Housing and marriage in Singapore
Assessing the impact of the primary public-housing market on marriage and consumption.
27 January 2022
In their paper “Tying the knot in a new home: consumption responses to a pro-marriage housing policy”, Agarwal et al. (2021) say that “people get married for different reasons, either for love or love of housing”.
Singapore is well-known for its pro-marriage housing policy. A massive public housing programme run by the country’s Housing and Development Board (HDB) has resulted in nearly 90% of the population living in generously subsidised apartments that go by the moniker, “HDB flats”. New public housing units sold directly by HDB on the primary market are popularly known as Build-to-Order (BTO) flats, in contrast with those transacted between private individuals in the resale market.
By and large, young Singaporeans are required to form a “family nucleus,” which entails getting married, before they can apply for a BTO flat, though singles are allowed to do likewise when they reach 35 years old.
So do couples get married in order to have a home of their own, or do they buy a home because they are getting married and want to live together? Is marriage an end in itself or merely a means to residential property?
And how does HDB’s policy stance on the family-nucleus prerequisite influence pre-marriage decision-making and post-marriage dynamics?
Agarwal et al. deep dive into the co-movements between the housing and marriage markets, and add to the existing literature by constructing a plausible case for a direction of causation between the two variables, namely, that favourable housing conditions encourage marriage.
Establishing correlation
Following the completion of the first batch of BTO flats in 2005, marriage rates showed a marked increase. The pre-2000s marriage rate of 45% for women aged between 25 – 29 rose to 55% in 2005 – 2009, and further climbed to almost 60% in 2010 – 2014. Between 2015 – 2019, the marriage rate hovered around a relatively high 55%.
Marriage rates among men in the 30 – 34 age bracket also followed a similar uptrend, rising from a low base of 22% in the pre-2000 period to nearly 40% after 2010.
These age brackets for women and men were selected based on stylised facts that females become competitive in the marriage market at a relatively young age (i.e. the 25 – 29 age group), while males choose to get married when they are financially prepared (i.e. 30 – 34 age group).
The rise in marriage rates coincided with the injection of BTO flats into the primary market in 2005 (five years after the inception of the BTO scheme in 2000), and a subsequent price decoupling from the HDB resale market in 2010. Prior to 2010, BTO flats had closely tracked prices in the secondary market. But as market dynamics began to jack up the prices of resale flats, the Singapore government swiftly intervened to keep primary public housing affordable and widely accessible.
Having established a correlation between housing and marriage, Agarwal et al. next explore whether there was a causative relationship between the two variables.
Evidence for causation
As the researchers say, the “formation and dissolution of marriages are sophisticated, and endogeneity further complicates the analysis”. Among other factors, marriage decisions are susceptible to influence from the labour market, human capital investment, location decision process and intergenerational impact.
Correlation between two variables does not necessarily engender a causative relationship, since these may be varying in conjunction with a third (and likely unknown) variable which is the actual cause. Correlation, while necessary for causation, is nonetheless insufficient for it.
To address the issue of endogeneity, Agarwal et al. employed a multi-pronged approach comprising (i) marginal analysis, (ii) identifying a quasi-experimental equivalent of an experimental control group via propensity score matching, and (iii) offering corroborating evidence on how the “treatment” (i.e. injection of BTO flats into the market) affects social norms.
Marginal analysis
Through ordinary least squares (OLS) estimates from simple regressions, the researchers found that one additional BTO project was followed by an increase of approximately 100 to 150 marriages over the next one to five years.
Similarly, an increase in every 100 BTO flats was associated with an increase in 15 to 20 cases of marriages, while a 1% increase in the standard deviation of annual BTO prices resulted in marriages increasing by 1,500 to 2,000 cases in the following one to five years.
These estimates were all statistically significant with high correlation coefficients. According to the researchers, “the high positive associations shed light on the marginal effect of the BTO supply and price variations in the marriage market”.
Quasi-experimental control
Working with a database of approximately 2 million Singaporean residents at least 20 years old from 1996 to 2018, the researchers deployed propensity score matching (PSM) to segregate their subjects into two groups, namely: BTO residents between 24 and 34 years old, and non-BTO residents (after meeting the minimum occupation period, MOP) in the same age bracket with similar demographics.
Agarwal et al. found that more subjects in the treatment group (BTO residents) were married, with a difference of 6.5 percentage points between the two groups. The differential was statistically significant at a high 1% level.
Corroborating evidence
In multicultural Singapore, different ethnic groups have different months of preference for getting married. From year 2000 onwards, especially after 2010, the researchers discovered a “fading” of marriage seasons – marriages became less dispersed throughout the year, and those changes coincided with the commencement of the BTO scheme at the turn of the century and the BTO supply expansion periods in 2001 and 2010, respectively. Apparently, the need to get married when a BTO flat is ready for occupation took precedence over customary norms.
With the above three sources of evidence taken in conjunction, a case for housing as a causative trigger for marriage could be made. Agarwal et al. suggest that the “BTO scheme is likely to spur marriage decisions” and, more specifically, that a pro-marriage housing policy, namely “the prerequisite in the BTO scheme to form a family nucleus”, would play an instrumental role.
Consumption friction
Using two years prior to moving into a BTO flat as the base year, the study found that “card spending” for married BTO residents below 35 years was 12.4% less on the overall, compared to the base year. Card spending was defined as expenditure charged to credit cards, debit cards and NETS accounts.
Out of the 12.4%, expenditure on “visible goods” – such as automotive, memberships, watches and jewellery, and cosmetics and spa – accounted for the bulk of the reduction at 35%, with statistical significance at the 1% level. Reductions in expenditure on necessities and entertainment were, however, relatively smaller and also statistically insignificant – the drop in expenditures there could have very well occurred by chance.
Married residents in non-BTO estates, on the other hand, spent 19.7% more on average vis-à-vis the base year. As additional corroboration, the study found that consumption patterns for singles (aged between 35 – 45) living in BTO flats were not significantly affected. The researchers therefore controlled for variable values – “live in BTO” vs “does not live in BTO” and “married” vs “not married” – so as to isolate “married couples living in BTO” as the reason for decreased spending.
Explaining consumption friction
An analysis of subjects with outstanding mortgage commitments showed that married BTO residents spend 41.6% more on mortgage payments than married residents in non-BTO estates. Considered in relation to their reduction in overall consumption and especially in visible goods, the evidence suggests a trade-off between mortgage and personal consumption among married BTO couples.
According to the researchers, “married BTO residents who tie the knot when buying new BTO flats bear a higher housing payment burden than their non-BTO counterparts who are not financially constrained and have smaller outstanding mortgage balances. With less discretionary income, married BTO residents are likely to cut consumption on visible goods to save for mortgage payment or other necessities.”
Agarwal et al.’s study contributes to the thin literature on the spillover effect between the housing and marriage markets.
(This working paper is co-authored by Professors Sumit Agarwal, Yi Fan, Wenlan Qian and Tien Foo Sing, at the National University of Singapore.)
