with my money on my mind: income, happiness and intrusive … · and positive (stevenson &...

24
1 With my money on my mind: income, happiness and intrusive financial thoughts Paul Dolan 1 and Robert Metcalfe 2 1 London School of Economics 2 University of Oxford Abstract Richer people are more satisfied with their lives than poorer people but are no happier with their daily experiences. We replicate these findings in the first ever panel dataset using the day reconstruction method (DRM). We find that richer people, however, are more satisfied with their days and less likely to be very unhappy. We also demonstrate that poorer people are more likely to report negative intrusive thoughts about money and having these thoughts is associated with lower life satisfaction and daily happiness. Interestingly, richer people are more adversely affected in the DRM by each negative thought about money. Intrusive thoughts have not been properly accounted for in reports of happiness to date. More research is needed on how positive and negative thoughts about money, health, family etc. affect happiness and how these thoughts may affect the relationship between happiness and its various determinants, including income. Keywords: intrusive thoughts; income; experienced utility JEL Classification: D0 Acknowledgements We thank Anja Göritz, Anthony Mee and Sabine Pahl for help with programming, translation and data collection. Andrew Oswald, Daniel Gilbert, Nick Powdthavee and Ivo Vlaev all provided helpful comments. Matthew White was involved in the design of the study and so is owed a very special thank you.

Upload: others

Post on 25-Jul-2020

7 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

1

With my money on my mind: income, happiness and intrusive financial thoughts

Paul Dolan1 and Robert Metcalfe2

1 London School of Economics

2 University of Oxford

Abstract

Richer people are more satisfied with their lives than poorer people but are no happier with their

daily experiences. We replicate these findings in the first ever panel dataset using the day

reconstruction method (DRM). We find that richer people, however, are more satisfied with their

days and less likely to be very unhappy. We also demonstrate that poorer people are more likely

to report negative intrusive thoughts about money and having these thoughts is associated with

lower life satisfaction and daily happiness. Interestingly, richer people are more adversely

affected in the DRM by each negative thought about money. Intrusive thoughts have not been

properly accounted for in reports of happiness to date. More research is needed on how positive

and negative thoughts about money, health, family etc. affect happiness and how these thoughts

may affect the relationship between happiness and its various determinants, including income.

Keywords: intrusive thoughts; income; experienced utility

JEL Classification: D0

Acknowledgements We thank Anja Göritz, Anthony Mee and Sabine Pahl for help with programming, translation and data collection. Andrew Oswald, Daniel Gilbert, Nick Powdthavee and Ivo Vlaev all provided helpful comments. Matthew White was involved in the design of the study and so is owed a very special thank you.

Page 2: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

2

1. Introduction

An increasingly prominent topic in economics is the relationship between income and happiness

(Blanchflower & Oswald, 2004; Easterlin, 2001; Frank, 1999; Frey & Stutzer, 2002; Layard,

2006). Most of the research has focussed on comparisons between the log of household income

and life satisfaction ratings, which ask respondents to consider their life as a whole (Dolan et al,

2008). Despite countless studies and several reviews, there is still no consensus about the nature

of this relationship (Clark et al, 2008) although the best evidence to date suggests it is moderate

and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011).

An alternative approach to measuring happiness, and one closer to Jeremy Bentham's original

conception of utility, is to measure experienced utility as the flow of feelings over time

(Kahneman et al, 1997). The Day Reconstruction Method (DRM) is an innovative approach to

measuring the experienced utility associated with daily activities. It asks people what activities

they were engaged in during the previous day, how long these lasted and how they felt during

them (Kahneman et al, 2004). Further work using the DRM has examined the difference between

affect and reward (White & Dolan 2009), and the difference between life satisfaction and affect

for those who are unemployed versus employed (Knabe et al, 2009). The activity-related

emotions can then be weighted by their duration to produce a profile of experienced utility over

the course of an entire day. From DRM data on Texan women, Kahneman et al (2006) found no

significant relationship between household income and experienced utility – despite respondents

with higher incomes reporting higher life satisfaction.

This discrepancy has been explained as an example of a focussing illusion whereby people focus

their attention on income when evaluating their lives but pay far less attention to it in the actual

experience of their lives (Kahneman et al, 2006; Schkade & Kahneman, 1998; Wilson & Gilbert,

2003). Richer women's actual experiences were possibly no better because, according to their

time use data, they spent more time in relatively less pleasant activities, such as work and

commuting. Kahneman et al (2006) conclude that "the belief that high income is associated with

good mood is widespread but illusory" (p.1908).

Page 3: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

3

Yet the DRM's attempt to capture experienced utility may have created a focussing effect of its

own by asking respondents to report their feelings when thinking about the activities in their

lives. As such, it may have neglected the way in which our attention drifts between current

activities and concerns about other things. Research in psychology suggests such "mind

wanderings" are frequent, occurring in up to 30% of randomly sampled moments during an

average day (Smallwood & Schooler, 2006). When these mind-wanderings repeatedly return to

the same issues, they are labelled intrusive thoughts and they often have a negative association

with our experienced utility (Watkins, 2008; Killingsworth & Gilbert, 2010). It has also been

found that such thoughts have a large impact on the valuation of health states (Dolan, 2010).

So far as we are aware, intrusive thoughts have been researched by psychologists in clinical

settings (Watkins, 2008). It is plausible that intrusive financial thoughts (IFTs) may be more

frequent amongst poorer people in the general population and there may therefore be a stronger,

positive relationship between income and experienced utility when we account for these

thoughts. IFTs have not been previously investigated in general population samples, which is

somewhat surprising given that money worries are one the most common everyday concerns

(MacGregor, 1991). Moreover, the DRM studies to date have been cross-sectional (Kahneman et

al, 2004; Knabe et al, 2010) and have therefore been unable to control for unobserved variables

that might lead to biased estimates of experience utility.

In a DRM panel study in Germany, we test whether IFTs are related to lower levels of happiness

(life satisfaction and experienced utility). Because people who have IFTs may be of a more

worried disposition to begin with, we also ask about intrusive thoughts for a range of other

domains (e.g. health). If IFTs remain significant controlling for a general tendency to report

intrusive thoughts then this would provide stronger evidence for the impact of IFTs. Our results

suggest that negative IFTs matter for happiness (as defined by all measures), in that those who

report IFTs have lower levels of happiness, even controlling for individual heterogeneity and the

tendency to report other intrusive thoughts. We also find that each IFT has a greater effect on the

experienced utility of richer people, and that including negative IFTs in a life satisfaction

regression reduces the size of the income coefficient, enabling us to reconcile the results between

life satisfaction and DRM Affect.

Page 4: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

4

2. Methods

Participants were recruited via a web-based internet panel run from a German University in

2007. Of the 1,825 people who accessed the study's site for more details, 625 (34%) completed

the study. This response rate is normal for this and other panels (Göritz, 2007) and provided a

highly heterogeneous sample. In 2008, 169 people who completed the DRM survey (see Table 1

for the summary statistics of the 169 panel sample). Household income was measured using nine

categories with increments of €20,000, so the mid-point of each category was used and raw

scores were log-transformed.

Following Kahneman et al (2004), respondents began by completing a diary of their previous

day. For each episode, they were asked to select what they were doing, from a list of 24 activities

(e.g. eating), say who they were with, from a list of twelve (e.g. boss), and how long the episode

lasted. They were also asked the extent to which they felt a number of emotions on scales from 0

(Not at all) to 6 (Very much). For consistency with Kahneman et al (2004), we calculate a DRM

Affect by weighting the average of the positive affect terms (happy, engaged, content) minus the

average of the negative affect terms (worried, tired, nervous). For consistency with Kahneman &

Krueger (2006) and Knabe et al (2010), we also calculate a DRM U-index, which is the

proportion of each person’s time engaged in an activity for which dominant (highest scoring)

emotion was negative. This is reverse coded in the data, so higher coefficients in the regressions

suggest a higher likelihood of being unhappy.

Respondents were then asked about intrusive thoughts: “So far you have described certain

episodes during your day and how you felt. However, perhaps certain thoughts kept popping into

your head during the day which had little to do with the activities you were engaged in. In this

section we would like you to say if you had any of these thoughts (positive or negative) during

the day." They were then asked: "Was there something positive or pleasant that kept popping

into your head during the day and which made you happy when you thought about it? yes or no.

If yes then what were these thoughts about?” There were ten domains to select from: finances,

work, holidays, family, health, dreams, travel, hobbies, events, and other thoughts. A similar

question then asked about the prevalence of any "negative or unpleasant" intrusive thoughts.

Page 5: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

5

Respondents were then asked about their life satisfaction and day satisfaction on a seven point

scale. We include day satisfaction since it is an evaluation of yesterday, which compares nicely

with the experiences of yesterday elicited using the DRM. It also provides a nice comparison

with life satisfaction since life satisfaction has no temporal restriction whereas day satisfaction

does. We use pooled regressions, random effects regressions and fixed effect regressions for the

analysis, but the latter may not be well defined because of only using the two time points. For all

analyses, all measures are on a zero to one scale. We use 5% significance levels throughout.

3. Results

Table 1 shows the summary statistics of the panel sample. It is clear that there are some

similarities across the two time periods. Over half of the sample does not have children; the

average household income is €42,392 in 2007 and €44,430 in 2008. The average level of

intrusive thoughts is similar across both time periods. Life satisfaction decreases from 2007 to

2008 but day satisfaction increases over the same time period – these results are not significant at

the five per cent level though. DRM Affect decreases over the same period and this is significant

at the five per cent level.

For the regressions that follow, all coefficients have been changed into a 0-1 scale to ease

comparability across the measures. The DRM U-index has been reverse coded to be consistent

with the other coefficients, so those coefficients that are positive suggest a lower likelihood of

being unhappy. Table 2 examines the initial relationship between income and the happiness

measures using pooled and random effects models. This is important to examine as we have the

first panel DRM dataset to examine the impact of income on experienced utility. For life

satisfaction, income is positive and significant for the pooled regressions, but not significant for

the random effects regressions, although the coefficient is positive. For day satisfaction, income

is positive and significant for both models. DRM affect has a positive relationship with income

but is not significant and DRM U-index is positive and significant for the pooled model. So

richer people tend to be more satisfied with their lives and days, and that richer people tend to

Page 6: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

6

have better experiences during the days. It is interesting that income is important to day

satisfaction, and more important than in the life satisfaction – income relationship.

We firstly go through the basic associations of the financial thoughts for the sample across both

time periods. Bar graph 1a illustrates the IFTs by income in 2007. Negative IFTs (light grey

(blue) bars) are more frequent than positive thoughts (dark grey (red) bars), and both are more

frequent for those with lower incomes. So for those who have annual incomes of €70,000 or less,

around 12-15% of this sample have some negative IFTs. Bar graph 1b illustrates the IFTs by

income in 2008 and shows that those who had very low incomes or very high incomes were

more likely to have negative IFTs. For those on annual incomes of €10,000 and €130,000, 20%

of this sample has negative IFTs, and around 7.5% of people have negative IFTs for incomes

between €30,000 and €90,000. As before, positive IFTs are more frequent for those with low

incomes.

Bar graph 2A shows the measures of happiness by income in 2007. There is no clear relationship

between happiness and income. For life satisfaction and day satisfaction, those on €90,000 have

the highest wellbeing. For DRM Affect, those on €110,000 have the highest wellbeing.

Interestingly, those on the highest income level (€150,000 and above), 30% have unhappy days

(in that the negative effect is at least the same as positive affect).

Bar graph 2B shows the measures of happiness by income in 2008. The satisfaction measures

increase slightly with income. For life satisfaction, those earning €110,000 have the highest life

satisfaction and highest day satisfaction. Those who have the highest DRM Affect are those who

are earning the highest incomes.

Tables 3a-3c examine the relationship between the measures of happiness and the negative and

positive intrusive financial thoughts. Table 3a examines the basic correlation between the

measures of happiness and IFTs, using pooled OLS. It is clear that negative IFTs are bad for

every measure of happiness apart from the U-index (where a negative coefficient means that the

person is more likely to be unhappy). The largest negative correlation is between IFTs and life

satisfaction. That is, having negative IFTs is associated with 17.4% lower life satisfaction. The

Page 7: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

7

positive IFTs associations are also interesting, since positive thoughts about finances are not

significantly positively associated with greater life satisfaction, day satisfaction, DRM Affect

and DRM U-index. This is consistent with recent research in this area (see Killingsworth &

Gilbert, 2010). Table 3b uses the same specification but the econometric structure is using the

panel nature of the data using random effects. There are similar results here from the pooled

regressions but the coefficients are not so large.

Table 3c uses the same specification but uses a fixed effects econometric structure. The

association between negative IFTs and life satisfaction is also reduced but it is still large: a 12%

reduction of life satisfaction if one thinks negatively about their finances. The sign of the other

measures of happiness change (from negative to positive) or are reduced through using the fixed

effects structure. Positive IFTs are positively associated with life satisfaction and day

satisfaction, but negatively correlated with the experience measures, i.e. DRM Affect and the U-

index, although the correlation is very small for the U-index.

Tables 4a – 4c examine the same relationship but include household income as a control variable

(in the same way as Table 2). The reason why income is included here is because we want to

determine the magnitude of IFTs when we control for the level of income of the individuals. It is

clear from these regressions that the sign and magnitude of the IFTs coefficients do not change.

So controlling for the level of income does not change the average magnitude of the impact of

IFTs on happiness. What is interesting here is that household income coefficients do not change

from those stated in table 2, in that income comes into the panel regressions as positive for

experiences. For life satisfaction, however, the coefficient on income is lower when negative

IFTs are included in the model (compare 0.042 in Table 2 with 0.035 in regression 1 in table 4a:

difference 0.007, p>0.05). This result supports the work of Dolan (2010) who finds that

controlling for thoughts about health reduces the impact of health states on health preferences.

Nonetheless, the coefficient on negative IFTs for life satisfaction is still overall large across all

models (12-17%), suggesting the importance of IFTs for life satisfaction.

Tables 5a – 5c include an interaction term between the IFTs and household income. We will

focus here on the interaction terms since they are demonstrating the interaction between income

Page 8: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

8

and the IFT. This will enable us to determine whether those who earn more income suffer more

from financial thoughts than those on lower incomes. So for negative IFTs, if the interaction

term (IFT x income) is negative, it means that the richer individuals are more affected by the

negative IFTs than poorer individuals, and conversely if the interaction term is positive, then

poorer individuals are more negatively affected than richer individuals. It seems that for life

satisfaction, the interaction term is positive and significant. So, poorer individuals who have

negative IFTs are more negatively impacted in life satisfaction assessments than those richer

individuals who have negative IFTs. For day satisfaction, we find the opposite associations. The

interaction term for negative IFTs is negative and significant. This suggests that richer

individuals who have negative IFTs are more negatively impacted than those poorer individuals

who have negative IFTs. For the DRM measures, the interaction terms are not significant.

Table 5c examines the same specification as Table 5b, but includes a fixed effect. It is clear that

the life satisfaction result (that less well off people are more negatively impacted by negative

IFTs) diminishes, but the negative coefficient on the interaction term for day satisfaction

increases in magnitude and is significant. From regressions (3) and (5), it is clear that negative

IFTs are worse for richer individuals than poorer individuals. For positive IFTs, they seem to

generally have a higher impact on richer individuals than poorer individuals, especially for day

satisfaction (see regression 3). For the U-index, positive IFTs are worse for richer individuals

than poorer individuals (see regression 8).

Tables 6a – 6c includes time-varying background variables of the respondent such as marital

status and whether they have children, and includes other intrusive thoughts – these are intrusive

thoughts about health, family, work, friends, and dreams. Therefore this helps us to pick up any

missing associations between financial thoughts and background variables and thoughts about

another domain. Life satisfaction and day satisfaction are still negatively impacted by intrusive

IFTs, and more so for richer individuals. So once we control for other negative thoughts about

other domains and time-varying background variables, financial worries become much more

prominent, especially amongst those who have money. The results for DRM Affect and DRM U-

index are similar although not significant.

Page 9: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

9

4. Discussion

Previous research in the economics of happiness has found a relatively robust, though moderate,

positive relationship between household income and life satisfaction (Clark et al, 2008;

Stevenson & Wolfers, 2008). Kahneman et al (2006), however, claimed that this relationship was

due to a focussing illusion i.e. the false belief that more money should be related to greater

happiness rather than being a reflection of the actual experiences of richer and poorer

individuals. Evidence supporting their claims came from a diary study of the previous day which

found that while income had a positive relationship to life satisfaction and people's estimates of

the amount of time spent in a good mood, there was no relationship with the actual amount of

happiness experienced.

The current research, using the same basic method with a different sample, has replicated these

findings. We suggest, however, that the diary method focuses the respondent's attention on

activities neglecting any intrusive thoughts they may have had. Since such thoughts can take up a

considerable amount of our time and are known to affect our experienced utility (Kane et al,

2007; Smallwood & Schooler, 2006; Watkins, 2008), this appears to have been an important

omission. Moreover, given that finances are also known to be one of the chief sources of daily

worries (MacGregor, 1991), it seemed plausible that experienced utility may be related to the

presence of intrusive financial thoughts (IFTs). In particular, we suspected that negative thoughts

about money would be related to lower levels of experienced utility across an average day.

Consistent with these predictions, negative IFTs were more prevalent among poorer respondents.

That the prevalence of these other thoughts was also negatively related to experienced utility

further supports our contention that experienced utility is composed of the thoughts that occupy

us as well as the activities we engage in. We report that having negative IFTs is associated with

around a 15% reduction in life satisfaction. That association is larger than most of the current

correlates of life satisfaction in the literature. The results suggest that while poorer individuals

report more negative IFTs, their impact on various SWB measures is larger for those individuals

who are richer. This result might go some way to explain the hedonic treadmill: If people who

have higher income are those who worry more about their income, then this negatively impacts

Page 10: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

10

on their SWB. Our method only focussed on frequency of thought and not on intensity. So it

might be that those who are richer have more intense negative thoughts. Or wealthy people might

be more affected by IFTs simply because they have fewer of them: if a person is not used to

worrying about money, each episode of worry may be less familiar and more alarming.

We also find that including negative IFTs in the life satisfaction regression reduces the income

coefficient. This is extremely important to the reconciliation between the results of life

satisfaction and DRM Affect. We propose that negative IFTs act as a potentially omitted

variable. Clearly, the sign of the omitted variables bias is the product of the sign of the

relationship between IFTs and life satisfaction, and IFTs and income. We find that negative IFTs

reduce life satisfaction (negative sign), and that we find a higher impact of IFTs for those on

higher incomes (positive sign). This implies that the sign of the omitted variables bias is

negative, i.e. controlling for ITFs should reduce the effect of household income on life

satisfaction – which is what we find. These IFTs might be very important for the reconciliation

of the results of income for life satisfaction and DRM Affect. Therefore, the inclusion of IFTs

into the model brings the larger income coefficient in the life satisfaction regressions closer to

the smaller income coefficient in the DRM Affect regressions. It is possible that IFTs are the

‘missing link’ in the relationship between happiness and income.

The upshot of this result is that people who worry or have more negative thoughts about money

believe that income is important to their life satisfaction. Moreover, people who do not

negatively worry about money believe that income is not that important to their life satisfaction.

This is consistent with the work of Dolan (2010) who found that the inclusion of health thoughts

in health preference regressions reduces the effect of health states on overall preferences. This is

also consistent with the work of Dolan and Metcalfe (2011), who found that the domains of life

that people think matter most (e.g. finances, health, work, etc) actually have lower satisfaction in

those domains. So our results here are consistent with this work in that those who think more

about negatively about finances (analogous to less satisfied with this domain) believe that

income matters more to their life satisfaction.

Page 11: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

11

The IFT results are consistent with related research in the field. Borooah (2005), for instance,

reports that needing to borrow money mid-week increases the chances of being "unhappy", and

Brown et al (2005) find that symptoms such as depression and anxiety are more widespread

among people with unsecured debts. In other words, day-to-day emotional experiences do seem

to be related to financial problems. The current research extends these findings by showing that

these problems can intrude on people's thoughts during their normal activities and can thus have

an effect on their utility. Even when we control for a host of other intrusive thoughts, IFTs are

still highly associated with life satisfaction and day satisfaction.

Related to this is the finding that positive thoughts about money are not necessarily positive for

people’s experienced utility. This is consistent with Killingsworth and Gilbert’s (2010) research

that showed that any mind wanderings have a negative impact on experiences. So, when

individuals think about money (both negatively and positively), this removes people from the

experience of the episode, and removes their ‘flow’ (Csíkszentmihályi, 1990). When individuals

are not fully engaged in the activity, their attention is directed to thoughts that might have little to

do with the activity. How such thoughts arise in some activities, and how interventions,

workplaces, public policies etc. can be constructed to direct attention to the activity as opposed

to mind wanderings, all require further investigation.

The intrusive thoughts could be involuntary (to some great degree outside of our control) as well

as voluntary (things that we choose to think about). We know very little about the relative impact

of different attentional types on intrusive thoughts. There is an increasing amount of work on

attention shaping choices, but mostly in relation to voluntary attention (see Chetty et al, 2009;

DellaVigna & Pollet, 2009; Hirschliefer, 2009). We do not suggest that our approach is the best

to elicit thoughts or understand attention, but future DRM type studies could certainly ask

respondents about thoughts and feelings before asking about them what they are doing and who

they are with. It may be that asking people about their main activity before asking them about

their mood draws their attention away from what they were thinking about and such a study

would allow us to say something about the importance of activity-related and general mood-

related focussing effects or attentional types (Dolan, 2010).

Page 12: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

12

The findings are also consistent with the notion that the relationship between income and

happiness is in part due to income's role as a "buffer", attenuating the impact of negative events.

Smith et al (2005), for instance, report that declines in happiness following disability were

smaller among wealthier individuals and argue that "money may not buy happiness, but it does

seem to buy people out of some of the misery associated with a decline in health status" (p.665).

Similarly here, income per se does not seem to directly affect experienced utility, but wealthier

people are less likely to have IFTs which are associated with more time spent in negative moods.

To paraphrase Smith et al (2005), money may not buy experienced utility, but it does seem to

buy people out of some of the misery associated with financial concerns. From our research, we

would like to change this by arguing that money may buy people out of financial worries, but if

people with high incomes have financial worries, then this impact is larger than the worries of

people who do not have money.

Our results do not challenge the focusing illusion as a general mechanism for explaining much of

the observed relationship between various demographic variables and measures of life

satisfaction. A substantial body of evidence has been gathered to suggest that life satisfaction

ratings are influenced by what is salient at the time of responding and intuitive beliefs about what

is "probably important" (Schwarz, 2007). Moreover, our own study may have introduced

focussing effects of its own. We are simply cautioning against concluding that income does not

show up in experienced utility when existing measures of experienced utility like the DRM focus

attention on activities and ignore the intrusive thoughts that affect us. Moreover, whilst fixed

effects rules out any omitted variables and individual heterogeneity in response, but we cannot

rule out for certain that people who are not “happy” go onto to have negative IFTs.

Importantly, such thoughts may also predict certain behaviours and so economists less interested

in happiness and more interested in behaviour might also give due consideration to intrusive

thoughts. People with chronically high levels of such thoughts, such as veterans and emergency

crew members with Post Traumatic Stress Disorder are, for instance, less likely to be engaged in

the labour market and tend to earn less if they are (e.g. Savoca & Rosenheck, 2000). Research is

now needed to explore the impact of more everyday intrusive thoughts on economically

important behaviours. For instance, our panel straddled the financial crisis, and we found that

Page 13: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

13

day satisfaction and DRM Affect were lower in 2008 as opposed to 2007 and, to lend some face

validity to our findings, that negative IFTs were higher for those at either end of the income

distribution in 2008 as opposed to 2007.

In any event, an individual's day is not simply made up of the activities they engage in but also

the thoughts that distract them. In considering the relationship between income and happiness –

and indeed between any set of determinants and happiness - future research should account for

the important role of intrusive thoughts in these associations. These thoughts can have a

powerful influence on people's moods and the present research suggests that income can affect

utility if we consider "intrusive financial thoughts" alongside "activity-focussed feelings".

Page 14: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

14

References Blanchflower, D. & Oswald, A. (2004). Well-being over time in Britain and the USA. Journal of Public

Economics, 2004, 88, 1359-1386. Borooah, V. K. (2005). How to assess happiness? A tale of three measures. Applied Economic Letters, 12,

191-94. Brown, S., Taylor K.B. & Wheatley Price, S. (2005). Debt and distress: Evaluating the psychological cost

of credit. Journal of Economic Psychology, 26, 642-66. Chetty, R., Looney, A. & Kroft, K. (2009), Salience and taxation: theory and evidence. American

Economic Review, 99, 1145-1177. Clark, A., Frijters, P. & Shields, M.A. (2008). Relative income, happiness and Utility: An explanation for

the Easterlin Paradox and other puzzles. Journal of Economic Literature, 46, 95-144. Csíkszentmihályi, M. (1990). Flow: The Psychology of Optimal Experience. Harper & Row: New York. DellaVigna, S. & Pollet, J. (2009). Investor attention and Friday earnings announcements. Journal of

Finance, 64, 709-749. Dolan, P. (2010). Thinking about it: thoughts about health and valuing QALYs. Health Economics. Dolan, P. & Kahneman, D. (2008). Interpretations of utility and their implications for the valuation of

health. Economic Journal, 118, 215–234. Dolan, P. & Metcalfe, R. (2011). Comparing measures of subjective wellbeing and views about the role

they should play in policy. Report for the UK Office for National Statistics, June 2011. Dolan, P., Peasgood, T. & White, M.P. (2008). Do we really know what makes us happy? A review of the

economic literature on the factors associated with subjective well-being. Journal of Economic Psychology, 29, 94–122.

Easterlin, R.A. (2001). Income and Happiness: Towards a Unified Theory. Economic Journal, 111, 465-

484. Frank, R. (1999). Luxury Fever: Money and Happiness in an Era of Excess. New York: Free Press. Frey, B. and Stutzer, A. (2002). Happiness and Economics: How the Economy and Institutions Affect

Well-Being. Princeton: Princeton University Press. Hirshleifer, D.A. (2009). The economics of attention: style and substance in the age of information.

Journal of Bioeconomics, 11, 99-102. Kahneman, D. (2000). Evaluation by moments: Past and future. In D. Kahneman & A. Tversky (Eds.)

Choices, Values and Frames (pp.693-708). Cambridge: Cambridge University Press. Kahneman, D., Krueger, A.B., Schkade, D.A., Schwarz, N. & Stone, A.A. (2004). A survey method for

characterising daily life experience: The day reconstruction method (DRM). Science, 306, 1776–1780.

Page 15: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

15

Kahneman, D., Krueger, A.B., Schkade, D.A., Schwarz, N. & Stone, A.A. (2006). Would you really be

happier if you were richer? A focusing illusion. Science, 312, 1908-1910. Kahneman, D., Wakker, P.P. & Sarin, R. (1997). Back to Bentham? Explorations of experienced utility.

Quarterly Journal of Economics, 112, 375-406. Killingsworth, M.A. & Gilbert, D.T. (2010). A wandering mind is an unhappy mind. Science, 330, 932. Knabe, A., Ratzel, S., Schob, R. & Weimann, J. (2010). Dissatisfied with Life but Having a Good Day:

Time-use and Well-being of the Unemployed. Economic Journal, 120, 867-889. Layard, R. (2006). Happiness and public policy: A challenge to the profession. Economic Journal, 116,

C24–C33. MacGregor, D. (1991). Worry over technological activities and life concerns. Risk Analysis, 11, 315-324. Pischke, J.-S. (2011). Money and happiness: Evidence from the industry wage structure. NBER Working

Paper No. 17056. Savoca, E. & Rosenheck, R. (2000). The Civilian Labor Market Experiences of Vietnam-Era Veterans:

The Influence of Psychiatric Disorders. Journal of Mental Health Policy and Economics, 3, 199-207. Schkade, D.A. & Kahneman, D. (1998). Does living in California make people happy? A focusing

illusion in judgments of life satisfaction. Psychological Science, 9, 340-346. Schwarz, N. (2007). Retrospective and concurrent self-reports: The rationale for real-time data capture. In

A. A. Stone, S. S. Shiffman, A. Atienza, & L. Nebeling (Eds.), The science of real-time data capture: Self-reports in health research (pp. 11-26). New York: Oxford University Press.

Smallwood, J. & Schooler, J.W. (2006). The restless mind. Psychological Bulletin, 132, 946-958. Smith, D. M., Langa, K.M., Kabeto, M.U. & Ubel, P.A (2005). Health, wealth, and happiness: Financial

resources buffer subjective well-being after the onset of a disability. Psychological Science, 16, 663-664.

Stevenson, B. & Wolfers, J. (2009). Economic growth and subjective well-being: Reassessing the

Easterlin Paradox. Brookings Papers on Economic Activity. Watkins, E. (2008). Constructive and unconstructive repetitive thought. Psychological Bulletin. 134, 163-

206. White, M. & Dolan, P. (2009). Accounting for the richness of our daily activities. Psychological Science,

20, 1000–1008. Wilson, T.D. & Gilbert, D. (2003). Affective forecasting. Advances in Experimental Social Psychology,

35, 345-411.

Page 16: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

16

Table 1: Summary statistics of the sample at t1 and t2 – this will include age, gender, income, IFTs (both positive and negative), SWB.   2007 2008 No children 57% 54% Cohabiting 30% 26% Married 40% 40% Divorced 1% 1% Separated 7% 7% Widowed 3% 4% Household income €42,392 €44,430 Negative intrusive financial thoughts 11% 10% Positive intrusive financial thoughts 3% 4% Life satisfaction 7.0 7.3 Day satisfaction 7.9 7.7 DRM Affect 0.52 0.45* DRM U-index 0.09 0.10

Page 17: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

17

Bar graph 1a: IFTs (positive (red) and negative (blue)) by income in 2007

Bar graph 1b: IFTs (positive (red) and negative (blue)) by income in 2008

Page 18: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

18

Bar graph 2A: SWB by income in 2007 (a) Life satisfaction (b) Day satisfaction (c) DRM Affect (d) DRM U-index

Page 19: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

19

Bar graph 2B: SWB by income in 2008 (a) Life satisfaction (b) Day satisfaction (c) DRM Affect (d) DRM U-index

Page 20: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

20

Table 2: SWB = f(Y) (1) (2) (3) (4) Pooled Pooled Pooled Pooled Life satisfaction Day satisfaction DRM Affect DRM U-index Log (HH income) 0.042* 0.061* 0.027 0.046* [0.017] [0.145] [0.020] [0.020] (5) (6) (7) (8) Random effects Random effects Random effects Random effects Life satisfaction Day satisfaction DRM Affect DRM U-index Log (HH income) 0.031 0.060* 0.023 0.032 [0.019] [0.160] [0.021] [0.022] Notes: each coefficient represents a separate regression. * represents significance at the five per cent respectively.

Page 21: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

21

Table 3a: SWB = f(IFTS): Pooled regressions (1) (2) (3) (4) (5) (6) (7) (8) Life satisfaction Life

satisfaction Day

satisfaction Day

satisfaction DRM Affect DRM Affect DRM U-index

DRM U-index

Negative IFTs -0.174* -0.172 -0.039 -0.013 [0.044] [0.373] [0.050] [0.052] Positive IFTs 0.053 -0.633 -0.108 -0.004 [0.078] [0.680] [0.086] [0.090]

* represents significance at the five per cent respectively. Table 3b: SWB = f(IFTS): Random effects regressions

(1) (2) (3) (4) (5) (6) (7) (8) Life satisfaction Life

satisfaction Day

satisfaction Day

satisfaction DRM Affect DRM Affect DRM U-index

DRM U-index

Negative IFTs -0.148* 0.009 -0.026 -0.031 [0.039] [0.035] [0.045] [0.047] Positive IFTs 0.083 -0.049 -0.110 -0.002 [0.073] [0.068] [0.082] [0.086]

* represents significance at the five per cent respectively. Table 3c: SWB = f(IFTS): Fixed effects regression

(1) (2) (3) (4) (5) (6) (7) (8) Life satisfaction Life

satisfaction Day

satisfaction Day

satisfaction DRM Affect DRM Affect DRM U-index

DRM U-index

Negative IFTs -0.116** 0.042 -0.005 -0.059 [0.046] [0.044] [0.054] [0.056] Positive IFTs 0.124 0.039 -0.113 -0.004 [0.090] [0.111] [0.104] [0.090]

* represents significance at the five per cent respectively.

Page 22: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

22

Table 4a: SWB = f(IFTS, income): Pooled regressions

(1) (2) (3) (4) (5) (6) (7) (8) Life satisfaction Life

satisfaction Day

satisfaction Day

satisfaction DRM Affect DRM Affect DRM U-index

DRM U-index

Negative IFTs -0.168* 0.003 -0.038 0.026 [0.044] [0.037] [0.050] [0.052] Positive IFTs 0.066 -0.036 -0.105 0.021 [0.078] [0.067] [0.087] [0.091] Log(HHincome) 0.035* 0.043* 0.061* 0.061* 0.026 0.026 0.047* 0.046* [0.017] [0.017] [0.015] [0.015] [0.020] [0.020] [0.020] [0.020]

* represents significance at the five per cent respectively. Table 4b: SWB = f(IFTS, income): Random effects regressions

(1) (2) (3) (4) (5) (6) (7) (8) Life satisfaction Life

satisfaction Day

satisfaction Day

satisfaction DRM Affect DRM Affect DRM U-index

DRM U-index

Negative IFTs -0.143* 0.024 -0.026 0.035 [0.039] [0.033] [0.046] [0.048] Positive IFTs 0.082 -0.031 -0.118 0.014 [0.073] [0.067] [0.084] [0.087] Log(HHincome) 0.025 0.033 0.061* 0.060* 0.021 0.021 0.034 0.033 [0.019] [0.019] [0.016] [0.016] [0.021] [0.021] [0.022] [0.022]

* represents significance at the five per cent respectively. Table 4c: SWB = f(IFTS, income): Fixed effects regressions

(1) (2) (3) (4) (5) (6) (7) (8) Life satisfaction Life

satisfaction Day

satisfaction Day

satisfaction DRM Affect DRM Affect DRM U-index

DRM U-index

Negative IFTs -0.118* 0.048 -0.006 0.053 [0.045] [0.041] [0.055] [0.057] Positive IFTs 0.145 0.021 -0.138 -0.001 [0.094] [0.107] [0.113] [0.118] Log(HHincome) -0.020 -0.014 0.049 0.045 -0.011 0.010 -0.071 -0.073 [0.036] [0.036] [0.033] [0.033] [0.045] [0.044] [0.046] [0.046]

* represents significance at the five per cent respectively.

Page 23: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

23

Table 5a: SWB = f(IFTS, income, IFTS x income): Pooled regressions

(1) (2) (3) (4) (5) (6) (7) (8) Life satisfaction Life

satisfaction Day

satisfaction Day

satisfaction DRM Affect DRM Affect DRM U-index

DRM U-index

Negative IFTs -1.575* 0.973* 0.404 -0.019 [0.537] [0.438] [0.620] [0.643] Positive IFTs -0.841 -0.572 -0.214 1.590 [0.040] [0.786] [1.091] [1.129] Log(HHincome) 0.019 0.040* 0.074* 0.059* 0.031 0.025 0.046* 0.051* [0.018] [0.018] [0.016] [0.019] [0.021] [0.020] [0.022] [0.021) IFT*Log(HHIncome) 0.138* 0.090 -0.096* 0.054 -0.043 0.011 0.004 -0.156 [0.052] [0.096] [0.043] [0.784] [0.061] [0.108] [0.063] [0.112]

* represents significance at the five per cent respectively. Table 5b: SWB = f(IFTS, income, IFTS x income): Random effects regressions

(1) (2) (3) (4) (5) (6) (7) (8) Life satisfaction Life

satisfaction Day

satisfaction Day

satisfaction DRM Affect DRM Affect DRM U-index

DRM U-index

Negative IFTs -0.965* 1.236* 0.609 0.218 [0.483] [0.387] [0.571] [0.597] Positive IFTs -0.336 -1.170 0.022 1.972 [0.926] [0.794] [1.057] [1.098] Log(HHincome) 0.018 0.031 0.074* 0.055* 0.029 0.021 0.035 0.039* [0.019] [0.019] [0.017] [0.016] [0.022] [0.021] [0.023] [0.022) IFT*Log(HHIncome) 0.080 0.042 -0.119* 0.115 -0.062 -0.014 -0.018 -0.195 [0.047] [0.092] [0.038] [0.080] [0.056] [0.105] [0.058] [0.109]

* represents significance at the five per cent respectively. Table 5c: SWB = f(IFTS, income, IFTS x income): Fixed effects regressions

(1) (2) (3) (4) (5) (6) (7) (8) Life satisfaction Life

satisfaction Day

satisfaction Day

satisfaction DRM Affect DRM Affect DRM U-index

DRM U-index

Negative IFTs -0.080 -3.202* 1.052 0.884 [0.563] [1.430] [0.684] [0.710] Positive IFTs 0.932 -1.170 0.842 3.133* [1.235] [0.794] [1.485] [1.529] Log(HHincome) -0.020 -0.010 0.028 0.055* -0.006 -0.006 -0.067 -0.061 [0.036] [0.037] [0.033] [0.016] [0.045] [0.045] [0.046] [0.046) IFT*Log(HHIncome) -0.004 -0.079 0.336* 0.115 -0.103 -0.010 -0.081 -0.315* [0.055] [0.124] [0.149] [0.080] [0.067] [0.149] [0.069] [0.153]

* represents significance at the five per cent respectively.

Page 24: With my money on my mind: income, happiness and intrusive … · and positive (Stevenson & Wolfers, 2008; Kahneman & Deaton, 2010; Pischke, 2011). An alternative approach to measuring

24

Table 6a: SWB = f(IFTS, income, IFTS*income, other background variables, other intrusive thoughts): Pooled regressions

(1) (2) (3) (4) (5) (6) (7) (8) Life satisfaction Life

satisfaction Day

satisfaction Day

satisfaction DRM Affect DRM Affect DRM U-index

DRM U-index

Negative IFTs -1.428* 1.117* 0.589 0.172 [0.539] [0.446] [0.622] [0.659] Positive IFTs -0.806 -0.239 0.213 1.471 [0.968] [0.807] [1.100] [1.145] Log(HHincome) -0.003 0.007 0.056* 0.049* 0.007 -0.002 0.037 0.032 [0.020] [0.020] [0.017] [0.017] [0.023] [0.022] [0.024] [0.023] IFT*Log(HHIncome) 0.123* 0.087 -0.107* 0.018 -0.056 -0.033 -0.011 -0.145 [0.053] [0.096] [0.044] [0.081] [0.061] [0.109] [0.064] [0.113]

* represents significance at the five per cent respectively. Table 6b: SWB = f(IFTS, income, IFTS*income, other background variables, other intrusive thoughts): Random effects regressions

(1) (2) (3) (4) (5) (6) (7) (8) Life satisfaction Life

satisfaction Day

satisfaction Day

satisfaction DRM Affect DRM Affect DRM U-index

DRM U-index

Negative IFTs -0.907 1.393* 0.826 0.994 [0.497] [0.401] [0.582] [0.758] Positive IFTs -0.497 -0.783 0.261 0.425 [0.929] [0.821] [1.071] [0.612] Log(HHincome) -0.004 0.004 0.066* 0.049* 0.010 -0.002 -0.062 0.027 [0.021] [0.021] [0.018] [0.018] [0.024] [0.023] [0.048] [0.025] IFT*Log(HHIncome) 0.074 0.058 -0.133* 0.748 -0.079 -0.039 -0.090 -0.035 [0.048] [0.127] [0.039] [0.083] [0.057] [0.106] [0.074] [0.060]

* represents significance at the five per cent respectively. Table 6c: SWB = f(IFTS, income, IFTS*income, other background vars, other intrusive thoughts): Fixed effects regressions

(1) (2) (3) (4) (5) (6) (7) (8) Life satisfaction Life

satisfaction Day

satisfaction Day

satisfaction DRM Affect DRM Affect DRM U-index

DRM U-index

Negative IFTs -0.042 1.847* 1.251 0.994 [0.595] [0.503] [0.719] [0.758] Positive IFTs 0.811 -3.016 0.517 3.016 [1.272] [1.608] [1.525] [1.599] Log(HHincome) -0.032 -0.026 0.051 0.024 0.013 0.009 -0.062 0.063 [0.038] [0.038] [0.034] [0.036] [0.046] [0.046] [0.048] [0.048] IFT*Log(HHIncome) -0.010 -0.070 -0.178* 0.318 -0.121 -0.066 -0.090 -0.303 [0.058] [0.127] [0.050] [0.167] [0.070] [0.153] [0.074] [0.160]

* represents significance at the five per cent respectively.