using social norms and pre populated forms to improve tax collection in manises

Enrique Belenguer Saborit © www.mints.es of 36

Using social norms and pre-populated forms to improve tax collection in Manises Enrique Belenguer Saborit https://es.linkedin.com/in/quiquebelenguer November 2015 Dissertation - LSE Exec MSc on Behavioural Sciences


INDEX

Abstract ............................................................................................................................. 3

I. INTRODUCTION: Background and Objective ...................................................... 3

II. LITERATURE: Theoretical Framework ................................................................ 5

II.a Social Norms ....................................................................................................... 7

II.b Elements that facilitate compliance .................................................................. 10

II.c National Context and Tax Morale ..................................................................... 13

III. THE EXPERIMENT ............................................................................................. 16

III.a Design and Methodology ................................................................................. 16

III.b Findings ........................................................................................................... 21

III.c Discussion ........................................................................................................ 27

IV. CONCLUSION AND LEARNING ...................................................................... 30

REFERENCES ............................................................................................................. 32


Abstract Tax compliance a major issue in Southern European countries at a time when budgetary deficit

is a key concern for many international organizations. Behavioural sciences can boost tax

compliance of those who are able to pay their taxes without compromising the financial

sufficiency of those who cannot. I decided to run an experiment using social norms and a pre-

populated form for enrolment in an instalment payment plan. My goal was to understand

whether social norms influenced behaviour in the direction I wanted and if facilitating a solution

to address a liquidity problem (i.e. by paying in instalments) improved tax compliance. All the

treatment groups had a positive and significant effect but, contrary to my expectations, social

norms did not have a significant effect – the effect was significant only in women. However, the

effect of receiving the letter was significant. The pre-populated form worked but when

combined with the social norm the effect was negative. Ex-post I have realized that this

combination could have sent a mixed message: pay now (social norm) vs pay later (instalment

plan). Furthermore, the ineffectiveness of social norms could be due to lower tax morale in

Mediterranean countries. Low intrinsic motivation to pay taxes may be tackled more effectively

with external regulations (punishments and coercive actions) rather than with internal

regulations (group identification).

I. INTRODUCTION: Background and objectives

In Southern European economies, governments are struggling to raise enough

resources to fund their activities and pay their debts. Tax compliance is therefore a key

goal. After all, in recent years, Southern European countries have been repeatedly cited

for their lax tax collection systems. Whilst much of the debate has been focused on

austerity, several critics addressed the low amount of taxes paid in these countries1. It

would be therefore reasonable to reinforce tax collection and fraud detection to reduce

outstanding payments that place a burden on public finances.

However reality is somewhat more complicated. Some individuals who fail to fulfil

their tax duties do so because they lack liquidity. Thus, it may be counterproductive to

force people to pay their taxes when they really need that money to make ends meet.

Forcing people in need to pay taxes would not only compromise the state’s role as a

safeguard of citizens’ well-being, but would also increase discomfort amongst people.

1http://www.economist.com/blogs/freeexchange/2012/09/tax-evasion-greece


Blanca Jimenez is the treasurer of the local council Manises (a city of 30,000

inhabitants in the metropolitan area of Valencia, Spain) and is therefore in charge of tax

collection. When we spoke about tackling tax compliance, she expressed some concerns

regarding the use of bank account seizing. Could it not be pointless to seize bank

accounts of individuals with financial issues?

I was attracted to the libertarian paternalist approach (Thaler & Sunstein, 2008)

which aims to influence behaviour without restricting choice. Several experiments have

used nudges to improve tax collection and I thought I could replicate some of them. I

was particularly inspired by the work of the Behavioural Insights Team (BIT) on tax

compliance (UK Cabinet Office - Behavioural Insights Team, 2012), and I decided to

run an intervention on outstanding tax payments using two mechanisms. The first was

social norms, and the second was pre-populated forms to facilitate enrolment in an

instalment payment plan.

I was excited with the opportunity. Not only might I share the case with other local

councils but the experiment might also trigger a wave of behavioural public policy in

Spain. Actually, in this sense my expectations were met during the first months of work,

as the experiment became the subject of several lectures and raised the interest of

councils across the country. Nevertheless, from my point of view, the research had a

further horizon. This study could add value to the growing literature on behavioural

sciences by analysing a) the effects of behavioural interventions in a Mediterranean

country and b) the effects of combining social norms with a pre-populated form. The

former element was the one that attracted most of my attention: Are social norms

effective across cultures?


II. LITERATURE: Theoretical framework

According to standard economic theory humans have internally consistent

preferences and choose the most preferable option (Sen, 1977) since we are able to

calculate all the trade-offs between the different options. With this utility maximization

framework, citizens assess whether they pay their taxes or not – or delay the payment –

based on the expected utility of each option – expected utility model (Allingham &

Sandmo, 1972; Gahramanov, 2009). Thus, paying taxes is just like any other decision in

economics: by assessing the costs and benefits of each option we decide whether

evading taxes or delaying the payment is worthwhile.

Standard models (Hashimzade, Myles, & Tran-Nam, 2013) predict the level of

evasion considering the penalty and the probability of getting caught. The value of each

option is the expected utility of the outcome multiplied by its probability. It is a decision

under uncertainty (Allingham & Sandmo, 1972) because subjects are fined only if they

are audited and the incompliant behaviour is detected. Ultimately, not paying taxes is a

gamble in which the subject chooses the option with the highest expected utility based

on his or her own degree of risk aversion.

In the case of late taxpayers, the decision is slightly different. The amount to pay is

already set and there is no need to audit. The authorities will fine the subject and seize

his or her bank accounts if needed. Thus late tax payment is not a decision under

uncertainty – unless the strategy is to expect the authorities to give up on chasing the

subject and allow the payment to expire. The only “rational” reason for paying late

would be if the late fee applied were less than the value of retaining the money for the

period until the fine were issued and the seizure were applied.

Whilst this is a clean model with an elegant mathematical solution, it may differ

considerably from reality. If tax evasion or deferral is a reasoned action based on clear

preferences, then tax intervention should focus on either informing about the

consequences and probabilities of being caught when failing to comply with tax duties,

or increasing fines and audit ratios. Any subject who was ignorant would thus know the

consequences and any sophisticated subject who lacked the incentive to pay taxes on

time would lose utility and change his or her behaviour.


In this Homo economicus’ world (in which the rational choice paradigm applies),

culture and ethics are unimportant. However, in our human world, despite having

incentives to cheat the tax system, people do not do it – at least not to the extent that

would be “rationally” optimal (Frey & Torgler, 2007; Hashimzade et al., 2013). In other

words, individuals are not looking to maximise their utility, at least not just in monetary

terms. Personal values also affect people’s relationship with taxes: tax morale. In fact

authors like Frey and Torgler (2007) have reported a high correlation between tax

morale and perceived tax evasion, which implies a reciprocal relationship: tax morale

increases when we perceive that people around us comply, and our perception of

compliance is linked to our own tax morale.

Thus, I can refute two key assumptions of the standard models that hinder our

understanding of how humans make tax-related decisions:

• Risk is approached from the expected utility theory (Allingham & Sandmo,

1972; Hashimzade et al., 2013) ignoring, for example, emotions

(Loewenstein, Weber, Hsee, & Welch, 2001) related to fear or our desire to

be liked.

• The issue is isolated: presented as an individual choice in which other

peoples’ behaviours do not affect the individual’s choice (Allingham &

Sandmo, 1972; Frey & Torgler, 2007; Hashimzade et al., 2013) thereby

neglecting the literature on descriptive and injunctive social norms and their

effect on human behaviour (Cialdini, Reno, & Kallgren, 1990; DellaVigna,

2009)

Furthermore the traditional approach ignores self-control problems derived from

non-standard preferences (DellaVigna, 2009). Some tax debtors do not intentionally

engage in incompliant behaviours. Lack of self-control and procrastination (Michael

Hallsworth, List, Metcalfe, & Vlaev, 2014) might be the cause. This is an important

issue, since procrastination itself could lead to future evasions, which maybe interpreted

as self-herding (Ariely, 2008) – we repeat our own past behaviours legitimizing post

facto what we did – or a negative promoting spill over effect (Paul Dolan & Galizzi,

2015) – once she has done it, the subject may think ‘what the heck’ and indulge in

further unethical behaviours.


Unsurprisingly a growing body of literature addresses the aforementioned issues

(Michael Hallsworth et al., 2014; Hashimzade et al., 2013) and seeks solutions to the

discrepancies between reality and economic models. In this study I focused on three

elements that appear in notable articles (Frey & Torgler, 2007; UK Cabinet Office -

Behavioural Insights Team, 2012) and address the concerns of the Manises tax officials:

a) social norms, b) elements that facilitate compliance and c) tax morale.

II.a Social norms

The use of social norms is one of the most popular and acclaimed ideas in

behavioural sciences. Essentially, we persuade people to behave in a certain way by

telling them what others do or what others think should be done. First, it is important to

clarify these two meanings of the social norm, differentiating the real behaviour from

the socially expected behaviour, as Cialdini et al (1990) suggest.

On the one hand, injunctive social norms – i.e. the norms that state what people

ought to do – are like a collective conscience that, to some extent, determines what is

acceptable and what is not – we are influenced by what we believe others think we

should do. For example, consider the statement, ‘The neighbours of this city disapprove

the use of recreational drugs’. The statement implies that the majority of citizens

disapprove of the use of recreational drugs, but not necessarily that they do not use them

in such way. Whilst the latter is implicit, the emphasis is on the moral mandate

determined by the neighbours. If you decide to use drugs in a recreational way, you are

against the collective idea of what is a right.

On the other hand, descriptive social norms state what people actually do. They

describe what is happening and not whether people think a certain behaviour is right or

wrong. For example, ‘65% of the neighbours in this city exercise for more than for

hours per week’. The statement describes what is happening. It is a fact. However,

whilst this statement describes reality, it implicitly reinforces the idea that people should

exercise for more than four hours per week. Consider the statement ‘35% of neighbours

in this city exercise less than ten hours per week’. Implicitly the statement suggests that

the neighbours should exercise more, but the reality is that they do not. The target

public would be aware that exercising for less than four hours per week is common,


albeit collectively undesirable. Such an effect was observed in the Arizona’s Petrified

Forest (Cialdini, 2003). To raise visitors’ awareness of the importance of not taking

with them the precious pieces of wood, a sign was installed explaining that large

amounts of petrified wood are stolen every year. Cialdini and his colleagues confirmed

that this was not a good idea since letting people know that the behaviour, whilst

undesired, is common significantly increases thievery.

These norms influence people’s behaviours mainly for two reasons. Firstly, such

statements make people aware of something they did not know. People may think a

certain behaviour is common when in it reality is not. Second, even if people are aware,

we are reminding them about this social norm in the decision context. We thus nudge

(Thaler & Sunstein, 2008) the decision maker to behave in the desired way because

people feel more comfortable as part of the majority (Martin, Goldstein, & Cialdini,

2014) and because people understand that if a given behaviour is common, it has to be

right.

According to the behavioural sciences literature, humans are affected by context cues

that help them make decisions, especially when facing some degree of uncertainty. If

we are unsure about the best choice, it is reasonable to choose whatever others are

doing. Common behaviour has the proof of the crowd – the approval of the majority –

which has always been part of popular wisdom: “When in Rome, do as the Romans do”.

A couple of factors should be considered when using of social norms. First, in order

to motivate behaviour, norms have to be activated – i.e. norms are salient and the

individuals are aware of them (Cialdini et al., 1990). Second, psychological distance

mediates the strength of the effect of social norms. According to construal level theory

(Trope & Liberman, 2010) psychologically distant objects – socially, physically or

temporally – are perceived as more abstract and less related to actions. This means that

social norms that portray people who are similar to the target audience are most

effective. The well-known experiment by Goldstein, Cialdini, & Griskevicius (2008),

run in a hotel to encourage environmentally friendly behaviours, endorsed this idea. A

card was left in the hotel rooms stating that the majority of guests had reused their

towels. This descriptive norm – a norm that describes what others do – had a positive

and significant effect. But when the social norms portrayed people similar to the target


audience – guests that had stayed in the same room as the subject – the effect was even

larger.

The use of social norms, particularly the use of descriptive social norms, has been

well-documented and validated in several fields such as energy consumption (Allcott,

2011), charitable donations (Frey & Meier, 2004) and hand washing (Lapinski,

Maloney, Braz, & Shulman, 2013). Due to its general acceptance and efficacy, social

norms are now key elements of some of the main behavioural science frameworks such

as MINDSPACE (P. Dolan et al., 2012) and EAST (Service et al., 2014)

The body of literature on the use of social norms for tax purposes is growing and

several of these studies shaped my experimental approach. A particular intervention by

the Behavioural Insights team, which sought to reduce late taxpayers debts using social

norms (UK Cabinet Office - Behavioural Insights Team, 2012), that has inspired my

experiment. In this intervention subjects in the treatment group received a letter

reminding them of their tax debts. The letter included the following statement: ‘Nine

out of ten people pay their taxes on time’. Psychological distance was also tested and

the narrower the geographical scope was – nation, postal code and town – the higher

efficacy. The experiment was a success. The subjects in the treatment group with social

norms and the narrowest geographical scope were 15% more likely to respond after

three months. Because of the original experiment’s success, I decided that the social

norm branch of my experiment would be a replication of this.

The BIT’s experiment was not the first to test social norms on tax collection. We can

find a similar experiment run in Minnesota in 1995 (Blumenthal, Christian, Slemrod, &

Smith, 2001; Coleman, 2007). The purpose was to remove the false belief that cheating

on taxes was common behaviour. It was a natural field experiment, in which researchers

sent letters to neighbours explaining that widespread tax evasion was a myth and that

over 90% of the taxpayers voluntarily paid their taxes. The experiment and the

subsequent replications showed that the inclusion of social norms in the letters exerted a

small but significant effect (Coleman, 2007). The authors used a Wilcoxon test to

measure the increase in the federal tax income and a nonparametric median test to

capture the differences in taxes paid in 1994 and 1995. Nevertheless, Blumenthal et al's

(2001) conclusions rejected the significance of the social norm’s effects. Running a


difference-in-differences analysis on the effect of social norms in the letters for this

same experiment, the authors rejected the idea that norms can affect ‘aggregate tax

compliance behaviour’. They found no evidence that overall taxes from one year to

another had increased because of any letter sent. They argued that this kind of message

may be misinterpreted because recipients could deduce that the tax department is unable

to detect some cheating. The researchers left a door open adding that if social norms

were delivered differently or more frequently, they might have an effect.

Nevertheless, most literature on social norms is positive about its effects. In addition

effects have not only been reported in the Anglo-Saxon countries but also in other

countries such as Peru (Del Carpio, 2013). The Peruvian study showed the effect of

social norms and the author cleverly spotted a complex set of mediating effects that

requires further study. Is this effect due to changes in beliefs about compliance? Is it

due to changes in the intrinsic motivation?

Given the widespread support for the effect of social norms, I should expect positive

significant effects in the treatment groups exposed to the social norm.

II.b Elements that facilitate compliance

People frequently avoid certain behaviours that they know are beneficial for them,

even when the outcome of that behaviour is greater than the effort required to perform

it. For example, people know that a pension plan will ensure financial security during

retirement, but they still fail to save enough money for their retirement. People also

know that doing regular exercise is good for their health but they just do not do it: there

is always an excuse for starting next week, next month or next year. Humans seem to

have some issues with striving for the goals they set themselves.

This not only hinders the achievement of our goals, but can even lead us to reassess

our goals and conclude that the behaviour is not worthwhile. In such cases, cognitive

dissonance takes over (Festinger, 1957). There is a gap between what we think should

be done and what we actually do, and that gap has to be bridged in our mind. So, instead

of modifying our behaviour, we may find ourselves thinking that we were not that eager

to behave in that way in the first place. For example, if we planned to start jogging but


we kept postponing it, our brain could become stressed for not doing what we think we

ought to. Changing behaviour is effortful, so we may instead say to ourselves that we

did not really like jogging anyway so we are going to find another sports activity that

better suits our preferences.

When we are uncertain about an outcome or the outcome seems out of reach, our

brain adopts a conservative mode, whereby it tries to reduce cognitive stress by doing

things as before. This is called status quo bias (Kahneman, Knetsch, & Thaler, 1991).

Thaler and Sunstein (2008) realized that workers faced complex decisions when

choosing a pension plan scheme. This complexity led to the inaction of not enrolling in

any plan, which was a terrible idea because it hindered the worker’s future wellbeing,

leaving workers without enough savings for their retirement. These authors suggested

changing the way people enrolled in such plans, making the choice easier by defaulting

to a pension plan they thought would suit the majority’s need. Thus, inactivity in this

new system meant enrolling in a good plan – a clever way of exploiting the status quo

bias. The question now is how can we make it easier for Manises citizens to make better

tax decisions?

Sometimes small details rather than big issues hinder our achievements by derailing

our efforts. These ‘frictional costs’ (Service et al., 2014) make the task effortful, and

removing such frictional costs may lead to unexpected but significant changes. The

literature contains several examples of effective interventions that make tasks easier.

Most examples focus on the use of default options – i.e. designing the choice such that

inaction implies a certain option that policy makers want to promote (P. Dolan et al.,

2012; D. G. Goldstein, Johnson, Herrmann, & Heitmann, 2008; Kunreuther & Weber,

2014). For example, countries where the organ donation form presumes consent and the

option is to opt out, instead of opt in, have higher rates of organ donation than countries

with opt in forms, despite being culturally similar (Abadie & Gay, 2006).

In the of late taxpayers in Manises, however, defaults were not a feasible. We could

consider that the inaction leads to bank account seizure, but this would mean a coercive

action, which the council was reluctant to promote. Nonetheless, I wanted to make

things easier once the payment was overdue, and the instalment payment plan was an

attractive solution for those with liquidity problems. I suspected that the instalment


payment plan would be successful because a) it transfers the economic effort to a future

self (Ariely, 2008; Frederick, Loewenstein, & O’donoghue, 2002; Thaler & Sunstein,

2008) and b) in the cases where subjects lack money, they earn some time without being

fined. Although the plan is available to all, taxpayers have to request it at the local

council, which may seem inconvenient or provide an easy excuse to postpone the

decision.

Could I facilitate this choice to improve mid-term tax compliance? Albeit less than

defaults, there is some literature supporting task simplification as an effective way to

influence behaviour. It may seem a rather obvious approach: If you want people to

choose something, just make that choice easier. It is indeed a simple approach but it

hides an interesting insight about effort and outcome: Sometimes the effort required is

small compared to the potential benefit of the outcome, yet people still do not choose

that option. In the pension plan example, it may even seem ridiculous that the little

effort of choosing a plan can overcome the immense benefit of enjoying a relaxed

retirement. For instance Leventhal, Singer, and Jones (1965) conducted an experiment

where students were given information about tetanus and then offered a shot against it

in a nearby building. Some of them were offered a map, whereas others were not. This

tiny nudge generated a significant increase in the number of students who got the shot.

As the authors concluded information alone is not enough to drive people to behave in a

certain way, but some elements that can help us steer them in the desired direction. Can

I steer Manises taxpayers to enrol in the instalment payment plan?

Once again, the work of the Behavioural Insights Team helped us apply these

concepts to tax collection. The ease with which actions can be taken is so important that

it forms part of the EAST (Easy, Attractive, Social, Timely) intervention framework.

Following this approach, the BIT ran an experiment in which the subjects were sent a

direct link to the form instead of being directed to a website that contained the form

(Service et al., 2014). This slight change meant a 4% increase in the response rate.

Furthermore, the pre-population of forms has led to successful tax collection policies

that increase tax collection, lower costs and reduce errors (Dohrmann & Pinshaw, 2009;

UK Cabinet Office - Behavioural Insights Team, 2012). Pre-population has been proved

to be an excellent device for simplifying tax compliance, and its use correlates with high

performance in tax collection, according to a McKinsey benchmark study (Dohrmann &


Pinshaw, 2009). The United Kingdom and Denmark offer useful examples of this

practice (HR Revenue and Customs, 2011). This seemed like the perfect approach for

my intervention. I could pre-populate an enrolment form and ask the taxpayer simply to

sign the form on the letter and to post it – the envelope was pre-stamped and pre-

addressed

II.c National context and tax morale

It is reasonable to expect reactions to social norms and pre-populated forms to differ

from one country to another due to culture and social characteristics. Historical reasons

may have shaped the way citizens understand taxes and their community duties. For

example, citizens in countries with older democracies and a stronger sense of

community should be more compliant with tax rules.

Remarkably, the most notable research on social norms has taken place in the United

States and the United Kingdom. Could I expect to observe differences in my experiment

because of culture? Although a detailed examination of cultural differences fell outside

the scope of my experiment, this research contributes to the literature on social norms

and ‘facilitating behaviours’ by providing evidence from a Southern European country.

Direct comparison is impossible, and I would need to run this experiment in other

Southern European cities, but a review of the literature on the impact of culture and

morale on tax collection is still worthwhile. This review not only frames my findings,

but also identifies cultural aspects that potentially require further research.

First, although contextual elements are powerful in shaping individuals’ decisions, as

humans, we use our knowledge and experience to make sense of the world. We are not

a tabula rasa and our brains are not just ‘automatic response’ machines that interact

with the surroundings. Instead the human brain has internal processes that assesses and

builds reality. From this constructivist approach, elements such as culture or social class

may be essential when analysing people’s tax behaviours.

Sharing a close relationship with the link between culture and taxes, tax morale

refers to an individual’s intrinsic motivation to pay taxes. Tax morale is especially

important when examining the limitations of the tax models based on the economics of


crime (Martinez-Vazquez & Torgler, 2009). These models, like most traditional

economics models, assume a cognitive capacity that humans lack. The decision to

commit a crime does not normally come from the careful evaluation of the benefits,

costs and probabilities of the different actions. Actually, traditional models based on the

expected utility and self-interest (Allingham & Sandmo, 1972) predict far too high non-

compliance. Comparing the probability of being caught and the fine for taxed evasion

with the savings of evading taxes, it is obvious that we should not pay taxes to the

degree that we do: Audits are uncommon and fines are small. Recognizing that this

approach just based on economic trade-offs is an unreliable representation of the human

condition, many authors (Frey & Torgler, 2007; Martinez-Vazquez & Torgler, 2009;

Wenzel, 2005) agree on the need for tax models to contain elements related to justice

and fairness. From this perspective, we can see that it is social motivation rather than

self-interest what that truly affects tax behaviour (Wenzel, 2005). Needless to say, the

idea of including social motivation clearly reinforces the use of social norms to tackle

tax-related issues.

Is Spaniards’ tax morale different from Anglo-Saxons’? According to Alm and

Torgler (2006) Spaniards have a lower intrinsic motivation for paying taxes than

Americans. Examining Spain’s historical and cultural context we can glimpse some

reasons: a young democracy and with a long history of picaresque attitudes and

behaviours. Furthermore, in Spain, the unemployment rate has reached 25% and

political corruption scandals are rife. According to the Centre of Sociological Research

these issues have been the two major concerns for Spaniards in recent years. Hence the

socioeconomic context may have affected the way citizens perceive their tax duties.

The current situation is comparable to the situation in the early 1990s, when Spain

experienced similar unemployment rates and major political corruption scandals

surfaced. This situation importantly affected Spanish tax morale, which decreased

according to Martinez-Vazquez and Torgler (2009).

Other studies have explored the potential effects of corruption and economic

downturn. Hofmann, Hoelzl, and Kirchler (2008) reported lower tax compliance when

the perceived compliance of the subject’s group drops – i.e. when there is corruption.

Wenzel (2005) emphasized that social norms work only when there is a strong


identification with the group – something that may be hindered by corruption and

unemployed/employed gap – and Michael Hallsworth et al (2014) acknowledged the

effect of a lack of liquidity on tax compliance, which we could apply to the current

economic situation. It is reasonable to think that the current socio-economic situation is

similar to the one the situation in the mid-1990s, when tax morale in Spain was indeed

low.

Finally, my experiment does not focus on evasion, but on the amount of money owed

to the local council, something subjects are certain of and cannot change. Nevertheless,

we may still assume that tax morale could influence self-indulgence regarding the delay

in tax payment. If tax morale relates to the internalisation of the motivation, according

to self-determination theory (Ryan & Deci, 2000), low levels of tax morale may be

more effectively dealt with external regulations, such as external negative rewards and

punishments, than with internal regulations such as self-identification and shared

values. This raises an interesting question: Are Spaniards more reactive to authority or

to the appeal of group identification? In Spain, coercive elements may have a stronger

effect than identification elements such as the social norm.


III. THE EXPERIMENT

III.a Design and Methodology

The best method to test my hypothesis was an experiment. Experiments allow us to

understand causality by setting a control group and controlling other effects through

randomization (Burtless & Burtless, 1995). If I were to run the intervention and just

compare it with past data, I could miss some contextual effects such as the economic

situation, a significant event or even the weather. I therefore designed an RCT in which

I not only tested the intervention using treatment groups that are independent – non-

interference – but also assigned subjects randomly to each group to avoid selection bias

(Duflo, Glennerster, & Kremer, 2007). Experiments are a powerful tool compatible with

my aims of understanding the effects of stimuli in a letter. After all it has been the main

tool in behavioural sciences and many researchers have used RCTs to study similar

research problems (M. Hallsworth, 2015; John, Sanders, & Wang, 2014).

I ran a natural field experiment (NFE) to prevent subjects from behaving differently

just because they were part of the experiment (experimenter demand effects,

randomisation bias, the Hawthorne effect or John Henry Effect). Unlike lab

experiments, natural field experiments overcome these biases, albeit at the expense of

losing control over contextual variables. Natural field experimentation is sound

approach that has been used extensively for assessing policy interventions (Burtless &

Burtless, 1995; Haynes, Service, Goldacre, & Torgerson, 2012).

The experiment was based on letters sent to late taxpayers of Manises. Together with

the council team, I varied the content of these letters to study how effectively social

norms and pre-populated forms resolved the problem of late tax payment.

First, to measure the effect of the descriptive social norm, we included a line of text

similar to the one used in the BIT experiment (UK Cabinet Office - Behavioural

Insights Team, 2012). This line of text stated the percentage of citizens of Manises who

paid their taxes on time – over 90% - phrased as ‘9 out of 10’. Second, because I wanted

to encourage the late tax payers to enrol in the instalment payment plan, I prepared a


pre-populated form that required the subject to complete his or her bank details and sign

the form.

I started by creating a database with all outstanding tax payments. Each payment was

linked to an individual, who was assigned to one the treatment or control group. I

established five groups:

> Group 1 was the control group. Individuals in this group did not receive any

letter.

> Group 2. Individuals in this group received a letter stating the descriptive

social norm.

> Group 3. Individuals in this group received a letter offering the instalment

payment plan with the pre-populated form.

> Group 4. Individuals in this group received a letter stating the social norm

and offering the instalment payment plan with the pre-populated form.

> Group 5. Individuals in this group received a letter that that neither stated the

social norm nor offered the instalment payment plan with the pre-populated

form. Having Group 5 allowed me to measure the effects of the other groups,

whilst controlling for the reminder effect of receiving a letter sent.

I clustered the payments by individuals – to ensure that each individual was exposed

to just one treatment – and randomly assigned each individual to a group using the

random number function in Excel. The groups had between 2,346 and 2,440

outstanding payments, which corresponded to around 700 individuals per group. The

sample was large enough to capture changes of half the standard deviation, with a 5% of

significance level and 80% power.


Once the database was ready, I prepared the letters. The front of the letter contained

a text reminding the subject about the outstanding tax payment and, depending on the

group, stating the social norm statement and/or providing the pre-populated form for

enrolling in the instalment payment plan. Box 1 shows the different variations of the

text in the letter that I prepared with Manises council. I used the regular tax collection

paper for the letters so the back of the letter contained a payment form and instructions

on how to pay the outstanding amount.

BOX 1

Dear Sir. / Madam.

According to our records you have an outstanding debt of 208.96 euros with the city of

Manises.

[IF IN GROUPS 2 OR 4] It is an effort that nine out of ten residents of the municipality

have already made and we are convinced so will do soon.

The payment is simple. Find attached the document for the total amount that you can pay

by card at the Municipal Income Collection Office or via deposit it at offices of Bank X.

[IF IN GROUPS 3 OR 4] However, if you need to pay in instalments, you can sign this

authorisation to start the instalment payment plan. Please complete your bank account

details.

[FORM WITH OUTSTANDING DEBT, NUMBER OF INSTALMENTS, AMOUNT

PER INSTALMENT, RECORD NUMBER ALREADY COMPLETED / BANK

ACCOUNT AND SIGNATURE TO BE COMPLETED]

Thanking you in advance for your attention. Kind regards.

*When referring to the ‘document’ they mean a piece of the letter that can be detached

with the outstanding payment information.

On 16 March 2015, the council IT person took an ‘image’ of the database and then

sent the letters. I agreed with the local council team that one month and a half should be

enough time for subjects to respond, so the IT person took another image of the

database on 30 April 2015 and created a database with the pre-experiment data and the

post-experiment data.

After cleaning and preparing the database, I first transformed the information of the

group variable into four dummy variables: G2, G3, G4 and G5. The dummy variables


identified each payment as pertaining to one of the treatment groups (G2 for group 2,

G3 for group 3 and so on, leaving group 1 as the baseline).

Second I created dichotomous variables to classify the possible outcomes: a) the debt

was fully paid without the instalment plan, b) the debts were fully paid via the

instalment plan, c) the debt was partially paid via the instalment plan or d) the subject

had not responded. The variables were defined as follows: C identified completely paid

debts – either by direct payment or by the instalment payment plan. R identified

responses either by paying or by enrolling in the instalment payment plan. Finally, F

identified subjects who were enrolled in the instalment plan. In addition to this I

created another set for the two instalment plan possibilities: enrolled and paid the whole

amount (F-PT) and enrolled and with outstanding debts (F-PP).

Third, because I wanted to understand the separate effects of the social norm, the

pre-populated form and the letter itself, merely defining the treatment and control

groups was not enough. Using the information of the groups in a regression would show

whether each of the groups had a significantly different response from the control

group. However, effects can become mixed and may be hard to identify in such

analysis. For example, I could not tell whether a significant increase in the response rate

in the Group 2, was due to the social norm or just having a letter from the local council

– as a reminder. Even if I were to compare the different coefficients of the group

dummy variables, I would not be sure about the significance. I needed to perform some

further analysis.

To control for these potential misunderstandings I created four more dichotomous

variables that classified the interventions based on the mechanism that used to influence

taxpayers’ behaviour (see in Table 1): a) letter, which identified the groups that had

received a letter (i.e. all but Group 1); b) social, which identified the groups that had

received a letter with the social norm (i.e. Groups 2 and 4); c) fform, which identified

the groups that had received a letter with the pre-populated form for the instalment

payment plan (i.e. Groups 3 and 4); and d) socialform, which identified the group that

received a letter with both the social norm and the pre-populated form (i.e. Group 4).


I thus obtained variables that would explain the effect of just sending a letter (letter),

the effect of the social norm (social), the effect of the pre-populated form (form) and the

combined effect of having both the social norm and the pre-populated form

(socialform). Regressing these variables against response or payment gave us a clear an

idea about the real effect of each mechanism.

TABLE 1

The relationship between the ‘mechanism’ variables and the treatment/control groups was as

follows: (being dichotomous variable, 1 makes the statement positive and 0 makes it negative):

Group 1 Group 2 Group 3 Group 4 Group 5

The subject has received a

letter (letter) 0 1 1 1 1

The letter received

contained the Social Norm

(social)

0 1 0 1 0

The letter received

contained the pre-

populated form (fform)

0 0 1 1 0

The letter received

contained both the social

norm and the pre-

populated form

(socialform)

0 0 0 1 0

For example, individuals in Group 3 a) received a letter that b) contained a pre-populated form.

Finally, I had information regarding the sex and date of birth of the subjects.

Although this information was missing for some subjects, it was worth analysing and I

created a dummy variable for sex (female) and a continuous variable for age calculated

as the numbers of days alive (age).


Having prepared the database, I could now analyse the data. To draw significant

conclusions from experiment I regressed the independent variables R, C and F against

the dependent the dummy variables of the groups – G2, G3 G4 and G5 – and with the

dependent variables letter, social, fform and socialform. I did these regressions

controlling for age and sex.

For this regression I used a logit model. As the independent variable was

dichotomous, I had to use a probit or logit model. I decided to use the logit model

because I could get some information regarding odds using the odds version of logit in

Stata.

III.b Findings

The regression of the independent variables R, C and F taking the dummy variables

for the treatment/control groups as dependent variables provided information about the

implications of being in each of the groups.

As FIGURE 1 shows, all group variables had a significant and positive effect on the

response rate for a 5% significance level (α = 0.05). Whereas the z value was high for

G2 and G3 – leaving virtually no doubt of the effect on R of being in such groups –, the

z value for G4 (z = 1.96) was borderline significant. Both control variables had positive

and significant effects. Hence, older debtors were more likely to respond, as were

female debtors.

FIGURE 2 shows that the treatment in Groups 2 and 5 had a significant effect on the

rate of paid debts, with larger coefficients than in the R regression. Understandably,

groups with letters that contained a nudge for enrolling in the instalment payment plan,

which implies deferring the payment, had lower rates of full debt payment. In this case

age and sex had positive and significant effects, which shows how important these

elements are for explaining the behaviour of the individuals in the sample.


FIGURE 1

FIGURE 2

FIGURE 3


Unsurprisingly being in Group 3 had a strong positive effect on the rate of enrolment

in the instalment payment plan (FIGURE 3). Being in Group 4, however, did not have

a significant effect. When I designed the experiment I assumed that the combination of

social norms and pre-populated forms would have a stronger effect than each treatment

separately. However, Group 4 had lower rates of response, payment and enrolment in

the instalment payment plan. Instead of enhancing compliance, the combination of

social norms and pre-populated forms seemed to confuse the subject. Ex-post the

finding is reasonable because the letter gave two contradictory messages: ‘Everyone

pays on time’ (social norm) and ‘You can delay your payment’ (pre-populated form).

Interestingly, in the case of the rate of enrolment in the instalment plan, whereas sex

still had a significant effect, the significance of age dropped substantially. Does this

mean that older people find it harder to use new payment systems?

Although the effects on R were positive and significant for all the groups and

positive for F and C, it is interesting to examine the mechanisms that affect behaviour.

To go beyond the analysis of the effect of being in each group, I regressed the variables

referring to mechanisms: social, fform, socialform and letter. My aim was to observe

whether the effect of the groups was due to the psychological mechanism or just to the

fact of having received a letter.

FIGURE 4 and FIGURE 5 show how the effect in Groups 2, 3 and 4 were due to

the letter. Social and fform variables were non-significant when regressing C and

significant only at α = 0.10 for R. I must admit that this was a pretty shocking denial of

my hypothesis, but it demonstrated that in the behavioural sciences, nothing is definitive

and everything requires testing.


FIGURE 4

FIGURE 5

FIGURE 6


As expected, the effect of combining social norms and the pre-populated form was

negative but non-significant for R and C. In contrast, the effect for F was negative and

significant. Hence, the combination of social norms and the pre-populated form for the

instalment payment plan had a negative synergy, thereby thwarting our goal. Bearing

this finding in mind, I ran regressions excluding Group 4 (social norms and pre-

populated forms) to understand whether this group was the cause for the poor results of

social norms. I found no significant differences.

Fortunately, whilst the use of social norm did not significantly affect any of the

independent variables, having a pre-populated form had a positive and significant effect

when measuring the rate of enrolment in the instalment payment plan (FIGURE 6). Hence, the letter containing the pre-populated form did indeed affect the rate of

enrolment in the plan. Facilitation is an effective mechanism for late taxpayers in

Manises.

Results for the control variables were unsurprising. The control variables had a good

explanatory power for almost all independent variables. Sex and age were significant in

predicting R and C; older people and women were more likely to respond and pay their

debts. For the prediction of F, however, only sex had a significant effect. As previously

suggested, this finding may be due to that fact that completing a form, such as the one I

sent, entails understanding new ways of payment that older people may find confusing.

The findings regarding sex raised questions. Did women have greater tax morale? If

so, did the social norm affect women differently from how it affected men? I realized

that men owed 72% of the original outstanding payments. I ran another regression on R

taking letter, social, fform and socialform as dependent variables, but only for the

women in the sample. To my surprise I observed a considerable change in p levels

(FIGURE 7): social norms did explain R for women (when regressing C, p was 0.07

not-significant at 5% but significant at 10%). The negative effect of the combination of

social norms and pre-populated form was clearly significant. Furthermore, for women,

receiving a letter (letter) was non-significant for R and for C. For women it was about

the content not the letter.


FIGURE 7 *logit regression of R omitting men

Overall receiving any kind of letter from the local authorities was what led the

recipients to respond and pay – unless the subject was a woman in which case the social

norm was the driver. In addition, receiving a letter with a pre-populated form for an

instalment payment plan led recipients to enrol in the plan. We can better understand the

implications of these findings by examining the odds. Running a logit with odds ratios

revealed that receiving a letter increased the odds of responding by a factor of 1.58 and

increased the odds of paying the whole debt by a factor of 1.66. Moreover, receiving a

letter with the pre-populated form increased the odds of enrolling in the instalment

payment plan by a factor 3.28.

Although these logit regressions yielded some significant results, the explanatory

power of the models was rather limited. The pseudo-R2 values were low in all

regressions, ranging from 0.0096 to 0.0268. These results imply high variation in the

independent variables due to other elements that are not part of my model. Therefore we

cannot fully explain what drives Manises late taxpayers’ behaviour.


III.c Discussion

For the local council, these interventions were a success. Payments increased and

taxpayers enrolled in payment plans, which was the goal of the intervention. The main

concern for the council was to increase citizens’ use of the instalment payment we

successfully achieved this goal using pre-populated forms. This way of addressing the

issue boosted compliance without coercion. Nevertheless, the general result for the

effect of social norms was disappointing because of its low significance.

It was no less surprising that social norms had a significant effect when the sample

was reduced to just women. Yet we should interpret this insight with precaution. Men

owed almost three-quarters of the original outstanding payment, so some other factor

may explain why the effect of social norms was significant for women. I first thought

this effect might be due to age. I ran a T test and found that women in the database were

significantly younger than men were. Could it be that younger people were more

sensitive to social norms? I created a variable dividing the sample into groups above

and below the mean age. Once again, however, the social norms effect was non-

significant for both groups. It may be that women have greater tax morale, or it may be

that there was something I could not see.

Although social norms worked for woman, 75% of individuals with outstanding tax

debts were men, so the effect of the social norms remained a concern to me. Perhaps I

found my results hard to believe because of the salience and the extensive use of this

element in the behavioural science literature. I even ran other tests to check the validity

of my findings. An ANOVA test of R and social only for those who had received the

letter would show whether the mean of the response rate (R) differed for people who

received a letter stating the social norm and people who received a letter that did not

state the social norm. The results, however, just confirmed that social norms did not

have a significant effect. The p value was 0.77 (FIGURE 8).


FIGURE 8

Accepting that the social norms did not have the expected effect, I can propose

several reasons to explain this finding:

a) As explained by Alm and Torgler (2006) Spaniards have a looser tax morale

than the citizens of the UK or the USA, and this may be hindering the effects of

social norms on tax compliance.

b) Today’s economy has undermined citizens’ capacity to pay taxes and they

cannot face their debts.

c) High levels of perceived corruption – recall that political corruption has in

recent years been the second biggest concern for Spaniards, after

unemployment – may have distorted how citizens perceive their tax duties.

d) The reminder effect was so strong that it overrode the social norm effect. In

other words, merely being reminded was sufficient motivation for the

individuals that might have been affected by the norm. Hence, the reminder

effect prevented us from observing the social norm effect..

All explanations seem reasonable but given that the letter affected taxpayer’s

behaviour, we can discount explanation b). Explanations a) and c) relate to cultural

values either structural or temporary, whereas d), in practical terms, means that social

norms are currently irrelevant for policy making in Manises.

There is little to add to explanation c). I would need to run further experiments

controlling for perceived corruption and analyse whether this variable could explain part

of the variance.


Delving into the implications of explanation a), we can argue that the very fact of

having a powerful letter effect implies that tax morale is indeed low. Tax morale

understood as internal motivation relates to, according to Self-Determination Theory

(Ryan & Deci, 2000), internal regulations such as valuing or identification, such as

social norms. If Spaniards were externally – rather than internally – motivated to pay

their taxes, the most effective interventions would be based on external regulations such

as punishments, fines or claims. Because of the youth of Spain’s democratic values, it

may seem to some extent legitimate to cheat the system. Yet when authority calls the

individual out, the individual reacts due to fear of punishment.

Some aspects related to the wording or salience of the phrases used in the letter may

causing a lower impact. For future research it may be interesting to change the wording,

the boldness of the sentence or even reinforce the message with a photo of some

neighbours – an effect that has been an effective way to influence behaviour (P. Dolan

et al., 2012).

Finally, combining these two elements (i.e. social norms and payment in instalments)

to improve tax compliance actually led to a decrease in the effect. I would not have been

able to reach this conclusion without the use of a variable measuring the combined

effect of social norms and the pre-populated form. In hindsight, we can see that whilst

the social norm emphasized the importance of paying on time, the pre-populated form

was encouraging just the opposite. I overlooked this contradiction when designing the

experiment, but, of course, everything is obvious once you know the answer (Watts,

2012).


IV. CONCLUSION AND LEARNING

I started with the simple idea of replicating the Behavioural Insights Team’s

experiment on tax collection and social norms. Together with the council I agreed on

adding a second element, namely the pre-populated forms to aid enrolment in an

instalment payment plan, I thought the social norm effect would eclipse any effect of

the form. Furthermore, I expected the combination of both social norm and the pre-

populated form to enhance the overall effect. I believed this latter treatment group

would clearly demonstrate that social norms push any behaviour in the direction I want,

in this case boosting the effect of the pre-populated form.

Reality is that, although all treatment group variables had a significant and positive

effect on increasing the response rate, the main reason for that effect was the letter

rather than the social norms – at least not for men, who made up the vast majority of the

debtors. Nonetheless, the pre-populated form significantly affected the rate of enrolment

in the instalment payment plan and could help to improve Manises’ tax collection

processes.

Failing to finding a significant effect from use of social norms was the most relevant

lesson I learnt from this dissertation: everything, absolutely everything, has to be tested

against reality. Having tested my hypothesis, I can now appreciate that social norms –

any behavioural mechanism I would say – do not work universally as a physical

principle does. Although it would be convenient for behavioural scientists to have

unequivocal rules, the vast range of variables that affect human behaviour makes

establishing hard and fast laws impossible.

The second key lesson I learnt from this project is that just measuring the effect of

the treatment groups is not enough to understand the mechanism behind such an effect.

As researchers we must disentangle all the mechanisms that could explain the behaviour

and analyse these mechanisms separately. In my case using a group that received a letter

with a neutral message helped me understand the real effect of social norms.

Combining the norm and the form had a significant effect in some of the models yet

the effect was always negative. Using both elements decreased the response rate. I now


believe this could be due to the recipient’s potential confusion when facing two

contradictory messages: pay now (social norm) vs. pay later (pre-populated form). It

would be interesting to conduct further research, using a social norm about the people

who enrolled in the plan and test whether this approach is improving the response rate

and tax collection.

More research is definitely needed to better understand the complex processes

behind social norms. As Del Carpio (2013) noted, different phenomena mediate the

effects of social norms. To understand if social norms fail to work because of factors

such as the economy, culture or perceived level of corruption, I would need to run

experiments in different Spanish cities or similar cultural regions – i.e. Mediterranean

cities –, run an experiment when the economic crisis is over and run an experiment

controlling for the level of perceived corruption.

From my point of view the most plausible reason why social norms did not work in

my experiment was a lower intrinsic motivation to pay taxes compared to other

countries – consistent with the findings on tax morale of Alm and Torgler, (2006). This

extrinsic motivation feature could lead to more effective interventions based on external

regulation elements (e.g. a letter from the local council which is a sign of authority)

rather than internal regulation elements (e.g. social norms which invoke group

identification.)

Overall, this has been a great experience. I have learnt that behavioural sciences is

not about implementing ideas read in an ‘airport book’ but rather about researching the

mechanisms that behind the hypothesis, testing them with experiments, improving the

hypothesis and test them again. There are many ways to improve our world if we

understand our decisions and create policies based on these insights. But we cannot

forget that there is no such thing as a universal truth in our field: after all… context

matters.


REFERENCES Abadie, A., & Gay, S. (2006). The impact of presumed consent legislation on cadaveric

organ donation: A cross-country study. Journal of Health Economics, 25(4), 599–620. doi:10.1016/j.jhealeco.2006.01.003

Allcott, H. (2011). Social norms and energy conservation. Journal of Public Economics, 95(9-10), 1082–1095. doi:10.1016/j.jpubeco.2011.03.003

Allingham, M., & Sandmo, a. (1972). Income Tax Evasion: A, 1, 323–338. doi:10.1016/0047-2727(74)90037-1

Alm, J., & Torgler, B. (2006). Culture differences and tax morale in the United States and in Europe. Journal of Economic Psychology, 27, 224–246. doi:10.1016/j.joep.2005.09.002

Ariely, D. (2008). Predictably Irrational. Harper Collins.

Blumenthal, M., Christian, C., Slemrod, J., & Smith, M. G. (2001). Do Normative Appeals Affect Tax Compliance? Evidence from a Controlled Experiment in Minnesota. The National Tax Journal, 54(1), 125–138.

Burtless, G., & Burtless, G. (1995). The Case for Randomized Field Trials in Economic and Policy Research. Journal of Economic Perspectives, 9(2), 63–84.

Cialdini, R. B. (2003). Crafting normative messages to protect the environment. Current Directions in Psychological Science, 12(4), 105–109. doi:10.1111/1467-8721.01242

Cialdini, R. B., Reno, R. R., & Kallgren, C. a. (1990). A focus theory of normative conduct: Recycling the concept of norms to reduce littering in public places. Journal of Personality and Social Psychology, 58(6), 1015–1026. doi:10.1037/0022-3514.58.6.1015

Coleman, S. (2007). The Minnesota Income Tax Compliance Experiment: Replication of the Social Norms Experiment.

Del Carpio, L. (2013). Are the neighbors cheating? Evidence from a social norm experiment on property taxes in Peru. Work. Pap., Princeton Univ., Princeton, NJ, (November).

DellaVigna, S. (2009). Psychology and Economics: Evidence from the Field. Journal of Economic Literature, 47(2), 315–372. doi:10.1257/jel.47.2.315

Dohrmann, T., & Pinshaw, G. (2009). The Road to Improved Compliance. McKinsey&Company.

Dolan, P., & Galizzi, M. M. (2015). Like ripples on a pond: Behavioral spillovers and


their implications for research and policy. Journal of Economic Psychology, 47(January), 1–16. doi:10.1016/j.joep.2014.12.003

Dolan, P., Hallsworth, M., Halpern, D., King, D., Metcalfe, R., & Vlaev, I. (2012). Influencing behaviour: The mindspace way. Journal of Economic Psychology, 33(1), 264–277. doi:10.1016/j.joep.2011.10.009

Duflo, E., Glennerster, R., & Kremer, M. (2007). Using Randomization in Development Economics Research: A Toolkit. Handbook of Development Economics, 4(07 Chapter 61), 3895–3962. doi:10.1016/S1573-4471(07)04061-2

Festinger, L. (1957). A Theory of Cognitive Dissonance. California: Standford University Press.

Frederick, S., Loewenstein, G., & O’donoghue, T. (2002). Time Discounting and Time Preference: A Critical Review. Journal of Economic Literature, 40(2), 351–401. doi:10.1257/002205102320161311

Frey, B. S., & Meier, S. (2004). Social Comparisons and Pro-social Behavior : Testing in a Field Experiment “ Conditional Cooperation .” The American Economic Review, 94(5), 1717–1722. doi:10.1257/0002828043052187

Frey, B. S., & Torgler, B. (2007). Tax morale and conditional cooperation. Journal of Comparative Economics, 35(1), 136–159. doi:10.1016/j.jce.2006.10.006

Gahramanov, E. (2009). The Theoretical Analysis of Income Tax Evasion Revisited. Economic Issues, 14(1974), 35–41.

Goldstein, D. G., Johnson, E. J., Herrmann, A., & Heitmann, M. (2008). Nudge your customers toward better choices. Harvard Business Review, 86(August), 2008.

Goldstein, N. J., Cialdini, R. B., & Griskevicius, V. (2008). A Room with a Viewpoint: Using Social Norms to Motivate Environmental Conservation in Hotels. Journal of Consumer Research, 35(3), 472–482. doi:10.1086/586910

Hallsworth, M. (2015). The use of field experiments to increase tax compliance. Oxford Review of Economic Policy, 30(4), 658–679. doi:10.1093/oxrep/gru034

Hallsworth, M., List, J. ., Metcalfe, R. D., & Vlaev, I. (2014). The Behavioralist As Tax Collector : Using Natural Field Experiments to Enhance Tax Compliance, 1–41.

Hashimzade, N., Myles, G. D., & Tran-Nam, B. (2013). Applications of behavioural economics to tax evasion. Journal of Economic Surveys, 27(5), 941–977. doi:10.1111/j.1467-6419.2012.00733.x

Haynes, L., Service, O., Goldacre, B., & Torgerson, D. (2012). Test , Learn , Adapt : Randomised Controlled Trials.

Hofmann, E., Hoelzl, E., & Kirchler, E. (2008). Preconditions of Voluntary Tax Compliance: Knowledge and Evaluation of Taxation, Norms, Fairness, and


Motivation to Cooperate. Zeitschrift Fur Psychologie, 216(4), 209–217. doi:10.1027/0044-3409.216.4.209

HR Revenue and Customs. (2011). Modernising the administration of the personal tax system : Tax Transparency for Individuals. [discussion document].

John, P., Sanders, M., & Wang, J. (2014). The use of descriptive norms in public administration: a panacea for improving citizen behaviours? Working Paper, 1–28.

Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1991). Anomalies: The Endowment Effect, Loss Aversion, and Status Quo Bias. Journal of Economic Perspectives, 5(1), 193–206. doi:10.1257/jep.5.1.193

Kunreuther, H., & Weber, E. U. (2014). Aiding Decision Making to Reduce the Impacts of Climate Change. Journal of Consumer Policy, 1–15. doi:10.1007/s10603-013-9251-z

Lapinski, M. K., Maloney, E. K., Braz, M., & Shulman, H. C. (2013). Testing the Effects of Social Norms and Behavioral Privacy on Hand Washing: A Field Experiment. Human Communication Research, 39(1), 21–46. doi:10.1111/j.1468-2958.2012.01441.x

Leventhal, H., Singer, R., & Jones, S. (1965). Effects of Fear and Specificity of Recommendation Upon Attitudes and Behavior. Journal of Personality and Social Psychology, 34(1), 20–29. doi:10.1037/h0022089

Loewenstein, G. F., Weber, E. U., Hsee, C. K., & Welch, N. (2001). Risk as feelings. Psychological Bulletin, 127(2), 267–286. doi:10.1037/0033-2909.127.2.267

Martin, S. J., Goldstein, N., & Cialdini, R. B. (2014). The small BIG: small changes that spark big influence. Profile Books Ltd. Retrieved from http://www.amazon.com/The-small-BIG-changes-influence/dp/1455584258

Martinez-Vazquez, J., & Torgler, B. (2009). The Evolution of Tax Morale in Modern Spain. Journal of Economic Issues, 43(1), 1–28. doi:10.2753/JEI0021-3624430101

Ryan, R. M., & Deci, E. L. (2000). Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. The American Psychologist, 55(1), 68–78. doi:10.1037/0003-066X.55.1.68

Sen, A. (1977). Rational Fools: A Critique of the Behavioral Foundations of Economic Theory. Philosophy & Public Affairs, 6(4), 317–344. doi:10.2307/2264946

Service, O., Hallsworth, M., Halpern, D., Algate, F., Gallagher, R., Nguyen, S., … Kirkman, E. (2014). EAST Four simple ways to apply behavioural insights, 53. Retrieved from http://www.behaviouralinsights.co.uk/sites/default/files/BIT Publication EAST_FA_WEB.pdf

Thaler, R. H., & Sunstein, C. R. (2008). Nudge. Focus.


Trope, Y., & Liberman, N. (2010). Construal-level theory of psychological distance. Psychological Review, 117(2), 440–463. doi:10.1037/a0018963

UK Cabinet Office - Behavioural Insights Team. (2012). Applying Behavioural Insights to reduce Fraud, Error and Debt, 38.

Watts, D. J. (2012). Everything is Obvious: How Common Sense Fail us. Crown Business.

Wenzel, M. (2005). Motivation or Rationalisation? Causal Relations Between Ethics, Norms and Tax Compliance. Journal of Economic Psychology, 26(4), 491–508. doi:10.1016/j.joep.2004.03.003

using social norms and pre populated forms to improve tax collection in manises

Documents