Review of policy leavers to combat poverty

Broadly speaking there are two main income-based transfers that government can use in tackling child poverty – cash transfers and in-kind transfers (such as childcare provision). There is debate among academics and research on the most effective approach with a large existing literature arguing the pros and cons of both approaches. While the focus of this paper is on the economic impacts of using cash-transfers, we review literature related to both cash and in-kind transfers.

Many studies employ regression techniques to measure the effectiveness of government transfers on child poverty in different countries. For example, using a linear regression, Bäckman and Ferrarini (Reference Bäckman and Ferrarini2010) analyses the relationship between policy cash transfers and child poverty rates in Europe, finding a clear negative relationship between the two. A similar study is also carried out for child poverty in Europe using the Eurostat and OECD SOCX databases by Nygård et al. (Reference Nygård, Lindberg, Nyqvist and Härtull2019). Again, the authors find a clear negative relationship between in-cash transfers and child poverty rates, but the authors also note that in some case the effectiveness of in-kind transfers is greater than cash transfers.

Bezze et al. (Reference Bezze, Canali, Geron and Vecchiato2020) analyse the social protection benefit system in Italy, finding that the country has a much higher proportion of in-cash benefits than in-kind when compared with the EU average, but is less effective in combating child poverty in disadvantaged families. Using a 120-family survey and discussions with social care workers, the authors find that, while cash-transfers do help to a point, families in mid to extreme poverty experience difficulties other than income that benefit greatly from professional service provision. Coppola and di Laurea (Reference Coppola and di Laurea2016) also argue that the Italian high level in-cash transfers system is ineffective in protecting the worse-off households.

There are several papers that investigate the impacts of both cash and in-kind transfers on child poverty in other countries. Using a microsimulation model and the national social survey 2013-2018 Sinisaar (Reference Sinisaar2021) evaluates the impact of an increase in family benefits in Estonia, finding that this resulted in reductions in absolute, relative and persistent poverty rates. But the author does note that the impact of benefits varies depending on the family type. Pac et al. (Reference Pac, Nam, Waldfogel and Wimer2017) compares the impact of different measures on child poverty rates in the US.

A key component of the literature has been on policy design and how this needs to be adapted for the specific country’s demographics. Using multilevel model analysis, Bárcena-Martín et al. (2018) analyses the different ways of social transfer targeting across 30 EU counties, finding targeting lower income households has larger impacts. Guio et al. (Reference Guio, Marlier, Vandenbroucke and Verbunt2022) also use a multilevel analysis for child poverty in 31 European countries using a range of micro and macro level determinants. They determine that a policy combination of both income support and social benefits is needed to combat child poverty effectively.

A number of studies compare individual countries’ current social transfer system with other similar neighbouring nations. For example, using the EUROMOD microsimulation model, Avram and Militaru (Reference Avram and Militaru2016) compare Romania and the Czech Republic to examine the extent to which child poverty reductions are driven by the policy design and size of benefit or interaction between policies and population characteristics. Salanauskaite and Verbist (Reference Salanauskaite and Verbist2013) carry out a similar study to compare Lithuanian design with Estonia, Hungary, Slovenia and the Czech Republic in reducing child poverty in single parent and large families.

Also explored in the literature is the effectiveness of transfers on reducing child poverty in certain family types and different economic regions. Payments to single parent households are evaluated in Van Lancker et al. (Reference Van Lancker, Ghysels and Cantillon2014) and Chzhen and in Bradshaw (Reference Chzhen and Bradshaw2012), finding the possibility of significant reductions in child poverty rates and that targets focussed specifically on single parents produce the most efficient results. Barrientos and DeJong (Reference Barrientos and DeJong2006) investigate cash transfers as an option in developing countries (South Africa & Latin America), with results similar to the European studies in that they prove to be an effective means of combating child poverty. Kumara and Pfau (Reference Kumara and Pfau2011) find some additional benefits in developing countries of cash transfers, such as an increase in school attendance.

The focus to date in the literature has been on the impact of cash transfers on individual households. While this is part of our investigation, we extend the analysis by considering the macroeconomic impacts of funding the cash transfers policy.

The remainder of this paper is structured as follows: the next section provides background on child poverty in Scotland and the SCP, followed by an outline of the data we use and our simulation methodology. We then present and discuss our results; and draw our conclusions in the final section.

Policy context

The statutory child poverty targets that this analysis focusses on are based on after-housing cost measures of household income, which are net of families’ cost of housing. They measure relative poverty – defined as the proportion of children living in households with income below 60% of the population median. Having an income below this level represents having a standard of living well below the average family in the UK. This is likely to translate into being unable to afford basic goods and services and being unable to participate in extracurricular or social activities, such as sports clubs or birthday parties, without cutting back on already constrained spending on essentials.

Although not the only measure of child poverty referred to in the Child Poverty (Scotland) Act, relative poverty is the most often used measure of poverty in analyses of household incomes. In the rest of this report, relative child poverty (using the after-housing cost measure) is referred to simply as child poverty.

The rate of child poverty has not changed significantly since the passing of the Act in 2017. Figure 1 shows relative poverty between 1995 and 2018 in Scotland based on three-year averages. The child poverty target levels are also shown in the chart by the dashed lines.

Figure 1. Child Poverty in Scotland (Relative poverty, after housing costs).

Source: Family Resources Survey, DWP

The figure shows that whilst child poverty still remains above the targets, it has not always been this high. There were a number of contributing factors to the decline in the early 2000s, including increases in social security for families with children and policies to incentivise and support labour market participation for parents. Many of these factors have been analysed in the existing literature on poverty (see Joyce and Sibieta, Reference Joyce and Sibieta2013).

The Scottish Child Payment was designed by the Scottish Government to help low-income parents with the costs of supporting their family. It originated as a weekly payment of £20 per child below the age of 16 (on full roll out) with a recent rise to £25 in November 2022. Although receiving the Scottish Child Payment may affect the need for some local council grants (for example, the Scottish Welfare Fund), it does not interact with any other UK or Scottish Government benefits that parents or household members might receive. It is similar to the child element in Universal Credit, but it does not have a limit on the number of children who can receive the payment. This makes it particularly effective for tackling child poverty, which is higher for larger families.Footnote ²

The Scottish Child Payment represents a direct transfer between the Scottish Government and households. It is also a policy instrument over which the Scottish Government has direct control. We therefore explicitly simulate scenarios in which the final child poverty targets are met through this channel and highlight the magnitude of the government investment required. With the change in the Scottish Child Payment, we then use the microsimulation model to solve for the changes to income tax bands that are sufficient to generate the tax revenue required to fully meet the net cost of this investment.

Where we model a Scottish Child Payment of £40 per week, for example, this amount is set in 2020/21 (the current financial year at the time of the analysis) and is then uprated in line with the Consumer Price Index to each of our chosen policy years, 2023/24 and 2030/31.

We model an increase in Income Tax rates to pay for the higher Scottish Child Payment, which reduces net incomes slightly. This in turn creates a small increase in Universal Credit entitlement, and thereby a small increase in entitlement to the Scottish Child Payment. Where we report results for the ‘fiscally neutral’ case, these interactions are taken into account.

Data & Methodology

Simulation methods

Our modelling approach employs two forms of simulations: one which changes individuals’ net income through alterations to the tax and benefit system – a microsimulation; and another which estimates the wider economic effects of these changes (for example, on GDP and employment) – a macrosimulation.

Microsimulation

The microsimulation model (based on the Institute for Public Policy Research’s Tax and Benefit model) was developed to forecast short-term changes in the UK income distribution and model the fiscal and distributional effects of tax and benefit policy. Initially the model establishes a baseline net household income distribution using the baseline household income dataset described in the previous section. As the poverty targets are set for the years 2030/31 the base year financial values are uprated – achieved using a combination of forecasts from the Office of Budget Responsibility (OBR), economic indicators and known policy changes. This analysis includes all policies announced up to March 2021.

To implement a policy change a counterfactual tax and benefit system or economic scenario is created. For this paper the counterfactual is increased income from the SCP which then impacts the assumptions of the model creating a new outcome. By comparing the counterfactual household outcomes to its baseline counterpart we can estimate the impact of the simulated change on incomes, rates of poverty, and government expenditure.

Macrosimulation

The macrosimulation proceeds in a similar manner to the microsimulation by setting up a baseline economy. We use a Computable General Equilibrium (CGE) model based on the AMOS (Harrigan et al., Reference Harrigan, McGregor, Dourmashkin, Perman, Swales and Ping Yin1991) family of models. The version of the model used here is based on the 2013 Scottish Input-Output (IO) table with 18 economic sectors. In addition to the 18 sectors/commodities within the model there are three internal institutions – households, firms and governments – and two external, the rest of the UK (RUK) and the rest of the world (ROW). Transactors are taken to be myopic. Scotland is treated as a small open economy so that RUK and ROW variables are treated as exogenous (1). Commodity markets are assumed to be competitive. Financial flows are not explicitly modelled, and the interest rate is assumed to be exogenous.

This AMOS framework has been used in a number of applications (e.g. Allan et al., Reference Allan, Lecca, McGregor and Swales2014; Figus et al., Reference Figus, Lisenkova, McGregor, Roy and Swales2018; Connolly et al., Reference Connolly, Eiser, Kumar, McGregor and Roy2021) and allows for a degree of flexibility in choice of model closures and parameters. Fundamentally, the model assumes that producers minimise cost using a nested multilevel production function. The combination of intermediate inputs with RUK and ROW inputs is based on Armington (Reference Armington1969). Output is produced from a combination of intermediates and value added, where labour and capital combine in a constant elasticity of substitution (CES) function to produce value added, allowing for substitution between these factors in response to relative price changes, i.e.

(1) $${Y_{j,t}} = {\left( {\alpha [EK]{_{j,t}}K_{j,t}^{{{\sigma - 1} \over \sigma }} + \beta {{[EL]}_{j,t}}L_{j,t}^{{{\sigma - 1} \over \sigma }}} \right)^{{\sigma \over {\sigma - 1}}}}$$

Where Y_j,t is the value added of sector j at time t and K and L are the stocks of labour and capital respectively. σ is the elasticity of substitution between labour and capital with share parameters α and β (β = 1 − α). [EK] and [EL] are the efficiency parameters for capital and labour, which initially are kept constant in our simulations.

There are four components of final demand in the model – namely, household consumption, investment, government expenditure and exports. Household consumption is assumed to be a linear function of real disposable income. Real government expenditure in the model is held constant, while exports are determined by an Armington function (Armington, Reference Armington1969) and accordingly are dependent on relative prices.

While the model can be run in dynamic mode, with periods interpreted as years, as both the SAM and behavioural relationships are benchmarked using annual data, here we focus primarily on long-run equilibria in which both capital stocks and population are optimally adjusted. The model is initially assumed to be in steady-state equilibrium, implying that with no exogenous disturbances, the model simply replicates the initial values over all subsequent time periods.

Capital stocks are fixed in the short run, but subsequently each sector’s capital stock is updated through investment, set as a fraction of the gap between the desired and actual (adjusted for depreciation) level of capital stock – in line with the Jorgenson (1963) neoclassical investment formulation. In the long-run, equilibrium investment is equal to depreciation and capital stocks are constant.

There is a single, imperfectly competitive labour market with perfect sectoral mobility. Workers bargain over their real consumption wage where their bargaining power is inversely related to the unemployment rate:

(2) $$ln({w^{s}}{cpi^{s}})=c-0.113(u^{s})$$

where w ^S is the net of tax nominal wage in Scotland, cpi ^s the Consumer Price Index, u ^S the Scottish unemployment rate, and c is a calibration parameter (Layard et al., Reference Layard, Layard, Nickell, Nickell and Jackman2005). Wage and employment changes are governed by the interaction of the wage curve given by (2) and the (general equilibrium) demand curve for labour (which is obtained by aggregating over all sectors’ labour demands).

While there is substantial support for this wage curve specification, there is evidence that wage rises have been very limited since the Great Recession of 2008, so we also allow for a simple fixed nominal wage case, which captures the limiting case of zero wage flexibility. At a regional level this is traditionally motivated in terms of a national bargaining system in which the region acts as a wage taker. It also corresponds to a traditional Keynesian view of the way that regional labour markets operate.

Labour force changes in the model are due entirely to migration as there is no assumed change in natural population. Migration in the model is determined by the real wage and unemployment rate differential between Scotland and the rest of the UK (RUK). We assume zero net migration in the base year (2013) and net migration flows re-establish this equilibrium.

The net migration function is given by:

(3) $$m = v - 0.08[ln({u^S}) - in({u^R})] + 0.06\left[ {ln\left({{{w^S}} \over {cp{i^S}}}\right) - ln\left({{{w^R}} \over {cp{i^R}}}\right) - \beta ln(1 - \tau )} \right]$$

In Equation 3, m is net migration from Scotland to RUK; ν is a calibration parameter to generate net zero migration in the base year and u the unemployment rate with the S and R superscripts representing Scotland and the RUK, respectively.

Households are disaggregated by quintile. The model takes the cost of the policy change from the microsimulation model, along with funding assumptions, to generate shocks to the macrosimulation model that take the form of changes in government transfers. In one scenario we explore the case where the policy change is unfunded with the region: there is external funding – perhaps by a central government committed to “levelling up” a “left behind” region. At least in the Scottish case, however, a more realistic scenario assumes that the increase in transfers to households has to be funded by an increase in income tax.Footnote ³ This fiscally neutral case produces estimates of the impact of the policy change on the wider economy, registering changes in, for example, Gross Domestic Product (GDP), employment, unemployment, capital stock, population, real wages, and CPI.

Discussion/Conclusion

As part of the Child Poverty Act 2017, the Scottish Government has committed to reducing the child poverty rate to 10% by 2030/31. The purpose of this paper is to explore the extent to which the Scottish Government would have to adjust the policy instrument that is most directly linked to its policy target of child poverty (and over which it has direct control, requiring no input from the Westminster Government) – namely, the Scottish Child Payment (SCP) – to meet these targets.

Many previous studies (e.g. Bäckman and Ferrarini, Reference Bäckman and Ferrarini2010; Nygård et al., Reference Nygård, Lindberg, Nyqvist and Härtull2019) take a microsimulation approach to analyse the effects of cash transfers on child poverty. In this paper we use a combination of micro and macro simulation models to analyse the effectiveness of the Scottish Child Payment in reaching the final target. This novel combination allows us to analyse the detailed distributional effects of the significant policy change, while also capturing its likely macroeconomic impacts. The microsimulation model is based on the Institute for Public Policy Research’s Tax and Benefit model, with outputs feeding into the AMOS modelling framework for macroeconomic analysis.

Acting as a direct cash transfer to households, we find that the 2030/31 Scottish child poverty can be met with a very large SCP payment of £165 per week – a very significant increase from its current level. It is very clear that, while the Scottish Government can in principle achieve its target through this mechanism, the scale of the required policy adjustment is dramatic. This increase in payment would require an additional £3 billion tax revenue to be raised in the realistic case of it having to be funded internally by the Scottish Government. In our modelling we assume the additional tax is raised through a 4 percentage point increase in income tax rates across all income bands.

From the macrosimulations we estimate the potential economy-wide impact of the policy, using the results of the microsimulation to identify the scale of the fiscal change. If the policy is funded externally, it has unambiguously positive effects on the economy. However, once we impose the restriction that the policies are funded through across the board increases in income tax rates – the mode of raising revenue to offset their costs most easily available to the Scottish Government – we find that, with inflexible wages, the consequences for economic activity are negligible in size, but positive. However, if wages are very sensitive to changes in income taxes we find that fiscally neutral Scottish Child Payment policy changes tend to have negative consequences for economic activity: the induced wage-push effect dominates any stimulus to demand. The macroeconomic outcomes of the policy are inversely related to the responsiveness of wage bargaining to the income tax changes required to fund the SCP.

While we demonstrate that the ambitious Scottish 2030/31 child poverty targets can be achieved in principle with sole use of the SCP policy instrument, this can only be done by incurring a very high fiscal cost. One key result of our analysis suggests that combining the SCP with other policies may provide a more promising route to achieving child poverty targets. Specifically, we find a non-linear response between the SCP and child poverty rates. Above payments of £80 per week there is a significant reduction in the marginal gain in child poverty rates for each additional £10 spend on the SCP. Accordingly, it becomes an increasingly expensive method of lifting children out of poverty, echoing the finding in Bezze et al. (Reference Bezze, Canali, Geron and Vecchiato2020) that high levels of cash transfers can be an ineffective way of combating child poverty. It may well be that other policy instruments such as those impacting rentals become a more efficient way of reducing child poverty than ever higher levels of SCP.

It is well known that there are certain circumstances, such as lone parent household or disability in the family, that increase the chances of a child living in poverty. One approach, for which the existing literature provides some support (Bárcena-Martín et al., Reference Bárcena-Martín, Blanco-Arana and Pérez-Moreno2018), is to reduce the costs of meeting child poverty targets by adopting a more targeted approach to direct transfers. So instead of a universal increase in SCP, other payments could be introduced directly to households with characteristics that are known to have a higher probability of being in poverty. Targeting these households should reduce the cost per child out of poverty.

In addition to direct transfers, there are other policy levers that may be used in conjunction with the SCP to meet the targets similar to the argument made by Zagel and Van Lancker (Reference Zagel and Van Lancker2022). The significant expense of childcare is a key barrier to work with many parents not able to work or working reduced hours, increasing the possibility of children being in poverty. Increased childcare provision would allow parents to either enter the workforce or increase the number of hours they can work. The effectiveness of such policy levers, and combinations of them, for Scotland should be explored in future work.