趨同與分歧：社會投資福利國家的聚類分析

Abstract

20世紀90年代福利國家再次陷入發展的十字路口。為了使戰後福利國家更好地適應當代知識經濟的風險和需求結構，社會投資成為指引福利國家改革的綜合範式與指南，引發了一場寂靜的「範式革命」。 當前對福利國家社會投資轉向的研究，在地理上分為兩個中心：它們要麼聚焦于歐洲福利國家，認為福利國家的社會投資轉向大致有「四個世界」；要麼聚焦于東亞福利國家，認為東亞福利國家的社會投資轉向趨向分化。並且，這些文獻：在概念操作化上，僅關注單一或幾個投資類政策工具的組合。在方法論上，對社會投資範式的衡量存在顧此失彼的瑕疵，未能突破定量與定性的隔閡，即使有部分的定量研究，也未能將政策支出和產出變數整合起來。在數據分析上，藉助較短時期的截面數據無法建立堅實的因果解釋機制。 為此，作者試圖去追問：針對既有研究在範式測量上的不足，在方法上有何改進的空間？在改進後的範式測量基礎上，以歐洲和東亞為中心的文獻是否有整合的可能？東亞福利國家與歐洲福利國家向社會投資範式轉向時存在怎樣的相似性與差異性？在研究結果上，如何解釋這種相似性與差異性，是否有理論創新的可能？ 通過對現有數據的盤點，作者將OECD社會支出資料（SOCX）、全球福利數據集（The Global Welfare Dataset，GLOW)及福利國家比較資料集（Comparative Welfare States Data Set，CWS）進行整合，最終形成了一個涵蓋19個樣本福利國家1990-2019年的面板數據。在對樣本福利國家的社會支出總體和分類描述的基礎上，利用標準化後的政策支出變數、政策產出變數以及政策支出-產出聯合變數進行了三次聚類分析，結果表明，除了日本、葡萄牙在聚類集群中的歸屬略有不穩定外，其它所選樣本福利國家相對穩定地聚集為「五個世界」，以政策產出變量以及支出-產出聯合變量聚類時，聚類結果近乎一致，特別是社會民主體制與自由主義體制相對穩定，而地中海體制的葡萄牙與東亞福利體制的日本的歸屬類別存在不穩定。它們代表著社會投資範式下不同的政策支出和產出模式，通過將上述聚類結果與此前福利國家的體制分類結合一起時，除日本、葡萄牙、荷蘭的歸屬有爭議外，福利國家的社會投資轉向與此前的體制分類大致一致。這一結果擴展了此前僅以個別歐洲或亞洲為中心的線性邏輯分類，並且證實了東亞福利國家在社會投資轉向上的分化。 為了深度了解社會投資轉向的差異，通過借鑒歷史制度主義關鍵概念和框架，作者以日本和韓國為典型的案例進行了比较分析，結果顯示：儘管日本和韓國有類似的制度遺續，特定的國際偶然事件開啟了兩國社會投資轉向的歷史窗口，在歷史窗口開啟後，儘管兩個國家都有著類似的政策學習，然而，當社會投資作為一種回應策略被採納後，不同國家的政治與經濟結構關係將對政治行動者在社會投資改革的深度和能力形成限制抑或機會，特別是兩國不同的政黨差異與競爭，以及民眾的態度均導致了兩國轉向的差異。

In the 1990s, the welfare state was again at a development crossroads. To better adapt the post-war welfare state to the risk and demand of the contemporary knowledge-based economy, social investment became a comprehensive paradigm and guide for welfare state reforms, triggering a silent ‘paradigm revolution’. Current research on the social investment transition of welfare states is geographically divided into two centers: either they focus on European welfare states, arguing that there are roughly ‘four worlds’ of the social investment welfare states, or they focus on East Asian welfare states, arguing that the East Asian welfare states tend to be differentiated in the social investment transition. Moreover, the literature, as for conceptual operationalization, focuses only on a single or a combination of several investment-type policy instruments. Methodologically, the measurement of the social investment paradigm is flawed, failing to bridge the quantitative-qualitative divide and integrating policy expenditures and output variables, even in the case of some quantitative studies. In terms of data analysis, it is not possible to establish a solid causal explanation mechanism using cross-sectional data over a relatively short period of time. In this regard, the authors try to ask: How can the measurement of the social investment paradigm be improved? Is there a possibility of integrating the literature between Europe and East Asia? What are the similarities and differences between East Asian and European welfare states in their transition towards social investment? How to explain such similarities and differences? And is there a potential for theoretical innovation? Considering the available existing data, the author integrates the OECD Social Expenditure Data (SOCX), The Global Welfare Dataset (GLOW), and the Comparative Welfare States Data Set (CWS), and finally forged a panel data covering the period 1990-2019 for 19 sample welfare states. Based on the aggregate and categorical descriptions of social expenditures of the sample welfare states, three clustering analyses using standardized policy expenditure variables, policy output variables, and joint policy expenditure-output variables are conducted. The results show that, except for Japan and Portugal, whose attribution to the clustered clusters is slightly unstable, the other 17 selected sample welfare states are stably clustered into the ‘five worlds’. They represent different patterns of social spending and output under the social investment paradigm. When these clustering results are compared with the previous regimes’ classification of welfare states, the two results are broadly consistent with each other. This result extends the last linear logic categorization centered only on Europe or East Asia. It confirms the divergence in the social investment turn of East Asian welfare states. To understand the differences in the shift to social investment, the authors conduct a comparative analysis of the cases of Japan and Korea by drawing on key concepts and frameworks of historical institutionalism. The results show that specific international contingencies opened a historical window on the shift to social investment in the two countries and that, even though similar policy learning took place in the two countries, the different countries’ politics were not always in line with each other’s views on the shift in social investment once it had been adopted as a strategy of response. However, when social investment is adopted as a response strategy, the structural political and economic relations of the two countries will either limit the depth and capacity of political actors to reform social investment or provide opportunities for them to do so. In particular, the different competition between political parties in the two countries and the attitudes of the public will contribute to the differences in the two countries’ transition.