# Tree Planting¶

## Overview¶

ImpactMatters generates estimates of impact — estimates that quantify the causal effect nonprofits have on social outcomes relative to cost. For example: a nonprofit offsets a year of personal carbon emissions for an average of \$80. Our estimates incorporate best principles in social science, described in our Impact Methodology.

## Description¶

Human-induced global warming has already had observable effects on the environment, including rising sea levels and more extreme weather events like heatwaves and droughts. Carbon dioxide, a by-product of fossil fuel combustion, is a major contributor to global warming. Planting more trees has been shown to successfully sequester carbon dioxide, partially offsetting carbon dioxide emissions from human activities.1,2

## Outcome¶

A year of personal carbon emissions offset

We measure the success of tree planting programs as the cost to offset a year of personal carbon emissions. The total amount of carbon (in metric tons) sequestered by a nonprofit’s tree planting program can be difficult to comprehend. The Y constant (below) converts metric tons of carbon sequestered by a nonprofit’s program into years of personal carbon emissions offset — a more relatable concept. The average American emits 18.6 metric tons of carbon a year. If one tree sequesters 18.6 metric tons of carbon in its lifetime, we would assume that planting one tree offsets a year’s worth of carbon emissions for one person.

## Methodology for estimating attributable outcomes¶

In algebra, we calculate the attributable outcomes of a tree planting program (O) as follows:

where:
T = Number of trees planted by the program
C = Carbon sequestered per tree, in metric tons
Y = Amount of carbon emitted by the average American in a year; constant value of 18.6 metric tons
D = Probability of displacement

The probability of displacement (D) is calculated as follows:

where:
e = Euler’s number (approximately 2.72), the base of natural exponential functions
A = Pervious (i.e., permeable) land available for planting
L = Total pervious land

We calculate attributable outcomes in four steps. First, we multiply the number of trees planted by a program (variable T) by the carbon sequestered per tree (variable C). For urban forestry programs, which plant trees in cities, variable C is always 0.06 metric tons of carbon sequestered per urban tree. This is based on the U.S. Environmental Protection Agency’s (E.P.A.) estimate that “the weighted average carbon sequestered by a medium growth coniferous or deciduous tree, planted in an urban setting and allowed to grow for 10 years, is 36.4 lbs of carbon [equivalent to 0.06 metric tons] per tree.”3 Different tree species grow, survive and sequester carbon at different rates. By taking the E.P.A.’s weighted average sequestration rate, we risk masking these differences. However, species by species analysis is not possible without more data from nonprofits on the species they plant.

For reforestation programs, which plant trees in areas that were once forested, we take an additional step to define the biome where the nonprofit is planting trees, selecting from a list of 14 biomes: (1) boreal, (2) montane grasslands; (3) temperate grasslands; (4) deserts; (5) temperate broadleaf; (6) tropical grasslands; (7) flooded grasslands; (8) temperate coniferous; (9) tropical moist forest; (10) Mediterranean forest; (11) tropical dry forest; (12) tundra; (13) mangroves; and (14) tropical coniferous. In our model, the amount of carbon sequestered per tree (variable C) varies by biome.4,5 For instance, each tree added to the boreal forest biome sequesters approximately 0.16 metric tons of carbon, whereas each tree added to the tropical dry broadleaf biome sequesters 0.59 metric tons of carbon.

For urban forestry programs, 0.06 metric tons per tree was calculated by multiplying the probability of survival of a tree seedling in each year by the amount of carbon sequestered per year, then summing up the results over 10 years. This yields the probability-weighted, cumulative amount of carbon sequestered by each seedling left to grow in an urban setting for 10 years. By contrast, our methodology for reforestation is based on biome-specific estimates of the carbon stock in a fully mature forest, where: (1) carbon stock is defined as the amount of carbon stored in a forest stand at a single snapshot in time, and (2) by our simplifying assumption, a fully mature forest is one containing trees that have been left to grow for 30 years.6 We assume that the carbon stock of a forest measured at year 30 is roughly equivalent to the cumulative carbon sequestered by newly planted trees over a 30-year period (i.e., equivalent to the sum of the difference in carbon stocks from year to year over 30 years).7 This assumption allows us to overcome data availability and accessibility limitations. Expressed mathematically:

Where t = year and c = carbon stock at year t.

In addition, for reforestation programs, we assume the carbon stock data we borrow from the literature need not be further adjusted for the probability of tree survival. Because carbon stocks are snapshots of real forests in nature, they include not only live trees, but also standing dead trees, down dead wood and forest floor debris. The carbon stock of a forest at year 30 should already account for both 30-year-old trees as well as any that died over that period (and even new trees if any pollination occurred naturally).

As a second step, we convert metric tons of carbon sequestered into person-years of carbon emissions offset by dividing by variable Y, the average annual amount of carbon dioxide emitted per person in the U.S., 18.6 metric tons.8

Third, we estimate the person-years of carbon emissions offset by the nonprofit’s intervention net of the “counterfactual” amount of carbon emissions that would have been offset in the absence of the nonprofit. Subtracting counterfactual outcomes from observed outcomes is essential to calculating the change that occurred because of the nonprofit’s program.

Take urban forestry programs as an example. In any given city, tree planting may be influenced by multiple actors, institutions and their interests: zoning laws and conservation codes designed by the municipality; conservation easements agreed upon between developers and cities; the by-laws of collective residential associations; the interests of private property owners; and nonprofit organizations advocating for a greener city. Against this complex backdrop, it is crucial to consider how the actions of one actor could affect those of others.

To estimate the outcomes attributable to a nonprofit, we estimate the extent to which its tree planting might displace that of others (e.g., city government, other nonprofits) (variable D). Consider an extreme scenario of full displacement: Nonprofit A plants a plot of land to its capacity (1,000 trees). Seeing this, the city government and other nonprofits no longer plant any trees on that plot or any other plot of land. The change in outcomes attributable to Nonprofit A is zero because it merely displaced trees that would have been planted by others. Consider the other extreme, zero displacement: No other actors besides Nonprofit A intended to plant on that plot of land, so Nonprofit A gets 100 percent of the credit for planting 1,000 trees. Finally, consider a scenario of partial displacement: Seeing Nonprofit A plant 1,000 trees, the city government and other nonprofits reduce their intended planting from 1,000 trees to 500 trees. This may be because they have been physically displaced from Nonprofit A’s plot of land and now can only afford to plant 500 trees on a more expensive alternate plot of land. Or, this may be because they feel less need to plant their initial goal of 1,000 trees now that they see Nonprofit A is planting trees as well.

Tree-planting nonprofits do not tend to report outcomes net of displacement. The research literature also does not provide estimates of the displacement effects of various tree-planting actors. We therefore estimate the counterfactual based on three intuitive assumptions about the relationship between the amount of land available to be planted with trees and probability of displacement:

1. Displacement is more likely the less available land there is to be planted with trees.

2. There are a limited number of tree-planting actors and trees they can afford to plant. At moderate to high percentages of plantable land, we assume the probability of displacement tends to be low. Only once the percentage of plantable land falls to a certain level does competition begin to be felt among the limited number of actors.

3. According to Watkins et al., given the different location constraints they face and different motivations they have for tree planting, nonprofits and city governments tend to plant on distinct parcels of land.9 We assume they tend to do so until forced to compete.

Based on these three intuitive assumptions, we assume the rate of displacement is generally low, except when the percentage of plantable land is very low. This is best captured by an exponential decay function (figure 1 below). An exponential function, by contrast to a linear function, captures the likely situation that displacement responds little to more untapped land when the amount of untapped land is already large (right side of graph), but displacement responds hugely to more untapped land when the amount of untapped land is small (left side of graph).

We use the above exponential decay function to calculate the probability of displacement for both urban forestry and reforestation programs, substituting in the percentage of total pervious (i.e., permeable) land left that is available for planting in the area where the nonprofit operates. For urban forestry programs, estimates of variables A and L come from the U.S. Forest Service’s i-Tree software.10 For reforestation programs, we use Bastin et al.’s estimates of potential canopy restoration area per biome versus total possible canopy area per biome.11

## Methodology for calculating cost¶

Below, we summarize the most important aspects of our methodology for estimating the costs of tree planting programs. For a detailed discussion of what sources of data we use, how we treat specific line items and accommodate variation in accounting practices, see Reference Manual on Data Analysis.

### Costs we include¶

ImpactMatters estimates cost-effectiveness from the perspective of a socially minded donor. This means we count all important costs associated with a program regardless of who incurs them. Generally, the key cost-bearing parties are: the nonprofit itself; organizations with which it partners to run a program; the government (taxpayers); and the nonprofit’s beneficiaries.

### Nonprofit costs¶

We report total costs, including costs paid out of pocket by volunteers. This includes all costs directly related to tree planting, such as procuring seeds or seedlings, maintaining tree nurseries or tree farms, transporting and planting trees, and pruning and watering.

We include tree maintenance expenses in our calculations. We make the simplifying assumption that, on average, all tree planting programs are operating at a steady state. For example, a nonprofit incurs costs in 2018 to plant new trees, but it also incurs costs to maintain trees planted in 2017. Assuming there is no significant change in the scale of its operations from one year to the next, we expect that the cost to maintain the 2017 trees is an appropriate substitute in our calculation for the future cost of maintaining the 2018 trees.

In some cases, nonprofits may group together the cost of planting trees with the costs of other activities not directly tied to planting trees, e.g., litter removal for community beautification. If the nonprofit has not itemized its costs such that we can subtract these unrelated activities, we apply a standard set of assumptions to isolate tree planting program costs. See Reference Manual on Data Analysis for more details of this calculation.

### Beneficiary costs¶

We assume beneficiary costs are \$0 unless they are reported. Some nonprofits charge beneficiaries a nominal fee for tree planting services, mostly in residential planting contexts. We deduct this revenue from the nonprofit’s program cost, then add this revenue amount to beneficiary costs.

Nonprofits tend to charge heavily subsidized fees for tree planting, so we assume that people who pay nonprofits to help them plant trees would not otherwise pay market rates for trees. Said differently, these are costs incurred as a result of the nonprofit’s intervention and therefore should be included in our calculation.

### Partner costs¶

We assume partner costs are \$0 unless they are reported.

## Methodology for calculating impact¶

To calculate the impact of a tree planting program, we divide the total program-related costs incurred by all cost-bearing parties by the total person-years of carbon emissions offset. Crucially, the numerator and denominator must match logically: The numerator reflects the costs incurred in generating the attributable outcomes reflected by the denominator.

## Cost-effectiveness benchmarks¶

To rate the cost-effectiveness of tree planting programs, we compare each program’s cost to offset a year of personal carbon emissions to the “social cost of carbon” as estimated by the U.S. government.12 Our approach follows that of the Environmental Protection Agency and other federal agencies, which use the social cost of carbon in weighing the costs and benefits of regulatory actions.

The social cost of carbon is the theoretical price of carbon if all “externalities” are “internalized” such that the cost of emitting is passed on to emitters. Externalities associated with carbon emissions include, but are not limited to, adverse changes in net agricultural productivity, human health, property damages from increased flood risk, and “ecosystem services” like nutrient cycling, from which humans otherwise freely benefit. To “internalize” these externalities, researchers estimate the monetary value associated with the damage they cause. The total monetized damage is the social cost of carbon: the amount of harm associated with an incremental increase in carbon emissions in a given year, expressed in dollar values.

The benefit of offsetting carbon emissions is therefore averting the social costs of carbon, which is estimated at \$50 per metric ton of carbon.13 We multiply this figure by 18.6 metric tons, the amount of carbon the average American emits in a year. The result is about \$931 — the social cost averted by offsetting a year of personal carbon emissions.

For a tree planting program to be cost-effective, its cost to offset a year of personal carbon emissions must be comparable to the \$931 benefit it generates. Specifically, we set our benchmarks for 4 and 5 stars as follows:

• 4 stars: Programs that offset a year of personal carbon emissions for 125 percent of the social cost of carbon averted (125% * \$931 = \$1,164).

• 5 stars: Programs that offset a year of personal carbon emissions for 75 percent of the social cost of carbon averted (75% * \$931 = \$698).

## Nonprofit checklist of data needed to calculate impact¶

The following data is necessary to estimate the impact of tree planting programs.

Table 1

Checklist item

Required from nonprofit?

Details

Program activities

Yes

A program is a set of goods or services provided by the nonprofit to a population of beneficiaries with the goal of improving one or more outcomes. Generally, a program consists of the same components delivered to each beneficiary, with only minor deviations across different settings.

Geography

Yes

We recommend nonprofits report the location where they plant trees, e.g., city or county for urban forestry programs, and country and biome for reforestation programs. We use this information to estimate the likely carbon sequestration rate of trees planted by the nonprofit.

Timeframe

Yes

We recommend nonprofits report annual figures that align with their fiscal year.

Number of trees planted

Yes

If a nonprofit plants trees in more than one location, we recommend it report the number of trees planted in each location separately.

Carbon sequestered per tree

No

If not reported by the nonprofit, we use estimates from the research literature of the average amount of carbon sequestered over a tree’s lifetime by trees planted in U.S. cities and 14 global biomes.

Counterfactual percentage of trees planted

No

If not reported by the nonprofit, we model the counterfactual as an exponential decay function of the amount of plantable pervious land left in the location or locations where the nonprofit operates.

Program cost

Yes

We recommend reporting total costs, including costs paid out of pocket by volunteers.

Beneficiary cost

No

We recommend reporting beneficiary costs if they are substantial. They can be estimated at \$0 if they are not substantial. Some nonprofits charge neighborhood residents a small fee for urban trees. We deduct this revenue from the nonprofit’s program cost, then add this revenue to beneficiary costs.

Partner cost

No

We recommend reporting partner costs if they are substantial. They can be estimated at \$0 if they are not substantial.

## Limitations of our analysis¶

### Oversimplification of the outcome¶

Emerging research suggests that not all carbon dioxide emissions are equal.14 While CO2 is generally well mixed in the atmosphere, data indicate that it appears in higher ratios in urban areas compared to background air, resulting in “urban CO2 domes,” clouds of carbon pollution that hover above cities. As a result, carbon emitted in a city may pose greater health risks than the equivalent emissions elsewhere. Conversely, emissions offset in a city may be more valuable than the equivalent emissions offset elsewhere. Our calculation values offset emissions equally, no matter the location. If research on urban CO2 domes is correct, we may be underestimating the effect of tree planting programs in cities and possibly overestimating the effect of tree planting programs in rural and uninhabited areas.

### Multiple important outcomes¶

We estimate the impact of tree planting programs on a single outcome, carbon emissions offset, chosen because it best aligns with the stated missions of the programs under analysis. However, tree planting programs may be achieving other secondary but important outcomes not related to carbon sequestration, such as increased property values, reduced expenditure on stormwater infrastructure, more shade from trees resulting in lower heating and cooling costs, and potential improvements to quality of life from neighborhood beautification.15

### Cost of reaching special locations¶

Planting trees in cities may be inherently more expensive than planting trees in rural and uninhabited areas. Urban forestry nonprofits may incur higher costs to secure rights to land and to transport saplings, and they may enjoy fewer economies of scale by planting fewer trees at lower tree density. Research by Jacobson suggests that urban forestry may be particularly worthwhile to combat urban CO2 domes. However, the mainstream consensus is that major greenhouse gases mix uniformly in the atmosphere.16 This implies that the location of emissions is independent from their climatological consequences. Given this uncertainty, we make no attempt to adjust upward the outcomes of urban forestry programs; higher costs of urban forestry are reflected in our calculations in their entirety.

### Specific counterfactuals¶

To understand the impact of a program, we ask the counterfactual question: What would have happened to beneficiaries if the program had not, counter to fact, been there to serve them? Because the vast majority of nonprofits have not conducted impact evaluations, we need to construct our own counterfactuals based on public data sources and the research literature. But in doing so, we risk masking variation in effectiveness across nonprofits. For instance, under our methodology, we apply the same counterfactual to all tree planting programs operating in the same biome. The counterfactual is a function of the plantable pervious land left in that biome: The less land there is still available for tree planting, the more likely other tree planters would have planted in on the nonprofit’s plot if the nonprofit had not existed. But consider the largest biome, the taiga or boreal forest, which accounts for 17 percent of the Earth’s land area.17 It is unlikely the same counterfactual applies to tree planting activities in all parts of the taiga, which spans parts of Russia, China, Mongolia, Finland, Sweden and Norway.

### Data quality¶

Our estimates rely on data made public by nonprofits on their websites, annual reports, financial statements and Form 990s. However, the data required for an ideal calculation of impact may not always be reported by nonprofits, e.g., the tree species planted by each nonprofit. Different species of trees sequester carbon at different rates and experience different growth and survival rates. Because data on nonprofits’ tree species are not always available to us, we must use carbon sequestration rates specific to U.S. cities and 14 global biomes rather than carbon sequestration rates specific to each type of tree species. As a result, our analysis necessarily simplifies a complex calculation. For more detail on our sources of data and how we interpret them, please see Reference Manual on Data Analysis.

### Representativeness of (analyzed) programs¶

We only issue ratings for nonprofits if we can perform analysis on 15 percent or more of the nonprofit’s total program budget. This approach means some nonprofits are rated on only some of their programs. The remaining programs, which we could not analyze, could be more or less cost-effective than the programs we analyzed.

Footnotes

1
2
3
4

Note that the research literature tends to focus on the amount of carbon sequestered per acre or hectare of forest, rather than per tree. We convert these estimates from the literature into carbon sequestered per tree by dividing by the approximate tree density in the biome in question. For instance, the literature finds that 178 million hectares of boreal forest can sequester 42.6 gigatonnes of carbon. At approximately 1,500 trees per hectare of fully forested boreal forest, that is roughly equivalent to 0.16 metric tons of carbon sequestered per tree (42,600,000,000 metric tons of carbon / (178,000,000 hectares * 1,500 trees per hectare)). Data on tree density come from Glick et al. (2016) Spatially-Explicit Models of Global Tree Density. However, Bastin et al. note in their supplementary materials: "Each value of tree density or carbon density reported in these studies was associated with a pixel-value of potential tree cover of 100 percent." We therefore adjust Glick et al.'s estimates of trees per hectare per biome to reflect 100 percent tree cover (as opposed to 63 percent on average reported by Glick et al. originally).

5
6

The assumption of 30 years is consistent with estimates by Loehle of the age at maturity of 159 North American tree species and Bastin et al.’s own claim of “several decades.” In reality, age at full maturity varies by species.

7

This effectively assumes the carbon stock of the forest stand at year zero is zero. Such an assumption is realistic in an afforestation scenario (foresting previously non-forest land), but may introduce minor upward bias in reforestation scenarios where the carbon stock of the forest stand at year zero is greater than zero. See, for instance, the differences between Appendix A and B of the U.S. Forest Service’s standard yield tables for reforestation and afforestation, respectively.

8
9
10
11
12
13

The Interagency Working Group estimated the social cost of carbon would be \$41 per metric ton of carbon in 2019, expressed in 2007 dollars. After adjusting for inflation, this is \$50 per metric ton of carbon in 2019 dollars.

14
15
16
17