Alternative population for derivation of poverty threshold As noted in Section 3, using the full population’s income distribution (and thus including groups such as pensioner households) for deriving the poverty threshold may have spillover effects on the estimated prevalence of in-work poverty. We therefore replicate the figures in Table 3 for deriving the poverty threshold based on the income distribution of our population of interest (i.e., workingage households excluding pensioner and self-employed households). For ease of comparison, this alternative poverty threshold is equivalent to 75 percent of the median of the full population. The resulting in-work poverty rate is 12.4 percent (and 14.8 percent when excluding WfF and AS). 15 Role of Working for Families and Accommodation Supplement We calculated the in-work poverty rate excluding income from WfF tax credits and AS. For our headline estimate based on a poverty threshold of 60 percent of the median equivalised income, inwork poverty prevalence increases from 7.0 percent to 9.2 percent. This illustrates the marked effect of such welfare policies. The impact is also visually represented by the difference between the two lines portrayed in Figure 1. Analysis by gender, children, region and time In-work poverty rate by gender We next break down the figures in Table 4 to determine the in-work poverty prevalence for adults in general, and separately for each gender. We find that 1,507,986 adults live in the above 725,313 working households. Out of this sample, 107,844 adults live in in-work poor households, which equates to an in-work poverty rate at the adult level of 7.2 percent. Decomposition according to gender reveals heterogeneity in the prevalence: 7.7 percent (59,967 adults) of females are associated with an in-work poor household, while for men this number is 6.6 percent (47,874 adults). 16 Additional analyses of in-work poverty rates by the gender of the household’s main earner can be found in Section 4.6 (Table 9). In-work poverty rate for children In terms of deriving the likelihood of children belonging to an in-work poor household, we further disaggregate the numbers in Table 4. Our sample of 725,313 working households translates into 567,069 children aged below 18. Out of these, 10 percent (56,823 children) live in an in-work poor household. In-work poverty rate over time To derive changes in the prevalence of in-work poverty over time 17, we use HLFS data over the period 2007 to 2017 to ascertain household units and link with income information from the Inland Revenue and WfF and AS (until 2012) for associated members within each household. As evident in Figure 2, when using HLFS data, we see an initial in-work poverty rate (including WfF and AS) of 8.2 percent in 2007, after which it rises to its peak (over our sample timeframe) of 9.4 percent in 2009, which coincides with the period post-global financial crisis. Afterward, there was a small Analysis by Bourquin et al. (2019) with the UK income distribution also includes in-work poverty rates when excluding pensioners from calculating the poverty threshold. The authors find a much smaller lift in the in-work poverty rate (the magnitude varies between +1 to +3 percentage points over the period 1994 and 2017). 16 Note that the 2013 Census did not provide other gender response options apart from male and female. 17 Note that we use a year-specific poverty threshold for each survey time point over the period 2007 to 2017. 15 Page 21