This post is about a research article1 I co-authored with Gilang Hardadi. Gilang is a former student from the JEMES program, who is now doing a PhD in industrial ecology at the University of Freiburg.
Just another student project?
I met Gilang in the autumn of 2015 when I agreed to supervise his third semester project. He was interested in quantitative assessment of so called social impacts of production and consumption activities, a research area where our students usually choose a qualitative inquiry approach. He was interested in working with databases and he had also some knowledge of programming in R. Also, Gilang was going to take an internship in New York City shortly after, where data analysis skills were required (sort of). I found this was a rather peculiar combination of skills and interests.
During that period the discussion on combining input-output analysis and social LCA was getting quite popular in the LCA community. My colleague Bo was finalizing an article on the social footprint2 and my colleague Jannick was involved in the final stages of the development of the multi-regional input-output database ExioBase.
After discussing these developments we decided to try an extension of ExioBase to social impacts. The database is already environmentally extended, in the sense that it is not simply an input-output database covering the exchanges between industrial sectors across multiple regions in the world, but it also includes (some of) the emissions generated by these sectors. This allows calculating e.g. carbon footprints per sector. However, the discussion at the time, and still now, was about possibilities for extending this theoretical framework to social impacts as well, in order to cover another pillar of sustainability or - to say it in LCA terms - to do social LCA.
So it all started as a semester project on somehow combining input-output analysis and social impacts. The first step was restricting the scope of the project idea to something manageable with the time and data available. After doing some background research and problem analysis, Gilang figured out that focusing on social impacts related to labor and related to the working conditions was the most feasible solution. Why?
Because this was a relatively easy impact pathway to cover (or at least partly cover): working conditions have an impact on productivity and well-being, which are respectively the instrumental/quantitative and intrinsic/qualitative components of the Human Health area of protection in the Life Cycle Impact Assessment theoretical framework. Moreover, ExioBase already included data on employment and working hours per sector, mostly taken from ILOSTAT, so we thought we could probably include other labor-related indicators as well, using the same data source.
But which indicators were relevant for social impacts related to the working conditions in different sectors? We decided to settle for unemployment and for occupational accidents. The former has an impact on productivity and well being, the latter on well being only. Additionally, we recalculated data for employment and working hours, in order to compare some of the new results with previous ones and thus validate the model on a reference.
From project to paper…remotely
Jannick kindly provided us with an older version of ExioBase that Gilang could use as starting point for the analysis. In theory…we could just take the data from one database (ILOSTAT) and smash them into the other database (ExioBase). In practice, however…the two databases have different spatial and industrial sector coverage! What Gilang did was basically a meticulous work of matching, disaggregating, fitting ILOSTAT data with ExioBase industrial classification of sectors, plus filling the data gaps for missing countries via collection of new data from the statistical agencies of these countries or interpolation from the other countries via regression analysis. On top of that, we performed a comparison of old and new values of employment and working hours. We also performed a calculation of social impacts measured in Quality Adjusted Lifetime Years (QALY) for the case of unemployment and occupational accidents, thus a real impact assessment step, inclusive of conversion into a monetary units, for social impacts.
All this didn’t really happen that smoothly. First Gilang did a base version of this analysis for the semester project. With some extra work, the study had the potential to be turned into a scientific article, maybe (I asked some colleagues for an opinion on this…Some said no, some said yes…). At last I noticed a special issue call in the [Journal of Industrial Ecology][http://onlinelibrary.wiley.com/journal/10.1111/(ISSN)1530-9290] and we decided to go for it, despite the short deadline (1 month!). Gosh, but we made it. We were not even half way yet: to comply with the major revisions request by reviewers we recalculated all results with an updated version of the database. On top of all this, the entire work was done investing substantial amounts of free research time and working remotely and time-lagged from various offices, cafés, libraries, internet points in Denmark, USA, Italy, Indonesia…
The proof of concept
What’s the scientific value of this work? The material result is a long list of values of unemployment and accidents data per sector level per country, plus an estimation of their impact and algorithms on how to repeat this analysis. There are some heavy assumptions behind these values, for example disaggregating the number of unemployed for a macro-sector to its industrial sub-sectors based on the total economic output of these sub-sectors. It’s also unrealistic that many readers will spend time checking one by one all these values.
The work was thus framed as a proof of concept were the focus is not much on the numerical values as such but on showing that this particular procedure is feasible. That is, we showed how it is actually possible to extend an existing input-output database to new indicators related to labor and working conditions, how it is possible to fill data gaps in various ways and validate the results, and to use this extension as starting point for the assessment of the specific social impacts all the way through endpoint values in monetary units.
Summing up on this experience, the real proof of concept was that combining teaching and research is possible, even when working remotely and even on strict deadlines. One regret though: we didn’t manage to drink a beer to celebrate the paper acceptance…