Research and analysis

Technical Report: Impact of Tax-Free Childcare on the labour market participation of working parents

Published 29 January 2026

Prepared by Ipsos for HM Revenue and Customs

Research report number: 843

March 2025

The views in this report are the author’s own and do not necessarily reflect those of HM Revenue and Customs

1. 1. Quantitative research

1.1 1.1 Introduction

This chapter provides technical details for the quantitative strand of research commissioned by HMRC to understand the impact of the Tax-Free Childcare (TFC) offer on the labour market participation of working parents and how this varies between families of different characteristics.  

It includes information about the sampling approach, recruitment, fieldwork and survey materials. 

1.2 1.2 Research overview 

The aims for this research were to understand: 

• the main factors in parents’ decisions about how many hours to work in a week 

• whether TFC users had any typical work or household characteristics and how these compare with parents not using TFC 

• the retention impact of TFC on the working patterns of parents 

• how TFC interacts with the 30 hours Free Childcare for Working parents (FCWP) offer and the possible impact of the expansion of this offer to nine-month-olds in September 2024   

• the role TFC plays in supporting sub-groups of interest 

The sub-groups of interest for this research were: 

• shift workers 

• single parents 

• parents of disabled children 

• parents of school-aged children 

• parents who are self-employed 

• parents living in Northern Ireland  

Shift workers were defined as those working outside 9am to 5pm on weekdays. They were a sub-group of interest for this research as these parents were expected to be less able to use alternative offers such as FCWP because nurseries tend to operate around 9am to 5pm on weekdays. 

Single parents were a sub-group of interest because they might rely more heavily on formal childcare. 

Parents of children with a disability were a subgroup of interest because they are expected to have additional barriers and needs relating to their child. They also receive double the TFC award in recognition of the higher level of needs of their children. 

Parents of school-aged children were a sub-group of interest because TFC may be the only childcare costs support offer they can use for these children. FCWP is only available for pre-school-aged children. 

Parents who are self-employed were a sub-group of interest because they may have different childcare needs and arrangements to employed parents. 

Parents living in Northern Ireland were a sub-group of interest because they were unable to benefit from new childcare offers due to the absence of an Executive. They may also recently have been impacted by the introduction of the Northern Ireland Childcare Subsidy Scheme (NICSS) in September 2024, which was part way through the fieldwork period.  

NICSS provides a 15% subsidy towards parents’ childcare bills (paid directly to providers registered with the scheme). Any parent receiving TFC, for a child who has not yet enrolled in primary school can apply for at least 12.5 hours per week of funded pre-school education. 

In total, we conducted 3,636 interviews with parents. 3,049 interviews were with parents who used TFC, of which 2,109 were TFC only users and 940 used TFC alongside a FCWP offer. We completed 587 interviews with parents who were eligible for TFC but did not use it at the time of the research. Of these interviews 438 were with non-users of TFC or FCWP and 149 interviews were with FCWP only users.  

1.3 1.3 Sampling 

There were four audience groups for this project and several sample sources were used. For the purposes of analysis, survey respondents were allocated to a group based on their answers to screening questions in the survey, even if this was different from the sample they were sourced from. 

All participants were screened at the beginning of the survey to ensure they were still eligible for TFC at the time of taking the survey. This included questions about income, age of children, health status of children, UK citizenship, and parental employment status. 

1.3.1 Users of TFC only 

Sample provided by HMRC from their database of TFC account holders. This sample included parents’ names, email addresses, telephone numbers, number of children, age of oldest child, and amount added to their TFC account. 

The TFC-only users sample size was 10,271 records. 

1.3.2 Users of TFC and FCWP 

Sample provided by HMRC from their database of TFC account holders and holders of Free Childcare for Working Parents (FCWP). These samples included parents’ names, email addresses, telephone numbers, number of children, age of oldest child, and amount added to their TFC account. 

The TFC and FCWP users sample size was 16,845 records. 

1.3.3 Users of FCWP only 

Sample provided by HMRC of FCWP account holders. This sample included parents’ names, email addresses, telephone numbers, number of children, age of oldest child, and amount added to their TFC account. 

The FCWP-only sample size was 14,298 records. This consists of 3,089 records in the original sample and 11,209 records in a reserve sample that was provided toward the end of fieldwork in order to increase responses. 

1.3.4 Users of neither TFC nor FCWP 

For the first pilot phase, a free-found sample purchased from Datascope was used during the pilot phases consisting of parents living in the UK with children aged 11 or under and with incomes of under £99,999 per parent. These characteristics were chosen as they are likely to indicate eligibility for TFC.  

This sample included parents’ names, telephone numbers, income, and age of children. This sample was found to be low quality; people contacted either claimed they did not have children or that their children were older than 11 years, and therefore ineligible for TFC so ineligible for the research. 

Due to this low sample quality, a different sample source for non-users of neither TFC nor FCWP was used for the second pilot and mainstage phases of the research. This sample source was recontacted participants from Ipsos’ Omnibus survey, which is a Ipsos’ nationally and regionally representative panel of 1,000 to 2,000 adults, with pre-collected background information.  

This sample consisted of parents living in the UK with children aged 11 or under and with incomes of under £99,999 per parent. These characteristics were chosen as they are likely to indicate eligibility for TFC. This sample included parents’ names, telephone numbers, income, and age of children.  

The non-TFC and non-FCWP user sample size was 2,400 records. 

1.4 1.4 Survey and questionnaire development 

The questionnaire started with screening questions to ensure respondents were eligible for TFC. The remaining questions covered:  

• background information 

• impact of TFC on parents’ labour market participation 

• use and impact of other childcare support offers  

Questions about background information included parents’ current childcare costs, what other support offers they used, what childcare they currently used, their work characteristics, and other demographic information.

Questions on the impact of TFC on parents’ labour market participation included things such as impact on parents’ work hours, work characteristics, household finances, and impact of hypothetical changes to the TFC offer on parents’ work hours.

Questions on the use and impact of other childcare support offers included the interaction between TFC and 30 hours FCWP for three to four-year-olds, and the anticipated impact of the expanded 15 hours FCWP for nine-month-olds on work hours and use of TFC.

Respondents were routed through the questionnaire, so they were only presented with questions that were relevant to them. 

The questionnaire was cognitively tested between 26 February to 1 March 2024. Ten cognitive interviews were conducted with participants with different work, household, and childcare characteristics.  

The purpose of the cognitive testing phase was to test that questions and their answer options were relevant and appropriate to the research. It also identified any difficulties participants may have in answering the survey. This included testing participants’ comprehension of questions, such as the ease or difficulty of recalling the information that questions were asking about, and the extent to which questions and possible question answers covered all circumstances.  

1.5 1.5 Pilot fieldwork 

1.5.1 Cognitive testing: 26 February to 1 March 2024 

Ten cognitive interviews were conducted via MS Teams or telephone. These interviews were held with both TFC user and non-TFC users in order to test questions intended for both audiences. 

Changes were made to the survey after the cognitive testing phase, such as removing or completely rewording difficult-to-interpret questions and lengthy questions that were laborious to comprehend and answer.  

1.5.2 First pilot phase: 2 to 17 April 2024 

Forty completed interviews were achieved. Twenty-seven of these were Computer-Assisted Telephone Interviews (CATI) and 13 were online.   

Fifty telephone interviews were planned, and 350 web survey invites sent out to test the survey and response rates. Due to poor quality of the sample, only 27 CATI interviews were achieved. This had a few potential causes. The non-user purchased free-find sample had an eligibility rate of 0.04% and the TFC user and FCWP user samples from HMRC and DfE had an eligibility rate of 38%.  

The survey length was longer than expected when administered over the phone (24 minutes rather than 20 minutes). The timing of this first pilot over the school closures during the Easter bank holidays 2024 may have impacted response rates as the target audience for this survey was parents of young children, who are likely to be busy at this time.  

These findings were reported to HMRC, and a decision was made to test the survey again after making major revisions to it in order to test the impact of these changes and re-test the response rate at a time not during school closures. 

The major changes made were aimed at improving comprehension and response rates. They included removing questions to reduce the overall length of the survey, rewording questions, and reformatting questions to ask the question of the respondents and their partner, if they had one, within the same question to aid flow and reduce completion time. 

A decision was also made to source further potential non-user participants from Ipsos’ Omnibus survey panel rather than purchase more sample from a sample supplier. This was done to increase eligibility and was expected to be more fruitful because the age of children, income, and working status of Omnibus panellists is regularly updated. 

1.5.3 Second pilot phase: 26 April to 5 May 2024 

Sixty-nine interviews were achieved, consisting of 35 CATI interviews and 34 online interviews. 

Seven-hundred web survey invites were sent out to test major changes made to the survey after the first pilot phase. 

This second pilot phase provided reassurance on the level of response and co-operation from participants, so it was decided the survey was ready to deploy on a large scale in mainstage fieldwork.  

Minor changes to the survey were made after the second pilot phase, such as small text amendments to shorten questions even further and rearranging questions so they were in a more logical order for respondents to answer.  

Because these changes were minor, respondents who completed the survey during the second pilot phase were included in the total number of overall completed interviews and analysed as part of this research. This is because they answered the same questions as respondents who completed during the mainstage, so the data from the second pilot and mainstage fieldwork are comparable.  

1.6 1.6 Mainstage fieldwork 

1.6.1 Mainstage fieldwork: 25 July to 3 October 2024 

A total of 3,568 completed interviews were achieved, consisting of 1,321 CATI interviews and 2,247 online interviews. 

CATI fieldwork began on 25 July 2024 targeting non-users from the Ipsos Omnibus recontact sample. The average CATI interview lasted 19 minutes. 

The online survey was launched on 26 July 2024. Invites were sent to 29,079 participants with an email address. 

We maximised response rates by: 

• sending branded advanced communications to introduce the research and to reduce concern about the legitimacy of the research 

• sending email reminders to people who had not completed the survey within one week then sent further reminder emails when responses began to slow down over time 

• sending SMS reminders as well as email reminders to maximise the chance that people would see the communication 

• offering the option to do the survey online or by telephone 

• reviewing daily updates of numbers of completed interviews in order to provide early warning of low response rates 

• offering convenient times to do a telephone interview such as evenings and weekends to accommodate for the schedules of working parents

1.6.2 Completed interviews 

1. Overall, 3,638 completed surveys were achieved. One person started the survey during the first pilot phase and finished during the mainstage, so their answers were removed from the data because the survey changed in key ways since the first pilot. 

2. Table 1, Table 2, and Table 3 below contain the numbers of completed surveys by mode and audience type. These are based on the 3,637 completed surveys that were used during data analysis.

Table 1: Number of completed surveys by mode

Mode	Number of completed surveys
Online	2,281
Telephone	1,356

Table 2: Number of completed surveys by user type

User type	Number of completed surveys
TFC user (TFC-only or uses TFC and FCWP)	2,839
Non-TFC user	798

Table 3 : Number of completed surveys by audience type

Audience type	Number of completed surveys
TFC-only user	1,411
TFC and FCWP user	1,428
FCWP-only user	453
Non-user of TFC nor FCWP	345

1.6.3 Mainstage fieldwork response rates 

At the time of the survey, large proportions of respondents in all sample sources were in a different audience type than indicated in the sample, so response rates are indicative and should be interpreted with extreme care.  

This was particularly so for non-users of TFC in the sample, of which a substantial proportion were users of TFC, FCWP, or both at the time of the survey. One possible explanation for this is people starting or ceasing to use certain offers since the samples were drawn. 

Toward the end of research, a quota was added to the CATI dialling to stop the interview if the respondent was a TFC only user as the numbers achieved for TFC only users far exceeded the target for this group. Therefore, more people ‘responded’ to the survey than are included in these calculations, having been screened out.  

Table 4 below shows the ‘completion rate’ which is based on the number of completed interviews divided by sample available for each audience type.  

Table 4 below details the counts of completed interviews and completion rate for each of the four audience types during the mainstage fieldwork. Figures for completed interviews are taken from participants’ answers to the survey and figures for sample size are taken from the sample.

Table 4 : Number of completed interviews and indicative completion rate for mainstage survey only

Group	Completed interviews (mainstage)	Sample size	Target	Completion rate (mainstage)
TFC-only	2,074	10,271	1,125	20%
TFC and FCWP	922	16,845	1,375	5%
FCWP-only	145	14,298	150	1%
Non-user	427	2,400	320	17%
Total	3,568	46,708	2,970	8%

1.7 1.7 Weighting and data processing

1.7.1 Weighting 

Weighting was carried out on the data to ensure the statistics reported were as representative as possible of the population of TFC users and non-TFC users. 

Separate weighting schemes were used for TFC users (‘TFC only users’ and ‘TFC and 30 hours users’) and for non-TFC users (‘30 hours only users’ and ‘non-users’). 

The TFC users weighting profile was taken from HMRC’s dataset on TFC account holders dated June 2024. TFC user data was weighted on the following variables: parents with a child with a registered disability, households with a self-employed parent, and region of the UK they reside in. 

The non-TFC users weighting profile was taken from the Family Resources Survey (FRS). The FRS is an annual report that provides facts and figures about the incomes and living circumstances of households and families in the UK. The non-TFC user data was weighted on the following variables: parents with a child with a registered disability, households with a self-employed parent, and region of the UK they reside in.  

The profiles were created to be the best approximation of the profile of eligible non-users based on the sources available. This profile was adults in the UK with at least one child aged 11 years old or under, not receiving benefits, where all parents in the household are employed, and where the individual income of each parent in the house does not exceed £99,999. 

The weighting efficiency for users was 93.3%. The Rim Weight is outside of the typical range for households with a disabled child and for Northern Ireland, but this was expected as these respondents were oversampled as they were from key subgroups of interest. 

The weighting efficiency for non-users was 85.7%. Because of low bases, figures were not reported on Scottish, Welsh, and Northern Irish non-users separately. These regions were combined into one weight for the ‘rest of UK’ equalling 13.34%, which improved the overall efficiency. 

The effective base size is reduced when weighting is added because the relative importance of answers of certain types of respondents is either increased or decreased in order to make the sample match the population as closely as possible. The weighting profiles for TFC users and non-TFC users can be found in Table 5 and Table 6 below.

Table 5 : Weighting profile for TFC users

Characteristic	Population figures (%)	Respondents: unweighted (%)	Respondents: weighted (%)
Child with a disability in the household	1.31	4.95	1.31
No child with a disability in the household	98.69	95.05	98.69
Household with at least one self-employed parent	14.02	15.05	14.02
Household with no self-employed parents	85.98	84.95	85.98
Age of youngest child is under 5	73.58	80.09	73.58
Age of youngest child is 5 or older	26.42	19.91	26.42
Northeast	3.50	3.38	3.50
Northwest	12.78	10.99	12.78
Yorkshire and The Humber	8.66	9.25	8.66
East Midlands	8.11	7.51	8.11
West Midlands	8.83	8.36	8.83
East of England	10.23	9.45	10.23
London	9.64	10.36	9.64
Southeast	16.37	15.64	16.37
Southwest	9.81	10.30	9.81
Wales	3.73	3.87	3.73
Scotland	6.00	6.56	6.00
Northern Ireland	2.34	4.33	2.34

Table 6 : Weighting profile for non-TFC users

Characteristic	Population figures (%)	Respondents: unweighted (%)	Respondents: weighted (%)
Child with a disability in the household	10.18	6.98	10.18
No child with a disability in the household	89.82	93.02	89.82
Household with at least one self-employed parent	12.31	18.06	12.31
Household with no self-employed parents	87.69	81.94	87.69
Age of youngest child is aged under 5	56.14	57.24	56.14
Age of youngest child is aged 5 or older	43.86	42.76	43.86
Northeast	3.31	3.92	3.31
Northwest	7.48	11.58	7.48
Yorkshire and The Humber	7.53	7.33	7.53
East Midlands	7.09	8.69	7.09
West Midlands	7.92	9.20	7.92
East of England	11.09	12.44	11.09
London	16.98	11.75	16.98
Southeast	17.04	18.40	17.04
Southwest	8.31	10.90	8.31
Rest of UK	13.25	5.73	13.25

1.7.2 Data processing 

Excel data tables were produced for each question in the survey and all cross-breaks were added to all tables. These cross-breaks included demographic information, household information, TFC use and impact, work characteristics, child characteristics, childcare use, and use of other childcare costs support offers. The data in these tables was weighted. 

An SPSS data file was also produced containing each variable and the weighting variable. 

Where a question asked for an answer regarding both the survey respondent and their partner, these answers were combined to produce a derived ‘household’ response variable. Many variations were calculated, such as whether both parents, one parent in a joint parent household, one parent in a single parent household, neither parent, or either one or both parents gave a certain answer. 

Statistical significance difference testing was performed on all subgroups to understand differences between subgroups and differences between each subgroup and the average. The test used was a z-test, which tests for significant differences between two proportions (percentage values).  

This test was used to determine whether the differences in proportions between subgroups were likely to be due to a genuine difference between groups rather than chance variation. The confidence level used for this research was 95%, meaning there is less than a 5% chance that results deemed significantly different differed due to chance. This is the standard level of confidence used in social sciences. 

Significance testing was not performed on comparisons involving subgroups with bases under 30 as these were deemed too small to make reliable inferences from. 

1.7.3 Coding 

Coding was applied to two questions that had open-ended answer codes. These codes were ‘other’ and asked participants to specify their answer in an open text box. These were a question that asked about barriers to using or difficulty using TFC, and a question that asked about reasons for using or potential reasons for future use of TFC. New codes were created on five or more instances of a similar response. Coded responses were included in data tables and SPSS data file. 

1.8 1.8 Interpretation of findings 

The 15 hours FCWP offer was expanded from two-year-olds to nine-month-olds from September 2024. This was towards the end of the survey period, so most parents who answered the survey answered questions about the expanded 15 hours FCWP for nine-month-olds offer based on what they expected to do rather than what they actually did. 

In the findings report, the survey data are rounded up to whole percentages. Therefore, in some cases, charts will appear to add to slightly more than 100%. For example, if the calculated estimates for a question are 20.2% and 20.4%, these percentages combined will be 40.6%, which will round to 41%. 

All sub-group differences that were reported in the presentation were statistically significant differences, with statistical testing relying on an assumption that the weighted sample achieved was as good as a random probability sample.  

2. 2. Qualitative research 

2.1 2.1 Introduction 

This chapter provides the technical and methodological details for the qualitative strand of research commissioned by HMRC to understand the impact of the TFC offer on the labour market participation of working parents and how this varies between families of different characteristics.  

This detail includes information about the sampling, recruitment, and fieldwork materials. 

The qualitative fieldwork was conducted between 14 October and 22 November 2024. 

2.2 2.2 Research overview 

The qualitative research aims aligned with the quantitative strand. These are outlined in section 1.2 in the quantitative chapter. 

2.3 2.3 Sampling and recruitment screening 

The qualitative sample was drawn from respondents of the quantitative survey who agreed to be recontacted. A total of 2,788 survey participants agreed to be recontacted, 77% of those who took part. 

Several primary and secondary quotas were set for fieldwork. These can be found in section 2.4. 

Participants were screened to check if their circumstances remained in line with the sample information. Participants were not sent an advanced email as they had opted-in to being recontacted about the research, with recruitment prioritising the primary quotas of the subgroups.  

2.4 2.4 Fieldwork and data management 

Sixty qualitative interviews were conducted. Forty-seven interviews took place with users of TFC, 13 interviews took place with non-TFC users. The interviews took place via Microsoft Teams or Telephone. The interviews took around 45 minutes to complete.  

Participants received a ‘Love2Shop’ voucher worth £40 as a thank you for taking part in the interview. Table 7 and Table 8 below show the primary and secondary recruitment quotas and the number of interviews achieved per quota.

Table 7 : Primary interview recruitment quotas and number of interviews achieved per quota

Characteristic	Description	Quota	Achieved
Non-TFC users	Non-Users of TFC	Min 10	13
TFC Users	Users of TFC	Max 50	47
Parents in Northern Ireland	Respondents in Northern Ireland	Min 8	10
Parents of children with a registered disability	Respondents who use it to fund childcare for a child with a registered disability	Min 6	6
Parents of children with additional needs, but not registered disabled	Respondents with a child they consider to have additional needs, but not registered disabled	Min 2	9
Self-employed parents	Respondents who are self-employed	Min 8	10
Single parents	Respondents who are single parents	Min 8	11
Shift workers	Respondents who are employed in work that has shift hours	Min 8	17
Parents of school aged children	Respondents with school aged child (4-11 and attending school)	Min 8	24

Table 8 : Secondary interview recruitment quotas and number of interviews achieved per quota

Characteristic	Quota	Achieved
Lives in Wales	Min 4	7
Lives in Scotland	Min 4	7
Lives in Northern Ireland	Min 10	10
Lives in England	Max 42	35
Two parents, both working full-time	Mix	38
Two parents, one full-time, one part-time	Mix	9
Two parents, both working part-time	Mix	2
Single parent working full-time	Mix	10
Single parent working part-time	Mix	1
Household income below £20,000	Mix	5
Household income between £20,001 and £50,000	Mix	34
Household income between £50,001 and £75,000	Mix	17
Household income between £75,001 and £99,999	Mix	3
Uses 15 or 30 hours FCWP	Min 15	19
Does not use 15 or 30 hours FCWP	Min 15	36
TFC helps parents work more	Min 10	14
Not applicable for those who do not use TFC	Not applicable	5
TFC does not help parents work more	Min 10	29
Male	Min 15	21
Female	Max 45	39

An analysis framework was used to compare responses across different participants, and within sub-groups, revealing patterns, contrasts, and relationships within the data. The framework consisted of a list of research themes.  

Qualitative notes were written into this analysis framework, including key quotes from participants. The framework approach was used to identify causal relationships, manage the complexity of the data, and focus analysis on the research objectives. 

2.5 2.5 Interpretation of findings 

Qualitative research approaches are used to shed light on why people hold particular views, or have particular experiences, rather than how many people have those views or experiences. These approaches are used to explore the nuances and diversity of experiences and the factors which shape or underlie them.  

The results are intended to be illustrative and explanatory, rather than statistically reliable. Parent types were sampled to provide a range of characteristics, situations and experiences and not intended to be statistically representative of the wider claimant population. 

It is not always possible in qualitative research to provide a precise or useful indication of the prevalence of a certain outcome or experience, due to the relatively small number of participants generally involved and because the sample is not intended to be statistically representative of the wider population. The findings reported here represent common themes emerging across multiple interviews.  

Sometimes, ideas can be mentioned a number of times in a discussion and yet hide the true drivers of experience. Additionally, a minority view can, in analysis, turn out to express an important view or trend. The value of qualitative work is to identify the issues which bear future investigation.  

In reporting the qualitative findings, we focus on exploring the breadth of experiences, and identifying the main themes, rather than the number of people who have expressed that thought. Any proportions used in qualitative reporting, such as ‘some’ or ‘many’, should always be considered indicative, rather than exact.

Where we pull out an example, insight or quote from one claimant, this is typically to illustrate findings that emerged more broadly across multiple interviews.