Labour and housing market precarity: What is the impact of time-related underemployment?

Abstract

Though unemployment has long been recognised as a threat to housing security, the impact of time-related underemployment remains neglected. This article takes an integrated approach, linking unemployment and underemployment within the context of a household analysis of housing insecurity. It draws on panel data from the Household, Income and Labour Dynamics in Australia (HILDA) survey and deploys an innovative measure that differentiates households in terms of employment hours deficits. The results confirm a significant relationship between total employment hours deficits within the household and housing payment arrears and reinforce the importance of integrating time-related underemployment into labour market and housing analysis.

Keywords:

• Australia
• housing insecurity
• precarity
• underemployment
• unemployment

Notes

1 Time-related underemployment is defined more fully below. Other forms of underemployment include skill related, where workers’ human capital is underutilized in their present jobs (Brown and Pintaldi, 2006). This article restricts its attention to the time-related dimension of underemployment.

2 Employees of eligible businesses received a flat rate A$1500 per fortnight that has been subsequently wound back in stages.

3 This is a refinement of the typology developed in Campbell et al., 2014.

4 If children over 15 years of age and living with their parents are taken out, the employment-rich category is even more important at 82.6% of the sample. There are, however, still 17% in the sample that have an inadequate amount of employment.

5 Outright owners are omitted, as are those living rent free and have zero recurrent housing costs. Among all households (with at least one LFP) the majority tenure is home purchase (43%), with private rental housing not too far behind (32%). Social housing accommodates the smallest share (2%). The omitted outright owners and households living rent free have tenure shares of 20% and 3% respectively. Thus, over three quarters of sampled households reside in tenures where recurrent housing costs must be met. Note that our sample also omits households containing only adults no longer active in the labour force. These households are typically in retirement and if homeowners they are generally mortgage free, and in social housing if renting. The sample tenure shares will therefore differ from those obtained from the population of all households.

6 Social housing tenants are only 2% of the sample as this sector is marginal in the Australian housing system. Rents are typically set at 25% of household income and the incidence of weighted arrears is 13.5%. While omitted from table 2 for space reasons, they are included in the regressions.

7 Household disposable income is equivalised using the modified OECD scale (Förster, 1994).

8 Each LFP is attributed the employment deficit value of the household he or she resides in.

9 The unit of analysis in table 1 is households while table 2 uses individual LFPs. The share of employment-rich households (77.3%) can therefore differ from the share of LFPs resident in employment-rich households (73.25). Table 1 also includes households in social rental housing a group omitted from table 2, and is another source of difference.

10 When a dependent variable is dichotomous ordinary least squares regression will result in inefficient estimates and predicted values outside the permissible 0 – 1 range. Maximum likelihood estimation of a logit model is a standard approach (see Wooldridge, 2009: pp. 575 – 580). The random effects model partitions the error term into two components: time invariant person specific and stochastic time varying components. For an introductory treatment of random effects logistic regression using panel data see Allison (2009: pp. 32 - 37).

11 In table 2 a slightly higher figure of 8.2% of purchasers are in arrears. The difference is due to weighting of observations in table 2 (see note a to table 2), and the omission from the sample used to estimate the arrears model of LFPs with one or more missing observations on control variables. The arrears share among renters in table 2 is slightly lower (at 14.3%) because in addition to the above reasons, table 2’s sample omits social renters.

12 For a dichotomous variable it is the ratio of odds occurrence in two groups - the odds for the group identified when the predictor takes the value one relative to the odds for the group identified when the predictor is zero. It is obtained from $e^{\widehat{{\beta }_i}}$ where $\widehat{{\beta }_i}$ is the coefficient estimate for dichotomous variable i. For a continuous variable, the transformation yields the odds of event occurrence when there is a one unit increase in the continuous right-hand side variable relative to the odds before the one unit increase.

13 Capital city location, lone person and couple with dependent children households are only significant in the tenant sample. Indigenous persons have statistically significantly elevated odds of arrears in the purchaser sample only.

14 Some of the household employment status categories have very small numbers in the purchaser sample. The unemployed single LFP category and the underemployed single LFP category are both present in only around 1% of observations (see online Appendix, Table A2).

15 In both renter and buyer samples the controls retain the same levels of statistical significance, and the odds ratio estimates are little different from those reported for model 1 (see online Appendix, Table A3).

16 Tenant households with only one underemployed LFP are again most prone to payment arrears, while the other two household employment groupings that feature a single LFP are also relatively more likely to fall behind on their housing payments.

17 In quarter 3, 2021 the percentage change in house prices on the same period of the previous year was 21.7% in Australia, 10% in UK, 16.5% in USA and 9.2% in EU countries. Rents have increased less steeply over the same time period: 0.3% in Australia, 1.4% in UK, 2.1% in USA and 1.1% in EU countries See: https://stats.oecd.org/Index.aspx?QueryId=98861