Classwork 6

REDD+ Experiment — Money Growing on Trees

Author

Byeong-Hak Choe

Published

October 6, 2025

Modified

October 9, 2025

Overview

You are a household in the beautiful but low-income tropical country of Ephoria.
You live in a rural area on the margin of a forest. Over the last few decades, expansion of agriculture and urban areas has reduced forest cover and put pressure on local resources.

Your small, traditional community survives mainly through:

Subsistence farming (gardens for food)
Harvesting wood and non-timber forest products (fruits, nuts, medicinal plants, bark, fish, and game)

Tropical forest ecosystem

Forest degradation from wood harvesting

Wood is used for cooking and heating, but contributes to deforestation and climate change.
If you don’t harvest wood, you must either do without or buy replacement goods on the market.

Your Card and Harvest Value

Each participant receives a playing card:

Number cards (1–10): harvesting value = $10 × card number
Face cards or Jokers: harvesting value = $0

Harvest values differ because of variations in household needs, family size, and access to forest areas.

Policy Context

To fight deforestation and raise revenue, Ephoria’s policymakers are experimenting with Payments for Ecosystem Services (PES) through an international REDD+ program.

Local households/communities sign PES contracts.
Foreign buyers purchase avoided-deforestation credits to offset emissions.

You will play through several contract periods, each representing a different policy environment.

Payment Mechanism (Real Incentives)

To make decisions realistic, two random participants will be selected at the end of the experiment to receive real payments based on their total earnings across all rounds:

\[ \text{Real Payment} = \frac{\text{Total Experimental Earnings}}{100} \]

Typically, this amounts to about $3–$9 per winner.

Your Earnings Formula

\[ (\text{Earnings}) = (\text{Farming Income}) + (\text{PES Payment}) + (\text{Forest Harvesting}) - (\text{Policing or Fines}) \]

Farming income of $70 is automatically granted to every household each period.

Income Components

Component	Description	Default Value
Farming Income	Automatic subsistence income from your land plot	$70
PES Payment	Payment for entering PES contract	$50
Forest Harvesting	$10 × card number (if not in PES)	varies

If total earnings < $75, your family cannot eat for a day (conceptually unpleasant!).

Result

Contract Periods (Treatments)

Period	Policy Context	Key Feature
0	Baseline	No PES (individual harvest only)
1	PES	Legal PES contracts only
2	PES + Illegal Harvest	Can cheat; subject to random audit
3	Community PES	Group decision and shared payment
4	Community + Illegal Harvest	Group and individual enforcement

Experiment Data

Code

df_0 <- read_csv("https://bcdanl.github.io/data/econ-340-redd-exp-cp0-2025-1006.csv")

df_0 |> 
  datatable()

Code

df_1 <- read_csv("https://bcdanl.github.io/data/econ-340-redd-exp-cp1-2025-1006.csv")

df_1 |> 
  datatable()

Code

df_2 <- read_csv("https://bcdanl.github.io/data/econ-340-redd-exp-cp2-2025-1006.csv")

df_2 |> 
  datatable()

Code

df_3 <- read_csv("https://bcdanl.github.io/data/econ-340-redd-exp-cp3-2025-1006.csv") |> relocate(community, .after = 2)

df_3 |> 
  datatable()

Code

df_4 <- read_csv("https://bcdanl.github.io/data/econ-340-redd-exp-cp4-2025-1006.csv") |> relocate(community, .after = 2)

df_4 |> 
  datatable()

Transformations by Contract Period

We compute earnings using the rules defined for each period.
(Recall: Earnings = Farming Income + PES Payments + Harvest − Fines/Policing.)

CP0 — Baseline (no PES)

Code

df_0 <- df_0 |>
  mutate(
    period = "Baseline",
    harvest = ifelse(harvest == "No", 0, 1),
    earning = 70 + harvest_value * harvest
  ) |>
  relocate(earning, .after = `submission-time`)

CP1 — PES (legal PES only)

Code

df_1 <- df_1 |>
  mutate(
    period = "PES",
    pes_payment = 50 * pes,
    harvest_payment = harvest_value * (1 - pes),
    earning = 70 + pes_payment + harvest_payment
  ) |>
  relocate(earning, .after = `submission-time`)

CP2 — PES + Illegal Harvest (audit risk)

If audited and cheating: lose PES (50), lose illegal harvest, pay $70 fine.

Code

df_2 <- df_2 |>
  mutate(
    period = "PES + Ill. Harvest",
    pes_payment = 50 * pes,
    harvest_payment = harvest_value * (1 - pes),
    # audit draw only for PES participants
    fine = ifelse(pes == 1, rbinom(n(), 1, 0.25), 0),
    base_earn = 70 + pes_payment + harvest_payment + illegal * harvest_value,
    penalty  = ifelse(fine == 1 & pes == 1 & illegal == 1, 50 + harvest_value + 70, 0),
    earning  = base_earn - penalty
  ) |>
  relocate(earning, .after = `submission-time`)

CP3 — Community PES (split rules)

Each community that joins PES has a total of $50 × (# members) to divide by the chosen split rule.

Code

df_3 <- df_3 |>
  mutate(
    period = "Community PES",
    pes = ifelse(pes_group == "PES", 1, 0),
    pes_payment = 50 * pes,
    harvest_payment = harvest_value * (1 - pes)
  ) |>
  group_by(community) |>
  mutate(
    additional = case_when(
      pes == 0 ~ 0,
      split_rule == "equal" ~ pes_payment,
      split_rule == "proportional_card" ~ (harvest_value + 10) * pes_payment * n() / sum(harvest_value + 10),
      split_rule == "floor_plus_proportional" ~ 0.75 * pes_payment + 0.25 * (harvest_value + 10) * pes_payment * n() / sum(harvest_value + 10),
      split_rule == "floor_plus_needs" ~ 0.75 * pes_payment + 0.25 * (1/(harvest_value + 10)) * pes_payment * n() / sum(1/(harvest_value + 10)),
      is.na(split_rule) | split_rule == "no_PES" ~ 0,
      TRUE ~ 0
    )
  ) |>
  ungroup() |>
  mutate(
    earning = round(70 + harvest_payment + additional, 2)
  ) |>
  relocate(earning, .after = `submission-time`)

CP4 — Community PES + Illegal Harvest (policing & audit)

Two randomizations (v1, v2) illustrate different audit outcomes.

Code

# --- Version 1 (seed = 2) ---
set.seed(2)
df_4_v1 <- df_4 |>
  mutate(
    period = "Community PES + Ill. Harvest",
    pes = ifelse(pes_group == "PES", 1, 0),
    pes_payment = 50 * pes,
    harvest_payment = harvest_value * (1 - pes)
  ) |>
  group_by(community) |>
  mutate(
    illegal_n  = sum(illegal),
    audit_prob = pmin(0.1 * illegal_n, 1),
    audit_draw = ifelse(pes == 1 & police == 0, rbinom(n(), 1, audit_prob), 0),
    audit      = ifelse(sum(audit_draw) > 0, 1, 0),
    additional = case_when(
      pes == 0 ~ 0,
      pes == 1 & police == 1 & split_rule %in% c("equal","compliers_only_equal","remove_benefits_illegal_equal") ~ pes_payment,
      pes == 1 & police == 1 & split_rule == "remove_benefits_illegal_proportional" ~ pes_payment * (harvest_value+10)/sum(harvest_value+10),

      audit == 0 & pes == 1 & police == 0 & split_rule == "equal" ~ pes_payment,
      audit == 0 & pes == 1 & police == 0 & split_rule == "compliers_only_equal" ~ ifelse(illegal==0, pes_payment, 0),
      audit == 0 & pes == 1 & police == 0 & split_rule == "remove_benefits_illegal_equal" ~ ifelse(illegal==0, pes_payment*n()/(n()-illegal_n), 0),
      audit == 0 & pes == 1 & police == 0 & split_rule == "remove_benefits_illegal_proportional" ~ ifelse(illegal==0, pes_payment*(harvest_value+10)*n()/sum((harvest_value+10)*(1-illegal)), 0),

      audit == 1 & pes == 1 & police == 0 & split_rule == "equal" ~ pes_payment,
      audit == 1 & pes == 1 & police == 0 & split_rule == "compliers_only_equal" ~ ifelse(illegal==0, pes_payment, -(illegal*harvest_value) - 70),
      audit == 1 & pes == 1 & police == 0 & split_rule == "remove_benefits_illegal_equal" ~ ifelse(illegal==0, pes_payment*n()/(n()-illegal_n), -(illegal*harvest_value) - 70),
      audit == 1 & pes == 1 & police == 0 & split_rule == "remove_benefits_illegal_proportional" ~ ifelse(illegal==0, pes_payment*(harvest_value+10)*n()/sum((harvest_value+10)*(1-illegal)), -(illegal*harvest_value) - 70),

      is.na(split_rule) | split_rule == "no_PES" ~ 0,
      TRUE ~ 0
    )
  ) |>
  ungroup() |>
  mutate(
    earning = round(70 + harvest_payment + additional + illegal*harvest_value - 5*police, 2),
    earning = ifelse(audit == 1, 0, earning)
  ) |>
  relocate(earning, .after = `submission-time`)

# --- Version 2 (seed = 1) for sensitivity ---
set.seed(1)
df_4_v2 <- df_4_v1 |>
  select(-earning, -audit_draw, -audit, -additional) |>
  group_by(community) |>
  mutate(
    illegal_n  = sum(illegal),
    audit_prob = pmin(0.1 * illegal_n, 1),
    audit_draw = ifelse(pes == 1 & police == 0, rbinom(n(), 1, audit_prob), 0),
    audit      = ifelse(sum(audit_draw) > 0, 1, 0),
    additional = case_when(
      pes == 0 ~ 0,
      pes == 1 & police == 1 & split_rule %in% c("equal","compliers_only_equal","remove_benefits_illegal_equal") ~ pes_payment,
      pes == 1 & police == 1 & split_rule == "remove_benefits_illegal_proportional" ~ pes_payment * (harvest_value+10)/sum(harvest_value+10),

      audit == 0 & pes == 1 & police == 0 & split_rule == "equal" ~ pes_payment,
      audit == 0 & pes == 1 & police == 0 & split_rule == "compliers_only_equal" ~ ifelse(illegal==0, pes_payment, 0),
      audit == 0 & pes == 1 & police == 0 & split_rule == "remove_benefits_illegal_equal" ~ ifelse(illegal==0, pes_payment*n()/(n()-illegal_n), 0),
      audit == 0 & pes == 1 & police == 0 & split_rule == "remove_benefits_illegal_proportional" ~ ifelse(illegal==0, pes_payment*(harvest_value+10)*n()/sum((harvest_value+10)*(1-illegal)), 0),

      audit == 1 & pes == 1 & police == 0 & split_rule == "equal" ~ pes_payment,
      audit == 1 & pes == 1 & police == 0 & split_rule == "compliers_only_equal" ~ ifelse(illegal==0, pes_payment, -(illegal*harvest_value) - 70),
      audit == 1 & pes == 1 & police == 0 & split_rule == "remove_benefits_illegal_equal" ~ ifelse(illegal==0, pes_payment*n()/(n()-illegal_n), -(illegal*harvest_value) - 70),
      audit == 1 & pes == 1 & police == 0 & split_rule == "remove_benefits_illegal_proportional" ~ ifelse(illegal==0, pes_payment*(harvest_value+10)*n()/sum((harvest_value+10)*(1-illegal)), -(illegal*harvest_value) - 70),

      is.na(split_rule) | split_rule == "no_PES" ~ 0,
      TRUE ~ 0
    )
  ) |>
  ungroup() |>
  mutate(
    earning = round(70 + harvest_payment + additional + illegal*harvest_value - 5*police, 2),
    earning = ifelse(audit == 1, 0, earning)
  )

Bind & Label

Code

all_cp <- bind_rows(df_0, df_1, df_2, df_3, df_4_v1) |>
  mutate(
    diff = earning - (70 + harvest_value),  # PES vs counterfactual non-PES
    period = factor(period,
                    levels = c("Baseline", "PES", "PES + Ill. Harvest",
                               "Community PES", "Community PES + Ill. Harvest"))
  )

all_cp_v2 <- bind_rows(df_0, df_1, df_2, df_3, df_4_v2) |>
  mutate(
    diff = earning - (70 + harvest_value),
    period = factor(period,
                    levels = c("Baseline", "PES", "PES + Ill. Harvest",
                               "Community PES", "Community PES + Ill. Harvest"))
  )

Visualizations

Illegal Harvest vs. Harvest Value (CP2 & CP4)

Higher harvest values imply a greater opportunity cost of conservation, leading to higher illegal harvest rates. However, this relationship weakens under community PES, as the consequences of punishment extend to other community members.

Code

all_cp |>
  filter(period %in% c("PES + Ill. Harvest","Community PES + Ill. Harvest"),
         !is.na(harvest_value)) |>
  group_by(period, harvest_value) |>
  summarise(illegal_rate = mean(illegal %in% TRUE, na.rm = TRUE),
            n = n(), .groups="drop") |>
  ggplot(aes(x = harvest_value, y = illegal_rate)) +
  geom_col(aes(fill = period), position = position_dodge2(preserve = "single"), alpha = .67) +
  geom_smooth(method = lm, se = FALSE, aes(color = period)) +
  scale_color_tableau() + scale_fill_tableau() +
  scale_y_continuous(labels = scales::percent) +
  labs(x = "Harvest Value", y = "Illegal Rate", fill = "Period", color = "Period") +
  theme_ipsum()

“What if you joined PES?” — Earnings Differences

Right of zero = PES helped vs. your non-PES counterfactual.
Left of zero = PES hurt, often due to audits of illegal harvest.

For this version below, only one community was audited in CP4 (Community PES + Illegal Harvest).

Code

all_cp |>
  filter(period != "Baseline") |>
  ggplot(aes(y = period, x = diff, fill = period)) +
  geom_boxplot(alpha = 0.85, outlier.colour = 'darkred', outlier.size = rel(2)) +
  scale_fill_tableau() +
  scale_x_continuous(breaks = seq(-175, 75, 25)) +
  labs(
    title   = "What If You (and Your Community) Joined the PES Program?",
    subtitle = "Distribution of Earnings Differences Between PES Participants and Their Counterfactual / Real Non-PES Incomes",
    x = "(PES Earnings) − (Non-PES Earnings)", y = NULL,
    caption = "(Non-PES = Farming Income + Harvest Value)\nPES earnings can be negative when audits catch illegal harvesting."
  ) +
  theme(legend.position = "none")

For this version below, two communities were audited in CP4.

Code

all_cp_v2 |>
  filter(period != "Baseline") |>
  ggplot(aes(y = period, x = diff, fill = period)) +
  geom_boxplot(alpha = 0.85, outlier.colour = 'darkred', outlier.size = rel(2)) +
  scale_fill_tableau() +
  scale_x_continuous(breaks = seq(-175, 75, 25)) +
  labs(
    title   = "What If You (and Your Community) Joined the PES Program?",
    subtitle = "Distribution of Earnings Differences Between PES Participants and Their Counterfactual / Real Non-PES Incomes",
    x = "(PES Earnings) − (Non-PES Earnings)", y = NULL,
    caption = "(Non-PES = Farming Income + Harvest Value)\nPES earnings can be negative when audits catch illegal harvesting."
  ) +
  theme(legend.position = "none")

Community Policing Impact on Earnings

For this version below, only one community was audited in CP4.

Code

all_cp |>
  filter(period == "Community PES + Ill. Harvest") |>
  mutate(police = ifelse(police == 0, "No Policing", "Policing")) |> 
  group_by(police) |>
  summarise(mean_earning = mean(earning, na.rm = TRUE),
            median_earning = median(earning, na.rm = TRUE)) |>
  ggplot(aes(x = factor(police), y = mean_earning, fill = factor(police))) +
  geom_col() +
  labs(
    title = "Average Earnings by Policing Decision (CP4)",
    x = "",
    y = "Mean Earnings ($)",
    color = "Period"
  ) +
  scale_fill_tableau() +
  theme_ipsum(base_size = 13) +
  theme(legend.position = "none")

For this version below, two communities were audited in CP4.

Code

# Community Policing Impact - v2
  # More community gets audited
all_cp_v2 |>
  filter(period == "Community PES + Ill. Harvest") |>
  mutate(police = ifelse(police == 0, "No Policing", "Policing")) |> 
  group_by(police) |>
  summarise(mean_earning = mean(earning, na.rm = TRUE),
            median_earning = median(earning, na.rm = TRUE)) |>
  ggplot(aes(x = factor(police), y = mean_earning, fill = factor(police))) +
  geom_col() +
  labs(
    title = "Average Earnings by Policing Decision (CP4)",
    x = "",
    y = "Mean Earnings ($)",
    color = "Period"
  ) +
  scale_fill_tableau() +
  theme_ipsum(base_size = 13) +
  theme(legend.position = "none")

Policing governance was effective in preventing illegal harvesting.

Table Summary

Below, click column headers to sort; use the search boxes to filter.

Community-level Summary (CP3–CP4)

What to look for: Which communities joined PES more? Did policing correlate with higher average earnings?

Code

comm_summary <- all_cp |>
  filter(period %in% c("Community PES", "Community PES + Ill. Harvest")) |>
  group_by(period, community) |>
  summarise(
    n              = n(),
    mean_earning   = round(mean(earning, na.rm = TRUE), 2),
    median_earning = round(median(earning, na.rm = TRUE), 2),
    pes_rate       = round(mean(pes, na.rm = TRUE), 3),
    illegal_rate   = round(mean(illegal, na.rm = TRUE), 3),
    policing_rate  = round(mean(police, na.rm = TRUE), 3),
    harvest_value      = round(mean(harvest_value, na.rm = TRUE), 2),
    .groups = "drop"
  )

DT::datatable(comm_summary, rownames = FALSE,
              options = list(pageLength = nrow(comm_summary), scrollX = TRUE))

Split Rule Preferences

What to look for: Did groups favor equal, proportional, or floor + top-up rules?

Code

df_rules <- all_cp |>
  ungroup() |>
  select(-`id-number`) |>
  filter(period %in% c("Community PES", "Community PES + Ill. Harvest"),
         !is.na(split_rule)) |>
  select(period, community, split_rule) |> 
  distinct() |> 
  count(period, split_rule) |>
  group_by(period) |>
  mutate(pct = round(100 * n / sum(n), 1)) |>
  arrange(period, desc(n)) |>
  ungroup()

DT::datatable(df_rules, rownames = FALSE,
              options = list(pageLength = 10, scrollX = TRUE))

Top & Bottom Earners (Total across all periods)

What to look for: Do high-card players dominate, or do PES/policing strategies change the leaderboard?

Code

totals <- all_cp |>
  group_by(`id-number`) |>
  summarise(total_earning = sum(earning, na.rm = TRUE), .groups = "drop") |>
  arrange(desc(total_earning)) |>
  mutate(rank = row_number()) |> 
  mutate(total_earning = round(total_earning, 2))

DT::datatable(bind_rows(
  head(totals, 5) |> mutate(group = "Top 5"),
  tail(totals, 5) |> mutate(group = "Bottom 5")
) |>
  relocate(group),
  rownames = FALSE,
  options = list(pageLength = 10, scrollX = TRUE))

Lottery (2 randomly selected students)

Real payout = Total experimental earnings / 100 (per winner).

Code

lottery_pool <- all_cp |>
  group_by(`id-number`) |>
  summarise(real_earning = round(sum(earning, na.rm = T) / 100, 2),
            .groups = "drop")

set.seed(1)
winners <- slice_sample(lottery_pool, n = 2)

DT::datatable(winners, rownames = FALSE,
              options = list(dom = 't', paging = FALSE))

Notes on Interpretation

Additionality: PES is additional only if it changes behavior relative to the non-PES baseline.
Verifiability/Audits: Illegal harvesting under PES can erase benefits (and trigger fines).
Community governance: Policing can deter cheating but costs resources; splitting rules shape perceptions of fairness and incentives.
Who benefits? High-card players often prefer harvesting; needs-based or floor-plus rules can shift surplus toward low-card members.

Experiment

Contract Periods (CP)

Period 0 — Baseline

PES contracts not available.
Decision: Harvest wood or not.
Earnings as defined above.

CP0 - Payment Recording

CP	Conditions	PES / Harvest?	Illegal Harvest?	Audited?	A: Farming	B: PES	C: Harvest	D: Fines	Earnings = A + C
0	Baseline	Harvest / No			$70		$

CP0 - Experiment

Period 1 — Payments for Ecosystem Services (PES)

PES contracts available at $50 income.
If you take PES, you cannot harvest.
- No illegal harvest is possible.

CP1 - Payment Recording

CP	Conditions	PES / Harvest?	Illegal Harvest?	Audited?	A: Farming	B: PES	C: Harvest	D: Fines	Earnings = A + B + C
1	PES	PES / No			$70	$	$

CP1 - Experiment

Period 2 — PES + Illegal Harvest

PES income = $50.
You may illegally harvest after signing PES.
Each contract audited with 25 % probability.
If audited and caught:
- Lose PES income
- Lose harvest earnings
- Pay fine of $70

CP2 - Payment Recording

CP	Conditions	PES / Harvest?	Illegal Harvest?	Audited?	A: Farming	B: PES	C: Harvest	D: Fines	Earnings = A + B + C – D
2	Illegal Harvest	PES / No	Y / N	Y / N	$70	$	$	$

CP2 - Experiment

Period 3 — Community Contract

Participants form a community of 4-5 members (randomly assigned).
Each community jointly decides:
- Whether to enter the PES contract.
- If the community joins PES, the community receives a total PES income equal to
  \[ \$50 \times (\text{number of households in the community}) \]
- The community must then decide how to divide this total PES income among its members.
If the community joins PES → no one may harvest.

CP3 - Payment Recording

CP	Conditions	PES / Harvest?	Illegal Harvest?	Audited?	A: Farming	B: PES	C: Harvest	D: Fines	Earnings = A + B + C
3	Community	PES / No			$70	$	$

CP3 - Experiment

Period 4 — Community Contract + Illegal Harvest

Same communities as in Period 3.
Community Decisions:
1. Join PES?
2. If yes → Police the group? (cost = $5 per household)
- If policing → no household harvests illegally.
- If no policing → each household privately decides Harvest illegally or not.

Audit Rule

Audits occur at community level only when joining PES.
Audit probability = 10 % × (# of illegal harvesters).
If audited and any illegal harvest is found:
- PES payments void for everyone.
- All harvests confiscated.
- Each household fined $70.

Order of Actions (Period 4)

Community decides whether to join PES.
If yes → community decides whether to police.
If no policing → households privately decide whether to harvest illegally.
Government may audit (probability as above).

CP4 - Payment Recording

CP	Conditions	PES / Harvest?	Illegal Harvest?	Audited?	A: Farming	B: PES	C: Harvest	D: Fines/Policing	Earnings = A + B + C – D
4	Comm + Illegal Harvest	PES / No	Y / N	Y / N	$70	$	$	$

CP4 - Experiment

Recording Sheet

CP	Conditions	PES / Harvest?	Illegal Harvest?	Audited?	A: Farming	B: PES	C: Harvest	D: Fines	Earnings = A + B + C – D
0	Baseline	Harvest / No			$70
1	PES	PES / No			$70	$	$	$
2	Illegal Harvest	PES / No	Y / N	Y / N	$70	$	$	$
3	Community	PES / No			$70	$	$	$
4	Comm + Illegal Harvest	PES / No	Y / N	Y / N	$70	$	$	$
TOTAL									$

Note: If you illegally harvest and are audited, cross out columns B and C for that round – both become zero.

Reference

Dissanayake, S. T. M., & Jacobson, S. A. (2021). Money growing on trees: A classroom game about payments for ecosystem services and tropical deforestation. The Journal of Economic Education, 52(3), 192–217. https://doi.org/10.1080/00220485.2021.1925183