Budget 2024: what shapes voter opinion on tax, borrowing and government approval?

�October 22nd 2024

2

​

• IPPR (Institute of Public Policy Research) is an independent charity working towards a fairer, greener and more prosperous society.�
• Persuasion UK is a new research initiative set up to study what is shaping public opinion on the issues that define British politics.�
• IPPR and Persuasion are separate organisations but have a strategic partnership, meaning they work together on projects of shared interest.

​

Who we are

3

​

• What is the ‘record’ that most voters will most reward and punish at the next election? That is, what is the most salient criteria they’re judging the government on and what does this say about upside and downside risk of different approaches to governing and economic policy?�
• What are people’s baseline attitudes to an array of possible wealth tax and borrow-to-invest ideas?
• What happens to these attitudes when they are exposed to different pro and anti messages? �
• What’s the best overall story for selling these changes?�
• To what extent, if at all, do popular progressive policies ‘accumulate’ in negative ways to the detriment of party brand? That is, is there an independent effect of the number of policies on the perception of a pary’s moderation, or is all the effect (if any) on the individual policies? �

As much as possible, to pay particular attention to results among swing voters (Con to Lab switchers; switchers to Lab generally; potential switcher groups next time, etc)

Research objectives

4

A series of experiments over September and October 2024:�

• A baseline survey of 4,000 UK voters.
• A RCT message testing experiment with 6,000 voters.
• One split-sample experiment looking at ‘policy accumulation’, with 12,000 voters
• A final experiment testing different scenarios on government approval ratings, with 6,400 voters

​

All of these were conducted via YouGov, with samples weighted to be nationally representative. Costs were shared between Persuasion and IPPR. �

The exact methodology of each experiment is explained in the results section of this deck.�

​

Approach

5

​

​

• Understanding what voters will most reward and punish government for: we showed people three random issue outcomes/trade-offs and asked them to imagine it formed the government’s record in 2029 and how how much they’d approve of such a government (plus they see a standard attack on this record). In the analysis phase we isolate the effect of each outcome/trade off on government approval to ‘reveal’ what is most important to voters.�
• Understanding baseline attitudes and dominant frames to wealth tax ideas and borrow-to-invest: straight-forward support/oppose followed by some paired statement testing. �
• Stress testing narratives: Randomised Control Trial (RCT) methodology used to observe the effect of different narratives on voter’s underlying attitudes. Results highlight persuasion effects, so far as they exist. �
• ‘Too much of a good thing?’ experiment: We take ‘in-principle popular but progressively coded’ policies and randomise only the number attached to the government’s brand that respondents see, then get respondents to fill in a survey. We see if the number of policies attached to Govt brand makes a difference to perceptions of moderation or competence.

​

Overview of methodology used in each section

6

Summary of findings

• Finding #1: The issue target voters will punish and reward Labour government on the most is public services. Tangible progress on this will be rewarded even if it comes with unpopular trade-offs (higher debt, taxes etc), while failure will be punished even if it comes with popular upsides (lower debts, deficits). Being good stewards or ‘book keepers’ of the economy is not sufficient. Crucial switcher groups are especially motivated by the NHS.�
• Finding #2: Both support for wealth taxation and the government’s political and economic brand is resilient to attack, should it adopt these ideas. �
• Finding #3. There is some evidence that a positive message around rebuilding the public realm, especially the NHS, and restoring fairness in ‘who pays’ can bolster Labour’s brand with crucial switcher groups. But it is mostly the ends, not the means, that move voters.�
• Finding #4: It is not all positive from a progressive perspective. Support for some policy ideas (eg reforming tax on pension contribution, raising NICs on employees) is more divided and vulnerable to opposition attack messages.�
• Finding #5: There is no firm evidence that the number of ‘popular but left coded’ ideas Labour adopts has an independent effect on the government’s underlying brand, albeit there are caveats on this in relation to Con-to-Lab switchers and overall research design. �

​

All research has limitations (nothing can perfectly recreate the information environment voters form opinions in) and should be taken with a pinch of salt, but we hold these conclusions with a fairly high degree of certainty.

7

Finding #1: The issue target voters will punish and reward Labour government on the most is public services.

8

Methodology��

• We constructed nine different possible policy outcomes that might form the basis of the government’s ‘record’ at the end of its first term. For each outcome, we put an associated trade-off. �
• Survey respondents were asked their current opinion of the Labour government. �
• Respondents then saw three of the outcomes/trade-offs at random and were told to imagine these were the government’s achievements at the end of the first term. �
• To try to keep things realistic, they were then shown a relevant opposition attack message on this record (‘broken promises’, ‘spent all the money’ or ‘they’ve changed nothing’ depending on trade-off seen).�
• Respondents were then asked, in light of all this, if such a Govt record shown would be acceptable or not - and, crucially, how they would approve of the government in this scenario. �

In the analysis phase, we can see what impact each outcome/trade-off has on government approval ratings.�

​

9

Methodology: outcomes/trade-offs tested�

​

10

Methodology: example of what was seen by respondents�

​

11

12

13

14

15

Finding #2: There is decent baseline support for both borrow-to-invest and some wealth taxation ideas, although support is more divided on policies that feel more ‘proximate’ to people.

​

Most tax rises are not intuitively seen as breaching Labour’s manifesto commitment.

16

Wealth policies tested�

​

• Raise capital gains tax (CGT) so it is the same level as income tax, meaning the tax paid on income from wealth (e.g shares, property) is the same as from work
• Cap the level of tax relief on pension contributions at 30%, limiting tax breaks on savings received by higher earners
• Introduce national insurance contribution (NIC) tax on investment income - such as dividends and rental income - in the same way as it is on regular income.
• Ending the exemption some shares get from inheritance tax (shares listed on the ‘Alternative Investment Market’ are currently exempt from inheritance tax)
• Capping the exemption on inheritance tax you get when transferring business and agricultural property within families (capping it at £500,000)
• Increasing the rate of stamp duty on ‘share buybacks’ - the tax large companies pay when they buy back their own shares from their investors
• Ending exemptions on stamp duty that some stock traders have when they buy UK government securities
• A new levy on the profits of tobacco companies
• End the freeze on alcohol duty, increasing tax on alcohol by 6.5%
• A new levy on the profits of gambling companies
• End the freeze on fuel duty, increasing tax on petrol
• Increasing national insurance tax by 2%, reversing the previous government’s cut to this tax (which people pay on their income)

17

Borrow-to-invest policies tested�

​

• Changing fiscal rules to allow government to borrow more to spend on:�
• General government expenditure
• Infrastructure
• Public services

18

19

20

21

Frame testing - methodology�

​

• Respondents saw two arguments, one pro and one anti, for each major thematic argument/narrative divide:

​

• On tax raising generally; wealth taxation specifically; pension tax reform; sin taxes; borrow to invest. �
• These arguments were randomly drawn from a long list of pro and anti. �
• In the analysis we see which argument most consistently won out; which pro argument - if any - consistently beat the anti, and vice versa.

22

23

24

25

26

27

• There is decent support for most ideas around wealth taxation - and on borrow to invest. �
• The obvious tension on taxation is voters are more wary of anything that they feel will hit them or people like them. �
• On borrow to invest, it needs to be linked explicitly to infrastructure or public services - clearly demarcated from day to day spending.�
• Arguments on the need to raise taxes generally, as well as sin taxes, can be won - but it’s a close run thing.. �
• On the plus side, wealth taxation and borrow to invest are far more comfortable territory - fairness and long-term investment.�
• Rebuilding public services and a general sense of fairness/egalitarianism pull people towards progressive positions on these matters. Anytime there is a sense ordinary people will be hit, it tends to push more people away.

Recapping findings from this section�

​

28

Finding #2: Both support for specific ideas on tax and borrowing - and the Labour government’s economic brand - is fairly resilient when stress tested against likely opposition attacks.�

​

Finding #3. There is some evidence that arguments for these changes tend to be more persuasive - and boost the government’s brand more - when framed around rebuilding the public realm, and/or fairness in asking more from those at the top when others are squeezed.�

​

Finding #4: However, some ideas (on pension tax relief, raising NICs) are more consistently vulnerable to opposition attack messages.

29

• We wanted to find out which messages best sell a combined package of wealth tax rises and borrow-to-invest. This needs to survive contact with aggressive attack messages and protect the underlying economic brand of those proposing it (in this case, the government). �
• We took a large group of voters and split them up into six sub-groups�
• Four groups saw a possible government message justifying wealth tax rises and borrow-to-invest changes. They also saw an attack message, either political in nature or business focused.�
• One group just saw an attack message, with no government message. �
• A control group saw nothing.

​

• All groups took the same survey at the end of the experiment, answering questions on their voting intention, perception of party’s economic competence and views on select tax and borrowing policies.�
• In the analysis phase we can see how the ‘outcome attitudes’ of different message groups differs to control. Any statistically significant difference can be attributed to the message they have seen. �
• This has the benefit of isolating persuasion effects in practice, rather than just asking people if they agree/disagree with a message. �
• It is not perfect (no controlled experiment can perfectly replicate the information environment voters operate in) but it is by far the best way of testing the effect of messages on underlying attitudes.

Methodology�

​

30

Methodology: summary of messages tested�

​

• Attack message. Respondents in every treatment group saw one of two possible attack messages. In one treatment, they only saw one of these messages but in the others they saw this plus a possible government message.�
• The first was a traditional political attack message explaining how these changed represented ‘same old Labour’ - whacking up taxes and borrowing at the first opportunity when they said they wouldn’t. It’s ordinary people who will pay the price, it argued. �
• The alternative attack message stimulated an opposing message from business groups. This argued that these changes represent a danger to the economy and will stifle growth and jobs at a time when the country needs it most. �
• Possible government message #1: cleaning up the last government’s mess. This argued that the last government left a £22bn ‘black hole’ in the finances and these measures were the only way to address this. �
• Possible government message #2: let’s rebuild our NHS. This message outlined the problems faced by the NHS as a result of under-investment and what this means for patients. It made the case for tax and borrowing changes as a way of raising money to fix this problem.

31

Methodology: summary of messages tested�

​

• Possible government message #3: let’s rebuild Britain. This was a broader message about the crumbling state of the public realm in the UK - touching on things like roads, rivers, public services, and so on. It made the case for tax and borrowing changes as a way of raising money to address this problem. �
• Possible government message #4: fairness/inequality. This message argued for the essential fairness of government asking more from those at the top, especially at a time when ordinary workers are squeezed. It pointed out the unjust nature of the current tax system, for instance arguing that Rishi Sunak pays less tax on his income than a nurse, etc.

32

Methodology: margins of error / minimum statistically significant effect �

​

Any difference in a message group that is smaller than this is just noise and not significant:�

​

• For all voters: 2.5% +/-
• For ‘Switchers’ - 10% +/-
• For ‘Defectors’ - 10% +/-
• For all non-Labour voters - 2.6% +/-

�

In the final results tables, any statistically significant effect in a progressive direction is in green; any significant effect in a non-progressive direction is in red. Anything statistically insignificant is in grey.

33

RCT results for all voters

34

35

36

Margin of error

37

38

Margin of error

39

40

Summary of RCT results for Labour switchers (voters who did not vote Labour in 2019, but switched to vote for them in 2024)

41

42

Summary of RCT results for Labour defectors (voters who voted Labour in 2019, but then did not in 2024)

43

44

Summary of this section

​

• The effect sizes across all groups are fairly small, at least for voters at large. The most striking thing is that not much changes, which suggests the ‘means’ of achieving different outcomes are not that vote moving. Even the attack messages on their own don’t move much. �
• However, there are some messages which perform better than others - especially among ‘retain and gain’ target voters.�
• One or some combination of these messages would work fine in protecting the government’s brand in a hostile environment:

​

• Rebuild our NHS
• Rebuild Britain (general)
• Fairness

​

45

Finding #5: There is no really clear evidence that ‘popular but left coded’ policies ‘accumulate’ against a party’s brand in negative ways. There is a some evidence for negative perception shift among Con-to-Lab switchers - albeit without damage to the Labour brand.

46

Methodology

​

​

• The approach here was to test the hypothesis that voters may like policies individually, but that a government adopting too many of them has negative effects on its underlying brand (especially it’s reputation for moderation).�
• To test this, it requires isolating the effect of the number of policies voters are exposed to. �
• To do this, we told a group of voters to imagine the Labour government had adopted a number of progressive wealth tax ideas - plus changes to fiscal rules to allow more borrowing to invest. We knew each of these would be individually popular. �
• However, some voters only 1 of these possible Labour policies, some only saw 3 and some saw 5. A control group saw none. �
• All voters then saw a standard opposition attack message.�
• All voters, no matter how many policies they saw, took the same outcome survey. �
• In the analysis phase we can see how the attitudes of those who saw more/less policies differ, if at all.

47

Policies tested

​

​

​

• Increases taxes on the wealthiest 1% of the population and invest the money into the NHS.�
• To redraw council tax bands so that those in the most expensive houses pay more council tax than today.�
• Raise capital gains tax (CGT) so it is the same level as income tax, meaning the tax paid on income from wealth (e.g shares, rental income) is the same as from work.�
• Allow the government to borrow more to invest in public services and infrastructure.�
• Increase the wages of front-line public sector workers like doctors, nurses and teachers by at least 1% above inflation each year from now until 2029.

48

Methodology: example of what was seen by respondents

​

49

Results

50

Results

51

Possible explanations for these results

​

​

​

Why is there not much impact between number of progressive policies and outcome attitudes?

​

There are a few possible reasons:

​

​

• It could be a problem with the research design (eg were respondents paying attention properly? Experiment also can’t replicate the effect of multiple news cycles etc)�
• People who respond negatively to Lab policy were already inclined to hold these views - i.e the number of policies is not making more people hold a negative view, just reinforcing the views of those who already held such opinions. �
• There just is no real relationship here. It’s the individual policy or policies that matter not the amount of them.�

52

Conclusions

53

Conclusions

​

​

​

• All research has limitations and this is no different. It’s hard to perfectly ‘simulate’ or re-create information environments that voters receive messages within, at least in a controlled environment. Research like this - which isolates the effect of messages/policies in practice and doesn’t just ask people what they think - is still better than relying solely on things like focus groups, standard surveys etc. But it all still needs a pinch of salt. �
• Nevertheless, we remain confident in our conclusions…�

54

Conclusions

• Finding #1: The issue target voters will punish and reward Labour government on the most is public services. Tangible progress here is rewarded even if it comes with unpopular trade-offs (higher debt, deficits etc), while failure is punished even if it comes with popular upsides (lower debts, deficits). Being good stewards or ‘book keepers’ on the economy is necessary hygiene but not sufficient. Crucial switcher groups are especially motivated by the NHS - they will forgive Labour a lot if they see meaningful improvement here.

• Finding #2: In stress testing, support for specific ideas on wealth taxation (eg raising CGT, capping IHT relief) - plus fiscal rule tweaks fiscal on borrow-to-invest - is resilient. Most importantly, the government’s economic brand does not appear to be negatively affected by political or business attacks on these ideas.

• Finding #3. There is some evidence - albeit limited - that a positive message around rebuilding the public realm, especially the NHS, and restoring fairness in ‘who pays’ can bolster Labour’s brand with crucial switcher groups. But it is mostly the ends, not the means, that move voters.�

• Finding #4: However, it is not all positive from a progressive perspective. Top line support for some progressive policy ideas (eg reforming tax on pension contribution, raising NICs on employees) is more divided and vulnerable to opposition attack messages, even in the face of strong Labour messaging. Likewise, opinion on ‘sin taxes’ divided - and fuel duty remains very divisive.

��

• Finding #5: The number of ‘popular but left coded’ ideas Labour adopts does not seem to have significant bearing on its underlying brand, albeit there are limits to how we tested this. Con-to-Lab switchers do seem to perceive Labour as slightly more left-wing the more of these ideas it adopts. But, at least in this experiment, there is no evidence that the number of policies alone drives underlying damage to the Labour brand or the government’s reputation for competence.

55

Example script, bringing in elements of all themes tested in this experiment:��