Mapping Social Policy in Economics Research: An Analysis of Core Journals

This study reviews the attention core economics journals have paid to the topic of social policy and economic performance based on 1,356 articles published between 1957 and 2020 using bibliometric analysis and data visualisation techniques. This approach allow us, based on the assessment of bibliometric data through co-citation, co-authorship and keywords networks analysis, to identify researchers, countries, institutions and research areas and trends offering useful information that facilitates future research and policy making. The emphasis is on the empirical literature given the theoretical controversies on the sign of the relationship and its usefulness for policy making. The most active country publishing in the core journals is the USA, with Sweden leading in terms of average citations. The top publishing authors and institutions are in the US, and China is the only developing country present.


INTRODUCTION
Social policy overlaps with other disciplines in the social sciences, including economics. The welfare state and the set of policies that governments use to promote welfare and provide social protection is not a new topic in economics research as far as its consequences for allocative efficiency and macroeconomic performance are concerned. [1][2][3][4][5][6][7][8] The deepening of some of the key challenges that the welfare state faces, such as globalisation, the fourth industrial revolution, demographic change or immigration, and more recently the COVID-19 pandemics and its foreseen impact on poverty and inequality, claims for a clearer understanding of where we stand in terms of the research on the relationship between social policy and economic performance. Policy decisions aimed at restructuring and rescaling the welfare state should be grounded in scientific research that allows to accommodate the potential impact of those decisions both at the social and the economic level, enabling policy makers to make more informed decisions. [9][10] M. Weir [11] defines the welfare state as "… a state that is committed to providing basic economic security for its citizens by protecting them from market risks associated with old age, unemployment, accidents, and sickness." The measurement of this wide and complex concept is done most often using social spending, [12] the main instrument of implementation of social policy. The literature suggests that the welfare state and associated social policy have the potential to influence economic performance, but there is no agreement as to the sign of this effect. [2,4,13,6] Theoretical arguments pose that a positive effect emerges when the welfare state creates the conditions for economic agents to make decisions that promote growth, such as taking more risks which enhance innovation or investing in physical and human capital. Nevertheless, the need to finance the welfare state introduces distortions in economic decisions that are detrimental to economic performance, for instance working less and with less effort or decreasing savings and in this way investment. A bibliometric analysis of applied studies can provide important guidance since in a context where there is no theoretical agreement on how social policy influences economic performance the issue becomes essentially empirical.
The main objective of this study is to conduct a research synthesis of the relationship between social policy and economic performance based on a descriptive overview that maps and characterizes this field of study in terms of relevant authors, Journal of Scientometric Research, Vol 10, Issue 2, May-Aug 2021 according to the number of times they cite each other) and co-occurrence of keywords (analyses the number of documents in which keywords occur together). Philipp Korom [21] provides additional technical details on the construction of maps by VOSviewer, a process that he describes as consisting of three steps, normalization, mapping, and clustering. Literature related to a specific field of study can be analyzed and synthetized using a range of different methodological approaches that include, among others, systematic literature reviews, narrative bibliographic analyses, meta-analysis, content/text analysis and SNA. [22] We focus on SNA because it is a useful methodology to find patterns, relationships in networks, and identify gaps that give directions on potential paths for future research, which is not always possible with other statistical and nonstatistical approaches. Nevertheless, the SNA implemented in this study has some limitations including the analysis of a single bibliographic database (WoS in our case), the clustering method (its interpretation depends on researchers' view) and the identification of bilateral links (even when there is only a unilateral relationship). Figure 1 summarizes the processes involved in our approach and the intended results. There are three main steps in the design of our analysis.
Step 1 involves identifying the relevant publications from the selected core economics journals (Table 1) with the aim of retrieving bibliographic information on key publications that deal with the topic under analysis. Having gathered a publications dataset, Step 2 next extracts and organizes the relevant bibliographic data for quantitative analysis e.g. author's name, cited references, the document's title, keywords and abstract. Finally, Step 3 corresponds to the construction of co-authorship, co-citation and keywords countries, institutions, research areas and trends, by applying social network analysis to bibliometric data. More specifically, our aim is twofold: (i) take stock of the literature in terms of authors' contributions, geographic clusters, institutions and keywords; and (ii) identify the most important research areas and trends within this general topic. Our aim is not to provide a synthesis of the evidence on the relationship between social policy and economic performance regarding the existence of a positive, negative or non-existing nexus based on published empirical studies, which would require other approaches, e.g. meta-regression analysis. Focusing on research published in 36 core economics journals reinforces the importance of economic analysis for policy making since these are likely the journals that influence policy makers the most. Researchers also pay particular attention to top journals. [14] Besides aiding policy makers in designing better informed social policy from an economic point of view, we believe that this analysis can help researchers to more easily access relevant literature, detect the most often and more recently investigated themes and, 'standing on the shoulders of giants', identify new approaches to the issue under analysis. These results from the fact that our analysis identifies networks of authors, countries, institutions and keywords published in high impact economics journals on the topic of social policy and economic performance. This valuable information may be used to strengthen the evidence from future studies and help to promote scientific-based social policy design.

METHODOLOGY AND DATA
Bibliometric analysis is a useful quantitative approach to summarise developments in a particular field of study and is based on the measurement and studying of scientific outputs. [15,16] In this study, we use bibliometric analysis and visualization methods to characterize the research on social policy and economic performance published in core economics journals over the past decades. This is an appropriate and useful approach to identify and visualize key researchers, countries and institutions and pinpoint main research themes and trends. We analyse bibliometric data applying social network analysis (SNA) which enables the mapping of current knowledge on a particular area of research by highlighting connections between authors, countries, institutions and topics, among other characteristics. [17] We use the VOSviewer (Visualisation Of Similarities) software for constructing and visualising bibliometric networks due to its powerful user graphic-interface and mapping visualisation capability. [18][19][20] This software allows mapping items and link them by their relatedness generating comprehensive and easy to interpret maps. These features include co-citation analysis (which highlights the relatedness of items based on the number of times they are cited together), co-authorship analysis (accounts for the number of co-authored documents), citation analysis (relates items  Table 1 for the list of 36 core economics journals and Table 2 for the search terms used in step 1. networks with the bibliometric data gathered in the previous step using VoS. The main result from this last step is pinpointing the most relevant (number of publications and citations) authors, countries, institutions and research areas and trends.
Due to their potential more influential role in policy making, we perform bibliometric analysis for core economics journals selected based on the respective high scientific influence (see Table 1). In this list the inclusion of the "Blue Ribbon Eight" journals is mandatory given its stability in rankings of academic journals in economics [23] and its wide scientific recognition. [24] Since the search restricted to these journals resulted in scarce information, we added relevant journals from other established rankings. Art Diamond [25] identified a set of core economics journals, known as the Diamond list, based on citations data from the Social Sciences Citation Index. This approach has been criticised for its arbitrary weighting criteria. M. Burton and E. Phimister [26] reassessed the original Diamond list by applying different ranking techniques, such as Data Envelopment Analysis (DEA). More recent influential studies [27][28][29] propose a new set of core journals. We take as a starting point the "Blue Ribbon Eight" and consider the number of times (at least two) each of the journals in the top 50 of each ranking is mentioned in the other rankings to select our list of core journals in economics, which resulted in the selection of 36 journals. Table 1  The articles were collected from the WoS database considered one of the most complete bibliographic data sources that provides the meta data on research outputs that is needed for bibliometric analysis, including title, authors, abstract, keywords, references, citations, as well as a longer time span. [30,31,21,32,33] Our main objective is to give an overview of research on the relationship between social policy and economic performance from an applied perspective in terms of publication and citation activity, prominent authors, countries and institutions publishing research on the topic and research areas and trends. We thus performed our search of the 36 journals using keywords associated with public social expenditure, its synonyms, and dimensions, and simultaneously references to their economic impact and the applied nature of the studies. Hence, our search included three main groups of terms/keywords: Welfare State, Economy and Empiric. For the first group, we selected the strict and generic way to obtain the set of articles of interest and in the cases of articles that do not mention this group in a generic way we decided to look for references to the main components of social spending. As far as the economy group is concerned, since the reference to economic performance can be done using quite diverse terms we tried to apply a broad set of associated words with a loose check with the support of the operator NEAR. For the last group, we decided to apply the set of words that most commonly appear in studies of an empirical nature. To restrict our search to those three groups we included the "AND" particle as a separator between each main terms/keywords group. To widen our search, we applied these keywords in a non-exclusive way within each subgroup (separated by the "OR" particle) and in different fields of the selected articles (title, abstract and keywords). See Table 2 for a summary of this search and the specific terms/words used.
The final search of relevant articles published in the 36 core journals in economics over the period 1957-2020 was performed on the 16 th of September of 2020 and retrieved 1,356 articles.

RESULTS AND DISCUSSION
The current study was designed to examine the attention core economics journals have paid to the topic of social policy and economic performance between 1957 and 2020 in terms of: (1) growth of literature as far as publication and citation activity is concerned; (2) core authors/researchers in the field (co-citation analysis); (3) prominent institutions and countries engaged in the research (co-authorship analysis); and (4) prominent research areas and trends (keywords networks). This valuable information may be used to strengthen the evidence from future studies on the topic and to help to promote scientific-based social policy design.

Publication and citation activity
The attention devoted by the selected journals to the topic is presented in Figure 2, containing annual data on the number of publications and citations. This raised concerns on the ability of significant parts of the population to recover from economic hardships and the role that social policy could play as a buffer, but also on the need for welfare state retrenchment resulting from the need for fiscal consolidation in many European countries [7,34,35] The deepening over the course of the new millennium of the challenges posed to the welfare state by globalisation, population ageing, migration flows or artificial intelligence (AI) and automation [36][37][38] can also have contributed to the steady interest on the welfare state and

Main terms/ keywords group
Bolean operator relative to the next main terms/keywords group

Search terms and bolean operators within each main group
Welfare State "Social expenditure*" OR "government transfer*" OR "Welfare State" OR "Social Security" OR "Social protection" OR "education expenditure*" OR "education spending" OR "health expenditure*" OR "health spending" OR "Public spending" OR "Government size" OR "social spending" OR "pension expenditure" OR "old-age retirement income" NEAR/3 Economy "growth" OR "GDP" OR "per capita income" OR "economic performance" OR "economic activity" OR "economic impact"

AND Empiric
"Empiric*" OR "applied" OR "estimate*" OR "regression"   articles), England (138) and Germany (93). These countries also tend to work more with each other, generating a stronger bond (and probably more publications) as reflected in the thicker lines connecting them. As for average citations, the top countries are Sweden with 51.47 citations per published article (despite having low scientific production/outputs) and the USA comes second with 48.18 citations. The location of the USA in the top positions is not surprising and is common to most research areas. [14] Sweden's position at the top confirms the long-standing interest on welfare state related issues in the Nordic countries. Germany is the country of birth of the conservative corporatist or Bismarckian welfare state regime and England is the landmark of the liberal welfare regime. It is thus not surprising that issues related to the welfare state and social policy raise particular interest in all these countries. The only countries not located in North America or Europe that appear in this ranking are Australia (31 articles), China (21), Israel (23) and Japan (21), and only China belongs to the group of developing economies. Since a multitude of developing countries are also suffering from increasing inequality and other important social and economic challenges (globalisation, automation and AI, deindustrialisation) that can leave a significant part of their population facing economic hardships and social exclusion, this result suggests that developed countries should support the developing regions in social policy and economic performance related research in order to better address those challenges on a global scale. for Robert Barro, despite not having linkages with every author in this network, he records the greatest link strength outside his cluster with Gary Becker. This could result from the fact that albeit he is not as versatile as other authors he manages to be more consistent in terms of co-citation. This cluster also seems to be the most influential one (more central). Given the contribution of (international) research collaborations to the advancement of knowledge by enabling more effective and efficient scientific efforts, our results suggest that the identified US based researchers can play a leading role in promoting research on the link between social policy and economic performance that better caters the needs of developing countries. Multi-country teams of researchers, including those most experienced based in the USA, can help to enhance the design and implementation of studies that provide more reliable results for developing countries.  does not mean that being productive is not important since the NBER, the MIT, CEPR and Harvard University also record high levels of average citations and these four institutions present, among those located at the top of the citation list, some of the strongest connections.

Keywords networks
Keywords or main terms co-occurrences are patent in Figure 6, where the node size depends on the number of occurrences of the respective term, curved lines connect terms (nodes) that occur together and the different colours identify clusters according to average keyword per year scores, i.e. the average publication year of the documents in which a keyword occurs 1    such as unemployment, sickness, incapacity, death, or old age. Some authors argue that because of globalisation, deindustrialisation and automation and AI, social policy should change its focus from social security to social investment. [40,41] Further research on the topic of social investment and economic performance could derive important implications for sustaining economic growth since human capital is a fundamental driver of growth and is mainly accumulated through investments in health and education. [42][43][44] Labour market related interventions (unemployment benefits and active labour market policies) and their economic consequences are other welfare programs that seem to have attracted some attention in core journals, reflected in the frequent use of the terms 'employment', 'work' (39 occurrences), 'unemployment' (25), 'labour supply' (20) or 'labour-force participation' (19). The complementarity or substitutability between the two types of labour market interventions is a source of controversy in economics research and, despite the attention already devoted to these topics, still needs further investigation. The word 'family' (25) also appears on the list but other welfare programs such as housing, incapacity related benefits or survivors' pensions are absent. Figure 6 shows not only the structure of the social policyeconomic performance analysis in economics research but also the time dynamics of this research. Cluster analysis gives us information on how the most important keywords changed over the years and allows for interesting conclusions. Around the year 2006, researchers' biggest concern lied on accumulation and family issues -main keywords: 'intergenerational transfers', 'overlapping generations', 'family', 'retirement', 'older men' and 'labour force participation'. Around the year 2008, the focus kept on retirement issues but also moved to human capital and government decisions -main keywords: 'retirement', 'pensions', 'human capital', 'government', 'model' and 'taxes'. In 2010, the focus revolves around education, health and social inclusion issues -main keywords: 'care', 'health', 'welfare', 'inequality', 'redistribution' and 'education'. In 2012, the focus changed to the labour market impact of the welfare statemain keywords: 'labour supply', 'impact', 'wages', 'employment', 'shocks' and 'returns'. The concern with retirement issues over most of the period seems a natural one given population ageing and the problems it poses to pensions financing systems. The move to human capital accumulation issues is in line with the previously discussed change in focus from social protection to social investment. In 2012, the harsh unemployment effects of the 2007-08 financial and economic crisis were visible and economics research accompanied the global concerns through the analysis of the role played by social policy. In sum, the research agenda changed slightly over time since 2006 until 2012 but results suggest an overlap between the dominant lines of enquiry and the most recent research agenda.
threshold of a minimum of 15 occurrences. Some labels may not be visible in Figure 5 to avoid overlapping labels.
Ten keywords stand out: 'model' (143 occurrences), 'retirement' (138), 'insurance' (130), 'taxes' (89), 'inequality' (74), 'consumption' (71), 'income' (68), 'health' (66), 'united states' (59) and 'behaviour' (57). Some top terms are not very precise in terms of research directions but reflect common approaches adopted in different economics research areas. The importance of the node 'model' is common to many fields in economics that include mathematical and/or econometric modelling to achieve clearer reasoning [14] . It is also an indication of the importance attributed to constructing adequate theoretical framework for understanding the relationship between social policy and economic performance, with studies on the topic developed around both theory and practice. Other more general use terms that reflect common approaches and concerns are 'fiscal policy', 'policy', 'government', 'political economy' and 'impact'. Social policy comprises instruments of fiscal policy that can be used as a buffer against shocks and as a means of structural reforms and these roles deserve further investigation. The term 'impact' reveals that it is important to assess qualitatively and quantitatively how social policy affects economic performance since governments must decide how to distribute scarce resources among different public spending areas and need to prioritise across welfare programs. [39] This assessment is especially important in the aftermath of the sovereign debt crises, and austerity measures that ensued, preventing an increase in public expenditure in many countries. Notice also that the only country that appears in Figure  6 is the United States. This finding suggests that core journals publish mainly articles that focus on the United States case, which is probably related to data availability issues, but is also the predominant location of the institutions of origin of the authors. It also suggests the need for more country specific and cross-country studies that reveal other realities, on the one hand, and that allow for generalisations, overcoming specificities of the US economy.
Other keywords highlight research on specific social policies, in particular retirement. The concern with this topic seems a natural one given population ageing and the challenges it poses to the welfare state and public finances sustainability and is thus likely to remain an important topic in the research agenda, at least for developed countries. Within different social policies it is interesting to notice that 'health', 'employment' (45 occurrences) and 'education' (30) appear in the top half of the ranking and 'human capital' (18) is also part of the list. Both health and education spending are classified as social investment, i.e. 'policies designed to strengthen people's skills and capacities and support them to participate fully in employment and social life' as opposed to social security/ protection that focus on providing protection against risks social policy and economic performance such as meta-regression analysis. This statistical technique analyses the results of quantitative studies to provide a more precise identification of the sign of effect of social policy on economic performance.

CONCLUSION
This study presents a portrait of the economics research on the relationship between social policy and economic performance by applying bibliometric analysis to 1,356 articles published in 36 core journals in economics between 1957 and 2020. Using literature mapping clustering tools based on the bibliometric analysis of the data collected allowed for a quantitative assessment and mapping of the research panorama, including the identification of the main themes, authors, countries and institutions in terms of quantity and relevance (average citations).
Research on the topic published in the core journals remained relatively stable over the period 2000-20 that includes the majority of publications (68%) for the whole period under analysis . After a drop in the early years of the 21 st century, the 2007-8 financial crisis and the associated Great Recession seem to have picked researchers and journals interest on the topic once again. The USA ranked first in the number of publications, followed by England and Germany, while Sweden is located at the top in terms of citations per published article. China is the only developing country that appears in the country ranking, suggesting the need for developed countries to support the developing regions in social policy and economic performance related research. The network of main keywords shows that the literature addresses the topic from different perspectives, but one key common element is the use of mathematical and/or econometric modelling. It also reveals that interest in publications on the theme in the core journals could be enhanced by research in the following topics: assess qualitatively and quantitatively how social policy affects economic performance due to the need to prioritise across welfare programs and public expenditure in general; country specific and cross-country studies that overcome the excessive focus on the US economy; examine how the change of focus of social policy from social security to social investment can impact economic performance; investigate further the economic impact of the varied welfare programs and their mechanisms of transmission since some schemes have been overlooked.
Although our review of the economics research on social policy is limited to 36 core journals in economics, we believe that it represents a useful roadmap for future research and policy making. In any case, there are likely other journals in economics that would deserve an in-depth analysis, namely those with a more applied scope and narrower subject coverage more directly related to welfare issues. The relationship between social policy and economic performance and related social outcomes has also raised a lot of interest among other social sciences, so it would be interesting to extend our database beyond the economics category. Directions for future research include also the application of alternative research synthesis approaches for evaluating and using scientific evidence on