Why global wood harvests aren’t emitting 3.5 to 4.2 Gt CO2 per year in net emissions.

Why global wood harvests aren’t emitting 3.5 to 4.2 Gt CO2 per year in net emissions.

Brent Sohngen (sohngen.1@osu.edu)

Part I: Good modeling matters, bad modeling matters more.

A recent article by Peng et al. (2023) called “The carbon cost of global wood harvests” published July 5, 2023 in Nature, suggested that economic models are not up to the task of measuring carbon emissions from wood product harvesting. The authors of that study calculate that wood harvesting will cause a net emission of 3.5 to 4.2 Gt CO­2­ per year over a 40-year period from 2010 and 2050. The authors claim to estimate this value from a counterfactual that assumes no harvesting at all. This supposed counterfactual is calculated via a biophysical model that compares the carbon flux after harvest in a regenerated stand plus the market products with the stand left alone.

The authors propose an interesting idea – comparing a world with timber harvests to a world without timber harvests – but their approach and model makes no sense. Peng et al. model 40 years of future timber harvests with a biophysical model (called the CHARM model) that uses only per capita income to determine how much timber gets harvested every year, what type of timber gets harvested every year and where it gets harvested. That’s right, they are modeling a market, but dispensing with the economics because, in their words, economic models are not “credible.” There are no costs to harvest wood in the model, no interest rates that affect investments or rotation ages, no equilibrium conditions, no setting of prices equal to marginal cost, no investments in new stocks, etc. They acknowledge economics is hard, so they ignore it, and instead deploy a set of arbitrary rules to consume wood, harvest trees, and regenerate trees.

Not surprisingly, their key result that there are 3.5 to 4.2 Gt CO2 in net emissions from wood harvesting is ridiculous.

Not surprisingly, this is not the first time this type of modeling has been deployed. After its creation in the early 1900s, the United States Forest Service famously started chasing a quixotic timber famine for much of the twentieth century. As shown by Clawson (1979), study after study by the US Forest Service found that US forests were growing far less than was needed for future timber harvests. In response to these “gap” models, which also ignored economics, the Forest Service created a huge timber harvesting operation that eventually met 15 % of the nation’s wood supply with federal timber – much of it old growth.

Thankfully, Darius Adams and Richard Haynes, who won the Marcus Wallenberg Prize in forestry this year, created an actual economic model to project timber harvests, prices, and forest stocks (Adams and Haynes 1980). They changed the dynamic. Whole posts could be written on the timber famine and its effects on US forest policy, but the upshot for the CHARM model is that most of us thought the idea of using purely physical models like this to predict future timber harvesting and forest growth were a thing of the past. But if we have learned anything from one of the most famous purely physical modeling exercises in the past –”The Limits to Growth” effort by Donnela Meadows and others in 1972 (Meadows et al. 1972) – purely physical modeling is quite the allure.


Part II: A closer look at the big numbers in Peng et al

(Hint: keep track of gross and net here)

It is incredibly unlikely that future timber harvesting would lead to net emissions of CO2 from forests of 3.5 to 4.2 Gt CO2 per year as claimed in Peng et al. Right now, land use, land use change, and forestry are a net global sink of 6.6 Gt CO2 per year (Nabuurs et al. 2022). Gross emissions from timber harvesting and deforestation are about 5.9 Gt CO2 per year, meaning forests and other land uses are pulling 12.5 Gt CO2 per year from the atmosphere in gross. Most of the gross emission is deforestation. Gross emissions from industrial wood production estimated by the Global Timber Model are about 1.6 GtCO2 this decade.

The Peng et al. study does include global wood fuel consumption, which we do not include in GTM.  Wood fuel is nearly half of all wood consumption globally, and its consumption is skewed heavily towards developing regions. We haven’t included it in GTM because it’s hard to know how much of this wood came from the scraps of timber cuttings, or deforestation. But the IPCC numbers above do include it.

So, here is a scorecard so far:

IPCC Gross emissions from all wood harvesting and deforestation                =             5.9 Gt CO2/yr

GTM estimated gross emissions from industrial wood harvesting                 =             1.6 Gt CO2/yr  

Potential gross emissions from wood fuel and deforestation                         =             4.3 Gt CO2/yr  


Gross emissions are rather large. But Peng et al. claim net emissions from just timber harvesting (not deforestation) are as big or bigger. How do they get to their rather large calculation of 3.5 to 4.2 Gt CO2 per year?

First, they ignore economics and construct a purely biophysical model. This will result in overestimating harvests and underestimating regrowth because the model will not harvest efficiently and will not regenerate efficiently. Seriously, have a look at Marion Clawson’s Science article in 1979. Dr. Clawson’s colleague at Resources For the Future, Roger Sedjo, got it right when he declared at a 1980 meeting at the International Institute for Applied Systems Analysis in Vienna, Austria:

“Many observers anticipate a growing scarcity of wood through the remainder of this century and into the next accompanied by an attendant rise in the relative price of wood products and the primary forest resource. Given these expectations it is certainly prudent to investigate the potential of plantation forests in meeting future demand and to recognize that the possibility of higher future real stumpage prices may provide incentives for forestry investments not previously economically justifiable.”

The idea that we are running out of trees and people will incompetently just watch it happen has been around a long time, but it is far from reality.  Today we get more than 40% of our wood consumption from plantations, of which there are over 130 million hectares globally (Mishra et al. 2021; McEwan et al. 2020; FAO 2020). People have responded to higher prices by planting trees as an investment. These trees suck up carbon and do it before the tree is harvested. They are not perfect environmentally, but they are renewable, as is the forestry sector as a whole (Mendelsohn and Sohngen 2019).

Biophysical models have no way to capture the behavioral response of landowners to market signals, like rising prices, so they ignore it. This means they get harvesting and regenerating wrong – by lots.

Second, the Peng et al. article is just an implementation of the incorrect argument by Searchinger et al. (2009) that emissions from timber harvesting and burning should be double counted. Favero, Daigneault, and Sohngen (2020) and recently Li, Sohngen, and Tian (2022) showed in different ways that Searchinger’s argument is wrong. Double counting emissions, in contradiction to the correct approach by IPCC, leads to less not more forests, just like higher taxes lead to less production of the good taxed. Peng et al. create a calculation of carbon emissions from harvesting which, they hope, will allow the emissions to then be counted a second time.

Third, Peng et al are making a normative judgment about which tons to count. They have a strange accounting procedure that starts counting gross emissions and gross sequestration at the time of the timber harvest rather than at the initial period in the model run. So they have decided to ignore the growth in forests that happens before trees are cut. Since a large (>40%), and growing, portion of wood cut by industrial markets is dependent on plantations planted for future harvesting, why not count this growth before the harvest? The reason is that this growth would negate a lot of the negative effect Peng et al. calculate, so they make a normative decision to ignore tree growth before harvesting. This convention is different from every other forest sector model.

Fourth, their approach to discounting is just strange. They use a mixture of positive discounting and no discounting together, in the same calculation. I don’t know what to make of that. I guess in the post-truth era, scientists now can do whatever they want. But their discounting amplifies their results and ignores how markets respond to changes in interest rates. So with this strange (also normative) approach, they get a bigger result.

Fifth, their counterfactual is unrealistic, and not just because it assumes no harvesting of wood. It’s weird because if they used an economic model, the carbon implications wouldn’t be so simple to calculate. There is a whole discussion out there about leakage when people stop harvesting trees to store carbon, and it has been around for quite a while (Murray, McCarl, and Lee 2004; Sohngen and Brown 2004). How can anyone do a scenario of no harvesting of wood without considering the market response?

Dave Wear and Brian Murray famously showed what happened when timber harvesting was stopped in federal forests in the United States (Wear and Murray 2004). For those who don’t want to read this really good paper, the short story is that they show the assumptions Peng et al make that you can evaluate the carbon consequences of a no harvest scenario by just looking at the site where you stopped logging are completely false. Peng et al. may try to argue that Wear and Murray is just a model result, so not real, but Wear and Murray is an empirical result, with real data and good statistics. Those of us who develop models of the forest sector create our models so that the same types of equilibrium conditions Wear and Murray rely on are met in our models. Peng et al. seem unaware of any of this, ignore links between timber stands over time and space, and ignore market equilibrium.

There are other problems with Peng et al., of course, but the ones above are the bigger ones. No doubt, the press will continue loving what Peng et al. estimate because it sounds big and problematic, when it’s just a restatement of an earlier incorrect argument. Hopefully, though, in the policy arena, real science will prevail.




Adams, Darius M., and Richard W. Haynes. 1980. “The 1980 Softwood Timber Assessment Market Model: Structure, Projections, and Policy Simulations.” Forest Science 26 (suppl_1): a0001-z0001.

Clawson, Marion. 1979. “Forests in the Long Sweep of American History.” Science 204 (4398): 1168–74.

FAO. 2020. “Global Forest Resources Assessment 2020 Main Report.” Rome: United Nations Food and Agricultural Organization. https://doi.org/10.4060/ca9825en.

Favero, Alice, Adam Daigneault, and Brent Sohngen. 2020. “Forests: Carbon Sequestration, Biomass Energy, or Both?” Science Advances 6: eaay6792.

Li, Rong, Brent Sohngen, and Xiaohui Tian. 2022. “Efficiency of Forest Carbon Policies at Intensive and Extensive Margins.” American Journal of Agricultural Economics 104 (4): 1243–67.

McEwan, Andrew, Enrico Marchi, Raffaele Spinelli, and Michal Brink. 2020. “Past, Present and Future of Industrial Plantation Forestry and Implication on Future Timber Harvesting Technology.” Journal of Forestry Research 31: 339–51.

Meadows, Donnela, Dennis L Meadows, Jorgen Randers, and William W Behrens III. 1972. The Limits to Growth. New York: Signet.

Mendelsohn, Robert, and Brent Sohngen. 2019. “The Net Carbon Emissions from Historic Land Use and Land Use Change.” Journal of Forest Economics 34 (2).

Mishra, Abhijeet, Florian Humpenöder, Jan Philipp Dietrich, Benjamin Leon Bodirsky, Brent Sohngen, Christopher PO Reyer, Hermann Lotze-Campen, and Alexander Popp. 2021. “Estimating Global Land System Impacts of Timber Plantations Using MAgPIE 4.3. 5.” Geoscientific Model Development 14 (10): 6467–94.

Murray, Brian C., Bruce A. McCarl, and Heng-Chi Lee. 2004. “Estimating Leakage from Forest Carbon Sequestration Programs.” Land Economics 80 (1): 109–24.

Nabuurs, GJ, R Mrabet, AA Hatab, M Bustamante, H Clark, P Havlik, J House, et al. 2022. “Agriculture, Forest and Other Land Uses (AFOLU).” In Climate Change 2022: Mitigation of Climate Change. Vol. Sixth Assessment Report. Intergovernmental Panel on Climate Change Working Group III. https://www.ipcc.ch/assessment-report/ar6/.

Searchinger, Timothy D., Steven P. Hamburg, Jerry Melillo, William Chameides, Petr Havlik, Daniel M. Kammen, Gene E. Likens, Ruben N. Lubowski, Michael Obersteiner, and Michael Oppenheimer. 2009. “Fixing a Critical Climate Accounting Error.” Science 326 (5952): 527–28.

Sohngen, Brent, and Sandra Brown. 2004. “Measuring Leakage from Carbon Projects in Open Economies: A Stop Timber Harvesting Project in Bolivia as a Case Study.” Canadian Journal of Forest Research 34 (4): 829–39. https://doi.org/10.1139/x03-249.

Wear, David N., and Brian C. Murray. 2004. “Federal Timber Restrictions, Interregional Spillovers, and the Impact on US Softwood Markets.” Journal of Environmental Economics and Management 47 (2): 307–30.




Short-Term Carbon Storage

Brent Sohngen, Department of Agricultural, Environmental and Development Economics, Ohio State University (sohngen.1@osu.edu)

For too many years, scientists and environmentalists have owned the discussion of short-term carbon storage, sowing confusion on an otherwise ordinary economic principle. The economic principle at play is renting versus owning.  Just about any asset, carbon included, can be rented or owned.

Consider this, when you fly to a vacation destination, you don’t have to buy a house because it is quite easy these days to rent one for the week. If you are an aspiring farmer who can’t afford the high price of buying farmland in the United States, you can join other farmers who annually rent about 40% of US farmland to produce crops. Chances are good that the last time you flew commercially, you did so on a leased aircraft just like the rich and famous do on small private jets. Short-term leases are ubiquitous, helping markets allocate goods and services throughout the economy.

Renting stuff works really well for other assets, why shouldn’t it work for the carbon asset stored in forests and agricultural soils?

The concept of renting carbon has been used to evaluate forest and agricultural carbon sequestration since the early 2000s. The economics of renting is straightforward. The price of any asset is determined as the present value of the stream of revenues associated with owning that asset, where the stream of revenues is the rent. In the case of carbon, the market price of carbon is the asset price. The rental value can be determined directly by using the discount rate.

If the price of carbon at time t is PC(t), and the annual rent is R(t), the economic relationship between the two is

R(t) = PC(t) – PC(t)*exp(-r)

Where r is the discount rate. When the carbon price is $50 and the discount rate is 5%, then the rent on that carbon is $2.44 per year.

Renting carbon is like buying it this year and selling it next year. If you buy a ton of carbon today on a market for $50, and sell it in one year (assuming no depreciation) for the same $50, and your discount rate is 5%, your economic costs of buying and selling that ton are exactly the same as the rental rate:

Costs of buying carbon and selling it a year later = $50  – $50*exp(-r) = $50 – $47.56 = $2.44

A recent paper a few colleagues and I wrote shows how storing carbon for one year like this has value, and how carbon stored for only a year can be used by companies to help them become carbon neutral (see Parisa et al., 2022: https://doi.org/10.1016/j.forpol.2022.102840).

In some cases, if a company wants to become carbon neutral, they may be able to purchase an offset credit from another company, based perhaps on renewable energy, nuclear energy, landfill methane capture, or some other method. However, a big source of relatively low-cost offset credits lies in forests and agricultural soils, both of which provide mainly temporary storage. Forests are temporary because they are susceptible to natural disturbance and future harvest, while agricultural soils are temporary because farmers frequently change their land use or management practices.

But now, with the study by Parisa et al. (2022), there is a clear pathway to treat short-term carbon storage on an equal basis with carbon emissions. To make sure that short-term storage and carbon emissions have equal value, Parisa et al. show that the straightforward answer is to hold multiple tons of short-term storage to equal 1 ton of carbon emission.

Parisa et al.’s paper works out the exact number of tons that need to be held for 1 year at a given discount rate to equal the value of 1 ton of C emissions from energy combustion. If the interest rate is 5%, then someone has to hold 20.5 tons for one year to have equivalent value as one ton emitted.

This means that a farmer who does conservation tillage this year and stores 41 tons for the year offsets the damages caused by 2 tons of CO2 emitted (20.5 tons for 1 year = 1 ton emitted and 41 tons for 1 year = 2 tons emitted). If the price of carbon is $50 per ton, then the farmer could be paid $100, or $2.44 per ton ($100/41 tons =$2.44 per ton), for their year of storage.

The math would work the same for trees, wetlands, or any other ecosystem warehouse of carbon storage. Under different discount rates, the annual rent for carbon would change, as would the number of tons that have to be held to equal a ton of emissions (see table below)

Table: Number of tons that need to be stored for 1 year to equal the value of 1 ton of CO2 emitted under alternative discount rates.

Discount rate

Tons stored 1 year





















By these calculations, if you have a farm or forest and you defer a timber harvest, reduce your tillage, or plant a cover crop, you now know exactly how much benefit your action provides society. Specifically, if your discount rate is 5%, and you hold 20.5 tons out of the atmosphere for just one year, you have offset the damages caused by 1 ton of your own or someone else’s emissions. With ecosystem storage (in forests, soils, grasslands, or wetlands) you only have to store the carbon for one year to have that benefit.

With short-term carbon storage, you can choose to adopt the new practice as long as you want, providing benefits the whole time. If you choose to store carbon tons for more than one year, you increase the carbon benefit you provide. Storing the carbon for 2 years provides the same benefit the second year as the first, meaning storing 20.5 tons for a second-year offsets the damages caused by 1 additional ton of your or someone else’s emissions. As a result, you can be paid the second year for the same tons. Similarly, storing it for 5 years means you can be paid the carbon price in each of the 5 years.

Moving towards efficient mechanisms to mitigate climate change with short-term storage like this is critical for solving the climate problem. Studies like Austin et al. (2020: https://www.nature.com/articles/s41467-020-19578-z) have estimated the costs of forest carbon storage assuming that markets properly price short-term storage in forests and agricultural soils. This and other similar studies show that there is quite a bit of potential to ramp up carbon sequestration on the landscape at low prices.

Unfortunately, the main crediting agencies, like Verra, American Carbon Registry (ACR), and the California Air Resources Board, have ignored the rental and short-term carbon storage approach in Austin et al. (2020) and Parisa et al. (2022). Instead, they have implemented approaches that rely on models of carbon rather than actual measured carbon, and approaches that rely on long-term contracts.

Environmental groups often bolster their arguments about the importance of fighting climate change using new estimates of the costs of forest carbon abatement in studies like Austin et al. (2020), and recent compilations of the earlier literature on costs such as in Griscom et al (2017) and Fargione et al. (2018). These studies make climate mitigation look cheap after all, suggesting that society should just get to it. However, many environmental groups then argue for crediting rules in the land-based sector that make land-based options hundreds of times more costly than estimated.

The results in Parisa et al. (2022) provide landowners and carbon markets with the assurance that their efforts to provide atmospheric benefits through short-term storage both work, and have atmospheric value. By providing a clear trade-off between short-term tons stored and carbon emissions, and basing the tradeoff on tons that are readily observed in ecosystems, offset markets can flourish. Ultimately, they can grow in scale to create the level of atmospheric benefits estimated in the many studies that have shown them to be low-cost options for climate mitigation.

Global Timber Model

This page hosts code, working papers, and lists of published papers developed with the Global Timber Model.  The Global Timber Model is a dynamic optimization model of global forests, used for analysis of policy questions.  Code for various papers will be deposited here and is freely available for use.  If you have questions, please contact Brent Sohngen (sohngen.1@osu.edu).