Housing affordability: a framing, synthesis of research and policy, and future directions

ABSTRACT

We synthesize the vast, international scholarly literature related to the growing problem of housing affordability. As a foundation for assessing both the causes of housing affordability problems and possible policies for alleviating them, we take a systematic, holistic perspective and specify nine structural relationships for a metropolitan area that comprehensively delineate the determinants of housing affordability. We discuss concepts and measures of housing affordability, evidence on the primary causes and effects of unaffordable housing, alternative policy approaches, potential future trends in affordability, and suggestions for further research. While focusing on the most recent, high-quality empirical evidence, we also rely on previously published reviews and seminal papers in order to provide historical perspective on how the literature has evolved. Our reconnaissance of the international literature since 2008 demonstrates that although evaluation research in the context of western, developed countries’ housing policies has dominated, increasingly such research has emerged from other countries.

Highlights

• We discuss how to frame housing affordability issues, synthesize evidence from the relevant literature, and propose directions for future research.

• We take a holistic perspective to understand causes of housing affordability problems and assess policy responses.

• Our review of the international scholarly literature encompasses the evidence from non-western, developing countries.

• Despite the extensive literature, there remain unexplored questions and emerging topics.

KEYWORDS:

• Affordability
• housing prices
• rent burdens
• construction costs
• affordable housing policy
• policy evaluation

Acknowledgement

The authors gratefully acknowledge the research assistance of Yun Yee Soh and Tatum Khoo.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 In addition, in the year prior to this writing there have been special issues of the Journal of Housing Economics and Regional Science and Urban Economics focusing on aspects of this topic.

2 Baer (1976), Hulchanski (1995), Stone (2006b) and Leishman and Rowley (2012) provide historical discussions of the evolution of the term ‘housing affordability’ and its measurement.

3 The challenge of defining affordability is not only technical but also is deeply embedded in cultural narratives of causation and policy agendas (Friedman & Rosen, 2019a).

4 In principle, the definition of housing affordability implicitly adds a dimension of physical adequacy of housing. For example, Bramley (1990:, p. 16) argues that households should be able to occupy housing that meets well-established (social sector) norms of adequacy (given household type and size) at a net rent which leaves them enough income to live on without falling below some poverty standard. Despite broad recognition of the importance of the right to adequate housing as a basic human right, the definition of adequate housing itself has not been a main scholarly concern potentially because it needs to be country-specific and highly contextualized.

5 To do otherwise would imply that even if one were homeless this would not be defined as a housing affordability problem if one were able to consume more than a minimally adequate amount of other goods and services. The converse argument is equally unreasonable: even superior housing is insufficient to make up for no food or clothing.

6 A similar option replaces the housing expenditure-to-income ratio with an absolute income residual (Thalmann, 2003).

7 Thalmann (1999) calls this a ‘type I misclassification.’ In his analysis of Swiss housing markets, Thalmann (2003) finds a substantial share of all households would be misclassified using these concepts of affordability.

8 There have been numerous articles critiquing measures of housing affordability, from which we draw here; see esp. Whitehead (1991), Linneman and Megbolugbe (1992), Hancock (1993), Bogdon and Can (1997), Thalmann (2003), Stone (2006b), Abelson (2009), Haffner and Heylen (2011), Myers and Park (2019).

9 Compare the non-spatial approaches of Yezer (1981), Lerner and Reeder (1987) and Thalmann (1999, 2003) with Bogdon and Can (1997).

10 This could be done by hedonic regression-based adjustments to prices, or by using repeat-sales price indices.

11 For example, the literature employs residual income amounts varying from 2/3 of the poverty line to well above it (cf. Kutty, 2005; Stone, 2006a, 2006b).

12 For example, the literature employs ratios between 25 and 50% of income (cf. Kearns, 1992; Stone, 2006a; Chen et al., 2010).

13 A similar complication exists for other unique types of tenure such as Jeonse in South Korea where a tenant gives a certain amount of deposit to the owners during the contract period and gets back the full deposit at the end of the contract.

14 Of course, household-level indicators can be aggregated to larger spatial scales to create aggregate measures. The purely aggregate measures suffer a variety of additional drawbacks beyond those discussed above. The actual distributions of households’ financial positions and the array of housing available for rent or sale by cost is easily obscured by use of medians. Variations in medians within the given spatial area for which the index is being calculated adds further noise. Routhier (2019) recently proposed a composite ‘index of rental housing insecurity’ the combines experiences of unaffordable, poor-quality and overcrowded housing, plus evictions.

15 For international comparisons of how policymakers have formulated housing affordability, see, e.g.: Edwards (1990) for Australia; Linneman and Megbolugbe (1992) for U.S.; Whitehead (1991) for UK; Hulchanski (1995) for Canada; Murphy (2014) for New Zealand.

16 For simplification here, we do not concern ourselves with the specification of temporal lags or whether levels or changes in factors are more important.

17 Occupancy price here is the annualized cost of what a household requires to contract with the owner of a vacant dwelling to secure that place of residence (either to rent or buy). This is simply annual rent plus utilities in the case of renter-occupied stock, and the annual cost of mortgage principal plus interest (net of taxes), utilities, real estate taxes, insurance, and maintenance in the case of the owner-occupied stock. This equivalent to the ‘short-run’ view of housing expenses (Haffner & Heylen, 2011: Table 2) and differs from the long-term user costs of owner-occupied housing as articulated by Quigley and Raphael (2004).

18 This claim has been empirically supported by Quigley and Raphael (2004) in the U.S. and Bramley (2012) in the U.K.

19 Here quality includes neighbourhood characteristics, public service/tax packages, local natural or human-made amenities and disamenities, etc.

20 For analyses of how $O is related to fundamental economic forces, compared to speculative ‘bubbles,’ see Nordvik (1995), Himmelberg et al. (2005) and Mikhed and Zemcik (2009).

21 This relationship arises because land prices equal the present discounted value of the expected future earnings from land and the discount rate is directly related to the real after-tax interest rate, which in turn is directly related to the user cost of housing, which in turn is inversely related to housing prices (Hendershott, 1988).

22 For evidence on this relationship, see Nordvik (1995).

23 The third and fourth points provide a conceptual basis for the empirical observation of great cross-metropolitan variability in the elasticity of housing supply (Green et al., 2005; Saiz, 2010).

24 Higher housing prices may have equity implications, but not efficiency ones, as implied when using the term externalities. How perverse if we could call the creation of many ‘living wage jobs’ a ‘market failure’ because it raised housing prices!

25 For fuller discussions of these theories, see: Meen and Andrew (1998), Meen (2002), Himmelberg et al. (2005), Clark and Coggin (2009). We do not include high costs of mortgage borrowing among these theories, since it only applies to homeowners and real interest rates have stayed generally low throughout the recent period of housing affordability crisis. Moreover, the theoretical effect of real interest rates on cost of homeownership is multifaceted and extremely complicated to explain succinctly; see Hendershott (1988).

26 This same logic can be applied to inter-neighbourhood price variations; Anenberg and Kung (2018).

27 They call these ‘superstar cities.’ Also see the complementary model of Behrens et al. (2014).

28 There are many publications we do not consider here that present putative causes of growing housing affordability problems, but do not attempt to parse quantitatively their respective contributions; see e.g., Bramley (1994) on the U.K., Yates (2008) on Australia. There is also a set of idiosyncratic quantitative studies that analyze various sorts of housing market processes, such as chains of household moves across types of neighbourhoods (Mast, 2019) and inter-neighbourhood price setting (Anenberg & Kung, 2018), from which implications can be drawn about the potential causes of housing affordability problems.

29 Also see Yates (2016) and Worthington (2012) for Australia; Gyourko and Tracey (1999); Quigley and Raphael (2004) for the U.S. Note, however, that there has been a surge since 2005 in construction costs (especially related to labour) in the U.S. (Romem, 2018b). It is likely that a good share of these rising construction costs are endogenous; i.e., they have been generated by upsurges in residential construction since 2010 (see relationships 7 and 8).

30 An interesting variant of this approach is by Bramley (2012), who estimated a multivariate model for subjectively (self) assessed mortgaged owner affordability problems.

31 Also see Chen et al. (2004) for Asia; Bourassa and Hendershott (1995) and Worthington and Higgs (2013) for Australia; Potepan (1996) and Hort (1998) for Sweden; Gyourko and Voith (1992), Abraham and Hendershott (1994) and Clapp and Giaccotto (1994) for the U.S. Meen (2002) provides a review of these and other studies.

32 In addition, some local land use regulations may be endogenous to home price appreciation (Malpezzi, 1999). For a strong counter to the notion that removing planning controls would substantially lower housing prices, see Meen, Gibb, Leishman, and Nygaard (2016).

33 An exception to this generalization is the work of Susin (2001) and Saiz (2003), who employed the natural experiment of the Mariel boatlift from Cuba to provide exogenous variation in population that could be used to estimate its impact on Miami housing prices.

34 We consider the issues of housing quality and affordability to be two sides of the same coin, as one can typically trade off the two. We recognize, however, that housing affordability issues are also related to homelessness and housing instability. The literature on these topics is so vast, however, that we do not consider it here; see Leventhal and Newman (2010), Dockery et al. (2010, 2013), Johnstone, Lee, Shah, Shields, and Spinks (2014), Maclennan, Ong, and Wood (2015).

35 For an excellent review of these consequences, see Gabriel and Painter (2018). There have been claims that lack of affordable housing will lead to an erosion of social capital, widespread political disaffection, and even a challenge to democratically elected governments (e.g., Wetzstein, 2017). We find such claims highly speculative and not supported by rigorous scientific investigations, thus we do not discuss them here.

36 For example, Chan et al. (2019) found that areas with higher housing costs increase the probability that young adults will either remain in or return to the parental home, and this may trap them in a place with fewer employment prospects (especially for black young adults).

37 One example is providing equal opportunities of income and wealth generation through homeownership promotion.

38 Also, see the Housing Studies’ Special Issue on After The Crisis: Housing Challenges in the United States and the United Kingdom (2016).

39 Even in Asian countries where government interventions in housing markets have been relatively stronger, some scholarly works have discussed the interactions between government intervention and private market responses. For example, evidence suggests that the price movement of public and private housing markets are closely interrelated, and policies that aim at any change in supply, financing, or regulations affecting one market have significantly influenced the other market in Singapore (Phang & Wong, 1997; Ong & Sing, 2002).

40 More recently, involving the non-profit organizations in the affordable housing provision has been increasingly popular. The rationale comes more from a normative ground (e.g. wiliness to serve poorer tenants, added community benefits) instead of pure economic efficiency (O’Regan & Quigley, 2000).

41 For example, see the work of the New Story-ICON collaboration at https://www.youtube.com/watch?v=SvM7jFZGAec

42 http://english.khan.co.kr/khan_art_view.html?artid=201606131848427&code=710100

43 See, for example, ttps://www.rics.org/east-asia/news-insight/future-of-surveying/surveying-technology/using-blockchain-to-own-a-home/; https://housingnews.org.ng/2018/11/21/challenges-of-mortgage-finance-and-the-way-forward/

44 We agree with the longstanding consensus in housing policy analysis that there is no uniformly ‘best’ strategy, but rather that choice depends on the particular cause of the affordability problem, institutional capacity, public goals, and market conditions. (Downs, 1990; Struyk, 1990; Weicher, 1990a, 1990b; Rothenberg et al., 1991; Galster, 1997),

45 For a broader range of criteria, see Galster (1997).

46 For an exception, see Harkness, Newman, Galster, and Reschovsky (1997).

47 There is also a significant variation in the extent of public housing stock and targeted income class and tenure between countries. For example, the share of public housing is 73% in Singapore while it is 7% in urban China, 7% in South Korea, and 5% in Australia. Public housing is homeownership oriented and main tenants are middle-class families in Singapore while main targets of public housing are low- and middle-income renters in most other countries.

48 For example, inclusionary zoning has been implemented in some US states including California, New Jersey, and Massachusetts, and Montgomery Country, Maryland as well as major cities in Canada, Australia, and UK. Also, see the special issue of the Journal of Housing and the Built Environment on inclusionary housing in planning and housing systems that specifically focuses on international comparisons.

49 The debate on the relative merits of supply-side and demand-side housing strategies has been longstanding; cf. Apgar (1990), Struyk (1990), Weicher (1990b), Barton (1996), Galster (1997), Yates and Whitehead (1998).

50 It has been recently estimated, for example, that over 75,000 dwelling units in New York City are intentionally held vacant over the long term (Mays, 2019).

51 ‘Web of Science’ allows access to the Social Science Citation Index (SSCI) which has been considered as a good instrument to determine research quality. However, we acknowledge that our focus on the SSCI listed journals may lead to some bias toward these journals’ preferences on certain topics, methods, and contexts.

52 Note that this study does not intend to investigate a complete population of evaluation studies of affordable housing policies. Despite the extensive search for publications using diverse keywords such as ‘effect’, ‘evaluations’, ‘policy implications’ ‘specific policy/program names’ ‘assessment’, papers retrieved for analysis may not be inclusive of all academic publications in affordable housing policy interventions.

53 In most countries, these policies do not even exist.