Data Analysis for Business

ASSESSMENT 2 GROUP REPORT

BEO6000 Data Analysis for Business

Victoria University

AssignmentTutorOnline

Table of Contents

Executive summary

We sought to gain a better understanding of the impact of COVID-19 on public company turnover in North America and Oceania by studying similarities between the regions and the industries most adversely affected. Statistical analysis enabled us to conclude that several of our hypotheses were incorrect and that very diverse results were reported upon comparing industries within and across regions, supporting the OECD’s assumption of the non-homogeneity of the economic COVID-19 experiences.

We established that many media indicators of similarity are misleading and lead to bias. We questioned the narrowness of our lens and whether qualitative analysis may be employed in future research to gain a deeper understanding under these very volatile conditions. We conclude that whilst vast amounts of data are available, knowledge and interpretations of the data has seemingly not been shared and responded to at an equal pace and transparency, hence the divergence in data we have established contributed to economic impact which has been more severe in some industries and regions.

Introduction

Our research below sets out our research objectives regarding the assessment of the adverse impact of COVID-19 on publicly listed companies from a regional and industry similarity perspective. We detail hypotheses per our chosen statistical methods. Our rationale for the statistical methods deployed as determined by the normality of the data, variables and questions are explained. Set procedures were followed to screen and clean our data in preparation for parametric and non-parametric tests; Wilcoxon and Kruskal-Wallis tests.

We set out our methodology in collecting our data turnover data, and the chosen data, sample size and variable structure and grouping variables best suited to our investigation. A discussion of our findings, limitations and recommendations for future research and improved methods may impact our assessment of the homogeneity of the data.

Research objectives : main objective and sub-objectives.

The main objective of the research is to ascertain whether the impact of COVID-19 ( measured in percentage change in turnover from calendar year 2019 to 2020 ) was similar in North America and Oceania across 3 sectors assumed to be most affected.

The sub-objective is to ascertain whether selected industries in North America were impacted in a similar way in Oceania i.e., research whether the segment affected most severely was the Hotel and Accommodation industry.

Lastly, we wanted to assess whether the impact of COVID-19 was similar across all selected industries within North America and also Oceania.

Hypothesis statements for testing

Null Hypothesis: The distribution of differences in operating revenue (turnover) for 3 selected industries in North America and Oceania is not symmetric around zero.

Alternative Hypothesis : The distribution of differences in operating revenue (turnover) for 3 selected industries in North America and Oceania is symmetric around zero.

Null Hypothesis : The impact of COVID-19 on the Hotel and Accommodation industry in North America was similar to the Oceania Market.

Alternative Hypothesis : The impact of COVID-19 on the Hotel and Accommodation industry turnover in North America was not similar to the Oceanian Market.

Null Hypothesis : That the distribution of percentage change in turnover due to the impact of Covid-19 in all three industries in North America are identical.

Alternative Hypothesis : That at least one of the industries is different.

Null Hypothesis : That the distribution of percentage change in turnover due to the impact of Covid-19 in all three industries in Oceania are identical.

Alternative Hypothesis : That at least one of the industries is different.

Methodology

Methodology in data collection : identification and explanation of rationale for the selected data collection method

We accessed the Orbis database to extract turnover figures for years 2019 to 2020 to determine percentage decline in 3 sectors most affected by Covid-19 in the North America geographic region : Automotive, Food and Beverage and Hotel Accommodation

Additionally, we accessed Orbis to extract turnover figures for years 2019 to year 2020 to determine percentage decline in 3 sectors most affected by Covid in Oceania. Our initial intention was to obtain deeper data from the Morningstar database: unfortunately, the sample size for Australian listed companies in all three of the sectors was insufficient, hence, we used Orbis data for both geographical regions.

In setting the sample size we needed to ensure that the sample size was representative and credible ( Hair et al 2019). We chose:

30 listed companies in the Hotel and accommodation sectors in North America and Oceania

30 listed companies in the Automotive and related sector in North America and Oceania

30 listed companies in the Food and Beverage Sector in North America and Oceania

We selected slightly more companies from the dataset to ensure we reached the target of 30 after data cleansing. The table below illustrates the final sample size selected after cleansing.

Diagram 1 : Data sample size from each region

The Data Structure consists of a primary variable (the dependent variable) for investigation.

The chosen variable, the percentage change in turnover from 2019 to 2020, is a continuous variable.

Two grouping variables

Two grouping variables were chosen : North America and Oceania. Within each geographic region distinct categories for investigating group differences of the primary variable were selected. The three distinct categories were significant industries where turnover was thought to be most severely impacted by COVID-19. A one-tailed test and a two-tailed test was performed.

The chosen industries are 1) the Broad Automotive Industry 2)Food and Beverage Industry and 3) Hotels and Accommodation

In summary:

Grouping Variable 1 – North America turnover % change (dependent )
Category 1-3 – 3 industries

Grouping Variable 2 – Oceania turnover % change ( dependent)
Category 1-3 – 3 industries

Data screening and cleaning:

Data was obtained from the Orbis Database. Insufficient data ( sample size of less than 30 for each of the industries) was available for Australia and the U.S.A as financial data is only made available for publicly listed companies; hence we extended our analysis to include data for North America and Oceania.

As the financial year end of Oceanian companies are mostly in June and in North America in December, our Orbis search returned mostly 2021 for Oceanian data and 2020 for North America. The second correction to our data was to use comparable data i.e with 2020 as the most recent year and 2019 set as year-1 for both regions.

Thirdly, we deleted company details where cells returned zero data. For example, some companies were only listed in 2020 and no 2019 data was available, or companies traded in 2019 but went bankrupt in 2020. Cases with missing information, and check & correct any error in data entry. Note that in a few isolated instances a zero is shown in company data but this is because the data is rounded up to $ Mil. We overcame currency differences by setting the currency in Orbis at LCU ( Local currency Unit) and comparing the percentage decline.

Fourthly, we removed outliers showing extreme changes to turnover, i.e more than 100% increase or decrease and elimination using z-scores. This occurred especially with a couple of newly founded or merged companies coming off a very low base.

Specify data analysis techniques, identifying and providing a rationale for the selected techniques.

The rationale for our selected techniques was to perform tests best suited to the shape of the data set and what our main questions are (Cronc 2019 p. 176) i.e dependent and independent variables considered. We needed to determine if revenue declines for one group (Oceania) is significantly different from the other (North America), also, we needed to know what the relationship between different variables ( the different industries ) are. We did not want to make predictions.

The shape of the data set is determined by the skewness and the kurtosis of the data : where these are close to zero parametric techniques should be used, and where the data is not close to zero nonparametric tests should be used. The mean, median and mode determines the skewness of the data. We performed one-tailed and two-tailed tests depending on the direction of the data tested i.e., when we are not testing one direction ( bigger or smaller ) a two tailed test is best suited. The correct P-value from the 2 tailed test was subsequently used for interpretation.

SPSS is very useful to obtain numerical summaries to analyse with descriptive statistics by computing skewness and kurtosis. In addition, where extreme outliers occur these should be removed using SPSS and Z-scores as this would impact on the normality.

When strength of association is tested, non-parametric tests are used. When looking for differences in groups using non-normal data, non-parametric tests are also used. As our data was not normally distributed, we used the Wilcoxon rank test as this is best suited to compare two paired groups where the data is not normally distributed and Kruskal-Wallis to test the strength of the relationships, testing three or more groups (industries) The Kruskal–Wallis test assesses whether an observation in one group is greater than an observation in another group (Mangiafico 2016). To determine which groups are different compared to others we conducted post-hoc testing.

Results and interpretations

Hypothesis 1

Null Hypothesis: The distribution of differences in operating revenue (turnover) for the data collection of 3 selected industries in North America and Oceania is not symmetric around zero.

Alternative Hypothesis : The distribution of differences in operating revenue (turnover) for 3 selected industries in North America and Oceania is symmetric around zero.

Data is not normally distributed because the skewness and kurtosis is not close to zero. Skewness statistic is considerably greater than 0 (left skewed ) and Kurtosis statistic is considerably larger than 0 (high peaked), Leptokurtic. Therefore, after checking the normality in data distribution we perform a non-parametric test: Wilcoxon rank sum test.

Diagram 2: Illustration of the frequency of the data of the two regions. Data is not normally distributed because the skewness and kurtosis are close to zero. Histogram ad normal curve, more “peaked” and thick left tail.

Statistical decision: We do reject the null hypothesis that the variances are equal because p-value (0.001) < 0.05.

Conclusion: At 5% level of significance, sample evidence does not support the null hypothesis that the distribution of differences in operating revenue (turnover) for the data collection of 3 selected industries in North America and Oceania is symmetric around zero. The samples lead us to conclude that the % change in turnover from 2019 to 2020 of North America and Oceania has significant differences across industries during the impact of COVID.

Hypothesis 2
Two Tailed Test

Null Hypothesis : The impact of COVID-19 on the Hotel and Accommodation industry in North America was similar to the Oceania Market

Alternative Hypothesis : The impact of COVID-19 on the Hotel and Accommodation industry in North America was not similar to the Oceanian Market

 

Data is not normally distributed because the skewness and kurtosis are not close to zero. Histogram ad normal curve, more “peaked” and thick left tail.

Therefore, after checking the normality in data distribution we perform a non-parametric test: Wilcoxon rank sum test.

 

Statistical decision: Do reject the null hypothesis because p-value (0.001) < 0.05.

Conclusion: At significance level of 5%, the sample does not support the null hypothesis that there is no significant difference in the operating revenue (turnover) for the Hotel and Accommodation industry between North America and Oceania.

Hypothesis 3

Null Hypothesis : That the distribution of percentage change in turnover due to the impact of Covid-19 in all three industries in North America are identical.

Alternative Hypothesis : That at least one of the population is different

 

Not normally distributed because the skewness and kurtosis is not close to zero. Therefore, after checking the data for normality we perform a non-parametric test: Kruskal-Wallis Test.

 

A total of 30 observations were collected for each industry in the North America market, therefore the mean ranks are not the same for all three sectors.

Statistical decision: Do reject the null hypothesis because p-value (0.001) < 0.05.

Conclusion: There is a significant difference in the mean rank between Hotel & Accommodation and Automotive, and Food & Beverage.

Post hoc method following a significant Kruskal-Wallis test

 

Hypothesis 4

Null Hypothesis : That the distribution of percentage change in turnover due to the impact of Covid-19 in all three industries in Oceania are identical.

Alternative Hypothesis : That at least one of the population is different

Not normally distributed because the skewness and kurtosis is not close to zero. Therefore, after checking the data for normality we perform a non-parametric test: Kruskal-Wallis Test.

Statistical decision: Do reject the null hypothesis because p-value (0.12 ) < 0.05.

Conclusion: At 5% significance level, the distributions of percentage change in turnover due to the impact of Covid-19 are not the same for the three industries in Oceania.

Post hoc method following a significant Kruskal-Wallis test

Conclusion: There is a significant difference in the mean rank between Food & Beverage and Automotive in the Oceania market.

 

Findings and discussion

The starting point of our research was to ascertain whether the regional economic impact of the COVID-19 in respect of the percentage change in geographical regional total turnover was similar in North America and Oceania for the period 2019 to 2020 for a select group of industries.

The reasoning behind our initial assumption was the similarity between the U.S. and Australian real GDP growth, as illustrated below – showing a similar 3.7% and 3.8% contraction in 2020, and the significant portion these national have on their respective geographical region’s economies. The business sector overall contributes on average 72% of GDP in the OECD ( McKinsey 2021). We assumed that because of the similarities of our chosen industries our regional data would not be as heterogeneous as proposed by the OECD ( OECD 2021) and other organisations.

Diagram 3 : Australia/ U.S Economic recovery post 2019.Source: Financial Times 2021

An analysis of the data from 180 listed companies in selected industries showed that our hypothesis was incorrect and that corporate revenues in North America and Oceania were not symmetric for the period affected by COVID-19 for 2019-2020.

Secondly, in terms of comparing Hotel and Accommodation turnover for the 2 regions we were expecting our data from the two geographical regions to be different: we expected the sector to follow the trends of the overall encompassing Travel and Tourism sector with the contributions to GDP falling by 3.5 and 5.1 in 2020 respectively for North America and Oceania per the illustration below.

Our independent sample test, which compared 30 North-American and 34 Oceanian companies, confirmed that this was also the case for the Hotel and Accommodation sector: as expected the global travel restrictions had a direct impact on hotel and accommodation provider turnover, but that the impact was not the same in the two regions. We recommend that the phenomenon is further researched to determine the extent of the aggravating role and impact the size of multinational hotel groups ( Such as Marriott, MGM and Hilton Hotel groups that suffered similar declines of around 50% in turnover year on year ) given the size of the U.S. economy as a whole.

Source : World Travel and Tourism Council 2021

Lastly, we compared the downturns between the three industries, and did so within each geographical region. The geographical regions were set as the independent variables, each with its own independent members or industries.

For the North American industries chosen, we argued that the 3 population medians would not be equal. Our assumptions were based on the S&P Global analysis of Probability of Default, among other things. The diagram below illustrates how driving factors such as reduced turnover due to reduced seating capacity, reduced leisure (which includes e.g., hotel accommodation) driven by domestic and international tourism factors/ travel restrictions and the automotive industry ( impacted by shortages of semiconductor supplies) render the responses very heterogeneous. Interestingly, the difference between the Auto parts and Leisure facilities PD levels demonstrated similar trends however the PD levels towards the end of 2020 were quite different at 4.59% and 7.3% respectively ( S&P Global 2020).

For both variables, the tests performed revealed that the medians of the industries were different, hence our assumptions were proved to be correct.

Source : S&P Global 2022

Similarly, our hypothesis regarding the Oceanian industries was influenced by the ABS’s National accounts which reflect the very divergent effect of COVID-19 on the various industries, as shown below for December 2020:

Source :ABS 2021

Recommendations

Our data is meaningful but not sufficiently insightful. We established that there are significant differences in how the Oceanian and North American businesses experienced the impact of COVID-19, furthermore, that there are significant differences between the ability of some industries to find ways to prevent or slow the downturn. While the differences are not clearly understood, vast amounts of data are available for deeper investigation.

The recommendation is that further research is now required to understand successful and unsuccessful government strategies to deal with the economic impact of COVID-19 in the same way that data enabled statisticians and economists to prevent a repeat of the Great Recession. The data was studied and analysed, and pre-emptive policies were put in place to prevent a repeat of e.g., the extreme government austerity and importance of stimulus. The data and analysis that emerges from COVID-19 will ensure that economies will be better prepared to deal with similar epidemics in the future.

Lisa Cook (Swonk et al 2020) suggests that economists are required to be “voracious consumers of whatever data are available. We must be able to use, analyze and interpret the data we have”.

Several economists speak to the urgency of new policy options : the initiatives that worked before may no longer be relevant. Reliance on data points for indicators are crucial as a faster and better understanding of the data from China as example could have helped North America and Oceania to better plan and prepare :to illustrate, very few foresaw the impact the shortage of semiconductors would have on the global auto industry.

Ellen Zenter (Swonk et al 2020) recommends that a deeper qualitative analysis is more useful in better understanding the COVID-19 impact due to the information overload, unpredictability and volatility of the data. For example, we do not fully understand why the hotel and accommodation sectors were so much more adversely affected in North America and the US despite the global reach of the multinational hospitality giants based in the US – this may point to the dangers of geographic centralisation but requires research.

Real-time government public and private data is readily available in many developed countries but has seemingly not been shared intra-industry, intra region or inter-continental fast enough. Bureaucracy has slowed down decision making and politicking in a pre-election year may have resulted in warning signals interpreted from data not being addressed in an upfront manner.

Limitations of this study

Our data was limited to public companies hence our lens may not be sufficiently wide to understand the impact of Covid-19 on SME’s compared to large organisations; if SME’s were better equipped to respond quicker due to flatter structures and faster decision making; and the effect of developing economies in both regions on the overall data of the geographical data. More data from private companies would mean that the research could be focussed on smaller segments : we were required to extract data and make assumptions for a wide region based on our need for more sample data rather than more meaningful interpretation.

Conclusion

The purpose of our study was to compare and understand the impact of Covid-19 on turnover in North America and Oceania. Orbis database was used to obtain data to conduct empirical statistical analysis. Upon assessment of 180 listed companies across the two different regions and adversely affected sectors, it can be concluded that both these regions experienced different levels of impact as a whole and also across their most impacted industries. Further research is required to better understand factors which may have a significant impact on our results e.g., the impact on SMEs; the impact in developing countries within the regions and the effect of data being available but not expeditiously shared between regions and industries.

References

ABS 2020, Australian National Accounts December 2020, viewed 20 January 2022,https://www.abs.gov.au/statistics/economy/national-accounts/australian-national-accounts-national-income-expenditure-and-product/dec-2020

Cronk, BC 2019, How to Use SPSS® : A Step-By-Step Guide to Analysis and Interpretation, Taylor & Francis Group, Milton. Available from: ProQuest Ebook Central. Viewed 1 February 2022.https://ebookcentral.proquest.com/lib/vu/reader.action?docID=5904712

Financial Times 2021, Australian Economy Powers out of Covid-19, viewed 30 January 2022, https://www.ft.com/content/ac98dd24-9edb-4618-a9af-5ab4cf892262

Hair, J , Page, M. & Brunsveld, N 2019, Essentials of Business Research Methods (4th ed.). Routledge. https://doi.org/10.4324/9780429203374.

Mangiafico, S 2016. Summary and Analysis of Extension Program Evaluation in R, version 1.19.10. Viewed 1 February 2022, rcompanion.org/handbook/. (Pdf version: rcompanion.org/documents/RHandbookProgramEvaluation.pdf.)

McKinsey 2021, A new look at how corporations impact the economy and households, viewed 30 January 2022,https://www.mckinsey.com/business-functions/strategy-and-corporate-finance/our-insights/a-new-look-at-how-corporations-impact-the-economy-and-households

OECD 2021, Managing the Crisis, viewed 30 January 2022,

https://www.oecd.org/coronavirus/policy-responses/the-territorial-impact-of-covid-19-managing-the-crisis-and-recovery-across-levels-of-government-a2c6abaf/?search=oceania#section-d1e1264

S&P Global 2021, Most and least impacted by COVID-19, viewed 30 January 2022, https://www.spglobal.com/marketintelligence/en/news-insights/blog/industries-most-and-least-impacted-by-covid19-from-a-probability-of-default-perspective-september-2020-update

Swonk, D, Cook, LD, Coronado, J, Morris, EK, Paulson, A, Poterba, JM, Sahm, C, Strain, MR & Zentner, E 2020, ‘Economists tackle the challenges of a pandemic’, Business Economics: The Journal of the National Association for Business Economics, vol. 55, no. 4, pp. 279–288, viewed 30 January 2022, <https://discovery.ebsco.com/linkprocessor/plink?id=3ecbc9bb-0273-3608-86e3-242d6267fbcd>

World Travel and Tourism Council 2021, Economic Impact, viewed 30 January 2022, https://wttc.org/Portals/0/Documents/EIR/EIR2021%20Global%20Infographic.pdf?ver=2021-04-06-170951-897

• Assignment status: Already Solved By Our Experts
• (USA, AUS, UK & CA PhD. Writers)
• CLICK HERE TO GET A PROFESSIONAL WRITER TO WORK ON THIS PAPER AND OTHER SIMILAR PAPERS, GET A NON PLAGIARIZED PAPER FROM OUR EXPERTS

QUALITY: 100% ORIGINAL PAPER – NO PLAGIARISM – CUSTOM PAPER