Economics of riparian beneficial management practices for improved water quality: A representative farm analysis in the Canadian Prairie region

Abstract

This paper examines the economics of adopting agricultural beneficial management practices (BMPs) to protect riparian areas in the Canadian Prairies. Monte Carlo simulation methods and cost/benefit analysis (net present value) are used to evaluate BMP adoption for a set of representative cropping and mixed farm operations in Alberta and Saskatchewan. BMPs considered in the analysis involve restoration, maintenance and/or protection of riparian areas and associated wetlands. Implementation of these BMPs by the representative farms is generally costly, resulting in negative net benefits. This is primarily due to opportunity costs from forgone returns and BMP establishment costs. The farm level results are combined with estimates of public values for riparian ecosystem services using the Pannell land use policy framework. Results indicate that in many cases, positive incentives, such as subsidies or conservation auctions, are appropriate policy responses to encourage increased production of riparian/wetland ecosystem services. Given the uncertainty in public and private benefit estimates, it may also be the case that no intervention is the optimal policy response in some instances. There is significant scope for further research in order to obtain better estimates of benefits and costs associated with riparian area management.

Ce travail étudie les gains économiques obtenus en adoptant de meilleures pratiques agricoles de management (MPAMs) pour protéger les zones riparienne des Rocheuses Canadiennes. Nous combinons les méthodes statistiques de Monte Carlo et une analyse coût/bénéfice (valeur présente nette) pour évaluer l’adoption de meilleure pratiques pour un échantillon représentatif de fermes céréalières dans l’Alberta et la Saskatchewan. Les MPAMs considérées dans notre analyse impliquent la restauration, le maintien et/ou la protection des zones riparienne et des marais qui leur sont associés. La mise en place d’une MPAM est en général couteuse et réduit les bénéfices nets. Cette réduction est en grande partie due aux coûts de mise en œuvre de la MPAM et à la perte nette d’exploitation qui résulte de la MPAM. En nous aidant du cadre règlementaire de l’utilisation des terres développé par Pannell, les résultats obtenus à l’échelle de la ferme sont combinés avec les estimations des valeurs publiques pour des services des écosystèmes riparien. Nos résultats montrent que dans de nombreux cas des incitations positives telles que des subventions ou des enchères de conservation sont nécessaires pour encourager la production de services obtenus grâce aux écosystèmes riparien. Dans certains cas, étant donnée l’incertitude, il est possible qu’il soit optimal de ne pas intervenir du tout. Notre étude montre aussi l’importance de continuer les travaux de recherche dans le management des zones ripariennes.

Introduction

Bordering nearly every stream, river and wetland is a riparian area, defined as the area adjacent to streams, lakes and/or wet areas where plant communities are predominately influenced by their relationship with water (Roath and Kreuger 1982). Riparian areas play a role in water-related ecosystem service production. They contribute to water quality by providing protection from stream bank erosion and the leaching/runoff of harmful substances into the stream. These areas also serve as a buffer for aquatic habitats provided by wetlands.

Riparian areas can potentially provide agriculturally oriented ecosystem services in the form of crops and pasture, but there is a risk of reducing levels of other ecosystem services when using riparian areas for these purposes. For example, the utilization of riparian areas for agricultural production can result in the loss of riparian vegetation and habitat, stream bank and channel damage, and reduced water quality. These impacts also adversely affect the long-run agricultural use of the riparian areas. The benefit or cost to farms of restoring or maintaining these areas is the focus of this study.

There has been increasing interest and concern with the maintenance of agricultural riparian areas in Canada. Evidence of this can be seen through research activities examining the relationship between agricultural production and riparian area health, such as the Watershed Evaluation of Beneficial Management Practices (WEBs) research program sponsored by Agriculture and Agri-Food Canada (AAFC). There have also been several public programs such as environmental farm plans and the National Farm Stewardship Program that are intended to encourage agricultural producers to protect or restore riparian areas through implementation of beneficial management practices (BMPs). BMPs are land uses or management practices that contribute to reduced environmental damage from agricultural production, and include practices such as nutrient management planning or the establishment of buffer strips (AAFC 2000).

The relationship between improved water quality and the use of BMPs to protect riparian areas has been established in the scientific literature (e.g. Stillings et al. 2003; Agouridis et al. 2005). However, economic benefits are also available to producers in using this forage-rich environment (Unterschultz et al. 2004). The uptake of riparian area BMPs will be dependent on net benefits or costs associated with adoption. Without policy intervention, uptake may be limited due to private costs exceeding private benefits. Decisions regarding appropriate policy instruments should be made based on both the magnitude and sign of public and private benefits from the adoption of BMPs (Pannell 2008). While there exists a literature examining public values of riparian areas (e.g. Liu et al. 2010; Gascoigne et al. 2011; Johnson et al. 2012), it is limited, particularly with respect to a Canadian setting. As well, little information exists regarding producer benefits or costs of adopting riparian area BMPs.

This paper examines economic tradeoffs associated with the adoption of agricultural BMPs to protect riparian areas in the Canadian Prairie region. The objectives are to (1) model the adoption of alternative land uses and/or management practices that may be considered as riparian area BMPs and examine the resulting impact on farm financial performance, and (2) place the results of this analysis into the context of a policy decision framework in order to discuss options to incorporate public benefits of riparian areas. Economic cost/benefit analysis is used to evaluate BMP implementation by representative cropping (cereals/oilseeds) and mixed (cropping and cow-calf enterprises) operations in Alberta and Saskatchewan. The use of multiple farms and sites allows for a more comprehensive analysis of agricultural BMP adoption because of the wide range and intensities of agricultural activities associated with the various operations.

Impact of agriculture on riparian ecosystems and ecosystem services

As noted above, riparian ecosystems contribute to ecosystem services. Ecosystem services are defined as “components of nature, directly enjoyed, consumed, or used to yield human well-being” (Boyd and Banzhaf 2007, 619). Many of these services are associated with water and wetlands, including aquatic populations, surface water quantity and quality, biodiversity and natural land cover. While the human benefits derived from these services (aesthetics, recreation, damage avoidance for health and property) are apparent, the value of these benefits is not as clear, due to their largely non-market nature. A wide range of wetland service values has been estimated (see review by Brander et al. 2006). Relatively few studies have estimated the values of riparian area ecosystem services.

Research examining the effects of agriculture on riparian ecosystems and associated water quality has been well documented. Riparian overgrazing from livestock can result in physical damage to channels and banks (Fitch and Adams 1998). Introducing livestock for as little as six weeks into a riparian area that has previously been rested for four years has been shown to cause erosion of banks into streams (Kaufmann and Kreuger 1984). Stream bank degradation contributes to reduced water quality due to increased sediment, along with chemical and fecal runoff. Platts and Wagstaff (1984) argue that, given their importance, riparian ecosystems should be identified and managed separately from upland ecosystems.

Belsky et al. (1999) summarize the major effects of livestock activity on riparian ecosystems and streams in the western United States, stressing the serious problems and consequences of grazing on riparian environments. These consequences include detrimental effects on vegetative cover and biomass, plant succession problems and a decline in riparian animal species. Several of these effects have negative implications for water quality. Increased bacterial counts and higher water temperatures from cattle defecation contribute to reduced fish spawning and survival.

Similar work has examined the effects of cropping agriculture on riparian areas. Studies that examine the relationship between cropping practices and runoff of sediment and/or agricultural chemicals into riparian and wetland areas include Sharpley (1995), Blanco-Canqui et al. (2009) and Schilling et al. (2010). Ongley (1996) summarizes the effects of crop activities on surface and groundwater, which include sediment in waterways resulting from tillage operations, and chemical runoff from fertilizer and pesticides.

The impact of cattle and cropping on riparian areas and related ecosystem services may be mitigated through the adoption of BMPs. Examples of BMPs related to riparian areas include fencing riparian areas, off-stream watering for livestock, buffer strips, manure management and grazing/crop rotation systems. Studies have established the effectiveness of these BMPs in maintaining or improving riparian area and associated waterway/wetland health (Platts and Wagstaff 1984; Miner et al. 1992; Sheffield et al. 1997; Rein 1999; Hoover et al. 2001; Boehm et al. 2004). Larsen et al. (1994) showed water quality benefits attributable to cattle manure being deposited away from streams, while Clary (1999) demonstrated that reduced grazing intensity in riparian environments resulted in improved stream bank stability, increased plant species richness and increased stream sediment transport efficiency. Similar results have been found in other studies examining the effects of livestock on riparian habitats, such as Fleischner (1994), Green and Kaufmann (1995) and Clark (1998).

While the environmental impacts of riparian area BMPs are well documented, limited information is available for these types of practices in terms of direct farm-level economic impacts of adoption. Stillings et al. (2003) and Yang and Weersink (2004) estimate the economic impacts of establishing off-stream watering sites and riparian buffers, respectively, but neither study examines adoption for producers in the Canadian Prairie region. Unterschultz et al. (2004) provide a farm-level economic analysis of riparian buffers and off-stream watering, but only for native range in southwestern Alberta. In this study, economic costs and/or benefits of on-farm implementation of BMPs intended to preserve and maintain riparian ecosystems and water quality are estimated for representative Canadian Prairie farm operations.

Methodology

This paper presents and discusses results from research done in three separate studies that examine the farm-level economics of riparian BMP adoption (Koeckhoven 2008; Dollevoet 2010; Trautman 2012). These studies use representative farm analysis and Monte Carlo simulation techniques. Stochastic simulation methods are used to capture the risky nature of agricultural production. Also, the farms are assumed to participate in public business risk management (BRM) programs, and incorporating variability in production and prices is necessary in order to properly model the impacts of these programs. The farms modeled and analyzed are representative of commercial cropping and mixed cow-calf farms in the Canadian Prairie region.

The farm models are built using @Risk software (Palisade Corporation 2010). Values for stochastic commodity production and prices are drawn from analyst-defined distributions and used to simulate values for financial performance of the representative farms. The process is repeated to generate distributions of outcomes. The simulation in these studies is dynamic in that performance for the representative farms is also modeled over a specified time period, measured in years. A baseline scenario with no BMPs is initially simulated for each farm. Various BMP scenarios are then modeled and compared to the baseline scenario to estimate the private economic benefit or cost of adoption.

Representative farms

Six representative farms are defined and modeled in the BMP analysis. A combination of agricultural census data, provincial agricultural statistics and expert opinion is used to define the characteristics for the different representative farms. These characteristics include farm size, enterprise selection and size, management practices, costs of production and relevant production parameters. All six farms use dryland production practices, including zero tillage. A summary of farm characteristics is provided in Table 1.

Table 1. Representative farm characteristics.

Farm type	Mixed	Mixed	Crop	Crop	Crop	Crop
Province	Alberta	Saskatchewan	Alberta	Alberta	Alberta	Alberta
County/rural municipality	Lethbridge	Silverwood	Forty Mile	Starland	Camrose	Smoky River
Soil type	Dark Brown	Black	Brown	Dark Brown	Black	Dark Gray
Farm size (ha)	5115	777	1295	1295	1036	777
Annual crops	Durum wheat	Spring wheat	Spring wheat	Spring wheat	Spring wheat	Spring wheat
	Spring wheat	Barley	Barley	Barley	Barley	Barley
	Barley	Canola	Canola	Canola	Canola	Canola
	Canola	Flax	Durum wheat	Summerfallow
			Summerfallow
Forage/pasture	Alfalfa-grass hay	Alfalfa-grass hay	None	None	None	None
	Barley silage	Native pasture
	Native pasture	Tame pasture
	Tame pasture
Beef herd (number of cows)	400	116	None	None	None	None

Four cropping operations are modeled in this study. All are located in Alberta. The primary characteristic used for defining the cropping farms is soil zone, with farms being defined for the Brown, Dark Brown, Black and Dark Gray soil zones. The Brown, Dark Brown Black and Dark Gray soil zones are all primarily chernozemic soils (AAFC 1998), which is the predominant soil order in the grassland regions of Western Canada. There is some potential for the Dark Gray zone to include luvisolic soils.

There is one farm per soil zone. The cropping farms vary in size, based on what is typical in each soil zone region, from 777 ha to 1295 ha (Table 1). As shown in Table 1, the base crop rotations for these farms consist of annual crops, with spring wheat, barley and canola being predominant. Summerfallow is utilized for the Brown and Dark Brown soil zone farms.

Two mixed beef/crop representative farms are modeled in this study. One is located in Lethbridge County in southern Alberta (Dark Brown soil zone), while the other is in the Rural Municipality (RM) of Silverwood in southeastern Saskatchewan (Black soil zone). Both of these farms have a beef cow enterprise, combined with production of annual cereal/oilseed crops and forage crops. The primary forage crop is alfalfa-grass hay, grown in rotation with the annual crops. The hay is assumed to be established by underseeding with barley, followed by seven years of hay production (Campbell et al. 1994; Entz et al. 1995). Land is also allocated to tame and native pasture. Tame pasture is land for grazing that has been established through the introduction and maintenance of forage species; native pasture is land that has not been cultivated or fertilized.

The beef enterprises for these two farms include the cow herd, plus replacement heifers and bulls. Calving occurs in February, with calves being weaned and sold in late fall. The herd grazes tame and native pastures, and crop residues from grain and oilseed production following crop harvest (i.e. aftermath grazing). During winter, cattle are fed primarily hay and barley silage, with estimates of daily requirements per animal being obtained from Alberta Agriculture and Rural Development (AARD). Beef herd parameters (conception, calving and weaning rates, death loss, culling rates, average daily gain and pasture stocking rates) are defined based on provincial data and expert opinion. Summary information about the two representative mixed farm operations is provided in Table 1.

Stochastic production

Stochastic yields are modeled for annual cereal/oilseed crops, forage crops and pasture. Historical yield data at the county level are obtained from crop insurance and other sources. Two methods are used to model crop yields. For the mixed beef/crop farms, growing season temperatures and precipitation are collected for nearby weather stations. Production functions are then estimated for yields. The explanatory variables are linear and quadratic terms for the ratio of growing season precipitation (GS, measured in mm) to cumulative growing degree days (GDD):(1)

where is the yield for crop j in year t, GS and GDD are defined as above, the αs are parameters to be estimated and ε is the error term. GDDs are calculated assuming a growing season of May to October. The (GS/GDD) ratio represents a proxy for water availability relative to water demand. Yields are estimated as a system of equations, and the equation error correlations are used in the simulation. Seemingly unrelated regression (SUR) methods (Woolridge 2002, 143–145) are used to estimate the system of equations.

For the cropping farms, yield values are drawn directly from crop yield distributions. This is done due to an inability to obtain defensible yield-weather functional relationships for all of the crop/farm combinations.

Within the simulation, draws are made from either GDD and GS distributions (which are then used to calculate stochastic yields) or from the yield distributions directly. The distributions are estimated independently, and correlations between GS and GDD or between yields for different crops are incorporated into the process of drawing values from the distributions using the RiskCorrmat function in @Risk. Crop yield data used in these studies are at the county level. However, using county yields results in a lower estimate of variability than would be typical at the farm level (Marra and Schurle 1994). An adjustment suggested by Marra and Schurle (1994) is used in these studies to adjust the variability in the yield distributions upward.

Given a lack of available data for forage and pasture yields, different methods are used to model stochastic yields for these activities. In the case of forages (barley silage and hay), historical correlations between forage yields and a reference annual cereal crop (barley) are obtained from AARD and used to calculate year-to-year changes in the forage yields based on the corresponding change in barley yield. Yield correlations between barley grain and greenfeed, barley silage and hay are 0.725, 0.725 and 0.3, respectively. The starting forage yield in year 1 is the historical average for the region. Annual hay yields also vary by the age of the stand, based on results from Leyshon et al. (1981).

A similar approach is used for tame pasture yields, based on the annual change in native pasture yield and the correlation between native and tame pasture yields. In this case, a correlation of 0.6 is used based on AARD expert opinion. For native pasture, a yield model based on data from Smoliak (1986) is used, where yield is a function of growing season precipitation.

Calf sale weight is not explicitly modeled as being stochastic, but calf weight gain is linked to pasture productivity. A certain amount of forage is required to achieve the target selling weight of 250 kg for weaned calves. This is largely provided through pasture production. In cases where pasture is insufficient to support the required weight gain, it is assumed that the calves are placed in drylots and fed until they reach the target weight. If pasture productivity is “higher than average”, it is assumed that the calves are sold at a weight that is higher than the target. Calf sale weight is thus indirectly stochastic, given this link to stochastic pasture productivity.

Stochastic prices

Beef and crop prices are modeled as being stochastic for all representative farms. Historical provincial level prices are collected from secondary sources, deflated to real values and tested for stationarity. Based on results for the Augmented Dickey Fuller test for stationarity, several of the price series used in the individual studies do not appear to be stationary. As discussed by Dixit and Pindyck (1994), Verbeek (2004) and Wang and Tomek (2007), however, there are limitations (related to both data and statistical power) associated with the use of this test. Andersson (2007) and Wang and Tomek (2007) suggest that agricultural commodity prices are generally stationary. As well, in the cropping analysis, a different test (Kwiatkowski-Phillips-Schmidt-Shin test) determined the majority of the price series to be stationary for the relevant crops. Based on this, time series models are estimated and used for all commodity prices in the analysis.

Tests are done, using the Akaike Information Criterion, to establish appropriate lag lengths for each commodity, and the price models are estimated using a system of equations approach. Individual equations within each system of price equations take the following form:(2)

where is the current price for commodity j, is the price lagged i times, ε is the error term and the βs are parameters to be estimated. In the case of the mixed beef/crop farms, separate cropping and livestock price equation systems are estimated. Random error draws and price equation error correlations are used, along with the time series model parameter estimates, to calculate annual commodity prices for the simulation analysis.

Due to a lack of market price data, hay prices are modeled as being deterministic, and set equal to a representative price based on information from the two most recent years in the analysis. Barley silage price is modeled to be directly and linearly related to the price of barley grain (AARD 2008).

Economic relationships and performance measures

A modified net cash flow (MNCF) measure is calculated for each year within the simulation model for the representative farms. MNCF is the sum of farm revenues and BRM payments, minus the sum of variable expenses and annual machinery depreciation expense. Debt servicing cash flows are not included in these calculations. This is done to avoid having capital structure decisions influence the model results with respect to the BMP scenarios. The analyses are all done on a before-tax basis.

Revenue includes returns from sales of crops and/or livestock. BRM program payments include crop insurance and AgriStability payments (AFSC n.d.). Variable expenses include those expenditures related to inputs used in the production of crop and livestock commodities. Variable expenses for crop and beef enterprises are calculated using budget information made available by AARD and AAFC. Costs included are seed, fertilizer, chemicals and machinery (repairs, fuel and lubricant) for cropping enterprises, and purchased feed and bedding for the beef enterprise. The machinery depreciation expense is calculated as the annualized cost of maintaining the farm machinery complement at the initial book value. As such, it represents a proxy for machinery replacement expenditures.

Net present value (NPV) is used as the performance measure to evaluate the impact of riparian BMP adoption scenarios. NPV represents a proxy for wealth, and scenarios are compared to determine whether a particular BMP has a positive or negative impact on financial performance. In general, NPV is equal to the present values of future net cash flows, summed over the relevant time horizon, minus any initial investment expenditures. For most of the NPVs calculated in these studies, there is no initial investment expenditure. As well, the NPVs are calculated assuming an infinite time horizon; that is, NPV in perpetuity. An infinite time horizon NPV is approximated by taking the MNCF in the last year of the simulation, treating it as a perpetual annuity, and calculating and adding the present value of this perpetual annuity to the NPV to obtain the NPV in perpetuity for each scenario. The last year of the simulation analysis varies by farm type: year 20 for the mixed farms and year 40 for the crop farms. The resulting NPV formula used in the analysis is as follows:(3)

where MNCF_t is the modified net cash flow for year t, T is the time horizon (20 for the mixed farms and 40 for the crop farms) and r is the risk adjusted discount rate. The discount rate used for all representative farms is 10% or 0.10.

BMP scenarios and analysis

A number of riparian BMPs are modeled in terms of adoption by the representative farms. These can be divided into two categories: restoration or maintenance of riparian area, and protection of riparian area with buffer strips or permanent cover. Both categories of BMPs are similar with respect to the provision of water-related ecosystem services. They provide protection for wetlands in terms of reducing the potential for runoff of sediment and agricultural chemicals reaching the wetlands, thus maintaining or improving water quality. As well, by maintaining or protecting riparian areas, wildlife and waterfowl habitat is provided. Other ecosystem services potentially provided by these BMPs include climate regulation (e.g. through carbon sequestration) and improved soil quality attributable to reduced soil loss.

The specific BMPs modeled vary by representative farm. The adoption of riparian area protection BMPs, in the form of establishing buffer strips or permanent cover, is modeled for cropping operations. Land adjacent to riparian areas is “retired” from annual crop production to provide a protective buffer. The difference between a buffer strip and permanent cover is that land allocated to permanent cover is permanently seeded to perennial forage (which is harvested as hay), while in the case of a buffer strip the land is left idle. In each case, the shape of the associated wetland area is assumed and a strip of land approximately 10 m wide is set aside as either a buffer strip or permanent cover.

Riparian area restoration/maintenance is modeled for the two mixed farm operations and, in both cases, restoration is implemented for riparian area adjacent to cropping and pasture areas. In the case of the southern Alberta farm, lotic wetlands are present that have a 9-m “band” of riparian area on each side. In the baseline scenario (no BMP adoption), the riparian areas are used for agricultural purposes, either crop production or pasture. The BMPs modeled involve restoring part or all of the riparian area by taking it out of agricultural production. Four versions of the restoration BMP are modeled: restoration of 25, 50, 75 and 100% of the total potential riparian area. The versions differ in the width of riparian area restored; for example, 25% restoration involves re-establishing a riparian area that is 2.25 m wide, 50% restoration represents re-establishment of a 4.5-m band of riparian area, and so on. For restored riparian area adjacent to pasture, fencing is installed to exclude cattle from the restored area. If the restored area is adjacent to land used for crop production, fencing is also installed along with an 11-m buffer strip or permanent cover area to protect the restored riparian area. The fencing is included in this case to exclude cattle during aftermath grazing.

Based on expert opinion, each cropped field for the base Saskatchewan mixed operation includes 10% riparian area. The nature of the BMP modeled in this case is maintenance of some or all of the riparian area; that is, not converting it to agricultural use. Three alternative maintenance scenarios are modeled for the Saskatchewan farm: maintaining 33% (⅓), 67% (⅔) and 100% of the initial riparian area. The modeling is slightly different in that the baseline scenario is maintenance of the initial riparian area. This baseline is compared with alternative scenarios that involve partial or complete conversion of riparian area to agricultural productive use (i.e. crop production). Given the nature of the land use change, there is an initial investment requirement in the form of clearing costs. If these scenarios result in positive (negative) private benefits, it is concluded that maintenance of the riparian area would generate equivalent negative (positive) net benefits.

One other BMP is modeled for the Alberta mixed farm operation: installation of off-stream watering (OSW) systems. Previous research (Miner et al. 1992) demonstrates that when OSW is provided in pastures, cattle spend significantly less time in or near wetlands. This results in less fecal contamination of wetland areas. Adoption of the OSW BMP is modeled for the southern Alberta mixed cropping-beef farm operation. No land use changes or other adjustments are made with the OSW adoption scenario. There is also an initial investment requirement for this BMP, represented by the expenditure to install the OSW equipment.

For each BMP adoption scenario, information required to model the impacts on relevant parameters is obtained from a variety of sources. Cropped and pasture areas are adjusted according to the amount of land removed from production due to restoration or protection of riparian areas. Provincial budgets and statistics are used to calculate costs of hay production for permanent cover, along with hay yields. Expert opinion is used to refine estimates of costs for riparian area restoration, fence installation and maintenance, and OSW system installation. A summary of the characteristics for the various BMPs is provided in Table 2.

Table 2. Characteristics of beneficial management practices (BMPs) modeled for the representative farms.

Farm	BMP^a	Prior land use^b	Width of affected area	Other relevant BMP information
Crop	Permanent cover	Cropping	10 m	Permanent cover established around riparian area and harvested for hay
Crop	Buffer strip	Cropping	10 m	Buffer strip established around riparian area; buffer strip retired from agricultural use
Alberta mixed	Riparian restoration + permanent cover	Cropping	9 m of riparian area	Riparian area is restored and protected by permanent cover which is harvested for hay and aftermath grazed
			11 m of permanent cover	Alternative scenarios modeled involving 25, 50, 75 and 100% restoration of potential riparian area
Alberta mixed	Riparian restoration + buffer strip	Cropping	9 m of riparian area	Riparian area is restored and protected by buffer strip; buffer strip is retired from agricultural use; fence installed to exclude cattle
			11 m of buffer strip	Alternative scenarios modeled involving 25, 50, 75 and 100% restoration of potential riparian area
Alberta mixed	Riparian restoration	Pasture	9 m of riparian area	Riparian area is restored and fence installed to exclude cattle
				Alternative scenarios modeled involving 25, 50, 75 and 100% restoration of potential riparian area
Alberta mixed	Off-stream watering system	Pasture	Not applicable	Off-stream watering system is installed in the pasture area; no changes in land use
Saskatchewan mixed	Riparian maintenance	Cropping	10% of cropped area	Assumes riparian area is present initially and some portion is converted to cropping use
				Alternative scenarios modeled involving 33, 67 and 100% maintenance of initial riparian area

^a “Restoration” refers to restoring riparian area; that is, converting from either cropping or pasture use. “Maintenance” refers to maintaining land as riparian area instead of converting to either cropping or pasture use. BMPs with a “+” are a combination of riparian area restoration and protection of the riparian area with permanent cover or a buffer strip.

^b Prior land use refers to the land use assuming no adoption of the BMP.

When modelling the alternative BMP scenarios, the assumption is made that the representative producers implement the specific change starting in the first year of the simulation. For most of the BMPs, then, the adjustments are “instantaneous”; for example, when the cropping operations implement a buffer strip BMP, it is assumed that all of the buffer strip areas are established at the beginning of year 1 of the analysis. However, for BMPs involving fencing, BMP implementation is completed over a three-year period to reflect the time required to construct the fence.

The impact of the alternative BMP adoption scenarios on farm performance is assessed using expected NPVs generated by the dynamic Monte Carlo simulation analysis. For each farm, the expected NPVs for the baseline scenario with no BMP adoption, and the BMP adoption scenario, are converted to annualized values or annuities. The perpetual annuity formula is used for this purpose:

where PV is the present value of the annuity, represented by the expected NPV in this case, r is the discount rate and A is the annuity or annualized NPV.

The difference between the annualized NPVs (A_BMP – A_baseline) is calculated as the annual net impact of the adoption scenario under consideration. A positive difference represents net positive private benefits associated with BMP adoption, while a negative difference represents a net cost. The change is then converted to an annual value per unit of land (ha) basis. This is done to express the benefit (or cost) in scale terms that are consistent with what would likely be relevant for policy implementation purposes (e.g. a tax or subsidy per hectare). Given the nature of the BMPs considered in the analysis, the area restored or converted is used as the basis for conversion. The exception to this is the OSW BMP. Given that this BMP does not involve any land use conversion or change, the difference in annualized NPV is converted to a per hectare of pasture basis, since the BMP affects that particular land use.

Results

Table 3 provides summary information associated with the baseline scenario results for the six representative farms. The level of performance, defined in terms of expected ending wealth (NPV) per hectare, and variability in performance (standard deviation of NPV per hectare) vary by type of farm and location. The highest performance levels are obtained for the Black and Dark Gray soil zone cropping farms, reflecting the greater production potential for these soil zones. Conversely, the weakest expected performance is demonstrated by the Brown and Dark Brown soil zone cropping operations. This is also consistent with relative yield and profit potential. Summerfallow is included in the crop rotation for both of these farms, which reduces the area allocated to marketable crops. The two mixed farms are “in between” with respect to expected performance, with the Saskatchewan farm demonstrating greater expected ending wealth per hectare. This is partly due to differences in productive potential attributable to soil zone. Variability in ending wealth is influenced to a large degree by production variability, which is linked to climate and soil productivity (both of which are correlated with soil zone), and, to a lesser extent, by diversification in enterprises.

Table 3. Summary of baseline simulation results with no beneficial management practice (BMP) adoption (2010 $CAD)^a.

Farm	Mean	Standard deviation	Annualized mean NPV^c
	(NPV^b)	($/ha)
Alberta mixed cropping-beef	$1134.55	$187.69	$113.46
Saskatchewan mixed cropping-beef	$1555.28	$375.38	$155.53
Alberta Brown soil cropping	$659.59	$270.51	$65.96
Alberta Dark Brown soil cropping	$853.84	$231.93	$85.38
Alberta Black soil cropping	$2770.38	$322.56	$277.04
Alberta Dark Gray soil cropping	$3144.86	$298.83	$314.49

^a The values in this table are calculated as the farm level value, divided by the total area of the farm.

^b NPV: net present value.

^c The annualized NPV values are calculated using the perpetual annuity formula; the mean NPV per hectare is multiplied by the discount rate (0.10).

Tables 4 and 5 report basic simulation results for the BMP scenarios. Protecting riparian and wetland areas on Alberta cropping operations through the use of permanent cover or buffer strips is costly, as indicated by the negative mean net benefits. The annual cost ranges from approximately CAD$21 to $346 per hectare converted from annual crop production (Table 4).

Table 4. Net benefits ($/ha/yr) of beneficial management practice (BMP) adoption for representative cropping farms (2010 $CAD).

	Mean net benefits
	($/ha/yr)^a
Farm	Permanent cover BMP^b	Buffer strip BMP^b
Alberta Brown soil zone	–$20.63	–$96.72
Alberta Dark Brown soil zone	–$27.59	–$115.84
Alberta Black soil zone	–$226.29	–$304.30
Alberta Dark Gray soil zone	–$283.14	–$345.74

^a Mean net benefits are calculated on a per-hectare basis based on the area directly affected by the BMP (i.e. restored, converted or maintained).

^b These BMPs involve establishing a buffer around undrained wetlands that are present on the representative farm operation, perennial forage in the case of permanent cover and idle land in the case of a buffer strip.

Table 5. Net benefits ($/ha/yr) of beneficial management practice (BMP) adoption for representative mixed cropping-beef farms (2010 $CAD).

Farm	Prior land use^a	BMP^b	Mean net benefits
			($/ha/yr)^c
			Restoration scenario
Alberta			25%	50%		75%		100%
	Cropping	Riparian restoration + permanent cover	–$202	–$199		–$198		–$198
	Cropping	Riparian restoration + buffer strip	–$459	–$469		–$473		–$475
	Pasture	Riparian restoration	–$544	–$461		–$434		–$419
	Pasture	Off-stream watering (OSW) system	–$0.98^d
			Maintenance scenario
Saskatchewan			33%		67%		100%
	Cropping	Riparian maintenance	–$213		–$202		–$201

^a Prior land use refers to the land use assuming no adoption of the BMP.

^b ”Restoration” refers to restoring riparian area; that is, converting from either cropping or pasture use. “Maintenance” refers to maintaining land as riparian area instead of converting to either cropping or pasture use. BMPs with a “+” are a combination of riparian area restoration and protection of the riparian area with permanent cover or a buffer strip.

^c Mean net benefits are calculated on a per-hectare basis based on the area directly affected by the BMP (i.e. restored, converted or maintained). In the case of the OSW BMP, the entire pasture area is used in the calculation. For the Alberta farm, the restoration scenarios refer to the percentage of total potential area that is restored to riparian area. The restoration scenarios do not apply to the off-stream watering system BMP. For the Saskatchewan farm, the maintenance scenario refers to the percentage of initial riparian area that is maintained.

^d The annualized change in net present value (NPV) for this BMP (i.e. the annual net impact for the farm business) is –$3,603.28.

The direction and magnitude of the impact for these BMP scenarios is largely due to the opportunity cost of removing land from agriculturally productive use. When permanent cover is compared with buffer strips, it can be seen that the net impact for the permanent cover BMP is partially offset by the value of forage production. The pattern across farms also reflects differences in profitability. For example, the net costs associated with adoption of these BMPs are greater for the Black and Dark Gray soil farms than for the Brown and Dark Brown soil farms.

Table 5 provides BMP adoption scenario results for the two mixed beef-cropping farms. For the Alberta representative mixed farm, adoption of riparian area restoration BMP scenarios generally results in negative expected net benefits. The annual impact, in net cost terms, ranges from approximately CAD$200 to almost $550 per hectare converted to riparian area. As shown in Table 5, the values for individual BMPs vary by restoration/maintenance scenario, that is, the percentage of potential or available riparian area restored or maintained.

The explanation for the Alberta farm results in Table 5 is similar to the BMP scenarios for the cropping operations. Land is removed from agricultural production as a result of these BMPs, and the opportunity cost of doing this contributes to the net impact. Also similar to the cropping farms, the impact for the permanent cover version of the BMP is lessened due to the value of harvested forage.

The net cost of BMP adoption for the Alberta mixed farm operations is further increased due to BMP establishment costs. For the case of riparian area restoration in pasture areas, installation of OSW is required since cattle no longer have access to water in wetland areas. Fencing is required for all riparian restoration scenarios in order to exclude cattle from the riparian and wetland areas. Costs are incurred by the producer for both installation and maintenance of OSW and fencing. The fencing costs also apply to areas that were initially cropped, as the cattle use those areas for aftermath grazing in the autumn season. As these fencing costs are spread over a greater area being restored, the mean impact per hectare affected is reduced. The exception is for the cropped area buffer strip BMP. In adopting this BMP, there is reduced area available for hay production, which results in a further cost to the producer for increased forage purchases.

The results for adoption of the OSW system BMP (Table 5) are significantly different than those for the other BMPs (i.e. minimal net cost per hectare), for two reasons. First, the impact on pre-adoption practices and land use is minimal. Beyond installation of the OSW system, there is no other change made (i.e. no land conversion or cattle exclusion). As discussed earlier, the intention with this BMP is to draw the cattle away from using the riparian and wetland areas by providing a stable and higher quality source of water in upland areas. Secondly, the overall impact (a net cost of approximately CAD$3600 annually) is allocated over the entire pasture area to obtain the value per hectare reported in Table 5.

The BMP modeled for the Saskatchewan mixed farm represents maintenance of riparian areas; not converting to agricultural productive use. The expected net benefits reported in Table 5 are opportunity costs of maintaining the area in its current form, as opposed to converting for agricultural use which involves clearing, leveling and (if necessary) draining. The results here are consistent with those for the Alberta mixed farm. If the cropped field riparian area is converted for use in crop production (annual crops or perennial forages), the direct effects on farm performance are positive – CAD$201 to $213 annually per hectare converted depending on the degree of riparian area converted (33% to 100%). Therefore, maintaining the riparian area results in an equivalent negative net benefit.

Policy discussion

The empirical results presented for the representative farm operations indicate that there are net private costs associated with the adoption of BMPs to create/maintain or protect riparian areas. In some cases, these costs are significant at over $500/ha/yr. If it is desirable from society’s perspective to maintain or increase the ecosystem services through riparian area management, including those associated with water quality, policy intervention is likely needed.

Given the negative effects on farm financial performance from adopting riparian area BMPs, what should be the nature of policy intervention, or should there be any policy intervention at all? One approach for considering this question is to use Pannell’s framework (Pannell 2008) for environmental policy analysis. The principle behind this framework is that the choice of policy instrument depends on the relative magnitude and sign of public and private net benefits associated with the desired land use change. Private net benefits refer to benefits accruing to the individual decision maker. Public net benefits are those benefits accruing to everyone else in society. A full explanation of the policy framework is provided by Pannell (2008).

The Pannell framework can be used to examine policy options for encouraging environmentally friendly land use changes, such as the riparian area BMPs examined in this analysis. For example, extension would be an appropriate policy instrument if a proposed land use change generates positive levels of both public and private net benefits. All that should be required to encourage the socially optimal land use in this case is education about private net benefits associated with the change. Conversely, the use of positive or negative economic incentive programs such as subsidies, fines or taxes may be appropriate in cases where the land use change generates negative net benefits for one or the other of public or private concerns, depending on the relative absolute magnitude of public versus private benefits. Private individuals are less likely to make publically optimal land use decisions in these cases.

For the current study of riparian area restoration or protection, the representative farm simulation results provide estimates of net private benefits. Obtaining estimates of public benefits is more problematic. Public benefits would be represented by the economic value of the ecosystem services provided by the restored/protected riparian areas. These benefits are difficult to quantify and many of the services provided have non-market values.

It is not surprising that there are relatively few studies that attempt to quantify the economic value of riparian ecosystem services, given the challenges associated with valuation. Further, there are no Canadian studies of this type. Studies by Gascoigne et al. (2011) and Johnson et al. (2012) are adapted for use to obtain estimates of public benefits associated with riparian areas. Both of these studies estimate values for ecosystem services associated with land use changes for the Prairie Pothole Region (PPR) of the Upper Midwestern United States. The majority of representative farms simulated in the current analysis are located in the Canadian portion of the PPR.

Gascoigne et al. (2011) estimate ecosystem service values associated with land conversion scenarios for the PPR in North and South Dakota. Relationships and estimates from published studies are used by the authors to generate biophysical and economic values for climate regulation through carbon sequestration, reduced soil loss and sediment deposition, and enhanced waterfowl habitat resulting from the land use changes. Their study examines a number of land use change scenarios, but the one used for the purposes of the current analysis is conversion from cropland to conservation reserve (CRP) and/or wetland reserve (WRP) programs. Conversion of land to CRP in particular is very similar to the BMPs where cropland is restored to riparian area or converted to buffer strips.

Johnson et al. (2012) estimate ecosystem service values for alternative land use scenarios in the Minnesota River Basin in Minnesota. Results from previous studies are used to calculate biophysical and economic values. The services considered by Johnson et al. (2012) are climate regulation through carbon sequestration, and reduced nitrous oxide emissions, and improved water quality through reduced phosphorus loading. The land use change considered in their study is conversion of cropland to buffers around rivers and streams. Three different land conversion scenarios are examined: establishing 25 m buffers, establishing 100 m buffers, and establishing 100 m buffers along with conversion of all low productivity agricultural land to perennial vegetation.

The land use change scenarios examined in Gascoigne et al. (2011) and Johnson et al. (2012) are not exactly the same as the BMPs modeled in the current analysis. They are similar, however, in that they involve removing land from agricultural use and setting aside areas for buffers. The conversion scenarios in the two studies are therefore comparable to the riparian area restoration and protection BMPs examined here. The geographic location (PPR) for the two US studies is also comparable to the locations of the representative farms. As discussed earlier, ecosystem services assumed to be provided by the BMPs in the current analysis include reduced runoff of sediment and agricultural chemicals, provision of wildlife/waterfowl habitat and climate regulation. These are consistent with the services considered in the Gascoigne et al. (2011) and Johnson et al. (2012) studies. The values estimated in those papers are therefore deemed appropriate for adaptation here as proxies for the ecosystem service values provided by the BMPs examined in this analysis.

Table 6 provides estimates of economic values for ecosystem services associated with the restoration of riparian areas, adapted from Gascoigne et al. (2011) and Johnson et al. (2012). The estimates for water quality, reduced soil loss/sedimentation and waterfowl habitat are taken directly from the results of the two studies. The two studies use different carbon values in their analyses. To obtain the values reported in Table 6, a carbon price of $21 per tonne of carbon dioxide equivalent (CO₂e) (2010 $US) is used to value the biophysical estimates for both studies. As discussed by Johnson et al. (2012), $21 per tonne of CO₂e was the central value for the social cost of carbon (SCC) chosen by a US government working group. All ecosystem values are then converted to 2010 $CAD.

Table 6. Estimates of annual public benefits for restored riparian areas (2010 $CAD)^a.

	Gascoigne et al. (2011)	Johnson et al. (2012)
	CRP/WRP	Buffer-25 m	Buffer-100 m	Buffer + upland
Riparian scenario^b		$/ha/year
Climate regulation^c	$80.60	$191.93	$174.66	$184.53
Water quality		$1,703.76	$753.37	$255.47
Reduced soil loss/sedimentation	$16.84
Waterfowl habitat	$37.98
Total public benefit	$135.42	$1,895.69	$928.03	$440.00

^a Results are adapted from Gascoigne et al. (2011, Tables 2 and 3) and Johnson et al. (2012, Tables 2 and A-1). The Johnson et al. (2012) results presented are for the scenario in their paper using a willingness-to-pay for water quality improvements of $131 per household (2001 $US) and a 2% annual increase in the social cost of carbon.

^b The riparian scenario modeled in Gascoigne et al. (2011) is a conversion of cropland to conservation (CRP) or wetland (WRP) reserve. In Johnson et al. (2012), three scenarios involve the retirement of agricultural lands, the creation of 25- or 100-m stream buffers by restoring land to perennial vegetation (Buffer-25 m and Buffer-100 m), and 100-m stream buffers and the retirement of all low productive agricultural land (Buffer + upland).

^c Climate regulation values are calculated using a social cost of carbon of $21 per tonne of carbon dioxide equivalent (CO₂e; 2010 $US). Values for both studies include carbon sequestration estimates. In addition, the Johnson et al. (2012) estimate includes the value of reduced nitrous oxide emissions.

The values presented in Table 6 can be interpreted as estimates of annual net public benefits associated with relevant land use changes. The range in annual public benefit values is significant, from CAD$135 per hectare to almost $1900 per hectare (Table 6). This variability reflects a number of factors, including differences in the scale of conversion, combinations of ecosystem services considered and methods used to calculate biophysical and economic values.

The results for the representative Alberta and Saskatchewan farms may be compared with estimates generated from the two ecosystem service studies (Gascoigne et al. 2011; Johnson et al. 2012). Based on this comparison, there is some justification for policy intervention to encourage the restoration and/or protection of riparian areas in the Canadian Prairie region. In particular, estimates of annual net private benefits for riparian area protection are in the range of –CAD$21 to –$283 per hectare. With public benefits in the range of CAD$135 to ~$1900 per hectare, Pannell’s framework would suggest that the use of positive incentives such as subsidies or conservation auctions represents an appropriate response. Similarly, the range of annual net private benefit estimates for riparian area restoration (–CAD$201 to –$544 per hectare) would suggest that in some cases positive incentives are warranted. Lending further credence to this conclusion is the fact that there are other ecosystem services, such as aesthetics and other dimensions of water quality, not considered in either of the two studies. Inclusion of these would increase the net public benefits associated with riparian areas.

The policy picture is not completely clear. There is significant variability in both public and private benefit estimates. If the upper range of net private benefits (i.e. the upper end of adoption cost estimates) is considered in combination with the lower end of the range for net public benefits, the conclusion would be that no intervention is required. The benefits to society are lower than the net private costs associated with adoption of the socially beneficial land use change.

There has been limited use of policy instruments by provincial or federal governments in Canada to encourage the restoration of riparian areas or establishing buffer strips. For example, stewardship programs were established through the federal and provincial governments’ Growing Forward agricultural policy framework. However, the majority of these programs involved cost-sharing arrangements for direct costs of undertaking environmental stewardship practices and did not include any consideration of private opportunity costs. One of the objectives of Growing Forward 2, the latest Canadian agricultural policy framework, is to encourage market-based solutions to environmental problems. This may imply greater use of instruments such as conservation auctions or other policy tools.

Conclusions

Simulation methods are used to model the adoption of riparian area BMPs for a number of representative cropping and mixed farm operations located in Alberta and Saskatchewan. While the empirical results vary significantly by BMP, it may be stated that in general, for these farms, the adoption of land use changes to restore/maintain and/or protect riparian areas comes at a significant annual cost to producers. These costs range from CAD$20 to over $540 per hectare annually. The main contributing factor to this result is the opportunity cost associated with removing land from agriculturally productive use. In some cases, there are also significant establishment costs incurred in implementing the BMPs. Thus, one conclusion of the study is that using riparian areas to protect wetlands and associated levels of wetland (and riparian) ecosystem services is costly and provides negative net private benefits.

The BMP simulation results for the representative farms are compared with estimates of public benefits associated with the retirement of agricultural land to riparian and/or buffer uses. These estimates are obtained by adapting empirical results from two studies (Gascoigne et al. 2011; Johnson et al. 2012) for the Prairie Pothole Region of the American Upper Midwest. Similar to the simulation results, there is significant variability in the estimates of ecosystem service values from these studies. The Pannell land use policy framework (Pannell 2008) is used to evaluate the results from a policymaking perspective. Based on this evaluation, there is support for the use of positive incentives such as subsidies, auctions or offset systems to encourage the protection of wetlands through the conversion of agricultural land to riparian area or the maintenance of current riparian areas. However, given the range of public and private benefits identified and discussed here, there may be situations in which the optimal policy option is to not intervene at all, but instead to look for other opportunities that provide a better “social rate of return” to public investment in supporting ecosystem service production.

Finally, the nature of the empirical results and discussion provided here suggests the need for further investigation of costs and benefits for riparian/wetland land use changes and ecosystem service production associated with these areas. While the body of literature is growing, there has still been limited analysis of private costs and benefits for producers who may consider adopting agriculturally related environmental stewardship practices. There has also been little work done to examine whether the financial costs and benefits of adoption accurately reflect producer needs for compensation – that is, their willingness to accept. Even less research has been done to quantify levels and values for relevant ecosystem service production associated with riparian and wetland areas in Western Canada. These all represent areas of need for research in order to make informed policy decisions.

Acknowledgements

The authors acknowledge the support from Agriculture and Agri-Food Canada (through the Watershed Evaluation of Beneficial Management Practices program), Alberta Crop Industry Development Fund, Alberta Pulse Growers and the Lower Souris Watershed Group. Additional data were provided by Alberta Agriculture and Rural Development and the Agriculture Financial Services Corporation.