Federal program since 2022 for private and municipal forest owners

• Flat rate of 85–100 €/ha/yr (forests ≤ 100 ha) for a package of 11–12 management criteria, e.g. no clear-cuts, ≥ 5 habitat trees/ha, deadwood, natural regeneration
• Whole forest, 10–20 years, certification required
• About 1.7 M ha enrolled: 60 % municipal, 40 % private (12 % of total private area)
• Budget rises to 200 M€/yr from 2027 (target ≥ 2 M ha)

What KWM does not tell us

• It pays for a bundle of practices, not for individual services
• It is tax-financed: we do not know what households would pay, and for which service
• It offers one rate: we do not know what small private owners need per service
• Recreation and timber are not part of the scheme

• R1: Does the population’s demand for forest ecosystem services match the supply offered by private forest owners?
  • H1: The aggregate willingness to pay of households covers the compensation demanded by private forest owners (for each forest ecosystem service for which owners require compensation).
• R2: Do the preferences of the general population and of private forest owners for forest ecosystem services differ?
  • H2a/b: The general population has a higher willingness to pay for habitat trees (a) and recreation infrastructure (b) than forest owners.
  • H2c: The general population has a higher willingness to pay for a reduction in timber use than forest owners.
• R3: Which socio-demographic and structural factors explain differences in preferences for forest ecosystem services?
  • H3: Private forest owners have different preferences than individuals from the general population with a similar socio-demographic and structural background.

• Vote (all): conditional and mixed logit → preferences, WTP, MRS
  • Pooled model with owner interactions; split-sample models
  • Timber as change vs. status quo, separate coefficients for increase and decrease
  • Habitat trees as increase Δ over the state status quo (program minimum ≥ 5)
  • Two models: A = owners + control group (H2, H3); B = follow-up group (demand)
• Enrollment (owners): two-part model
  • Part 1, join? Conditional and mixed logit → WTA
  • Part 2, how much? Share of own forest enrolled, fractional logit
  • Habitat trees as gap between program minimum and own baseline; KWM participation and coverage as interactions
  • Supply curve: part 1 × part 2, predicted for all owners and payments
• Matching (H3): owners vs. comparable non-owners (PSM / ML) → split-sample mixed logit
• Design: Bayesian efficient design for binary choices incl. payment \(p\); power simulation

Vote (all respondents)

\[ U_{nt} = \alpha + \boldsymbol{\beta}_n' X_t + \beta_{cv} \log cv_{nt} + \kappa\, FU_{nt} + (\beta_{levy} + \tau_{levy} D_n)\, levy_t + \varepsilon_{nt}, \qquad U^{SQ}_{nt} = \varepsilon^{SQ}_{nt} \] \[ \boldsymbol{\beta}_n = \boldsymbol{\beta} + \boldsymbol{\tau} D_n + \boldsymbol{\rho} \log cv_{nt} + \boldsymbol{\eta}_n \]

Model A (owners + control group): main votes, \(cv_n\) as ASC interaction. Model B (follow-up group): main votes + follow-ups (\(FU_{nt}=1\)).

Enrollment, part 1: join? \[ U^{J}_{nt} = \alpha^{J} + \boldsymbol{\gamma}_n' \tilde X_{nt} + \delta\, p_t + \omega \log cv_n + \varepsilon^{J}_{nt}, \quad U^{no}_{nt} = \varepsilon^{no}_{nt}, \quad \boldsymbol{\gamma}_n = \boldsymbol{\gamma} + \boldsymbol{\lambda}\, KWM_n + \boldsymbol{\nu}_n \]

Enrollment, part 2: how much? (if joined) \[ s_{nt} = \Lambda\big(\theta_0 + \boldsymbol{\theta}' \tilde X_{nt} + \theta_p\, p_t + \theta_{cv} \log cv_n + \boldsymbol{\varphi}' z_n\big) + u_{nt} \]

For a program \(X\) and a payment \(p\):

1. Join: predict for every owner \(n\) the probability of joining, \(P_n(p)\), from the mixed logit (own baseline, KWM status, averaged over draws)
2. Share: predict the share of their forest that owner \(n\) would enroll if joining, \(s_n(p)\), from the fractional logit
3. Hectares: expected enrolled area of owner \(n\) \(=\; P_n(p) \cdot s_n(p) \cdot area_n\)
4. Scale up: sum over owners and scale to the private forest area \(A_{PF}\) \[S(p) = A_{PF}\cdot \frac{\sum_n w_n\, P_n(p)\, s_n(p)\, area_n}{\sum_n w_n\, area_n}\]
5. Curve: repeat for \(p = 25, 50, \dots, 400\) €/ha → \(S(p)\); confidence bands by Krinsky–Robb
6. Inverse: \(p^*(A)\) = payment needed to enroll area \(A\) → compared with the demand curve (next slides)

H1: public demand covers owners’ compensation, per attribute \(X\) with \(WTA_X > 0\), per federal state

\[ \underbrace{WTP_X(cv) \cdot N_{HH}}_{\text{benefit, €/yr}} \;\geq\; \underbrace{\big[p^*(X{+}1) - p^*(X)\big]\cdot cv\,A_{PF}}_{\text{extra cost, €/yr}}, \qquad S\big(p^*(X);X\big) = cv\,A_{PF} \]

• \(p^*(X)\): payment at which the supply curve reaches the target area (previous slide)
• Logit: \(p^*(X{+}1)-p^*(X) \approx WTA_X = -\gamma_X/\delta\), the vertical shift of the supply curve per unit of \(X\)
• \(WTP_X(cv) = -(\beta_X + \rho_X \log cv)/\beta_{levy}\): value of \(X\) depends on the area covered
• Tested at \(cv = 40\,\%\); also: up to which coverage does H1 hold? CIs by Krinsky–Robb
• Timber: decrease (\(\Delta^{-}\)) and increase (\(\Delta^{+}\)) tested separately

H2: households have a higher WTP than owners (one-sided, in WTP space)

\[ WTP^{HH}_{k} > WTP^{FO}_{k}, \qquad WTP^{HH}_{k} = -\frac{\beta_{k}}{\beta_{levy}},\quad WTP^{FO}_{k} = -\frac{\beta_{k}+\tau_k}{\beta_{levy}+\tau_{levy}} \]

• \(k\): habitat trees (H2a), recreation infrastructure (H2b), timber decrease \(\Delta^{-}\) (H2c)
• Both groups answer the same vote cards; \(\tau\) are the owner interactions in model A (owners + control group)
• Full public sample if the main votes of the follow-up group show no preference differences
• WTP ratios remove differences in error scale between the groups

H3: the owner difference remains after controlling for characteristics \(z_n\)

\[ \beta_{kn} = \beta_k + \boldsymbol{\pi}_k' z_n + \tau_k D_n + \eta_{kn}, \qquad \exists\, k:\ \tau_k \neq 0 \]

• \(z_n\): socio-demographic and structural characteristics (e.g. age, education, income, wealth, nature relatedness, rural residence)
• Robustness: owners vs. comparable non-owners (PSM / ML matching) → split-sample mixed logit

• AGDW: funding levels
• Bavarian Forest Owner Portal: new plantings
• Saxony funding guideline
• FNR: KWM budget increase to 200 M€/yr (18.09.2026)
• FNR: KWM background and criteria
• FNR: KWM payment calculation scheme
• BWI 2022 result database: harvest 2012–2022 by state and ownership (table 6.01)
• BWI 2022 result database: habitat trees (theme „ökologisch bedeutsame Baummerkmale“, job L247of)