Quantitative Research Methods in Finance

2018/2019

Individual Research Project

DUE DATE: 22 April 2019 at 12 noon

 

Please read this document carefully before commencing the project.

 

 

Choose one key regressor that may determine one of the following firm-level outcomes:-

1. Innovation, as measured by research and development expenditures scaled by total assets
2. Stock liquidity, as measured by stock turnover (trading volume scaled by number of shares outstanding)

Discuss whether and how your chosen determinant (regressor) may affect your chosen outcome (dependent variable). State and test your research questions and hypotheses using a cross-sectional regression model on two samples of firms, with at least 50 firms in each sample. In addition to appropriate control variables, include at least one dummy variable and one interaction term involving the dummy variable in your analysis.

Present your results and discussion in a 2,500-word report. A. What do you need to submit?

1. A 2,500-word report (see Section B for more details).
2. A zip file containing all the data and Stata do/log files that allow a complete replication of all your results (see Section C for more details).

B. Your report

Think of this report as a “mini research paper”. You will need to perform very similar steps for your dissertation; therefore, use this research project as practice run.

Try to follow the structure and writing style of “modern” research paper in Finance (see, for example, papers you have seen in class and those you will come across during your literature review).  The final report should be 2,500 words (+/- 10% as per School regulations). Note that high quality projects are usually concise.^[1]

Your report should have the following sections:-

1. Introduction and hypothesis development (600 words)
   1. Motivate why your research questions you are pursuing is important
   2. Discuss what have been done before (i.e. a brief review of the literature)
   3. Outline your hypothesis. What do you expect to find based on prior theoretical and empirical works.
   4. Provide a summary of your results. Discuss the most important things you find. Focus on the economic implications of your results.^[2]
   5. Discuss your contribution to the literature. How does your work differ from what have been done before, and how your results add to the current understanding of the topic.^[3]
2. Empirical model (400 words)
   1. Specify your regression models as equations.
   2. Justify the functional forms of your variables and also your choice of control variables. The discussion in this section should be heavily based on our current understanding regarding the economic relationship between these variables (i.e. the literature!).
3. Sample (400 words).
   1. Describe your data sources and how you construct your sample.
   2. Discuss your treatment of any data issues (e.g. how you deal with outliers, missing values, etc.).
   3. Present summary statistics of all your variables and a correlation matrix. Discuss whether what you see are consistent with what you would expect.^[4]
4. Results and discussion (1,000 words)
   1. Present your regression results
   2. Discuss your findings in terms of both statistical and economic significance. Relate your findings to your hypotheses as well as prior literature.
   3. Discuss whether your results can reliably be used as evidence for/against your hypothesis. Can there be alternative explanations? Is there any potential violation of OLS assumptions? Carry out specification tests and/or robustness checks as necessary.
5. Conclusion (100 words): Summary of your key findings and takeaway messages.

 

READ THIS: Content of tables/graphs (along with their titles and descriptions), footnotes, and appendices are not included in the word count. However, all key arguments must be presented in the main body of the report.

Key findings must be included in the main report as professionally formatted tables and graphs. Tables and diagrams must have a title and relevant description. Graph axes must be clearly labelled. Readers should be able to understand all tables and diagrams without referring to the main text.^[5]

Although this coursework is classified as a “report”, you are expected to write in complete sentences and well-structured  paragraph (and not sentence fragments in bullet points), with introduction and conclusion sections. Headings and subheadings are encouraged.

The report must be presented in PDF format. You should use Times New Roman 12-point font, setup your page size to A4 paper, 1-inch (2.54 cm) margins from each side (top, bottom, left, right), 1.5 line spacing, justified margins with footnotes on the same page rather than endnotes. Use the same paragraph format throughout. Number all your pages, tables and diagrams sequentially. These constraints are imposed so as to achieve a standard professional text format and the work may be penalised if it violates them.

C. Data and Stata do/log files

Put everything is a single .zip file and submit the file electronically. Do not include the print-outs of your do/log files in your report.

The original data files (e.g. the original .dta file you download from Compustat, or .xls file with variables from Capital IQ) must be included in the submission and the .do file you submit must show how the data is cleaned, transformed, and merged into the final data file that you use to conduct your analysis.

Please annotate your do files, such that other people besides you can understand them, and that the Stata results can be linked to those results in the report. See tutorial do files for examples.

 

D. Data

You can use any data sources for your analysis. Below are some of the databases that we have available in the School:-

 

• Compustat has financial report data that you will require to construct most of the variables such as firm size, profitability (ROA, ROE, ROS, etc.), leverage, R&D expenditure, capital expenditure, cash holding, etc. The data is only available for US firms.
• CRSP has stock price/return data and return on commonly-used proxies for the market portfolio. You may need this to compute annual stock returns and/or stock return volatility. The data is only available for US firms.
• Capital IQ and Bloomberg have both financial report and stock market data for international companies.

 

Other databases that you may want to be aware of are Execucomp (executive compensation data) and ISS Riskmetrics (director data). Both of these are available on WRDS.

You are asked to conduct your analysis on two groups of firms. You can define the two groups in any way you like. Therefore, think strategically how this may fit into your analysis and what sort of results may be sensible. Note that we do not require you to estimate two separate regressions (but you can if you think it is useful for your analysis).

Our suggestion would be that you do not make your sample too large e.g. all US firms on Compustat with data available between 1925-2018, but instead focus on a small subset of firms, e.g. technology firms between 2001-2005, or manufacturing and service firms in 2008-2012.^[6][7] For technology firms, the two groups for your analysis may be young and old firms (divide your sample into two groups by age). Or you may want to see whether the results differ between manufacturing and service firms (divide your sample into two groups by industry classification). You may also think how dummy variables and interaction terms may be useful in this context.

Note that you are asked to conduct a cross-sectional analysis. To do this, you may obtain the data for the time period of your choice and collapse the observations into cross-sectional observations (with each variable containing the average value over the time period).

When completing this project, several issues will arise with regards to sample construction and data analysis. Most of the time, there is no one correct way of dealing with these issues. What we are assessing is your ability to decide and justify your decisions pertaining to how you deal with these issues.

E. Source materials, references and bibliography

Source materials for this assignment should come from high quality academic sources. High quality academic sources in this context include the followings: articles published in high-quality peer- reviewed journals, working papers (i.e. unpublished articles) from leading academics in the area^[8], books from respectable publishers.

Factual data (such as market size, trading volume, etc.) should come from institutions or agencies that at least arguably can be deemed respectable. Students should avoid referencing dubious internet sources (especially Wikipedia and Investopedia).

Academic articles can be found through online databases such as Scopus, ScienceDirect, or Google Scholar. Note that although all published articles have gone through a peer review process to ensure accuracy and academic rigor, quality and reliability of the content in the articles may still vary.

The Chartered Association of Business Schools provides a list of business-related academic journals and a rating for each journal. This is called the ABS list and is widely used in the UK.^[9] As a rule of thumb, journals rated 3 and above are deemed of acceptable quality. Examples of high-quality journals include top 5 Economic journals (American Economic Review, Econometrica, Journal of

Political Economy, Quarterly Journal of Economics, and Review of Economic Studies) and top 5

Finance journals (Journal of Finance, Journal of Financial Economics, Review of Financial Studies, Journal of Financial and Quantitative Analysis, and Review of Finance). All these journals are rated 4 or 4* by ABS.

References should be made in-text (not in the footnote) according to the Harvard style of referencing.^[10] A bibliography should be included at the end of the report (but before the appendix).

F. Student interaction and discussion board

Throughout this project, you will encounter many common problems and issues (data handling, Stata commands, et cetera) and may sometimes need help. As long as the submitted work is wholly your own, you are encouraged to interact with each other and help each other.

To facilitate this interaction, we have set up a discussion board for this project. This is on LEARN > Discussion Board > Individual Project. You can post questions and issues you encounter on this discussion board to solicit answers/solutions from your colleagues. If you post a question and later find an answer, you can post that on the board to help others. You are also encouraged to share any interesting ideas, papers, best practices, and how-to’s on this board. The discussion board will be moderated by the instructors at least once a week.

Please post the questions about the project on the discussion board. As the purpose of this board is to be the first point of contact for support, any questions that are directly emailed to Angelica, Ben, or Maria will be responded with the following message:-

 

“Thank you for your email. Could you please ask this question on the Discussion Board?”

 

Please do not collude and work together on the same project. Whilst it may be possible that two individuals independently approach the same research question from a similar angle, it is likely to be very rare that their data set and/or empirical models are identical or very similar. This does not mean that we will automatically accuse anybody of plagiarism if such cases occur, but these will certainly raise suspicions.^[11]

 

 

Marking rubrics

	Fail	Satisfactory	Good	Very good	Excellent
Literature review and hypothesis development	Characterised by irrelevance, brevity and/or superficiality; no clear link between the literature and hypotheses; no or little use of appropriate and relevant content in the work; inappropriate and/or narrow range of references used	Hypothesis drawn from the literature; uses some appropriate and relevant content to develop simple ideas in some parts of the work; may make omissions and/or includes irrelevant material;  narrow range of references	Hypothesis drawn from the literature; applies knowledge without much integration and synthesis of material;  uses appropriate and relevant content to develop and explore ideas through most of the work; adequate references are drawn	Hypotheses clearly and logically drawn from the sophisticated synthesis of the literature; material organised in a clear and logical form;  evidence of integration and occasional indication of synthesis;  uses appropriate, relevant and compelling content to explore ideas with the context of the discipline and shape the whole work; adequate references are drawn	Hypotheses clearly and logically drawn from the sophisticated synthesis of the literature coupled with evidence of independent insight;  uses appropriate, relevant and compelling content to illustrate mastery of the subject, conveying the student’s understanding and shaping the whole work; draws on a wide, relevant literature base
Empirical analysis and discussion	Inappropriate model construction, which does not demonstrate any link to theoretical and empirical evidence. Inaccurate interpretation of the results; demonstrate no understanding of estimation results.	Appropriate model construction and estimation, which show some understanding of theoretical works and evidence in the data; may contain flaws; demonstrate superficial understanding of the estimation results.	Appropriate model construction and estimation, which show some understanding of theoretical works and evidence in the data; accurate interpretation of the results; may contain flaws; demonstrate some understanding of the estimation results.	Appropriate model construction and estimation, which are drawn from both theoretical works and evidence in the data;  accurate interpretation of the results; demonstrate some understanding of the estimation results including some nuances regarding the methodology, the empirical setting, and any potential limitation.	Appropriate model construction and estimation, which are drawn from both theoretical works and evidence in the data;  Compelling conclusion drawn from estimation results; demonstrate sophisticated understanding of the estimation results including nuances regarding the methodology, the empirical setting, and any potential limitation.
Use of language	Employs words that are inappropriate or unclear in context;  sentence structure is inadequate for clarity; errors in use of English are seriously distracting;  uses language that often impedes meaning because of serious errors in usage	Choice of words and sentence structures are adequate to convey basic meaning;  errors in use of English are noticeable and distracting;  uses language that sometimes impedes meaning because of errors in usage	Usually employs the correct choice of words for the context and sentence structure is reasonably effective;  presence of errors in use of English are not overly distracting;  uses language that generally conveys meaning to readers with clarity, although writing may include errors	Correct choice of words for the context and sentence structure is effective;  presence of a few errors in use of English are not distracting;  uses straightforward language that generally conveys meaning to readers; language has few errors	Employs appropriate words fluently;  develops concise good use of English; balances a variety of sentence structures effectively; uses graceful language that skilfully communicates meaning to readers with clarity and fluency and is virtually error-free
Use of figures / tables	Inappropriate or incorrect figures / tables used	Figures and tables are not well embedded in the report’s narrative and may be poorly labelled, explained or of limited relevance	Figures and tables are correct but not well used in the report’s narrative	Figures and tables are very well labelled and add to understanding in the narrative	Outstanding use of (possibly original) figures and tables that enhance narrative
Quality of referencing / bibliography (incl. referencing style)	Incorrect use of referencing style and/or presentation of bibliography/references;	Evident errors in the use of references in body of text; minor flaws in presentation of bibliography / references	Bibliography/references correctly done;  some errors in referencing in text	Bibliography/references correctly done; referencing in text is done correctly;  no errors in referencing style	Bibliography/references correctly done; referencing is to publishing standard

 

[1] The following documents provide some tips for writing a good report: (1) Nikolov, P. (2013). Writing tips for economics research paper. Available at: http://www.people.fas.harvard.edu/~pnikolov/resources/writingtips.pdf; and (2) Cochrane, J. (2005). Writing tips for PhD students. Available at: https://faculty.chicagobooth.edu/john.cochrane/research/papers/phd_paper_writing.pdf. Not only are these papers useful for this project, they can also be useful when writing your dissertation.

[2] Modern academic papers in Finance usually provide a ‘preview’ of the results and markers of your dissertation will expect to see this. Therefore, we ask that you do the same for this project.

[3] This is going to be difficult, because of the limited scope of your empirical work. However, for this assessment, the contributions do not need to be big or many, they just need to be clearly stated. This is also one of the elements of your dissertation that will carry a substantial weight and you will benefit if you practice early on.

[4] Students are often not sure what to discuss in this section. You may, for example, compare your summary statistics with those from other similar studies and discuss whether what you obtain is similar (and, if not, what are the potential reasons). This is usually a good exercise, as you can check whether there is anything wrong with your data collection and variable construction.

[5] See http://jfe.rochester.edu/table11.pdf for examples of professionally formatted tables and graphs. Note how these tables are graphs are generally “self-contained” i.e. the readers can understand what these tables and graphs are trying to say without having to refer to the main text of the paper.

[6] Working on a small sample also means that it is feasible to hand-collect and/or manually clean some variables in a reasonably short period of time. See e.g. Nguyen D. D., Hagendorff, J., and Eshraghi, A. (2018). Does a CEO’s cultural heritage affect performance under competitive pressure? Review of Financial Studies, 31(1): 97-

[7] . https://academic.oup.com/rfs/article/31/1/97/3835432

[8] Whether or not an academic is considered “leading” is obviously subjective. Generally, working papers of academics who have published in high-quality journals and/or work in well-respected institutions are likely to be more rigorous.

[9] The most recent list is available at https://charteredabs.org/academic-journal-guide-2015/.

[10] For more information, see http://www.docs.is.ed.ac.uk/docs/Libraries/PDF/SEcitingreferencesHarvard.pdf.

[11] See https://www.ed.ac.uk/files/atoms/files/codeofstudentconduct.pdf for the University’s Code of Student Conduct, and https://www.ed.ac.uk/academic-services/staff/discipline/academic-misconduct for Academic misconduct procedures.