The Global Welfare Dataset
(GLOW) is a cross-national panel dataset that aims at
comparative social policy research on the Global North and Global South. The database includes 381
variables on 61 countries from years between 1989 and 2015. The database has four main categories of
data: welfare, development, economy and politics.
GLOW is an outcome of a comparative welfare politics research project, Emerging Welfare (emw.ku.edu.tr), funded by the European Research Council (ERC Grant number: 714868, Principal Investigator Dr.Erdem Yörük).
The data is the result of an original data compilation assembled by using information from several international and domestic sources. Missing data was supplemented by domestic sources where available. We sourced data primarily from these international databases:
- Atlas of Social Protection Indicators of Resilience and Equity – ASPIRE (World Bank)
- Government Finance Statistics (International Monetary Fund)
- Social Expenditure Database – SOCX (Organisation for Economic Co-operation and Development)
- Social Protection Statistics – ESPROSS (Eurostat)
- Social Security Inquiry (International Labour Organization)
- Social Security Programs Throughout the World (Social Security Administration)
- Statistics on Income and Living Conditions – EU-SILC (European Union)
- World Development Indicators (World Bank)
However, much of the welfare data from these sources are not compatible between all country cases. We conducted an extensive review of the compatibility of the data and computed compatible figures where possible. Since the heart of this database is the provision of social assistance across a global sample, we applied the ASPIRE methodology in order to build comparable indicators across European and Emerging Market economies. Specifically, we constructed indicators of average per capita transfers and coverage rates for social assistance programs for all the country cases not included in the World Bank’s ASPIRE dataset (Austria, Belgium, Bulgaria, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Luxembourg, Netherlands, Norway, Poland, Portugal, Romania, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, and United Kingdom.)
what is GLOW
what is EMW
Our recent review of the literature (Yörük et al 2019) on welfare modelling identifies a significant advancement from Esping-Andersen’s original Three Worlds study (1990), with a commendable effort by several scholars who aimed at extending the welfare regime analysis to other cases around the world by utilizing different methods. Yet, this review illustrates that the selection of variables used in this effort has however fallen short of fulfilling this ambitious commitment of extending the debate on welfare regime types both methodologically and geographically.
We have shown that the use of genuine “welfare policy variables” is mostly limited to studies focusing on OECD countries or those analysing Esping-Andersen’s original 18 country cases. One of the reasons behind this narrow focus could be the lack of available datasets on welfare policies for non-OECD countries. For the studies on non-OECD countries, our analysis reveals that developmental outcome variables, and to a lesser extent, contextual and political variables, are commonly used in place of welfare policy variables, creating a issue that we identified as the “variable selection problem”. Despite various implicit and explicit links between welfare policies and developmental outcomes, interchangeable use of these two large variable categories is harmful as it may blur the boundaries between comparative developmental analysis and comparative welfare regime analysis. Also, despite the increasing variety of the methods utilized, the literature has been increasingly detached from the original study of Esping-Andersen and under-utilization of the variables used in Esping-Andersen’s original study implies a discontinuity in the welfare regime literature. We also observe that, across the methodological spectrum, studies utilizing data reduction methods or mixed-regression methods have under-utilized welfare policy variables compared to developmental outcome variables. This poses a challenge to the literature because it can potentially create a validity problem across different studies.
The welfare regimes literature falls short of using genuine welfare policy variables—and within such variables, especially the original Esping-Andersen ones. We construe this as a problem because underuse of welfare policy variables leads to a validity issue, in that the measurements intended for the welfare state do not give results for welfare efforts when developmental, contextual, or political variables are used. Instead, students of the welfare regime literature must confine themselves to the exclusive use of welfare policy variables in operationalizing welfare state regimes, and this effort has to be undertaken especially for non-OECD countries, where other types of variable currently dominate existing empirical studies.
Therefore, we suggest that in order to operationalize welfare state regimes, only welfare policy variables should be used. Also, Esping-Andersen’s original variables should be used in the Global North and South to maintain validity and reliability. In this endeavor, the main criteria for selection of variables should be theoretical requirements of conceptualization, rather than data availability, which is naturally expected to impose certain constraints.
Availability of data stands as one of the main challenges confronting the scholars undertaking welfare modelling research when making their decision on methodological orientation and case selection, and such choices in return constrain the quality of the prospective findings from these studies. Thus, utilization of new datasets with a wider coverage of nations as well as welfare policy variables could help explore other regions/country cases for welfare modelling research without validity problems. Currently, there is a mismatch between existing datasets and the theoretical requirements of the literature, which increasingly aims to make global points through large-scale comparisons. This can only be resolved by developing new datasets that contain welfare policy variables for both Western and non-Western countries.
With the release of GLOW, scholars of non-Western welfare regimes can pay more attention to the possibility that a global welfare regime theory and classification is viable. We believe that this new database and the resulting availability of new welfare policy variables will help enable the scholars of welfare regimes in the West and non-West to conduct analyses with the variables that would fit into scholar’s theoretical concerns and to help overcome the variable selection problem in the Three Worlds literature.
References to the codebook should be made as:
Yörük, Erdem, Gabriela Ramalho Tafoya, İbrahim Öker, Ali Bargu, Alper Şükrü Gençer, Rahmi Çemen, Fuat Kına, Çağrı Yoltar, Burak Gürel, Murat Koyuncu. 2019. “Global Welfare Dataset, Version 2019.” Available at https://glow.ku.edu.tr/download
- Esping-Andersen, Gosta. 1990. The Three Worlds of Welfare Capitalism. Princeton, N.J.: Princeton University Press.
- Yörük, Erdem, Öker, İbrahim, Yıldırım, Kerem and Yakut-Çakar, Burcu, (2019) “The Variable Selection Problem in the Three Worlds of Welfare Literature”, Social Indicators Research.