Purpose: The article revisits the negative effect of welfare state schemes on crime rates with a focus on the pathways of that effect. Social support theory (SST) and institutional anomie theory (IAT) are two related theoretical approaches to explain why – and how – the welfare state might prevent criminal behavior. This article tests the relevance of these theories for cross-country and over-time variation in homicide rates with more precise disaggregated welfare state indicators than used previously. Methods: We use panel regressions with country fixed effects and data for 18 OECD countries between 1990 and 2011. Disaggregated cross-national social expenditure and benefit generosity data is used to discriminate between pathways of welfare state influence on cross-national homicide. Results: The welfare state suppresses crime particularly through social support via generous unemployment benefits. Overall decommodification, the key measure to test IAT, however, does not have any effect on homicide. Conclusions: Only some welfare state interventions matter for homicide rates.

Introduction

Thinking about the connection between social programs and crime is as old as the modern welfare state itself. Famously, German criminologist Franz von Liszt stated already in 1898 – i.e. about a decade after Chancellor Bismarck had introduced the first national social insurance schemes in the world – that social policy “represents the best and most effective crime policy” (von Liszt, 1899: 22). In 1939 President Franklin D. Roosevelt claimed that “[t]hrough a broad program of social welfare, we struck at the very roots of crime itself” (cited in Fishback, Johnson, & Kantor, 2010: 715). In more recent decades, the public discourse and public policy in the United States (US) rather reflected the opinion that generous welfare policies do not reduce, but may even foster criminal activity and that stricter criminal prosecution would be the best way to reduce crime (Brown, 2016; Cullen, Wright, & Chamlin, 1999). Punitive responses to crime became widespread (Garland, 2001), together with harsher views on welfare (Wacquant, 2009). This stands in contrast to the criminological literature which, since the 1980s, has taken the idea of a negative association between welfare and crime more seriously and empirically tested it, first for the United States and later also crossnationally. These studies overwhelmingly show that more generous welfare policies are associated with lower levels of crime, hence unambiguously confirming Franz von Liszt’s early intuition. Yet, the literature is also marked by two interrelated shortcomings: First, while there exists a variety of theoretical approaches to explain the relationship, their empirical implications at the hypothesis level are often hard to distinguish. Second, at the operational level, we find that scholars often employ the same, highly aggregate measures of welfare state size such as total social spending to test different theories, although a range of much more specific indicators are available. Hence, while evidence of a negative effect is by now sufficiently well-established, we still do not know how the welfare state may reduce crime across countries. The view of the welfare state in this literature is quite simplistic and does not consider that different social policy schemes work in very different ways. We therefore need to disaggregate the welfare state to better understand which elements of social protection are responsible for the effect (see Worrall, 2005: 368).