1. Environment
  2. Methodology
Download icon

The Complexity of Incorporating Carbon Social Returns in Farm Afforestation: A Microsimulation Approach

  1. Mary Ryan  Is a corresponding author
  2. Cathal O’Donoghue
  1. Teagasc, Ireland
  2. National University of Ireland, Ireland
Research article
Volume 14(1) Spring 2021
Cite this article as: M. Ryan, C. O’Donoghue; 2021; The Complexity of Incorporating Carbon Social Returns in Farm Afforestation: A Microsimulation Approach; International Journal of Microsimulation; 14(1); 102-134. doi: 10.34196/ijm.00232
6 figures and 16 tables

Figures

Figure 1
Download asset Open asset
Structure of the Forestry Land use Change Microsimulation Model
Figure 2
Download asset Open asset
Total carbon (C) storage curves for unthinned (u) and thinned (t) Sitka spruce yield class 8, 16 and 24
Figure 3
Download asset Open asset
Carbon (C) Sequestration/Loss for No Thin (NT) Yield Class 18 over 200 years
Figure C1
Download asset Open asset
Comparison of Above Ground tCO 2 (C-ForBES and Teagasc Forest Carbon Tool) over one rotation
Figure C2
Download asset Open asset
Impact of different approaches to determine the wood inputs to HWP on SW, WBP, PPB Carbon Liberation Curves for NIR (2018) WBP, PPB Carbon Liberation Curves for NIR (2018) approach, C-ForBES harvest losses (HL) and C-ForBES harvest losses + wood energy losses (HL Energy)
Figure C3
Download asset Open asset
Harvested Wood Flow Carbon Liberation Curves: Comparison of NIR (2018) (SW, WBP, PPB) with Bateman & Lovett (2000) (Bateman SW, Bateman WBP, Bateman PPB)

Tables

Table 1
Agricultural Carbon Emission Factors and GHG equivalent conversions
Energy CO 2 Energy CH 4 Energy CH 4 CH 4 Agri CH 4 Total CO 2 Total CO 2e
tCO 2/€ tCH 4/€ tCO 2e/€ tCH 4/head/yr tCH 4/head/yr tCO 2e/head tCO 2e/LU
Dairy 0.124 0.00012 3.129 3.129
Cattle 0.051 0.00013 1.308 2.975
Sheep 0.006 0.00001 0.152 0.684
Horses 0.020 0.00015 0.544 1.238
Pigs 0.006 0.00003 0.262 1.176
Poultry 0.000 0.000001 0.00590 0.006
Fuel 0.003 0.00000154 0.00000015
Fertiliser 0.000032
Crops 0.00000041
CO 2 equiv conversion factor 1 25 298 25 298

  1. Source: Common Reporting Framework 2015[6]

Table 2
Private Returns to Planting SS Forest in 2015 (annual equivalised (AE) of Average NPV per ha) at 5% discount rate
Yield Class Average AE of NPV (€ h a − 1 )
24 563
20 473
18 425
14 355

  1. Source: C-ForBES

  2. Carbon Models, Assumptions and Validation

Table 3
Average Private Returns (Gross Margin € ha −1) to Agriculture by Soil Code and Farm System (2015)
SC YC Specialist Dairy Cattle Rearing Cattle Other Sheep Tillage
1 24 1566 1029 996 954 1098
2 24 1181 855 882 1066 1186
3 20 1145 766 764 740 1103
4 20 1082 656 878 859 803
5 18 670 497 550 471
6 14 584 583 356 681
All All 1237 784 867 811 1114

  1. Source: Teagasc NFS (2015)

  2. Note: SC: Teagasc NFS Soil Code (1 – Best, 6 – Worst), YC: Forest Yield Class

Table 4
Economic Components of Agriculture and Forestry (2015) (Annual Equivalised NPV per ha (€)
Soil Code Agriculture Forestry
Mkt Gross Margin /ha (A) Subisides /ha (B) Mkt Gross Margin /ha (©) Subsidies /ha (D)
€ € € €
SC1/YC24 1200 366 224 306
SC2/YC22 792 388 224 306
SC3/YC20 803 342 154 302
SC4/YC18 731 351 154 302
SC5/YC16 356 314 124 300
SC6/YC14 258 326 52 298
Average 878 359 155 296
Table 5
Average Annual Equivalised Social Return to Planting one hectare of Unthinned Forest (2015) Displacing Agriculture
Soil Code Private Return10 Agriculture Forestry Social Return
(C+D) – (A+B) (€/ha) E11 F12 C – (A+B) + (F-E)*P13 (€/ha)
tCO 2 Value (P) €0 tCO 2 tCO 2 €20 €32 €100 €163
SC1 1036 -9.2 14.9 -556 -268 1365 2878
SC2 651 -8.4 14.9 -185 95 1680 3148
SC3 690 -7.5 11.8 -304 -72 1239 2455
SC4 627 -7.4 11.8 -242 -11 1298 2511
SC5 246 -4.5 10.8 59 242 1281 2243
SC6 234 -4.9 7.8 19 171 1031 1828
Total -8.0 13.4 -359 -103 1347 2691

  1. Note: No-thinning assumed; BEF Factors from the 2015/18 National Inventory Reports are used;

  2. All values are discounted using a 5% discount rate. Sensitivity Analyses

Table 6
Comparing the Annual Equivalised tCO 2 for Thinned Forests by Soil Type if all thinnings are used for energy
Soil Code Equal Allocation from Thinnings and Clearfell to Energy Thinnings only Allocated to Energy
1 10.77 9.60
2 10.77 9.60
3 9.04 8.16
4 9.04 8.16
5 9.14 8.37
6 5.37 4.83
Average 9.98 8.95

  1. Note: 2015/18 National Inventory Report Assumptions

Table 7
Total Annual Equivalised tCO 2 Forest Carbon Sequestration for Mineral and Peat Soils by Yield Class
Yield Class Mineral Peat
24 12.69 10.56
24 12.69 10.56
20 10.44 8.32
20 10.44 8.32
18 9.34 7.21
14 7.60 5.48
Average 11.56 9.43

  1. Note: For simplicity purposes in peat calculations, these values are calculated using EFs from NIR (2012)

Table 8
Share of Farms with greater social returns from Forestry than from Agriculture (Annual Equivalised NPV-2015) for a range of Discount Rates
Discount Rate 0 1 2 3 4 5 6 7
Carbon Value (€): 32
Soil Code
1 1.000 0.998 0.958 0.846 0.720 0.643 0.597 0.563
2 1.000 1.000 0.991 0.940 0.923 0.895 0.853 0.824
3 1.000 1.000 0.955 0.868 0.805 0.755 0.736 0.724
4 1.000 1.000 0.969 0.922 0.844 0.814 0.779 0.741
5 1.000 1.000 0.979 0.932 0.919 0.912 0.912 0.903
6 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000
Total 1.000 0.999 0.969 0.892 0.820 0.774 0.739 0.712

  1. Distribution of Private and Social Returns

Table 9
Share of Farms with a positive Private and Social return to Forestry by BEF Assumption and inclusion of Forest Subsidy
Private Return Social Return
Carbon value (€) 0 20 32 100 163
BEF (NIR, 2012)
Incl Farm Subsidy 0.324 0.264 0.427 0.943 0.998
Excl Farm Subsidy 0.551 0.577 0.672 0.983 0.999
BEF (NIR, 2015)
Incl Farm Subsidy 0.324 0.304 0.466 0.965 0.999
Excl Farm Subsidy 0.551 0.594 0.697 0.991 0.999

  1. Impact of substitution of trees for livestock

Table 10
Area of livestock production that could be offset by one hectare of SS, at different livestock densities (LUha−1), with/without accounting for land use change (LUC) by soil code
Soil Code Stocking Rate Area (ha)offlivestock displaced by one ha of forest Area (ha) of livestock displaced by one ha of forest LU displaced per Ha of Forest LU displaced per ha of Forest LU per Ha of Forest
LU per Ha No LUC (not accounting for displaced emissions) With LUC (accounting for displaced emissions) No LUC With LUC(No Thin) With LUC (Thin)
1 1.57 1.62 2.62 2.56 4.13 3.03
2 1.52 1.76 2.76 2.67 4.18 3.03
3 1.33 1.59 2.59 2.12 3.45 2.69
4 1.36 1.59 2.59 2.17 3.52 2.75
5 0.87 2.38 3.38 2.06 2.93 2.29
6 0.90 1.61 2.61 1.44 2.34 1.85
Average 1.41 1.68 2.68 2.38 3.79 2.84
Table A1
Enteric Fermentation and Nitrous Oxide Emission Factors
CH 4 N 2 O
Enteric Fermentation Manure Management Manure Management (kt)
Emission Factor (kg/head CH 4/yr) (kg CH 4/head/yr) (kg/N 2 O/head)
Dairy 113.41 10.30 0.12
Cattle 46.39 4.43 0.13
Sheep 5.61 0.39 0.01
Horses 18.00 1.99 0.15
Pigs 1.33 5.04 0.03
Poultry 0.00 0.22 0.00
Deer and Goats 25.00 1.62 0.12

  1. Source: Common Reporting Framework http://www.epa.ie/pubs/reports/air/airemissions/ghg/nir2015/

Table A2
Average LU per Ha by Soil Code
Soil Code Dairy Cattle Sheep
SC1 2.09 1.74 1.62
SC2 2.03 1.48 2.13
SC3 2.00 1.42 1.66
SC4 1.84 1.35 1.25
SC5 1.58 0.73 0.42
SC6 1.14 1.69
Average 2.00 1.45 1.28

  1. Source: C-ForBES

Table A3
Average tCO2 per hectare by source of emissions by Soil Code
Soil Code Dairy Cattle Sheep Fuel Fertiliser Crops
SC1 1.77 4.00 0.21 0.000207 0.001297 0.000019
SC2 0.96 4.00 0.79 0.000190 0.001071 0.000018
SC3 1.15 3.79 0.56 0.000221 0.000937 0.000008
SC4 1.31 3.58 0.38 0.000219 0.000853 0.000003
SC5 0.36 1.42 0.41 0.000080 0.000303 0.000002
SC6 0.00 2.80 0.68 0.000107 0.000378 0.000001
Average 1.24 3.57 0.45 0.000191 0.000993 0.000012

  1. Source: C-ForBES

Table B1
C-ForBES Modelling Assumptions
Model component C-ForBES parameters and assumptions Comparisons with parameters/assumptions in NIR (2018) and Bateman and Lovett (2000) Additional notes
Scale of analysis Per hectare net carbon storage or emissions in a given year. Bateman and Lovett (2000): Per hectare net carbon storage or emissions in one year.NIR (2018): National Carbon Stock Change (CSC) over time.  
Period of analysis 200 years. To accommodate approx. 4 rotations of SS (dep on YC) to allow for consideration of half-life of Harvested Wood Products (HWP). Replanting is assumed in the year following harvest. Bateman and Lovett (2000) model extends to 1000 yearsNIR (2018): Annual CSC 1990 – 2016  
Yield Class (YC) Yield classes14-24 are analysed. YC14 is lower bound for eligibility for afforestation grants and subsidies. NIR (2018) reports across all historic and current YCs.Bateman and Lovett (2000) estimate YC 6 – 24.  
[1] C-ForBES Forest ManagementAssumptions derived from Teagasc FIVE (see Ryan et al., 2018) [1a] Tree species:Forest Investment and Valuation Estimator (FIVE):Sitka spruce (SS), which is the most commonly planted conifer in Ireland.To reduce complexity a simplifying assumption is made to model one hectare of pure SS (see [2] and [11]) as cost, growth and price data in FIVE are not as robust for mixtures of broadleaf and conifers.However, productive area is modelled at 85% to account for broadleaf component. In this analysis therefore, carbon sequestration from broadleaves is not modelled. NIR (2018) estimates CSC for all species planted in Ireland.Bateman and Lovett (2000) model and map C storage for SS and Beech in Wales “The forest must contain a minimum of 15% broadleaves by area. This can comprise: broadleaves planted in broadleaf GPC plots of minimum width; and/or broadleaves planted as part of the 'at least 10% diverse' requirement for GPC 3; and/or additional broadleaves planted for environmental and landscape reasons”.www.teagasc.ie/forestry/Grants
  [1b] Forest Yield:Merchantable Timber Volume (MTV)(Edwards and Christie, 1981) NIR (2018) is based on a range of models including FORCARB (based on Edwards and Christie, 1981), CARBWARE & Carbon Budget Model (CBM) (NIR, 2019).Bateman and Lovett (2000): MTV from (Edwards and Christie, 1981) and combine with data on carbon storage in Sitka spruce (Cannell and Cape, 1991) to plot thin and no-thin carbon storage curves.  
  [1c] Thinning:Marginal Thinning Intensity (MTI)15 @5 year intervals from Edwards and Christie (1981) NIR (2018)/Bateman and Lovett (2000) also use this static thinning assumption.  
  [1d] Rotation:‘Reduced rotation’ = (Age of max MAI – 20%)16 (Phillips, 1998; Anon, 1977, ). NIR (2018)rotation = max MAI from CARBWARE (Black, 2016).Bateman and Lovett (2000) estimates felling year (F) based on age of max NPV for given species, YC and discount rate – see Bateman (1996)  
[2] ForSubs model (Ryan et al. (2016) Forest subsidy:General Planting Category GPC3 (10% Diverse Conifer, e.g. Sitka spruce and 10% broadleaves)€510/ha/year - paid annually for 15 years for first rotation only GPC 3 – 10% Diverse Conifer/Broadleaf:Comprises of a mix of Sitka Spruce/Lodgepole pine together with at least 10% Diverse conifer (approved conifer other than SS/LP). Broadleaves adjacent to roads and watercourses may also form part of this 10% www.teagasc.ie/forestry/grants
[3] CostsTeagasc FIVE (Ryan et al., 2018) Ground preparation, fencing, planting, maintenance, insurance, replanting.  
[4] Timber prices Coillte (State forestry body) 10 year average timber pricesAnnual timber price series published annually by Irish Timber Growers Association (ITGA) (see Symons et al. (1994); Teagasc (2019))  
[5] NPV discount rate The conventional discount rate used for forestry in Ireland is 5% (Clinch, 1999). Bateman and Lovett (2000) also use a discount rate of 5% for forest NPVs.  
[6] Soil organic matter (SOM) Analysis assumes land use change on grassland on mineral soil only with no change in SOC.In the sensitivity analysis of planting on mineral or peat soils, the coefficients from NIR (2012) are used. NIR (2018) assumes (a) no carbon stock change on the planting of forests on mineral soils and (b) a mean organic soil EF of 0.59 t C/ha/year over the first rotation (50 years) as organic soils are not a source following successive rotations (Byrne and Farrell, 2005).Bateman and Lovett (2000) assume long term net gain of soil carbon (50 t C ha-1 on mineral soils) or loss (750 t C ha-1 on peat soils) occurring within 200 years. NIR (2018) categorises Irish soils into three major groupings based on soil carbon characteristics. All mineral soils are grouped together, while all organic soils with an organic layer greater than 30 cm are classified as peat. Finally, organic soils with an organic layer less than 30 cm are classified as peaty/mineral.CARBINE (Temperli et al., 2020): V Changes in soil carbon are assumed to take place in response to land use change. Magnitude and changes over time are estimated according to soil type (texture) and major land use category.
[7] Early growth We use a logistic function to interpolate early growth and the growth in 5 year intervals recorded in Edwards and Christie (1981) models. NIR (2018) uses a modified expo-linear growth function (Monteith, 2002) to simulate early annual growth.Bateman and Lovett (2000) fitted an S shaped curve to Edwards and Christie (1981) data  
Carbon mass of Sitka spruce (SS) [8] Basic density 0.387 (NIR, 2012, p. 123 ) [9] Carbon fraction 0.5 (NIR, 2015, p. 123 )  
Biomass – above ground [10] Biomas Expansion Factor (BEF) follows NIR (2018) methodology. NIR (2015). A dynamic BEF is used in this analysis based on species, yield class and growth phase. Ranging from a value of 2 to 1.68 for lower YCs (14 & 16), 3 to 1.68 for YCs 18 & 20, 4 to 1.68 for the most productive YCs (22 & 24). A constant BEF of 1.68 is utilised once stand volume is equal to or greater than 200m 3 ha− 1 Bateman and Lovett (2000).estimated functional relationships for livewood. MTV is related to total woody volume (TWV) by allowing for branchwood, roots, etc. (Corbyn et al., 1988). NIR (2018): Based on the model developed by Dewar and Cannell (1992), (Kilbride et al., 1999) used a static value of 1.3 for all species, age and yield classes, while the 2012 NIR uses a value of 1.64. However, since the allocation of biomass between different forest components is dependent on species, yield class and the growth phase of the forest, current estimates of sink capacity have been revised to use age and species-specific BEF values that include the below ground fraction.
[11] Productive area 85% of the area taken out of agriculture is classified as productive area due to mandatory areas of biodiversity enhancement (ABE), set-back distances for roads, rivers, houses, fencing, unplantable terrain etc., (Ryan et al., 2018). In scaling up, NIR (2018) applies a 10% area reduction to account for open spaces.  
Biomass – below ground [12] Ratio of below ground to above ground biomass: 0.2Country specific ratio (NIR, 2015)  
DOM Litter [13 ] LLF represents the transfer of carbon from the above ground pool to the litter pool. It is simulated using derived leaf/needle biomass (LB) and the foliage turnover rates (Ft) from Thorne and Fingleton (2006):LLF = LB × F t The F t rate is assumed to be 6.7 years for conifer crops and 1 year for broadleaf crops (Thorne and Fingleton, 2006). Needle biomass is calculated according to the equation defined in Annex 3.4.A.4 of NIR (2018):LB = 0.025 × AB + 0.089 × exp (−0.003 × AB)The litterfall LLF is assumed to decompose at a rate of 14% per year (NIR (2018) p 222). NIR, 2018 – p197: Biomass carbon losses from the above ground biomass pool are calculated based on harvest (Ltimber), harvest residue (LHR), litter fall (LLF), above ground losses due to mortality (Lmort(AB)) and fire (Lfire):Ltimber is calculated based on the above ground biomass removed from harvest,LHR includes the harvest residue representing all stems and branches with a DBH less than 7cm and litter left on site after timber is removedLLF reflects the transfer of carbon from the AB pool to the litter poolNIR (2018): Equation from NIR (Thorne and Fingleton, 2006) (needle turn- over is 6.7 years for conifers and annually for broadleavesCalculation of matter from equation in NIR, 2018Decomposition = 14% decline/yr p222 NIR, 2018 Bateman and Lovett (2000): litter is not modelled CARBINE (Temperli et al., 2020): Litter is not modelled
  Deadwood [14]Inflow is 1.6%Decomposition rate is 14% decline /year (Carbware) NIR (2018) Mortality:Growth, harvest and mortality derived from Edwards and Christie (1981) described by Black et al. (2012).Net deadwood stock changes (CDW) are derived from carbon inputs associated with timber extraction residue (Ltr), timber from mortality (Mtimber), dead roots from mortality (Lmort(BB)), roots from harvest (LHRroot) and carbon loss due to decomposition of the new and previously existing deadwood pool (DDW):Biomass carbon losses from the below ground biomass pool are calculated as the sum of losses due to death of roots after harvest (LHRroot), natural mortality of roots (Lmort(BB)) and root death following fire (Lfire).Bateman and Lovett (2000): Assume 5 year oxidation of deadwood  
[15] Harvest losses We assume differential harvest losses for each harvest as per Teagasc FIVE.Ist Thin – 14% loss of Merchantable Timber Volume (MTV)2nd Thin - 12% loss of MTVSubsequent Thin - 9%Clearfell/Final harvest – 5% NIR (2018) assumes static harvest losses of 4% Morison et al. (2012) include HL in HWP as they may not be immediately oxidised
Wood fuel oxidation [16] 34%In the 2017 Wood Flow report, 34% of forest biomass is used as wood fuel (Knaggs and O’Driscoll, 2017).  
HWP allocations [17] ForBES follows Pingoud & Wagner (2006) model (2006 IPCC guidelines)Assumes the same allocation of wood to the value chain from thinnings and from final harvest (scenario analysis to examine alternative scenario).Allocation of SW, WBP and PPB to HWPSW: 52%, WBP 48%.PPB: zero (no longer any paper production). Phimmavong and Keenan (2020) model (2006 IPCC guidelines) The UK CARBINE model (Temperli et al., 2020) allocates MTV to HWP pool (long-lived sawnwood, short-lived sawnwood, particleboard and paper), with remainder to waste.
Saw-milling losses [18] SW: 50%WBP: 41%  
IPCC conversion factor [19] Conversion factor from C to CO 2: 3.67 3.67 A cost of USD 1 per tonne of carbon dioxide is equivalent to a cost of USD 3.67 per tonne of carbon. OECD
Carbon Valuation [20] Carbon values applied as per Irish Government shadow price of carbon for 2019, 2020, 2030, 2040.Future carbon sequestration and emissions are discounted at 5%. NIR (2018) does not discount future carbon sequestration or emissions.Bateman and Lovett (2000) discount carbon at 5% and include scenario analysis for discount rates of 2 – 12%  
Annual Equivalised (AE) NPV per hectare [21] A E = r . N P V 1 - ( 1 + r ) - n Assume no agricultural income in year of planting  
Table C1
Harvested Wood Flow Carbon Liberation Curves: Comparison of NIR (2018) (SW, WBP, PPB) with Bateman & Lovett (2000) (Bateman SW, Bateman WBP, Bateman PPB)
Bateman 0 0 0 1
Harvest Loss 0 1 1 0
Energy 0 0 1 0
Sawnwood (SW) 0.255 0.242 0.160 0.253
Wood-based panels (WBP) 0.186 0.177 0.117 0.072
Paper and paper board (PPB) 0.020 0.019 0.012 0.015
Table C2
Comparing the Annual Equivalised tCO2 for Thin and No Thin by Soil Code/Yield Class for 2012 and 2015/18 Biomass Expansion Factor (BEF) National Inventory Report Assumptions
Soil Code (SC)/Yield Class (YC) 2012 BEF 2015/18 BEF Ratio 2012 BEF 2015/18 BEF Ratio
No Thin No Thin No Thin Thin Thin Thin
SC1/YC24 12.69 14.86 1.17 8.46 10.77 1.27
SC2YC/24 12.69 14.86 1.17 8.46 10.77 1.27
SC3/YC20 10.44 11.83 1.13 7.58 9.04 1.19
SC4/YC20 10.44 11.83 1.13 7.58 9.04 1.19
SC5/YC18 9.34 10.75 1.15 7.43 9.14 1.23
SC6/YC14 7.60 7.80 1.03 5.14 5.37 1.04
Average 11.56 13.37 1.16 8.04 9.98 1.24

  1. Note BEF – Biomass expansion Factor

Data and code availability

The data used in this model is publicly available. The farm level data (National Farm Survey) is accessible via the Irish Social Science Data Archive, while the growth curves used in the model are available online in the referenced documentation. All other parameters used in the model have been documented in the extensive documentation provided within this and referenced papers.

The models have been coded in Stata. While they are not currently stored in the public domain (primarily because they haven’t been documented to a standard to facilitate such dissemination), we are happy to share code for the algorithms and indeed have documented the main algorithms and modelling choices made in this paper and other cited references.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)