1. Approach to Calculating Take-up
Take-up is a measure of the extent to which people who are eligible for a benefit receive it. It is calculated by comparing the number of recipients of a given benefit to the population of people who are eligible for that benefit. We refer to those people who are eligible for a given benefit and receive it as benefit recipients. We refer to the total number of people who meet the eligibility criteria for a benefit, whether or not they claim it, as the eligible population.
Figure 1. Composition of the Eligible Population
Claim the benfit, Do not claim the benefit, Total Eligible Population
Take-up is expressed as a ratio (or percentage), where the number of benefit recipients is divided by the eligible population. Figure 2 is an infographic that is an illustrative example of how take-up is calculated (the figure of 70% is for illustrative purposes).
Figure 2. Illustrative Example of How Take-Up is Calculated
Eligible: Benefit Recipient, Eligible: Benefit Recipient or Non-Recipient, 70% Take-Up Rate
Figure 3 is a flowchart which summarises our approach to calculating estimates of take-up rates across a range of low-income benefits which we present in the second Take-up Strategy report.
Figure 3. Flowchart of Steps to Calculate Take-Up
- Use outturn Management Information data from Social Security Scotland regarding number of applications;
- Restrict the period which we cover to consider only successful applications that have observed a full application window;
- Sum to give number of benefit recipients;
- Use outturn demographic data on the population of interest e.g. the number of births, deaths or population estimates;
- Use microsimulation (or outturn data) to calculate the proportion of the relevant population that are eligible for a given benefit;
- Apply eligibility proportion to outturn demographic data to give the number of eligible people;
- Sum over the same time period as that used for eligible recipients to give eligible population;
- Calculate take-up rate as: Take-up rate = Eligible Recipients/Eligible Population.
Taking benefit recipients first, we often use Management Information (MI) data as it typically aligns better with our estimates of the eligible population. It can also help us to identify cohorts of clients that had a full application window to apply for the benefit, which is an important consideration when measuring take-up. This is because we cannot say that an eligible person has not taken up their eligibility until the full length of the application window has been observed. Therefore, we ensure that take-up is not misrepresented when we exclude from the calculation application data that do not meet this criteria. There are some limitations associated with using MI data. Generally, MI data are not always as robust and reliable as official statistics; more specific limitations associated with the MI data used for each estimate are explained in the relevant sections.
Whilst outturn data can be used to calculate the number of benefit recipients, we need to estimate the size of the eligible population for each of the benefits that we provide an estimate of take-up for in this strategy. We produce our own estimates of the size of the eligible population, rather than using those produced by the Scottish Fiscal Commission (SFC), where available. This is because the SFC focus on future years and typically publish eligibility estimates extending back to the financial year prior to that which their latest forecast is published in. For example, Scotland’s Economic and Fiscal Forecasts, published in August 2021, includes eligibility estimates from 2020-21 onwards. Our focus lies in a retrospective measure of the size of the eligible population in the period prior to 2020-21. Given this difference in focus, it would be more appropriate for the purposes of calculating take-up estimates to use our own estimates of the size of the eligible population. Box A provides a detailed overview of the steps followed to calculate the size of the eligible population for low-income benefits.
Box A. Steps to calculating the size of the eligible population for low-income benefits
There are two main steps we use to estimate the size of the eligible population. Firstly, we use outturn demographic data on the population of interest e.g. the number of births, deaths or population estimates (typically produced by National Records of Scotland).
The second step is to use microsimulation modelling to calculate the proportion of the relevant population that are eligible for a given benefit. Microsimulation applies tax and benefit rules to household survey data, which is representative of the population, and can be used to predict entitlement to low-income benefits. The Department for Work and Pensions (DWP) similarly incorporate microsimulation into their approach to estimating take-up of income-related benefits.
We have used the latest version (A2.51) of the microsimulation model ‘UKMOD’ in this analysis. This incorporates the tax and benefit rules from the UK Government’s Spring Budget in 2021, as well as the Office for Budget Responsibility’s (OBR) assumptions on the unemployment impact of COVID-19 from their Economic and Fiscal Outlook in March 2021. We also use the latest, 3 year pooled version of the Family Resources Survey (FRS) that is available for use in UKMOD, which includes data up to 2018/19. We have made a change to the default version of UKMOD A2.51, and that is to use the OBR’s latest Welfare Trends Report (March 2021) forecast of the rate of transition from legacy benefits to Universal Credit. We use UKMOD to estimate the percentage of people who are likely to be eligible for each benefit. These are known as our eligibility proportions.
An important limitation of UKMOD is the accuracy with which it predicts benefit caseload. When compared with outturn data, the forecast caseload is typically lower. The reason for this difference lies in the FRS data that UKMOD uses, which grosses up sample observations to yield estimates for the overall population. Grossing up sample observations to match population values is a process called calibration. For example, the FRS estimate for the number of children aged 0 to 9 will match ONS population estimates for the number of children aged 0 to 9. Calibration is a complex process that, in the FRS, accounts for demographic and geographic factors, but not benefit caseload. We account for this issue by multiplying the number of people that UKMOD predicts to be eligible for each benefit by adjustment factors. The adjustment factors are different for each benefit, but generally are equal to the ratio of outturn qualifying benefit caseload to UKMOD predicted qualifying benefit caseload, respectively for each benefit. These adjustment factors are calculated on a financial year basis and applied dependent on whether the predicted qualifying benefit is Tax Credits, Universal Credit, Housing Benefit or Pension Credit. This overall process improves the accuracy of our eligibility estimate.
The following sections, which cover each of the benefits that we provide an estimate of take-up, explain the methodology and associated limitations for calculating the take-up rate for the respective benefits in more detail. Box B provides an overview of key considerations of the take-up analysis more broadly.
Box B. Key considerations of the analysis
A key consideration for this analysis is that the take-up estimates are only initial as they rely on a methodology that is still in development stage and may change in future publications.
Some of the low-income benefits, like Scottish Child Payment, have not yet reached their ‘steady state’. This occurs when growth in the number of benefit recipients flattens, and the number of benefit recipients settles at its natural level. Calculating take-up prior to this is not incorrect from a methodological perspective – it would represent take-up of the benefit at that point in time. However, it could misrepresent the ‘natural’ level of take-up as we would expect this to increase until the steady state is reached.
The take-up estimates presented throughout this report should be treated as initial analysis of take-up and not Official Statistics. It would not be appropriate at this stage to attempt to badge this type of analysis Official Statistics because of the early stage we are at in developing our approach to estimating take-up and the methodology and the fact that the underlying data may change in future publications. However, the developed methodological approach and associated limitations have been made transparent and the spirit of the Code of Practice has been followed where possible.