Housing,urban growth and inequalities: The limits to deregulation and upzoning in reducing economic and spatial inequality

Introduction: Housing is no longer a local issue

Housing market failures can imperil local economic growth and generate problems such as segregation, long commute times, deteriorating quality of life, homelessness and barriers to social mobility for certain populations. In recent years these worries have increased in the metropolitan areas of many countries. In developed countries, housing regulation through planning and zoning has long been considered primarily the domain of local policy, with national policy in a supporting and guiding role. This notion is now changing because, in the mainstream view, sharply rising housing prices and declining affordability in metropolitan areas are the result of overly-restrictive zoning and onerous regulation of construction. Restrictive local housing policies are increasingly seen as central to the magnification of social and spatial inequalities at the national scale. This connection emerges, it is argued, because local housing policies create barriers to people from less dynamic regions moving to more dynamic ones. Housing, in this view, is no longer a local issue: it is central to debates about national growth and the effects of globalisation on communities, and can become a source of populist anger from those who are locked out of prosperous metropolitan regions.

In this article, while agreeing that housing policy can have important impacts on economic growth and social equity, we will critique this mainstream academic view of how housing fits into the broader picture of economic growth and inequalities in the age of globalisation. Specifically, we suggest that housing supply is more important by city scale (among neighbourhoods) than it is to shaping inter-regional migration, the size of cities or the performance of national economies. The barriers to migration to prosperous economic areas consist less of housing and much more of the skill composition of urban labour demand. The affordability crisis within major urban areas is real, but it is due less to the over-regulation of housing markets than to the underlying wage and income inequalities, and to a sharp increase in the value of central locations within metro areas, as employment and amenities concentrate in these places. We posit that this school of thought is simultaneously diverting attention from the real problems of lagging regions, while overestimating the potential demographic, income distribution and productivity effects of less restrictive zoning on prosperous regions.

The ‘housing-as-opportunity’ view of inter-regional inequality

Since the beginning of the current round of globalisation in the 1970s, two types of inequality – inter-personal and geographical – have risen (Milanovic, 2016). But while inter-personal inequalities have attracted the most attention (e.g. Piketty, 2014), the parallel rise of inter-regional inequality has remained somewhat under the radar. Yet, it is becoming clear that regional inequalities have also turned a corner. In the US, the inequality of income per person among metropolitan areas was 30% higher in 2016 than it was in 1980 (Ganong and Shoag, 2017). In Europe and in a process driven by fast-growing capital regions, a long period of regional convergence dating back to 1900 has been replaced since 1980 by divergence (Rosés and Wolf, 2018).

A number of recent changes are behind the widening of inter-territorial inequality. First, inter-regional migration is declining: in the United States it has fallen to half of its century-long average up to 1980, and it is more spatially selective by skill level (Giannone, 2017). Second, labour force participation rates also have a higher inter-regional variance in the EU and the US than since the 1930s Great Depression. Moreover, the jobs generated in lagging-behind and declining areas have lower average skill levels and wages than those created in more prosperous metropolitan areas (Di Cataldo and Rodríguez-Pose, 2017). Generally, the employment relocating from core areas to other regions is in more routine occupations, with lower wage levels (Autor and Dorn, 2013). Moreover, technology (automation) is suppressing some routine activities, and import competition keeps unskilled wages down (Autor et al., 2015). Though there are some bright signs in lagging-behind regions, such as the expansion of warehousing and logistics jobs linked to the internet economy, wage and skill levels are stagnating and employment is being challenged by robotisation. Overall, the divergent new geography of income and jobs is also becoming a divergent new geography of opportunities (Storper, 2018).

What can be done to reverse the increasing polarisation of income, jobs and opportunity? Although multiple solutions have been proposed since the 1950s, as the Economist (2016) put it, ‘orthodox economics has few answers to the problem of regional inequality’. In recent years, however, one position deeply embedded in urban economics has also become dominant: the only place-based policy that stands a chance of making a difference involves lifting the barriers to migration from lagging regions to leading metropolitan areas, as the way out of the current predicament of divergence and inequality.

The barrier that must be lifted in order to make this happen is, according to this view, insufficient housing construction in prosperous areas due to local restrictive zoning in those regions. The places where policy is needed are therefore not the lagging and falling-behind regions, but the prosperous areas, which are perceived as afflicted by the disease of NIMBY-ism (Not-in-My-Back-Yard). Undoing NIMBY-ism would allow people from other regions, who are deemed excluded by high housing prices and low availability in prosperous places, to move to prosperity (thus, a place-based policy leads to a people-based outcome).

Along these lines, a host of academic papers (e.g. Ganong and Shoag, 2017; Gaubert, 2018; Glaeser and Gottlieb, 2008; Hsieh and Moretti, 2015, 2017; Ihlanfeldt, 2007; Katz and Rosen, 1987; Kline and Moretti, 2014; Quigley and Raphael, 2005; Saiz, 2010) have made a set of inter-linked claims:

Taken together, these claims amount to an ambitious and comprehensive vision of how the space-economy works to link the local and the far away. They form the basis of what is a growing mainstream policy consensus that centres on reducing housing restrictiveness in prosperous areas, asserting that massively expanding the land zoned for housing and its permissible densities is a meaningful instrument to confront and change the current inter-regional divergence of incomes, employment and opportunity, to reduce income inequality and increase housing access for low-income people in prosperous metropolitan regions.

We call this perspective the ‘housing as opportunity’ policy school of thought. This perspective has come to dominate academia and captured the public imagination with its claims about the benefits of housing deregulation to prosperous and less prosperous areas alike and to the national economy as a whole. Hsieh and Moretti (2017), in estimations we shall criticise below, assert that the US economy would be far bigger if housing were unregulated and that the larger agglomerations would increase in size, as the less-skilled currently languishing in stagnating regions would migrate to them. Hence, inter-personal and inter-territorial inequality would decline, in a win-win scenario of rising productivity and prosperity for all. Ganong and Shoag (2017) claim that inter-state income convergence would have been at least 10% greater since 1980 had it not been for restrictive housing regulations in prosperous areas. Glaeser (2017) endorses their claims that ‘America’s most important, hence costly regulations, are land use controls’, and the New York Times editorial board and its progressive columnists, such as Paul Krugman, have backed the view that NIMBY-ism, in the form of neighbourhood housing regulations, is strangling national economic growth, by making prosperous metro areas like citadels unassailable to those with lower skills and education. ¹

This is not without consequences. As indicated by Romem (2016: 1), ‘cities confronting growth pressure face a trade-off between accommodating growth through outward expansion, or accepting the social implications of failing to build enough new housing’. Thus, local housing and zoning regulations have become not just a local, but a national planning issue, and not just in the United States. They have long been the concern of a national policy debate in the UK. Interestingly, the academic consensus unites, on the one hand, groups that emphasise the supposed social justice aspects of reducing housing regulation, assuming it to help the less-skilled and reverse a long association of zoning with racial exclusion, and, on the other, mainstream economists, who posit that regulation is inefficient. As can be imagined, both are applauded by land developers, who are strong supporters of the California-based YIMBY (Yes-in-My-Backyard) movement. ²

A corollary to their emphasis on housing in prosperous regions is a systematic rejection of place-based policies in less prosperous regions. ³ In their view, large and densely populated cities are the only possible motors of economic activity, and promoting economic activity in less-developed areas through public intervention leads, at best, to market distortions and, at worst, to a waste of public resources on ‘troubled areas’ (Glaeser and Gottlieb, 2009: 1014). Furthermore, because promoting development in lagging areas implies ‘severe equity efficiency trade-offs’ (Kline and Moretti, 2014: 656), ‘subsidizing poor or unproductive places is an imperfect way of transferring resources to poor people’ (Kline and Moretti, 2014: 656).

In their thinking, the alternative –‘place-based policies’ for lagging-behind areas – is a suboptimal solution, unlikely to have any discernible economic impact (Glaeser and Gottlieb, 2008; Leunig and Swaffield, 2007). ‘Local economic policies that are meant to increase production in a particular area […] seem to be either extremely expensive or ineffective’ (Glaeser and Gottlieb, 2008: 203). Moreover, the ‘mobility of people and capital can complicate the effects and potentially undo most or all of the gains from such redistributive policies’ (Neumark and Simpson, 2014: 12).

Our purpose in this article is to scrutinise this mainstream view about housing construction in prosperous areas as a route to greater prosperity and equality. Our point of departure is that housing markets are not like standard markets: aggregate increases in supply do not necessarily translate in any straightforward way into decreases in price, because the internal plumbing of housing markets – succession, migration and occupation patterns – is full of frictions, sunk costs, barriers and externalities that make the effects of aggregate supply increases highly uneven and, in many cases, involve unintended or contradictory effects. From this point of departure, our critique argues that the first three claims of the mainstream view above are reasonable but require considerable nuance, while the remaining four are implausible. Through our critique of these claims, we conclude that the housing as opportunity school is diverting attention away from the tasks necessary to address the problems of lagging regions and inter-regional inequality. It also exaggerates the effect of housing in contributing to the overall rise of inter-personal inequality and socio-spatial segregation. Finally, by implication it diverts attention away from the real need to address housing affordability for low- and moderate-income groups already residing in the prosperous metropolitan regions.

Does economic growth come mostly from city size or from urban specialisation?

The housing as opportunity school sees housing as the key element in determining inter-regional population mobility and the geographical pattern of real incomes. City size and density are crucial for economic growth. The corollary assumption is that restricting immigration limits productivity growth, by preventing the unskilled from matching to better job opportunities that supposedly exist for them in prosperous city-regions. The basic set-up here derives from spatial equilibrium theory, which holds that city size and population growth are the only important factors for economic growth, because once the conditions are in place for population growth, jobs and output growth will follow (Glaeser, 2008). This model explicitly rejects using income, per capita income or the wage structure as measures of urban performance. Therefore, we need to scrutinise the relationship between housing and land supply and urban population growth.

However, the connection between city size, urban population growth and economic growth is far from straightforward (e.g. Frick and Rodríguez-Pose, 2018; Polèse, 2005). Even in the case of the US, the country that has traditionally been used to prove this connection, economic polarisation has coincided with the absence of a close link between city size and economic growth. Figure 1 displays the link between the size of US metropolitan statistical areas (MSAs) and their economic performance per capita between 2001 and 2016. Taken as a whole, the relationship is, at best, tenuous. While some large MSAs such as Portland, Los Angeles, San Francisco, Boston or Seattle have done well, economic growth per capita in other large agglomerations during the same period, including Atlanta, Phoenix and Las Vegas, has been negative. The link between initial city population and economic growth during the same period is non-existent (Figure 1).

OPEN IN VIEWER

Figure 1.

The relationship between city size and per capita economic growth in the US, 2001–2016.

The same applies to population growth. Although some cities with more unregulated housing markets have witnessed greater population growth (Figure A1), this has not necessarily been translated into economic growth (Figure 2). Some cities with relatively unregulated housing markets, such as San Antonio, Dallas-Fort Worth and Houston, have experienced high levels of both population and economic growth. However, rapid population expansion has not resulted in economic growth in Las Vegas, Orlando, Phoenix or Atlanta. In contrast, many cities with highly regulated housing markets, such as New York, Boston, Portland and San Francisco, have enjoyed high absolute levels of economic and population growth in the past few decades (with the exception of Portland’s population growth). Overall, since the turn of the century there has been no connection between population change and economic growth across US cities (Figure 2).

OPEN IN VIEWER

Figure 2.

The link between population growth and the growth of GDP per capita in US MSAs, 2000–2016.

For present purposes, this means that, even if substantial deregulation of housing markets were to reshape migration and population distribution, more national economic growth would not necessarily follow automatically. This is because urban productivity and incomes appear to rise up to a certain point, but are also shaped by what a city specialises in and by the absolute size of a particular specialised agglomeration of firms (Kemeny and Storper, 2014). Thus, the US has a bigger urban productivity surplus than does Europe, and Europe has many more middle-sized cities than the US, but this difference is less due to the preponderance of larger urban areas in the US and more to their greater specialisation, an outcome of the lower barriers to trade within the US urban system. Not enough is currently known about the relative contributions of size and specialisation to incomes and productivity, but it is a great leap to predict unlimited positive relationships of the latter to metropolitan size.

Is inter-regional migration shaped principally by housing prices?

Mainstream theories have used low levels of housing regulation in the Southeast and Southwest of the US as an explanation for population growth in these areas (Glaeser, 2008; Graves, 1983; Roback, 1982). Most such models assume that housing and amenities rank highly on preference functions, via a further assumption that ‘jobs follow people’ (Muth, 1971). Their canonical image is the US Sunbelt development of the post-war period, which involved high domestic migration (rural to urban within the South; and Industrial Midwest/Northeast to South and West). But the assumption that these migration streams were motivated by cheaper housing in the developing areas finds no historical proof in that literature.

A more plausible explanation is that such migration was unleashed by de-agglomeration of routine manufacturing from its Northeastern-Midwestern heartland, combined with job-market deregulation in the form of the Taft-Hartley Amendments to the National Labour Relations Act in the 1940s, which made unionisation more difficult in the resulting ‘right-to-work’ states. This prolonged the cheapness of labour during the rural-to-urban transition in the post-war period. Hence, an explanation for the growth of these regions starts with the movement of jobs, fuelled by a deregulated labour market, rather than with unregulated housing markets. Massive migration to California, by contrast, clearly was not due to cheap housing, as Californian metropolitan housing prices remained well above national averages for the best part of a century.

Deepening a perspective based on employment as the key factor behind urban growth and decline, from the late 1950s and through the 1970s unemployment increased rapidly in cities in the Northeast and Midwest, as manufacturing went through a three-phased process of de-agglomeration to the South, technological change and, finally, globalisation. This is what is vernacularly known as ‘the new geography of jobs’ and the ‘great inversion’ (Moretti, 2012). It was not highly-regulated housing markets that led to population loss in these cities and regions, but employment decline. Later, with the advent of the new economy, selective parts of the old economy, such as Boston, Washington and New York (as well as London, UK), reinvented themselves and rebounded from population decline, again in spite of highly-regulated housing markets.

The difference between the two cases of population change is the type of jobs and the point in industrial maturity that generated them (Norton and Rees, 1979). In the rapidly-growing cities of the American Sunbelt, average skill and wage levels have for decades been lower than in cities such as Washington, Seattle and San Francisco. One of the forces behind the rising gap in housing prices between high-cost – e.g. New York or San Francisco – and low-cost – e.g. Orlando or Phoenix – regions is the widening differences in the wages and wealth of those seeking housing (Romem, 2018). Hence, differences in housing prices are not uniquely determined by the level of in-migration (aggregate demand) but also by its composition. ‘Composition’ here refers to the wage and income structure of the population. In areas that have grown principally due to jobs with routine skills and moderate wages, housing prices are lower than in metropolitan areas that draw in the highly-skilled and highly-paid. A different case is weak aggregate demand, explaining why housing is also inexpensive in most middle-sized cities of the Rustbelt – Buffalo, Milwaukee, South Bend, Syracuse – which people leave in spite of low housing costs. Lack of jobs and a weak geography of opportunity are the main culprits. In growing Sunbelt cities, by contrast, employment growth has taken place mainly in middle- to low-wage jobs. This trend is so strong that, as Autor (2019) puts it, ‘the economic advantages of dense cities are disappearing for low-skilled workers’. In 1950, denser urban areas offered higher wages for both educated and less educated workers. Today, when wages are adjusted for density, workers without a college degree draw little advantage from locating in large cities. Similar trends are being uncovered in Europe. Bjerke and Mellander (2019) find that moving from a rural to an urban area in Sweden has no positive effects on the movers’ salaries, with the only exception of the highly-skilled. Though housing costs in dense areas compound the disadvantage to low-skilled workers, reducing housing costs would not, under any scenario, erase the basic facts of the labour market in dense urban areas for these workers, which stem from fundamental changes in economic geography, as we shall argue later in this article.

A more recent generation of models in the housing literature centres on the geography of employment. Gaubert (2018) argues that firm sorting and housing are strongly tied. In her model, cities are undersized today because firms cannot capture all the potential gains to workers’ productivity from potential agglomeration economies. In her view, wages in large cities are inflated by excessive housing costs due to regulation. Thus, if housing supply increased, this would flatten the inter-regional wage curve, attracting more firms into big cities and, consequently, endogenously creating more agglomeration externalities. The key mechanism would be increasing the skilled labour supply, resulting in bigger and more productive and specialised cities. This model affirms our view that urban growth – whether induced by greater housing supply or other factors – primarily involves skilled workers enjoying high urban wage premiums (Autor, 2019; Bjerke and Mellander, 2019). Figure 3 buttresses our view that it is the fundamentals of economic geography more than house prices that are at work, by showing graphically the weak relationship between changes in home values, expansion of the developed residential area and the presence of immigrants in US cities.

OPEN IN VIEWER

Figure 3.

Urban land area development, house prices and in-migration in the largest metropolitan areas of the US (1990–2017).

The relationship between residential expansion and low housing value appreciation emphasised in the housing as opportunity literature appears only in a small number of Southern cities, highlighted in Figure 3 by a black square. In Charlotte, Raleigh, Nashville, Atlanta, Jacksonville, Las Vegas and Orlando, a rapid increase in the developed land area between 1990 and 2010 has indeed come with moderate increases in house prices. But affordable housing has hardly been a magnet for immigrants, as the number of immigrants from outside the US in these cities remains low in comparison with the rest of urban America. On the whole, these cities are outliers, rather than the norm.

The norm, in fact, does not exist, reflecting that there is much more than housing regulation driving the relationship between housing expansion, affordability, mobility and urban growth. A number of cities with stringent zoning and high house price increases (those denoted by a red star in Figure 3) have continued to attract population and large numbers of immigrants. Miami, with almost 40%, tops the rank of metropolitan areas with more foreign immigrants. Los Angeles comes second; San Francisco is fifth; San Diego and New York are not far behind. Another group of US cities – including Indianapolis, Cincinnati Greensboro, Columbus, Louisville and Kansas City, indicated by a green diamond – have experienced rapid housing growth and, while prices have remained low, they have failed to attract immigrants. In other expansive cities – Austin, Phoenix, Salt Lake City, Denver, Dallas, San Antonio and Tampa (denoted by a yellow pentagon) – rapid housing growth has not led to greater affordability. While the share of immigrants is greater in these cities than in areas with more affordable housing, as a magnet for migrants they lag behind the more expensive places where the housing stock has hardly grown (red star in Figure 3). Finally, cities like Cleveland, St. Louis, Pittsburgh, Milwaukee, Detroit and Hartford (the blue circle in Figure 3) have neither expanded nor witnessed strong increases in house prices nor (with the partial exception of Hartford) attracted migrants. Their economic performance (with the exception of Pittsburgh) has also been below par.

Figure 4 shows a positive relationship between house price growth and population growth, though with considerable dispersion at the level of individual metropolitan areas.

OPEN IN VIEWER

Figure 4.

House price vs. population growth, %, 1990–2010.

Supply adjustments can be made by using new ‘develop-able’ land or by changing housing stock on existing land (for example, through in-fill or greater density). In Figure 5, a strongly positive relationship is observed between population growth and expansion of a developed area, consistent with the way that many rapidly growing American metro areas expand on their periphery.

OPEN IN VIEWER

Figure 5.

Population growth vs. increase in developable area 1990–2010.

But once we consider the connection between ‘develop-able’ areas and house prices, as in Figure 6, there is no relationship. Rapidly developing urban areas expanding outward on their urban fringe may benefit from low land prices on the develop-able fringe, which in turn lowers their average housing prices. However, we cannot capture this effect in declining urban areas (such as Rustbelt cities), which have plenty of vacant land available. In any event, in large and mature urban areas, the metropolitan fringe is already far away from the core and long occupied (and sometimes has hit natural geographical barriers), raising commuting times. That is why policies to increase supply are directed to already-developed land in the urban core, with its strong structurally high land prices, in addition to the declared policy goals of developing near public transit. Why are aggregate supply changes in this type of metropolitan core area unlikely to reduce overall housing costs through social and spatial trickle-down effects?

OPEN IN VIEWER

Figure 6.

House price growth vs. increase in developed land area, %, 1990–2010.

In what follows, we argue that the missing element in determining housing prices and affordability in these cities is the structure of jobs and incomes, not aggregate supply policies.

Incomes and urban size drive housing prices

The difference between expensive and expansive urban areas is the income and wealth underlying the structure of housing demand. The share of very high-income households in the San Francisco Bay Area increased from 17% to 27% of the total in the 2001–2013 period (Bronstein, 2017). This was driven by the concentration of high-skill, high-wage employment in agglomerated core industries of the Third Industrial Revolution. ⁴ Real incomes in high-wage and high-amenity metropolitan areas, after accounting for housing costs, are on average 15% higher than in lower-wage metropolitan areas (Kemeny and Storper, 2012). These prosperous regions generally have high levels of income inequality, resulting from a growing gap between the wages of the high- and the low-skilled (Baum-Snow et al., 2017). Low-skilled jobs are largely filled by international migrants, because low-skilled domestic workers have largely stopped migrating to these areas. Foreign migrants have a variety of housing strategies, ranging from high densities, overcrowding and substandard conditions, to long-distance commutes. In spite of these supposed barriers, both unskilled and skilled workers keep on moving to high-income cities (Lindley and Machin, 2004). Another element of population growth in large agglomerations is the young. The young are not yet at the top of the skill–wage hierarchy, but are willing to put up with difficult conditions in the short run to build up their experience on the job escalator (De la Roca and Puga, 2017; Glaeser and Maré, 2001; Jayet, 1983).

Let us return to the case of the less-skilled domestic worker population, whose inter-regional mobility is said by the mainstream housing view to be impeded by housing costs. Autor (2019) reveals that the hourly wages of less-skilled adults in the US, which formerly rose steeply with density, no longer do so, whereas the hourly wages of the skilled are ever more strongly related to density. This has contributed to a widening rural–urban divide in skills: the share of working-age population with a college degree is now 20 percentage points higher in urban than in rural places. In 1970 that gap was just five percentage points. Several decades ago, mid-skilled work was clustered in big cities, while low-skilled work was most prevalent in the countryside. No longer; the mid-skilled jobs that remain are more likely to be found in rural areas than in urban ones.

The wave of inter-regional divergence in the location of different types of job since the 1980s (innovative/agglomerated/non-routine versus routine) has affected the geography of housing prices. Formally, this is modelled within the New Economic Geography as a spatial split between the prosperous metropolitan areas, which agglomerate innovative high-wage industries scarcely affected by spatial competition, and other territories whose industrial mix is dominated by activities that are strongly tradeable, involve routinised work and are subject to global competition (Venables, 2018).

The agglomeration effect in prosperous metro areas, such as Boston, San Francisco, New York or most large European cities, involves an intra-metropolitan dimension as well: housing preferences are increasingly geared towards access to centrally-located urban amenities, transportation and employment. This phenomenon has driven up housing prices in the centre of prosperous metropolitan regions, such that distance from urban centres imposes an increasing penalty on house prices (Partridge et al., 2009). This is an inversion from patterns in the mid- to late 20th century, when skilled employment agglomeration forces were weaker. The demand for more central locations is, by contrast, weaker in successful Sunbelt metropolitan regions, such as Atlanta or Houston, with weaker residential urban cores. Moreover, skilled workers today are no longer moving to the suburbs at the same point in the life cycle as in previous generations. According to Autor (2019), there has been a 50–75% decline in the outmigration rate of prime age adults since the 1990s. This may be due to the longer and steeper opportunity ladders in cities today (De la Roca and Puga, 2017), as well as to the higher time costs of commuting in major metropolitan areas. The result is greater competition for inner metropolitan locations than hitherto, reinforcing the notion that less restrictive zoning may gentrify inner metropolitan areas but do little about housing affordability for the less-skilled.

The domestic migration slowdown: Housing or skills? Kept out or trapped outside?

Inter-regional migration in the US – usually taken as the canonical case of a geographically-fluid system of inter-regional population adjustment – has declined to half the average level that prevailed between 1880 and 1980, and has remained low since (Goetz et al., 2017). A greater share of the population is spatially ‘trapped’, suffering from barriers to mobility to opportunity. But to what extent is this a consequence of planning restrictions and lack of affordable housing, as opposed to the nature of employment opportunities and the skills required to access them, in dynamic cities?

Skill-biased technical change has a distinctive geography, consisting of the concentration of skilled jobs in certain regions, mainly large – but not always the largest – metropolitan areas. Part of this new geography reflects the increasing divergence in the returns to education (Giannone, 2017) and the changing nature of skills. Individuals thus increasingly face multiple challenges in the new economy: paying for the education to obtain entry-level formal skills and accessing the jobs that provide experience by mastering the soft codes and conventions that get them into networks and allow them to acquire the right kinds of skills (DeLong, 2016).

Under these circumstances, moving to big cities provides no immediate benefits for workers without a college education (Autor, 2019). While building more affordable housing in core agglomerations would accommodate more people, the collapse of the urban wage premium for less-educated workers means that the extra housing would mostly attract additional skilled workers. Consequently, as the prospects for improving wages in core areas are poor and the opportunity ladder has shrunk, the choice for low-skilled workers to stay put is rational (Autor, 2019). In brief, the decline in inter-regional migration has multiple sources, including the new geography of skills and wages, ageing, the changing nature of skills, social networks, negative housing equity for some and – far down the list of causes – housing restrictions in prosperous areas.

In this light, urban economic models emphasising the role of housing supply in inducing or preventing inter-regional mobility make unrealistic assumptions about migration. Three main points about inter-regional migration need to be stressed: it is not costless; housing markets have more of an income distribution effect than a migration effect; and the political influences on both the labour supply (migration) and labour demand side (housing regulation) are extremely complex and not amenable to the simple aphorisms of many urban economic models. We discuss each of these in turn.

Many migration arguments of the housing as opportunity school start with the decline in aggregate inter-regional migration. But migration is still happening. It has just become more selective and spatially separated by skill. Skilled individuals continue to migrate to the most dynamic places and use them as ‘escalator’ regions (De la Roca and Puga, 2017; Fielding, 1992). This is happening everywhere in the developed world. University graduates from the North of England flock to London and the South-East after graduation, regardless of whether they studied in northern or southern universities (Faggian and McCann, 2008, 2009). Similar processes are in evidence in Italy (Biagi et al., 2011), Sweden (Eriksson and Rodríguez-Pose, 2017) and Australia (Corcoran et al., 2010). And the drive towards opportunity is not limited to the highly-skilled. In the US, the skilled move between skilled cities (Diamond, 2016; Giannone, 2017). This is essentially ‘brain exchange’, and is different from the classical mass migration of manual workers to industrial cities of the mid-20th century, as in the Sunbelt migration in the US or the 1960s movement from the Italian Mezzogiorno to Lombardy, Piedmont, Switzerland and Germany.

In Europe, low-skilled migrants continue to move in large numbers. One-third of Romanians between the ages of 25 and 35 live outside Romania (World Bank, 2017). Lithuanians and Poles in the UK mostly conduct low-skilled activities in London, regardless of their level of education (Parutis, 2014). This may reflect the fact that European border-free movement is more recent than in the US, where the Sunbelt migrations of the post-war period already resettled many low-skilled migrants. In both Europe and the US, those not moving are those who either cannot move – because of the growing skill divide between large cities, on the one hand, and towns and rural areas, on the other – or do not want to move.

There are variations on this theme for each country. For example, in East Germany there is a significant skill advantage of young migrants – mainly women – over stayers (Hunt, 2006: 1032). The migration difference between the skilled and unskilled young is reproduced in the UK (Faggian and McCann, 2009) and Sweden (Eriksson and Rodríguez-Pose, 2017). In addition to the young who have failed to acquire new economy skills, many non-migrants are older, including those who never migrated from traditional industrial areas or did so during the mid-20th century industrial de-agglomeration wave. But that was a generation ago.

The current domestic migration slow-down affects also middle-aged professionals. People in this group often started their careers on the experience ‘escalator’ of the city and moved back to medium-sized cities to cash in on their acquired experience (De la Roca and Puga, 2017; Eriksson and Rodríguez-Pose, 2017). They moved ‘back’ in search of a better quality of life for their families, a different set of amenities, more security and lower housing costs (Whisler et al., 2008).

Hence, large groups are caught in a ‘spatial trap’ that prevents them from moving to more dynamic areas. Life cycles, strong family ties, emotional and material attachment to place and lack of employment opportunities in more dynamic areas for less-skilled and/or older workers limit the propensity of people in lagging-behind and declining cities and regions in the developed world to migrate.

Lack of affordable housing in metro areas may play a role in all of this, as in extreme cases of negative equity (housing bubbles in certain areas, long-term depopulation in others). But its influence will be small. The housing as opportunity school has traditionally assumed that migration is costless – or, in the words of Glaeser and Gottlieb (2008: 159), that ‘migration is cheap enough to make consumers indifferent’– but the reality is that for those in a ‘spatial trap’, migration is anything but costless and is not a realistic and/or viable option. Hence, neither subsidised housing, nor any reasonably imaginable price effect of supply changes induced by less restrictive zoning, would overcome the skills and equity barriers or the differences in perceptions about opportunity that these populations face in the new economy.

Data and measurement: An air of unreality

Until recently, most of the papers on housing, migration and economic performance have avoided offering counter-factual scenarios for population distributions, the size of metropolitan areas and employment levels of the less-skilled that would come about in a world of re-formed housing policy, focusing on the housing price effect (Glaeser and Ward, 2009; Ihlanfeldt, 2007; Quigley and Raphael, 2005; Saiz, 2010). Those papers have mostly relied on the Wharton Index, which is a turn-of-the-21st-century survey of about 2600 municipalities. The Index highlights those municipalities that have terms such as ‘growth control’ in their statutes, and shows the responses of municipal planning directors and other officials about ‘perceived’ regulatory pressure. The models using the Wharton Index associate an average effect of housing prices with migration elasticity, but they do not differentiate the supposed effect according to incomes, wage levels or skill levels. There is generally no direct identification of how housing regulation affects housing supply. This corresponds to the wide variations in housing regulation in relation to housing supply change and, especially, to the fact that many Northeastern and Midwestern municipalities with weak regulation experience limited new housing construction. Adding to this weakness, in order to characterise regulation at the MSA level, the Wharton Index tends to be aggregated up from the municipalities that control zoning to metropolitan area levels (at which housing markets operate), without weighting for different municipal areas within metro areas (Storper et al., 2015).

A more recent wave of research ventures into estimating how housing supply deregulation affects, variously, population growth and city sizes, inter-regional income inequality, income distribution and national economic productivity and output (e.g. Hsieh and Moretti, 2017). Though some of these authors demur about their own work by claiming to produce only ‘instructive’ simulations, such ventures influence the policy debate. They are picked up by the media as academic proof of the potential benefits of deregulation for housing supply. In Hsieh and Moretti’s (2017)‘full adjustment scenario’– as housing construction becomes unshackled from regulation in prosperous metro areas – New York gains 787% in employment, while the job base is multiplied by five in San Francisco-San Jose. The employment loss in Flint, Michigan, is, by contrast, 98%. Even in their ‘intermediate’ (and thus supposedly more realistic) scenario, New York enjoys 179% population growth and San Jose 149%, and Flint loses ‘only’ 77% of its jobs. Housing deregulation, they claim, would generate US$1.4 trillion annually in additional GDP, through a combination of wage gains and transfer of excessive landowner rents to worker salaries. The benefits of housing deregulation to the national economy would also be highly territorially uneven, as almost all would come from three large metropolitan areas.

Yet, the authors admit that their simulations rest on unrealistic assumptions of perfect mobility and do not consider all the conditions required for a Detroit car factory worker to move to the San Francisco Bay Area or any other New Economy region. Ganong and Shoag (2017) claim that housing regulation has contributed to an approximately 10% greater inter-regional income divergence effect than would otherwise be the case. These claims about the magnitude of potential effects of housing regulation on prices, output, income, productivity and population are implausible, especially when the full costs of migration are taken into account.

In Europe, where planning regimes are, on average, stricter than in the US, tight housing restrictions have also not prevented population growth. The highest-income European regions with the most expensive housing are those attracting the most people (Figure 7).

OPEN IN VIEWER

Figure 7.

Population growth in European regions by income levels, 2001–2014.

The European example, in concert with the analyses of Diamond (2016) and Giannone (2017) and with more realistic estimates of the price of new housing in prosperous metropolitan areas, indicates that, if housing deregulation were to substantially increase inter-regional migration – which is, as we have argued above, improbable – the migration would mostly accelerate the transfer of skilled workers from less prosperous into prosperous regions. Simulations for London suggest this would be the outcome of authorising construction in the Green Belt. A chain reaction would be triggered: those close to the Green Belt would shift further into London as others arrive on the London periphery (Szumilo, 2017).

Three consequences of such feedbacks are expected. First, in contrast to Ganong and Shoag (2017) and Hsieh and Moretti (2017), greater, rather than less, inter-regional skill and income divergence would be the outcome, although the magnitude of this phenomenon may be small. Second, any further emptying out of skills and talent in lagging regions will further degrade their future capacity to improve their economic performance and widen the development gap. And third, the intra-metropolitan movements of the skilled will further increase intra-regional spatial-neighbourhood inequalities, while simultaneously increasing the commuting times of the less-skilled.

The effects of upzoning: Gentrification without affordability

Hsieh and Moretti (2015) argue that NIMBY-ism in prosperous cities, and strong housing regulation more generally, redistribute income from workers to rentiers and accentuate income inequality (presumably placing the landowners higher in the intra-regional income distribution), which, in turn, inhibits aggregate growth. These claims are more plausible than those made about inter-regional migration and convergence, but they still require considerable nuance.

To begin with, there is a strong and strengthening correlation between regional per capita income and the Gini coefficient on income of urban agglomerations. Prosperous metropolitan areas such as San Francisco, Boston and New York are more unequal, in the aggregate, than less prosperous ones such as Provo, Utah – although the American Deep South represents an exception. Welfare and income redistribution systems, among other factors, also play an important role in levels of inequality. As Musterd et al. (2017: 1070) indicate, ‘segregation levels of the better-off and the worst-off are still lower in metropolitan Europe than in the largest metropolitan areas in the United States’.

Some of the intra-regional inequality in prosperous cities may indeed be due to a net transfer of income from non-owners to owners, as housing prices increase more rapidly than wages and other prices. However, the assumptions in the models behind these assertions are often too simple and the data too aggregated to shed light on this relationship. In the San Francisco Bay Area, the households with an income above US$150,000 increased by 80% from 1990 to 2015, with their proportion of the total growing from 17 to 27 per cent (ABAG, 2017: 9). These populations have pushed up housing prices and gentrified a small number of neighbourhoods in central areas that were formerly quite poor. This rising inequality in incomes powerfully affects the housing available to less-skilled lower-income workers because the existing housing stock of prosperous city-regions has been upgraded, generating an affordability crisis for lower- and middle-income households.

The powerful effects of income inequality rather than aggregate supply emerge from recent analysis of IPUMS (Integrated Public Use Microdata Series) data. Popov (2019) finds that in all of the top 100 US metropolitan areas, housing costs are growing more for those in the bottom half of the national income distribution than for those in the top half. Income inequality has risen in 45 of the top 50 metro regions, and fallen in only 19 of the top 100. Housing costs have actually fallen for the top quartile of the national income distribution in virtually all metro areas, but they have strongly risen for the bottom half. This is not a new problem for the bottom half, which is paying about the same share of its income for housing today as in 1980. The difference is the fall in the housing costs for the top earners. Part of this is attributable to the landowner bonus that figures prominently in Hsieh and Moretti (2017), because top income earners have a higher proportion of owners. But even for renters, top income households show a decline in income going to housing costs, while the bottom half of households that are renters show an increasing share going to housing costs, in a result consistent with Freemark’s (2019) detailed analysis of Chicago.

Building on these data, we argue that policies such as blanket upzoning principally unleash market forces that serve high-income earners, therefore reinforcing the effects of income inequality rather than tempering them. Combes et al. (2018) find that the elasticity of land prices between the centre and periphery of metro areas is non-convex and rises with urban size. Supply changes must then be large and central to bend this curve appreciably and thereby generate trickle-down effects to other areas. Thus, upzoning at a regional scale mainly triggers new housing construction in the neighbourhoods where skilled workers want to live: the already-gentrifying areas and the extensive boundary zones between them and other neighbourhoods. More skilled workers in the upper quarter of the income distribution can thus live in the metropolitan core. Moreover, through filtering, producing housing for high-income households has the positive effect of preventing them from directly outcompeting low-income households for older and lower quality housing stock. But upzoning generally involves replacing older and lower quality housing stock in areas highly favoured by the market, effectively decreasing housing supply for lower-income households in desirable areas. This is gentrification.

There is also virtually no evidence that substantially lower costs trickle down to the lower two-thirds of households or provide quality upgrading of their neighbourhoods. The evidence is that they enhance displacement in neighbourhoods at the boundary of higher-income inner metropolitan areas. This means, once again, declining rents for the highest earners and rising ones for the poorest – in San Francisco, Atlanta, Nashville, Chicago, Philadelphia, Denver, Pittsburgh and Washington, as reported in the Washington Post (Stein, 2018). A boom in luxury construction in desirable inner-city areas has not eased housing market competition for cheaper properties. And while there is more evidence of filtering, this seems to have also stalled.

Let us now expand on intra-metropolitan housing choice dynamics in the face of increasing inter-personal inequality. Income inequality in prosperous metro regions strongly affects less-skilled workers, forcing them into painful arbitraging of their residential locations within such regions, usually to outer suburbs. This often involves long commuting times and high transport expenses that affect the quality of their lives disproportionately compared with higher-income workers; barring that, it involves subject status downgrading in order to live in more central but less amenity-rich neighbourhoods.

A different type of arbitraging is at work for less-skilled foreign immigrants. Inter-generational social mobility is higher in the more prosperous metropolitan areas (Chetty et al., 2014), making them magnets for less-skilled immigrants. But while foreign immigrants are willing to accept poor living conditions (higher house prices in lower quality neighbourhoods), low-skilled domestic migrants are less inclined to leave their places of origin, as they already have a higher relative social status than they could achieve in more prosperous metropolitan areas.

In any event, all types of lower-income households in prosperous regions pay the price of ‘displacement’ in competing with higher-wage workers, who benefit from upzoning to gentrify neighbourhoods, as they occupy their newer, higher quality housing. None of the extant models or simulations provide realistic estimates of how much new housing would result from upzoning in prosperous regions – or of the realistic geographical distributions of such new supply; the magnitudes of intra-metropolitan sorting of the skilled to new housing stock and inter-metropolitan increases in skilled in-migrants; or their effect on housing competition (see also Freemark, 2019).

Hence, without active policies to help low-income households and their neighbourhoods, the less-skilled would not benefit from the blanket upzoning policies prescribed by the mainstream literature. As indicated by Jacobus (2019), if upzoning leads mainly, as is overwhelmingly the case, to building ‘only high-end housing, everyone may see some benefit, but most of the benefit will flow to the rich’. This is evident as both more un-regulated (Houston, Phoenix, Orlando) and highly-regulated and supposedly NIMBY-ist (Boston, New York, San Francisco, London, Paris and most large European cities) housing markets feature high levels of housing segregation by income and increasing commuting times, especially for low-income residents.

Segregation by income, race, national origin and other vectors, of course, has manifold structural causes (Boustan, 2017; Sampson, 2012, 2018). Even in cities with strong ‘mixity’ policies (such as Paris), market forces push in an opposite direction, although renter protections can slow down the gentrification and segregation processes (with other side effects). Regulation and other policies are often supported by residents who wield political power to enforce homogeneous neighbourhood quality and resist dis-amenities, with the intended or unintended outcome of segregation. Overturning these regulations, however, has little to do with any general liberalisation of housing markets. In Chicago, for example, it has been found that upzoning has had unintended consequences, such as raising housing prices without necessarily triggering additional construction of newly permitted dwellings (Freemark, 2019). Highly deregulated Atlanta or Houston are also more segregated than most highly regulated cities. Indeed, the policy mix that would be required to reduce segregation in the name of providing better access to jobs and transportation, reducing commuting times and getting access to better schools and amenities for lower-income groups has largely eluded research, even though there is evidence that when lower-income groups access higher quality neighbourhoods there are strong positive effects on childhood development (Chetty et al., 2015). As a whole, the housing as opportunity school has failed to properly internalise that the intra-urban housing market is highly segmented and that different large spatial and structural factors affect the characteristics of within-city submarkets (Jacobus, 2019; Watkins, 2001). For our purposes here, upzoning is not the kind of delicate and complex policy mix that is required to address inter-personal inequality in our cities. Most importantly, undifferentiated aggregate supply policies do essentially nothing to abate the underlying structural causes of the housing crisis in prosperous metro areas that we have identified: high demand from highly-skilled, high-income people; increasing income inequality; and a rise in construction and land costs due to the growth and maturation of metropolitan regions and demands for a higher quality urban environment. The targeted policies that would be needed to reduce spatial-economic segregation may involve increased regulation and other forms of public intervention into the housing market, exactly the opposite of the deregulation approach. The evidence from cities with active public/social housing programmes (such as New York, Paris and London) is that this requires high public subsidies for construction of affordable housing.

The uses and misuses of theory

The housing as opportunity school has become vocal and assertive about the political and policy uses of its research, but, as we have argued, its research is, in our view, not scientifically solid enough to merit this assertiveness. The reasons for this are:

Many of these weaknesses stem from the underlying spatial equilibrium model that is used with few questions in much of urban economics today. The field needs an enriched and more realistic approach, fully incorporating the geography of labour demand and ranked preferences (e.g. Schwartzman, 2017). Households consider not only the average cost of housing when contemplating mobility, but, first and foremost, the type of jobs available given their skills. In today’s environment, less-skilled domestic workers avoid big, expensive cities not simply because of high average housing prices there. They could secure some type of housing in these vast metropolitan markets, as most external migrants do. Nevertheless, the declining urban wage premium for internal less-skilled migrants, combined with uncertainty about the future of their income, as well as their likely high commute times and subjective status downgrading (such as having to co-locate with immigrant groups whom they consider to be of lower social status than themselves), shape their decisions not to move to prosperous cities. There is no realistic housing supply expansion in prosperous metropolitan areas that could address the employment and residential utility requirements of less-skilled domestic workers and enable them to move to prosperous regions in any great numbers.

In the face of what remains an insufficiently developed scientific case, however, the political uses of the housing as opportunity position have become quite prominent. Opposition to more lax planning regimes – and to theories that promote the development in the London Green Belt or constructing on the park lands that encircle the hills of the San Francisco Bay Area – comes not only from rich, rentier land owners, but also from ordinary citizens who appreciate green spaces in their daily lives, as well as dedicated environmentalists. Yet these groups are now depicted as NIMBYs opposed to social justice and this representation is backed up by prestigious academic authorities. In lagging regions and in the populist – and, increasingly, the mainstream – media, residents of prosperous regions are portrayed as erecting ramparts to keep out the less fortunate (e.g. Economist, 2019; Edsall, 2018; Guilluy, 2014). There is little consideration of the fact that the high-skilled workers, who are the main political constituency of the YIMBY movements, might be more motivated by self-interest than social justice. Part of the mainstream academic literature may also have become – wittingly or unwittingly – a stalking horse for developers whose primary interest is not in reducing socio-spatial inequalities or spreading prosperity. Serious affordability policies, which inevitably involve public subsidies and regulation, as well as measures to finance them, are curiously absent from the literature, with its focus on deregulation.

It is also worrying that policies aimed at promoting place-sensitive development in the left-behind regions, where large numbers of individuals are becoming increasingly spatially trapped, continue to be dismissed out of hand as deadweight loss subsidies ‘targeting heavily distressed areas into which outsiders are unlikely to migrate’ (Kline and Moretti, 2014: 657).

It is our view that too much is being promised to policy-makers about the supposed potential benefits of housing market de-regulation. At the same time, in the rush to promote an oversimplified vision of ‘densify near transit stops’, too little consideration is being given to the policies that would promote affordability for the right people in the right places. Moreover, planning deregulation and housing construction in prosperous regions – while interesting issues – are not going to solve the problem of areas lagging behind. However, an excessive focus on these issues at the expense of serious and sustainable development strategies can fuel economic, social and political distress and anger in declining and lagging areas that can threaten the very foundations on which economic activity, both in less developed and more prosperous areas, has been erected in recent decades (Rodríguez-Pose, 2018). It is vital to keep considering the important role for regulation and other forms of public intervention in combating the severe socio-spatial inequality that afflicts prosperous metropolitan areas today. And, to return to our introductory discussion, it is ever more vital to consider that a complex array of problems contributes to the current stagnation of less prosperous regions, notably the structural changes in the spatial distribution of employment, agglomeration forces and the types of skills that are in demand today.