State-led Gentrification and the Changing Geography of Market-oriented Housing Policies

Abstract

Governments in a wide range of contexts have long pursued policies of social mixing to disperse poverty concentrations, attract middle class residents, and manage disadvantaged neighbourhoods. Drawing on longitudinal and spatial housing data for the case of Amsterdam, this paper shows that the dominant instruments to facilitate social mixing have changed over time. Policy focus has shifted from large-scale urban renewal projects and the demolition of social rental housing to the sale of existing social rental dwellings. The changing nature of tenure restructuring also brings about a changing geography: while urban renewal was mostly concentrated in post-war neighbourhoods of socio-economic decline, social housing sales are increasingly concentrated in inner city neighbourhoods where already existing gentrification processes are amplified. These shifts need to be considered within their wider policy context. Local policies increasingly focus on catering to the preferences of middle class households, while welfare state restructuring and national austerity measures push policies that cut back on social rental housing. Thus, this paper demonstrates that the demise of social rent has accelerated under conditions of market-oriented housing restructuring, and increasingly occurs in high demand neighbourhoods where current housing policies push gentrification.

Key Words:

• State-led gentrification
• Social mixing
• Housing policy
• Social-rental housing
• Housing liberalization

Introduction

Housing policies form a key instrument for steering urban development. Across contexts, homeownership is ideologically pushed as the “superior” tenure form (Ronald 2008), as it is assumed to have a positive influence on individual homeowners and society at large. At the urban or neighbourhood level, homeownership is also expanded to address a plethora of issues. These issues include social and physical neighbourhood problems, the economic strength of cities, and the provision of housing and urban milieus attractive to middle class residents. Tenure mixing policies typically entail the introduction of more expensive owner occupied dwellings and more affluent residents to targeted neighbourhoods, frequently at the cost of affordable rental housing for lower income tenants. These policies may therefore be considered part of state-led gentrification (Smith 2002; Uitermark, Duyvendak, and Kleinhans 2007; Lees 2008; Bridge, Butler, and Lees 2012). State policies may amplify already existing gentrification processes, or alternatively seek to spark gentrification in disadvantaged neighbourhoods.

The expansion of homeownership in existing urban neighbourhoods is generally coupled with a reduction in the number of rental dwellings, particularly cheaper or rent-regulated stock. There are different policy approaches for doing so: rental dwellings may be demolished and replaced with new-build owner occupied dwellings. Alternatively, rental dwellings may be brought onto the market and converted into homeownership. Furthermore, it is not only important to consider how changes in tenure composition come about, but also where tenure restructuring¹ is pursued. The promotion of homeownership at the cost of rental units may be focused on “disadvantaged” neighbourhoods in an attempt to mitigate or counter developments perceived as negative (Uitermark 2003). Urban policies may also seek to facilitate market processes by deregulating the housing stock and expanding homeownership in those neighbourhoods that are already in high demand. Although much research attention has been paid to tenure mixing policies, little is in fact known about how the instruments and geography of tenure restructuring may be interrelated, and how both may have been subject to change over time. Focusing on social housing reductions, this paper provides rare insight into the changing geography of state-led gentrification under conditions of market-oriented housing restructuring. Drawing on the case of Amsterdam (the Netherlands), this paper sets out to investigate these topics by answering the following question:

Where and how are policies of tenure mixing pursued, particularly reductions in social housing, and how has this changed over time?

This research question is answered for the period 1998–2015 describing both overall changes as well as specific tenure transitions on the basis of longitudinal and spatial housing data available from the Amsterdam Federation of Housing Associations (AFWC), and the municipal statistical department (OIS Amsterdam). Furthermore, attention will be paid to how these changes have contributed to a changing accessibility of the social rental housing stock. The question of how and where tenure restructuring policies are pursued, and how this changes over time, may be related to broader contextual and policy factors. This paper therefore not only describes these dimensions and shifts but also contextualizes them by linking housing dynamics to urban and housing policies, as well as to macro trends and societal debates regarding social housing. Amsterdam is an interesting case through which to investigate this question because it has a rich history of providing affordable social rental housing to a large segment of its population. However, urban policies that promote homeownership at the cost of social rental housing have also been a permanent component of the national and local political landscape since the 1990s (Aalbers 2004; Uitermark 2009). Furthermore, the local government has embraced gentrification as a positive policy instrument, as stated in the city’s most recent housing policy whitepaper (Van Gent 2013).

This paper progresses as follows: first, the theoretical section further elaborates on welfare state restructuring, the various reasons for tenure mixing policies, and how these different rationales may also be reflected in the geography and instruments of tenure restructuring. Furthermore, it specifically highlights scholarly work on the linkages between social mixing and state-led gentrification. Second, this paper further excavates the context of the Amsterdam case before presenting long-term and spatial data on overall shifts in Amsterdam’s housing tenure composition, paying specific attention to the sale and demolition of social housing. Subsequently, these changes are contextualized by linking them to national policies and macro developments.

Literature

Tenure and Welfare State Restructuring

Tenure restructuring may be pursued for a range of different reasons, using different instruments at different locations. Urban and neighbourhood level policies intersect here with national policies and broader trends of welfare state restructuring. Most European welfare states have, through policies of de-regulation and re-regulation, shifted their attention and resources over recent decades towards enabling market forces (Brenner, Peck, and Theodore 2010). Market-oriented reforms have solidified the privileged position of private property and stimulated the private accumulation of wealth (Aalbers and Christophers 2014). Homeownership has substantially grown under conditions of welfare state restructuring, as it is ideologically cast as the most desirable form of tenure (Ronald 2008). With welfare safety nets dismantled, individuals are encouraged to secure future welfare through asset accumulation (Doling and Ronald 2010). Homeownership is promoted as an effective means for households to accumulate assets, as they pay off mortgage debt and raise house prices. Under conditions of crisis-related austerity, private accumulation may become even more important as state spending on public services is reduced.

In contrast, welfare state restructuring as well as austerity measures have hit rental housing segments in various ways. The share of rental housing, and particularly rent-regulated social rental housing, has decreased in most countries. These policies work towards a residualization of the social rental stock (Forrest and Murie 1983; Van der Heijden 2002). The underlying assumption is that as private ownership is considered more desirable, social housing should only cater to those with few or no options to acquire property themselves. Processes of social housing residualization cater to a dualization of the housing market, which entails that different housing market segments increasingly come to serve different strands of the population (Kemeny 2001): a small social rental sector for low-income households, while owner occupancy but also private rent caters to households higher on the socio-economic ladder. Welfare state restructuring not only seeks to reduce the size of social rental housing, but policy reforms are generally also geared towards the de-regulation and re-regulation of rental segments. Examples of such reforms are the gradual erosion of tenant rights, a reduction in state subsidies, and efforts to give private landlords and investors more leeway (Huisman 2016; Kadi and Ronald 2016).

Various rounds of welfare state restructuring and austerity measures also have a more indirect impact on urban and housing policies. As a consequence of reforms, it has become more difficult for states to intervene in traditionally key domains since they increasingly lack the means and power to do so. This makes it ever harder to address the structural causes of issues such as deepening socio-economic inequalities and social exclusion. Consequently, it has been suggested that states increasingly shift their attention and efforts to specific areas or neighbourhoods where they are still able to manage and control the local effects and outcomes of these issues (Uitermark 2014). Local urban and housing policies may thus play a crucial role, as the following sections will highlight.

Tenure Restructuring and Disadvantaged Neighbourhoods

At the urban level, a key reason for policy-makers to engage in policies of tenure mixing is to achieve a change in population composition and reduce levels of residential socio-economic and/or ethnic segregation. High levels of segregation are often assumed to have additional negative consequences for those living in areas with a high concentration of poverty (Wilson 1987). The assumed underlying causal mechanisms for these additional effects are, for example, a lack of positive role models living in a neighbourhood, a negative work ethic, the lack of useful local social networks, and the stigmatization of neighbourhoods (Wilson 1987). These assumptions form the core of the “neighbourhood effects” thesis (Sampson, Morenoff, and Gannon-Rowley 2002; Van Ham et al. 2012). Although evidence for the existence of substantial neighbourhood effects remains mixed, the thesis has had a major impact on urban policies (Ostendorf, Musterd, and De Vos 2001). It provides a clear-cut legitimation for direct intervention in the social and physical structure of disadvantaged neighbourhoods to facilitate the introduction of more expensive housing and middle class residents, and to disperse lower income residents. In addition, it has been suggested that the neighbourhood effects thesis allows states to address incremental differences in poverty without addressing the key institutional arrangements responsible for producing structural inequalities (Slater 2013). Although it is beyond the scope of this paper to discuss these projects at great length, examples from the US include the Moving to Opportunity experiment (De Souza Briggs, Popkin, and Goering 2010), and the demolition and renewal of public housing estates through the HOPE VI program (Hackworth 2003; Popkin et al. 2004). In Western Europe, policy responses generally entail the wholesale or partial demolition or restructuring of impoverished neighbourhoods (Kleinhans 2004; Andersson and Musterd 2005; Bolt, Phillips, and Van Kempen 2010). Dutch urban renewal policies have been particularly ambitious in altering the housing composition of “disadvantaged” urban neighbourhoods in an attempt to dissolve poverty concentrations and counter the corresponding accumulation of problems (Uitermark 2003).

Given the mixed evidence for neighbourhood effects and the generally modest effects of urban restructuring policies on individuals’ life outcomes (Kleinhans 2004), states may have different rationales to invest in social mixing. Uitermark (2003) argues that in neighbourhoods where large poverty concentrations exist and the share of unemployment is relatively high, local bureaucrats may feel that they lack the means to address local problems. The introduction of middle class households that are generally supportive of state policies is supposed to make neighbourhoods easier to manage and amenable to government interventions. Social mixing thus becomes a strategy to ensure or bring back social order and control over deviant spaces (Uitermark, Duyvendak, and Kleinhans 2007). Such policies of social mixing through tenure restructuring therefore take place in areas where poverty concentrations exist and the negative effects of segregation seem most visible. These will generally be neighbourhoods that are considered to be low in the urban hierarchy. Here, governments may intervene to reverse negative developments and counter market processes of income decline.

Tenure Mixing and Gentrification

Policies of tenure mixing and social mixing have frequently been associated with state-led gentrification processes, because social rental housing and low-income tenants typically have to make way for more expensive owner occupied dwellings and a higher income clientele (Lees 2008). Furthermore, in certain cases, it may be considered a form of “gentrification by stealth”, as the language of social mixing conceals the class-based changes induced by gentrification (Smith 1996, 2002; Bridge, Butler, and Lees 2012). Social mixing may bring about the involuntary displacement of longer term tenants, although in the Dutch context, the relocation process is guided and compensated for, which mitigates some of the negative effects (Kleinhans and Kearns 2013; Posthumus 2013). In contexts where tenant protection is strong, rent regulation is in place, and relocation is guided, social mixing policies may still have a substantial impact as the structural decrease in affordable rental dwellings available to lower income residents effectively excludes them (Slater 2009). These shifts are especially likely to have a detrimental impact on the housing position of low-income housing market outsiders, as well as those low-income households in direct need of housing (Kadi and Musterd 2015).

The strong and proactive involvement of local states in pushing gentrification is one of the defining features of contemporary “third wave” gentrification, setting it apart from the preceding wave. State involvement has also been important in allowing gentrification processes to spread geographically: away from the inner neighbourhoods of major cities into secondary cities, as well as to neighbourhoods further from the urban core (Hackworth and Smith 2001; Smith 2002; Hackworth 2002). Governmental strategies that facilitate gentrification are often pursued for – or legitimized by – economic reasons. In some contexts, attracting middle class residents is important for municipalities to increase their local tax base. However, also in welfare state arrangements where tax revenues play little to no role – as in the Dutch case – attracting higher income and higher educated residents may be an important part of local authorities’ entrepreneurial strategies to enhance their city’s competitive position (Harvey 1989). These policies are often heavily inspired by Richard Florida’s Creative Class thesis and seek to provide attractive, “authentic” urban environments with which to lure in the creative class (Florida 2002; see also Peck 2012). Large-scale urban restructuring that includes the demolition of old housing stock and the construction of new dwellings may be at odds with the ambition to provide such environments. In this case, local authorities may opt instead for other interventions in the housing stock that better suit this goal. Particularly in cities and neighbourhoods where the social rental housing stock is relatively large, local authorities can play a crucial role in spurring gentrification through encouraging the sale of social rental housing or the removal of rent controls (Hamnett and Randolph 1984; Andersson and Turner 2014; Boterman and van Gent 2014). By bringing former social rental housing onto the market, these dwellings may become accessible to higher income households, since eligibility for social rental housing is generally restricted to lower income households. Tenure conversions as part of governmental gentrification strategies may therefore concentrate in neighbourhoods where demand from higher income households is already present and can be catered to. Consequently, the social rental stock that functions as a brake on gentrification in these neighbourhoods is gradually eroded (Van Gent 2013).

Housing and Geographical Context

This paper charts how Amsterdam’s housing composition has changed over time, where this is done, and how. It focuses particularly on shifts in the social rental stock owned by housing associations. Housing associations are semi-private not-for-profit institutions and are regulated by the state to focus on their official core task of providing affordable rental housing to low-income households. Housing associations traditionally rent out the vast majority of their stock in the rent-regulated sector where maximum rents are determined on the basis of a legally binding point-scoring system that takes into account various dwelling characteristics (e.g. size, type of dwelling, energy efficiency, general quality), and more recently also location. All rent-regulated dwellings have a monthly rent of below roughly €700 (subject to yearly incremental changes).² Exceptions notwithstanding, since 2011 only households with an income up to €35,000, or €39,000 for certain families, are eligible for social rental housing allocated via a central waiting list. The average waiting time for a social rental dwelling is around nine years, although active search period is typically shorter. Dwellings scoring a sufficient number of points can – but do not have to be – be “liberalized”, which means that rents can be raised above the aforementioned maximum of €700 per month. Above this threshold rents are no longer regulated and can be freely determined. Furthermore, maximum income criteria and waiting lists do not apply to rent-liberalized dwellings, meaning owners can decide on allocation. In Amsterdam, liberalized rents are typically relatively high and can far exceed the €700 liberalization threshold due to high levels of demand. This segment therefore primarily accommodates middle or higher income households able to shoulder high rent burdens. Next to the sale and demolition of social housing units, rent liberalization represents another way through which the number of rent-regulated dwellings accessible to lower income households can be reduced, although liberalization is only possible when dwellings are vacated by the previous tenants.

This paper presents data on changing patterns of social rental housing sales and demolition, as well as overall tenure shifts in Amsterdam. Specific attention is paid to the changing geography of these tenure shifts. Because information on sales and demolition are only available at the borough level, I group together Amsterdam’s four central boroughs (Centre, East, West, South) and its three more peripheral boroughs (North, New West, and Southeast) (Figure 1). This distinction roughly captures the division between Amsterdam’s gentrifying central city and its struggling urban periphery (Hochstenbach and van Gent 2015). Data on the overall changes in Amsterdam’s tenure composition are available on a lower spatial scale, enabling the definition of a more fine-grained neighbourhood typology based on dominant building period and the traditional status of the neighbourhood (Figure 1). The first neighbourhood type comprises the central city – including the seventeenth century canal belt – and the Old South district. The neighbourhoods belonging to this type have always or for a long time been affluent. The second neighbourhood type is the 1800–1920 belt directly surrounding the inner city. Most of these neighbourhoods experienced socio-economic decline throughout the second half of the twentieth century. The same goes for the third neighbourhood type in the 1920–1940 belt. Both the 1800–1920 and 1920–1940 belts are characterized by relatively large shares of social housing and relatively small apartments. However, both belts currently represent the city’s gentrification frontiers, with the process in a generally more advanced stage in the nineteenth century neighbourhoods (Hochstenbach and van Gent 2015). The fourth neighbourhood type encompasses the post-war expansions to the city, which were built as a response to housing shortages as well as the low quality of housing that then dominated Amsterdam’s central city. Although initially home to mostly (lower) middle class households, the post-war neighbourhoods have experienced long-term processes of socio-economic downgrading (income decline and growing concentrations of poverty) and are now often low on the urban neighbourhood hierarchy. The fifth typology includes newly built areas (constructed after 1990) as well as rural areas. These areas have been combined into a single type because they are both primarily middle class residential areas and are characterized by above average shares of owner occupation.

Figure 1. Neighbourhood typology in Amsterdam based on building period and urban milieu. Source: OIS Amsterdam, own adaptation.

Amsterdam’s Changing Tenure Composition

Since the 1901 Housing Act, not-for-profit housing associations have played a key role in Amsterdam’s urban development. During different time periods, housing associations built large numbers of affordable dwellings to accommodate lower and lower middle class residents, including entire neighbourhoods in the “Amsterdam School” architectural style during the interbellum period, and large modernist estates in the 1960s and 1970s (e.g. Bijlmermeer in the southeast). The 1970s and 1980s also saw housing associations and governments active in the urban renewal of inner city neighbourhoods, buying up private rental dwellings and replacing slum housing with new social rental housing complexes for the original residents, following the ideal of “building for the neighbourhood” (Uitermark 2009). As a consequence of this historical development path, Amsterdam represents a city where social rental housing has long been the dominant tenure form. Yet from the 1990s onwards, the promotion of homeownership has gained greater currency at the national level, at the cost of social renting. A key event in this regard was the cutting of financial ties between housing associations and governments in 1995, officially eradicating state support for social housing³ (Boelhouwer and Priemus 2014; Aalbers, van Loon, and Fernandez 2017). Housing associations did, however, remain important players in urban development, as is evidenced by their important role in cooperating in various urban renewal policies initiated since then (Uitermark 2014). Despite the push for homeownership and the reduced financial and political support for social rental housing, the share of social rental housing in Amsterdam remained rather stable during the late 1990s and peaked in 2002 at 205,301 dwellings (55.1% of the city’s total housing stock). However, since then the social rental stock has steadily decreased in size. In 2014 there were a total of 181,882 social rental dwellings, or 45.6% of the total stock. In contrast, the number of owner occupied dwellings increased from 42,199 in 1998 (11.6% of the total stock) to 113,694 in 2014 (28.5%) (Figure 2). To compare, at the national level – where policies promoting homeownership are also in place – the share of owner occupation stands at 57%, while social and private rent constitute 31 and 12% of the stock respectively (figures for 2012) (CBS 2014).

Figure 2. Tenure composition of the Amsterdam housing stock. Source: OIS Amsterdam; own adaptation.

Changing Instruments and Geography of Tenure Restructuring

The two dominant instruments used to reduce the size of the social rental stock are demolition and sales. In 1997, various stakeholders, including the Amsterdam housing associations and the municipal government, signed a first “Social Housing Sales Covenant” allowing housing associations to sell part of their property to individual households (Aalbers 2004). After a slow start, the number of yearly social housing sales quickly increased after 2002 (Figure 3), partly because sales became an explicit local policy goal. Only 434 dwellings were sold in 2002, but this number increased to 2402 sales in 2005. The onset of the global financial crisis in 2008 had a profound impact on the Dutch housing market, as the number of housing sales sharply decreased, as well as the average sales prices (Ronald and Dol 2011). Yet social housing sales quickly picked up again after 2010, showing substantial year-on-year increases, until reaching a peak of 2682 sales in 2014.⁴ Although the en bloc sale of multiple dwellings, for example post-renovation, also occurs, most sales are of individual dwellings in existing complexes. The vast majority of sales occurs after the previous tenants have vacated the dwelling: in 2015, only 11% of sales were to the sitting tenants (AFWC 2016).

Figure 3. Number of existing housing-association dwellings sold per year 1998–2015. Source: AFWC (2016); own adaptation.

In response to increasing concerns about the scarce availability of social rental housing and growing socio-spatial inequalities in the city (also see Gemeente Amsterdam 2013), the municipality, housing associations, and tenants organizations agreed upon the ambition to sell no more than 2000 dwellings per year for the 2015–2019 period, considerably below the 2014 figure (HV Amsterdam, AFWC, and Gemeente Amsterdam 2015). The number of sales did indeed decrease after 2014, although a total of 2337 sales were recorded in 2015, still considerably greater than the targeted maximum. Figure 3 also highlights how the geography of social housing sales has shifted in recent years. In the years prior to the global financial crisis, the number of sales was somewhat higher in the urban periphery than in the city’s central boroughs. However, from 2009 onward sales have increasingly concentrated in the city’s central boroughs. In 2014, 66% of social housing sales occurred in one of the central boroughs. While these spatial contrasts became less pronounced in 2015, still 59% of social housing sales occurred in the central city.

In contrast to growing social housing sales figures, large-scale urban renewal has become more difficult due to the decrease in allocated funds. The number of demolished social rental dwellings saw a steep decrease after the onset of the 2008 financial crisis: from 1814 in 2007 down to 622 in 2014, before again increasing to 1304 in 2015. The decrease in demolished dwellings is most notable in the urban periphery where urban renewal projects were previously typically concentrated (Figure 4). There is thus a general trend towards the increasing importance of the sale of dwellings vis-à-vis urban renewal in order to achieve tenure and social mixing, because state funding for urban renewal has become scarcer and ambitions to pursue such policies have been scaled down. Demolition did increase again in 2015 (although an important explanation for this is the planned removal of temporary student housing), which could indicate that demolition as part of urban renewal may pick up again with economic recovery. Nevertheless, the expectation is that the shift away from large-scale urban renewal is more structural.

Figure 4. Number of housing-association dwellings demolished per year 2007–2015. Source: AFWC (2016); own adaptation. Note: the steep increase in demolished dwellings in 2015 in central boroughs is mainly due to the removal of temporary student “container” housing (located in Houthavens).

So far, two clear and interrelated trends come to the fore: first, social rental housing sales have strongly increased, while demolition as part of urban renewal has decreased; second, tenure restructuring, which used to be concentrated in the city’s peripheral boroughs, is now increasingly concentrated in Amsterdam’s central boroughs. Before contextualizing these shifts, Table 1 and Figure 5 further unravel the changing geography of tenure restructuring. Table 1 describes the tenure composition of each neighbourhood type in 1999 and 2014. It shows that over this period, the share of social rental dwellings saw the strongest decrease in the post-war neighbourhoods (from 69% to 57%), but also decreased in the 1800–1920 and 1920–1940 belts (to 43 and 47% in 2014, respectively), the city’s current gentrification frontiers. Shifts in the share of social rental housing may, however, also be caused by new additions to the stock as part of new-build developments. Decreasing shares therefore do not necessarily reflect decreasing absolute numbers of social housing.

Table 1. Tenure composition per neighbourhood type in 1999 and 2014.

Neighbourhood type	1999				2014
	Owner occupied (%)	Social rent (%)	Private rent (%)	Total (N)	Owner occupied (%)	Social rent (%)	Private rent (%)	Total (N)
Central city, affluent south	23	32	45	50588	33	28	38	56271
1800–1920 belt	9	52	40	74511	27	43	30	75300
1920–1940 belt	8	54	39	94593	23	47	30	98399
Post-war 1945–1990	14	69	17	125537	27	57	15	131635
New build (>1990), rural	40	33	28	20950	45	33	22	35250
Amsterdam	14	54	32	366979	29	46	26	398565

Note: Percentages may not add up to 100% due to rounding.

Source: OIS Amsterdam.

Figure 5. Percentage change (compared to 1999; 2004; and 2009) in the number of housing association dwellings. Source: OIS, own adaptation.

It is thus imperative to also consider changes within the social rental stock. Figure 5 does so by breaking down the time span 1999–2014 into three even time periods, and analysing the percentage changes in the social rental stock per neighbourhood type per time period. These overall shifts are the product of both subtractions (sales, demolitions) and additions to the social rental stock. It shows that the total number of housing association dwellings slightly increased between 1999 and 2004, mainly due to the long-term requirement to include at least 30% social rental housing in new-build developments. In the post-war neighbourhoods, the number of social rental dwellings showed the strongest decrease, although this decrease still only stood at 1.8% between 1999 and 2004. Yet between 2004 and 2009, the social rental stock clearly decreased in size in the city as a whole (−4.3%). This decrease was especially marked in the city’s post-war neighbourhoods, where large-scale urban renewal was concentrated (−7.9%), as well as in the city’s nineteenth century belt (−8.3%). In other words, during this period, the decrease in social rent was concentrated in the city’s periphery where income downgrading took place, but also at the gentrification frontier.

The most notable spatial shift among the analysed time periods comes to the fore when comparing the 2004–2009 and 2009–2014 periods. The latter period saw a more intense decrease in the size of the social rental stock in Amsterdam overall (−5.3%). However, the decrease in post-war neighbourhoods slowed as a consequence of stagnating urban renewal. In contrast, particularly the 1920–1940 belt saw its social rental stock decrease at an increasing pace (−7.1%), while the decrease in the nineteenth century neighbourhoods remained more or less stable compared to 2004–2009 (−8%). Hence, during the last analysed time period, the decreases in the size of the social rental stock shifted particularly to these gentrifying belts. The size of the social rental stock has, however, remained relatively stable in the city’s most affluent neighbourhoods, namely the central city and affluent south. On the one hand, this reflects the already relatively small size of the social rental sector in these neighbourhoods. On the other hand, it also reflects the ambition to keep hold of these dwellings. These more fine-grained spatial analyses thus highlight that the decrease in social rental dwellings has not simply been focused on the city’s most up-market neighbourhoods where housing values are highest, but rather on the surrounding gentrification frontiers.

Current housing policies and tenure restructuring have had a particularly strong impact on the accessibility of this stock. Social housing sales, demolition, rent liberalization, and low residential mobility rates – itself a consequence of scarce housing availability, but also of the ageing of the social rental population – are all responsible for a marked 36% decrease in the number of new social housing allocations between 2007 and 2015: in 2007, 9517 rent-regulated dwellings were rented out to new tenants through the official allocation system; this decreased to 6050 in 2015 (Figure 6; AFWC 2016). Thus, although the social rental stock remains relatively large and overall shifts in tenure composition have been rather gradual (Figure 2), the accessibility of the social stock has diminished at a rapid pace. As a result, divides between housing market insiders (those who enjoy tenant protection) and outsiders (those struggling to find housing) are reinforced (Kadi and Musterd 2015). Interestingly, this decrease occurred rather evenly across the central and peripheral boroughs.

Figure 6. Number of new social-rental housing allocations per year in central and peripheral boroughs. Source: AFWC (2016), own adaptation. Notes: (1) These figures exclude student housing and direct allocations outside the central allocation system. (2) In some years a small number of allocations have an unknown location and are not included in this figure (highest number of unknown was 165 in 2013).

Contextualizing Shifting Patterns of Tenure Restructuring

To understand the two key interrelated trends of increasing social housing sales and the shift of focus towards Amsterdam’s inner neighbourhoods, they need to be situated within the broader policy context. This section pays particular attention to local urban and housing policy goals and rationales, as well as to the influence of the wider national and international political and economic context.

Previous urban housing policies were primarily focused on achieving a “balanced” social mix, not only by attracting more affluent households to disadvantaged neighbourhoods, but also by retaining upwardly mobile households by offering them the opportunity to progress along their housing career within the same neighbourhood (see Aalbers 2004). Amsterdam’s current housing policies are, on the other hand, increasingly focused on enhancing the housing opportunities of middle income households.⁵ This group earns too much for social housing but also struggles on the increasingly expensive and exclusionary Amsterdam owner occupied market. Because former social housing dwellings are often sold at comparatively low prices, they may be affordable and accessible to these middle income households. Housing associations therefore argue that the sale of social rental housing is important to offer middle income households, young people, and starters a place in the city (Woon Amsterdam 2015). In so doing, the sale of social housing is discursively recast as contributing to housing associations’ traditional role of providing housing to groups in need. In the same vein, the large body of existing social rental dwellings is cast as harmful to the housing position of various housing groups – most notably young and upwardly mobile households. Two aldermen of the local conservative-liberal VVD party, in the run-up to the 2014 municipal elections, echoed this position with the statement that the large social rental stock is responsible for keeping “talent outside the city walls” (Wiebes and van der Burg 2014; author translation; also see Hochstenbach 2015). Although this is a political statement, the official municipal housing memorandum (Gemeente Amsterdam 2009) speaks in a similar vein of an oversupply of affordable rental housing out of sync with the city’s actual population (see Uitermark 2009; Van Gent 2013 for critical reflections). Consequently, the sale of social housing is cast as the logical – or indeed social – way to go.

In addition to the sale of social rental dwellings, recent years have seen the introduction of several other policies that aim to enhance the housing position of middle income households. Most importantly, the local government has committed to increasing the number of rental dwellings in the rent-liberalized segment, where waiting lists and maximum income criteria are absent (D66, VVD, and SP 2014). This is done particularly to help young upwardly mobile households that earn too much for social rental housing but also struggle to buy. Especially since the global financial crisis, access to homeownership has decreased among younger middle income households due to stricter mortgage lending criteria and growing insecurities on the labour market (McKee 2012; Forrest and Hirayama 2015; Hochstenbach and Boterman 2015; Lennartz, Arundel, and Ronald 2016). The liberalization of dwellings owned by housing associations represents an important instrument for the expansion of the rent-liberalized sector. Current agreements stipulate that a maximum of 1000 dwellings per year may be “liberalized” (HV Amsterdam, AFWC, and Gemeente Amsterdam 2015). Although housing associations retain ownership, these dwellings essentially move from being rent-regulated and reserved for low income households to being of higher rent and aimed at higher income groups. While the number of rent-liberalized housing association dwellings stood at 3680 in 2008 – only 1.9% of the dwellings owned by housing associations – policy shifts enabled a rapid growth to 14,053 dwellings in 2016 (7.7% of the stock) (AFWC 2008, 2016).⁶ This shift may be considered a way of gearing the social rental stock towards serving other purposes or population groups.

A further reduction in the number of “regular” social rental dwellings is envisioned through the experimentation with new policies that will ultimately allow housing associations to rent out up to one-third of their stock to young people using temporary five-year contracts (HV Amsterdam, AFWC, and Gemeente Amsterdam 2016). The sale of social housing, rent liberalization, and the introduction of temporary contracts are all policy measures to enhance the housing opportunities of middle income, relatively young, and upwardly mobile households. This is partially the consequence of the national government’s enforcement of EU regulations in 2011, which stipulate that at least 90% of the social housing stock should be rented out to households earning below the €35,000 income limit, creating a housing shortage for those households earning more (Boelhouwer and Priemus 2014). While these policies may improve the housing opportunities of young households, they will also contribute to a residualization of the remaining regular stock with the consequence that housing opportunities for low income groups worsen (cf. Van Duijne and Ronald 2016).

Although these policy changes explain the increasing number of social housing sales over time and the concomitant rise of other policies such as rent liberalization to reduce the number of “regular” social rental dwellings, they do not explain the changing geography of tenure restructuring. The overall shift towards Amsterdam’s centrally located gentrifying neighbourhoods is the result of a combination of factors. Municipal housing policies aim not only to enhance the housing opportunities of middle income households, they also seek to meet their preferences. The municipality spends much resources on making the city more attractive to middle class households – especially the creative class that includes groups of “new urbanites” that value urban amenities – by spurring commercial and residential gentrification (see Van Gent 2013). This is considered essential for the city’s competitive position: policy-makers are concerned that not catering to these groups will contribute to them moving elsewhere. In the residential domain, this means accommodating middle class demand for housing in the gentrifying neighbourhoods surrounding the historical centre. Many of these middle income households are also either single-person households or childless couples. The many relatively small apartments located in the nineteenth and early twentieth century belts are thus considered particularly suitable to accommodate these household types. Because of their small size, these apartments are also relatively affordable to middle income households.

The gentrifying neighbourhoods located in the belts surrounding the city centre furthermore provide a specific spatial setting where burgeoning market demand is combined with the continuing presence of relatively large – but decreasing – shares of social rental housing. Selling off part of their stock in these neighbourhoods allows housing associations and local governments to combine the goals of attracting and accommodating middle class households with the aim of altering the population composition of often still low-income neighbourhoods. At least initially, the sale of social housing contributes to more social mixing in these neighbourhoods. This also constitutes an important reason why sales are concentrated in the gentrifying neighbourhoods rather than in the central city or affluent south, where shares of social housing are already comparatively low: despite financial incentives to sell the valuable stock located in the most expensive neighbourhoods, the local state and housing associations are committed to preserving centrally located social housing (HV Amsterdam, AFWC, and Gemeente Amsterdam 2016).

The changing geography also relates to financial rationales. Housing associations work with a revolving fund, which means that extra income generated from increased rents or from social housing sales should be used for social purposes such as maintaining the existing stock or new-build developments. However, consecutive national government coalitions have sought to roll back the operations of housing associations, with the 2013 “landlord levy” being a recent example (Boelhouwer and Priemus 2014). The levy is a state-imposed austerity measure that taxes housing associations in order to transfer increased income from rents, and is part of a broader policy goal to set housing associations on a more sober trajectory (Nieboer and Gruis 2014). Although the social housing stock is still rather large in the Netherlands, and even more so in Amsterdam, these new policies work towards the gradual residualization of the social housing stock (Van Duijne and Ronald 2016). Public spending on social housing is being trimmed and housing associations have to focus their activities on a more narrowly defined group of low income households (Van der Heijden 2002; Boelhouwer and Priemus 2014). These policies are partly the consequence of a longstanding ideological push towards homeownership (Ronald 2008) and crisis-related austerity measures (Boelhouwer and Priemus 2014). However, they are also the product of a loss of public support for housing associations. In the years before and after the onset of the 2008 global financial crisis, the public legitimacy of housing associations suffered several blows due to a range controversies. Examples include high salaries for managers, high profile cases of fraud, failed flagship developments, and perhaps most significantly the heavy losses suffered by the housing association Vestia due to speculation with derivatives (Aalbers, van Loon, and Fernandez 2017). Not only did these cases directly impact the financial position of housing associations – Vestia ultimately had to be bailed out for €2 billion, partly paid for by other housing associations (Aalbers, van Loon, and Fernandez 2017) – but they also fostered public support to reduce the role of housing associations in general.

Importantly, these trends have not only led to a decline in the social rental stock, but have also had particular spatial consequences, as described in the empirical sections above. It has become increasingly difficult for housing associations to engage in large-scale urban renewal projects, which tend to concentrate in the urban periphery. This is partly as a result of the financial crisis and related austerity measures, but also due to government policies that seek to restrict the role of housing associations. At the same time as housing associations have become financially more restricted due to government policies and financial mismanagement, they may become more inclined or be forced to sell off the more valuable property they possess. This could spur the further reduction of social rental housing in gentrifying locations, where selling becomes ever more profitable.

Conclusion

This paper has shown that questions of how and where local states pursue policies of tenure restructuring are closely interrelated, as well as how these questions are influenced by their broader societal and policy contexts. The case of Amsterdam documents various shifts in urban housing policies. The demise of social rent has accelerated under conditions of market-oriented housing restructuring. Furthermore, while policies used to focus on urban renewal and the demolition of social rental dwellings, they now primarily operate through the sale of existing social rental dwellings. The global financial crisis has spurred this shift, although it had been set in motion beforehand. Market-oriented restructuring has also brought about a shift as current housing interventions focus to a greater extent on centrally located high demand neighbourhoods. These shifts demonstrate the contextual and temporal contingency of state-led gentrification within a context of welfare state and housing restructuring.

Particularly in the Dutch welfare state, tenure mixing policies have traditionally formed an important means to disperse poverty concentrations in order to mitigate potential negative neighbourhood effects (Galster 2012) and to manage disadvantaged neighbourhoods (Uitermark, Duyvendak, and Kleinhans 2007). Yet, austerity and the transition towards financially more restricted housing associations have made it more difficult to engage in costly urban renewal to alter the tenure and social mix of disadvantaged neighbourhoods (cf. Uitermark, Hochstenbach, and van Gent, 2017). Selling social housing then emerges as a more affordable alternative, but also serves other goals such as expanding the housing stock available and attractive to middle class households through the sale and liberalization of social rental housing. Furthermore, while urban renewal policies were typically aimed at sparking gentrification processes in neighbourhoods of low and declining socio-economic status (see Uitermark, Duyvendak, and Kleinhans 2007; Aalbers 2011), current policies are increasingly geared towards accelerating gentrification processes in neighbourhoods of burgeoning demand. This has important theoretical implications: it demonstrates that welfare state restructuring and changing urban housing policies influence the form, intensity and geography of state-led gentrification.

In recent decades, Western welfare states have gone through successive waves of market-oriented restructuring. Rather than replacing state regulation with market forces, restructuring typically entails the reorientation of state resources to promote private property and private accumulation (Peck and Tickell 2002; Brenner, Peck, and Theodore 2010; Aalbers and Christophers 2014). Likewise, Amsterdam’s tenure mixing policies are increasingly imbricated with “market logics”. However, the local government and not-for-profit housing associations have not simply become the agents of private capital. Despite rapidly increasing sales, the state continues to play a key role in determining where and how many units may be sold. Mounting concerns about housing affordability and socio-spatial inequalities re-open public and political debate about the desirability and necessity of selling social housing. Furthermore, sales do not concentrate in the city’s most expensive areas but instead are focused on gentrifying areas where market-oriented restructuring can still be combined with the goal of social mixing. Hence this paper suggests that gentrification is not simply unleashed, but that market-oriented restructuring efforts are, to some extent, balanced with policies of social mixing.

Nevertheless, by selling social rental housing in these high demand neighbourhoods, existing barriers to gentrification are removed here. The gradual decline of Amsterdam’s social rental stock has distinct spatial outcomes, as the effects of tenure restructuring are increasingly felt in the gentrifying neighbourhoods. Consequently, access to these areas of privilege becomes increasingly reserved to those who possess sufficient financial resources. Although Amsterdam remains to date a rather mixed city, current housing policies open up space for increasing socio-economic inequalities, socio-spatial divisions, and the segmentation of tenure forms serving different strands of the city’s population. Housing exclusion and high rent burdens therefore become more pressing issues as low income households are confronted with a lack of affordable housing options. These developments highlight how market-oriented housing policies can gradually take hold and forge greater socio-spatial divides.

Acknowledgements

I would like to thank Jannes van Loon, Wouter van Gent, Sako Musterd, Richard Ronald, Justin Kadi and Jeroen van der Veer for feedback on previous drafts. Also thanks to two anonymous reviewers for their useful comments. Data in this paper are from the AFWC (Amsterdam Federation of Housing Associations) and OIS Amsterdam (Research, Information and Statistics Amsterdam).

Notes

1. I use the term “tenure restructuring” here to denote all policies that seek to change the tenure composition of certain areas.

2. The point system also applies to private landlords, and means that part of the private rental stock is also rent-regulated. This paper focuses, however, on shifts in the social housing stock owned by housing associations.

3. State support continues in other ways though: housing associations for instance continue to be able to borrow at favourable interest rates from state banks (Aalbers, van Loon, and Fernandez 2015).

4. To compare, the total number of all sales in Amsterdam (not just social housing sales) only started to increase after 2013.

5. A uniform definition of what constitutes a middle income household does not exist. Here, it indicates a household that earns more than the maximum income set for eligibility for social housing (approximately €35,000). The maximum (gross household) income to qualify as middle income is often set at €43,000 or €50,000.

6. This increase is not only the product of rent liberalization, but also of new-build developments and of incremental rent increases for incumbent tenants.