Trees or Trotters?

TREES OR TROTTERS? JASMINA BEHAN, KIERAN MCQUINN, AND MAURICE J. ROCHE A real options model is used to explain why Irish farmers have been slow to switch from traditional farming to forestry despite numerous government incentives. In the theoretical model our results depend on profits from traditional farming relative to forestry. Under reasonable parameterisations of this profit ratio we show that it is optimal for farmers to stay in farming for six years before switching to forestry. In a subsequent empirical dynamic panel data model, the error correction model also predicts that it would take about six years for a change in the profit ratio to fully affect the number of hectares planted. Key words: error correction, forestry, panel data, real options Contact author Dr. Maurice J. Roche Department of Economics, National University of Ireland Maynooth, County Kildare, Ireland. Telephone: 353-1-7083786 Fax: 353-1-7083934 Email: [email protected] Jasmina Behan is an economist with FAS, The Training & Employment Authority, 27/33 Upper Baggot Street, Dublin 4, Ireland. Kieran Mc Quinn is an economist with the Central Bank of Ireland, Dame Street, Dublin 2, Ireland. Maurice J. Roche is a senior lecturer at the Department of Economics, National University of Ireland Maynooth, Co. Kildare, Ireland. The research was support in part by Research Stimulus Funds allocated under the Irish National Development Plan 2000-2006. The views expressed in this paper are the personal responsibility of the authors. They are not necessarily held by FAS, the Central Bank of Ireland or the ESCB. The Irish government, since 1986, has been aggressively encouraging farmers to switch from traditional agriculture1 to forestry. This switch has been promoted through a scheme of grants and premium payments initiated by the European Union (EU) under the Agricultural Development Programme for the West of Ireland – referred to as the “Western Package”. The government’s policy to increase forestry cover has been further influenced by the adoption of the Kyoto Protocol (see UNFCCC (1997) for details) in which national targets of greenhouse gas emission levels are established for participating countries. Significant attention has begun to focus on the ability of the agricultural/forestry sector to reduce the level of greenhouse gases through abatement and sequestration strategies (see for example, McCaul and Schneider (2000) for the US, Behan and McQuinn (2003) for Ireland). Currently, Ireland at 9% has the lowest percentage of land covered by forestry in the EU. The government’s strategic plan for forestry has the objective of increasing forest cover to 17% by 2030 (Department of Agriculture Food and Forestry (1996)). This implies a target of a national annual afforestation rate of 20,000 hectares (ha). Despite the level of aid given to farmers to encourage them to plant trees, the national annual average private afforestation rate has increased from almost less than 1,000 ha per annum in 1986 to a little over 14,000 ha per annum in 2001. This is 30% below he desired government target. In this paper, we ask why are Irish afforestation rates below target and what can the government do to encourage farmers to meet this objective? In order to address these questions real options theory to analyse the farmer’s production/investment decision and to motivate a statistical model for afforestation rates. Under realistic parameterisations of the theoretical model, we show that it is 1 optimal for farmers to remain in traditional agriculture for approximately 6 years before switching to forestry. Increasing the present value of a stream of profits from forestry relative to that of traditional agriculture (we label this the profit ratio) reduces waiting time. Thus, one might expect a positive relationship between the numbers of hectares of trees planted and the profit ratio. The statistical model uses dynamic panel data methods that allow for cross sectional heterogeneity in the long-run cointegrating vectors and in the dynamics associated with short-run deviations from these cointegrating vectors. We find that a 1% increase in the profit ratio leads to a 2.2% increase in the afforestation rate in the long-run and that it takes about 6 years for full adjustment to occur. The econometric results seem to conform to our predictions from the real options model. Our results indicate that the Irish government would need to increase the profit ratio by approximately 20% to increase the number of hectares planted by farmers from 14,000 ha per annum to the target of 20,000 ha per annum. Background Since the independence of the Irish State in 1921 until the mid 1980s the number of hectares of forest planted per annum was small and has been almost exclusively public i.e. conducted by state run agencies. Given an average afforestation rate of 5,000 ha per annum, forest cover had only increased from 1 to 5 percent over the period 1921 to 1980. It was not until the 1980s that private planting by farmers began to significantly contribute to the level of forest cover. A more aggressive approach was adopted by the national government enhanced by funds provided for forestry by the then EEC (“Western Package” in 1981 and Regulation 297/85). Forestry was 2 promoted as a profit generating economic activity. Given this stated policy shift, a series of financial incentives, grants and premia, were introduced at a national level to encourage farmers to consider forestry as an alternative farm enterprise. The values, in euro, of planting grants and premia are presented in Table 1. The planting grants and premia have increased by over 300% since 1986. The forestry grant is paid in two instalments. The first instalment is paid in the first year after the plantation is established and it amounts to 75 percent of the total grant aid. The remaining 25 percent of the grant is paid after four years, when the plantation is assessed as to whether it complies with the Irish Forest Service’s2 standards in terms of stocking levels, fertility of the plantation (level of nourishment), fencing, drains, formative shaping etc. The forestry premium is paid to farmers annually, for a period of 20 years. Under current legislation, farmers are obliged to replant their land, which implies that planted land is bound in forestry in perpetuity. The premium is not paid for reforestation. There are also additional schemes, which the farmer-forester can avail of such as a road construction grant and reconstruction of woodlands scheme. In Figure 1 we present afforestation rates for five regions in Ireland over the period 1986-2001.3 The planting conducted by farmers has become the key driver for forestry expansion in recent years, with farmer planting currently accounting for 93 percent of total plantings. While there has been a steady increase in regional afforestation rates since 1986, there have been peaks and troughs. These swings broadly coincide with major changes in forestry premia and reforms under the EU’s Common Agricultural Policy (CAP). However, despite the financial incentives, the 3 uptake by farmers has not been sufficient to meet the national target of 20,000 ha per annum. We present the profit ratio in Figure 2 (see the Appendix for details on the construction of this ratio for each region). The major increase in the ratio since 1994 is due to two main reasons. The first is that the forestry premium was greatly increased in 1994 under the EU Forestry Development Programme (see table 1). The second is that cattle prices fell significantly over the 1994-1999 period due to the BSE crisis. There would appear to be a similar general upward trend in both regional afforestation rates and the profit ratios. We motivate such a relationship using real options theory in the next section. The Real Options Model For most countries in the EU and North America it is the issue of reforestation (planting on clear felled land) that is primarily of interest to policymakers. In the Republic of Ireland however, the first clear fell of farmer-owned forests is expected to occur beyond 2020. Therefore, in this paper we wish to model a farmer’s decision to switch from traditional agriculture to forestry. Qualitative studies suggest that Irish farmers are reluctant to plant their land with trees and two main reasons are cited (see by Gardiner and Ni Dhubhain (1994), Gillmor (1998), Clinch, McCormack and O’Leary (2000), Frawley and Leavy (2001)). The first is that profits from forestry tended to be smaller than those in agriculture. The second is farmers were worried about the costs and risks associated with forestry and the long-term nature of forestry investment. We try to capture these findings with a suite of real options models. 4 Consider a farmer’s decision to switch from traditional agriculture to forestry. We assume that the farmer is a profit maximizer and operates in a competitive market with certainty. At time t, let A and F be profits from traditional agriculture and forestry respectively. For simplicity we assume that profits from traditional agriculture and forestry grow at constant rates,  and , respectively. The farmer’s constant discount rate is r (r> and r> by assumption). We assume that the farmer faces three options with regard to his/her production decision and chooses the option that maximizes the present discounted value. The first option is to stay in traditional agriculture forever. The value of land per hectare for option 1, V1(0), at time t=0 is given by (1) 1() 0 (0)     rt A VAedt r The second option is to start in traditional agriculture and then switch to forestry at an optimal time T* and remain in forestry forever after that time. In this case the investment decision is deemed irreversible given that farmers are obliged under Irish law to replant their land after its clear felled (see Tegene, Wiebe and Kuhn (1999) and Behan and Wiemers (2003)). There is an annualised cost C per hectare associated with this irreversibility. This is due to the fact that the value of forestry land is less that that of agricultural land. The value of land per hectare for option 2, V2(0), at time t=0 is given by (2) * ** ** 2()() 0 () () (0) 1                 Trt rt rt TT rT rT rT VAedtFedtCedt AF ee rr *   C e r 5 The third option is to convert idle land to forestry and remain in it forever. The value of land per hectare for option 3, V3(0), at time t=0 is given by (3) 3() 00 (0)       rt rt FC VFedtCedt rr  We calculate the value of the three options using parameters from Behan and Wiemers. The value of profits per hectare for dry stock farming and forestry is estimated to be €128 and €390 respectively for 2001. We assume that only dry stock farming should be taken into consideration as it is seen as the agricultural enterprise most likely to compete with forestry. In the last decade, profits from traditional agriculture have been more or less constant. We assume that the growth rate  will be zero. In contrast, profits from forestry have grown at fast and fluctuating rates over the 1986-2001 period. This has been due to the aggressive intervention by the government over the same period. Following discussions with industry representatives (Coillte4, Forest Service, Teagasc5 forest advisors) profits from forestry are expected to grow at more modest rates. We assume that the growth rate  will be 2%. The annualised difference in the value of agricultural and forestry land is estimated to be €320 per hectare (see Behan and Wiemers). The discount rate is assumed to be 5% per annum.6 The discounted values of options 1, 2 and 3 are €2440, €6750 and €6600 respectively. Under this parameterisation, discounted profits are maximized using option 2. In this case, it is optimal for a farmer to wait 6.3 years before switching to forestry. Increasing the profit ratio reduces this waiting time. The Statistical Model 6 The increased focus of Irish government policy on afforestation has also lead to some recent empirical studies exploring the factors determining Irish forestry levels. Barrett and Trace (1999) investigate the relationship between hectares of forest planted and the annual forestry premium, agricultural subsidies net of Rural Environment Protection Scheme (REPS) payments, forestry land prices and relevant CAP policy measures. The fact that none of these variables are found to be statistically significant is probably due to the small sample size used in the analysis.7 McCarthy, Matthews, and Riordan (2002) using panel data analysis relates hectares of forest planted to the forestry grant and premium levels, forestry margins, agricultural margins and area entered in REPS. All variables are found to be statistically significant except agricultural margins. The authors suggest that forestry may be planted on marginal agricultural land, and therefore is relatively impervious to agricultural margins. International empirical investigation of land allocation between forestry and traditional agriculture is characterised by a paucity of efforts. Alig, Adams and McCarl (1998), in a review of agriculture sector models, comment, “sector models used to analyse policies have not included full land base interaction between the forest and agriculture sectors.” In presenting an optimal control model of carbon sequestration and energy abatement, Sohngen and Mendelsohn (2003) state that a “dynamic integrated model that captures both forestland and agricultural land is clearly needed”. What approaches do exist are typified by the net present value approach of Alig, Adams and McCarl, where area allocation is determined by the relative present value of future agriculture and forestry gross margins. These approaches, in line with those of Barret and Trace, McCarthy, Matthews, and 7 Riordan, and Behan and McQuinn, assume a single period instantaneous adjustment by producers to changing market and policy signals. Such a static approach is particularly inappropriate in an Irish context where qualitative research testifies to the reluctance of Irish producers to consider forestry as an alternative to traditional agricultural practices. As noted previously, Gardiner and Ni Dhubhain, Gillmor, Frawley and Leavy, and Clinch, McCormack and O’Leary all present survey data detailing the cultural reticence of Irish producers in adopting farm forestry. Producer behaviour is likely to be characterised by a multi-temporal adjustment to market and policy signals – dynamics, therefore, we contend, should be a key component of any area response function. Using techniques developed by Pedroni (1999, 2000), we present a dynamic panel data model that investigates the short and long run relationships between afforestation rates and the profit ratio. Recall, that the real options models suggests that afforestation rates at time t per region i (Yi,t) depend on the profit ratio per region (Xi,t) and that changes in the profit ratio may take time to take effect. Thus, we hypothesise that the following error-correction model should adequately capture this relationship (4)   ,,1,1,, 1,... 1,...       it it i it it it YYX XetTiN There are T=16 annual observations covering the period 1986-2001 and N=5 regions. We follow Pedroni (1999,2000) and allow the error term to include fixed effects and allow the long-run slope coefficients to vary across regions. We can test whether the latter are equal using Pedroni (2000). The group mean panel t-test statistics developed by Pedroni (1999) allow one to test for the null of a unit root against the alternative hypothesis of a stable root that does not assume a common value across regions. The tests are based on the panel 8 0 0.5 1 1.5 2 2.5 0 1 2 3 4 5 6 7 8 9 101112131415161718192021222324252627282930 Years after shock Percentage change in hectares palnted Figure 3. The impulse response of the numbers of hectares planted following a 1% increase in the profit ratio Table 1. Irish Grant and Premium Payments Per Hectare for 20% Diverse Conifer Plantings on Enclosed Land (euro) Source: Afforestation Grant and Premium Schemes, Forest Service, Department of the Marine and Natural Resources Years Grant Premium 1986-1988 630 94 1989-1991 1,397 109 1992-1993 1,397 127 1994-1997 1,905 286 1998-1999 2,286 343 2000-2001 2,857 391 Table 2. Group Mean Panel Unit Root and Cointegration Tests Test Variable Test-statistic Panel Unit Root Test Afforestation rate -1.47 Panel Unit Root Test Profit ratio -0.65 Panel Cointegration Test -3.11* Notes: The tests are based on the panel group mean fully modified OLS estimator and are statistic number 4 in Table 1 in Pedroni (1999). A * indicates significance at the 1% level. 15 Table 3. Group Mean Panel Estimator of the Cointegrating Coefficient Individual tests Mid-East 1.52 (2.49) [-1.13] North-West 1.80 (2.19) [-0.49] West 1.89 (2.88) [-0.48] South-West 3.18 (4.68) [1.44] South-East 2.53 (3.32) [0.43] Group test 2.18 (6.95) [-0.10] Notes: The t-statistics for H0: i =0 are in the round parentheses and the t-statistics for H0: i =2.2 are in the square parentheses. The tests are based on the panel group mean fully modified OLS estimator in Pedroni (2000). Table 4. Panel Error-correction Model Model 1 Model 2  i,t 1 i,t 1 Y2.2*X   -0.36 (-3.80) -0.42 (-5.13) i,t X 0.25 (0.60) i,t 1 X  1.06 (1.82) Mid-East -0.03 (-0.26) -0.08 (-0.54) North-West 0.07 (0.44) 0.05 (0.31) West 0.28 (2.44) 0.29 (2.29) South-West 0.43 (2.87) 0.45 (3.13) South-East 0.24 (2.36) 0.26 (2.47) Notes: The Newey-West (1987) t-statistics are in parentheses. The fixed effects dummy variables are represented by their regional names. 16 References Alig, R.J., D.M. Adams and B.A. McCarl. “Impacts of Incorporating Land Exchanges between Forestry and Agriculture in Sector Models.” Journal of Agricultural and Applied Economics 30(1998):389-401. Barrett, A. and F. Trace. “The Impact of Agricultural and Forestry Subsidies on Land prices and Land Uses in Ireland.” Policy Research Series No. 35. The Economic and Social Research Institute, 1999. Behan, J. and E. Wiemers. “Farm Forestry and the Real Option Value”, paper presented at the Agriculture Economics Society conference, University of Plymouth, Devon. Behan, J. and K. McQuinn. “Projecting Net Greenhouse Gas Emissions from Irish Agriculture and Forestry.” Quarterly Economic Commentary, Spring, 44-54, The Economic and Social Research Institute, 2003. Clinch P., A. McCormack and T. O’Leary. “Afforestation in Ireland: Regional Differences in Attitude.” Land Use Policy 17(2000):39-48. Department of Agriculture, Food and Forestry, Growing for the Future: Strategic Plan for the Development of the Forestry Sector in Ireland. 1996. Frawley, J. and A. Leavey. Farm Forestry: Land Availability, Take-Up Rates and Economics. Teagasc, Dublin, 1999. Gardiner, J., and A. Ni Dhubhain. “Farmers Attitudes to Forestry.” Irish Forestry 51(1994):19-26. Gillmor, D. “Trends and Spatial Patterns in Private Afforestation in the Republic of Ireland.” Irish Forestry 55(1998):10-25. Heavey, J., Roche, M. and Burke, T. National Farm Survey, Teagasc, Dublin , 1997. 17 McCarthy, S., Matthews, A. and B. Riordan. Determinants of Private Afforestation in the Republic of Ireland. Trinity Economic Papers Series 2002/3. McCaul B.A. and U.A. Schneider. “U.S. Agriculture’s Role in a Greenhouse Gas Emission Mitigation World: An Economic Perspective.” Review of Agricultural Economics 22(2000):134-59. Newey, W.K., and K.D. West. “A Simple, Positive Semi-definite, Heteroskedasticity and Autocorrelation Consistent Covariance Matrix.” Econometrica 55(1987):703-06 Pedroni, P. “Critical Values for Cointegration Tests in Heterogeneous Panels with Multiple Regressors.” Oxford Bulletin 61(1999):653-670. Pedroni, P. “Fully Modified OLS for Heterogeneous Cointegrated Panels.” Nonstationary Panels, Panel Cointegration and Dynamic Panels, Advances in Econometrics, 15(2000):93-130. Sohngen B. and R. Mendelsohn. “An Optimal Control Model of Forest Carbon Sequestration.” American Journal of Agricultural Economics 85(2003):448-457. Tegene A., K. Wiebe and B. Kuhn. “Irreversible Investment under Uncertainty: Conservation Easements and the Option to develop Agricultural Land.” Journal of Agricultural Economics 50(1999):203-219. United Nations Framework Convention on Climate Change (UNFCCC). Kyoto Protocol to the United Nations Framework Convention on Climate Change (1997). http://www.unfcccc.de/resource/docs/convkp/kpeng.html 18 19 Endnotes 1 In Ireland traditional agriculture is mainly dairy and livestock. 2 Based in the Department of Communications, Marine and Natural Resources. 3 These regions are the mid-east, north-west, west, south-west and south-east. They are used for regional classification by the Irish Central Statistical Office (CSO). 4 Coillte Teoranta (The Irish Forestry Board) is the leading Irish company operating in forestry and related businesses. 5 Teagasc is the Irish Agriculture and Food Development Authority. 6 The average ex post real interest rate on long-term Irish government bonds has been 5% over the 1986-2001 period. 7 The number of observations was 17. 8 REPS is a co-financed scheme between the Irish Exchequer and the EU designed to reward farmers for carrying out their farming activities in an ‘environmentally friendly’ manner. In 2002, over €200 million was paid out to Irish farmers through the scheme. 9 The national envelope is a separate discretionary fund available to the national exchequer for expenditure within agriculture. It is accumulated by reducing the total amount of direct payments receivable/payable by a certain proportion.