A Review of Well-Being Valuation for Sports, Culture and Leisure Activities

Abstract

It is widely acknowledged that engagement in sports, as well as in cultural and leisure activities benefits people’s well-being. However, what remains unclear is the monetary value of this contribution. For creating sustainable policies that withstand austerity measures, it is crucial to have a better understanding of the value of these activities. This scoping review provides the first overview of studies that estimate the value of leisure activities by using the compensating variation approach exclusively. The purpose of the review is to identify methodological issues to detect knowledge gaps and to investigate the conduct of research. Records were retrieved from several scientific databases and Google Scholar. To analyze the results, all studies were summarized for country, scope, source, database, sample, measures, control variables, design, common bias and outcomes. The search resulted in eleven original studies of which five were commissioned reports delivered by academics. Important methodological issues were the diverse use of control and well-being variables and the endogeneity and selection biases that contributed to a wide range of monetary values. Because of the variability in their conduct, methodological standardization is required to reach a consensus on the contribution of sports and cultural and leisure activities to society.

1. Introduction

For centuries, philosophers have considered well-being to be the ultimate goal for human action [1]. Consequently, the understanding, measurement and improvement of well-being have been major goals for individuals, as well as governments [2]. However, well-being outcomes are not yet commonly used in policy evaluation despite the fact that those measurements of well-being are claimed to be valid, and that the government has control over factors that are positively correlated with well-being levels [3,4,5]. Well-being has been positively associated with good health, better social relationships and work performance [6]. Furthermore, sports and cultural and leisure activities have been commonly studied in relation to well-being levels [7]. As these types of activities provide opportunities to meet life values and needs [8], can improve health [9] and establish important social connections [10,11], evidence suggests that engagement in sports, culture and leisure has a positive and causal relationship with well-being [12,13,14,15,16,17].

The infrastructure for sports, culture and leisure activities plays a critical role in facilitating and promoting these positive outcomes. Smart, accessible and affordable infrastructure and facilities engage people in leisure activities and through that enhance the quality of life [18]. Because infrastructure depends a lot on public investment, the benefits of engagement in sports, culture and leisure are used to justify government spending [17,19,20]. Surprisingly, only a few attempts have been made to assign a monetary value to these benefits [21]. This valuation, turning effect measures into monetary units, could be salient, as it provides policy makers with a means to improve the decision-making process of allocating scarce public resources. For this reason, concrete evidence can help to create sustainable policies for leisure activities and support government funding over the years, especially in times of austerity and public spending cuts.

A novel and promising approach to valuating the benefits of non-market goods, such as sports, culture and leisure, is the compensating variation approach (CVA), sometimes referred to as the shadow pricing method or well-being valuation (WV). In this method, the contributions of a non-market good can be assessed and monetarized by the amount of money that is needed to remain at the same level of well-being as when the goods are absent [22]. So, the marginal rate of substitution (MRS) is the ratio of the effect that the non-market good has on well-being (βq) to the effect of an increase in income on well-being (βm) [23].

The CVA approach is a more sophisticated way to identify the value of the non-market goods than deriving it from hypothetical situations or market transactions in proxy markets [24]. The technique does not rely on any direct forecasting or evaluation, as individuals are not asked to value the non-market good directly and therefore it does not evoke strategic or desirable answers [16,25].

An important limitation of the CVA is that the method requires a reliable research design for estimating the causal effect of the non-market good on well-being, as well as the effect of a unit increase in income on well-being. Without that, CVA can result in very biased outcomes [23]. Because governments rely on those results to justify public spending and make decisions for evidence-based and sustainable policies, it is important to obtain accurate results. More crucially, researchers, experts and consultants in the field of leisure should be aware of the potential bias that can occur when using specific research designs, so that they are careful when making certain claims on public budgets. Therefore, to have a better understanding of the research designs using the CVA approach, this scoping review has been carried out. The research question is: What are the methodological issues encountered when using the compensation valuation approach for estimating the value of sports, culture and leisure activities?

To our knowledge, there has been just one review that looked at the monetary value of sports activities. Orlowski and Wicker performed a rapid evidence assessment of peer-reviewed studies on monetary valuation in sports research [21]. They identified eight different valuation approaches, including the CVA method. The current review focused exclusively on the CVA method and included more leisure-related activities, such as cultural activities, because of the limited evidence that was found in the previous review. Moreover, this study also included non-peer-reviewed reports to give a broad overview of the state of research, to look at the techniques that have been applied and to compare them with peer-reviewed studies. Combining academic and consulting perspectives thereby facilitating the advancement of research in this field.

2. Materials and Methods

To evaluate the current state of the literature, a scoping review was performed. Scoping literature reviews can provide a preliminary assessment of the potential size and scope of available research literature and identify the nature of existing research evidence. Moreover, scoping reviews can identify gaps in the existing literature [26]. Scoping reviews also allow for the inclusion of grey literature, which are materials and research produced by organizations outside of traditional academic publishing. The inclusion of grey literature helps to identify potential biases and provides a better overview of the state of research. All studies that used the compensating variation approach or well-being valuation for any leisure, sports or cultural activity were considered for this review.

The review was not registered beforehand but registries, such as Open Science Framework were scanned for potential projects on this subject. At the time of the search, none were found.

2.1. Search Strategy

The literature search was conducted conforming to the preferred reporting items for systematic reviews and meta-analyses [27]. The search was carried out in two steps: First, peer-reviewed articles were identified using the Gale Academic Onefile, Web of Sciences, Business Source Ultimate, Medline, SPORTDiscus, JSTOR and ScienceDirect databases. Second, grey literature articles were identified using Google Scholar. Both steps were conducted in September 2022. The search strategy used the following keywords and phrases in the title, abstract or subject: (“compensation variation” OR “well-being valuation”) AND (“sport OR “physical activity” OR “culture” OR “leisure” OR “recreation”). For scientific databases, truncation was used to broaden the search to include various word endings and spellings.

2.2. Inclusion and Exclusion Criteria

All 747 records from Google Scholar were imported and combined with the records from the scientific databases using an automation tool. Following the removal of duplicates, four studies were removed because they were not written in English and three studies could not be retrieved. The remaining 814 eligible studies were scanned by title and abstract, by one researcher, for the following inclusion criteria in order to narrow the results:

• Included studies must have any sports, cultural or leisure activity included as a variable in the research,
• Included studies must have used CVA as empiric design,
• Included studies must contain new empiric evidence.

Studies about volunteering were excluded because some versions of volunteering occur outside the leisure domain, for instance caregiving, and have an economic component to the work being carried out.

This approach resulted in eight papers that matched the inclusion criteria. Three additional studies were identified through a manual search of the reference lists of the included studies in October 2022. The search strategy is summarized in Figure 1.

2.3. Data Extraction and Synthesis

To analyze the results, all studies were summarized by year, country, scope and source. Next, research methods were analyzed by database, sample, SWB measures and design, and addressing common problems, such as endogeneity bias and the selection bias. Further analysis was carried out by comparing the different control variables. Finally, the outcomes of the studies were presented as the effect coefficient on SWB (βq) and the compensating monetary value (MRS). Only statistically significant variables (p < 0.05) were included. The monetary value was converted to euro EUR(EUR) using historic exchange rates of the year the study was published using the website Xe.com. No correction for time effects were made.

Owing to a lack of standardization, the sports, culture and leisure variables were measured in different ways. In order to classify and compare, the measured variables were categorized in binary engagement variables on a weekly, monthly or yearly basis. Because some studies used other variables, such as frequency or intensity, those were classified as ‘other’.

3. Results

The eleven included studies were published between 2010 and 2022 [28,29,30,31,32,33,34,35,36,37,38] (

Table 1. Summary of the included studies.

#	Authors, Year	Country	Scope	Source:
				Peer- Reviewed	Commissioned Report
[28]	Marsh et al., 2010	England	Sports and Culture	No	Yes
[29]	Downward and Rasciute, 2011	England	Sports Participation	Yes	No
[30]	Fujiwara, 2013	England	Museums	No	Yes
[31]	Fujiwara et al., 2014	England	Sports and Culture	No	Yes
[32]	Downward and Dawson, 2016	England	Sports	Yes	No
[33]	Orlowski and Wicker, 2018	Germany	Sports Participation	Yes	No
[34]	Fujiwara, 2018	England	HACT Social Value Bank	No	Yes
[35]	Lemyre et al., 2018	Canada	Social activities	No	No
[36]	Simetrica and Jacobs, 2018	New Zealand	Sports	No	Yes
[37]	Shi et al., 2019	Canada	Health behavior	Yes	No
[38]	Thormann et al., 2022	Germany	Sport participation	Yes	No

). The first five originated from England [28,29,30,31,32]. Starting in 2017, results from Germany [33,34,35,36,37,38], Canada [35,37] and New Zealand [36] were also published. Five studies were published in a peer-reviewed academic journal [29,32,33,37,38]. Lemyre et al. published their results at a statistics symposium [35]. Other studies were commissioned reports from private consulting firms delivered by academic researchers [28,30,31,34,36].

3.1. Datasets and Design

All studies used a large cohort dataset with more than 10,000 observations (

Table 2. Datasets and designs of the included studies.

#	Database	n	SWB Measures	Correction for Income or Selection Bias
				Design	Other
[28]	British Household panel survey	>10,000	Life Satisfaction (1–7)	Observational
[29]	Taking Part survey	28,117	Happiness (1–10)	Observational	HTOP for selection bias
[30]	Taking Part survey	>100,000	Happiness (1–10)	Observational	Income coefficient from previous study
[31]	Understanding Society survey	>40,000	Life Satisfaction (1–7)	Observational	Income coefficient from previous study
[32]	Taking Part survey	14,913	Happiness (1–10)	Instrumental variable for selection bias
[33]	German socio-economic panel	191,828	Life Satisfaction (0–10)	Instrumental variable for selection bias
[34]	Taking Part survey	>100,000	Life Satisfaction (1–7)	Instrumental variable for income bias
[35]	General social survey	15,390	Life satisfaction (0–10)	Instrumental variable for income bias
[36]	Active NZ survey	52,188	Life Satisfaction (0–10)	Observational	Income coefficient manually adapted by previous study
[37]	Canadian Community Health survey	74,577	Life Satisfaction (0–10)	Observational	Mediation analysis for income coefficient
[38]	German socio-economic panel	30,861	Life Satisfaction (0–10)	Instrumental variable for selection bias

Note: HTOP is thresholds ordered probit model.

). Four studies did not specify how many respondents were included [28,30,31,34]. Whereas most datasets were used once, the Taking Part survey in England was used four times [29,30,32,34] and the German Socio-Economic Panel was used twice [33,38]. Six studies pooled multiple years of cross-sectional datasets to obtain a bigger sample [29,30,31,33,34,36]. Although all datasets aimed to be a representative sample of the whole national population and some were published by national statistics organizations, no details were provided about possible sampling biases.

The SWB measurement differed greatly. The 11-point life satisfaction scale was used most, but only in five studies [33,35,36,37,38]. This was phrased as “All things considered, how satisfied are you with your life as a whole?”. The 7-point life satisfaction scale was used thrice [28,31,34] and was formulated as “How dissatisfied or satisfied are you with your life overall?”. The 10-point happiness scale was also used in three studies [29,30,32] and was worded as “On a scale from 1 to 10, can you indicate to what extent you consider yourself to be a happy person?” or “Taking all things together, how happy would you say you are?”.

Six studies used an observational design in which a regression analysis accounted for observable heterogeneity (Table 2). To address some unobserved heterogeneity, Downward and Rasciute used a thresholds ordered probit (HTOP) model [29]. This kind of heterogeneous model allows for different respondents to have different coefficients, as some discrete variables have a certain ordering based on other observable characteristics of individuals. Another way of dealing with unobserved heterogeneity is the use of an instrumental variable design. Three studies used the availability of facilities or clubs as instruments to estimate the causal interference between sports and exercise and SWB [32,33,38]. This was carried out because potential unobservables could select individuals to engage in that kind of activity and the effect could also be reversed, as happy people might engage more (selection effect). Thormann et al. compared the instrumental approach with standard regression and “seemingly unrelated regression” and found that the effect of sports on SWB was ten times bigger when they corrected for this selection effect [38].

However, as these three studies might give a more accurate estimate of the causal effect of sports, culture and leisure activities than other studies, another endogeneity issue was not addressed. The CVA technique also needs the estimated income effect on SWB to calculate the monetary value. That effect may be biased as well. Two studies used a different instrumental variable approach to overcome the specific endogeneity bias related to income. By exploiting the randomness of lottery wins [34] or spouse income [35] as an instrument, these two studies showed that the income coefficient was much larger than predicted in standard linear regression. Fujiwara used the control function approach in a 3-stage well-being model to make sure the estimations for the effect of income and the effect of leisure participation on SWB were unbiased [34]. The control function allows one to derive estimates of the sample average partial effect (APE) for income for anyone in the sample, instead of the local average treatment effect (LATE), for only a subsample, resulting from a standard instrumental approach.

Four studies controlled for the endogeneity of income in a different way. Shi et al. performed a mediation analysis by removing each control variable from the full model, one by one [37]. Three studies addressed the endogeneity of the effect of a unit increase in income on SWB by using the coefficient of a previous study that exploited lottery wins as an instrument. For two studies this was achieved with the same national population [30,31], but for the study conducted by Simentrica and Jacobs in New Zealand, the coefficient of the income effect came from a different country (England) and was manually adapted to reflect the country-specific effect of income on SWB [36].

3.2. Control Variables

Marsh et al. did not specify which control variables were used to statistically control for differences between respondents [28]. To control for confounding, all other studies have used at least gender and employment (

Table 3. Control variables of the included studies.

	Studies
Control variables	[28]	[29]	[30]	[31]	[32]	[33]	[34]	[35]	[36]	[37]	[38]
Gender	NA	X	X	X	X	X	X	X	X	X	X
Employment	NA	X	X	X	X	X	X	X	X	X	X
Geographical area	NA	X	X	X	X	X	X	X	X	X
Education	NA	X	X		X	X	X	X	X	X	X
Marriage	NA	X	X	X	X	X	X	X		X	X
Housing	NA	X		X	X	X	X	X	X	X
Age	NA	X		X	X	X	X	X	X		X
Children	NA	X	X	X	X	X	X		X		X
Health	NA	X	X	X		X		X		X
Ethnicity	NA	X	X		X				X
Social connections			X	X			X	X
Religion	NA		X
Personality traits	NA	D,S,V	S,D	V,C,I	V	*	C		S	S

Note: Ref. [28] did not specify what control variables were used (NA). Personality traits were smoking (S), drinking (D), volunteering (V), caring duties (C), insurance (I). * Ref. [33] used panel data to avoid individual-specific personality traits.

). Education, marriage and housing variables were used in eight of ten studies. Religion, as well as most lifestyle factors, was controlled for only in one or two studies. Interestingly, in four studies [32,34,36,38], health was deliberately omitted because of the indirect effect on the sports and physical activity variables. This decision was made because physical activity can improve health and, through that, SWB. Therefore, the researchers claim that controlling for health would lead to an underestimation of the true contribution of engagement in sports to SWB. Orlowski and Wicker included individual-specific time-invariant effects from the panel data to avoid bias caused by unobservable personality traits or individual characteristics [33].

3.3. Outcomes

The eleven studies had 65 different variables of engagement in sports, culture and leisure activities. Thirteen measured engagements (yes/no) on a weekly basis, four measured them on a monthly basis and twenty-four measured them on a yearly basis. Six studies had also twenty-four other variables, for example: engagement in categories (intensity or frequencies) or disaggregated results to identify whether the form of participation affects well-being.

The twenty-four variables measuring engagement (yes/no) on a yearly basis, ranged between 0.24 (scale 0–10) for participating in any culture activity at least once in the past year to even negative coefficients for playing musical instrument/writing music and participation in fitness (−0.06; scale 1–7). As for the monetary value (MRS), outcomes for engagement at least once a year ranged between EUR 21,307 for participating in any sports at least once in the last 12 months and EUR 1606 for participating in fitness at least once in the past 12 months. A full list of all variables, coefficients and monetary values can be found in the online Supplementary Material (see Table S1).

The design of the study seems to relate with the outcomes (see

Table 4. Range of effect coefficients (βq) and monetary values (MRS).

#	Scope	Correction for Bias	Range βq (Scale)	Range MRS in EUR per Year	Note
[28]	Sports and Culture	None	0.019–0.025 (1–7)	EUR 10,341–12,639
[29]	Sports participation	HTOP for selection bias	0.0312–0.0374 (1–10)	EUR 21,307–26,183	(1)
[30]	Museums	Income coefficient from previous study	0.039–0.088 (1–10)	EUR 1771–3811
[31]	Sports and Culture	Income coefficient from previous study	−0.058–0.078 (1–7)	−EUR 1.606–EUR 2036
[32]	Sports	Instrumental variable for selection bias	0.191–0.247 (1–10)	EUR 47,592–60,760	(2)
[33]	Sports participation	Instrumental variable for selection bias	−0.079–0.206 (0–10)	−EUR 5652–EUR 14,232	(3)
[34]	HACT social value bank	Instrumental variable for income bias	NA	EUR 1601–5991	(4)
[35]	Social activities	Instrumental variable for income bias	0.080–0.243 (0–10)	EUR 1746–4948
[36]	Sports	Income coefficient manually adapted from previous study	NA	EUR 900–3319
[37]	Health behavior	Mediation analysis for income coefficient	0.08–0.14 (0–10)	EUR 14,026–21,745	(3)
[38]	Sport participation	Instrumental variable for selection bias	1543 (0–10)	EUR 20,687	(3)(5)

Note: (1) The βq and MRS was calculated by multiplying the outcome per minute by the average number of minutes of the whole sample (334 min for low intensity; 29 min for meeting health guidelines). (2) The βq and MRS was calculated by multiplying the outcome per minute by the average number of minutes of the whole sample (412.88). (3) The monetary values are multiplied by 12 (months) or 52 (weeks) to indicate the monetary value over one year. (4) Outcome values are obtained from community investment value calculations from the social value bank by HACT and Fujiwara. (5) The βq and MRS was calculated by multiplying the outcome per hour by the average number of sport hours of the whole sample (17.73).

). Studies that addressed the selection effect by using an instrumental variable had larger coefficients than other studies and resulted in higher monetary values [32,33,38]. Ref. [33] had a wide range from small to large coefficients because the variables measured different intensity categories in which less engagement resulted in negative estimates and monetary values. Other studies that addressed the endogeneity of income by using an instrument or coefficient from previous studies [30,31,34,35] were similar to the studies that had no instrument [28,29,37], but the monetary values were much lower.

To illustrate the methodological issue of endogeneity,

Table 5. Participation in sports or physical activity at least once a week or once a month.

Study	Leisure Variable	βq (Scale)	EUR Currency per Year	Country	SWB Measures	Correction for Bias
[28]	Practicing sports at least once a week	0.025 (1–7)	EUR 12.639	England	Life Satisfaction (1–7)	None
[29]	Participated in any sports in the last four weeks	0.0374 (1–10)	EUR 26.183	England	Happiness (1–10)	HTOP for selection bias
[30]	Practiced sports or physical activity in the last four weeks	0.04 (1–10)	EUR 1.816	England	Happiness (1–10)	Income coefficient from previous study
[32]	Participated in sports activity in the last four weeks prior to the interview (1)	0.231 (1–10)	EUR 45.420	England	Happiness (1–10)	Instrumental variable for selection bias
[34]	Participated in moderate exercise that raised your heart rate and resulted in sweat at least once a week	NA	EUR 4.741	England	Life Satisfaction (1–7)	Instrumental variable for income bias
[35]	Participated in any sports once a week	0.160 (0–10)	EUR 3.374	Canada	Life Satisfaction (0–10)	Instrumental variable for income bias
[38]	Participated in sports at least once a week (2) (3)	1.543 (0–10)	EUR 20.687	Germany	Life Satisfaction (0–10)	Instrumental variable for selection bias

Note: (1) The βq and MRS was calculated by multiplying the outcome per minute by the average number of minutes of the whole sample (412.88). (2) The monetary values are multiplied by 12 (months) to indicate the monetary value over one year. (3) The βq and MRS was calculated by multiplying the outcome per hour by the average number of sport hours of the whole sample (17.73).

shows seven studies that all measured participation in sports or physical activity at least once a week or once in the past four weeks. Refs. [28,29,30] estimated a small effect coefficient. Nevertheless, the monetary value was very different in each study. Using an income coefficient from a previous study lowers the monetary value to a great degree. Furthermore, other studies that used an instrumental variable for the endogeneity of income had lower monetary values [34,35]. Studies that addressed the selection bias by using an instrumental variable approach had larger effect coefficients and higher monetary values than those resembling variables [32,38].

4. Discussion

The field of well-being valuation for sports, culture and leisure is still young and developing, but with just eleven studies between 2010 and 2022, it is doing so at a relatively slow pace. It could be that the scientific limitations of the CVA method are hard to overcome and that obtaining the required data to address possible biases is challenging. However, the sheer volume of studies in the grey literature indicates the need for valuation studies to justify subsidizing these kinds of leisure activities.

As observational datasets are not in the best position to estimate causal effects, different approaches were undertaken to avoid methodological issues. The endogeneity issue seems to be most pressing. This is also pointed out by Lemyre et al. [35] and Fujiwara [24]. They argue that the monetary values will decrease substantially if the causal effect of income on SWB is fully accounted for. The changes in magnitudes arise because the monetary value of a non-market good (MRS) decreases when the effects of income rise [23]. The peer-reviewed studies did not address the endogeneity biases of income but instead focused on the selection effects. So, this scoping review highlights the importance of looking beyond the academic literature to learn that the endogeneity issue of income has large consequences for the outcomes.

Moreover, not controlling for the selection effect, as is observed in most grey literature, led to an underestimation of the effect of engagement in sports, culture and leisure on SWB. This is because the effect coefficient in the numerator (βq) seems to be much higher in studies that used an instrumental variable approach to control for the selection bias [38]. This is consistent with the results from Forrest and McHale, who suspect a disproportional number of individuals that engage in sports and exercise have unobserved characteristics that negatively influence SWB [17].

This review also showed that there are big differences in the control observables being used. Interestingly, in three studies, health was deliberately omitted because of the indirect effect on sports and physical activity variables. Indeed, when health is not a control variable, and therefore the indirect effect through health is included in the coefficient, the value of engagement in sports and physical activity will be larger than many other leisure activities in the same study. However, as healthy people are more likely to participate in sports, this approach also may be subject to a selection bias [16].

It is not clear to what extent the different SWB measures influenced the results, because of the various methods of conducting research. However, there is a vivid debate concerning which measurement captures the people’s well-being best [16]. As happiness scales survey people’s affective state and try to gauge people’s moods at the very moment, they are more related to engagement in sports, culture and leisure [7]. On the contrary, SWB measures an evaluative judgement, such as life satisfaction, which is more closely linked to income [6]. Therefore, the SWB measures can have an impact on the outcomes.

Limitations and Future Research Directions

The strength of this scoping review is that it included both peer-reviewed and grey literature databases. In this way, by carrying out an in-depth synthesis of the research designs and outcomes, it was possible to determine differences between studies and learn from their principles. However, the wide range of research approaches to avoid biases made it difficult to compare outcomes and fully assess the methodological issues. For example, it is not clear to what extent the different SWB measures influenced the results. Furthermore, a lot of grey literature did not provide all estimates or covariates for a comparison of outcomes. Finally, this review only included grey literature results from Google Scholar and records were only scanned by one researcher. However, because of the limited evidence on this topic, many studies referred to previous studies and, when missing, those were manually added to the review using the citation search.

For researchers in the leisure field, this review is a call to address important biases in order to obtain a better consensus regarding the value of sports, culture and leisure for public policies. To date, not one study has controlled for both the endogeneity of the effect of a unit increase in income on SWB and users’ selection of engagement in sports, culture and leisure. Future research should use two separate instruments in a three-stage regression to account for these two common biases [24]. Additionally, there should be scientific effort to identify the true causal effects using different approaches, such as propensity score matching techniques, difference-in-difference methods, regression discontinuity design or semi-experimental settings. This would be worthwhile, as the outcomes can increase the accountability of future public investment in many non-market goods and help to increase government quality by creating sustainable policies that are evidence-based [21]. It is important to see if the results from these different techniques are robust for real-world policy and decision-making. Moreover, the role of smart, accessible and affordable infrastructures and facilities in SWB levels should be further explored. For example, innovation in data and technology to improve the occupancy rate of public infrastructure can benefit society if this also affects subjective well-being.

Lastly, when observational datasets are used in future research, it would be good to develop a set of minimum control variables that need to be considered when measuring and reporting about the effect of sports, culture and leisure activities on well-being. The twelve different control variables found in eleven studies could be the starting point for future studies. Health could be omitted if the research design addressed the selection bias, for example by using an instrumental variable approach. These efforts would increase the opportunity to compare different outcomes and reach a consensus about the value of sports, culture and leisure activities for society.

5. Conclusions

The aim of this scoping review was to identify the methodological issues encountered when using the compensation valuation approach for estimating the value of leisure activities. The diverse use of control and well-being variables, as well as endogeneity and selection biases, were important methodological issues that led to a wide range of monetary values. Because of the variability in their conduct, methodological standardization is required to reach consensus on the contribution of these types of activities to society. It is recommended that future studies at least address these prevalent biases. This can be carried out, but not limited to, by employing a three-stage regression utilizing two separate instruments and including the twelve control variables that have been found in previous studies. These efforts help to create sustainable policies for public funding in sports, culture and leisure activities that can endure austerity measures or, even better, help to promote our nations’ well-being levels.