International Review of Applied Economics: Being Derived From

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

This article was downloaded by: [The University Of Melbourne Libraries]

On: 10 October 2014, At: 06:48


Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

International Review of Applied


Economics
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/cira20

Does sport make you happy? An


analysis of the well‐being derived from
sports participation
a b
Paul Downward & Simona Rasciute
a
Institute of Sport and Leisure Policy, School of Sport and
Exercise Sciences , Loughborough University , Leicestershire, LE11
3TU, UK
b
Department of Economics , Loughborough University ,
Leicestershire, LE11 3TU, UK
Published online: 09 May 2011.

To cite this article: Paul Downward & Simona Rasciute (2011) Does sport make you happy? An
analysis of the well‐being derived from sports participation, International Review of Applied
Economics, 25:3, 331-348, DOI: 10.1080/02692171.2010.511168

To link to this article: http://dx.doi.org/10.1080/02692171.2010.511168

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or
arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014
International Review of Applied Economics
Vol. 25, No. 3, May 2011, 331–348

Does sport make you happy? An analysis of the well-being derived


from sports participation
Paul Downwarda,* and Simona Rasciuteb
a
Institute of Sport and Leisure Policy, School of Sport and Exercise Sciences, Loughborough
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

University, Leicestershire LE11 3TU, UK; bDepartment of Economics, Loughborough


University, Leicestershire LE11 3TU, UK
International
10.1080/02692171.2010.511168
CIRA_A_511168.sgm
0269-2171
Original
Taylor
02010
00
Dr
p.downward@lboro.ac.uk
000002010
PaulDownward
and
& Article
Francis
(print)/1465-3486
Francis
Review of Applied
(online)
Economics

This paper examines the impact of sports participation upon the subjective well-
being of individuals. Encouraging participation in sports activity is now an
important public policy issue, as it is argued that there are benefits in terms of
health and well-being to individuals as well as to society through externalities.
Controlling for personal and socio-demographic characteristics affecting well-
being, this paper examines if participation in, and the frequency and duration of,
67 sports activities affects well-being. The form in which sports participation takes
place is also investigated by examining if social-interaction sports produce more
well-being. This paper demonstrates that sports participation has a positive affect
upon the subjective well-being of the population and, moreover, estimates its
monetary value. The effects are larger if one allows for social interactions.

Keywords: ordered choice model; heterogeneous thresholds; happiness; sports;


willingness to pay measures

JEL Classifications: C25, I31, D31, J12

1. Introduction
Encouraging mass participation in sport is now an important public policy issue. Claims
are made that sport, as a form of physical activity, can help to reduce the physiological
health costs to the individual emanating from obesity, cardiovascular diseases, high
blood pressure, hypertension, diabetes, strokes, heart disease, osteoarthritis and
osteoporosis. Moreover, participation in sport can also help to promote psychological
benefits by reducing anxiety and depression (Department of Health 2004; WHO World
Health Day 2002; DCMS/Strategy Unit 2002; Scully et al. 1999).
In addition to these individual costs and benefits, policymakers have argued that
sports participation will have positive externalities involving a reduction in the health-
care costs to society. These were estimated at £1,890 billion in 2002 for the UK
(DCMS/Strategy Unit 2002). Further, claims are made that that sports participation
will generate other positive externalities by promoting social inclusion and reducing
crime (DCMS/Strategy Unit 2002; Collins and Kay 2002; Nicholls 2007; Coalter
2007; Downward, Dawson, and Dejonghe 2009). It is by no means clear that any of

*Corresponding author. Email: p.downward@lboro.ac.uk

ISSN 0269-2171 print/ISSN 1465-3486 online


© 2011 Taylor & Francis
DOI: 10.1080/02692171.2010.511168
http://www.informaworld.com
332 P. Downward and S. Rasciute

these latter impacts are guaranteed. Coalter (2007) argues that there is a lack of strong
cumulative evidence to support such claims about the effects of sports participation.1
Despite these policy targets and discussions, there have been few studies using
large-scale data analysis to examine whether or not sport has a positive impact on the
overall well-being of individuals. The literature also tends only to focus on aggregate
sports participation and does not allow for the frequency and intensity of participation.
This paper seeks to rectify this gap in the literature.
The impact of life events upon subjective well-being has developed into a substan-
tial literature in economics (Clark et al. 2008). This paper addresses the impact of the
choice to participate in sport in England, as a ‘life event’, upon a measure of subjective
well-being defined as happiness. The paper proceeds as follows. In Section 2 the
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

economic literature addressing subjective well-being is briefly reviewed to highlight


the gap in research addressed. Clarification about the definition of subjective well-
being employed in the research is offered. Section 3 presents details of the data
employed, while Section 4 outlines the econometric model estimated. Section 5
presents and discusses the results. Conclusions follow in Section 6.

2. The well-being literature


There is now a detailed body of evidence on the determinants of subjective well-being,
as stated directly by research respondents, in the economics literature. This form of
research is more traditionally associated with psychology but there is a history of
seminal work by economists (Van Praag and Frijters 1999; Kahneman, Diener, and
Schwarz 1999). The motivation of the economics literature is to measure the utility of
individuals (Gardner and Oswald 2006; Kahneman and Krueger 2006; Shields and
Wheatley Price 2005).
In a detailed review of the literature, Dolan, Peasgood, and White (2008) note that
the approach taken by economists is typically distinct from the psychology literature
in basing its claims upon relatively large-scale secondary data sets. Dolan, Peasgood,
and White (2008) also identify the datasets employed in the research, as well as the
main determinants of well-being. The determinants include variables that measure
dimensions of respondent incomes, their personal characteristics, their socially devel-
oped characteristics, how they spend their time, their attitudes and beliefs about their
lives and about others, and, finally, their relationships to the wider economic, social
and political environment.2
What is particularly significant for the current paper is that their review notes that
whilst ‘a review of the broader literature on exercise and well-being has recently
appeared … little use has been made of large datasets and thus there seems to be an
important gap in the research here’ (Dolan, Peasgood, and White 2008, 104). Since
the publication of the review, a few studies have examined aspects of exercise based
on an economic perspective and large-scale data. Becchetti, Pelloni, and Rossetti
(2008) identify positive impacts of relational goods on well-being, of which participa-
tion in sport generally is identified as an example. Likewise Lechner (2009) examines
how an ordinally defined aggregate sports variable affects health and well-being.
Rasciute and Downward (2010) explore the impact of walking and cycling on well-
being as well as participation in forms of physical activity generally. What is lacking,
is a more detailed examination of the impact of sports as exercise. This is the objective
of this paper, in which a variety of forms of participation are investigated, including
their intensity. Further, an attempt is made to identify how different sports activities,
International Review of Applied Economics 333

particularly those that have a more relational character than others, affect well-being.
This is not possible in the Becchetti, Pelloni, and Rossetti (2008) study. Before
discussing how these issues are investigated, the remainder of this section briefly
discusses how the economics literature theorises and measures well-being, and the
conceptions that are employed in this research.
Despite the fact that the economics literature generally identifies well-being with
utility, a variety of specific conceptions of utility and its measurement appear in the
literature. From a theoretical perspective, this paper follows Frey’s (2008) argument
that ‘experienced’ rather than ‘decision’ utility is measured. This reflects a distinction
between the axiomatic approach to utility in textbook (neoclassical) microeconomics,
and an approach that recognises utility as a broader hedonistic concept that is not neces-
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

sarily grounded in objective (or eudaimonic) criteria and also captures altruistic motives
and social interactions rather than just individual self-interest. In the context of the
measurement of utility, the literature indicates that respondents either indicate the value
of their well-being on a single-item scale, or on multi-item measures as components
of well-being. Dolan, Peasgood, and White (2008) note that the two most common single-
item scales comprise statements either about the respondents’ happiness with life as
a whole, or their satisfaction with life as a whole. Some recent studies, which measure
well-being as happiness, include Blanchflower and Oswald (2008), Golden and Wiens-
Tuers (2006) and Shields, Wheatley Price, and Wooden (2009). In contrast, studies by
Winkelmann and Winkelmann (1998), Winkelmann (2005), Gardner and Oswald
(2006) and Frijters, Johnston, and Shields (2008) measure well-being as life satisfaction.
To measure well-being through an aggregation of multi-item components, studies
such as Brown, Taylor, and Wheatley Price (2005) and Gardner and Oswald (2006)
have made use of the General Health Questionnaire 12 Score (GHQ12), developed by
Goldberg (1972). To derive this score, respondents rank their experience of symptoms
relative to their usual experience on 12 items using a four-point scale.3 In the current
research, a single-item measure of happiness is employed.

3. Data
To investigate the impact of participation in sports upon individual well-being, data
from the first tranche of the Taking Part Survey from 2005, now lodged in the UK
Data Archive for public access, is analysed. This was a three-year repeated cross-
section survey commissioned by the Department for Culture Media and Sport and
conducted by the British Market Research Bureau. One individual from a randomly
sampled household was interviewed aged 16 years or older. The first tranche of data
comprised 28,117 respondents. Data on 67 sports activities were investigated identi-
fying participation or not over the last 12 months prior to the interview, in the last four
weeks prior to the interview, the frequency in times of participation in the last four
weeks, as well as the typical time in minutes of participation in these activities.4
Well-being in this data is investigated by the question ‘Taking all things together,
how happy would you say you are?’ Respondents then have to assign a value between
1 and 10 to this question with ‘1’ indicating extremely unhappy and ‘10’ extremely
happy. Table 1 presents the frequency of responses associated with this variable
for the 16,627 of cases for which there were eligible responses, which shows that well-
being answers are skewed towards the upper end of the distribution.5
The independent variables included in the analysis are based upon the available deter-
minants in the dataset that are also identified in the well-being literature.6 Table 2
334 P. Downward and S. Rasciute

Table 1. Frequencies of happiness.


Scale Value Frequency Percent
1 101 0.0061
2 103 0.0062
3 288 0.0173
4 359 0.0216
5 1292 0.0777
6 1232 0.0741
7 3082 0.1854
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

8 5002 0.3008
9 2870 0.1726
10 2298 0.1382
Total 16627 1.0000

Table 2. Variable descriptions and descriptive statistics.


Variable
Variable Frequency % Type Variable Description
single 5474 32.92 Nominal Single = 1, 0 = otherwise
married 8012 48.19 Nominal Married = 1, 0 = otherwise
separated 2283 13.73 Nominal Separated = 1, 0 = otherwise
widow 858 5.16 Nominal Widowed = 1, 0 = otherwise
white 14202 85.42 Nominal White = 1, 0 = otherwise
Asian 1240 7.46 Nominal Asian = 1, 0 = otherwise
black 780 4.69 Nominal Black = 1, 0 = otherwise
Otheth 405 2.44 Nominal Other ethnic origin = 1, 0 = otherwise
NorthEast 1523 9.16 Nominal North East = 1, 0 = otherwise
NorthWest 1697 10.21 Nominal North West = 1, 0 = otherwise
yorks 1689 10.16 Nominal Yorkshire = 1, 0 = otherwise
EastMid 1536 9.24 Nominal East Midlands = 1, 0 = otherwise
WestMid 1873 11.26 Nominal West Midlands = 1, 0 = otherwise
East 1833 11.02 Nominal East England = 1, 0 = otherwise
London 2061 12.40 Nominal London = 1, 0 = otherwise
SouthEast 2481 14.92 Nominal South East = 1, 0 = otherwise
SouthWest 1934 11.63 Nominal South West = 1, 0 = otherwise
working 11025 66.31 Nominal Working = 1, 0 = otherwise
student 530 3.19 Nominal Student = 1, 0 = otherwise
keephouse 1129 6.79 Nominal Keep house = 1, 0 = otherwise
retired 2503 15.05 Nominal Retired = 1, 0 = otherwise
illnotwork 509 3.06 Nominal Ill and can’t work = 1, 0 = otherwise
othwk 427 2.57 Nominal Other work = 1, 0 = otherwise
he 7002 42.11 Nominal Higher education or equivalent = 1, 0 = otherwise
alevel 3260 19.61 Nominal A Levels = 1, 0 = otherwise
apprentice 889 5.35 Nominal Apprentice = 1, 0 = otherwise
olevel5 3134 18.85 Nominal 5 GCSEs = 1, 0 = otherwise
otheredu 2342 14.09 Nominal Other education = 1 0 = otherwise
International Review of Applied Economics 335

Table 2. (Continued).
Variable
Variable Frequency % Type Variable Description
Gender 7752 46.62 Nominal Male = 1, 0 = female*
drinkdaily 1743 10.48 Nominal Drink alcohol every day = 1, 0 = otherwise
drink4to6 1748 10.51 Nominal Drink alcohol 4 to 6 days a week = 1, 0 = otherwise
drink1to3 5759 34.64 Nominal Drink alcohol 1 to 3 days a week = 1, 0 = otherwise
drinkless1 4511 27.13 Nominal Drink alcohol less than 1 day a week = 1, 0 =
otherwise
notdrink 2866 17.24 Nominal Don’t drink alcohol = 1, 0 = otherwise
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

smkdaily 3453 20.77 Nominal Smoke every day = 1, 0 = otherwise


smk4to6 98 0.59 Nominal Smoke 4 to 6 days a week = 1, 0 = otherwise
smk1to3 246 1.48 Nominal Smoke 1 to 3 days a week = 1, 0 = otherwise
smkless1 140 0.84 Nominal Smoke less than one day a week = 1, 0 = otherwise
notsmoke 12690 76.32 Nominal Don’t smoke = 1, 0 = otherwise
Ticknet 4578 27.53 Nominal Undertaken any voluntary work =1, o otherwise
spclose 15358 94.05 Nominal Sports facilities are close = 1, 0 = otherwise
anysport4 10248 61.63 Nominal Participated in any sport in the last 4 weeks =1, 0
= otherwise
anysport12 13044 78.45 Nominal Participated in any sport in the last 12 months =1,
0 = otherwise
Mean S.dev
numsport12 3.01 3.25 Cardinal Number of sports activities undertaken in the last
12 months
numsport4 1.48 1.86 Cardinal Number of sports activities undertaken in the last 4
weeks
numtimes4 7.56 12.82 Cardinal Number of times anysport participated in the last 4
weeks
totalmins4 610.607 1410 Cardinal Minutes of participation in any sport in the last 4
weeks
Age 43.57 16.35 Cardinal Age in years
GenHealth 4.10 0.85 Ordinal General level of health 5 = very good to 1 = very
bad
nadult 1.98 0.85 Cardinal Number of adults in the household
nchild 0.65 0.99 Cardinal Number of children in the household
Income 18.01 13.86 Cardinal Total gross annual personal income £000s

provides variable definitions and also their sample characteristics. This includes frequen-
cies and sample proportions for nominal variables and means and standard deviations
(S.dev) for cardinal variables. The variables include the personal characteristics such
as age, gender, ethnicity, education levels, marital status, occupational status, drinking
and smoking behaviour as well as general health. Regional variables are also included
as indicative of other general ‘external’ factors that can affect well-being. The proximity
of sports facilities to respondents is also included to control for the availability of some
aspects of sports supply, though it should be remembered that not all sports participation
takes place in dedicated facilities. Lastly, but of most significance for this paper, a number
of measures of sport are included to capture the ‘life event’ of sports participation.
336 P. Downward and S. Rasciute

Both aggregate and disaggregated measures of sports are employed. In the former
case this is to examine the overall significance of sport to well-being. In the latter
case the aim is to identify if the form of sports participation affects well-being. It is
hypothesised that there will be an additional impact upon well-being from activities
that have more potential for social interactions, and hence cross-participant externali-
ties. Team sports and sports in which it is more obvious that an opponent is required,
such as racquet sports, etc, are consequently classified as social interaction sports,
whilst others are classified as non-social interaction sports. It is recognised that this is
a fairly arbitrary and exploratory distinction as the official data do not identify the
specific form with which activity takes place.7
Table 3 lists the sports for which data are collected and from which the aggregate
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

and disaggregate measures of participation were derived. Activities regarded as


social-interaction sports are indicated in Table 3 by the reference ‘SI’ in brackets next
to the variable name. The table also reports the percentage frequency of participation,
also in brackets, based on participation in the activity in the last four weeks or not.
This indicates the relative scale of participation in each case, which is typically low
(see Downward, Dawson, and Dejonghe 2009). The measures of sports participation
employed in the research include dummy variables to identify if any sport was under-
taken over the last 4 weeks prior to the interview, and if any sport was undertaken over
the last 12 months prior to the interview. Variables also measure the number of sports
undertaken in these periods, as well as variables measuring the total number of times
sport was undertaken over the last 4 weeks prior to the interview and the total time
spent on sport in the last four weeks. This is a measure of the intensity of sports partic-
ipation as it reflects not only the frequency of participation over the period but also
accounts for the duration of activity. Measures of these variables were also identified
for the social interaction and the non-social interaction activities.

4. Estimation
As a unique but cross-section dataset is employed in this research, it is important to
recognise that it precludes analysis of any dynamic effects. Further, as with much
of the literature, the analysis proceeds with the caveat that causal claims are weak.
The correlations may suffer from some endogenity problems stemming from, for
example, health being a component of well-being. Consequently the paper follows
the specifications that populate the literature surveyed in Section 2. Nonetheless, in
the estimated results all specifications appear to be robust in respect of the health
variable.
A variety of estimators have been used to examine well-being. The dependent vari-
able has been treated as both a cardinal and ordinal measurement. In this paper an
ordinal estimator is employed as it is felt that this most consistently captures the char-
acteristics of utility. This implies that utility is ordered, lacks a conditional mean, and
that marginal rates of substitution, for example given by the ratio of coefficients, has
meaning rather than the direct (interpersonally comparable) estimates of marginal
utility implied by coefficients in cardinal estimators (Greene and Hensher 2010; Frey
2008; Boes and Winkelmann 2004).
Following Greene and Hensher (2010), the random utility model for individual i is:

yi* = β ′xi + ε i , i = 1,L, N . (1)


International Review of Applied Economics 337

Table 3. Sports activities.


Swimming or diving indoors (18.46%) Table tennis (SI) (1.61%)
Swimming or diving outdoors (4.58%) Track and field athletics (0.27%)
BMX, cyclo-cross, mountain biking (0.94%) Jogging, cross-country, road running
(6.31%)
Cycling for health, recreation, training, Angling or fishing (1.89%)
competition (10.99%)
Cycling to get to places (5.26%) Yachting or dingy sailing (0.81%)
Bowls indoors (1.00%) Canoeing (0.60%)
Bowls outdoors (0.60%) Windsurfing or boardsailing (0.19%)
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

Tenpin bowling (4.00%) Ice skating (0.89%)


Health, fitness, gym, conditioning activities Curling (SI) (0.04%)
(16%)
Keepfit, aerobics, dance exercise, exercise bike Golf, pitch and putt, putting (6.03%)
(8.00%)
Judo (0.05%) Skiing (0.61%)
Karate (0.40%) Horse riding (1.43%)
Taekwando 0.19%) Climbing/mountaineering including indoor
climbing (0.74%)
Other Martial Arts including self defence, tai Hill trekking or backpacking (1.54%)
chi (1.01%)
Weight training including body building Motor sports (0.84%)
(3.81%)
Weightlifting (1.55%) Shooting (0.97%)
Gymnastics (0.34%) Volleyball (SI) (0.37%)
Snooker, pool, billiards exclude bar billiards Orienteering (0.18%)
(8.61%)
Darts (3.36%) Rounders (SI) (0.57%)
Rugby League (SI) (0.14%) Rowing (0.38%)
Rugby Union (SI) (0.69%) Triathlon (0.08%)
American football (SI) (0.08%) Boxing (0.63%)
Football indoors including 5-a-side and 6-a-side Waterskiing (0.12%)
(SI) (3.07%)
Football outdoors including 5-a-side and 6-a- Lacrosse (SI) (0.02%)
side (SI) (6.82%)
Gaelic sports (SI) (0.02%) Yoga (3.17%)
Cricket (SI) (1.85%) Fencing (SI) (0.06%)
Hockey (SI) (0.40%) Frisbee (0.05%)
Archery (0.38%) Trampolining (0.19%)
Baseball/softball (SI) (0.19%) Rambling/walking for pleasure (0.10%)
Netball (SI) (0.49%) Skittles (0.17%)
Tennis (SI) (3.03%) Pilates (0.16%)
Badminton (SI) (3.32%) Any other water sport (0.35%)
Squash (SI) (1.63%) Other sport (1.57%)
Basketball (SI) (1.35%)
338 P. Downward and S. Rasciute

In this equation, the dependent variable represents the underlying random utility, or
latent variable, in which continuous latent utility yi* is observed in discrete form
through the censoring mechanism:

yi = 0 if µ −1 < yi* ≤ µ0 ,
= 1if µ0 < yi* ≤ µ1,
= 2 if µ1 < yi* ≤ µ 2 ,
=L
= J if µ J −1 < yi* ≤ µ J .
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

(2)

The vector xi is a set of K covariates that are assumed to be strictly independent of εi;
β is a vector of K parameters. The thresholds divide the range of utility into cells that
are then identified with the observed ratings of happiness. An important feature of
the ordered choice model is that the threshold parameters have no obvious interpreta-
tion, although they can indicate something about the distribution of preferences of
individuals (Greene and Hensher 2010).
As these threshold values are unlikely to be the same for all individuals, it is likely
that there will be unobserved heterogeneity across individual rankings. If this is not
entirely random but conditional on the covariates of well-being, this would produce
biased estimates if it is not accounted for. This is accommodated for by the use of the
Heterogeneous Thresholds Ordered Choice model. In this model, threshold parame-
ters depend on variables such that µi = δ′i zi, where zi is a subset of the variables
included in xi. This and other models accounting for heterogeneity are reviewed in
great detail in Greene and Hensher (2010).8
The existing literature applying ordered choice models to investigate well-being
has tended to concentrate empirical discussion upon estimated coefficients. As noted
earlier, this is problematic. The effect of a change in one of the variables in the model
depends on all the model parameters, the data, and which probability (cell) is of inter-
est. Therefore, one possibility is to compute partial effects to give the impacts on
the specific probabilities per unit change in the covariate.9 The partial effects in the
ordered choice models are expressed as

∂ Pr ob ( y = j x i )
δ i ( xi ) =
∂x i
[ ]
= f ( µ j −1 − β ′x i ) − f ( µ j − β ′x i ) β ( 3)

Alternatively, ratios of coefficients are meaningful and useful substitutes for partial
effects (Greene and Hensher 2010; Boes and Winkelmann 2004). This means that
‘shadow price’ measures can be calculated for the effects of events on well-being. In
the well-being literature, the relationship between income and well-being has been
used to put a monetary value on non-traded goods, in the context of cost-benefit anal-
ysis (Boes and Winkelmann 2006). When a specific factor has a negative effect on
individual’s well-being, the shadow price can be understood in terms of how much
income should increase to offset the negative effect of the factor, in order to keep the
person at the same level of well-being. Similarly, when a specific factor has a positive
International Review of Applied Economics 339

effect on individual’s well-being, the shadow price can be understood to be how much
income an individual would be willing to give up in order to obtain that non-traded
good. Oswald and Clarke (2002) describe the measures as ‘compensating values’ for
life events.

5. Results
Seven different specifications of the model are estimated and presented in Table 4. For
illustration, specification 1 presents the results of a basic Ordered Probit model, where
all of the explanatory variables are included irrespective of their statistical signifi-
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

cance, and the sports impact investigated is the number of times in which sport was
undertaken in the last four weeks (numtimes4). The threshold estimates are presented
in the lower part of the table and these were treated as homogeneous. Relaxing this
assumption, specifications 2–7 then report refined estimates from a Heterogeneous
Thresholds Ordererd Probit (HTOP) model. Thresholds are explicitly modelled to
account for heterogeneity by including individual characteristics as covariates of the
thresholds. The statistical significance of the threshold parameters and covariates and
better model fit indicated by the log-likelihood, pseudo R-squared and Chi-squared
values suggest that the HTOP model is most appropriate. Parsimonious versions of
this model are derived by excluding insignificant variables. Specification 2 includes
the variable that takes into account if a respondent participated in any sports over the
last 12 months prior to the interview. Specification 3 includes the variable to account
for the participation in any sports over the last four weeks prior to the interview.
The variable controlling for the number of sports activities undertaken in the last 12
months is added to Specification 4, while Specification 5 accounts for the number of
sports activities undertaken over the last four weeks. Specification 6 contains the vari-
able that takes into account the typical time in minutes of participation in all sports
activities. Finally, Specification 7 controls for the number of times participation in any
sport activity takes place in the last four weeks.
Table 4 only presents estimated coefficients. However, as discussed in Section 4,
as neither the sign nor the magnitude of the estimated coefficients are necessarily infor-
mative, further post-estimation is carried out as reported in Table 5. This is applied to
all of the aggregate and disaggregated sports variables. In the latter case, variables iden-
tifying social interaction and non-social interaction sports were included in the esti-
mations as it is possible that individuals could participate in combinations of the sports.
For brevity, only the estimated partial effects for the highest value of the happiness
variable are presented in each case and for the sports covariates. They are shown to be
positive. Because the partial effects of all the potential values of the happiness variable
can only change sign once, this suggests that the probabilities of levels of happiness
are redistributed such that the event of sport participation, however defined, increases
the probability of higher levels of happiness (as measured). Notably the probabilities
are greater for the social interaction sports measures. This suggests that the form of
sport participation is associated with happiness, and that this is consistent with viewing
some aspects of sports participation as the consumption of relational goods.
The t-tests in Table 5 illustrate that the aggregated measures of sport have a signif-
icant association with happiness. This is also the case for the social interaction sports
and some of the non-social interaction sports. Nonetheless, the disaggregated measures
are all jointly significant as indicated by the Likelihood ratio tests reported in the last
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

Table 4. The estimated coefficients for the ordered probit model with heterogeneous thresholds.
340

1 2 3 4 5 6 7
Variables Coef t-stats Coef t-stats Coef t-stats Coef t-stats Coef t-stats Coef t-stats Coef t-stats
Constant 1.0852 8.20 0.6158 4.28 0.6319 4.52 0.6465 4.62 0.7296 5.29 0.7377 5.36 0.7305 5.30
Single 0.2642 5.74 0.3552 5.23 0.3524 5.19 0.3534 5.20 0.3525 5.20 0.3536 5.22 0.3544 5.22
Married 0.6188 14.32 0.6537 10.97 0.6597 11.08 0.6666 11.16 0.6619 11.10 0.6630 11.20 0.6638 11.13
Separated 0.2705 5.91 0.2427 5.37 0.2437 5.40 0.2427 5.37 0.2426 5.37 0.2413 5.34 0.2432 5.38
Income 0.0026 3.55 0.0079 5.51 0.0077 5.41 0.0077 5.33 0.0082 5.71 0.0082 5.74 0.0082 5.71
White 0.0810 1.53 0.2046 3.59 0.2079 3.64 0.1965 3.43 0.2130 3.73 0.2150 3.81 0.2165 3.79
Asian 0.0170 0.23
Black 0.1140 1.78
P. Downward and S. Rasciute

NorthEast 0.3016 5.00 0.2035 4.29 0.2058 4.33 0.2037 4.28 0.2033 4.28 0.2024 4.27 0.2011 4.23
NorthWest 0.1313 2.60
Yorks 0.2120 4.19 0.0884 3.10 0.0888 3.12 0.0897 3.14 0.0888 3.12 0.0894 3.12 0.0886 3.11
EastMid 0.1664 3.24 0.1769 2.67 0.1797 2.72 0.1767 2.66 0.1833 2.77 0.1827 2.76 0.1843 2.78
WestMid 0.1640 3.29
East 0.1342 2.67
London −0.0127 −0.22
SouthEast 0.0849 1.54
Working 0.0236 0.44 0.2752 5.83 0.2771 5.88 0.2778 5.88 0.2835 6.01 0.2860 6.00 0.2848 6.04
Student −0.0300 −0.43
Housekeeper −0.0416 −0.70
IllnotWork −0.1205 −1.72
Unemployed −0.3000 −4.38 −0.2359 −4.71 −0.2382 −4.75 −0.2376 −4.73 −0.2367 −4.71 −0.2393 −4.78 −0.2380 −4.74
Retired 0.1578 2.58 0.1984 4.89 0.1960 4.83 0.1982 4.89 0.1982 4.89 0.1967 4.85 0.1968 4.86
HE −0.0748 −2.85 −0.0577 −3.00 −0.0593 −3.07 −0.0613 −3.17 −0.0595 −3.08 −0.0574 −2.98 −0.0596 −3.09
Alevel −0.0297 −1.04
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

Table 4. (Continued).
1 2 3 4 5 6 7
Variables Coef t-stats Coef t-stats Coef t-stats Coef t-stats Coef t-stats Coef t-stats Coef t-stats
Apprenti 0.0320 0.76
Olevel5 −0.0563 −1.98 −0.0442 −1.97 −0.0438 −1.95 −0.0437 −1.94 −0.0445 −1.98 −0.0448 −1.99 −0.0446 −1.98
Gender −0.0572 −3.16 −0.0425 −2.43 −0.0457 −2.60 −0.0497 −2.81 −0.0499 −2.82 −0.0544 −3.07 −0.0500 −2.84
Drink4to6 −0.0918 −2.59 0.1255 1.98 0.1199 1.89 0.1192 1.88 0.1304 2.06 0.1298 2.05 0.1316 2.08
Drink1to3 −0.0286 −0.98 0.1611 3.78 0.1559 3.66 0.1527 3.57 0.1625 3.81 0.1625 3.81 0.1628 3.82
DrinkLess −0.0659 −2.17
NotDrink −0.0111 −0.33 −0.1103 −2.09 −0.1122 −2.13 −0.1108 −2.10 −0.1214 −2.31 −0.1245 −2.36 −0.1222 −2.32
Smk4to6 0.0083 0.08
Smk1to3 −0.0567 −0.83
SmkLess 0.0048 0.05
NotSmoke 0.1048 5.00 0.3421 8.23 0.3403 8.17 0.3371 8.09 0.3474 8.36 0.3507 8.45 0.3472 8.36
GenHealth 0.3170 27.53 0.3244 32.99 0.3228 32.85 0.3226 32.88 0.3230 32.96 0.3216 32.86 0.3215 32.86
LillHarm −0.0338 −1.51
SpClose .0.0000 0.15
Nadult 0.0549 4.98 0.0576 5.24 0.0570 5.19 0.0555 5.04 0.0565 5.13 0.0549 4.99 0.0554 5.03
Nchild −0.0056 −0.59
Age −0.0259 −7.82 −0.0308 −8.77 −0.0306 −8.74 −0.0295 −8.31 −0.0313 −8.91 −0.0313 −8.95 −0.0313 −8.93
AgeSquared 0.0003 8.90 0.0003 9.34 0.0003 9.35 0.0003 9.11 0.0003 9.18 0.0003 9.17 0.0003 9.17
Work_Asia −0.1166 −1.76 −0.1728 −3.68 −0.1715 −3.65 −0.1707 −3.64 −0.1706 −3.63 −0.1676 −3.57 −0.1684 −3.58
Work_NE −0.1319 −2.19 −0.1733 −2.94 −0.1753 −2.98 −0.1740 −2.95 −0.1723 −2.93 −0.1746 −2.96 −0.1723 −2.93
Work_SE −0.0098 −0.20 −0.0671 −2.21 −0.0677 −2.24 −0.0689 −2.28 −0.0673 −2.22 −0.0665 −2.20 −0.0658 −2.17
Work_Lon 0.0418 0.77 −0.1050 −3.14 −0.1045 −3.12 −0.1052 −3.14 −0.1053 −3.14 −0.1061 −3.17 −1.1062 −3.17
International Review of Applied Economics

Work_SW 0.0862 1.56 −0.0627 −1.91 −0.0634 −1.93 −0.0652 −1.99 −0.0638 −1.95 −0.0631 −1.92 −0.0628 −1.92
Tosckmetd 0.0618 3.34 0.1376 3.34 0.1349 3.28 0.1243 3.00 0.1487 3.62 0.1494 3.64 0.1500 3.65
341
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

Table 4. (Continued).
342

1 2 3 4 5 6 7
Variables Coef t-stats Coef t-stats Coef t-stats Coef t-stats Coef t-stats Coef t-stats Coef t-stats
NuTimes4 0.0027 4.03 0.0027 4.38
AnySpo12 0.1492 3.13
AnySpo4 0.1787 4.54
NumSpo12 0.0265 4.54
NumSpo4 0.0128 2.72
TotalMin 0.0017 4.94
Estimates of Threshold
P. Downward and S. Rasciute

Mu(1) 0.2816 13.58 −1.4153 −13.54 −1.4049 −13.49 −1.3950 −13.40 −1.3703 −13.21 −1.3718 −13.23 −1.3701 −13.22
Mu(2) 0.6741 36.07 −0.5605 −8.75 −0.5498 −8.72 −0.5395 −8.58 −0.5140 −8.25 −0.5153 −8.27 −0.5146 −8.26
Mu(3) 0.9494 58.22 −0.2282 −4.20 −0.2176 −4.10 −0.2071 −3.91 −0.1812 −3.48 −0.1824 −3.50 −0.1818 −3.49
Mu(4) 1.5007 117.71 0.2209 4.81 0.2313 5.22 0.2426 5.50 0.2683 6.23 0.2670 6.20 0.2677 6.21
Mu(5) 1.8327 161.07 0.4171 9.56 0.4275 10.22 0.4391 10.55 0.4646 11.45 0.4633 11.41 0.4641 11.43
Mu(6) 2.4328 235.61 0.6918 16.65 0.7020 17.73 0.7141 18.14 0.7396 19.32 0.7384 19.28 0.7391 19.30
Mu(7) 3.2719 313.17 0.9777 24.19 0.9876 25.74 1.0000 26.23 1.0260 27.72 1.0248 27.67 1.0256 27.69
Mu(8) 3.9032 304.86 1.1506 28.74 1.1602 30.58 1.1728 31.12 1.1989 32.78 1.1979 32.73 1.1989 32.75
Threshold Covariates
Single 0.0498 2.50 0.0479 2.41 0.0479 2.41 0.0484 2.43 0.0495 2.49 0.0490 2.46
Married 0.0292 1.73 0.0306 1.82 0.0324 1.92 0.0311 1.85 0.0312 1.85 0.0310 1.84
Income 0.0018 3.97 0.0018 3.88 0.0018 3.90 0.0019 4.25 0.0019 4.25 0.0019 4.24
White 0.0653 3.37 0.0670 3.46 0.0648 3.34 0.0697 3.59 0.0697 3.60 0.0695 3.58
EastMid 0.0474 2.32 0.0483 2.37 0.0477 2.34 0.0500 2.45 0.0501 2.46 0.0502 2.46
Working 0.0614 4.22 0.0623 4.30 0.0627 4.31 0.0644 4.44 0.0648 4.46 0.0647 4.46
Drin4to6 0.0586 2.90 0.0572 2.83 0.0576 2.85 0.0612 3.03 0.0608 3.01 0.0613 3.04
Drink1to3 0.0487 3.52 0.0473 3.43 0.0463 3.34 0.0497 3.59 0.0493 3.57 0.0497 3.60
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

Table 4. (Continued).
1 2 3 4 5 6 7
Variables Coef t-stats Coef t-stats Coef t-stats Coef t-stats Coef t-stats Coef t-stats Coef t-stats
NotDrink −0.0639 −3.44 −0.0645 −3.48 −0.0649 −3.49 −0.0681 −3.67 −0.0684 −3.69 −0.0679 −3.66
NotSmoke 0.0920 6.45 0.0920 6.45 0.0911 6.38 0.0950 6.67 0.0951 6.67 0.0948 6.65
Age −0.0014 −2.92 −0.0015 −3.05 −0.0015 −2.91 −0.0019 −4.12 −0.0019 −4.08 −0.0019 −4.11
Ticknetd 0.0273 2.06 0.0270 2.04 0.0251 1.88 0.0329 2.49 0.0331 2.51 0.0334 2.53
AnySpo12 0.0502 3.10
AnySpo4 0.0510 3.91
NumSpo12 0.0063 3.40
Loglikelihood −29583.25 −29461.74 −29456.54 −29458.38 −29463.79 −29455.99 −29458.79
Pseudo R-sq 0.03896 0.0429 0.04308 0.043016 0.04284 0.04309 0.043
Chi-squared 2398.558 2641.578 2651.979 2648.295 2637.485 2653.07 2647.477
The dependent variable, happiness, takes values between 1 and 10, where ‘1’ indicates extremely unhappy and ‘10’ extremely happy.
International Review of Applied Economics
343
344 P. Downward and S. Rasciute

Table 5. Partial effects and WTP measures.


Partial Willingness to Log-likelihood
Effects Pay Measures t-stats Ratio Index
AnySport12 0.0312* 18.8861 3.13
AnySport4 0.0374* 23.2078 4.54
NumSport12 0.0059* 3.4416 4.54
NumSport4 0.0026* 1.5610 2.72
TotalMinutes 0.0004* 0.2152 4.94
NumDays4 0.0006* 0.3293 4.38
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

NumSIsport12 0.0031*** 1.9367 1.70 6.74


NumNotSIsport12 0.0011 0.6835 1.38
NumSIsport4 0.0053*** 3.2250 1.84 8.02
NumNotSIsport4 0.0018 1.1000 1.52
AnySIsport12 0.0065*** 1.9316 1.70 8.18
AnyNotSIsport12 0.0031 0.6835 0.74
AnySIsport4 0.0086*** 5.2025 1.72 8.87
AnyNotSIsport4 0.0077** 4.7468 2.16
TotalSIminutes 0.0007* 0.4375 2.92 25.48
TotalNotSiminutes 0.0003* 0.1875 3.78
NumSIdays 0.0008*** 0.5063 1.87 17.28
NumNotSIdays 0.0005* 0.3165 3.49
* Significant at 1% level ** Significant at 5% level *** Significant at 10% level.
Partial effects are reported for the highest outcome of the dependent variable
The critical value of Chi-squared at 5% significance level is 5.99
The dependent variable, happiness, takes values between 1 and 10, where ‘1’ indicates extremely unhappy
and ‘10’ extremely happy.

column of Table 5. Table 5 also presents the willingness to pay values of sports partic-
ipation, which are obtained by dividing the respective sports coefficients by the coef-
ficients on income in the particular regression. The results echo the partial effects in
indicating that the value of happiness is enhanced by sports participation and that this
is particularly the case with social interaction sports. Taking the aggregate measures
as an example, the results indicate that, on average, a person values participation in at
least one sport over the year at about £19,000 per year with the figure rising to about
£23,000 if participating in any sport took place in the period of the last four weeks.
The results also show that an additional sport participated in relative to their portfolio
of activities is valued at about £1600 per year if this occurs during the period of four
weeks before the interview and about £3500 per year in the period of 12 months before
the interview (specifications 5 and 4 respectively). Finally, on average, an extra minute
of sports is valued at approximately £215 per year if the sport is undertaken in the last
four weeks, while an extra session in which sport is participated in, during the last four
weeks, is worth approximately £329 per year.10
The disaggregated results indicate that evaluations applying to the period of the
last four weeks are typically higher than those for the last 12 months and that generally
the social interaction values exceed those attached to the aggregated versions of the
variables. The results also show that the aggregate ‘anysport’ variable for participation
in the last 12 months is greater than that for the last four weeks. In addition, both of
International Review of Applied Economics 345

the aggregate ‘anysport’ variables have much higher values overall than the other
participation variables.
The generally higher values of participation in the last four weeks compared to the
last 12 months could suggest that there is some form of depreciation of values associ-
ated with longer time horizons being considered by respondents, which could be
consistent with the dynamic adjustment to happiness identified in the literature. The
effects of marital status on happiness, for example, have been found to be complex.
Based on panel data, it is shown that becoming married generates a positive ‘shock’
to happiness that eventually returns to previous levels after about 5 years. In contrast,
the reduction in happiness from suffering the bereavement of a partner dissipates more
slowly over 8 years (Lucas et al. 2003). Likewise complex lead and lag effects are
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

identified for divorce, with females being more affected than males (Clark et al. 2008)
and those that remarry recovering levels of happiness after divorce (Johnson and Wu
2002). In contrast, Gardner and Oswald (2006) identify that whilst divorce can reduce
happiness, allowing for higher levels of stress before divorce, implies that the apparent
reductions in happiness are actually a return to normal levels of happiness. Further,
Stutzer and Frey (2006) identify that there are selection effects present, such that it is
happier singles that are more likely to opt for marriage. Frijters, Johnston, and Shields
(2008) synthesise such findings, arguing that events such as separation have anticipa-
tion, selection and adaptation effects. Although speculative in the case of sport, it
could be that the benefits derived from participation, in terms of, say, physical or
mental health impacts, deteriorate without further investment as could be understood
with reference to standard human capital theory (Grossman 1972, 1999).
In the case of both of the aggregate ‘anysport’ variables yielding much higher
values associated with happiness, the results could suggest that a more complex form
of aggregation than suggested in the modelling exists. It could be that the individual
gets greater benefits from participating in both social interaction and non-social inter-
action sports than either separately. In this respect it might be hypothesised that there
are increasing returns to participation in sport generally. Clearly both of these issues
require further investigation.

6. Conclusion
In both the UK and elsewhere, encouraging citizens to engage in sports activity is now
an important public policy issue. It is argued that there are benefits in terms of health
and well-being to individuals as well as to society through externalities. This paper
demonstrates that sports participation has a positive association with the subjective well-
being of the population, as measured by happiness and, moreover, estimates its monetary
value. The results indicate that in the aggregate, on average, a person values participation
in sport to be £19,000–23,000 per year, while an additional sport participated in relative
to their portfolio is valued at about £1600–3500 per year. Furthermore, on average, an
extra minute of sports, best interpreted as a discrete change corresponding approxi-
mately to a session (see Note 10), is valued at £215 per year, while an extra direct session
at £329 per year. Further variance in these results can be noted if one considers the
social interaction nature of some forms of sport. In general, social interactions based
in sport can raise well-being further. There is also the suggestion that the value of sport
can vary over time and that its value through multi-sport participation may be complex.
Nonetheless, the positive values identified provides some impetus to public policy
looking to promote well-being through sport but, of course, it remains that future
346 P. Downward and S. Rasciute

research is required that unpicks more specifically how and to whom the benefits
accrue, how they link to any external benefits and costs, and the consequent levels of
funding that might legitimately be committed to support sport as a policy objective. It
may also be the case that the measured benefits are smaller than for activities that the
individual chooses instead of participation in sport. This would help to explain the
relatively low levels of sport participation. However, the above research suggests a
method by which an evaluation of these alternatives could be undertaken as well as
applied to evaluate specific sports. This would increase the accountability of the use
of public investment.
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

Acknowledgements
An earlier version of this paper was presented at the EASM conference in Amsterdam 12 to 16
September 2009. The authors acknowledge useful feedback from this event. Very helpful
comments and feedback from two anonymous referees and the editor of this journal have also
considerably improved the paper.

Notes
1. Four main reasons are given that qualify the evidence generated from evaluation studies.
These are: the lack of clear definition of sport, and the form of its participation, inappro-
priate sampling and longitudinal evidence, imprecision about the sufficiency of sport
achieving policy goals and a limited literature base.
2. Personal characteristics include age, gender, ethnicity, and personality. Socially developed
characteristics include education, health, type of work and unemployment. How individu-
als spend their time includes activities such as hours worked, commuting, caring for others,
community involvement and volunteering, exercise, and religious activities. Attitudes and
beliefs towards others include attitudes towards personal circumstances, trust, political
persuasion and religion. Relationships include marriage and intimate relationships, having
children, seeing family and friends. Finally, the wider environmental determinants include
income inequalities, unemployment, inflation, welfare systems, the degree of democracy,
climate, the natural environment, safety and urban deprivation and urbanisation. Examples
of the literature on the impact of physical exercise on well-being from smaller-scale studies
are Scully et al. (1999) and Biddle and Ekkekakis (2005).
3. This scale is included, for example, in the British Household Panel and the Health Survey
for England. The binary recoding and scale items can be adjusted to emphasise positive and
negative symptoms.
4. Both time periods are now used for data collection on sports participation in the UK. Whilst
it is felt that face validity of measures over a four-week period will be greater, the seasonal
nature of sport means that four-week data collection can miss activity out of season. Time
periods referred to in data collection vary internationally (see Downward et al. 2009).
5. This sample is based on the elimination of missing cases from the happiness variable as
well as across all of the covariates in the analysis so that different model specifications
have the same sample size.
6. Because datasets vary, the literature does not employ the same determinants. This naturally
raises the issue of bias in comparing results. However, the results obtained in this study for
the covariates used are qualitatively similar to the literature.
7. It is possible that a respondent could classify themselves as having played football when in
fact they have been kicking a ball about by themselves. Likewise, activities such as martial
arts may be undertaken in classes as part of a group but with the potential to be an individ-
ual within this group and, further, without it being necessary to have an opponent. Of
course, the fact of participating in the activity in a class makes the activity different from,
say, running alone outside. The same could be true of weight training in a fitness centre. It
is clear that the degree of social interactions could vary and requires further exploration.
8. These include the latent class and random parameters models. A very important point to
note is that they are not commensurate with the Generalised Ordered models emanating from
International Review of Applied Economics 347

Williams (2006) and Boes and Winkelmann (2004), which reject the parallel regression
assumption that the behavioural parameters are constant across discrete outcomes and model
separate regressions for each outcome, implying that the magnitude of a well-being covariate
depends upon the level of well-being considered. As Greene and Hensher (2010) show, they
do not impose the restriction that the sum of probabilities is one across the outcomes as all
parameters can vary freely (which requires parallel regressions) and moreover implies that
the probability that the random variable (well-being) is equal to a particular outcome is
defined as when well-being is at that value. In other words, the data generating mechanism
collapses onto the observed data. Parameter variation is thus defined across the outcomes
rather than the individuals for whom the outcomes are possibilities.
9. Under certain conditions it might be regarded that a positive (negative) coefficient is
connected with a reduction (increase) in the probability in the lowest cell and an increase
(reduction) in the probability in the highest cell. With the single crossing feature of the
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

model, such that some probabilities fall and some rise, one can imply that probabilities
have shifted in a particular direction.
10. The result for minutes should be interpreted with caution. For each sporting activity the
questionnaire asks ‘How long do you normally spend in minutes’ on the activity.
Responses are typically clustered as discrete values, for example on 60 minutes, 120
minutes and so on, with some small frequencies of values in between. The ‘minute’
changes are therefore most likely identifying approximate discrete ‘session’ changes, hence
accounting for the similar, if smaller, values to those of sessions investigated directly.

References
Becchetti, A., A. Pelloni, and F. Rossetti. 2008. Relational goods, sociability, and happiness.
Kyklos 61, no 3: 343–63.
Biddle, S.J.H., and P. Ekkekakis. 2005. Physical active lifestyles and well-being. In The
science of well-being, ed. F.A. Huppert, N. Baylis and B. Keverne. Oxford: Oxford
University Press.
Blanchflower, D.G., and A.J. Oswald. 2008. Is well-being U-shaped over the life cycle?
Social Science & Medicine 66: 1733–49.
Boes, S., and R. Winkelmann. 2004. Income and happiness: New results from generalised
thresholds and sequential models. IZA Discussion paper, No 1175.
Boes, S., and R. Winkelmann. 2006. The effect of income on positive and negative subjective
well-being. Socioeconomic Institute. University of Zurich, Working paper No 0605.
Brown, S., K. Taylor, and S. Wheatley Price. 2005. Debt and distress: Evaluating the psycho-
logical cost of credit. Journal of Economic Psychology 26: 642–63.
Clark, A.E., E. Diener, Y. Georgellis, and R.E. Lucas. 2008. Lags and leads in life satisfac-
tion: A test of the baseline hypothesis. Economic Journal 118: F222–43.
Coalter, F. 2007. A wider social role for sport. London: Routledge.
Collins, M., and T. Kay. 2002. Sport and social exclusion. London: Routledge.
Department of Health. 2004. At least five a week: Evidence on the impact of physical activity
and its relationship to health. A report from the Chief Medical Officer. http://www.
dh.gov.uk/en/Publicationsandstatistics/Publications/PublicationsPolicyAndGuidance/
DH_4080994
DCMS/Strategy Unit. 2002. Game plan: A strategy for delivering government’s sport and
physical activity objectives. London: DCMS/Strategy Unit.
Dolan, P., T. Peasgood, and M. White. 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.
Downward, P.M., A. Dawson, and T. Dejonghe. 2009. Sports economics: Theory, evidence
and policy. London: Butterworth-Heinemann.
Frey, B.S. 2008. Happiness: A new revolution in economics. Cambridge Massachusetts: MIT
Press.
Frijters, P., D.W. Johnston, and M.A. Shields. 2008. Happiness dynamics with quarterly life
event data. IZA Discussion Paper, No 3604, July.
Gardner, J., and A.J. Oswald. 2006. Do divorcing couples become happier by breaking up.
Journal of the Royal Statistical Society (A) 169, no. 2: 319–36.
348 P. Downward and S. Rasciute

Goldberg, D. 1972. The detection of psychiatric illness by questionnaire. Oxford: Oxford


University Press.
Golden, L., and B. Wiens-Tuers. 2006. To your happiness? Extra hours of labor supply and
worker well-being. Journal of Socio-Economics 35, no. 2: 382–97.
Greene, W.H., and D.A. Hensher. 2010. Modelling ordered choices: A primer. Cambridge:
Cambridge University Press.
Grossman, M. 1972. On the concept of health capital and the demand for health. Journal of
Political Economy 80: 223–55.
Grossman, M. 1999. The human capital model of the demand for health. National Bureau of
Economic Research, Working Paper 7078.
Johnson, D.R., and J. Wu. 2002. An empirical test of crisis, social selection, and role explana-
tions of the relationship between marital disruption and psychological distress: A pooled
time-series analysis of four-wave panel data. Journal of Marriage and Family 64: 211–24.
Downloaded by [The University Of Melbourne Libraries] at 06:48 10 October 2014

Kahneman, D., E. Diener, and N. Schwarz, eds. 1999. Well-being: The foundations of hedonic
psychology. New York: Russell Sage Foundation.
Kahneman, D., and A.B. Krueger. 2006. Developments in the measurement of subjective
well-being. Journal of Economic Perspectives 20, no. 1: 3–24.
Lechner, M. 2009. Long-run labour market and health effects of individual sports activities.
Journal of Health Economics 28, no. 4: 839–54.
Lucas, R.E., A.E. Clark, Y.Y. Georgellis, and E. Diener. 2003. Re-examining adaptation and
the set point model of happiness: Reactions to changes in marital status. Journal of
Personality and Social Psychology 84: 527–39.
Nicholls, G. 2007. Sport and crime reduction: The role of sports in tackling youth crime.
London: Routledge.
Oswald, A., and A. Clarke. 2002. A simple statistical method for measuring how life events
affect happiness. International Journal of Epidemiology 31, no 6: 1139–44.
Rasciute, S., and P. Downward. 2010. Health or happiness? What is the impact of physical
activity on the individual. Kyklos 63, no. 2: 256–70.
Scully, D., J. Kremer, M. Meade, R. Graham, and K. Dudgeon. 1999. Physical exercise and
psychological well-being: a critical review. British Journal of Sports Medicine 32: 11–20.
Shields, M.A., and S. Wheatley Price. 2005. Exploring the economic and social determinants
of psychological well-being and perceived social support in England. Journal of the Royal
Statistical Society (A) 168, no. 3: 513–37.
Shields, M.A., S. Wheatley Price, and M. Wooden. 2009. Life satisfaction and the economic
and social characteristics of neighbourhoods. Journal of Population Economics 22: 421–43.
Stutzer, A., and B.S. Frey. 2006. Does marriage make people happy, or do happy people get
married? The Journal of Socio-Economics 35: 326–47.
Van Praag, B.M.S., and P. Frijters. 1999. The measurement of welfare and well-being: The
Leyden approach. In Well-being: The foundations of hedonic psychology, ed. D. Kahneman,
E. Diener and N. Schwarz. 413–32. New York: Russell Sage Foundation.
WHO World Health Day. 2002. http://www.who.int/world-health-day/brochure.en.pdf.
Williams, R. 2006. Generalised ordered logit/partial proportional odds models for ordinal
dependent variables. The STATA Journal 6, no 1: 58–62.
Winkelmann, R. 2005. Subjective well-being and the family: Results from an ordered probit
model with multiple random effects. Empirical Economics 30: 749–61.
Winkelmann, L., and R. Winkelmann. 1998. Why are the unemployed so unhappy? Evidence
from panel data. Economica 65, no. 257: 1–15.

You might also like