Guidelines for Writing Term Papers and Dissertations in Economics Christoph Siemroth Department of Economics University of Essex First version: October 2016 Last updated: February 21, 2017

Abstract The goal of this document is to provide some guidelines on writing academic papers and dissertations on the undergraduate and master’s level.

1

Introduction

Why do universities assign papers and dissertations to students as a requirement to obtain a degree? The idea is that students should demonstrate the ability to independently research and answer questions in an academically rigorous manner. This includes showing an awareness of the academic literature by citing, discussing and possibly using prior work. Writing a paper is an opportunity for students to delve deeper into topics that have not been covered in lectures or were only mentioned briefly. The assignment gives the freedom to explore your own interests further and to specialize. Which topic did you always want to know more about or which question did you always want answered? Here’s your chance to do it and get a good grade on top. However, it is also more responsibility than taking an exam since you decide (within limits set by the program) what the topic and question is.

2 2.1

Finding a topic and defining a question or goal Topic

(You can skip this section if you have already been given a topic.) Students know their own interests best. Still, often students struggle to find topics or “research questions.” Or, if they have a rough idea about the topic, they have no idea how to exactly define the goal of the paper. These are some guidelines about the choice of topic. While this is not a requirement, framing topics in terms of questions is usually a good idea, because it makes the paper’s goal very clear (answer the question). Topics like “Financial stability in the US” don’t tell the reader exactly what to expect. “Did financial stability improve after the Dodd–Frank act in the US?” on the other hand sets the topic and goal of the paper. A title I like a lot from a rather controversal figure is “Who Lies on Surveys, and What Can We Do about it?” The title nicely poses an important question (Who lies on surveys thus invalidating the method?) and tells the reader the goal of the paper (provide a solution to the problem). But make sure that the question you pose can and will be answered in the paper. Your supervisor will not be happy if you ask “How to maximize your investment return?” and then do not deliver. This applies in particular to data projects where the data and analysis should help you to say something about the question. Also, make sure the topic or question is specific enough. Thus, “Are stocks any good?” is too vague. Good for what? Good for whom? A better topic might be “Are stocks a viable investment class for retirement savings?” Even if the goal of the paper does not appear to be a question—say you want to use some computer science technique to find an algorithm that would have maximized the stock

2

return over the last 10 years—then it can typically be rephrased as a question: “Which trading algorithm would have maximized stock returns in 2000-2010?”. The topics of papers and dissertations should be specific enough so that these can be realistically tackled in the set time and word limit. If a broad question can only be answered unsatisfactorily in the given time, then it is better to opt for a more narrow topic. Your supervisor should be able to help you define realistic goals, or you have some understanding from a course or a previous paper. And do not apologize when you impose bounds on your topic. Either no apology is necessary (if you have reasonable bounds) or it won’t help you (if you made it too easy for yourself). In term papers and in particular dissertations, the expectation is that papers do not merely restate what other authors have done (i.e., provide merely a descriptive text), but also that students advance arguments, defend them, and derive their own conclusions. Thus, the paper should have a discursive part with some assessment based on the knowledge from the literature and/or data/model. This may be done in the form of evaluating whether previous work is applicable in a certain case or problem, or indeed by criticizing other authors’ conclusions or methods.

2.2

Data project vs theory project vs literature review

Especially on the undergrad level, a dissertation does not have to be a data project. A dissertation can also a review of the literature in light of a certain question, i.e., identify the relevant studies, critically assess those, and use them to answer your question. This is what you might have to do in your next job, so being able to go through academic papers/books and find the answers that you need is a valuable skill. Plus, a project where you use existing research to answer a question is better than a project where you use poor data and an ill-applied statistical test that doesn’t answer the question. Use common sense: What is more likely to give a satisfactory answer, a data analysis done within 2 weeks by an undergrad or a conclusion based on the available studies and papers written by professional researchers? I do not want to discourage a data project, but you should not start a data project for the sake of doing a data project: It has to contribute to answering a question. In a literature paper/dissertation, the goal is to survey the literature, decide what is relevant and what isn’t, and then critically assess previous work to answer your question. The goal should be to find the right answer to your question given the work (read: evidence) available in the literature. Thus, a good literature paper/dissertation requires a critical, argumentative part—weighing the evidence, evaluating its relevance for the present question, making a conclusion and defending it—otherwise it is just retelling what others have done without any additional input from you. Make your contribution/addition very clear from the start. For example, in the introduction you can write “the paper surveys the studies on X in developing countries followed by a critical evaluation and discussion of whether

3

X [answers the research question]. I conclude that X is typically ineffective in developing countries because of a,b,c, but may help if d.” Being clear about this is important because your independent contribution in a literature paper is less obvious and salient compared to a data project (where you analyzed data!). So make clear how your paper/dissertation differs from a mere summary of the literature.

3

Getting started

Once you have your topic, you can start googling content and perhaps glance at the related wikipedia pages to get a broad overview (if you don’t already have one). Wikipedia is not a source that you can cite or use as “evidence” in your paper—it is to be used strictly as background information—but it can help you get an overview or find cite-able sources such as research studies. A good start into the academic literature is to find academic books on the topic that often summarize and explain previous research (e.g., Oxford University Press books, where experts in the field often disseminate their knowledge in a less technical manner), or survey articles in academic journals that specifically aim at getting people up to speed on the topic. In economics, the Journal of Economic Surveys might have an overview article on your topic if you are lucky. The Journal of Economic Perspectives and Journal of Economic Literature also provide overview articles. Other journals publish survey articles occasionally, just search for “[topic] survey” or “[topic] overview”. A good search engine to find such academic books and articles is Google Scholar which restricts the search to more reputable sources (compared to google which searches everything). Once you found a relevant article you are in. Then it’s easy because you just have to follow the citation trail. Author A who did work on your topic will reference authors B and C who have done related work with a quick explanation. Authors B and C cite others in turn. This will help you identify the relevant studies in the field. To find the relevant studies conducted after B and C published theirs, you can use Google Scholar to see who cited B and C.

4

Structure of the paper/rough outline

Journal articles tend to have abstracts, i.e., brief summaries of the main message of the paper (usually no more than 150 words). This is not necessary for student papers, but it can help make the main message clearer if the title does not already do so. There is no need for keywords or JEL codes; this is done by journals to help select editors/referees and to improve search routines, none of which is relevant here. It can be a good idea to put a table of contents at the beginning of the paper. It helps the reader and yourself to understand the paper’s structure in one glance. If you include it,

4

then do not assemble it manually but use your word processor’s function to generate it. A typical paper structure is as follows: 1. Introduction 2. Literature 3. Data/Method (if empirical) or Model (if theoretical) 4. Results 5. Discussion [may be merged with Results] 6. Conclusion 3., 4. and 5. form the main part, sometimes these can be merged together in one section. In a pure literature paper/dissertation, a typical structure can be 1. Introduction 2. Literature: What is X?/What are previous studies and results on X? 3. Evaluation/discussion: use 2. to answer your question 4. Conclusion More important than these labels, the paper or dissertation must have the following key elements: (1) what’s the goal or question of the paper/dissertation, (2) what is the related work in the literature, (3) what do you do to answer the question/achieve your goal (data/model/literature), (4) what’s your answer (arguments/evidence!).

4.1

Introduction

The most important piece of information that has to be included in this section is: What is the research question/goal of the paper? It doesn’t have to be pretty; you can explicitly state “the goal of the paper is X” or “the question this paper answers is Y”. Moreover, you should typically provide some motivation for the research question in this section (unless it is obvious): Why is the research question important? Where can we use your results? The remainder of the introduction is typically a short overview of the contents of the paper and in particular the results/conclusions.

5

4.2

Literature

This section should answer: What has been done before in the relevant literature? What are related questions and topics that have been tackled? How does it relate to the question posed in this paper? You may also use the section to reference important methodological contributions in the literature, but typically these are better referenced later when you discuss your own method (empirical or theoretical model, if you are not doing literature paper). The density of references should be very high in the literature section compared to the others. This section refers to some articles in high quality journals or possibly academic books. This section typically does not cite non-fiction popular books or newspapers (these may be used for motivation or background in the introduction), and certainly not blogs or wikipedia. As a rule of thumb, the farther you progressed in your studies (e.g., undergrad vs postgrad), the more you are expected to cover the state of the art research in academia relating to your question. If your paper is about something in a particular country, then do not only reference authors from that country. In all likelihood researchers from other countries have made important contributions to your question as well. This section is not about citing everything remotely related to your question. Your task is to find the most relevant work and explain how it is relevant for your work/question. Thus, do not merely enumerate the references with sparse information that could be obtained from the abstracts; demonstrate that you read the articles by for example also covering the assumptions/data of the papers and discussing/comparing them. An in-depth literature review is not just name-dropping. You can present definitions or define important concepts relating to your topic in this section (if not already done in the introduction), especially if understanding these is necessary to understand the literature. After all, there is no point in reviewing the literature on microfinance if the reader has no idea what microfinance is.

4.3

Data/Method or Model

In this section you should explain how you intend to answer the question you posed in the introduction. If it is an empirical analysis, then which data do you use, where did you get it from, perhaps discuss whether it is adequate to answer the question and what its shortcomings are (e.g., missing data problems). If it is a theoretical paper, you should formalize the problem and ideally state all important assumptions. In particular, who are the agents, what are their actions and payoffs, what is the information structure, and what is the equilibrium concept (if any)? In terms of method, if you use something standard like OLS regression or Nash equilibrium, then there is no need to explain it in detail. But it is usually a good idea to demonstrate to the markers that you understand these methods. In OLS, this can be done 6

by running a robustness check (what changes if you use 2SLS or ordered logit instead?) or including alternative ways of calculating the standard errors (clustering, bootstrapping etc.), and explaining why these robustness checks improve on the original specification. In theoretical work, you can demonstrate your mastery of the method for example by explaining how a change in assumptions would change the results. If you use a “non-standard method”, say a fixed effect negative binomial regression or the quantal response equilibrium concept, then you should explain what it is and how it differs from the standard methods (possibly with justification or motivation why you go with the non-standard method). It is absolutely fine and in fact desirable to use non-standard methods if these are called for, just remember that you have to do a bit more explaining.

4.4

Results

Do not lose sight of your research question. This section should have something that addresses it and hopefully answers it. If you work with data, do not put the results of regression analyses and similar in the appendix; it should be right here where you write about the results. Also, do not just run a bunch of regressions. Interpret them. How do they help to answer your question or solve your problem? The same goes for theory: What do the model results imply for your question? Thus, the results section should not be merely an enumeration of results. Sometimes the interpretation is separated in a section called “Discussion”. In this case, it is fine to postpone the discussion and interpretation, but it has to be somewhere in the paper!

4.5

Conclusion

This section can be brief.1 You should explain briefly what the question was, how you addressed it, and what your answer is. Do not summarize the entire paper, just go for the main point. One option to end the paper is to give an outlook: Which new questions emerge from your analysis? What are related interesting questions that you didn’t cover in this paper? Another option is to briefly remark on how your question/results may be relevant in other countries/fields/contexts. But don’t wildly speculate about things beyond your analysis. 1

Indeed, one author argues it is obsolete (Cochrane, 2005). While you shouldn’t go that far, it is an otherwise excellent writing guide.

7

5 5.1

Writing, language, and orthography General remarks

Use your best English and, if you are not a native speaker, perhaps have someone else proofread your paper. While you will not lose major points for slight language errors, your mark can suffer if the language degrades to a point where it is unclear what you are saying. Also, make sure to avoid spelling mistakes. Poor orthography gives the impression of a rushed job and may call into question the quality of the content. Use the spell check of your word processor to catch obvious mistakes. Typically it is a good idea to put away the paper/dissertation once it’s finished for a week and then read it once more to check for problems in language or clarity. More often than not you will spot passages where even you—the author—will not know what the heck this means. Then fix it. Many students don’t realize that first drafts have weaknesses, and the elegant argument you thought you made is not understandable to anybody but yourself. There is a lot of implicit knowledge in our thought processes that is not explicitly written down in the paper. Hence, the reader who lacks this implicit knowledge may not be able to make sense of what you wrote. Keep that in mind when writing and stick to simple sentence structures and language. It helps to find another student and read each other’s papers to identify unclear paragraphs.

5.2

Writing

There are many writing guides on the web. Use them if you have never seen one. The university also provides courses on writing skills. These are skills that should be useful in many non-academic jobs as well. A few points on this here. First, most student papers are single authored, but surprisingly many use “we” instead of “I”. While there are no strict rules at university, it is probably good practice also for your next job to start using “I” if you are the sole author (you want full credit for your work, right?). In a famous anecdote, a physicist used “we” in a paper when a colleague noted that he was the sole author and would have to rewrite the entire paper for journal submission. Too lazy, the physicist included his cat under an alias as co-author so that “we” was the correct form. So why not do it right from the start. Second, don’t include personal views or experiences (don’t: “When I read these papers I was surprised that. . . ”)—papers and dissertations are not to be confused with your diary. Try to adopt a more objective language (“It may be surprising that. . . ”) or better yet, blame someone else (“Authors like X and Y frequently conclude that. . . ”). Third, and this is advice given in many writing guides, try to avoid passive forms whenever possible. Don’t: “First, an analysis was carried out, then a conclusion was

8

reached. . . ” (I wonder who did that. . . ), better: “First, I analyzed the time series for non-stationarity. . . ”.2 Fourth, use subsections to divide your paper into well defined parts. Give sections and subsections descriptive titles. A proper structuring of the text and informative titles improve readability and allows readers to quickly find what they are looking for.

5.3

Plagiarism and references

Don’t plagiarize. This is one of the things that can get you into serious trouble. And it is nowadays very easy to spot these things digitally. So if you read something that seems smart and you wish you had come up with it on your own, do not rephrase and pretend it was your idea. And by no means copy sections from other writings without reference (i.e., without acknowledging the author and adding the source). Always properly cite your sources. It is perfectly fine to include many references in the paper. Indeed, it demonstrates that you are familiar with the literature and you are expected to cite the relevant works in the literature. But do not just enumerate the references. Make sure you demonstrate that you actually read and understood them. So don’t write things like “Some authors like Smith (2000), Baker (2001), Miller (2002) have done work in this area.” Be more specific what they have done and possibly how. Again, as mentioned above, try not to merely summarize what others have done but also critically assess and discuss it, for example: “Mason (1999) found a negative impact of A on GDP growth in OECD countries, but this finding may not be applicable in Tanzania, which as a developing country is structurally very different as shown by Taylor (2000).” There are many citation styles and different disciplines favor different ones. In law, for example, scholars tend to put the full source in a footnote. In economics, and the social sciences more generally, we tend to cite author surnames and year of the publication, as exemplified by this random reference that otherwise has nothing to do with this guide (Page and Siemroth, 2016). The full reference is then given at the end under the heading “References” or similar.

5.4

Formal requirements

Adhere to them, especially word/page limits. Moreover, a paper should have a cover page that provides the following information: • Title • Author name (unless there is blind marking) 2

An unnamed US comedian made a similar point in an entirely different discussion where “sex was had”. Don’t write like that unless you have comedy aspirations.

9

• University/affiliation • Some kind of a date (e.g., submitted 19 October 2016, October 2016) • Purpose (e.g., MSc dissertation, term paper for course ECxyz, etc.) For some term papers the administration requires a university cover sheet/form when you hand in, make sure you add that as well. Include page numbers. For readability, use 1.5 spacing or less in your text. Mainly because double spacing looks ugly and wastes space. 12pt font size, 1.2 spacing, a serif font (e.g., not Comic Sans), and text justify (rather than left aligned text) looks most professional. Use footnotes rather than endnotes—nobody wants to flip to the end to read them—and use them sparingly.

6

Common mistakes • Abbreviations: Don’t use abbreviations for terms that you need only once or twice in the paper, but they do help for frequently used terms. In this case do not forget to define the abbreviations. While economists should have heard of GDP, not everybody knows what SMEs are. • Statistical vs economic significance in regressions: High p-values need not imply a relevant magnitude of the effect, just that we can be reasonably confident that the effect is not zero (it may still be very small). Conversely, low p-values need not imply that the factor is irrelevant, it just means that we cannot be statistically confident that there is a non-zero effect. • Don’t write which software you use to run OLS regressions. The reader doesn’t care whether you use Stata or invert the data matrices with pencil and paper (though I would advise against the latter). Just do it right. If you use non-standard techniques, then you should explain the steps you take or what the technique does, but typically not in the form of “And then I use Stata’s nbreg command to estimate a negative binomial regression model. . . ”.

Bibliography Cochrane, J. H. (2005): “Writing Tips for Ph. D. Students,” https://faculty. chicagobooth.edu/john.cochrane/research/papers/phd_paper_writing.pdf. Page, L. and C. Siemroth (2016): “An experimental analysis of information acquisition in prediction markets,” Games and Economic Behavior.

10