The impact of the 2008 economic crisis on imprisonment in Europe

Abstract

We explore how the worldwide economic downturn of the early 2000s affected imprisonment rates across Europe. We test three hypotheses: (i) the recession caused an increase in incarceration rates directly, regulating the excess in labour supply; (ii) it did it indirectly, by affecting crime; (iii) its effects varied according to the institutional context – countries’ welfare states and criminal justice traditions. We use cross-national panel data to fit fixed, random and mixed-effects models and to explain variations in incarceration rates within and across countries during 12 years. The results show that the economic crisis had multiple effects on imprisonment and that these were moderated by the institutional context, increasing it in countries with less comprehensive welfare states and more punitive penal traditions and decreasing it in countries with penal-welfarist policies.

Keywords

Economic crisis imprisonment penal policies social policies

Introduction

The impact of socioeconomic conditions and, specifically, of economic crises on imprisonment rates has been the subject of much research – see the next section for a review. Until recently, the consensus was that incarcerations rose with deteriorating economic conditions owing to higher unemployment and poverty (Chiricos and DeLone, 1992; Pratt and Cullen, 2005). However, some studies now suggest that, during the recent worldwide recession, incarceration rates dropped (D’Amico and Williamson, 2015; Uggen, 2012), or that there were important cross-national variations (Dünkel, 2017; Wenzelburger, 2018).

Among the researchers finding a relationship between economic crises and incarceration, many follow Rusche and Kirchheimer’s thesis (1939) that during recessions repression and harsher penalties replace routine channels for maintaining social order (Jacobs and Carmichael, 2001). Others argue that the relationship is indirect, via changes in crime (Rosenfeld, 2014; Rosenfeld and Messner, 2013). And still others make the relationship conditional on the institutional context in which it operates (Lacey 2010; Lappi-Seppälä, 2011; Sutton, 2004; Wenzelburger, 2018).

In this article we investigate the impact of the 2008 economic crisis on imprisonment rates across Europe, using cross-national panel data over a 12-year period. We test whether the recession produced direct or indirect changes in incarceration rates and if they depended on the institutional context.

The article is structured as follows. In the next section, we review the theoretical approaches to studying economic conditions and incarceration, distinguishing studies that explore a direct or indirect link between the two from those that consider the impact of the institutional context. This discussion helps us develop the research objectives and hypotheses. Next, we present the data, variables and methodology. The presentation of the results follows. The final section discusses the research implications.

Economic crisis and imprisonment

As noted, the scientific literature on the impact of economic downturns on imprisonment posits either an indirect effect via variations in crime rates, a direct effect via harsher punishments, or differential effects depending on the institutional context. We next briefly review the three approaches.

Indirect effects

In general, criminologists have posited a negative (countercyclical) association between economic conditions and crime (Bushway et al., 2012; Pratt and Cullen, 2005; Pratt and Lowenkamp, 2002) and a positive one between crime and incarcerations, at least for some crimes (Aebi, Linde and Delgrande, 2015; Uggen, 2012). Hence, they expect declines and increases in crimes and incarcerations during, respectively, expansionary and recessionary periods.

The main explanation for the criminogenic relationship between economic downturns and crime is economic, linking the rise in (property) crimes and incarcerations to the increasing costs of engaging in legitimate activities (Vold et al., 1998). Another could be provided by Agnew’s (1992) strain theory: economic crises cause increases in inequality, poverty and unemployment, which produce a state of tension and more (violent) crime.

The alternative hypothesis expects a positive relationship. Criminality ‘increases and decreases in the same manner as other economic activity’ (Vold et al., 1998: 115), and the returns from crime are curtailed during recessions (Rosenfeld 2009). Furthermore, contractions reduce the opportunities to commit crimes – in line with Cohen and Felson’s (1979) routine activities theory.

The research supporting each hypothesis is inconclusive (Michalowski and Carlson, 1999; Vold et al., 1998; Zimring, 2007), partly because it stems from different methodological designs – cross-sectional or longitudinal – and measurements (see Chiricos, 1987, and Pratt and Lowenkamp, 2002, for discussions). As noted by Rosenfeld (2009), the debate somewhat faded after Cantor and Land (1985) introduced longitudinal data and new indicators to measure economic cycles (for example, changes in salaries and GDP per capita). However, it did not vanish, for several reasons.

First, associations between changing economic conditions and incarceration rates continued to be observed for some crimes. Most studies found thefts to be countercyclical, increasing with recessions and decreasing with expansions, and violent crimes to be pro-cyclical (Arvanites and Defina, 2006; Bushway et al., 2012; Cantor and Land, 1985; Jacobs and Richardson, 2008). However, some authors found violent crimes to be countercyclical (Fajnzylber et al., 2002; Rosenfeld, 2009).

Second, there remain significant methodological differences among the studies using longitudinal data. Some apply times-series analyses for single geographical units (Cantor and Land, 1985; Pratt and Lowenkamp, 2002; Rosenfeld, 2009); others, semi-experimental designs contrasting periods of varying duration (Bushway et al., 2012); and still others use fixed- or random-effects models comparing different geographical units (Fajnzylber et al., 2002; Jacobs and Carmichael, 2001). Some authors log the variables (Arvanites and Defina, 2006; Jacobs and Richardson, 2008; Rosenfeld, 2009); others, transform them into relative measures based on first or second differences (for example, Cantor and Land, 1985; Rosenfeld, 2014); and still others use moving averages or lagged variables with different time specifications (for example, Jacobs and Richardson, 2008; Pratt and Lowenkamp, 2002).

The debate on the relationship between crime and incarceration rates is also controversial (Rosenfeld, 2014). In the 1990s, the consensus was that the rise in US incarceration rates was due to rises in drug crimes (see Blumstein and Beck, 1999, but also Arvanites and Defina, 2006, and Beckett and Western, 2001, for opposite conclusions). However, in the 2000s this view was abandoned as incarceration rates continued to rise despite a sustained drop in crime (Raphael and Stoll, 2007). In Europe, the increases in prison rates during the 1990s and 2000s were often explained as being driven by more violent crimes (Aebi, Linde and Delgrande, 2015; D’Amico and Williamson, 2015), although not in all countries (Dünkel, 2017; Lappi-Seppälä, 2011). An overall conclusion could be that there is a positive, if weak, relationship between (violent) crime and incarceration. Greenberg (2001) argued that dismissing this association would be against common sense and might simply reflect researchers’ methodological choices.

The difficulty in estimating the relationship between crime and punishment stems from its complexity. Variations in imprisonment rates respond both to the ‘breadth’ and to the ‘intensity’ of prison use (Young and Brown, 1993: 14). The former reflects how many people are admitted to prison; the latter, the mean length of stay (Lappi-Seppälä, 2011). Each reacts differently to alternative crimes, criminal policies (Tonry, 2007; Von Hofer, 2003) and governments’ financial capacities (Lappi-Seppälä, 2008). Finally, estimates of the effect of crime on imprisonment must consider the opposite effect of imprisonment on crime (Zimring, 2007), to avoid its underestimation (Liedka et al., 2006).

Direct effects

In contrast to the previous line of enquiry, Rusche and Kirchheimer’s (1939) thesis is that imprisonment responds directly to an economic rationality. The capitalist state uses imprisonment during recessions to discipline and regulate workforce excess with harsher punishments that keep criminals off the streets for longer. Penal institutions play an important role in helping capitalism meet its accumulation needs (Lessan, 1991; Michalowski and Carlson, 1999), although perhaps not as a conscious strategy. Thus, for Melossi (1987) the rising imprisonment rates associated with recessions respond to changes in ‘punitive motive’ in the moral climate that surrounds the penal system, which fluctuates with business cycles.

The generally positive association between unemployment and imprisonment has been used to support this argument ; see De Giorgi (2013) and Sutton (2012) for recent reviews. The unemployment rises accompanying downturns destabilize the social order, legitimizing the state’s use of harsher penalties – for example, fewer early releases, newly mandatory prison terms for some crimes, longer sentences for others – to combat crime (Jacobs and Carmichael, 2001).

The argument also applies to the relationship between inequalities and imprisonment. Apart from Arvanites and Asher’s (1998) early research comparing rates of incarceration across US states, which showed inequalities having no impact on imprisonment, most research has found that higher inequalities are associated with greater use of prison (Beckett and Western, 2001; Cavadino and Dignan, 2006; Phelps and Pager, 2016; Sutton, 2012; Western and Pettit, 2010).

Price changes are also used to proxy business cycles. Some researchers expect inflation, which often accompanies expansionary times, to make people poorer, by devaluing wages and increasing the relative incentives for acquisitive crimes (Chiricos, 1987; Lessan, 1991). Others expect the opposite, because inflation redistributes wealth from creditors to debtors (Sutton, 2012). The evidence is inconclusive. Lessan (1991) and Rosenfeld (2014) found inflation to have criminogenic effects on incarceration; Sutton (2004, 2012), the opposite.

Besides sharing similar methodological problems as those mentioned above for other lines of enquiry, research on the relationship between economic conditions and incarceration suffers from the endogeneity of unemployment in the imprisonment function, especially in countries with high incarceration rates. According to Western and Beckett (1999), imprisonment increases unemployment in the long term, owing to prisoners’ decreased employment expectations. In the short term, it reduces it by removing the incarcerated from unemployment official statistics. Greenberg (2001), Jacobs and Kleban (2003) and Kling (2006) argue that the extent to which incarceration reduces unemployment has been exaggerated, especially in countries where incarceration rates are low.

Institutional effects

The complex relationship between economic conditions, crime and incarceration led a group of comparativist criminologists to argue that countries react differently to similar economic stimuli, depending on their institutions (Downes and Hansen, 2006; Lacey, 2008; Rosenfeld and Messner, 2013; Sozzo, 2017; Sutton, 2004; Tonry, 2007). There are two approaches within this view, emphasizing cross-national differences either in penal traditions or in their welfare states. Both are clearly interrelated (Lacey 2008).

Criminal justice tradition

There are different criminal justice traditions, with different levels of punitiveness, that moderate the association between economic conditions and imprisonment. These differences largely depend on whether a country’s legal culture is rooted in the common law or the civil law traditions (D’Amico and Williamson, 2015). Some criminologists further divide civil law into Romano-Germanic and Nordic law (DeMichele, 2010). These traditions affect how permeable the system is to outside pressures – more in common law countries with majoritarian electoral systems that revolve around ‘median’ voters who elect judges and prosecutors directly (Lacey, 2008; Savelsberg, 1994; Tonry, 2007). There are also differences in the notion of justice that predominates in each – oriented to victims’ reparation, crime control or criminals’ rehabilitation (Tonry, 2007).

The empirical evidence shows that common law countries incarcerate more people for longer times. Countries in the Romano-Germanic tradition have moderate levels of imprisonment. Nordic law countries have the lowest levels of imprisonment (D’Amico and Williamson, 2015; Lacey, 2010).

Welfare state

Garland (1985) argued that welfare and criminal policies are interrelated because they share the same goal – managing social marginality. Messner and Rosenfeld (1997) also showed that, in liberal countries with less regulated markets, social relations are commodified and collective values undermined, promoting crime.

For some scholars the key factor mediating the relationship between welfarist policies, crime and punishment is the level of social public spending (Downes and Hansen, 2006; Greenberg, 2001). This has been shown to reduce thefts and homicides (Savage et al., 2008) and incarcerations (Sutton, 2012). Beckett and Western (2001) showed that it could be an effective tool and an alternative to imprisonment to manage social problems.

Whereas some authors (Lacey, 2008) assess the effect of welfarist policies on incarceration by comparing corporatist and bureaucratic societies with liberal and market-oriented ones, most rely on the larger typology of welfare states developed by Esping-Andersen (1999) (for example, Cavadino and Dignan, 2006; Lacey, 2010; Lappi-Seppälä, 2008). Countries with universal and generous welfare provisions – such as traditionally social-democratic, North European countries – have lower rates of incarceration. In typically Anglo-Saxon countries with liberal traditions and minimalist (means-tested) welfare states, incarceration rates are higher. Central and West (Continental) European welfare states share the bureaucratic logic of the Nordic countries, but are more conservative and distribute benefits unequally, based on contributions (Esping-Andersen, 1999). Incarceration rates are here slightly higher than in Northern Europe. In South European (Mediterranean) countries, where family support plays a key welfare role, imprisonment rates approach the Anglo-Saxon levels. Finally, Baltic and Eastern countries show the highest, if rapidly declining, rates of imprisonment (Dünkel, 2017) owing to their transitional regimes, poor networks of social support and more ‘authoritarian’ democracies (Fenger 2007; Lappi-Seppälä, 2008, 2011; Stamatel, 2009).

Countries’ welfares states and justice traditions are strongly correlated because both dimensions vary according to common political institutions (Lacey, 2008; Melossi, 1987). Common law countries tend to have liberal, minimalist welfare states. Countries with Nordic law traditions have generous and universal welfare states. Countries with Roman-Germanic legal traditions have heterogeneous welfare states, ranging from contributory, to familistic and to hybrid systems (Downes and Hansen, 2006).

Research question and hypotheses

Our main research question is whether or not changing economic conditions alter prison population rates. Concretely, we ask if the 2008 economic crisis increased incarceration rates across the EU. The severity of this crisis should facilitate testing the following three hypotheses derived from the previous discussion, by making the direct, indirect and institutional effects of economic downturns sharper.

(H1) Indirect effects. The economic crisis increased imprisonment rates owing to rises in crime associated with deteriorating living conditions.

(H2) Direct effects. The economic crisis led to an increase in incarceration rates owing to the state’s higher punitiveness and harsher sentences.

(H3) Institutional effects. The association between economic crisis, crime and imprisonment is moderated by a country’s welfare and criminal justice system. The recession’s direct and indirect effects should affect incarceration rates more markedly in Anglo-Saxon, South European and East European countries, which have less comprehensive or efficient welfare states and traditionally more punitive penal systems, than in Continental and North European countries, which have more generous welfare states and bureaucratic penal systems.

Methods

The data consist of annual aggregate statistics for the EU’s 28 member states for the period 2002–15. This 14-year period comprises phases of economic growth (before and after the recession) and contraction. The statistical sources are varied and specified for each variable in Table 4 in the Appendix. We have 28 countries × 14 years = 392 cases. In the analyses, we treat countries as analytical cases rather than as demographic units, and hence each country carries the same weight.

We perform three sets of analyses. First, we apply fixed-effects models for panel data in order to explain countries’ incarceration rates. We enter sequentially two groups of time-changing variables measuring, respectively, the economic and the crime conditions in each country and year, with the aim of isolating the recession’s direct and indirect effects on incarceration. Estimates for each independent variable in these fixed-effects models are between-country averages of the effects estimated within countries using time variations. They are overall estimates of the recession’s direct and indirect impact on incarceration, net of countries’ idiosyncrasies.

In the second set of analyses, we run a simplified version of the most complex model estimated before, and compare it with a similar model estimated with a random intercept specification using a Hausman test. This is a first evaluation of the importance of institutional, cross-regional differences in the recession’s impact on incarceration. Next, we run another random intercept model where we add a time-invariant variable measuring the region that each county belongs to, based on the five-class typology of welfare and penal states presented above. In this model, estimates are regional averages of the between-country averages calculated in each country from time variations in the dependent and independent variables.

Finally, we estimate a set of mixed-effects models with interaction effects between region and the independent variables to investigate whether regional differences in contraction’s effects on incarceration differ according to regional economic and crime characteristics.

In all models, standard errors are clustered within countries. This is recommended when inertia effects are likely (Cameron and Trivedi, 2009), as when working with crime trends (Fajnzylber et al., 2002).

In measuring countries’ yearly economic and crime conditions we use absolute magnitudes (for example, rate of thefts × 100,000). Only for GDP do we further consider the relative percentage changes from the previous year. The values for this variable are inverted: positive values express contraction’s intensity; negative ones, expansion’s.

To deal with endogeneity – the possibility that incarceration rates may affect socioeconomic conditions and crime – we use moving averages of independent variables’ values in the current and previous two years, instead of their current values (Jacobs and Richardson, 2008). Suppose that incarceration affected the independent variables (for example, crime rates) immediately, by incapacitating prisoners, whereas the independent variables’ effects were lagged owing to other intervening factors (Young and Brown, 1993). Then, independent variables’ lag effects on incarceration could be estimated net of their concurrent, endogenous associations, by entering both the concurrent and the lag values of the same variables. Similar and consistent results are obtained with moving averages, because endogeneity is diluted when the independent variables are averaged over several years.¹ Moving averages also smooth independent variables’ effects on incarceration rates, making them less responsive to trendless fluctuations. The solution de facto restricts the range of the period analysed, because the first two years (2002 and 2003) provide the input for calculating only the first moving average.

The dependent variable is the Prison population rate per 100,000 inhabitants, a stock statistic estimating the yearly number of inmates, which is a function of the number of prison entrances and sentence length (Aebi and Kuhn, 2000). Cross-national comparisons of incarceration rates are problematic because, for example, juveniles or on-remand prisoners are omitted in some countries, but it is the most reliable index of punishment in cross-national research (Blumstein and Cohen, 1973; Lappi-Seppälä, 2008).

Our main independent variable is Yearly percent contraction in GDP, which measures the downturn’s severity. This relative measure captures better than absolute levels of GDP per capita the recession’s non-linear effects on incarceration – but see Rosenfeld (2009) and Sutton (2004) for examples that use absolute measures. Drops in wealth can have similar or larger effects on incarceration depending on whether they exceed a minimum threshold below which individuals’ ability to make ends meet or governments’ capacity to incarcerate criminals is severely undermined. Thus, Lappi-Seppälä (2008) found that the impact on imprisonment of wealth variations is stronger in poorer than in richer countries, and Jacobs and Kleban (2003) found that richer countries have higher rates of incarceration. The problem is mitigated using relative measures of wealth, since the same absolute drop yields higher percentage contractions in poorer contexts, but this cannot address the threshold effects. Hence, we use both measures. Finally, the variable Contraction’s length (number of consecutive years without economic growth immediately preceding the current year) aims at assessing whether or not its duration has additional effects on incarceration.

GDP contraction is just one of the many indicators of deteriorating economic conditions used in comparative research. Accordingly, we add three more indicators: Level of unemployment (percentage of non-working individuals actively seeking a job at year’s end); Deflation (negative changes in consumer prices over one year); and the Gini index or degree of income inequality (proportion of individuals who should change their income for all to have the same amount). In all cases, the values are averaged for the current and previous two years.

We use three variables to measure crime incidence in each country and three-year period – the indirect mechanism though which, according to hypothesis H1, economic conditions affect incarceration rates. The Rate of intentional homicides (per 100,000) is among the most reliable international indicators of crime, because homicide reporting is less affected by criminal justice traditions (LaFree, 2005). Many authors – for example, Fajnzylber et al. (2002), Rosenfeld (2009) or Sutton (2004) – use them as proxies for the rates of violent and general crimes. The Theft rate (per 100,000) excludes property crimes producing damages or injuries. Finally, the Total crime rate (per 100,000) is the least amenable to cross-national comparisons because the same behaviour may be categorized as a crime or a misdemeanour in different countries (for example, soft drugs use) or as not punishable (for example, abortion). We treat it as a residual measure of unspecified crimes not captured by the other rates.

Finally, with the variable Region we expect to capture the main institutional effects on incarceration, especially in interaction with other variables. We group countries into five regions based on Esping-Andersen’s original three-worlds welfare state typology (1990), later revised by himself and others and resulting in the addition of the Mediterranean and Eastern variants (Esping-Andersen, 1999; Lappi-Seppälä 2011). The regions are also meant to capture countries’ punishment and legal traditions (DeMichele, 2010; Lacey, 2010). The Anglo-Saxon region comprises Ireland and the UK. Denmark, Finland, Norway and Sweden make up the Northern region. Austria, Belgium, France, Germany, Luxemburg and the Netherlands comprise the Continental group. Southern Europe comprises Malta, Cyprus, Greece, Italy, Portugal and Spain. Finally, Bulgaria, the Czech Republic, Croatia, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovenia and Slovakia form the East European region.

Results

We start the presentation of the results with descriptive analyses showing, graphically, regional trends between 2002 and 2015 in our main variables. Figure 1 shows trends in imprisonment and crime.

The upper-left graph shows important cross-regional variations in incarceration rates over time. In Eastern and Northern Europe the rates were, respectively, twice as large, and between half and two-thirds as small, as elsewhere. Nordic countries’ low rates became more pronounced after 2005, irrespective of the recession. Eastern Europe’s higher incarceration rates declined before and after the recession, remaining higher than in the rest of Europe at all times. In Continental, Anglo-Saxon and Southern Europe, imprisonment rates rose similarly until the start of the recession, but diverged afterwards. In Continental Europe, they remained stable during the recession; in the Mediterranean and Anglo-Saxon countries, they increased. After the recession ended in 2012 all regions sustained a decline in incarceration rates. In short, only in East European, Anglo-Saxon and South European countries was the recession accompanied by increases in incarceration rates.

Figure 1’s upper-right graph suggests that Eastern countries’ high rates of incarceration are a function of their high rates of violent crimes (as proxied by the homicide rate), which carry longer prison sentences. The fact that the theft and total crime rates remained low over time reinforces this explanation (see bottom graphs). However, homicides declined markedly over time, approaching the levels of the other European regions at period’s end. This decline is inconsistent with the rise in incarceration rates during the recession and suggests that the real reason for Eastern Europe’s high rates of incarceration is that prison sentences where harsher for all crimes, also during the recession.

Regarding the relationship between crime and incarceration rates, the Nordic countries stood opposite to Eastern Europe, displaying the highest rates of thefts and total crimes and the lowest incarceration rates. This supports the widespread view that these countries have a lenient, anti-prison stance towards punishment (Lappi-Seppälä, 2008; Sutton, 2004).

Anglo-Saxon and Continental countries had similarly low rates of homicides, fluctuating around a slightly declining trend. They also had similarly moderate theft rates that declined consistently over time – slightly more in Anglo-Saxon Europe. However, the total rate of crimes followed different directions in each region. In Continental Europe, it declined slightly over the period. In the Anglo-Saxon countries, it rose markedly before the recession, only to decline less intensely afterwards. Since these countries experienced declining rates in all crimes during the recession, here crime rises cannot explain the increase in incarcerations.

Finally, Southern Europe had rates of thefts and total crimes as low as Eastern Europe’s, and intentional homicide rates almost as high as Northern Europe’s. It is the only region experiencing upward trends in intentional homicides, thefts and incarcerations during the recession, which started to decline in 2012.

Turning now to trends in socioeconomic conditions, the upper-left chart in Figure 2 shows deep wealth inequalities between European regions. Eastern Europe is substantially poorer than Anglo-Saxon, Northern and Continental Europe (in this ascending order), and moderately poorer than Southern Europe. These regional differences, combined with Eastern and Southern Europe’s lower rates of thefts and general crimes, give support to an economic argument that links wealthier nations’ higher crime rates to the higher returns obtained by criminals from illegal activities (Soares, 2004; Zimring, 2007). Alternatively, richer countries could have better means to detect, process and record crimes, or more highly trusted police and penal institutions that encourage victims to report crimes.

The economic argument applies less clearly to wealth trends. The first three charts in Figure 2 show that only Southern Europe experienced a sustained period of wealth loss after 2007, which led to steeper and longer-lasting rises in unemployment. Elsewhere, the recession was shorter and less onerous for employment. Eastern Europe recovered its pre-recession levels earlier than Continental and Northern Europe. In the Anglo-Saxon countries, the recovery was also slower but it accelerated after 2012, reaching the richest regions’ levels by 2015.² In these richer regions, inequalities rose with the recession and stayed constant afterwards. In the others, they dropped with the recession, possibly indicating an equalization towards the bottom, owing to the demise of the lower middle classes (Whelan et al., 2016), and increased afterwards. Figure 2’s last graph shows drastic and ubiquitous price contractions accompanying the recession, in contrast to the 1970s recession, when prices in production inputs rose markedly (Perri and Quadrini, 2018). The EU’s introduction of common and stricter monetary policies in 2010 explains the subsequent convergence in price changes towards zero. In short, Figure 2 documents the recession’s extraordinarily wide and deep impact in Europe, especially in the South.

Although illuminating, the descriptive evidence allows for the testing of neither the validity of the economic argument nor, more generally, of our hypotheses. The hypotheses on the recession’s direct and indirect effects on incarceration rates can be tested more effectively with fixed-effect models that use the pooled sample of countries and consider only time variations within them. Their results are reported in Table 1.

Table 1.

Explanatory models of incarceration rates.

	M1	M2	M3	M4	M5	M6	M7	M8
Mean rate of GDP contraction in last three years
Rate of GDP contraction in current year	−60.06	−53.75	−44.84	−53.89	−58.11	−25.30	−31.33	−36.14
	(44.45)	(35.54)	(43.85)	(45.39)	(44.23)	(35.11)	(39.92)	(43.57)
Length of recession (No. of years)	1.341**	1.286**	1.094	0.96	1.076⁺	0.822⁺	0.943⁺	0.998**
	(0.526)	(0.462)	(0.672)	(0.647)	(0.588)	(0.423)	(0.46)	(0.477)
Mean GDP per capita in last three years (in ’000 euros, differences from overall mean)		−0.631	−0.282	0.444	0.433	1.937	2.875	2.933
		(1.77)	(2.408)	(2.457)	(2.434)	(1.799)	(1.927)	(1.844)
Mean rate of unemployment in last three years (differences from overall mean)			0.428	1.023	0.974	1.455⁺	1.398⁺	1.234
			(0.997)	(1.089)	(1.037)	(0.827)	(0.782)	(0.771)
Mean rate of deflation in last three years (differences from overall mean)				−2.125**	−1.821+	−1.064	−1.176	−1.09
				(0.894)	(0.947)	(1.397)	(1.331)	(1.308)
Mean Gini in last three years (differences from overall mean)					−1.558	−2.020**	−2.369**	−2.382**
					(1.494)	(0.907)	(0.905)	(0.898)
Mean violent crime rate per 100,000 in last three years (differences from overall mean)						6.176	5.477	5.464
						(8.037)	(7.6)	(7.556)
Mean property rate per ’000 in last three years (differences from overall mean)							13.4	16.34
							(9.803)	(9.8)
Mean total crime rate per ’000 in last three years (differences from overall mean)								−2.975⁺
								(1.741)
Constant	135.4***	135.6***	135.2***	134.8***	135.0***	135.1***	135.4***	135.5***
	(0.711)	(0.937)	(1.512)	(1.493)	(1.356)	(1.536)	(1.522)	(1.486)
Observations	335	335	335	335	335	335	335	335

Notes: Standard errors in parentheses; standard errors are clustered within countries. Results from fixed-effects models with time nested within countries.

p < .10, **p < .05, ***p < .001.

The models in Table 1 are increasingly complex. We start with one explaining incarceration rates with basic indexes of economic conditions. We continue by specifying these conditions with indicators of unemployment, deflation and inequality. The final models add potential crime mechanisms.

The starting model (M1) regresses countries’ yearly rates of incarceration on the contraction’s severity and length. Only the estimate for the recession’s length is significant and positive. Longer recessions are associated with higher incarceration rates. This effect remains significant even after controlling for crime mechanisms in models M6 to M8, confirming that the recession’s length had a direct, criminogenic impact on imprisonment rates.

Models M3 to M5 specify the associations between economic conditions and incarceration, by considering the role of unemployment, deflation and inequality. Model M3 shows that higher unemployment in the current and last two years was associated with higher incarceration rates in the present year. The effect is not significant but becomes so (at the .1 level), and its magnitude larger, after controlling for homicides in M6. This shows that unemployment had a direct effect on incarceration, partly obscured by an indirect effect via reductions in violent crimes leading to imprisonment. Unemployment’s other indirect effects increased incarcerations, First, unemployment multiplied the number of thefts punishable with prison, those typically committed by recidivists (Durose et al., 2014) – see the change in unemployment’s coefficient after controlling for thefts in M7. Second, unemployment decreased the proportion of crimes least punishable with prison, an indirect consequence of the increase in thefts punished with prison – see the change in unemployment’s coefficient in M8 after controlling for total crime rates. Considering unemployment in M3 makes the effect observed in M2 for the recession’s length smaller, because lengthier recessions produced higher unemployment. In sum, long recessions fuelled unemployment, sparking direct and indirect effects on incarceration. These effects were inconsistent. Unemployment’s direct effects increased imprisonment; its indirect effects increased or decreased it depending on the crime mechanism through which they operated – property or violent crimes, respectively.

Deflation’s effect on imprisonment in M4 is negative and significant. Higher deflation in the present and last two years is associated with lower incarcerations in the current year. Deflation is negatively associated with the contraction’s severity and length – see changes in these variables’ coefficients after adding deflation in M4. This seems counterintuitive because deflation was highest at the recession’s peak in 2009 (see Figure 1). However, it also increased after 2012, when the economy recovered and imprisonment rates declined. The negative association between deflation and incarceration is likely to apply to this post-recessionary period and the pre-recession period of high inflation and rising incarceration rates. Deflation also affected imprisonment negatively via reductions in property and violent crimes – see M6 and M7. This suggests that deflation had a depressing effect on illegal markets, lowering the incentives for committing crimes (Rosenfeld, 2014).

The coefficient for deflation loses strength and significance after controlling for the Gini coefficient in M5, signalling that part of deflation’s lessening effect on imprisonment occurred through concurrent rises in inequality – confirming inflation’s equalizing impact (Sutton, 2012) – and inequalities’ direct, negative and significant effects on imprisonment. However, inequalities also affected imprisonment indirectly and in the opposite direction, fuelling increases in property and violent crimes. The effect is associated with the recession’s length and unemployment. Hence, inequalities had two opposite effects on imprisonment, associated with either expansionary or recessionary times: a direct, negative effect; and an indirect, criminogenic effect via higher property and violent crimes.

The relationships between the criminological variables are as expected, with the rates for all crimes being positively inter-correlated – see changes in their coefficients from M6 to M8. However, their net impact on imprisonment differs. Higher property and violent crimes (non-significantly) led to higher incarceration rates; higher total crime rates led to fewer imprisonments (the effect is significant only at the .1 level).

Overall, it appears that the recession had both direct and indirect effects on incarceration rates, confirming hypotheses H1 and H2. Both were more clearly associated with the recession’s length than with its severity. The direct effects were linked to rising unemployment. The indirect effects were also linked to rising unemployment but were inconsistent regarding imprisonment rates. One path led to fewer incarcerations via fewer violent crime; the other, to more incarcerations via higher inequalities and rises in thefts and other crimes.

We now turn to assessing the presence of institutional effects in Table 2.

Table 2.

Stylized models: Results from fixed- and random-effects models with time nested within countries.

	M9a^a,c	M9b^b,c	M9c^a,d	M10^a,d
Mean rate of GDP contraction in last three years	49.8	49.8	41.13	42.94
	(59.59)	(39.50)	(41.31)	(61.37)
Mean GDP per capita in last three years (in ’000 euros, differences from overall mean)	0.847	0.847	−1.389**	0.523
	(2.096)	(0.989)	(0.482)	(1.016)
Mean rate of unemployment in last three years (differences from overall mean)	0.631	0.631	0.362	0.654
	(0.807)	(0.408)	(0.401)	(0.687)
Mean Gini in last three years (differences from overall mean)	−2.115	−2.115**	−0.798	−1.198
	(1.517)	(0.784)	(0.770)	(1.513)
Mean total crime rate per 100,000 in last three years (differences from overall mean)	−2.885	−2.885⁺	−4.081**	−2.826
	(2.124)	(1.726)	(1.682)	(2.003)
Mean property rate per ’000 in last three years (differences from overall mean)	17.34⁺	17.34***	9.927**	14.97⁺
	(10.06)	(4.503)	(4.111)	(8.980)
Anglo-Saxon^e				−0.679
				(30.57)
Central/Western^e				−31.08
				(30.33)
Eastern/Baltic^e				77.00**
				(28.03)
Nordic^e				−73.64**
				(35.20)
Constant	134.9***	134.9***	135.0***	119.0***
	(0.991)	(1.229)	(7.89)	(13.18)
Observations	335	335	335	335

Notes: Standard errors in parentheses.

Standard errors are clustered within countries.

Standard errors are not clustered within countries.

Model specifications are fixed-effects.

Model specifications are random-intercepts.

Coefficients for regions are relative to Southern Europe.

p < .10, **p < .05, ***p < .001.

Models M9a to M9c present a ‘stylized’ version of M8 in Table 1 using various methods for estimating standard errors (robust and clustered within countries, in M9a; regular in M9b and M9c) and coefficients (M9a and M9b use a fixed-effects specification; M9c, a random-effects one). For simplicity, all models include only one variable to measure the recession’s severity and duration – a moving average of the relative contraction in GDP over the current and last two years. The models exclude all non-significant effects in M8, unless they become significant after excluding other effects (GDP per capita and unemployment rate) or after applying other model specifications (homicide and theft rates).

M9a in Table 2 yields similar results to Table 1’s more complex M8, using the same fixed-effects statistical specification and clustered standard errors. However, it is simpler and helps detect institutional effects. A first evaluation of these effects can be achieved by comparing M9a with another that uses a random-intercepts specification, with a Hausman test. The test cannot be performed with robust standard errors. Thus, M9a was re-estimated in M9b with regular standard errors, and compared with another with the same standard errors but a random-effects specification (M9c). They differ significantly (χ²_{(6 df)} = 40.42, p = .000) – the impact of the economic and criminological variables on incarceration rates must be estimated controlling for countries’ idiosyncrasies, which so far we did with the fixed-effects specification.

We could also have controlled for national differences by running a random-effects model with dummy variables for each country. In M10, we approximate this solution by adding dummy variables for each region, using Southern Europe as the baseline. M10 estimates how much the five regions defined by their typical welfare states and penal traditions differ in incarceration rates, net of differences in economic and crime conditions. The coefficients for the effects of these conditions are similar to the ones obtained without controlling for regions in M9a using a fixed-effects specification, indirectly validating our country groupings. (An exception is the effect for the Gini coefficient, which is cut almost by half.) M10 confirms that Eastern Europe had significantly higher rates of incarceration than Southern Europe, Anglo-Saxon, Continental and Northern Europe, and that these institutional differences, displayed in Figure 1 in gross form, were not due to regional differences in socioeconomic and criminal conditions.

M10 does not evaluate whether the recession’s effects varied by region. This is done in M11 of Table 3, which adds an interaction effect between GDP’s contraction and region.

Table 3.

Models with region and interaction effects between region and other variables: Results from random- and mixed-effects models with time nested within countries.

	M11	M12	M13	M14	M15	M16	M17	M18
Mean rate of GDP contraction in last three years	250.6***	246.8***	350.8***	288.6***	284.6***	276.5***	226.2**	278.4***
	(41.51)	(41.26)	(50.66)	(67.88)	(68.29)	(78.94)	(100.3)	(71.51)
Mean GDP per capita in last three years (in ’000 euros, differences from overall mean)		−0.184	4.828	7.336⁺	6.809⁺	6.899⁺	7.508⁺	5.473⁺
		(0.990)	(3.072)	(3.835)	(3.666)	(3.716)	(4.420)	(2.820)
Mean rate of unemployment in last three years (differences from overall mean)				1.105	1.089	1.084	0.938	0.611
				(1.047)	(1.023)	(1.023)	(1.063)	(1.038)
Mean Gini in last three years (differences from overall mean)					−0.898	−0.879	−1.146	−1.804
					(1.402)	(1.384)	(1.295)	(1.185)
Mean total crime rate per 100,000 in last three years (differences from overall mean)						−0.664	−3.793	−6.639**
						(2.744)	(2.583)	(3.039)
Mean property rate per ’000 in last three years (differences from overall mean)							15.57	−33.94***
							(9.727)	(9.753)
Anglo-Saxon	5.251	7.628	3.055	−0.131	5.255	5.701	−28.19	−4.748
	(24.58)	(27.44)	(31.41)	(34.89)	(35.48)	(37.04)	(41.69)	(33.75)
Central/Western	−12.14	−8.388	−26.15	−30.51	−35.87	−32.59	−51.66⁺	−12.27
	(11.14)	(19.63)	(20.06)	(23.87)	(23.45)	(21.22)	(30.42)	(25.21)
Eastern/Baltic	71.65**	69.78**	12.73	22.81	19.05	17.99	19.1	113.7
	(22.96)	(24.41)	(62.45)	(68.93)	(64.34)	(64.27)	(64.60)	(76.18)
Nordic	−46.13***	−42.70**	−25.67	−32.53	−41.74	−35.89	−63.09	3.923
	(8.558)	(20.64)	(31.37)	(36.29)	(35.99)	(38.80)	(39.77)	(44.19)
Anglo-Saxon # Mean rate of GDP contraction in last three years	−135.0**	−136.7**	−270.5***	−222.8***	−239.0***	−214.7⁺	2.598	62.1
	(41.58)	(45.07)	(51.93)	(58.81)	(61.11)	(118.3)	(167.6)	(156.4)
Central/Western # Mean rate of GDP contraction in last three years	−350.7**	−344.5**	−451.2**	−388.7**	−376.8**	−373.7**	−301.8+	−360.1**
	(155.1)	(161.7)	(160.3)	(171.6)	(172.6)	(171.9)	(166.4)	(130.2)
Eastern/Baltic # Mean rate of GDP contraction in last three years	−221.1**	−215.1**	−276.7**	−250.4**	−248.8**	−240.4**	−175.3	−182.0**
	(112.4)	(96.63)	(91.39)	(88.10)	(90.97)	(115.0)	(114.5)	(83.00)
Nordic # Mean rate of GDP contraction in last three years	−376.3***	−370.0***	−447.7***	−394.5***	−373.0***	−369.5**	−273.2⁺	−349.1**
	(65.90)	(69.91)	(82.97)	(85.23)	(104.8)	(115.5)	(150.7)	(172.6)
Anglo-Saxon # Mean GDP per capita in the last three years			−5.972⁺	−8.581**	−8.472**	−8.220**	−5.44	−1.139
			(3.114)	(3.963)	(3.879)	(3.736)	(4.244)	(3.426)
Central/Western # Mean GDP per capita in the last three years			−4.812	−7.223⁺	−6.530⁺	−6.669⁺	−6.739	−4.710⁺
			(3.169)	(3.817)	(3.642)	(3.718)	(4.413)	(2.775)
Eastern/Baltic # Mean GDP per capita in the last three years			−8.312	−9.495**	−9.258**	−9.391**	−9.839**	−6.254⁺
			(5.320)	(4.772)	(4.610)	(4.630)	(4.996)	(3.694)
Nordic # Mean GDP per capita in the last three years			−6.996**	−9.271**	−8.430**	−8.595**	−8.524⁺	−6.998**
			(3.314)	(3.937)	(3.721)	(3.854)	(4.426)	(2.891)
Anglo-Saxon # Mean property rate in the last three years								58.11***
								(12.90)
Central/Western # Mean property rate in the last three years								52.50***
								(13.42)
Eastern/Baltic # Mean property rate in the last three years								107.8***
								(26.71)
Nordic # Mean property rate in the last three years								43.81***
								(12.21)
Constant	112.8***	112.3***	126.5***	131.8***	132.4***	131.3***	136.7***	99.22***
	(8.302)	(8.747)	(18.94)	(23.05)	(21.77)	(20.65)	(23.68)	(14.50)
Observations	335	335	335	335	335	335	335	335

Notes: Standard errors in parentheses are clustered within countries. Coefficients for regions’ main and interaction effect are expressed relative to Southern Europe.

p < .10, **p < .05, ***p < .001.

M11 excludes all other independent variables in M10, which are added (alone or in interaction with region) in subsequent models of Table 3. This strategy aims at explaining ‘away’ the regional differences in the recession’s impact in M11. For simplicity, rather than relying on the coefficients estimated in Table 3, in Figure 3 we show the partial effects of the recession for all five regions before and after controlling for the variables and interactions entered throughout Table 3.

The upper-left chart in Figure 3 restates that the recession’s effects on incarceration were most detrimental in the Southern and Anglo-Saxon regions, increasing them. In Eastern Europe, the recession increased incarceration rates just slightly. In Continental and Northern Europe, imprisonment rates declined with the contraction. (Only the differences between the Nordic and Southern and Anglo-Saxon regions are significant, and only at the extremes of the contraction scale.)

Cross-regional differences in the recession’s effects on incarceration increase substantially after controlling for regional differences in wealth and the interaction between wealth and region, due to changes in Southern and Eastern Europe’s estimates. The recession’s effect becomes stronger in both regions because, as M12 in Table 3 clarifies, the drop in incarceration rates produced by wealth contractions was more marked than in the richer regions (in Northern and Continental Europe, it actually led to absolute rises). One possibility is that the recession affected the system’s capacity to keep current levels of incarceration more negatively in poorer than in richer regions. Were it not because of these differences, incarcerations would have risen even more sharply in Southern and Eastern Europe.

In contrast, regional differences in the recession’s impact on incarceration become smaller after controlling for regional unemployment trends (controlling for differences in unemployment’s effects is unnecessary because they are insignificant). This shows that unemployment increased more markedly in Southern Europe than elsewhere (and, less markedly, also in Anglo-Saxon and Eastern countries), pushing incarceration rates upwards. Controlling for regional differences in inequality trends does not change this picture because they were stable, like their effects.

Finally, controlling for cross-regional differences in property crimes and their effects makes the recession’s impact on incarcerations rise in the Anglo-Saxon region and (less markedly) in Eastern Europe. The reason is that here property crimes declined during the recession and these crimes were more severely punished with prison than in Southern Europe. After discounting this, the more punitive reaction of the Anglo-Saxon and Eastern regions to the recession is more clearly revealed.

Overall the results support hypotheses H1, H2 and H3. They confirm that the recession affected imprisonment rates directly and indirectly, and that European regions with alternative welfare and legal systems displayed different punitive regimes and reacted with different degrees of punitiveness to the recession.

Before discussing these findings in the final section, we tested whether or not they could be a consequence of our grouping heterogeneous countries within five regions. We based this grouping on theoretical grounds and partly validated it in Table 2, when we showed that estimates were similar in the random-effects model that controlled for regional differences in incarceration rates rather than in the fixed-effects model that controlled for country differences. We now assess whether actual country trends in incarceration rates are within the 95% confidence intervals predicted for each region in M18 of Table 3 (the most complex).

Figure 4 shows the predicted and actual incarceration trends in each region. Only outliers are shown (Figures 5 to 9 in the Appendix show trends for all countries), that is, countries where incarceration rates fell outside the predicted 95% CI for three or more years. Some regions are very heterogeneous, both regarding the countries’ average incarceration rates and trends. This is especially true of East European countries, where three sub-groups can be clearly identified based on average levels of incarceration: Baltic countries with very high average incarceration rates; Balkan countries with the lowest averages; and Eastern Continental countries with mid averages. We discuss the implications in the next section.

Discussion

We aimed to investigate whether or not the deep recession that hit Europe after 2007 increased incarceration rates, and the role played by crime. Past research on the relationship between economic conditions, crime and imprisonment had generally omitted this recession and approached the matter from alternative angles. Some scholars had investigated whether downturns could be an excuse to discipline and regulate the workforce with harsher punishments, emphasizing the direct links between recessions and incarceration (for example, Jacobs and Carmichael, 2001; Rusche and Kirchheimer, 1939). Others had analysed how contractions might alter choices between legitimate and illegitimate activities, focusing on the indirect links between recessions and incarceration through crime (for example, Cantor and Land, 1985; Rosenfeld, 2009). And still others had explored whether the relationship between economic downturns, crime and incarcerations depended on the institutional context and countries’ penal and welfare traditions (for example, Lacey, 2008; Sutton, 2004; Tonry, 2007).

We evaluated the validity of the three approaches using the latest available trends in incarceration and crime (total, homicides, thefts) rates, and in wealth, unemployment, deflation and inequality in the EU’s 28 countries from 2002 to 2015. We applied different methodologies to investigate each approach – fixed-effects models that analyse time variations within countries, to explore the recession’s direct and indirect effects; random and mixed models that rely on within- and between-country variations, to explore the institutional hypothesis.

The results of the fixed-effects analyses identified two ways in which the recession affected incarceration across Europe – directly or indirectly, via crime changes. Both were associated with rising unemployment and its effects on incarceration.

We found that rising unemployment directly led to higher incarceration rates, adding to the evidence that, net of crime, labour surplus is significantly related to prison population (Chiricos and Delone, 1992; Sutton, 2012). We can only speculate on the mechanism explaining the effect. We saw that some scholars assign to imprisonment a disciplining effect over the labour force during times of labour surplus. Others saw it as resulting from higher ‘punitive motives’ accompanying mounting social anxieties (Melossi, 1987). However, other explanations are possible. For example, incarceration rates might have risen through fewer early releases aimed at shielding offenders from the deteriorating economic conditions, or through unpaid fines owing to criminals’ lower resources to avoid prison.

Rising unemployment also increased incarceration rates indirectly, via higher property crimes. However it also decreased them, by depressing – in combination with deflation – the illegal activities that foster violent crimes (Rosenfeld, 2009). Our findings confirm that the effects of economic conditions on crime are countercyclical regarding property crimes and pro-cyclical regarding violent ones (Bushway et al., 2012; Cantor and Land, 1985).

We could not find evidence linking recessions to crime and incarcerations through changes in inequalities and prices (net of unemployment). We found significant inverse associations between inequality and deflation, and between crime and incarcerations, but they seemed to be based on a contrast between the periods preceding and following the recession. The results suggest that the combination of deflation and high inequalities after 2012 had a lessening effect on incarcerations because it depressed the conditions fostering criminal activities, but also because it curtailed (poorer) governments’ capacity to maintain high numbers of prisoners (Sutton, 2012). This may help integrate past contradictory findings connecting these variables to incarcerations (Lessan, 1991; Phelps and Pager, 2016; Sutton, 2012).

We have unveiled important institutional differences across Europe in how the 2008 recession impacted crime and incarceration. We hypothesized and confirmed that in Northern Europe, which has inclusive and generous welfare states and penal systems oriented towards criminals’ rehabilitation, it unleashed less punitive reactions. We found the same lenient response in the corporatist countries of Continental Europe, as against the more punitive response found in the Southern, Eastern and Anglo-Saxon regions. Our results support some scholars’ argument that the main institutional factor explaining variations in punitive responses across Europe is the degree of coordination of a country’s economy and of bureaucratization of its penal system, which alter the costs of harsher punishments (Lacey, 2008; Savelsberg, 1994; Tonry, 2007). Other findings back this reading. Northern and Continental Europe’s lenient response to the recession was unmediated by variations in economic conditions – which were less deep than elsewhere – or in crime – which remained stable. Instead, they seemed to follow secular trends towards lower punitiveness.

In the more market-exposed regions of Europe, incarceration rates increased with the recession. In Eastern Europe – the region with the highest incarceration rates overall – they did it just slightly; in Anglo-Saxon countries, and especially in Southern Europe, they did it more significantly. These regions’ higher exposure to market forces is shown in the role played by unemployment in explaining the rise in incarceration – higher in Southern Europe, where the recession was longer and unemployment skyrocketed. The strict monetary policies imposed by the EU on member states – visible in their price convergence after 2009 - favoured the more coordinated economies of Central and Northern Europe but had high human costs in the EU’s periphery. As noted, it is hard to identify the mechanism explaining a region’s punitive response. However, in Southern Europe’s case it seems plausible that the combination of unemployment levels exceeding 20 percent and of limited public safety networks may have made imprisonment a tool for sheltering inmates from destitution, rather than for disciplining a labour force with little negotiating power. This interpretation is reinforced if we consider that here thefts were less commonly punished with prison. Had it not been because of governments’ dwindling economic resources, prisons could have been used more frequently as shelters.

In contrast, Anglo-Saxon countries’ response to the recession fits better the punitive role assigned by the literature to imprisonment in liberal states with common law traditions (National Research Council, 2014; Wacquant, 2001). Here, property crimes had higher probability to be punished with prison. Were it not because they decreased markedly with the recession and the contraction of illegal markets, the punitiveness of the Anglo-Saxon response would be more visible. Whether it ensued from purposely harsher punishments or it was the unintended consequence of insufficient welfare provisions remains to be seen. There is evidence that both mechanisms operated in the UK and Ireland during the recession (Roberts and Ashworth, 2016).

Eastern Europe stood in between the Anglo-Saxon and South European cases. Like in Southern Europe, in Eastern Europe the loss in wealth and in the governments’ capacity to maintain high numbers of prisoners limited the rise in imprisonment, which was equally linked to unemployment’s increase. However, like the Anglo-Saxon region, the East European region was more punitive in its response to property crimes and benefited similarly from their reduction during the recession.

Our sensitivity analyses showed that Eastern Europe is the least homogeneous region and that only one sub-region of Central European countries fits the average regional profile. The heterogeneity of the East European ‘construct’ was noted previously (Dünkel, 2017), and casts doubt on our treating Eastern countries as a single region, especially now that the transitional political stage has ended. However, we defend its utility, like most generally the utility of using regions rather than countries in cross-national analyses, based on its heuristic value. It is a tool that, without being optimal, can help uncover substantively important institutional differences.

Our article raises questions about the mechanisms that explain the association between recessions and incarceration rates, which will be explored in the future. However, it provides clear evidence about the variety of these effects and on how countries’ welfare and penal institutions moderate or exacerbate them.

Footnotes

Appendix

Table 4.

Variables and sources.

Variable	Source
Prison population rate (per 1,000,000 inhabitants)	Aebi (2005)
	Aebi and Delgrande (2015)
	Aebi, Tiago and Burkhardt (2015)
	Aebi, Tiago and Burkhardt (2017)
Gross Domestic Product per capita (euro per inhabitant)	Eurostat (2018a)
Gini coefficient of equivalized disposable income (scale from 0 to 100)	Eurostat (2018b)
Unemployment by sex and age – annual average (percent of active population)	Eurostat (2018c)
Inflation, consumer prices (annual percent)	World Bank (2018)
Theft rate (per 100,000 inhabitants)	UNODC (2018a)
Total crime rate (per 100,000 inhabitants)	Aebi et al. (2006)
	Aebi, et al. (2010)
	Aebi et al. (2014)
	UNODC (2018b) [used to impute missing values from 2012 to 2015]
Intentional homicide rate (per 100,000 inhabitants)	WHO (2018)
	UNODC (2018c) [used to impute missing values from 2014 to 2015]

Acknowledgements

We thank Elena Larrauri and the rest of the members of the Research Group in Criminology and Criminal Justice (Pompeu Fabra University) for their comments and suggestions on previous drafts of the article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/ or publication of this article: This research was supported by the project (Consecuencias socio-demográficas de la Gran Recesión: ¿Nuevas tensiones en las relaciones de clase y de género? – CSO2016-80484-R) funded by the Spanish Ministerio de Economía y Competitividad.

Notes

References

Aebi

(2005) Council of Europe Annual Penal Statistics, SPACE 1, Survey 2004. Strasbourg: Council of Europe.

Aebi

Delgrande

(2015) Council of Europe Annual Penal Statistics, SPACE 1, Survey 2013. Strasbourg: Council of Europe.

Aebi

Kuhn

(2000) Influences on the prisoner rate: Number of entries into prison, length of sentences and crime rate. European Journal on Criminal Policy and Research 8(1): 65–75.

Aebi

Linde

Delgrande

(2015) Is there a relationship between imprisonment and crime in Western Europe? European Journal on Criminal Policy and Research 21(3): 425–446.

Aebi

Tiago

Burkhardt

(2015) Council of Europe Annual Penal Statistics, SPACE I, Survey 2014. Strasbourg: Council of Europe.

Aebi

Tiago

Burkhardt

(2017) Council of Europe Annual Penal Statistics, SPACE I, Survey 2015. Strasbourg: Council of Europe.

Aebi

Aromaa

Aubusson

Cavarlay

Barclay

Gruszczyñska

von Hofer

Hysi

Jehle

J-M

Killias

Smit

Tavares

(2006) European Sourcebook of Crime and Criminal Justice Statistics - 2006 (3rd edition). The Hague: WODC.

Aebi

Aubusson

Cavarlay

Barclay

Gruszczyñska

Harrendorf

Heiskanen

Hysi

Jaquier

Jehle

J-M

Killias

Shostko

Smit

Þórisdóttir

(2010) European Sourcebook of Crime and Criminal Justice Statistics - 2010 (4th edition). The Hague: WODC.

Aebi

Akdeniz

Barclay

Campistol

Caneppele

Gruszczyñska

Harrendorf

Heiskanen

Hysi

Jehle

J-M

Jokinen

Kensey

Killias

Lewis

Savona

Smit

Þórisdóttir

(2014) European Sourcebook of Crime and Criminal Justice Statistics - 2014 (5th edition). Helsinki: HEUNI.

10.

Agnew

(1992) Foundation for a general strain theory of crime and delinquency. Criminology 30(1): 47–88.

11.

Arvanites

Asher

(1998) State and county incarceration rates. American Journal of Economics and Sociology 57(2): 207–222.

12.

Arvanites

Defina

(2006) Business cycles and street crime. Criminology 44(1): 139–164.

13.

Beckett

Western

(2001) Governing social marginality welfare, incarceration, and the transformation of state policy. Punishment & Society 3(1): 43–59.

14.

Blumstein

Beck

(1999) Population growth in US prisons, 1980–1996. Crime and Justice 26: 17–61.

15.

Blumstein

Cohen

(1973) A theory of the stability of punishment. Journal of Criminal Law and Criminology 64(2): 198–207.

16.

Bushway

Cook

Phillips

(2012) The overall effect of the business cycle on crime. German Economic Review 13(4): 436–446.

17.

Cameron

Trivedi

(2009) Microeconometrics with STATA. College Station, TX: StataCorp LP.

18.

Cantor

Land

(1985) Unemployment and crime rates in the post-World War II United States: A theoretical and empirical analysis. American Sociological Review 50(3): 317–332.

19.

Cavadino

Dignan

(2006) Penal policy and political economy. Criminology and Criminal Justice 6(4): 435–456.

20.

Chiricos

(1987) Rates of crime and unemployment: An analysis of aggregate research evidence. Social Problems 34(2): 187–212.

21.

Chiricos

Delone

(1992) Labor surplus and punishment: A review and assessment of theory and evidence. Social Problems 39(4): 421–446.

22.

Cohen

Felson

(1979) Social change and crime rate trends: A routine activity approach. American Sociological Review 44(4): 588–608.

23.

D’Amico

Williamson

(2015) Do legal origins affect cross-country incarceration rates? Journal of Comparative Economics 43(3): 595–612.

24.

De Giorgi

(2013) Punishment and political economy. In: Simon

Sparks

(eds) The SAGE Handbook of Punishment and Society. London: SAGE, 40–59.

25.

DeMichele

(2010) A Regime Theory Approach to Cross-National Incarceration Rate Variation: Three Worlds of Western Punishment, 1960–2002. American Society of Criminology, San Francisco, USA. URL (accessed 5 February 2019): https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1535265 (accessed 29 August 2017)

26.

Downes

Hansen

(2006) Welfare and punishment in comparative perspective. In: Armstrong

McAra

(eds) Perspectives on Punishment: The Contours of Control. Oxford: Oxford University Press, 133–154.

27.

Dünkel

(2017) European penology: The rise and fall of prison population rates in Europe in times of migrant crises and terrorism. European Journal of Criminology 14(6): 629–653.

28.

Durose

Cooper

Snyder

(2014) Recidivism of Prisoners Released in 30 States in 2005: Patterns from 2005 to 2010. Special Report. Bureau of Justice Statistics, US Department of Justice. URL (accessed 4 February 2019): https://www.bjs.gov/content/pub/pdf/rprts05p0510.pdf.

29.

Esping-Andersen

(1990) The Three Worlds of Welfare Capitalism. Cambridge: Polity.

30.

Esping-Andersen

(1999) Social Foundations of Postindustrial Economies. Oxford: Oxford University Press.

31.

Eurostat (2018a) Gross Domestic Product at market prices – Chain linked volumes (2010), euro per capita [nama_10_pc]. URL (accessed 4 February 2019): http://ec.europa.eu/eurostat/data/database.

32.

Eurostat (2018b) Gini coefficient of equivalised disposable income – EU-SILC survey [ilc_di12]. URL (accessed 4 February 2019): http://ec.europa.eu/eurostat/data/database.

33.

Eurostat (2018c) Unemployment by sex and age – annual average [une_rt_a]. URL (accessed 4 February 2019): http://ec.europa.eu/eurostat/data/database.

34.

Fajnzylber

Lederman

Loayza

(2002) Inequality and violent crime. Journal of Law & Economics 45(1): 1–39.

35.

Fenger

(2007) Welfare regimes in Central and Eastern Europe: Incorporating post-communist countries in a welfare regime typology. Contemporary Issues and Ideas in Social Sciences 3(2): 1–30.

36.

Garland

(1985) Punishment and Welfare: A History of Penal Strategies. Aldershot: Gower.

37.

Greenberg

(2001) Novus ordo saeclorum? A commentary on Downes, and on Beckett and Western. Punishment & Society 3(1): 81–93.

38.

Jacobs

Carmichael

(2001) The politics of punishment across time and space: A pooled time-series analysis of imprisonment rates. Social Forces 80(1): 61–89.

39.

Jacobs

Kleban

(2003) Political institutions, minorities, and punishment: A pooled cross-national analysis of imprisonment rates. Social Forces 82(2): 725–755.

40.

Jacobs

Richardson

(2008) Economic inequality and homicide in the developed nations from 1975 to 1995. Homicide Studies 12(1): 28–45.

41.

Kling

(2006) Incarceration length, employment, and earnings. American Economic Review 86(3): 863–876.

42.

Lacey

(2008) The Prisoners’ Dilemma: Political Economy and Punishment in Contemporary Democracies. New York: Cambridge University Press.

43.

Lacey

(2010) Differentiating among penal states. British Journal of Sociology 61(4): 778–794.

44.

LaFree

(2005) Evidence for elite convergence in cross-national homicide victimization trends, 1956 to 2000. The Sociological Quarterly 46(1): 191–211.

45.

Lappi-Seppälä

(2008) Trust, welfare, and political culture: Explaining differences in national penal policies. Crime and Justice 37(1): 313–387.

46.

Lappi-Seppälä

(2011) Explaining imprisonment in Europe. European Journal of Criminology 8(4): 303–328.

47.

Lessan

(1991) Macro-economic determinants of penal policy: Estimating the unemployment and inflation influences on imprisonment rate changes in the United States, 1948–1985. Crime, Law and Social Change 16(2): 177–198.

48.

Liedka

Piehl

Useem

(2006) The crime-control effect of incarceration: Does scale matter? Criminology & Public Policy 5(2): 245–276.

49.

Melossi

(1987) Political business cycles and imprisonment rates in Italy: Report on a work in progress. Review of Black Political Economy 16(1–2): 211–218.

50.

Messner

Rosenfeld

(1997) Political restraint of the market and levels of criminal homicide: A cross-national application of institutional-anomie theory. Social Forces 75(4): 1393–1416.

51.

Michalowski

Carlson

(1999) Unemployment, imprisonment, and social structures of accumulation: Historical contingency in the Rusche-Kirchheimer hypothesis. Criminology 37(2): 217–250.

52.

National Research Council (2014) The Growth of Incarceration in the United States: Exploring Causes and Consequences. Committee on Causes and Consequences of High Rates of Incarceration, Travis

Western

Redburn

Editors. Committee on Law and Justice, Division of Behavioral and Social Sciences and Education. Washington, DC: The National Academies Press.

53.

OECD (Organisation for Economic Co-operation and Development) (2016) Irish GDP up by 26.3% in 2015? URL (accessed 4 February 2019): http://www.oecd.org/sdd/na/Irish-GDP-up-in-2015-OECD.pdf.

54.

Perri

Quadrini

(2018) International recessions. American Economic Review 108(4–5): 935–984.

55.

Phelps

Pager

(2016) Inequality and punishment: A turning point for mass incarceration? ANNALS of the American Academy of Political and Social Science 663(1): 185–203.

56.

Pratt

Cullen

(2005) Assessing macro-level predictors and theories of crime: A meta-analysis. Crime and Justice 32: 373–450.

57.

Pratt

Lowenkamp

(2002) Conflict theory, economic conditions, and homicide: A time-series analysis. Homicide Studies 6(1): 61–83.

58.

Raphael

Stoll

(2007) Why Are So Many Americans in Prison. Institute for Research on Poverty – Discussion Paper no. 1328-07. URL (accessed 4 February 2019): https://www.irp.wisc.edu/publications/dps/pdfs/dp132807.pdf .

59.

Roberts

Ashworth

(2016) The evolution of sentencing policy and practice in England and Wales, 2003–2015. Crime and Justice 45(1): 307–358.

60.

Rosenfeld

(2009) Crime is the problem: Homicide, acquisitive crime, and economic conditions. Journal of Quantitative Criminology 25(3): 287–306.

61.

Rosenfeld

(2014) Crime and inflation in cross-national perspective. Crime and Justice 43(1): 341–366.

62.

Rosenfeld

Messner

(2013) Crime and the Economy. London: SAGE .

63.

Rusche

Kirchheimer

(1939) Punishment and Social Structure. New York: Columbia University Press.

64.

Savage

Bennett

Danner

(2008) Economic assistance and crime: A cross-national investigation. European Journal of Criminology 5(2): 217–238.

65.

Savelsberg

(1994) Knowledge, domination, and criminal punishment. American Journal of Sociology 99(4): 911–943.

66.

Soares

(2004) Development, crime and punishment: Accounting for the international differences in crime rates. Journal of Development Economics 73(1): 155–184.

67.

Sozzo

(2017) The renaissance of the political economy of punishment from a comparative perspective. In: Melossi

Sozzo

Brandariz

(eds) The Political Economy of Punishment Today: Visions, Debates and Challenges. New York: Routledge, 37–64.

68.

Stamatel

(2009) Correlates of national-level homicide variation in post-communist East-Central Europe. Social Forces 87(3): 1423–1448.

69.

Sutton

(2004) The political economy of imprisonment in affluent western democracies, 1960–1990. American Sociological Review 69(2) 170–189.

70.

Sutton

(2012) Imprisonment and opportunity structures: A Bayesian hierarchical analysis. European Sociological Review 28(1): 12–27.

71.

Tonry

(2007) Determinants of penal policies. Crime and Justice 36(1): 1–48.

72.

Uggen

(2012) Crime and the Great Recession. Stanford, CA: Stanford Center on Poverty and Inequality, Recession Trends Series.

73.

UNODC (United Nations Office on Drugs and Crime) (2018a) Theft at the national level, number of police-recorded offences. URL (accessed 21 May 2018): https://data.unodc.org/.

74.

UNODC (United Nations Office on Drugs and Crime) (2018b) Assault, robbery, motor vehicle theft, theft, total sexual violence and intentional homicide – number of police-recorded offences. URL (accessed 21 May 2018): https://data.unodc.org/.

75.

UNODC (United Nations Office on Drugs and Crime) (2018c) Intentional homicide at the national level, number of police-recorded offences. URL (accessed 21 May 2018): https://data.unodc.org/.

76.

Vold

Bernard

Snipes

(1998) Theoretical Criminology (4th edition). New York: Oxford University Press.

77.

Von Hofer

(2003) Prison populations as political constructs: The case of Finland, Holland and Sweden. Journal of Scandinavian Studies in Criminology and Crime Prevention 4(1): 21–38.

78.

Wacquant

(2001) The penalisation of poverty and the rise of neo-liberalism. European Journal on Criminal Policy and Research 9(4): 401–412.

79.

Wenzelburger

(2018) Political economy or political systems? How welfare capitalism and political systems affect law and order policies in twenty western industrialised nations. Social Policy and Society 17(2): 209–226.

80.

Western

Beckett

(1999) How unregulated is the US labor market? The penal system as a labor market institution. American Journal of Sociology 104(4): 1030–1060.

81.

Western

Pettit

(2010) Incarceration & social inequality. Daedalus 139(3): 8–19.

82.

Whelan

Nolan

Maître

(2016) The Great Recession and the changing intergenerational distribution of economic stress across income classes in Ireland: A comparative perspective. Irish Journal of Sociology 25(2): 105–127.

83.

WHO (World Health Organization) (2018) SDR, homicide and intentional injury, all ages, per 100 000. URL (accessed 4 February 2019): https://gateway.euro.who.int/en/indicators/hfa_185-1793-sdr-homicide-and-intentional-injury-all-ages-per-100-000/.

84.

World Bank (2018) Inflation, consumer prices (annual %). URL (accessed 4 February 2019): http://data.worldbank.org/indicator/FP.CPI.TOTL.ZG.

85.

Young

Brown

(1993) Cross-national comparisons of imprisonment. Crime and Justice 17: 1–49.

86.

Zimring

(2007) The Great American Crime Decline. New York: Oxford University Press.