Model Specification

This sub-section will present the baseline model specification for the panel data regression. As earlier mentioned, the purpose of introducing a panel regression is to analyze the factors determining stock return fluctuations during the event period. This sub-section will start by identifying the dependent variable, which we aim to explain. After that, the independent variables chosen to explain the dependent variable will be identified. Lastly, the final baseline model specification and additional sub-specifications will be defined. The data applied for each regression variable and their measurements will be presented in section 6.1.4.

5.2.1.1. Dependent Variable

The dependent variable has to be a measure of stock return movements. Furthermore, as we aim to analyze the difference in stock performance across sectors, the dependent variable must be on company level. Therefore, cumulative abnormal returns (CAR) have been chosen as the dependent, defined in section 5.1.6.1.

5.2.1.2. Independent Variables

The purpose of the independent variables is to explain the variation in the dependent variable, CAR. Similarly, the selection of the independent variables depends on how each variable is expected to explain the fluctuation in CAR during the event period.

COVID– 19 cases:

As previously mentioned in section 2.1.3, the first COVID-19 cases in Scandinavia (second in the case of Sweden) were followed by a rapid increase in new cases. The outbreak and spread of COVID-19 created fear and uncertainty across the Scandinavian countries, as the severe consequences were evident from other countries, e.g., Italy. The increasing spread of the virus indicated a higher probability of severe consequences and the continuation of these. Hence, creating more uncertainty and depressing private consumers’

expectations for the future. Therefore, the continuation and severity of the pandemic are expected to play a significant role in investor behavior and thus abnormal returns. The number of new COVID-19 cases is therefore selected as an independent variable in the explanation of CAR. The number of new cases gives the public, companies, and governments an indication of how the pandemic is evolving and what could be expected in the long run. Therefore, it would be expected that increasing COVID-19 cases will lead to a significant decrease in CAR. However, it should be noted that specific sectors might have benefitted from the pandemic, e.g., the Technology sector, and the continuation of the pandemic will thus have the opposite effect in these sectors.

Restrictions:

As earlier mentioned, to limit the coronavirus’s impact on public health, local governments took extraordinary measures to control the spread of the disease. However, these restrictive measures were at the expense of disrupting economic activity and induced crucial challenges for companies. As the restrictions included limitations on public gatherings, traveling, and total or partial lockdowns, companies faced difficulties supplying their products and services to consumers. Therefore, the restrictions implemented by the governments are expected to have a significant impact on private investors’ expectations for companies’

performance, and thus CAR. However, the impact of introducing a federal restriction on CAR is ambiguous.

On one side, introducing new restrictions is expected to decrease CAR, as it indicates significant challenges and disruptions for companies. On the other side, the implementation of new restrictions indicates that the spread of the disease is being controlled and thus that the pandemic will not last long. Hence, it could also be expected that the restrictions have had a positive impact on CAR. The announcement of new national restrictions is therefore chosen as an independent variable.

Fiscal policy measures:

To mitigate the restrictions’ negative impact on the economy, the local governments introduced extraordinary fiscal policy measures (section 2.2.6.1). As earlier mentioned, the fiscal policies included, among others, stimulus packages, salary compensation, and extraordinary loan opportunities targeting both companies and private households. As the fiscal policy measures assist companies through the pandemic’s financial challenges, introducing a fiscal policy measure is expected to influence investors’ assumptions regarding if companies will overcome the pandemic. Therefore, COVID-19 related fiscal policy announcements are selected as an independent variable for the explanation of CAR. Hereunder, it would be expected that a new extraordinary fiscal policy measure will increase abnormal returns, as these measures should make investors believe that it will be easier for companies to survive the pandemic.

Monetary policy measures:

Complementary to the governments’ extraordinary fiscal policy measures, the local central banks also introduced extraordinary monetary policies. As previously mentioned in section 2.2.6.2, the central banks mainly implemented monetary policies to alleviate companies’ liquidity issues during the pandemic. Similar to the fiscal policy measures, the monetary policy measures are also expected to have helped companies through the difficulties caused by the pandemic. Hence, a new monetary policy measure is expected to impact investors’ view of the future positively and likewise positively affect CAR. Announcements of COVID-19 related monetary policy measures are therefore included as an explanatory variable in the model.

Volatility Index (VIX):

As most Scandinavian companies work internationally, they are also dependent on the economies of other countries besides their local countries. As COVID-19 is a global pandemic, the measures taken to limit the spread of the virus and subsequent economic recession are global phenomena. Therefore, the global market’s uncertainty during the first wave is expected to significantly impact Scandinavian companies’ performance.

More specifically, it is expected that increased expected volatility in the global stock market leads to a decrease in CAR for Scandinavian companies, as it indicate fear amongst investors. Therefore, the CBOE Volatility Index (VIX) will be included in the baseline model to measure the global volatility.

Sector dummies:

Additionally, to test if there is a significant difference between stock performance across sectors, sector dummies will be included in the model. The sector dummies will indicate how a sector has been impacted by the pandemic relative to other sectors. Hence, this variable will test main hypothesis 2. It is hereunder expected that the Consumer Staples and Technology and Telecommunications sectors are some of the sectors that performed best during the pandemic, as they benefitted from the COVID-19 induced market conditions.

Contrary, the Consumer Discretionary sector is expected to have performed significantly worse than several other sectors, as the introduced restrictions had severe consequences for this sector.

Country dummies:

Finally, country dummies for Denmark, Norway, and Sweden will be included in the panel data regression.

Including the country dummies enable testing whether there was a significant difference between the CARs in the Scandinavian countries. The difference in stock performance across the three countries is expected to be significant, given the different strategies and different economic impacts. More specifically, it could be expected that the Swedish stock market have performed significantly different than its neighbor markets, due to the country’s different strategies towards the pandemic. Hence, this variable will test main hypothesis 3.

5.2.1.3. Final Baseline Model Specification

Given the above selected dependent variable and independent variables, the baseline model specification is the following:

𝐶𝐴𝑅_#,% = 𝛽_*+ 𝛽_!𝐶𝑜𝑣𝑖𝑑19𝑐𝑎𝑠𝑒𝑠_5,%+ 𝛽_"𝑀𝑜𝑛𝑒𝑡𝑎𝑟𝑦𝑝𝑜𝑙𝑖𝑐𝑖𝑒𝑠_5,%+ 𝛽₆𝐹𝑖𝑠𝑐𝑎𝑙𝑝𝑜𝑙𝑖𝑐𝑖𝑒𝑠_5,%+ 𝛽₇𝑅𝑒𝑠𝑡𝑟𝑖𝑐𝑡𝑖𝑜𝑛𝑠_5,%

+ 𝛽₈𝑉𝐼𝑋_%+ 𝛿₂+ 𝛿₊+ 𝛿_#,%+ 𝑢_#,+,% (14) Where 𝛽_* is the intercept, which is included depending on the choice of estimation model, as some panel data estimation models are unable to estimate time-invariant components. 𝐶𝑜𝑣𝑖𝑑19𝑐𝑎𝑠𝑒𝑠_5,% is the number of new COVID-19 cases in a given country c at day t. 𝑀𝑜𝑛𝑒𝑡𝑎𝑟𝑦𝑝𝑜𝑙𝑖𝑐𝑖𝑒𝑠_5,% is a dummy variable indicating if there has been a monetary policy announcement in a given country c at a given day t. 𝐹𝑖𝑠𝑐𝑎𝑙𝑝𝑜𝑙𝑖𝑐𝑖𝑒𝑠_5,% is a dummy variable indicating if there has been a fiscal policy announcements for given country c at day t. Similarly, 𝑅𝑒𝑠𝑡𝑟𝑖𝑐𝑡𝑖𝑜𝑛𝑠_5,% is a dummy variable indicating if there has been a restriction announcement in country c at day t. 𝑉𝐼𝑋_% is the volatility index at time t. 𝛿₂ and 𝛿₊ are vectors of sector dummies and country dummies, respectively. Additionally, in order to control for time-varying specific effects a vector of company-date dummies is included in the model. Lastly, 𝑢_#,+,% is the idiosyncratic error term. The measurement of each variable will be presented in chapter 6.

As earlier mentioned in section 2.1, the three Scandinavian countries handled the first wave of the pandemic quite differently. On the one hand, the Swedish government had a more simple strategy towards the pandemic, as milder restrictions and a broad range of monetary and fiscal policies were introduced. On the other hand, the Danish and Norwegian countries introduced stricter restrictions, a broad range of fiscal policies, and only a few monetary policy measures. Hereunder, the severity of the pandemic also differed across the countries, and thus also the impact of it. Therefore, considering the difference of the pandemic and cultural differences across countries, it could be expected that investors might have reacted differently in each country. To examine whether investors on the three Scandinavian stock markets have reacted differently during the pandemic, we also run three sub-regressions for each country. Hence, the baseline model specification will additionally be estimated for Denmark, Sweden, and Norway separately.