Research and analysis

Analysis of the drivers of trust in HMRC

Published 28 May 2025

This report details the analysis of findings from the Trust and Fairness quantitative survey undertaken with over 3000 individual and small business taxpayers.

HM Revenue and Customs (HMRC) Research Report 794.

Research conducted by Verian (formerly Kantar Public) between April and May 2022.

Disclaimer: The views in this report are the authors’ own and do not necessarily reflect those of HMRC.

1. Glossary

1.1 Components and sub-components

The outputs of the factor analysis. Each component is a variable which summarises a set of sub-components of trust. The technical term used for these estimated variables is typically factors, throughout the report we have chosen to use the more intuitive term component.

1.2 Correlation

A statistical procedure that measures the strength of association between 2 variables.

1.3 Confidence intervals

A measure of the uncertainty around a population estimate that has been derived from a sample. Confidence Intervals come with a probability value, 95% is most usual, which reflects the degree of confidence that the (unknown) true population value lies within the interval.

1.4 Factor analysis

A statistical technique used to reduce a large number of associated variables into a smaller number of uncorrelated dimensions.

1.5 Imputation

Imputation in statistics refers to the procedure of using alternative values in place of missing data. Missing information can introduce a significant degree of bias and can make processing and analysing the data more difficult, which is why it is so important to mitigate. Regression models were used to impute the missing data points based on other more complete information provided to other questions.

1.6 Individuals, Small Businesses and Agents Customer Survey (ISBA)

An annual survey that measures customers’ experience of interacting with HMRC, their perceptions of compliance, and HMRC’s reputation for the 3 main customer groups (individuals, small businesses and agents).

1.7 Likert scale

Rating system used in questionnaires to measure attitudes, opinions, or perceptions. The survey on which this report is based relied particularly on a 5-point Likert scale, ranging from strongly disagree (1) to strongly agree (5), to measure agreement with various statements on customer attitudes and perceptions.

1.8 Ordinary Least Squares (OLS)

Ordinary Least Squares regression (OLS) is a common technique for estimating coefficients of linear regression equations which describe the relationship between one or more independent quantitative variables and a dependent variable.

1.9 Outcome variable

The variable that a regression model is explaining (based on the predictors).

1.10 Pay As Your Earn (PAYE)

The system of tax collection where an employer or pension provider deducts Income Tax and National Insurance contributions before they pay wages or pension and remits the amounts deducted to HMRC.

1.11 Predictor variables

The variables included in the regression model to explain/predict variation in the outcome variables.

1.12 Principal Components Analysis

A method that reduces the dimensionality of a dataset by transforming a large set of variables into a smaller set that contains most of the same information.

1.13 p-value

The probability of obtaining a result equal to or rarer than the result observed. It is used to determine whether a difference is significant, and if it is, to what level of confidence.

1.14 Regression modelling

A statistical technique used to model the relationship between an outcome variable and predictor variables.

1.15 Regression coefficients

The main output from regression modelling. For numerical predictor variables, the regression coefficient indicates the change in the outcome variable associated with a one unit increase in the predictor. For categorical predictor variables, the regression coefficient indicates the change in the outcome variable associated with each category in relation to a reference category.

1.16 Self Assessment (SA)

A system HMRC uses to collect Income Tax. Tax is usually deducted automatically from wages, pensions and savings. People and businesses with other income must report it in a Self Assessment tax return and pay any tax due to HMRC. Self-assessment may also be required in order to declare other sources of untaxed income, or because the individual meets certain criteria, such as earning over £100,000 or claiming child benefit with an income over £50,000. These types of taxpayers are referred to as non-business Self Assessment customers throughout this report.

1.17 Significance testing

A statistical test to determine whether relationships observed in the survey sample are likely to exist in the population from which the sample is drawn.

1.18 Weighting

Weighting is a process by which data is adjusted to reflect the known population profile. It’s used to balance out any significant variance between actual and target profile.

2. Executive summary

2.1 Introduction

In July 2020, HM Revenue and Customs (HMRC) and HM Treasury (HMT) published a 10-year strategy to build a trusted, modern tax administration system^{[footnote 1]}. International evidence shows that a trusted, modern tax system that makes it easy for taxpayers to pay the tax they owe is a cost-effective way of securing and sustaining future revenues. Trust is important because evidence suggests that increased trust in the tax system gives rise to increased and voluntary compliance^{[footnote 2]}. Greater levels of trust will likely strengthen the ability to successfully collaborate and make changes to modernise tax collection.

Overall levels of customer trust in HMRC are measured using the Individuals, Small Businesses and Agents survey. Such quantitative research does not allow an understanding of the factors that influence trust, hence HMRC commissioned a Trust and Fairness survey to identify quantitively which factors are most important in building trust and whether this varies by demographic and firmographic variables. This report details the analysis of findings from this survey, undertaken with over 3000 individual and small business taxpayers between March and May 2022.

The results of the research will empower HMRC to take an informed and targeted approach to building trust across different customer groups.

This study covered 3 customer groups:

1. Pay As You Earn (PAYE) customers
2. non-business Self Assessment (SA) customers
3. small business customers

2.2 Methodology

The sampling, weighting and data collection approaches varied between the customer groups, but a similar questionnaire and analytical approach was used for all groups. Fieldwork took place between 16 March 2022 and 27 May 2022. All PAYE customers and some non-business SA customers completed an online survey. Most non-business SA customers and all small business customers completed the survey over the telephone.

Interviews were carried out with 1,297 PAYE customers, 538 non-business SA customers and 1,313 small business customers.

The survey included questions on 30 sub-components of trust. Factor analysis and regression models were run on the survey data to identify the high-level components of overall trust and better understand the influencers of trust. This analysis has helped to understand:

• the key underlying components of trust that summarise a wide range of sub-components identified from previous research
• the importance of each component of trust in relation to overall trust, and differences according to customer characteristics
• what influences overall trust and each individual component of trust (for example, demographic or firmographic characteristics, experiences dealing with different taxes and HMRC, personal beliefs and exposure to information about HMRC), and how this differs by customer subgroup

Although this analysis measures the strength of association between overall levels of trust in HMRC, the key components of trust, and the influencers of trust, it cannot test the explanations or mechanisms behind the relationships identified in this report.

This report contains analysis on the relationship between survey answers, rather than reporting the survey findings for each question. Detailed responses to each question can be found in the data tables accompanying this report.

2.3 Underlying components of overall trust

The survey included 30 questions that reflected the sub-components of trust identified in academic literature and previous research. Factor analysis was used to summarise these 30 sub-components into a smaller number of underlying components. Five main components were identified, each describing a set of sub-components of trust that covered:

1. institutional fairness and reputation
2. transparency and clarity
3. competency of HMRC staff
4. how HMRC responds when things go wrong
5. use of power and lack of reciprocal trust

A regression model was run to understand how strongly each of these components was associated with overall trust. For each of the 3 main customer groups that were surveyed, perceptions of institutional fairness and reputation – component 1 – had the strongest association with overall trust. A change in the component describing institutional fairness and reputation was linked to a change in overall trust which was:

• twice as large as a change in the components describing transparency and clarity, or competency of HMRC staff
• four times as large as a change in the components describing how HMRC responds when something goes wrong, or use of power and lack of reciprocal trust

The force with which HMRC is perceived to use its power and the perceived lack of reciprocal trust – component 5 – had a negative association with overall trust, which means that higher levels reduce trust. The association of the other 4 components with overall trust was positive, which means higher levels increase trust.

The models identified certain demographic and firmographic characteristics associated with overall trust. For PAYE customers, institutional fairness and reputation had a stronger positive association for those who had no interactions with HMRC compared with those who had one or more. For non-business SA customers, competency of HMRC staff had a stronger positive association for those that had 5 or more interactions compared with those who had 4 or fewer.

For small businesses, the component that described institutional fairness and reputation had a stronger positive association with overall trust for those that outsourced at least some tax work to an agent, compared with those that did all tax work in-house. Also, among small businesses, the component that described competency of HMRC staff had a stronger positive association for those that did all tax work within the business.

2.4 Influencers of overall trust and the components of trust

Regression models were run to measure the associations between the influencers of trust and overall trust, as well as influencers of trust and the 5 identified components of trust.

Some of the influencers analysed in this research are related to customer experiences. Delivering positive customer experiences when dealing with HMRC was identified as one of the most important influencers of trust. For all 3 customer groups, a positive experience dealing with HMRC was associated with higher:

• overall trust in HMRC
• perceptions of institutional fairness and reputation (trust component 1)

This association was strong even among respondents who considered taxes difficult to deal with.

For all 3 customer groups, a positive experience when taxes were not considered difficult to deal with, and higher confidence in dealing with taxes, were both associated with improved perceptions of:

• transparency and clarity (trust component 2)
• competency of HMRC staff (trust component 3)

When errors are dealt with unsatisfactorily by HMRC there was a strong association with lower levels of overall trust for the PAYE and non-business SA customer groups. HMRC making errors and dealing with them unsatisfactorily was linked with poor perceptions of how HMRC responds when things go wrong for all 3 customer groups. The negative association was smaller for those who experienced errors that were perceived to have been dealt with well.

For all 3 customer groups, higher levels of agreement that HMRC goes after the wrong people had a negative association with:

• overall trust in HMRC
• perceptions of institutional fairness and reputation (trust component 1)
• perceptions of how HMRC responds when things go wrong (trust component 4)

A perception that HMRC goes after the wrong people was associated with use of power and a lack of reciprocal trust for all 3 customer groups.

2.5 Demographic influencers of trust

Some demographic characteristics were found to influence overall trust and the 5 components of trust, even after controlling for the effects of other variables.

For PAYE customers these differences were:

• being in paid employment was associated with worsened perceptions of transparency and clarity
• being placed on furlough was associated with increased levels of overall trust
• for those aged below 65 years old, each one-year increase in age associated with higher overall trust. This was particularly strong for those aged 16 to 24 years old
• receiving Child Benefit was associated with worsened perceptions of how HMRC responds when things go wrong
• being in receipt of Marriage Allowance was associated with improved perceptions of transparency and clarity

For non-business SA customers demographic differences included:

• having a higher qualification was associated with higher levels of overall trust
• an increase in age was found to be associated with lower levels of overall trust and worsened perceptions of institutional fairness and reputation

For small business customers the differences were:

• businesses that had a higher number of employees had higher levels of overall trust in HMRC, mainly driven by improved perceptions of the competency of HMRC staff
• running the business alone as a self-employed individual was associated with higher levels of overall trust, driven by improved perceptions of institutional fairness and reputation
• self-employed small businesses perceived lower levels of transparency and clarity, compared to those not self-employed
• higher education qualifications were significantly associated with worsened perceptions of institutional fairness and reputation, as well as lower perceived use of power

2.6 Applying the model of trust to real life scenarios

The model of trust derived in this research can be used to measure the impact that different ‘real life’ scenarios have on overall trust.

If customer satisfaction increases, customer trust in HMRC also increases across all three customer groups. In more detail if the proportion of unsatisfied PAYE or non-business SA customers (customers who gave the lowest overall scores for their experience with HMRC) reduced by half, 10% of customers (in the corresponding group) would move +1 point in the 5-point Likert scale used to measure trust in HMRC. The same reduction for small businesses would have a lesser impact on levels of trust (7% would move +1 point). If the proportion of unsatisfied customers reduced by 25%, the proportions of customers that would move +1 point in the Likert scale for overall trust would also halve.

A reduction in HMRC errors might have a higher positive impact on trust for Non-business SA customers than for the other two customer groups. In more detail, if customers’ experience of errors reduced by 50%, the impact might be higher among non-business SA customers (around 7% would move +1 point in the Likert scale for overall trust), than PAYE customers (circa 4%) and small businesses (1% to 2%).

Exposure to positive information about HMRC, for example, friends, family or the media, was associated with higher levels overall trust and improved perceptions of institutional fairness and reputation for all customer groups. If the tone of the information the customer was exposed to was rated +1 point higher, 10% of PAYE customers, 13% of non-business SA customers and 8% of small business customers would move +1 point in the 5-point Likert scale used to measure trust in HMRC.

The view that HMRC goes after the wrong people was associated with lower overall trust for all 3 groups. If all customers move 1 point down in the 5-point Likert scale for ‘HMRC goes after the wrong people’, 8% of PAYE customers, 12% of non-business SA customers and 7% of small business customers would move +1 point in the 5-point Likert scale used to measure trust in HMRC.

For PAYE customers in particular, confidence in dealing with taxes was associated with higher levels of overall trust. If confidence scores improved by +1 point in the 5-point Likert scale for PAYE customers, 14% would move +1 point in the 5-point Likert scale used to measure trust in HMRC.

3. Introduction

3.1 Research context

In July 2020, HMRC and HMT published a 10-year strategy to build a trusted, modern tax administration system^{[footnote 3]}. International evidence shows that a trusted, modern tax system that makes it easy for taxpayers to pay the tax they owe is a cost-effective way of securing and sustaining future revenues4. Trust is important because evidence suggests that increased trust in the tax system gives rise to increased and voluntary compliance^{[footnote 4]}. Greater levels of trust will likely strengthen the ability to successfully collaborate and make changes to modernise tax collection.

3.2 Research objectives

Previous international research using qualitative methods has identified various sub-components of trust and influencers of trust in tax authorities. HMRC also gathers quantitative data on overall customer trust broken down by various customer groups as part of the Individuals, Small Business and Agents (ISBA) Customer Survey^{[footnote 5]}. However, there is a lack of detailed quantitative evidence regarding trust in the UK tax administration, which this research aimed to address. Its core objectives were to:

• explore links between influencers of trust, components of trust, and overall trust in HMRC
• measure the relative contributions of the components of trust on overall trust and the relative contributions of a range of influencers on overall trust and its components
• understand how customer characteristics and experiences of HMRC impact the relative importance of components of trust and trust influencers
• identify the influencing factors with the greatest tractability and potential to improve trust in HMRC

This report does not provide descriptive information related to current levels of trust among HMRC customers. An overview of current levels of trust in HMRC among individual and small business customers can be found in the ISBA Customer Survey. Rather, this report quantifies the links between trust and its influencers. However, descriptive statistics generated from this survey data are included in the accompanying data tables.

3.3 Research methodology

This study covered 3 customer groups:

1. PAYE customers who paid income tax directly from their salary, wages, or pension
2. non-business SA customers who had completed a Self Assessment return, but who were not self-employed or in a business partnership
3. small business customers, comprised of those self-employed or in a partnership paying income tax via Self Assessment and registered small businesses, with fewer than 20 employees and an annual turnover of less than £10 million

The sampling, weighting and data collection approaches varied between the customer groups, but questionnaire design and the analysis approaches were similar. These methodological details are summarised in this section, but further detail is in the technical annex.

3.4 Sampling and weighting

PAYE customers

The sample of PAYE customers was generated by screening a representative sample of the general population from Kantar Public’s ‘Public Voice’ panel. In total, 1,380 interviews were completed with PAYE customers. Respondents with a large volume of missing data (for overall trust and the sub-components of trust) were excluded from the statistical modelling (see section A.3.1.1 for additional details). Following this exclusion, the modelling was based on 1,297 cases.

The screened sample was weighted to be representative of the general UK population. Individuals that were not eligible PAYE customers were removed from the survey dataset. This left a representative sample of PAYE customers. Full details on weighting are provided in the technical annex (section A.2).

Non-business SA customers

Sample for non-business SA customers was primarily drawn from HMRC administrative data. It was supplemented with a sample of customers identified through ‘Public Voice’. In total, 563 interviews with non-business SA customers were completed. Respondents with a large volume of missing data (for overall trust and the sub-components of trust) were excluded from the statistical modelling (see section A.3.1.1 for additional details). Following this exclusion, the modelling was based on 538 cases.

Weights were derived to ensure the sample was representative of the population of non-business SA customers. Full details on weighting are provided in the technical annex (section A.2).

Small business customers

Small business customers were drawn from the HMRC Self Assessment database and the Inter-departmental Business Register (IDBR). A small number of small businesses were sourced from previous participants of HMRC’s ISBA Customer Survey. In total, 1,391 surveys with small businesses were completed. Respondents with a large volume of missing data (for overall trust and the sub-components of trust) were excluded from the statistical modelling (see section A.3.1.1 for additional details). Following this exclusion, the modelling was based on 1,313 cases.

Weights were derived to ensure the achieved sample matched the estimated population profile. Full details on weighting are provided in the technical annex (section A.2).

3.5 Data collection

The survey questionnaire was developed by Kantar Public in collaboration with HMRC. Questions were cognitively tested prior to the start of fieldwork. The questionnaire was almost identical for the 3 customer groups. Differences between the questionnaires were to capture relevant demographic or firmographic questions appropriate to the customer groups. It took approximately 25 minutes to complete.

All Public Voice surveying was conducted online, while small businesses and the majority of non-business SA customers were interviewed using Computer Assisted Telephone Interviewing (CATI).

Telephone fieldwork was conducted between 16 March 2022 and 27 May 2022, and online fieldwork was conducted between 15 April 2022 and 10 May 2022.

3.6 Approach to analysis

Factor analysis and regression models were run on the survey data to derive the components of overall trust and better understand the influencers on levels of trust.

Firstly, factor analysis was conducted on 30 sub-components of trust (identified through previous desk-based and qualitative research) to summarise these into 5 components of trust. These 5 components were based on the survey data rather than existing theory or qualitative work.

Secondly, a regression model was run to examine how each of the 5 components of trust were associated with overall trust with socio-demographic variables used as controls. This stage of modelling helped to understand the importance of each component in relation to overall trust, as well as the demographics linked to differences in trust not explained by the 5 components.

Thirdly, regression models were run to explore the associations influencers of trust have with overall trust and each of its 5 components. This helped to understand the relative strength of aspects that influence trust and may help to provide focused areas for action in the future.

Missing data points were imputed prior to all modelling. Further details about the imputation approach used are included in the technical annex (section A.3.1.1).

Notes on regression modelling

Regression modelling is used to explore how different predictor variables are associated with an outcome of interest while keeping all other variables in the model fixed. This is particularly important when predictor variables are correlated (for example, the number of employees a small business has and its turnover). This approach allows us to test whether associations that would be identified in simple cross tabulations still hold once other variables are controlled for.

While regression modelling can be used to identify associations between predictor variables and the outcomes of interest, it is not possible to derive causality (with absolute certainty) from estimates produced by analysis of cross-sectional data. There may also be different forms of bias present that could impede the identification of the true association of predictors with the outcome of interest. Where associations between overall trust or the components of trust and influences or demographics have been identified, these will not necessarily all be causal, and some effects may be a result of other underlying components that we have not measured as part of this research.

3.7 Other reporting notes

Many of the questions in the survey to measure sub-components of trust asked customers to rate their customer experience and perceptions of HMRC using a 5-point Likert scale, where 5 was the most positive response and one was the least positive response. Respondents could also say ‘don’t know’ or ‘not applicable’ (‘not applicable’ was only an option at selected questions. For example, if the customer was asked to rate aspects of their experience dealing with HMRC, they could say the question was not applicable if they had no experience).

Where customers rated HMRC, they were asked to give an overall opinion about all their experiences of dealing with HMRC over the previous 12 months.

All differences reported in this report are statistically significant. Differences identified through regression modelling are reported at either the 90%, 95% or 99% confidence level. However, differences reported in the descriptive statistics are statistically significant to the 95% confidence level.

4. Underlying components of overall trust

4.1 Introducing the 5 components of trust

Deriving the components

The survey questionnaire included 30 questions reflecting the sub-components of trust identified in academic literature and previous research. Some of these questions were highly related to one another. Using questions or variables that are highly correlated with each other would yield unreliable (likely biased) results. To prevent this, factor analysis was conducted to understand the relationships between sub-component questions and to summarise them into 5 unrelated variables that could be used for modelling – the 5 components of trust. Further details on the factor analysis can be found in the technical annex (section A.3).

Sub-components of trust described by each component

Principal Components Analysis was used to determine the number of components. Results are included in Table 1.2 of the accompanying regression output tables. A 5-component solution was chosen. A sixth component was identified for the non-business SA customer group which described the statements ‘I would be expected to respond to HMRC quicker than they would respond to me’ and ‘HMRC ensures that my data and personal information is treated confidentially’. However, this component was not found to be significantly associated with overall trust and was excluded in further analysis.

Analysis was conducted independently for each of the customer groups, but the outputs were similar and can be summarised together for each component. While some sub-components were correlated to more than one component of trust, in this section each variable is associated with only one to ease interpretation. The full output tables from the factor analysis are included in Table 1.1 of the accompanying regression output tables.

Institutional fairness and reputation (component 1)

The first trust component, institutional fairness and reputation, describes the sub-components of trust that centre around ‘fairness’, ‘respect’, ‘responsible’ and ‘reputation’. The sub-components with the strongest associations with this component were broadly similar for all 3 customer groups. They were (in order):

1. HMRC treats customers fairly
2. HMRC is an organisation with a good reputation
3. HMRC applies penalties and sanctions equally for all of its customers
4. HMRC is an organisation that people respect
5. HMRC acts responsibly when carrying out its duties
6. HMRC ensures all of its customers pay or receive the correct amount of tax
7. HMRC treats you as honest
8. HMRC would take account of my circumstances when applying penalties and sanctions

Transparency and clarity (component 2)

The transparency and clarity component describes the sub-components of trust related to the way HMRC communicates with customers and the information provided to make it easy to deal with tax affairs. The sub-components with the strongest associations with this component for all 3 customer groups were (in order):

1. HMRC communicates in a way that is easy for me to understand
2. HMRC makes it easy for me to meet my tax obligations
3. HMRC provides a service that is tailored to my needs
4. HMRC provides me with useful and relevant information

Competency of HMRC staff (component 3)

The competency of HMRC staff component describes the sub-components of trust related to the interactions customers have with HMRC, and particularly questions regarding HMRC staff. The sub-components with the strongest associations with this component for all 3 customer groups were (in order):

1. HMRC staff are professional
2. HMRC staff are knowledgeable
3. HMRC staff are friendly
4. HMRC ensures that my data and personal information is treated confidentially

How HMRC responds when things go wrong (component 4)

This component describes the sub-components of trust that relate to the way HMRC responds if a mistake or error is made. The sub-components with the strongest associations with this component for all 3 customer groups were (in order):

1. HMRC would admit if they made a mistake
2. HMRC would apologise if they got things wrong
3. HMRC would put things right if they made a mistake
4. HMRC would support me if I made a mistake

Use of power and lack of reciprocal trust (component 5)

This component describes the sub-components associated with the force with which HMRC uses its power and is also linked to the reciprocity of trust in the relationship between HMRC and the customer. The sub-components with the strongest associations with this component for all 3 customer groups were (in order):

1. HMRC uses its powers in a very forceful way
2. HMRC tries to catch people like me out

4.2 Modelling associations between the components of trust and overall trust

A model was run to understand how each of the 5 components of trust was associated with overall trust with socio-demographic variables used as controls. This analysis helps to understand the importance of each component in relation to overall trust. It also identified the characteristics of customers with remaining differences in levels of trust after accounting for individuals’ views on the components of trust.

An Ordinary Least Squares (OLS) regression was conducted. Modelling was repeated separately for each of the 3 customer groups. Respondents with high levels of missing data (overall trust and sub-components of trust questions) were excluded from the analysis, and lower levels of missing data were imputed, as detailed in the technical annex (section A.3.1.1).

The models without controls explained a lot of the variation in the overall trust variable and there was only a small improvement in explanatory power after adding a range of control variables. Most of the variation in levels of overall trust could be explained by the 5 components summarising the sub-components of trust. The full list of the control variables used in the second version of the models can be found in the technical annex (Table A.14). They include socio-demographics such as age, gender, ethnicity, working status and income, as well as firmographic data including number of employees, turnover, and length of time trading.

The output for each model was a set of coefficients which indicated the strength of the relationship between the predictor variables and overall trust. For a categorical predictor variable, the coefficient represents the difference in the predicted value of overall trust between the category for which the predictor variable is equal to 1 (for example, Yes) and the category for which the predictor variable is equal to 0 (for example, No). For a continuous predictor variable, the coefficient represents the difference in the predicted value of overall trust for each one-unit change in the predictor variable, assuming all other predictor variables are held constant.

Only significant results have been reported, which means that variables not included in the findings were not found to have a significant association with overall trust. The full data tables displaying coefficients for all variables and the two versions of the model can be found in Table 2.1 of the accompanying regression output tables.

4.3 Importance of each component in relation to overall trust

Table 4.1 shows the regression coefficients for each of the components that describe the sub-components of trust. The control variables were also included in this version of the model.

For all customer groups, the component that described institutional fairness and reputation had the strongest association with higher levels of overall trust. This was followed by the components describing transparency and clarity, and competency of HMRC staff. How HMRC responds when things go wrong and its use of power had a lower, albeit still significant, association with trust. The component that described the force with which HMRC uses its power and the lack of reciprocal trust had a negative association with trust. This means that as the predictor values increased, trust in HMRC decreased.

A one-point change in the institutional fairness and reputation component was linked to a change in overall trust twice as large as one-point changes in either the transparency and clarity or competency of HMRC staff components. It was 4 times as large than a one-point change in the components describing how HMRC responds when something goes wrong, or use of power and lack of reciprocal trust.

Figure 4.1 Regression coefficients for each component of trust by customer group

Base: PAYE customers (1,297), non-business SA (538), small business (1,313).

R-squared: PAYE customers (0.680), non-business SA (0.631), small business (0.522).

Significance levels: p<0.10, **p<0.05, **p<0.01

4.4 Additional analysis by sub-groups

The models identified certain demographic and firmographic characteristics that are associated with overall trust, outside the 5 components analysed in section 4. Subsequent regression modelling was used to explore whether the relationship between overall trust and the components of trust varied between different customer sub-populations, based on the number of interactions they had with HMRC and whether they had agent representation. These subpopulations were chosen based on differences observed in the descriptive statistics. Further, examination of differences according to number of interactions was based on a theoretical proposition. The suggestion was that more direct contact with HMRC gives taxpayers more information to form their belief on trust, while fewer interactions mean they may be more affected by external influences.

Findings are reported in this section by customer group. The output tables containing coefficient estimates from the regressions can be found in Tables 2.2 and 2.3 of the accompanying regression output tables.

PAYE customers

The PAYE customer group was split into two categories based on the level of contact they had with HMRC over the last 12 months (no interactions or one or more interactions), and the models were rerun for each group. The analysis found that component 1, which described the sub-components of trust centred around ‘fairness’, ‘respect’, ‘responsible’ and ‘reputation’, had a stronger positive association for those who had no interactions compared with those who had one or more (see table in accompanying regression output tables).

No significant differences in the importance of the 5 components of trust were identified when comparing PAYE customers who handle at least some of their own tax affairs in-house compared to those who outsource tax affairs completely.

Non-business SA customers

The non-business SA group was split into two categories based on the level of contact they had with HMRC over the last 12 months (0 to 4 interactions or 5 or more interactions), and the models were rerun for each group. The analysis found that component 3, describing competency of HMRC staff, had a stronger positive association for those that had 5 or more interactions compared with those who had 4 or fewer (see table in accompanying regression output tables).

No significant differences in the importance of the 5 components of trust were identified when comparing non-business SA customers who handle at least some of their own tax affairs in-house compared to those who outsource tax affairs completely.

Small business customers

The small business customer group was split into two categories based on their use of an agent to deal with taxes; all tax work done within the business or outsourcing at least some tax work.

The analysis found differences for two of the components. Component 1, which described the sub-components of trust around ‘fairness’ and ‘reputation’, had a stronger positive association with overall trust for those that outsource at least some work, compared with those that do all tax work in-house (see Table 2.3 of the accompanying regression output tables).

Component 3, which described the sub-components of trust related to competency of HMRC staff, had a stronger positive association for those that do all work within the business (see Table 2.3 in accompanying regression output tables). For those that do all work within the business, it is the second most important component (after component 1 relating to institutional fairness and reputation), whereas for those that outsource at least some work it is the third most important component (after component 1 relating to institutional fairness and reputation, and component 2 relating to transparency and clarity).

No significant differences in the importance of the 5 components of trust were identified when comparing small businesses with lower contact (0 to 4 interactions) to those with more frequent contact (5 or more interactions). Similarly, no significant differences were identified when comparing self-employed and not self-employed within the small business group.

5. Influencers of overall trust and the components of trust

5.1 Modelling associations between influencers and trust

Statistical models were run to measure the associations between the influencers of trust and overall trust, as well as influencers of trust and the 5 components of trust (introduced in section 4.1). This analysis helps to understand what influences trust overall, each component of trust individually and may help to provide focused areas for action and further research. The analysis is exploratory, and the findings will support HMRC to anticipate the impact of any changes to services it provides.

Summary of the modelling approach

OLS regression modelling was used to explore how the influencers of trust are associated with overall trust in HMRC and each of the 5 component scores summarising the sub-components of trust.

Modelling was repeated for each of the 3 customer groups. Respondents with high levels of missing data (overall trust and sub-components of trust questions) were excluded from the analysis, as detailed in the technical annex (section A.3.1.1).

Two versions of each model were produced. The first version included just the influencer variables, and the second version included a range of control variables (demographics for individuals, and firmographics for small businesses). The control variables used in the model were the same as those used in the model to understand how each of the 5 components of trust were associated with overall trust. They are listed in Table A.14.

Influencers of trust included in the models

The predictor influencers were grouped into 8 categories:

1. behaviours and attitudes towards compliance, including risk aversion
2. attitudes around social norms on compliance, including perceived compliance and the perception that HMRC goes after the ‘wrong’ people
3. perceived opportunities and risks of non-compliance
4. customer experience, including dealing with errors made by HMRC
5. the level and mode of interaction with HMRC, including contact channels used and the number of interactions with HMRC
6. exposure to the opinions of others, including sources of influence and how positive those opinions were
7. having the right understanding of HMRC’s functions, including HMRC’s role in setting tax rules and administering COVID-19 schemes
8. tax literacy and confidence, including use of professional help when dealing with taxes

Further detail on how each of these influencers was derived using survey data is in the technical annex (section A.4).

5.2 Influencers associated with higher levels of overall trust and improved perceptions of the components of trust

A table showing the complete numerical results of the analysis covered in this section can be found in the Table 3.1 of the accompanying regression output tables.

Customer experience

A positive customer experience when dealing with HMRC, even when taxes were difficult to deal with, was associated with higher levels of overall trust for all 3 customer groups (PAYE, non-business SA and small businesses). It was also associated with improved perceptions of institutional fairness and reputation (component 1) for all 3 customer groups, improved perceptions of transparency and clarity (component 2) for PAYE and non-business SA, and improved perceptions of competency of HMRC (component 3) for PAYE and small businesses. For small businesses it was also associated with improved perceptions of how HMRC responds when things go wrong (component 4).

Among customers who did not find taxes difficult to deal with, a positive customer experience was also associated with improved perceptions of institutional fairness and reputation (component 1) for small businesses which are not self-employed, while an association with transparency and clarity (component 2) was identified for self-employed small businesses only. Competency of HMRC staff (component 3) was linked to positive customer experiences for non-business SA who did not find taxes difficult to deal with.

Among those with a negative experience, finding taxes difficult to deal with was not directly associated with overall trust but was associated with improved perceptions of transparency and clarity (component 2) among non-business SA customers and small business customers. For small businesses it was also associated with lower levels of perceived use of power (component 5).

Behaviours and attitudes towards compliance

Among PAYE customers and small businesses, a more compliant attitude towards tax was associated with improved with overall trust as well as with improved perceptions of transparency and clarity (component 2) and competency of HMRC staff (component 3). Among small businesses the association with perceptions of transparency and clarity (component 2) was only significant among those who were not self-employed.

For PAYE customers a more compliant attitude also was associated with improved perceptions of institutional fairness and reputation (component 1). Among non-business SA customers, and small business customers who were self-employed it was associated with a perception of lower levels of use of power and lack of reciprocal trust (component 5).

Tax literacy and confidence

Levels of confidence in dealing with taxes was associated with higher levels of overall trust for PAYE customers and small businesses, and improved perceptions of transparency and clarity (component 2) for all 3 customer groups. Among PAYE customers, levels of confidence in dealing with taxes was associated with improved perceptions of institutional fairness and reputation (component 1) and competency of HMRC staff (component 3).

For PAYE customers low levels of confidence were mitigated by having someone else, such as an accountant, friend or relative deal with their tax affairs, which was associated with a higher overall trust rating. This was also associated with improved perceptions of transparency and clarity (component 2), and competency of HMRC staff (component 3) among PAYE customers.

Exposure to the opinions of others

Being exposed to positive information about HMRC from others, was associated with higher levels of overall trust and improved perceptions of institutional fairness and reputation (component 1) among all 3 customer groups. It was associated with improved perceptions of competency of HMRC staff (component 3) among PAYE and non-business SA customers, and with improved perceptions of how HMRC responds when things go wrong (component 4) among PAYE customers and small businesses.

The level and mode of interaction with HMRC

For non-business SA customers, the variety of channels used to contact HMRC was associated with overall trust. Using 3 or more channel types was associated with a higher overall trust rating and improved perceptions of institutional fairness and reputation (component 1), compared with using two or fewer channels. The types of channels covered included searching for information on gov.uk, using the Business Tax Account or Personal Tax Account, using another online HMRC service, using online commercial software, telephone, post, face-to-face contact, receiving an email from HMRC or receiving a text message from HMRC.

The use of the telephone to contact HMRC was significantly associated with higher levels of overall trust among small business customers that were not self-employed. It was also associated with improved perceptions of the competency of HMRC staff (component 3) among non-business SA customers and small businesses, and with lower levels of use of power (component 5) among non-business SA customers. The telephone was also associated with improved perceptions of institutional fairness and reputation (component 1) for small businesses that were not self-employed only.

There were some contact channels that were not significantly associated with overall trust directly but were significantly associated with the components of trust, including:

• use of an online commercial software which was associated with improved perceptions of transparency and clarity (component 2) among PAYE customers and small businesses
• receiving a text from HMRC which was associated with improved perceptions of transparency and clarity (component 2) among PAYE customers and improved perceptions of the competency of HMRC staff (component 3) and lower levels of use of power (component 5) among non-business SA customers
• searching for information on gov.uk which was associated with improved perceptions of transparency and clarity (component 2) among small businesses who were not self-employed

Making contact with HMRC via the customer’s preferred contact channel was associated with lower levels of use of power (component 5) among non-business SA customers and small businesses.

Perceived opportunities and risks of non-compliance

Among small businesses who were not self-employed, higher agreement that there is more opportunity for tax non-compliance among businesses like their own was associated with higher levels of overall trust. Higher agreement that there is more opportunity for tax non-compliance was also associated with improved perceptions of the competency of HMRC staff (component 3) among all small businesses.

Having the right understanding of HMRC’s functions

Other influencers that were not directly associated with overall trust included wrongly believing that HMRC is responsible for setting tax rates, which was associated with improved perceptions of institutional fairness and reputation (component 1) among all 3 customer groups (although among small businesses this was only significant among the self-employed), and improved perceptions of transparency and clarity (component 2) among small businesses.

5.3 Influencers associated with lower levels of overall trust and worsened perceptions of the components of trust

Attitudes around social norms on compliance

Higher levels of agreement that HMRC goes after the wrong people was associated with lower levels of overall trust for all 3 customer groups^{[footnote 6]}. It was also associated with worsened perceptions of institutional fairness and reputation (component 1), worsened perceptions of how HMRC responds when things go wrong (component 4) and an increase in the view that HMRC uses power and lack of reciprocal trust (component 5) among all 3 customer groups.

Customer experience

HMRC making errors and dealing with them poorly was associated with lower levels of overall trust for PAYE and non-business SA customers when compared with HMRC not making any errors. The negative association was smaller but still significant when HMRC made errors but dealt with them well (compared with HMRC not making any errors).

For all 3 customer groups, HMRC making errors and dealing with them poorly (compared with HMRC not making any errors) was also associated with worsened perceptions of how HMRC responds when things go wrong (component 4). However, this association was only significant among non-business SA customers when HMRC made errors but dealt with them well.

HMRC making errors and dealing with them poorly (compared with HMRC not making any errors) was also associated with worsened perceptions of institutional fairness and reputation (component 1), and perceptions of transparency and clarity (component 2) among PAYE customers. These associations remained significant, although smaller, when HMRC made errors but dealt with them well.

Exposure to the opinions of others

Being influenced by the media when forming attitudes towards HMRC was associated with lower levels of overall trust among PAYE customers (compared with not being influenced by the media). This was also associated with worsened perceptions of institutional fairness and reputation (component 1) among PAYE customers and small businesses, and use of power and a lack of reciprocal trust (component 5) among small businesses.

Being influenced by friends and family when forming attitudes towards HMRC was associated with worsened perceptions of transparency and clarity (component 2) among PAYE customers, and an increased perception of use of power (component 5) among small businesses. For the small business group, professional advisors influencing attitudes towards HMRC was also associated with worsened perceptions of how HMRC responds when things go wrong (component 4).

Perceived opportunities and risks of non-compliance

Perceiving a greater risk associated with non-compliance, such as a greater likelihood of being caught or receiving more severe sanctions, was associated with lower levels of overall trust among PAYE customers. This was mainly driven by the association of perceived risks of non-compliance with higher perceived use of power (component 5), which was significant for all 3 groups (except small businesses not self-employed).

The level and mode of interaction with HMRC

The level and type of interaction with HMRC was also important. Having more frequent contact with HMRC was associated with lower levels of overall trust and worsened perceptions of institutional fairness and reputation (component 1) among non-business SA customers. Using 3 or more different channel types to contact HMRC was associated with worsened perceptions of how HMRC responds when things go wrong (component 4) among small business customers who were not self-employed.

The only channel used to contact HMRC that was significantly associated with lower levels of overall trust was searching for information online using gov.uk among non-business SA customers. Searching for information online using gov.uk was also associated with worsened perceptions of transparency and clarity (component 2) among small businesses who were self-employed.

Use of the Business Tax Account or Personal Tax Account was associated with worsened perceptions of institutional fairness and reputation (component 1) among PAYE customers. It was also associated with worsened perceptions of how HMRC responds when things go wrong (component 4) among non-business SA customers, and among small businesses that were not self-employed. The reasons for this association are unclear and could be explored with further research. Receiving an email was associated with worsened perceptions of competency of HMRC staff (component 3) among PAYE customers. Receiving a text was associated with worsened perceptions of how HMRC responds when things go wrong (component 4) among small businesses that were not self-employed.

Having the right understanding of HMRC’s functions

Influencers that were not directly associated with overall trust included wrongly believing that HMRC is responsible for setting tax rates, which was associated with increased perceptions of the use of power (component 5) among all non-business SA customers and small businesses.

6. Demographic influencers of trust

Some demographic characteristics were found to influence overall trust and the 5 components of trust, even after controlling for the effects of other variables. This section highlights these demographic influencers through exploration of the characteristics that appeared significant in descriptive statistics, and whether they were still significant in the regression models of trust.

6.1 PAYE customers

Working status

PAYE customers in paid employment were more likely to disagree that HMRC was a trustworthy organisation compared with those not in paid employment. However, when all influencers were controlled for using the regression modelling, being in paid employment was not associated with different levels of overall trust. The analysis by components revealed it was associated with worsened perceptions of transparency and clarity (component 2). This means that the link between employment and overall trust can be explained by the influencers considered in this research, although it also seems to directly impact customers’ perceptions of transparency and clarity.

Being placed on furlough during COVID-19

There was no difference in ratings of HMRC’s trustworthiness between PAYE customers who had been placed on furlough during COVID-19 and those who had not. However, being placed on furlough was associated with increased levels of overall trust when all other influencers were controlled for in the modelling, implying some link between being placed in the furlough scheme and higher levels of trust.

Age

Older PAYE customers (aged 65 or older) were more likely to agree that HMRC was a trustworthy organisation compared with those aged under 65. However, when all influencers were controlled for using the regression modelling, an overall increase in age by one-year increments was independently associated with higher overall trust among those aged below 65 years old. This association was particularly strong for those aged 16 to 24 years old.

Receipt of Child Benefit

PAYE customers who had received Child Benefit were more likely to disagree that HMRC was a trustworthy organisation compared with those who had not received Child Benefit. Further analysis revealed such association can be mostly explained by the set of influencers considered in the previous section, although receiving Child Benefit was associated with worsened perceptions of how HMRC responds when things go wrong (component 4) even after controlling for all influencers.

Receipt of Marriage Allowance

PAYE customers who had received Marriage Allowance were more likely to disagree that HMRC was a trustworthy organisation compared with those who had not received Marriage Allowance. However, when all influencers were controlled for using the regression modelling, receiving Marriage Allowance was not associated with different levels of overall trust. The analysis by components revealed it was associated with higher perceptions of transparency and clarity (component 2).

6.2 Non-business SA customers

Qualifications

There was no variation in how much non-business SA customers agreed or disagreed that HMRC was a trustworthy organisation according to qualification level. However, when all influencers were controlled for, the models found that having a higher qualification was associated with higher levels of overall trust.

Age

There was no variation by age when non-business SA customers were asked how much they agreed or disagreed that HMRC was a trustworthy organisation. However, when all influencers were controlled for using the regression modelling, an increase in age by one-year increments was found to be associated with lower levels of overall trust and worsened perceptions of institutional fairness and reputation (component 1).

6.3 Small business customers

Number of employees

There was no variation when exploring descriptive statistics by numbers of employees among small business customers and how much they agreed or disagreed that HMRC was a trustworthy organisation. However, the regression modelling showed that businesses that had a higher number of employees had higher levels of overall trust in HMRC after all influencers were controlled for. This was mainly driven by higher levels of perceived competency of HMRC staff (component 3).

Self-employed status

There was no variation in levels of overall trust in HMRC between small business customers that were self-employed and ran the business alone compared with those who were not self-employed. However, the regression modelling showed that when all influencers were controlled for, running the business alone as a self-employed individual was associated with higher levels of overall trust. This association seems to be driven by improved perceptions of institutional fairness and reputation (component 1, positively associated with being self-employed, even for those running the business with a partner), although the analysis also revealed that self-employed small businesses perceived lower levels of transparency and clarity, compared to those not self-employed, after controlling for all influencers.

Qualifications

Small business customers who had qualifications below university degree level were more likely to agree that HMRC was a trustworthy organisation, compared with small business customers who had a university degree. The association with overall trust faded away after controlling for all the influencers. However, higher education qualifications were still significantly associated with lower perceptions of institutional fairness and reputation (component 1), as well as lower perceived use of power (component 5).

Gender

Male small business customers were more likely to disagree that HMRC was a trustworthy organisation compared with those who were female. However, when all influencers were controlled for using the regression modelling, gender was not significantly associated with overall trust or any of the components of trust.

7. Applying the model of trust to real life scenarios

This final section in this report summarises the key findings from this research, by simulating different scenarios using the model of trust derived in this research, to measure the effect it might have on overall trust. The simulated scenarios in this section are based on the regression models that are discussed in section 5. These scenarios assume a causal relationship between the influencer variables and overall trust. However, this is an assumption that cannot be tested with the data from this survey. This analysis is intended to illustrate the magnitude of the regression coefficients more clearly, by displaying them in terms of the number of customers that may be affected by specific improvements or changes in different influencer variables.

For all scenarios described in this section, the ‘point estimate’ shown in the table is equivalent to the percentage of customers moving +1 point in the overall trust 5-point Likert scale (for example, moving from strongly disagree to disagree, or from neither agree nor disagree to agree). The lower and upper bounds are the 95% confidence intervals. Results which cross ‘0’ (where the direction of the effect is uncertain) are indicated by an *.

7.1 Improving customer experiences

This research has highlighted the importance of a positive customer experience in building trust. Having a positive experience when dealing with HMRC is associated with higher levels of overall trust for all 3 customer groups, even if tax affairs are difficult to deal with. If tax affairs are easy to deal with and the experience with HMRC was positive, this is associated with improved perceptions of:

• institutional fairness and reputation (component 1)
• transparency and clarity (component 2)
• competency of HMRC staff (component 3)

There is scope to improve in this area. Around half (51%) of PAYE customers rated their experience of dealing with HMRC 4 or 5 out of 5 (good), although ratings were slightly higher for non-business SA customers (60%) and small businesses (69%).

Table 7.1 and Table 7.2 shows how trust might be impacted if there was a reduction in ‘unsatisfied’ customers (those that rated their overall experience 1 or 2 out of 5) by 50% and 25% respectively.

For example, if the proportion of unsatisfied PAYE customers reduced by half, 10.1% of customers would move +1 point in the 5-point Likert scale used to measure trust in HMRC. There is a similar finding for non-business SA customers (10.2%), but such a reduction would have a lesser impact on levels of trust among small businesses (6.9%). If the proportion of unsatisfied customers reduced by 25%, the proportions of customers that would move +1 point in the Likert scale for overall trust would also halve.

Table 7.1 Scenario 1: reduce unsatisfied customers by 50%

	PAYE	Non-business SA	Small business
Point estimate	10.1%	10.2%	6.9%
Lower bound	6.3%	4.8%	4.4%
Upper bound	13.9%	15.5%	9.4%

Table 7.2 Scenario 2: reduce unsatisfied customers by 25%

	PAYE	Non-business SA	Small business
Point estimate	5.1%	5.1%	3.4%
Lower bound	3.2%	2.4%	2.2%
Upper bound	7.0%	7.8%	4.7%

7.2 Improving the way HMRC deals with errors

This research has found that errors that are dealt with badly by HMRC are associated with lower levels of overall trust for the PAYE and non-business SA customer groups (when compared with making no errors). The negative relationship is evident for:

• overall trust
• perceptions of institutional fairness and reputation (component 1)
• perceptions of how HMRC responds when things go wrong (component 4), for all 3 customer groups
• perceptions of transparency and clarity (component 2), for PAYE customers only

The worsened associated perceptions were smaller for those who experienced errors that were perceived to have been dealt with well. For small businesses, dealing with errors well was associated with improved perceptions of the competency of HMRC staff (component 3).

Table 7.3 shows how trust might be impacted if errors were reduced by 50%, evenly split between whether they were dealt with well and poorly, and Table 7.4 shows the impact on trust if HMRC dealt with all errors well. The findings are similar for both scenarios. The impact might be higher among non-business SA customers (around 7% would move +1 point in the Likert scale for overall trust), than PAYE customers (circa 4%) and small businesses (1% to 2%).

Table 7.3 Scenario 1: reduce overall errors by 50% (evenly split among dealt with well and poorly)

	PAYE	Non-business SA	Small business
Point estimate	4.4%	6.9%	1.8%
Lower bound	2.6%	3.8%	0.0%*
Upper bound	6.1%	10.0%	3.5%

Table 7.4 Scenario 2: dealing with all errors well

	PAYE	Non-business SA	Small business
Point estimate	4.4%	6.7%	1.0%
Lower bound	2.4%	1.5%	-1.0%*
Upper bound	6.4%	12.0%	3.1%

7.3 Exposure to information about HMRC

Exposure to positive information about HMRC in the last 12 months from, for example, friends, family or the media, was associated with higher levels of overall trust and improved perceptions of institutional fairness and reputation (component 1) across all 3 groups.

For PAYE and non-business SA customers it was associated with improved perceptions of the competency of HMRC staff (component 3), whereas for small businesses it was linked with improved perceptions of how HMRC responds when things go wrong (component 4).

Attitudes towards HMRC that were influenced by the media were associated with lower levels of overall trust and worsened perceptions of institutional fairness and reputation (component 1) for PAYE customers.

Table 7.5 to Table 7.7 show how trust might be impacted under different scenarios when exposed to information about HMRC, and the overall tone of the information was rated more positively (on a 5-point scale where 1 is least positive and 5 is most positive).

If the tone of the information the customer was exposed to was rated +1 point higher, 10.0% of PAYE customers, 12.9% of non-business SA customers and 8.3% of small business customers would move +1 point in the 5-point Likert scale used to measure trust in HMRC.

The other two scenarios measured were:

• improve the scores of those who rated the overall tone of information about HMRC negatively (1 or 2) by +1 point on the 5-point Likert scale
• improve the average score of the overall tone of information about HMRC a customer was exposed to by half point

They both would have less of an impact than the first scenario by around a half.

Table 7.5 Scenario 1: increase in overall tone of information about HMRC customer was exposed to by +1 point in the 5-point Likert scale (where 5 is positive)

	PAYE	Non-business SA	Small business
Point estimate	10.0%	12.9%	8.3%
Lower bound	6.8%	6.5%	4.5%
Upper bound	13.3%	19.2%	12.0%

Table 7.6 Scenario 2: improve scores of those who rate overall tone of information about HMRC exposed to negatively (1 or 2) by +1 point in the 5-point Likert scale

	PAYE	Non-business SA	Small business
Point estimate	4.7%	5.3%	2.4%
Lower bound	3.1%	2.7%	1.3%
Upper bound	6.2%	8.0%	3.5%

Table 7.7 Scenario 3: improve the average score of the overall tone of information about HMRC customer was exposed to by half point

	PAYE	Non-business SA	Small business
Point estimate	5.2%	6.8%	4.6%
Lower bound	3.5%	3.4%	2.5%
Upper bound	6.9%	10.1%	6.7%

7.4 Perception that HMRC goes after the wrong people

Around half of PAYE customers (54%) agreed that HMRC goes after the ‘wrong’ people, which is slightly higher than the same proportions of Non-business SA customers (43%) and small businesses (41%).

The view that HMRC goes after the wrong people was associated with lower overall trust for all 3 groups, but the association is small. The perception that HMRC goes after the ‘wrong’ people was also associated with worsened perceptions of:

• institutional fairness and reputation (component 1)
• how HMRC responds when things go wrong (component 4)

It had a positive association with the component related to use of power and lack of reciprocal trust (component 5), which is a negatively framed perception.

Table 7.8 to Table 7.10 show how trust might be impacted under the scenarios related to ratings for ‘HMRC goes after the wrong people’ whereby:

• all customers move 1 point down in the 5-point Likert scale
• reduce by 1 point those that agreed (4 or 5 out of 5)
• shift those that agreed (4 or 5 out of 5) to the neutral score (3 out of 5)

In summary, the measured impact on trust is highest if all customers move 1 point down in the 5-point Likert scale, though shifting those that agreed (4 or 5 out of 5) to the neutral score (3 out of 5) was broadly similar. The impact might be higher among Non-business SA customers for all 3 scenarios compared with PAYE customer and small businesses.

Table 7.8 Scenario 1: all customers move 1 point down in the 5-point Likert scale

	PAYE	Non-business SA	Small business
Point estimate	7.6%	12.0%	6.6%
Upper bound	12.5%	18.3%	10.2%
Lower bound	2.7%	5.7%	3.0%

Table 7.9 Scenario 2: reduce by 1 point those that agreed (4 or 5 out of 5)

	PAYE	Non-business SA	Small business
Point estimate	4.4%	6.1%	3.3%
Upper bound	7.2%	9.3%	5.1%
Lower bound	1.5%	2.9%	1.5%

Table 7.10 Scenario 3: shift those that agreed (4 or 5 out of 5) to the neutral score (3 out of 5)

	PAYE	Non-business SA	Small business
Point estimate	6.6%	9.2%	5.1%
Upper bound	10.9%	14.0%	7.9%
Lower bound	2.3%	4.4%	2.3%

7.5 Improving customer confidence in dealing with tax

Fewer than 2 in 5 PAYE customers (38%) feel confident to deal with taxes (rate their confidence 4 or 5 out of 5), which is lower than for both non-business SA customers and small businesses (both 52%).

For PAYE customers in particular, confidence in dealing with taxes was associated with higher levels of overall trust, but also with improved perceptions with components of trust across all customer groups. These associations were with:

• overall trust (PAYE customers and small businesses)
• institutional fairness and reputation (component 1) among PAYE customers
• transparency and clarity (component 2) among all customer groups
• competency of HMRC staff (component 3) among PAYE customers

Among PAYE customers, confidence was also associated with perceptions that HMRC uses power and lack of reciprocal trust (component 5).

Among the PAYE group, there was an association between those with low confidence in dealing with taxes who completely relied on others to handle their tax affairs and higher levels of overall trust. There was no association with overall trust for those who were confident and used professional help for their taxes.

Table 7.11 to Table 7.13 show how trust might be impacted if customers’ ratings of confidence dealing with taxes improved, including:

• improve confidence score by +1 point in the 5-point Likert scale
• all customers rate their confidence at least 3 out of 5
• improve average confidence by half a point in the 5-point Likert scale

If confidence scores improved by +1 point in the 5-point Likert scale for PAYE customers, 14.1% would move +1 point in the 5-point Likert scale used to measure trust in HMRC. The measured impact on trust was lower for the other two scenarios, but in all 3 scenarios the impact would be higher for PAYE customers compared with non-business SA and small businesses.

Table 7.11 Scenario 1: improve confidence score by +1 point in the 5-point Likert scale (except those who are fully confident)

	PAYE	Non-business SA	Small business
Point estimate	14.1%	6.1%	6.9%
Lower bound	7.1%	-2.7%*	1.2%
Upper bound	21.1%	14.9%	12.7%

Table 7.12 Scenario 2: all customers rate their confidence at least 3 out of 5

	PAYE	Non-business SA	Small business
Point estimate	7.2%	2.7%	3.3%
Lower bound	3.6%	-1.2%*	0.6%
Upper bound	10.8%	6.5%	6.0%

Table 7.13 Scenario 3: improve average confidence by half a point in the 5-point Likert scale

	PAYE	Non-business SA	Small business
Point estimate	8.0%	3.9%	4.5%
Lower bound	4.0%	-1.7%*	0.8%
Upper bound	12.0%	9.6%	8.2%

---

1. HMRC (2020). Building a trusted, modern tax administration system. https://www.gov.uk/government/publications/tax-administration-strategy/building-a-trusted-modern-tax-administration-system ↩

2. OECD (2010). Understanding and Influencing Taxpayers’ Compliance Behaviour. Also, Muehlbacher S, Kirchler E, Schwarzenberger H (2011) Voluntary versus enforced tax compliance: Empirical evidence for the “slippery slope” framework October 2011 SpringerLink ↩

3. HMRC (2020). Building a trusted, modern tax administration system. https://www.gov.uk/government/publications/tax-administration-strategy/building-a-trusted-modern-tax-administration-system ↩

4. OECD (2010). Understanding and Influencing Taxpayers’ Compliance Behaviour. Also, Muehlbacher S, Kirchler E, Schwarzenberger H (2011) Voluntary versus enforced tax compliance: Empirical evidence for the “slippery slope” framework October 2011 SpringerLink ↩

5. Kantar Public on behalf of HMRC (2022). Individuals, Small Businesses and Agents Customer Survey 2021 ↩

6. In the survey the statement was ‘HMRC go after ordinary people instead of the real tax dodgers’. Throughout the report this is referred to as HMRC going after the wrong people. ↩