First, some short background for readers who might find this all very confusing. The annual American Economics Association meetings calls for papers early in the year, with submissions due around April. People either submit a paper they have already written, or an abstract for something they hope to have written by time of the meetings the following January. Around the end of July, people learn if their session was accepted, and if it was one of the few sessions accepted for the papers and proceedings (Pamp;P). This is a strange format, since sessions for the Pamp;P are chosen by the program committee before many of the papers are written, and before they have been presented. Papers then do not go through any sort of peer review process ex post, but are simply sent by the authors to the journal and published in May (over a year since submission). The papers still have some prestige (notoriously they used to be published in an issue of the American Economic Review, and there was much discussion of whether people disclosed it was the Pamp;P issue or not - now it is a separate journal called AEA Papers and Proceedings). The papers also get seen by a lot of people, and can garner a large number of citations as a result. But they are very short papers (4-5 pages) and the AEA advises "is recommended that authors not simply create abbreviated versions of longer papers, but instead, use this opportunity to publish an extension of an aspect that has been left out of the full-length paper".
So for a development economist, particularly one with work on an impact evaluation that is a full paper and which the authors hope to publish elsewhere, here are my suggestions for strategies for writing a Pamp;P. This is mostly relevant for the AEA Pamp;P, but other journals have also done a Pamp;P at times (e.g. the World Bank Economic Review has done Pamp;P volumes attached to ABCDE conferences).
Approach 1: Write a paper mostly about program take-up, or about some interesting descriptive details from your baseline survey
This strategy works well if you are early on in the process of an impact evaluation, and are presenting work-in-progress at the conference. It is probably the most common approach. The paper can describe your setting and intervention, and then examine who takes up the intervention. Or it can describe some correlations between baseline characteristics and some other outcome of interest. Some examples of such papers include:
Wage Subsidies for Microenterprises (my paper with Suresh de Mel and Chris Woodruff), published in the 2010 AER Pamp;P - which looked at take-up for an experiment that is now finally forthcoming at AEJ Applied. From this year's AEA Pamp;P (gated): Iacovone et al. look at the take-up decision of women in an on-going business training experiment; Hardy and Kagy use their baseline data to examine the profits gap between male and female owners and show it persists even after controlling for lots of characteristics that they measure in their baseline; Bernhardt et al. use data from their ongoing impact evaluation on an intervention to boost female labor participation to examine how husbands and wives have different views of women's work, what correlates with these beliefs, and what these beliefs are correlated with; Breza et al. who look at take-up of a commitment savings account under different treatment conditions, and whether this take-up provides a signal to managers about these workers. This strategy can also be used with descriptive data from your rounds of follow-up data. An example is Buehren et al. who use their baseline data from a savings experiment to look at the cross-sectional correlates of having a bank overdraft, and then their time series data from their multiple rounds of follow-ups to look at how overdrafts are correlated with salary delays in the time series.
Approach 2: Write a paper that describes a new measurement tool or method developed for your impact evaluation
This can work well in providing a place to have more details about a survey technique or measurement tool that you have worked hard to develop, and be a more prominent place to describe and discuss it than burying it as appendix 8 of your longer paper. An example is: Reducing Information Asymmetries in the Youth Labor Market of Jordan with Psychometrics and Skill Based Tests (this is a paper I wrote for the 2015 WBER Pamp;P with Matthew Groh and Tara Vishwanath). It describes a psychometric testing approach we developed for job seekers, and provides some evidence that these measures have useful signal about labor market outcomes. This was then useful background for the main paper on this experiment.
Approach 3: Write a bigger picture paper that takes a broader look at the problem that motivates your paper
This is your chance to lay out some key facts and big picture ideas that you would typically want to use as introduction to your paper, or as extra slides in the motivating part of a seminar talk. These papers can potentially generate a lot of citations, since they can then serve the same purpose for others. An example is: Why do Firms in Developing Countries Have Low Productivity? (a paper I wrote with Nick Bloom, Aprajit Mahahan and John Roberts for the 2010 AER Pamp;P). This was the Pamp;P for our work on a management experiment in India. It takes a step back from the main paper's focus on management, and discusses several theories for low productivity and what our and other work had to say about them. It now has 183 citations, so is serving that motivation citation role for others too.
Approach 4: Dig deeper into your main paper, perhaps by exploring heterogeneity or outcomes you hadn't looked at in the main paper.
This works best when the main paper is already published, and you are then writing a Pamp;P after learning this happy news. This is the chance to explore additional questions of policy or research interest that didn't have a home in your main paper. An example is: Is personal initiative training a substitute or complement to the existing human capital of women? Results from a randomized trial in Togo (this is a paper I wrote with Campos, Frese, Goldstein, Iacovone, Johnson and Mensmann for this year's 2018 Pamp;P), and took the experiment on personal initiative training we had published earlier in the year in Science, and investigates a question we had then been asked several times by people looking to apply it - did the program work better for less-educated/less-skilled or more-educated/more-skilled women?
Approach 5: Use this forum to share lessons from a failed experiment or part of the experiment
Here there is no concern about cannibalizing another paper, but instead the Pamp;P can serve as a way to share lessons from an experiment that failed. An example is Fischer and Karlan in the 2015 AER Pamp;P, who discuss constraints to firm growth based on an experiment they attempted to do in the Philippines, where ultimately sample sizes and take-up were low.
Approach 6: Report Short-Term Results from Your Evaluation
This makes complete sense, right? - get those early results out there, and then when the long-term results are available, publish those separately. In this year's AEA Pamp;P, an example is Janzen et al., who report short-term results of a livestock transfer and training program in Nepal. But, unfortunately, I think it highly likely that if you publish .5 year results, then come back and try to publish the 2, year results, many referees and journals will feel that they are being asked to publish supplementary analysis to an existing paper, and this will make it hard to publish. So those short-term results may be better disseminated in working paper form only I'm afraid.
Unfortunately, there is a chance that a Pamp;P can reduce the chances of publishing your main paper if it presents even any results on outcomes from the main experiment. So be cautious on this My co-authors and I wereburnt once on this early in my career, when this paper published as a conference paper in the New Zealand Economic Papers was noted as part of a reason to reject this migration lottery paper, even though the emphasis was quite different(the main paper was eventually published in the JEEA). So congrats if you get the opportunity to write a Pamp;P, but be careful how you use it.