3 days at the Oxford CSAE conference
The CSAE Conference 2016: Economic Development in Africa was held in Oxford this week (20th – 22nd March 2016). This high profile conference provided key opportunities for development economists to share their latest research.
From EDI perspective, what shall we remember about this conference?
- EDI Tanzania. It was great to see data collected by EDI being used in several presentations. This reveals the importance researchers put in data quality. EDI has been collecting data over 250 000 respondents over the past 13 years. At EDI academic rigor and policy relevance complement each other, which explains why the data we collected were used in so many research articles. During sessions, data quality and measurement issues regularly appeared to be of high concern, confirming that, at EDI, we should continue to push the frontiers of information provision.
- Measurement issues. There was a lot of great research on measurement issues. Many presentations provided significant insights on how to measure poverty, income, agricultural labour, etc. One particular highlight was the presentation on Measuring Household Labor on Tanzanian Farms by Amparo Palacios-Lopez (World Bank) and co-authors. With a survey experiment, they offer a stimulating discussion on how to measure family farm labour and therefore labour productivity in the Tanzanian context. They randomly assigned households to one of the four following categories (i) households reporting agricultural labour in weekly in-person visits (ii) households reporting agricultural labour in weekly phone surveys (iii) households reporting agricultural labour in a single post-harvest recall survey, per the Tanzania National Panel Survey and (iv) households reporting agricultural labour in a shorter version of the Tanzania NPS post-harvest recall survey.Their results provide significant contributions to the literature on labour recall modules and on the agricultural productivity gap.
- Data gap. During a plenary session, the World Bank presented its report “Poverty in a Rising Africa“. This report stresses the lack of good quality data in African countries making it difficult to track poverty trends. The presentation was concluded with the following “Better data can lead to better decisions and better lives”.The natural conclusion of this, is that more efforts and resources should be allocated to high quality surveys in African countries.
- Data quality. With the rise of open data initiatives, more and more data sets will be available to development economists. Are all data sets of high quality? Certainly not. But how to measure the quality of a data set is not an easy question. During the World Bank plenary session, two criteria were underlined, the presence of outliers and missing consumption data. This question should be further discussed among data users and data providers. Even with the most sophisticated econometric techniques, bad data cannot lead to insightful results (and therefore useful policy recommendations).
- PAPI versus CAPI. Some researchers still rely on PAPI (Pen-and-Paper Interviewing) which makes data more prone to measurement errors. CAPI (Computer-Assisted Personal Interviewing) is the way forward and the interested reader should be convinced by reading the following article. Simply put, CAPI is the first step in cost-efficient, high quality data collection (to illustrate, have a look at the work done by my Surveybe colleagues). Adding validations, capturing images and GPS coordinates, using paradata for quality control activities, etc. are some functionalities which without a doubt help survey managers in improving the quality of their data.
By attending to the conference and talking to participants, as an EDI researcher, I am all the more convinced of the absolute necessity to pursue research on data quality and measurement issues. Data is what we, development economists, rely on and we must ensure that we are always working towards improving its quality.