Tackling child poverty delivery plan: forecasting child poverty in Scotland

Child poverty projections for Scotland independently produced by Howard Reed at Landman Economics and Graham Start at Virtual Worlds Research.


1 Scottish Parliament 2017b, Explanatory Notes, page 2

2 Scottish Parliament 2017b, Policy Memorandum, pages 1-2

3 Scottish Parliament 2017b, Policy Memorandum, page 2

4 The freeze follows a three-year period under the previous Conservative-Liberal Democrat Coalition UK Government where most benefits and tax credits were uprated by only 1 per cent per year, below the rate of CPI inflation.

5 See Appendix 1 below for a discussion of the sources of these projections.

6 Possible reasons for slow growth in the UK economy as a whole include the collapse in productivity growth following the 2008-09 recession, a forecast decline in immigration post Brexit, shocks from Brexit itself, and secular changes such as the increasing share of service industries and the rise of the 'gig economy'.

7 Except absolute poverty (measure 2), since the poverty line there does not move as median income changes over our simulation period.

8 One interesting reason for the divergence between the OBR and SFC forecasts that could be relevant to us is that OBR continue to use the ONS central population forecast, which does not assume any impact of Brexit, whereas the SFC one uses one of the NRS's pessimistic forecasts, which assumes a fall of 50% in EU net migration.

9 The results in Chapter 3 contain a robustness check – running the child poverty projections using the OBR forecasts for the whole of the UK including Scotland – to check what difference it makes to the child poverty results if we use OBR forecasts for the Scottish economy as well as the rest of the UK. The results show that using OBR forecasts for Scotland makes almost no difference to our poverty forecasts.

10 One interesting reason for the divergence between the OBR and SFC forecasts that could be relevant 1to us is that OBR continue to use the ONS central population forecast, which does not assume any impact of Brexit, whereas the SFC one uses one of the NRS's pessimistic forecasts, which assumes a fall of 50% in EU net migration.

11 The model was originally developed in 2008-09 by Landman Economics for the Institute for Public Policy Research ( IPPR) and was subsequently shared with researchers at the Resolution Foundation which provided additional funding for improvements in the model functionality and performance. The model is now used by all three organisations. Other comparable models in use in the UK include the distributional analysis models used by HM Treasury and the Department for Work and Pensions; the Institute for Fiscal Studies ( IFS)’s TAXBEN model; and the UK version of the Euromod model which is hosted at the University of Essex.

12 The impact of assuming 100% take-up of means-tested social security payments on the result is tested in Chapter 3 as a robustness check.

13 The Family Resources Survey and the Living Costs and Food Survey are both ‘repeated cross-sectional’ surveys rather than panel surveys; they interview a new set of households each year rather than conducting repeat interviews with the same set of households over a number of years.

14 The FRS data for 2016/17 are scheduled to be released on 22 nd March 2018.

15 Although the 2016-17 tax year is in the past, because the most recent FRS data currently available is 2015-16, for the purposes of the empirical analysis in this report, 2016-17 (and 2017-18) are forecast years rather than historical years.

16 A third technique – artificial ageing – was considered for use in this project but rejected as too time-consuming given the available time and budget to complete the project.

17 The uprating index used for the absolute poverty measure in the UK Department for Work and Pensions' Households Below Average Income statistics is now a version of the Consumer Prices Index ( CPI), and the Scottish Government also uses CPI in the Child Poverty (Scotland) Act. See Levell and Skingle (2016) for details.

18 Although material deprivation indicators were introduced in the 2007-08 FRS survey, the questions were changed in 2010-11 so 2010-11 to 2015-16 is the longest run of data that we have on a consistent basis.

19 Analysis of the Scottish Family Resources Survey data for 2015/16 suggests that only around 3 per cent of households with children receive Carers’ Allowance.

20 It should be noted that one of the objectives of the Best Start Grant is to incentivise families to spend on certain child-related items, some of which are part of the questions used in the FRS to measure material deprivation. Thus, it is possible that the introduction of the Best Start Grant could result in reduced poverty under Measure 3 due to a decrease in measured material deprivation. The modelling used in this report is unable to capture this effect, however.

21 Intuitively, this means that there is a 95% chance that the true value is somewhere between our upper and lower limits. See Giles (2011) for more detail on this.

22 The statistical significance of the increase in poverty measure 3 over the forecast period (at the 95% level) arises because the upper confidence interval for 2015/16 is below the lower confidence interval for 2030/31.

23 Reed and Portes (2018), Chapter 9, shows that the roll-out of Universal Credit leads to an average gain for one-earner couples with children compared to the previous benefit and tax credit system which UC replaces.

24 The IFS study does not present forecasts for combined material deprivation and relative poverty, or for persistent poverty.

25 See (Creedy 2003) for a very good, largely non-technical discussion of the techniques involved, (Creedy and Tuckwell 2003) for a discussion of an application similar to ours, and (Deville and Sarndal 1992) for the statistical background.

26 consistent with the Child Poverty bill, we use financial years (April-March) throughout. Our forecasts for population and household composition, however, use calendar years, which we interpolate. The differences are minor, however.

27 (see Appendix 1.1 for the full list)

28 Over the period 2012/13 to 2015/16, 37% of lone parents in Scotland were in AHC relative poverty, as against 18% of all adults (Scottish Government 2017f, Table 2a).

29 Note this implies participation and unemployment rates are independent of migration and fertility levels.

30 The initial weight is the ratio of the total number of households forecast for the year to the count of FRS households.

31 The take-up regression for Housing Benefit does not include an owner occupier dummy variable because Housing Benefit can only be claimed by tenants.


Back to top