 83(1), 171-222. http://dx.doi.org/10.1016/j. jfineco.2005.12.002. Mahmood, I., & Shah, S.Z.A. (2015). Negative relationship between risk and return: A contrary view. Pakistan Journal of Commerce and Social Sciences, 9(2), 336-343. Merton, R. (1973). An intertemporal capital asset pricing model. Econometrica, 41(5), 867-887. Nageri, K.I. (2019a). Evaluating good and bad news during pre and post financial melt- down: Nigerian stock market evidence. Studia Universitatis Babeș-Bolyai Oeconomi- ca, 64(3), 1-22. http://dx.doi.org/10.2478/subboec-2019-0012. Nageri, K.I. (2019b). Evaluating volatility persistence of stock return in the pre and post 2008-2009 financial meltdown. Copernican Journal of Finance & Accounting, 8(3), 73-91. http://dx.doi.org/10.12775/CJFA.2019.013. Nelson, D.B. (1991). Conditional heteroskedasticity in asset returns: A new approach. Econometrica, 59(2), 347-370. Kamaldeen Ibraheem Nageri116 Neumann, M., Bobel, I., & Haid, A. (1979). Profitability, risk and market structure in West German industries. Journal of Industrial Economics, 27(3), 227-242. Pastor, L., & Veronesi, P. (2006). Was there a Nasdaq bubble in the late 1990’s? Journal of Financial Economics, 81(1), 61-100. Raputsoane, L. (2009). The risk-return relationship in the South Africa stock market. Paper presented at the 14th Annual Conference of the African Econometrics Society on Econometric Modelling in Africa, Abuja, Nigeria, 1-13. Salisu, A.A., Ebuh, G.U., & Usman, N. (202). Revisiting oil-stock nexus during COVID-19 pandemic: Some preliminary results. International Review of Economics & Finance, 69, 280-294. http://dx.doi.org/10.1016/j.iref.2020.06.023. Salisu, A.A., & Vo, X.V. (2020). Predicting stock returns in the presence of COVID-19 pandemic: The role of health news. International Review of Financial Analysis, 71, 101546. http://dx.doi.org/10.1016/j.irfa.2020.101546. The Nigerian Stock Exchange (NSE) (2020). Nigerian Stock Exchange, https://www.nse. com.ng. World Health Organisation (WHO) (2020). WHO Director-General’s opening remarks at the media briefing on COVID-19, https://www.who.int/director-general/speeches/ detail/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid- 19---11-march-2020 (accessed: 03.07.2020). Zhang, D., Hu, M., & Ji, Q. (2020). Financial markets under the global pandemic of COVID-19. Finance Research Letters, 36, 101528. http://dx.doi.org/10.1016/j. frl.2020.101528.