Welfare effect of substitute sites for coastal recreation – evidence from the Baltic Sea

Figures & data

Figure 1. The data structure.

Table 1. Descriptive statistics for the variables in the logit and negbin models.

		All countries (n = 2642)	Finland (n = 1257)	Germany (n = 827)	Latvia (n = 558)
Dependent variables	Description	Mean (Std.dev)	Mean (Std.dev)	Mean (Std.dev)	Mean (Std.dev)
Substitute	1 =  Perceived substitute site exists, 0 = No perceived substitute site	0.43 (0.50)	0.47 (0.50)	0.52 (0.50)	0.20 (0.40)
Trips to the Baltic Sea per year	Number of annual recreation visits, 1–365	13.10 (42.51)	16.85 (47.06)	10.93 (47.26)	7.32 (12.63)
Independent variables
Travel cost (TC)	Stated two-way travel costs in PPP-adjusted EUR, 0–1000	72.93 (119.02)	57.88 (104.60)	117.32 (150.98)	35.31 (51.17)
TC*Substitute	Interaction term: Travel cost (TC) * Substitute in PPP-adjusted EUR, 0–1000	32.80 (93.24)	26.52 (81.72)	58.22 (123.47)	6.57 (31.73)
Distance	Euclidean distance from home to the Baltic Sea coast in 100 kilometres, 0–7.33	1.14 (1.42)	0.57 (0.73)	2.09 (1.77)	0.60 (0.66)
Income	Monthly net income in 1000 PPP-adjusted EUR, 50–5000	1.63 (1.06)	1.55 (0.90)	2.28 (1.12)	0.83 (0.59)
Age (negbin model)	In years, 18–79	46.13 (15.37)	45.70 (17.19)	48.49 (11.87)	45.70 (17.19)
Middle aged (logit model)	1 = 30–64 years, 0 = 18–29 years and over 65	0.68 (0.47)	0.56 (0.50)	0.88 (0.33)	0.66 (0.47)
High education	1 = University or polytechnic, 0 = Lower education	0.35 (0.48)	0.42 (0.49)	0.33 (0.47)	0.24 (0.43)
Number of facilities	1 = Many at the most often visited site, 0 = None/some	0.45 (0.50)	0.40 (0.49)	0.65 (0.48)	0.26 (0.44)
Number of activities	1 = Two or more activities typically participated to at the most often visited site, 0 = One activity	0.74 (0.44)	0.62 (0.48)	0.85 (0.36)	0.62 (0.48)
Water contact	1 = Respondent engages in swimming or diving at the most often visited site, 0 = Does not engage in swimming or diving	0.33 (0.47)	0.16 (0.37)	0.39 (0.49)	0.63 (0.48)
Recreation/CES	1 = Recreation is an important cultural ecosystem service (26–100 points out of total 100), 0 = 0–25 points	0.54 (0.50)	0.53 (0.50)	0.42 (0.49)	0.75 (0.43)
Only purpose recreation	1 = Recreation is the only purpose of the visit, 0 = Recreation is not the only purpose	0.43 (0.49)	0.37 (0.48)	0.38 (0.49)	0.62 (0.49)
Recent visit	1 = Last visit made within the last 12 months, 0 = Over 12 months ago	0.76 (0.43)	0.81 (0.40)	0.68 (0.47)	0.78 (0.41)

Table 2. Logistic regression model for the existence of substitute sites (dependent variable = 1 if the respondent had a substitute site for Baltic Sea recreation).

Variable	Coefficient	p-value
Constant	−0.830***	0.000
Distance	−0.108**	0.018
Recent visit	0.299***	0.006
Number of activities	0.491***	0.000
Number of facilities	0.200**	0.028
Water contact	−0.178*	0.087
Middle aged	0.339***	0.001
Education	0.513***	0.000
Finland	−0.211*	0.072
Latvia	−1.467***	0.000
Number of observations	2417
Log likelihood	−1523.96
Restricted log likelihood (constant only)	−1656.80
Log-likelihood ratio	265.67	<0.001
Wald test	224.19	<0.001
Pseudo $R^2$	0.0802
Bayesian information criterion	3125.82
Hosmer-Lemeshow test	35.59	<0.001
Overall predictive capability %	62.6

Table 3. Model 1 - TC model without substitutes (dependent variable = number of recreation visits to the Baltic Sea per year).

	Germany		Finland		Latvia
Variable	Coefficient	Z value	Coefficient	Z value	Coefficient	Z value
TC	−0.012***	(17.92)	−0.013***	(11.72)	−0.015***	(5.45)
Income	0.222***	(4.21)	0.183***	(3.33)	0.323***	(3.57)
Age	0.015***	(3.17)	0.002	(0.65)	−0.004	(1.18)
Recreation/CES	−0.007	(0.06)	0.445***	(4.93)	0.200*	(1.74)
Number of activities	0.519***	(3.23)	0.662***	(6.75)	0.448***	(2.93)
Only purpose recreation	−0.493***	(4.26)	0.154	(1.60)	−0.243**	(2.21)
Water contact	0.295**	(2.42)	0.048	(0.37)	0.399***	(3.52)
Constant	1.224***	(3.92)	1.728***	(11.74)	1.391***	(5.56)
ln(alpha)	0.426***	(8.02)	0.453***	(10.32)	−0.170**	(2.28)
Log likelihood	−1872.1785		−2940.9948		−1237.9547
McFadden's Pseudo $R^2$	0.0790		0.0372		0.0283
N	647		879		426

*p < 0.05,

  **p < 0.01,

  ***p < 0.001

Table 4. Model 2 - TC model with substitute dummy (dependent variable = number of recreation visits to the Baltic Sea per year).

	Germany		Finland		Latvia
Variable	Coefficient	Z value	Coefficient	Z value	Coefficient	Z value
TC	−0.012***	(17.72)	−0.013***	(11.93)	−0.015***	(5.54)
Substitute	−0.288***	(2.68)	−0.202**	(2.22)	0.420***	(3.62)
Income	0.234***	(4.43)	0.191***	(3.47)	0.276***	(3.06)
Age	0.017***	(3.52)	0.001	(0.45)	−0.002	(0.71)
Recreation/CES	0.003	(0.02)	0.418***	(4.59)	0.204*	(1.80)
Number of activities	0.490***	(3.07)	0.688***	(6.98)	0.410***	(2.71)
Only purpose recreation	−0.480***	(4.17)	0.186*	(1.92)	−0.248**	(2.31)
Water contact	0.288**	(2.37)	0.013	(0.10)	0.421***	(3.80)
Constant	1.266***	(4.06)	1.835***	(11.79)	1.272***	(5.13)
ln(alpha)	0.417***	(7.82)	0.448***	(10.21)	−0.203***	(2.70)
Log likelihood	−1868.5974		−2938.5456		−1231.1678
McFadden's Pseudo $R^2$	0.0807		0.0380		0.0336
N	647		879		426
LR test between models 1 and 2	7.16***		4.90**		13.57***

*p < 0.05,

  **p < 0.01,

  ***p < 0.001

Table 5. Model 3 - TC model with substitute dummy and travel cost - substitute interaction (TC*Substitute) (dependent variable = number of recreation visits to the Baltic Sea per year).

	Germany		Finland		Latvia
Variable	Coefficient	Z value	Coefficient	Z value	Coefficient	Z value
TC	−0.013***	(14.02)	−0.011***	(7.61)	−0.013***	(4.17)
Substitute	−0.485***	(3.49)	−0.075	(0.64)	0.652***	(3.79)
TC*Substitute	0.003**	(2.25)	−0.004*	(1.70)	−0.012*	(1.89)
Income	0.226***	(4.28)	0.186***	(3.40)	0.282***	(3.13)
Age	0.016***	(3.47)	0.002	(0.66)	−0.003	(0.78)
Recreation/CES	0.010	(0.09)	0.422***	(4.63)	0.203*	(1.80)
Number of activities	0.435***	(2.69)	0.673***	(6.81)	0.387**	(2.55)
Only purpose recreation	−0.474***	(4.14)	0.160	(1.63)	−0.250**	(2.33)
Water contact	0.283**	(2.34)	0.038	(0.29)	0.423***	(3.82)
Constant	1.443***	(4.48)	1.762***	(10.99)	1.255***	(5.07)
ln(alpha)	0.409***	(7.66)	0.445***	(10.13)	−0.210***	(2.79)
Log likelihood	−1866.0785		−2937.1068		−1229.4301
McFadden's Pseudo $R^2$	0.0820		0.0385		0.0350
N	647		879		426
LR test between models 1 and 3	12.20***		7.78**		17.05***
LR test between models 2 and 3	5.04**		2.88*		3.48*

*p < 0.05,

  **p < 0.01,

  ***p < 0.001

Table 6. Consumer surplus estimates and predicted number of visits for recreation on the Baltic Sea coast (CS estimates are in PPP adjusted 2017 euros).

	Germany	Finland	Latvia
Model 1 (TC)
CS per visit (€) [95% CI]	83.3 [74.2, 92.4]	79.5 [66.2, 92.9]	66.9 [42.9, 91.0]
Predicted number of visits per year [95% CI]	8.7 [7.7, 9.7]	13.2 [11.9, 14.4]	6.9 [6.2, 7.6]
CS per person per year (€)	725	1049	462
Model 2 (TC, Substitute dummy)
CS per visit (€) [95% CI]	84.0 [74.7, 93.3]	78.0 [65.2, 90.8]	66.6 [43.1, 90.2]
Predicted number of visits per year, no substitute [95% CI]	10.1 [8.5, 11.8]	14.5 [12.6, 16.5]	6.1 [5.4, 6.8]
Predicted number of visits per year, has substitute [95% CI]	7.6 [6.5, 8.7]	11.9 [10.3, 13.4]	9.3 [7.5, 11.2]
CS per person per year (€), no substitute	848	1131	406
CS per person per year (€), has substitute	638	928	619
Model 3 (TC, Substitute dummy, TC*Substitute)
CS per visit (€), no substitute [95% CI]	74.4 [64.0, 84.8]	89.5 [66.5, 112.6]	79.4 [42.1, 116.6]
CS per visit (€), has substitute [95% CI]	95.6 [78.9, 112.4]	67.3 [53.3, 81.4]	41.1 [22.7, 59.6]
Predicted number of visits per year, no substitute [95% CI]	11.3 [9.1, 13.4]	13.6 [11.5, 15.6]	5.8 [5.1, 6.6]
Predicted number of visits per year, has substitute [95% CI]	6.9 [5.8, 8.1]	12.6^a [10.7, 14.5]	11.2 [8.0, 14.4]
CS per person per year (€), no substitute	841	1217	461
CS per person per year (€), has substitute	660	848^a	485

^a Calculated based on the variable Substitute, which is not significant in the model.

  Confidence intervals for the per-trip consumer surplus are calculated using the delta method in Stata.