Effective research project management practices (continuously updating)

Since I am almost done with my first year of PhD (apart from a research proposal which is due three weeks from now) and I am exempted from qualifiers, I have started my own research project on household migration decisions in Indonesia.

I have realized that good project management practices are essential to productive researchers. This (continuously updating) post will jot down some useful tips from my view.

I: General Project Management Practices

  • Make plans for the implementation of research projects and monitor progress on a regular basis.
  • Create separate folders for research projects and research methods. The latter is more of a toolbox and not necessarily research-topic-specific, and separating it out makes it more readily accessible.
  • Have a readme file for every folder. Five years from now, you are going to forget what “figure1″ means, and you need a master file documenting (at the proper level of detail) what each file is about.

II. Data Management Using Stata

While reading this great book on data management with Stata, I took down the following notes which I found to be most helpful.

  • create a master do file to link all steps in data cleaning and analysis.
  • ask a friend to check my do files to make sure others can understand.
  • create documentation for each project
    • document key information that you may forget over time
    • labeling data is not the only way; comments in do files, a journal, flow diagrams, etc can be useful.
    • pick something that works for you and stick with it.
  • add “version” at the top of each do file to make sure it will be executed accordingly.
  • separate intermediate work from final work by
    • regularly checking the old/new versions of the documents and pruning the unused ones
    • create a separate folder for the intermediate work, or
    • use version control software
  • develop routines: use the “research methodology” folder; decompose big tasks into smaller ones; be aware of the steps to identify a research topic, find relevant literature and gaps in the literature, and data analysis.
  • back up your data: use a hard drive and save it in Dropbox.
  • A few notes about macros
    • can be used to set “default” options for regressions, e.g. “local regopts noheader beta vce(robust)”
    • identify multiple assertions without halting the do file: assert …,rc0
    • locals can be combined by defining a new local
  • Accessing saved results
    • use “quietly summarize” and refer to the results r() to immediately generate new variables.
    • for estimation results, use “ereturn list” to list the estimation results.
    • use “statsby” to save estimation results in a new dataset.

HT: SS.

Julie Cullen on the Unintended Consequences of Education Reform

Julie Cullen at UCSD visited Duke this week and gave a talk this morning on the unintended consequences of education reform. Her talk is a general summary of the lessons drawn from three papers on 1) the impact of fiscal incentive on student disability rates, 2) school gaming under a performance accountability system, 3) strategic high school choice under Texas’ top ten percent plan.

These papers ask different specific questions, but they all address broadly the mechanisms through which education reforms can change individual behavior and create unintended consequences. The intuition is that whenever eligibility to funding/benefits are conditional on some manipulable measures, those measures will be manipulated. This same logic is applied to examine how fiscal incentives, performance standards, and policies aimed at expanding college access have distorted incentives of schools.

Although I am not very interested in the economics of education, I can imagine the same approach being used to assess other types of government policies, e.g. welfare program participation.

Using mobile phones to collect (otherwise ignored) data!

While I was doing a literature review on migration, I saw an interesting new article on inferring internal migration patterns from mobile phone usage data in Rwanda. The author Joshua Blumenstock, a professor in the information school at the University of Washington, has quite a few interesting projects using mobile phone data for policy evaluation in developing countries.

He also has a few courses on data analysis and visualization, including an introduction to data science and another covering core methods in data science, both of which might be appealing for a broad audience.

New Directions for Migration Research

If you are interested in economic research on migration, I highly recommend this presentation at the Center of African Economies 2015 conference. A great deal of fresh, puzzling recently discovered data patterns and exciting directions for future research from a group of young promising economists. The following are some quick notes. I encourage you to watch the whole presentation for your own inspiration.

Michael Clemens on Skilled Migration

– Future research questions: 1) look at surveys with details about migrants: what do they learn? How well do they assimilate into the local labor markets? 2) How does participation in the global labor market affect education systems in the origin country? 3) How should we model human capital externality (this is hard!)?

– Incorrect language can mislead our discussion on migration. “Brain drain”? Think about calling women’s labor force participation rate “family abandonment rate”. How would that feel?

Melanie Morten on Migration in Sub-Saharan Africa

– Facts: much migration in Sub-Saharan Africa in rural-urban, rural-rural, and urban-urban; there are large regional differences in migration flows.

– Apart from migration across sectors, we should think about labor flows within sector across areas as well.

– Potential reasons for the lack of migration in the presence of large wage gaps: 1) selection; 2) people care more than wages (amenities/quality of life, cost of living, congestion, etc), which is the compensating differentials argument; 3) it’s costly to migrate.

– There’s evidence for all three channels mentioned above. This points to the need for further understanding of the costs of migration and potential roles of policy to foster a better match between skill and location.

– Policies that are space based (e.g. subsidies) might prevent people from migration by making it easier to stay in low-productivity places. (I found this particularly interesting).

Clement Imbert on Seasonal Migration

– Seasonal migration is especially important for risk coping in certain areas with agricultural lean seasons (India, Bangladesh, etc); but the contexts are similar.

– Data: NSS data, REDS data, RICE survey.

– Two sectors: construction (spot market) and manufacturing (network/referral matters).

– Need to consider the general equilibrium effects of seasonal migration (on urban wages, amenities, etc).

– Little data and work done on seasonal migration in Africa.

Q&A:

1. How to study forced migration due to climate change or conflicts?

Morten: This is challenging to model because migration is not a choice. Current research has focused on the impact of large migrant inflows on the local economy. See Ran Abramitzky‘s work. In the case of climate change, which is a permanent shock, political economy might matter and the distributional impacts are important. See Esteban Rossi-Hansberg‘s recent work (he presented at Duke two weeks ago and it was fascinating!).

2. Are migrant non-migrant wage differences only seen in developing countries?

Morten: No. It seems people value amenities differently. In developing countries, the benefits can be similar with those in developed countries, but the costs can be very high. See my paper on nonparametric estimation of migration costs (my comment: gravity model in trade?). Thinking about differences between developing and developed countries can be useful for policy intervention.

Clemens: There is a broader literature on “Why don’t people make profitable investments?” There’s a lot to explore in developed countries — why do people stay in the Appalachia when they can move to Miami?

3. How do property rights affect models on migration?

Morten: This is a first-order consideration. See paper in AER: as people get formal property rights, they migrate more. Part of the lack of migration might be due to property rights.

4. What’s the role of information on migration?

Morten: Experiments. Need to find high quality data on wages that people can trust. There are many open questions and possibilities to extend existing models.

Imbert: Information seems to be more easily available for lower-skilled jobs.

IZA Evaluation Dataset Survey

IZA just published this new panel data set tracking unemployed individuals in Germany.

The IZA Evaluation Dataset Survey (IZA ED Survey) is a novel panel survey which tracks the employment history, behavior and individual traits of a large, representative cohort of individuals. The IZA ED Survey covers a panel of 18,000 individuals who registered as unemployed at the Federal Employment Agency in Germany between June 2007 and May 2008. The individuals were interviewed up to four times over a time span of three years, starting at their entry into unemployment. This data allows the researchers to observe dynamics with respect to individual and labor market characteristics during the early stage of unemployment, as well as tracking long-run outcomes. Within the survey, information on labor market activities, ALMP (Active Labor Market Policy) participation, migration background, search behavior, ethnic and social networks, psychological factors, cognitive and non-cognitive abilities, attitudes and preferences was recorded. Its large sample size of individuals entering unemployment, in combination with its broad set of variables and the measurement of unemployment dynamics offers many new perspectives for empirical labor market research.

A detailed description can be found here. Seems like a good resource for labor economists.

New Data Repository on Randomized Controlled Trials

Data from Randomized Controlled Trials conducted by IPA/J-PAL are available from Harvard:

http://thedata.harvard.edu/dvn/dv/socialsciencercts

The availability of more data on RCTs will hopefully spur more meta analysis across studies and shed light on the design of effective development policies in different contexts.