Choosing a topic that can answer an economic research question is challenging. Some tips:
- Ripped from the headlines rarely makes a good economic paper. You will be using data to determine causation or correlation. Sometimes a similar event can be used. Topics such as artificial intelligence may make a good policy paper but not a good economic one due to lack of data.
- Literature Review: Your JIW should use primarily scholarly sources. Start with Econlit (the database of the American Economic Association). Econlit indexes major journals, working papers, conference proceedings, dissertations, and chapters in critical books. It takes a long time for scholarly literature to appear. Preprints are called working papers in economics and major ones are indexed in Econlit. You are your own research team and have limited time. Many articles are written over a couple of years and involve many people gathering and cleaning the data. Some starting places: see https://libguides.princeton.edu/econliterature/gettingstarted
- Outside of finance and some macroeconomic data, most data will not have many points in time. Data determines the methods used. While a linear regression can be great for time series data, it is likely not what you will use for survey data.
- Keep in mind the characteristics of data:
- Microdata: survey or administrative data about an entity (person, family, establishment).
- Longitudinal or panel study: same group of individuals is interviewed at intervals over a period of time. This can be very useful to observe changes over time. Keep in mind when using a long running longitudinal dataset that the panel generally is not adding new participants so may not reflect today’s demographics.
- Cross-sectional study: data from particular subjects are obtained only once. While you are studying different individuals each time, you are looking at individuals with similar demographic characteristics. Demography is typically rebalanced to reflect the population.
- Summary statistics: aggregated counts of survey or administrative data.
- Typically around a 2 year time lag from the time the survey data is collected to the time of release. The Economic Census and Census of Agriculture take about 4 years for all data to be released. Many surveys never release the microdata.
- Very little subnational data is available and is often restricted when available. State level macro data for the United States is more prevalent. City level data is often a case study or only available for very large cities.
- Many micro-level datasets are restricted. It is not uncommon to wait a year before getting permission or denial to use the data. Each organization has its own rules.
- Historical data in electronic format prior to 1950 is rare. Most governmental links provide current data only.
- What is measured changes over time. Do not assume modern concepts were tracked in the past. Definitions of indicators often change over time.
- Data cannot be made more frequent. Many items are collected annually or even once a decade. Major macroeconomic indicators such as GDP tend to be quarterly but some countries may only estimate annually.
- What exists for one country may not exist for another country. Data is generally inconsistent across borders.
- Documentation is typically in the native language.
- Always look at the methodology. The methodology section is one of the most important parts of the paper. Someone should be able to replicate your work. Describe the dataset and its population. Describe how the data was subset, any filters used, and any adjustment methods. While you are likely not trying to publish in American Economic Review or Journal of Finance, these are the gold standards. See how they layout the articles and in particular the methodology and data sections.
- The basic question to ask when looking for economic data is "who cares about what i am studying?" Unfortunately, the answer may be no one. Ideally, look for an organization that is concerned with your research as part of its mission. Examples include the International Labor Organization or the Bureau of Labor Statistics focusing on labor research; the International Monetary Fund or the Board of Governors of the Federal Reserve System focusing on monetary and fiscal concerns; the World Bank focusing on development; and the World Health Organization focusing on health. This does not mean these organizations collect data on all topics related to that field.
- Find a topic for which there is literature and data but allows room to add a contribution. Topics such as sports and music are popular due to personal interests but may not make good research topics due to lack of data and overuse.
More tips:
- Data is typically not adjusted for inflation. It is usually presented in current (nominal) currency. This means the numbers as they originally appeared. When data has been adjusted for inflation (constant or real), a base year such as 2020 or 1990 will be shown. If a base year is not provided, then data is current and therefore not adjusted for inflation. If given a choice, choose current dollars. Data is often derived from different datasets and many will use different base years. Adjust everything at the end. It is easier than doing reverse math!
- While most datasets are consistent within the dataset for currency used such as all in US Dollars or Euro or Japanese Yen or each item in local currency, some will mix and match. LCU is a common abbreviation meaning local currency units. Consider looking at percent changes rather than actual values. If adjusting use the exchange rate for each period of time, not the latest one.
- Economic indicators may be either seasonally adjusted or not seasonally adjusted. This is very common for employment and retail sales. Unless something says it is seasonally adjusted, it is not. Be consistent and note in methodology.
Librarians are here to help! Librarians can help to devise a feasible topic, assist with the literature search, and choose appropriate data. Your data may fall into multiple categories. Think of the primary aspect of your topic in terms of first contact. Do not email librarians individually. If unsure who to contact either put all that apply on same email or email just one. If that person is not the best, they will refer you.