Skip to main content

The New Politics of Welfare: Towards an “Emerging Markets” Welfare State Regime

Periodic Reporting for period 2 - EmergingWelfare (The New Politics of Welfare: Towards an “Emerging Markets” Welfare State Regime)

Reporting period: 2018-07-01 to 2019-12-31

Can we say that emerging market economies are developing a new welfare regime? If so, what has caused this? This multimethod and interdisciplinary project seeks to answer these questions by investigating the politics of contemporary welfare state development in emerging market economies. It builds separate databases for welfare and political protests and then employs quantitative, qualitative and comparative methods to describe and explain the politics of welfare in select emerging markets, namely Argentina, Brazil, China, India, South Africa and Turkey.
During the neoliberal period, the countries in question have developed extensive social welfare programs. They have expanded existing social security programs and introduced generous social assistance programs for the poor. The project is based on the idea that this welfare development in emerging markets is not only a quantitative expansion, but it corresponds to a radical qualitative shift in the history of the welfare state. The project argues that (i) emerging markets are forming a new welfare regime that differs from welfare regimes of the Global North on the basis of expansive and decommodifying social assistance programmes for the poor (Hypothesis 1) and (ii) this new welfare regime is emerging principally as a response to the growing political power of the poor as a dual source of threat and support for governments (Hypothesis 2).
The project challenges and expands the state-of-the-art in different literatures by developing novel concepts and approaches. The project’s first contribution is to the existing welfare regimes literature. This literature has created a “race to discover new regime types”, but it has radically slanted towards Western or OECD countries in case selection. Except for a few studies, literature has ended up with six welfare regimes for around 20 developed countries, while mostly ignoring the rest of the world. Although a few “expansionist” scholars believe in the global coverage of the welfare regime perspective, most of these expansionists have focused only on the Global South, and not compared it with Western countries. Also, existing expansionist studies have mostly identified “geographically based cluster of states”, without considering the possibility that geographically, culturally and historically distant emerging markets are converging into a new welfare regime as a result of shared political dynamics. Finally, due to data limitations, these expansionist scholars have often times failed to use welfare policy variables. Instead, for clustering purposes, they have utilized developmental variables as a proxy, such as GDP per capita, immunization rate, life expectancy etc. as well as other variables that represent broader economic context or political dynamics. This has created serious validity problems and made it hard to compare studies on the Global South with the ones analysing Western countries. All these have undermined the possibility of reaching at a global welfare state theory. This project contributes to this literature by illustrating the contemporary global welfare state regime structure where emerging markets join as separate clusters.
The project’s second contribution is to the contemporary welfare state development and contentious politics literatures. The welfare state development literature, especially in the non-west, has been dominated by structuralist explanations, underestimating the effect of political factors—particularly the grassroots politics. The dominant structuralist paradigm emphasizes demographic and economic changes and argues that welfare reform is essentially a natural result of aging, labour informalization, unemployment, globalization, deindustrialization, the rise of poverty and the rise of the service sector. The project challenges this structuralist paradigm and investigates the political causes of recent welfare systems changes in emerging markets. The project argues that these political causes have determined the actual trajectory of welfare policy changes, mediating structural pressures, shaping specific policies and regulating the way benefits are distributed. The project contributes to the literature by generating quantitative and qualitative comparative data on welfare and politics and by testing whether this welfare state expansion is caused by particular changes in grassroots politics.
The project will show that the poor are not passive victims of recent economic transformations, but instead are active political agents that shape welfare policies. Thus, the project will not only will take us beyond the state-of-the art of welfare state studies, but will also be highly politically relevant. The number of the poor and the problem of social exclusion have been growing all around Europe, due to economic crises, immigration, deindustrialization. European welfare states have been struggling to deal with this situation. It is likely that existing poverty and social exclusion in Europe will be radically politicized with the intervention of political actors ranging from radical Islamist to the far right. As a result, the social and political problems most strongly occasioned by emerging markets may prevail in Europe as a permanent political crisis. Hence, if the hypotheses of this project hold true, i.e. political power of the poor leads to a new welfare regime, it can be expected that similar dynamics lead to the development of similar welfare policies across Europe, as well.
In order to test its main hypotheses, the project has been organized around four work packages. Work package 1 (WP1) is designed to understand the global welfare state regime structure and its transformation. As part of WP1, the project team has built the first comparative Global Welfare Indicators Dataset (GLOW) and conducted cluster analysis confirming the first hypothesis of the project that emerging markets have developed a new welfare regime different from the welfare regimes of the Global North. You can find detailed information about the GLOW on www.glow.ku.edu.tr. (alternative url: https://glow-emw.herokuapp.com/). By using the GLOW, we employed a cluster analysis of 52 Emerging Market and OECD countries and find four overarching welfare state regime families. These results confirm H1, demonstrating the emergence of a new welfare regime in emerging markets.
To test the second hypothesis about the political causes of welfare provision, the project designed three work packages. In work package 2 (WP2), the project’s computer science team builds a software that automatically generates the first Global Contentious Politics Database (GLOCON) for emerging markets. This work package is ground-breaking as it is the first study to employ advanced computing techniques such as artificial intelligence, natural language processing and machine learning to extract protest data from online news sources. We have managed to develop generalizable computational tools that operate independent of the context and fully-automate the event extraction process from a variety of sources from different countries and in different languages. As part of WP2, the project will generate contentious politics databases for each case country, enabling the project to identify protest trends over time and statistically analyse their effects on welfare provision in these countries. The main methodology we apply is the state-of-the-art deep learning methodology, based on transfer learning. More information about GLOCON, including initial results can be found here: https://drive.google.com/drive/folders/1SBMsLnSIvHEU6MAWBNYe2KbAKTtk6h8A

In work packages three and four (WP3 & WP4), the project establishes causal relations between politics and welfare policy outcomes for each case country. While WP3 uses advanced statistical methods (cross-sectional regression analyses and pooled cross-sectional time-series regression analyses) on longitudinal, large-N datasets. The project team has so far conducted cross-sectional regression analyses and pooled cross-sectional time-series regression analyses for Brazil, China, India, South Africa, Mexico, Turkey as well as at a global scale. The results of these analyses indicate a historically situated causal process between protest events and politicized identities and social assistance (level of social assistance increases with the level of protest activities of the poor over time after controlling for country-level structural factors), providing strong evidence for H2.

The team has been working on a total of 30 manuscripts, some of them have already been published at leading academic venues, such as World Development, Governance, Social Policy & Administration, Social Indicators Research and European Review. The PI has signed a book contract from University of Michigan Press for a book from the project. These analyses have been presented at international conferences, (ASA, ISA, APSA, MPSA, EASP etc). The PI has edited a special issue for the leading political science journal Governance on the topic of the project. He is invited to contribute one chapter for Oxford University Press, one for Palgrave and one for Edward Elgar Publishing.

WP4 focuses on qualitative analysis of official policy documents and parliamentary proceedings to demonstrate the significance of grassroots political activity in welfare policy making. We reviewed the scholarly literature on social policies and politics of the poor and we examined policy-making processes of case countries and designed different document collection strategies for each country context to create the most extensive official document databases possible. We have completed literature review for all case countries, document collection for four case countries (Brazil, India, South Africa and Turkey) and the coding and analyses of two countries (Brazil and Turkey). We authored two articles based on the results of our analyses and presented these papers at international conferences.
The combination of computational, quantitative and qualitative methods is unprecedented in comparative research and seeks to open up new horizons for social sciences and also computational social sciences.

As part of WP1, the project team has built the first Global Welfare State Indicators Dataset (GLOW), collecting and harmonizing data from 52 countries, for the period between 1985-2015 and for 381 variables. As a contribution to the literature, the project team has collected welfare policy variables that used to be available only on advanced OECD countries. As such, the GLOW enables comparisons between OECD and emerging markets using welfare policy variables, rather than developmental variable proxies, allowing cluster analysis at a global scale.

By employing cluster analysis methods on the GLOW, the project has confirmed its first hypothesis. Welfare regimes literature to a large extent focuses on Western countries, while those few studies analyzing the non-West concentrate on geographical or cultural patterns that are believed to generate non-Western welfare regimes, falling short of producing a global welfare regimes theory and analysis. The project contributes to this literature by demonstrating that emerging markets are forming a new welfare regime that differs from welfare regimes of the Global North, based on analyses that rely exclusively on welfare policy variables made available by the GLOW.
The project team’s efforts to generate a global protest events database (GLOCON) for emerging markets (to test H2) has also significantly contributed in the advancement of the state-of-the-art in computational linguistics. Cross-context generalizability in text processing is an unresolved task in the fields of machine learning and natural language processing (NLP). Existing text processing tools do not perform well on data other than the one used to train and validate them. The project’s WP2 team has addressed this problem by generating a Gold Standard Corpus (GSC) that utilizes the domain knowledge of social scientists and relies on optimization of annotation processes. This GSC provides a basis for experimentation and comparability. Coupled with computational techniques (such as transfer learning, domain adaptation, data augmentation) the project’s work bears the potential of resolving the generalizability problem in the literature and improving cross-country, cross-language and cross-source usability of ML and NLP tools.
The project team has at the same time provided strong support for the second hypothesis by using available administrative data and micro data (quantitative analyses) and by examining policy documents (qualitative analysis). In different parts of the world, we have captured a global pattern in which social assistance programs are being used to contain social unrest and especially violent unrest. This is a ground-breaking finding.

For the remaining 2.5 years of the project, we expect to see the following results:
1. We will update GLOW dataset with more data points.
2. We will finish the GLOCON dataset and release it for the scientific community.
3. We will make important contributions in the field of natural language processing by developing a cross-language cross-country machine learning tools.
4. GLOCON dataset will provide us with the monthly and district level panel data with several key characteristics. We will test our main political hypotheses with panel data analyses. We have already confirmed our hypotheses with existing micro and macro data, but we will now establish them with additional time series data.
5. I will publish my book on Turkish welfare regime politics from the University of Michigan Press. I will send my new book proposal on the comparative politics of welfare to a top university press.