Crime,punishment and segregation in the United States: The paradox of local democracy

Democracy in the USA: Non-partisan electoral competition in multiple arenas

Voter affiliations – and hence the strategies which candidates use in seeking election – are defined in the USA in terms of the policies and even personalities of current office-seekers or office-holders to a much greater extent than in the more disciplined political systems of the other Anglo-Saxon countries: in the US system, characterised by weak party discipline, it pays leaders, as individual candidates for office, to appeal directly to voters. In this context, policies likely to secure independent votes by appealing to median voter interests have become a key preoccupation for political leaders. To the extent that criminal justice is identified by politicians as just such an issue (here crime and the fear of crime potentially enter the political picture; Enns, 2014), this sets up, loosely speaking, a ‘prisoners’ dilemma’ (Lacey, 2008) in which candidates become locked into a strategy which they dare not abandon because of the electoral advantage, particularly vis-a-vis ‘floating voters’, which they fear would accrue to the other side. Key examples at the national level would be Richard Nixon’s War on Drugs (later amplified by Ronald Reagan) and Bill Clinton’s enthusiastic support for the death penalty. The key impact of the electoral ‘prisoners’ dilemma’ dynamic in the USA seems likely, however, to be at the state and local levels.

The impact of political institutions on criminal justice varies significantly as between different levels of electoral competition (Miller, 2008). But these levels are much more numerous and differentiated in the USA than in almost any other advanced democracy (Soskice, 2009). The radically extensive and extraordinarily decentralised quality of US democracy sets up, we suggest, two dynamics which, particularly in a world of relatively widespread anxiety about crime, strongly shape the formation of crime policy. First, it implies that the ‘prisoners’ dilemma’ is reproduced through very frequent elections at state, county and municipal levels, significantly increasing its impact. Second, it implies an accountability gap: individuals seeking election at the local level have an interest in advocating popular policies the costs of which do not necessarily fall on the electoral constituency (Boggess and Bound, 1993; Campbell, 2012: 306; Soskice, 2009; Stuntz, 2001; Zimring, 2012: App. B). Increased resort to imprisonment, where the political benefit accrues largely to local politicians while the costs fall on the state, would be a key example. This accountability gap is a common consequence of the negative externalities of local decision making, another example being expenditure on the education of those who subsequently move elsewhere.

In relation to the key policy areas relevant to poverty, segregation and crime and punishment, it would be hard to exaggerate the distinctiveness of the US tendency to organise governmental and executive power through electoral mechanisms at the local level (see Table 1). Though state politics and the varying institutional structure of states are undoubtedly of great importance in understanding crime policy (Barker, 2009), there is strong reason to think that the local level of the county or city, where in any case many state policies have to be implemented and interpreted – a level which is more laborious to research, and hence less fully understood – has been of equal or even greater significance. And if weak party discipline and personal platform-domination has largely characterised national and state-level politics, this is yet more true of local politics. Here actors with key roles in the criminal process – mayors, judges, district attorneys, sheriffs, to name only the most obvious – are, in stark contrast to other liberal market economies, often elected, and hence subject to direct electoral discipline; and their electoral campaigns depend on an extensive practice of radio and television advertising focused on individual record or policy commitments rather than on party platforms. Beyond this, the US practice of electing officials – County Commissioners, School Boards, Treasurers and so on – reaches deep into institutions at one or more remove from the criminal justice system, in which a median voter orientation will be likely, under certain conditions, to bring concerns about crime and punishment into play. These locally elected officials in turn appoint police chiefs and zoning boards. The resulting electoral cycle effects on policy areas such as police hiring, rates of prosecution, rates of dropping charges and conviction rates (particularly in property and drug cases: Dyke, 2007) and even judicial decision making on sentencing and case disposal are amply attested in a developing political science literature (Berdejó and Yuchtman, 2013; Brooks and Raphael, 2002; Huber and Gordon, 2004; Levitt, 1997; Shepherd, 2009) ² as well as in ethnographic research (Bogira, 2005: 311–336).

The collapse of Fordism and the rise in violent crime and punishment from the late 1960s to early 1990s: The polarising and reinforcing dynamics of local electoral democracy in education zoning and crime policy

In this section, we develop a testable thesis about how the dynamics of decentralised politics reacted to the social and economic upheavals of the collapse of Fordism and led to two decades of sharply rising violent crime and incarceration.

The ultimate locus of most democratic decision making in residential zoning, public schools and law and order (policing, public prosecution and non-appellate state justice) in the USA is the city or county; this entails that local autonomy has tremendous power, and that decisions reflect the interests and preferences of median voters. The decisive voter in local elections is likely to be a home owner with strong concerns (Fischel, 2001, 2004) to segregate the poor residentially – in the suburbs to keep the poor out, and in the cities to push the poor into their own enclaves (Beckett and Herbert, 2009); to keep property taxes low if public schools are bad (de facto segregated) or high if they are good, in order to maintain property values; and in any case to promote de facto educational segregation. The exclusionary implications of homeowners’ concern with property values is amply attested in the minutes of homeowner and neighbourhood associations in relation to zoning boards. These local political dynamics in zoning and public education explain the background conditions of densely segregated housing and de facto segregated and less resourced education for poor ethnic groups; as employment fell for less-skilled male workers in Fordist plants, so unemployment and poverty grew in these neighbourhoods – reinforcing the local politics of residential and educational segregation.

Another ‘local’ policy area is policing, with cities having independent police forces and in consequence senior appointments and decisions about police practices being made at local level. Refracted through mayoral and council elections broad police practices are responsive to local democracy. To take the most salient policy example, given residential segregation, homeowner voters will have reason to favour effective policing outside disadvantaged, high-crime areas and limited resources for policing in these areas (where victims are poor and are not median voters). Moreover in so far as these areas have high levels of violent crime and established gangs, they are extremely costly to police effectively. One can readily see how this may become a self-reinforcing process: with ineffective policing, violence is liable to increase, further eroding trust in the police and increasing the costs of ‘community’ policing. In this context, there is a greater temptation to turn to minimal, or to aggressive, highly intrusive policing tactics (Goffman, 2014; Meares and Kahan, 1999).

These policies on zoning, education and policing have strong implications for violent crime, both in highly segregated poor (largely black and Hispanic) tracts and via its spillover into middle class areas. In a powerful recent study, Krivo and Petersen have shown how violent crime is strongly related to highly segregated disadvantaged areas, with large ethnic populations and with a high proportion of high school dropouts, in large ex-manufacturing cities (Krivo et al., 2009; Peterson and Krivo, 2010). In an important finding, Krivo and Peterson also show that middle class areas in cities with high proportions of such poor high violence areas suffer from spillover of violent crime. Indeed, the violent crime rate in the most advantaged white neighbourhoods in cities with high segregation is equal to the rate in the least advantaged white neighbourhoods in low segregation cities (Petersen and Krivo, 2010). ³

Why did violent crime rise so dramatically from the late 1960s to the early 1990s in these disadvantaged areas? Violent crime was relatively low in the early and mid-1960s, then began rising. It levelled off briefly from 1980 to 1983 and then resumed a rapid rise in the late 1980s until the early 1990s since when it has fallen (Figure 2). Our argument is that the major trends over time are shaped by the availability of reasonable employment for males from disadvantaged backgrounds in segregated neighbourhoods. In the early to mid-1960s, the Fordist regime provided, directly and indirectly, employment even for those with weak educational background and low social capital. This was because semi-skilled employment on assembly lines required physical skills but limited analytical or social skills, and unionisation was relatively easy to impose on Fordist employers, because of the assembly line system. Moreover the semi-skilled had considerable power within unions because they were necessary to assembly line continuity. So even the disadvantaged shared in high negotiated wages. Fordism gradually collapsed as a competitive system from the late 1960s through to the early 1990s. The implications of this can be seen from data in Figure 3 on real hourly earnings of male workers in 10th and 20th percentiles of the labour force. OPEN IN VIEWER

OPEN IN VIEWER

Figure 4.

Change in aggravated assault rate and in population, seven largest non-southern cities, 1990–2009 (aggravated assault rates, UCR data (Uniform Crime Reporting statistics, available at: http://www.ucrdatatool.gov/data); population change, census data and population estimates).

What explains the collapse in 10th/20th percentile earnings illustrated in Figure 3? The 1970s and 1980s saw a continual middle class exodus from inner cities, which in itself reduced availability of low-skill service employment in the inner city. As Fordist jobs were largely eliminated by automation, new jobs increasingly followed middle classes away from city centres – and increasingly demanded analytic and social skills. And as middle classes moved away, so socio-employment networks – which could link those in the inner city with employment opportunities – declined. This in turn led to the collapse in inner city unskilled earnings. Many of the resulting group of unskilled, unemployed men were black Americans who had moved relatively recently, in the middle decades of the 20th century, from the Jim Crow South to the North in search of work and better opportunities – before becoming surplus to the requirements of the labour market in the 1970s (Wilson, 1987, 1996). The Civil Rights Act was less than a decade old when the economic collapse overtook the industrial cities; persisting discrimination and segregation in both housing and education meant that the educational and social disadvantages which African Americans had brought north with them remained ineffectively tackled.

Why did this cause the huge increase in violent crime? We suggest that there were two types of causes: lack of (or inappropriate) public policy responses; and individual responses by disadvantaged males. Taking policy causes first, our suggestion is that these developments made homevoters more determined to maintain residential segregation; and local democracy allowed these interests to produce zoning ordinances and their enforcement through zoning boards, with accentuated segregation contributing to the increase in violence and the social disorganisation which produces it. Second, public education opportunities for disadvantaged youth were determined through school boards; here again, as disadvantaged youth turned to violence and the alternative economies offered, for example, by gangs, homevoters became more concerned to reinforce de facto educational segregation, especially for high schools. Third, homevoter interests were also decisive in blocking the investment in serious foot patrols or ‘community’ policing in poor areas, which accordingly experienced either a vacuum of legitimate law enforcement or aggressive, military-style policing, often on an intermittent basis (Meares and Kahan, 1999). In terms of individual responses, disadvantaged youths accordingly found themselves in dangerous weakly policed environments, and street gangs (which had been unimportant before mid-1970s) developed as protection devices, using violence to protect their turf. This increased with their engagement in the drug market, encouraged by the ‘War on Drugs’, and was further increased by gang instability as leaders were incarcerated and new leaders needed to establish a reputation for toughness. Gangs further became involved in legitimate as well as illegitimate social control, and teamed up with institutions like block clubs, community associations and churches to provide local goods and security – hence with public local support and toleration of their illegitimate activities, albeit within limits (Pattillo-McCoy, 1999: chs 4 and 5). While there is a significant amount of ethnographic work to support this argument (Hagedorn, 1998, 2008; Patillo, 2007; Venkatesh, 2008), comparable aggregate data on gangs is poor. However, such data as exist confirm that this was a period of major upswing of gangs (US Department of Justice, 2011); and recent research has confirmed that this was also a major period of growth of prison gangs which were important in areas with big cities (Skarbek, 2014).

The juxtaposition of the extremes of wealth and poverty produced by the USA’s distinctive form of capitalism, along with the normative force of the ‘American Dream’ combined with the impossibility, for many, of realising it, have been argued to produce a criminogenic anomie (Lafree, 1998; Messner and Rosenfeld, 2007). The dominant explanation of the relationship between segregation, poverty and crime lies in ‘disorganisation theory’ (Hagan and Peterson, 1995a; Sampson and Groves, 1989; Sampson and Wilson, 1995) – the argument that deprivation and in particular concentrations of deprivation undermine the capacity of communities to sustain norms of order. Crucially for our thesis, much of that disorganisation in the USA is traceable to political decisions at the local level (Hagan and Peterson, 1995b: 15, 23–24; Massey and Denton, 1993: 14, 153–160). As Krivo and colleagues spell out the irony here, the greater the fragmentation and polarisation, the greater the need for coalition-building to resolve problems, yet the lower the capacity to engage in it (Krivo et al., 2009). In the face of these developments, it is hardly surprising if the middle class median voter reacted by interpreting crime and indeed poverty as matters of individual responsibility, thus legitimating their own resistance to voting for expensive strategies such as the improved housing, better schools and proactive policing (Alexander, 2012: 216).

Thus we believe the remarkable upsurge in violent crime through the 1970s and 1980s stemmed from three factors: ineffectively policed, poor (largely ethnic) residentially and educationally segregated neighbourhoods – the consequence of local democratic decisions reflecting the interests of middle class homeowners; the collapse of Fordist employment and the devaluation of less-skilled labour which led to mass unemployment of young ethnic males in these segregated neighbourhoods and sharply reduced returns to legal low-skilled work; and, as social disorder developed, the rapid growth of gangs, initially as self-protection institutions, where violence was a consequence of turf war and drug-dealing. These factors as we see it were self-reinforcing as the economy of the inner cities weakened.

A key element of our underlying argument is that violent crime drives punitiveness which in turn drives the rate of change of incarceration. We have shown (Figure 2) that (at least in terms of correlations) this is true at the national level. Moreover Miller (2015) has shown the relevance of violent crime for the political salience of crime and punishment. Unfortunately we do not have data for the punitiveness variable at the state or local level, and data for incarceration only at the state level. So while there is considerable ethnographic and electoral cycle research evidence (already cited) that prosecutors and district judges are responsive to electoral concerns about spillover violence, this is an area which needs to be further researched before it can be pinned down. What is known however is that high violent crime in poor neighbourhoods in a city produces a spillover in violent crime to middle class neighbourhoods; and that this spillover effect is magnified by increasing levels of violent crime (Petersen and Krivo, 2010). Given the relation between spillover violent crime and the political salience of punishment to which Miller (2015) has pointed, this seems to offer a potential explanation for the rapid increase in incarceration during the decades of rising violence. Moreover when violence stopped increasing and began to fall, so too did the increase in incarceration (Figure 2). Bearing in mind that the change in incarceration is (loosely) the variable on which policy-makers can act – by prosecuting and imprisoning offenders and then by hastening or not their release – this seems a relevant result. Again it is shaped by local democracy.

We might summarise our thesis by saying that when the poor have a strong presence (Wilson, 1987), these reinforcing policies lead to a ‘bad’ equilibrium; conversely, when the poor are hardly present ‘nice’ communities result from the same dynamics. Our hypothesis would explain how the ‘bad’ equilibrium created fertile conditions for persisting (racially patterned) ghettos, poor schools and sporadic and/or militarised policing which further attenuated the social structures which would have been capable of promoting social norms and order (Patillo, 2007). The thesis suggests that residential exclusion further leads to attenuated networks which create further barriers to integration and social mobility (Royster, 2003; cf. Charles, 2006; Pattillo-McCoy, 1999), and effects damaging black exclusion from pluralist politics (Massey and Denton, 1993: 153–160). The perverse incentives here are proportionate to the level of disadvantage: the worse the disadvantage, the greater the incentive for middle class voters to opt for segregation. Evidence for the politically self-sustaining quality of this equilibrium, even independent of race, can moreover be found in ethnographic research such as Pattillo’s rich study of the gentrification of a downtown Chicago area, Black on the Block, in which middle class blacks who have invested in housing exhibit much the same concern as their suburbanite white counterparts about the impact of the lifestyles of their less advantaged neighbours for property values (Pattillo, 2007: chs 2 and 6). Furthermore the electoral success of a significant number of black leaders depends on the concentration of black voters which is produced by residential segregation – a factor which may have diluted the incentive for black elites to fight segregation (Massey and Denton, 1993: 213–215).

The crime decline and the deceleration of punishment since 1990

Our argument so far suggests that the dynamics of voting and of electoral competition in the decentralised US system have implied powerful polarising forces, which have given added political weight to both criminogenic and punitive dynamics. The poor schools, residential segregation and zoned policing policies produced by local autonomy have, we suggested, combined with the economic and social upheavals generated by the prolonged collapse of Fordist semi-skilled employment for males to produce social disorganisation. And social disorganisation is associated with high levels of crime (Sampson, 1987; Sampson and Groves, 1989; Sampson and Wilson, 1995). But two obvious questions can be raised about this thesis. First, how can the marked decline in violent crime since the 1990s be explained? Second, while the marked decline in violence has indeed led to a corresponding decline in our measure of punitiveness (Enns, 2014: see Figure 2), how are we to explain the continuing rise in the rate of incarceration until very recently?

So why did violent crime start to decline in the 1990s? The gradual development of a knowledge economy combined with the prolonged period of growth and low inflation in the USA and elsewhere from the early 1990s to the financial crisis (the ‘Great Moderation’) generated a major population shift back especially into the largest cities where growth in high skill industries has been concentrated. Florida’s (2002) important work on the ‘Creative Classes’, as well as Glaeser and colleagues’ work on skill clusters (Chatterji et al., 2013), help to explain the gentrification of inner city areas during this period. We hypothesise that this co-movement of economic growth and of young professionals into the inner city led to declining crime for three interrelated reasons. First, as Figure 3 demonstrates, it decisively raised real hourly earnings of unskilled male workers (10th percentile) as a corollary of the need for unskilled work to complement the emerging skill clusters. Second, with rising population in city centres, actual residential segregation declined between 1990 and 2000 (Rothwell and Massey, 2009). This in turn transformed the political dynamics, in that the new residents wanted effective policing of these areas including city centres. Recent research by Sharp (2013) has shown econometrically across a range of metropolitan areas that the key variables in determining the importance of ‘Order Maintenance’ policing are, first, the percentage of ‘creative classes’ or other measures of post-industrial development and, second, police use of surveys of citizen concern on crime and disorder to guide police deployment. Our tentative argument is therefore that violent crime declines (in the cities where it does) for fundamentally political-economic reasons; and that these political, economic and demographic developments are already having significant effects on the dynamics of local democracy.

If the new political-economic dynamics have had decisive implications for crime, however, the polarising dynamic of US local politics remains evident. For those unable or unwilling to play by the rules of the newly refurbished cities, or whose presence promotes feelings of insecurity among residents or those spending money on property investment, retail or leisure activities, new kinds of zoning regulation have been created. These regulations take the form of local civil, criminal and planning laws to ‘banish’ the troublesome from middle class areas (Beckett and Herbert, 2009).

We can loosely test this broad explanation, since not all very large cities benefited from falling violent crime from the 1990s. The effects upon crime are suggested by Figure 4: of the largest cities outside the South, New York, Boston, Los Angeles, Chicago and Washington, DC all saw falling violent crime from the early 1990s on, while their population grew; Philadelphia, ⁴ Detroit, Milwaukee and Baltimore did not have declining violent crime (at least until the 2000s), and did have continuing population decline.

It can be seen from Figure 2 that the decline in violent crime since the early 1990s has been associated with a fall in the Enns (2014) punitiveness index, which measures changing attitudes to the rate of change of incarceration over time, and compares this with both public attitudes to crime and rates of violent crime. Incarceration has gone on rising, so does that mean that the link between punitiveness and policy responses via incarceration has been broken? There is good reason to think otherwise: for the continuing increase in the imprisonment rate is driven by many factors which are not readily within policy-makers’ control – notably the continuing impact of increased numbers of long term prisoners on the global number. So the relevant policy response which we would expect to see if there is indeed a link between crime, punitiveness and imprisonment is rather an effort to change the rate of incarceration, by operating either on new admissions to prison or on releases. As can be seen in Figure 2, the rate of change in imprisonment, after rising sharply through the 1970s and 1980s, began to fall equally fast. This suggests that the effect of crime on punitiveness and the effect of punitiveness on policy on imprisonment have remained very much in place, albeit that institutional path dependence and the accretion of commercial interests in the prison system may well have slowed the deceleration of imprisonment to some extent (Gottschalk, 2014: 25–74). Furthermore, the rate of change remained positive until recently, with rising incarceration (although significantly lower for violent crime). This is an area for future research, but some part of the explanation is likely to do with the more intensive policing of urban centres (often oriented to ‘banishment’ of the homeless or otherwise troublesome: Beckett and Herbert, 2009).

Centralised vs local democracy: US crime, punishment and democracy in comparative perspective

We have already suggested that the power of homevoters – people particularly motivated to vote out of concern with ensuring policies which protect the value of their primary asset – (Fischel, 2001, 2004, 2005), in the US system of local democracy may hold an important key to understanding the distinctive dynamics of crime and punishment. And we know that ‘homevoters’ turn out to vote in much higher numbers than others (Hajnal and Trounstine, 2005), resulting in a more advantaged median voter at the local level. Particularly given low turnout and homevoter interests, median voters at the local level are hence likely to be considerably to the right of median voters at the federal or even state level (Cutler et al., 1999; Hajnal and Trounstine, 2005: 516–517). They are also more likely to be significantly concerned about the impact of crime on property values and the quality of local services such as education than general survey respondents. It is therefore a significant problem that most evaluations of the political impact of concern about crime are based on generalised survey evidence (e.g. Beckett, 1997). Our hypothesis is that the homevoters who have so much power in local elections are in a position to co-opt the local state, which in effect operates in their private interests.

But why should this ‘homevoter’ effect be so powerful in the United States, leading to unique levels of residential segregation (Johnston et al., 2007; Peach, 1996; Reardon and Bischoff, 2011)? Two considerations are important, and differentiate the United States from comparable countries such as the United Kingdom. The first has to do with the structure of the labour market and of pensions provision. The motivation of the median voter is almost certainly strongly affected by concern about property values in other countries where home ownership is widespread, as is the case in most liberal market economies. But in the USA the home represents not only the largest personal investment for many middle class families, but also their pension pot, as compared with more complete public and occupational pension provision in the less flexibilised liberal market economies. Growing rates of home ownership through the second half of the 20th century have increased the relevant dynamics (Simon, 2010).

The second consideration follows from the fact that the main mechanism by which segregation is achieved is that of zoning – a key form of autonomous political power at the local level ever since the Supreme Court ruled on the legality of zoning ordinances in 1926. ⁵ By contrast to the localised basis for zoning, education and policing decisions in the USA, all the key rules governing zoning, police organisation and practice, justice, public prosecution and the education system are made primarily at the national or provincial level in other liberal market economies such as Canada, the UK and New Zealand (Table 1). This may help to explain the striking differences between the USA and other liberal market economies which have also struggled with issues of inner city decline and unemployment following the collapse of Fordist production in the 1970s. Despite the ‘prisoners’ dilemma’ effects attendant on national governments’ competing for median voters, more centralised systems, our thesis would suggest, avoid the negative externalities of local decision making characteristic of the USA. This is because of an important difference in the typical interests of median voters – and hence of politicians – at national as opposed to local level, even leaving the demographic differences just discussed aside. The idea that people are more likely to vote for public goods at the national than at the state or local levels seems counter-intuitive given evidence that more homogeneous groups are more likely to vote for collective goods (Alesina et al., 1999). However, it makes sense if one assumes a mobile society in which people vote at the national level for goods from which they will benefit wherever they live. Conversely, at the local level people will be inclined to vote only for things from which they can be guaranteed to benefit – hence voters are less likely to vote for long term investment, notably in education, where the benefits accruing may then move elsewhere. Local voters, in short, have an incentive to ‘capture’ the benefits of social policy by restricting their support to policies from which they are sure to benefit. Under conditions of relative homogeneity, this becomes a less pressing concern. But the extensiveness of local electoral government means that arrangements which are in the interests of politically powerful swing voters – themselves drawn from more advantaged groups – can more readily be inscribed in public policy.

If it is true that the positive externalities – the expectations that they may benefit from widely diffused goods – mean that voters are more likely to vote for public goods at higher levels of government, especially the national level, it would follow that localities make a poor basis for long term policy making let alone redistribution. (Indeed many progressive reforms at the state level through US history have been motivated by the recognition that local decision making performs poorly in terms of redistribution: Zackin, 2013: 103.) And while there has been some progress in the USA, especially at the state level, in mounting legal challenges to impose uniform standards in education – notably in decoupling school spending from local property tax revenues (a strategy which itself ultimately depended on state enforcement (Corcoran et al., 2003)) – housing and zoning policies remain strongly shaped by local interests, with devastating effects for efforts at desegregation, in both class and race terms. Even in the area of education, litigation strategies have had mixed success (Douglas, 2005; Frankenberg and Orfield, 2012), with network-based inequalities rooted in factors such as private/parental contributions to school infrastructure and variations in teacher standards persisting even in the face of some political successes at the state level in increasing overall per capita spending. The continuing battle to counter voting restrictions likely disproportionately to exclude the disadvantaged is testimony to the power of the relevant interests here.

The structure of local government and in particular of ‘city trenches’ in the USA is well known to have accorded power to ‘in-groups’ (typically those with property, strongly associated with race, for obvious reasons: King, 2007). Early city politics ran on ethnic lines, and provided a strong basis for reproduction of group/sectoral identities and ethnic separation (Katznelson, 1981: 104ff, 80ff). Newer local political structures such as school boards have in many ways reproduced the old group-based Tammany Hall structures. ⁶ The scope of local autonomy is such that huge regional and intra-state differences of what would in more centralised countries be stabilised by public provision and national regulation can emerge. In the UK, for example, policing, education and planning all go forward within a national legislative framework, with modest provision for local control/variation. In Canada, these policies are largely framed at Province level, while even voter preferences for more localised city government have on occasion been overridden by provincial legislatures in the interests of better co-ordination of policy (Mitchell-Weaver et al., 2000: 865). In the ‘balkanised’ (Miller, 2010; Orfield, 2002) local government of the USA, multiple jurisdictions do not have to consider effects on other jurisdictions. Yet more importantly, the fact that elections are decided by a relatively small group of ‘median’ voters means that the ‘truly disadvantaged’ are rarely heard at the ballot box even if they vote, and that this problem becomes more intractable the greater the degree of inequality and concentration of disadvantage. Again, this is confirmed by empirical research: Pattillo’s (2007) black ‘middlemen’ lost power due to demographic changes and the collapse of Fordism.

That the voices of the disadvantaged tend to be muted in electoral politics, and that that disadvantage itself was accentuated by the collapse of Fordism is, of course, true in all the industrialised democracies. What is special about the USA, however, is the degree to which the structure of the political system has allowed these widespread facts to issue in distinctively polarising policy. Our thesis implies that local autonomy in the USA means that wealthier groups can opt out of collective problems via the construction of gated communities or the purchase of private education or private security: they can even incorporate within a new city with its own zoning laws. But, yet more importantly, local autonomy means that the local state itself can be invoked for similar structural purposes. In other words, zoning decisions, public housing policy, policing and school funding can be organised in the interests of the middle classes who swing elections. Hence to the extent that their interests have become more different, and that they view their fate as more and more independent from that of the disadvantaged, support for redistribution to the latter becomes ever harder to motivate. Moreover this would not be changed by a greater emphasis on local democracy or a change in constituency boundaries. For even granting Miller’s (2008) finding that high victimisation/high crime groups such as poor inner city blacks have a more sophisticated view of crime policy than do more privileged groups, to implement the sorts of policies which that sophisticated view would endorse – better housing, education and employment – would require resources. And these resources in turn depend to a significant extent – particularly in the wake of the federal funding cuts and squeezed state budgets of recent years – on a local tax base which would be severely attenuated. There is an irony here: in this markedly anti-statist democracy, it seems that the local state has become a powerful medium for realising private interests. Our suggestion is that these decentralisation effects, though they vary according to specific state structures (Barker, 2009), are sufficiently pervasive to constitute a key part of the explanation for American exceptionalism.

One further aspect of decentralisation bears on our explanatory hypothesis about the USA’s distinctive problems of criminal violence, punitiveness and polarising social policy: the coordination problem implicit in reliance on legal enforcement, often at the local level, to implement national or state policy. As the Great Society and New Deal Programmes of the mid-20th century attest – an era, incidentally, which saw drops in the homicide rate (Hall and McLean, 2009) – the USA can develop significant national policies in areas such as education, housing and criminal justice. President Obama’s Affordable Care Act provides a more recent example. But the federal government can push forward nationwide policy objectives only within certain constitutional constraints and under certain conditions: either where a party with a clear programme has control of Congress as well as the Presidency, or where there is cross-partisan consensus. Where there is neither this power nor this consensus at the federal level, everything turns on states or localities. And even when federal initiatives are brokered, implementation largely rests on action at the state and local levels (Feeley and Sarat, 1980). In the absence of powerful agencies or plentiful funding to provide economic incentives, implementation furthermore has to be triggered either by legal enforcement or by local political will (Lacey and Soskice, 2016).

Law is often regarded as a particularly salient resource to counter the discriminatory effects of public power in the USA. But local autonomy means that law can often be subverted to the interests of the relatively advantaged. For the democratic choice at the local level of local judges and district attorneys blunts what has been seen as the primary resort of minorities whose rights have been abused by the political will of local majorities, namely the US Constitution and the Constitutions of the several states. ‘Legal adversarialism’ (Kagan, 2001) has undoubtedly put tools into the hands of litigants with the resourcefulness (and resources) to challenge outcomes such as educational segregation or housing discrimination. However, the democratic choice of justices at the local level makes effective implementation of standards against local majority will difficult, even leaving aside pathologies of legal adversarialism such as high costs, delays, patchy impact and ineffectiveness at the level of implementation (Douglas, 2005; Frankenberg and Orfield, 2012; Frankenberg et al., 2003). And this in turn – notwithstanding, and indeed in some ways evidenced by, the limitations of the various programmes rolled out under the aegis of the New Deal and the Great Society in the 20th century – relates to an underlying structural difficulty for the Presidency to establish micro-governing, rule-implementing federal bureaucracies across the USA (Feeley and Sarat, 1980). The US system moreover features lower status bureaucracies, and a lower overall level of trust in expertise, than other comparable countries (Savelsberg, 1994, 1999). Lessons from school segregation and civil rights history show that while litigation strategies can achieve real progress, they are both costly and divisive: they provide an adversarial framework for policy implementation, while individual case-based legal remedies or even class actions are rarely effective to resolve structural or coordination problems (Aaronson, 2015; Douglas, 2005; Kagan, 2001). Hence this distinctively US translation of political activism into legal strategy has significant disadvantages. It is no surprise that law has come to assume such dominance in the US system: it makes sense that the more individualistic and fragmented the society, the more likely it will be to resort to legal enforcement which does not depend on compromise and negotiation, hence bypassing structural problems of coordination. But these problems, inevitably, reproduce themselves at the implementation level, with key recent examples including the lack of enforcement powers under the Fair Housing Act (Massey and Denton, 1993: 14–15, 187, 223–230, 206 ff) and the notoriously long-running Gautreaux litigation in Chicago (Pattillo, 2007: 110–114; Peterson and Krivo, 2010: ch. 5). These difficulties have become more acute in an era of declining funding for agency enforcement (Pierson and Skocpol, 2007).