Western Economics Forum - Western Agricultural Economics Association

Fall 2012: Volume 11, Number 2

A Journal of the Western Agricultural Economics Association

Western Economics Forum

Farm & Ranch Management Marketing & Agribusiness Natural Resources & the Environment Policy & Institutions Regional & Community Development

Western Economics Forum Volume XI, Number 2 FALL 2012

Table of Contents Dawn Thilmany, Eyosiyas Tegegne and Brett Hines Farmers Markets and Direct Marketing in the Western US: Market Trends and Linkages with Food System Issues ........................................................................................ 1 Ryan Mortenson, Jay Parsons, Dustin L. Pendell and Scott D. Haley Wheat Variety Selection: An Application of Portfolio Theory in Colorado ................................... 10 Nicholas S. Brown and Phil Watson What can a comprehensive plan really tell us about a region?: A cluster analysis of county comprehensive plans in Idaho ........................................................ 22 David T. Taylor, Thomas Foulke and Archie Reeve A Case Study in Habitat Equivalency Analysis: The Pacific Connector Gas Pipeline ........................................................................................... 38

The Western Economics Forum

A peer-reviewed publication from the Western Agricultural Economics Association Purpose One of the consequences of regional associations nationalizing their journals is that professional agricultural economists in each region have lost one of their best forums for exchanging ideas unique to their area of the country. The purpose of this publication is to provide a forum for western issues. Audience The target audience is professional agricultural economists with a Masters degree, Ph.D. or equivalent understanding of the field that are working on agricultural and resource economic, business or policy issues in the West. Subject This publication is specifically targeted at informing professionals in the West about issues, methods, data, or other content addressing the following objectives: Summarize knowledge about issues of interest to western professionals To convey ideas and analysis techniques to non-academic, professional economists working on agricultural or resource issues To demonstrate methods and applications that can be adapted across fields in economics To facilitate open debate on western issues Structure and Distribution The Western Economics Forum is a peer reviewed publication. It usually contains three to five articles per issue, with approximately 2,500 words each (maximum 3,000), and as much diversity as possible across the following areas: Farm/ranch management and production Marketing and agribusiness Natural resources and the environment Institutions and policy Regional and community development There are two issues of the Western Economics Forum per year (Spring and Fall). Editor – Send submissions to: Dr. Don McLeod Editor, Western Economics Forum Dept. of Ag & Applied Economics University of Wyoming Dept. 3354 1000 E. University Avenue Laramie, WY 82071 Phone: 307-766-3116 Fax: 307-766-5544 email: [email protected]

Western Economics Forum, Fall 2012

Farmers Markets and Direct Marketing in the Western US: Market Trends and Linkages with Food System Issues Dawn Thilmany1, Eyosiyas Tegegne1 and Brett Hines1 Farmers markets have a rich history in the market development of agriculture in the United States, and they are re-emerging as a key community-based option for local food marketing. They represented an important community food distribution system long before the rise of the retail agribusiness system and began to re-emerge (following years of decline) after the passage of the Farmer-to-Consumer Direct Marketing Act of 1976. Some argue that they are now integral part of the food community linking consumers and producers through business and social relationships. Others view markets as an appropriate distribution channel for entrepreneurial and small farmers who strive to establish a loyal customer base through personal selling and quality differentiated (vs. low margin commodity) marketing strategies. The increasing number of direct market channels, including farmers markets, are growing at a persistently high rate since the mid-1990s. These suggest several other bigger food system discussions: the role of farmers markets and producer-focused value chains as “engines of grassroots economic development”, direct markets as a mechanism to address access to healthier foods, and as a market access point for small, beginning or socially disadvantaged producers who might otherwise face barriers to conventional supply chains. The growth and contribution of farmers markets and other direct marketing channels has led to more studies in the literature and encouraged us to update an earlier study on farmers markets and direct marketing in the Western US. To organize our discussion on markets, we will first attempt to understand the supply side (number of producers and marketing channels) and demand side factors (growth in sales, as well as number and types of potential consumers), as well as how direct marketing activities influence the communities where such activities are established and growing. The major objective of this article is to summarize the findings of recent analyses pertaining to farmers markets and direct marketing by agricultural producers, including the US Ag Census, the USDA’s Agricultural Marketing Service study of farmers markets and community-based studies. The second purpose of this article is to present a brief review of the literature on the role of farmers markets on key public issues such as community development and dietary and health outcomes. By synthesizing what has been reported on farmers markets and direct marketing, we can assess the potential role of such channels in the region, and whether the role is expanding into new realms of rural and economic development.

Direct Marketing by Producers Data on direct marketing from the 2007 Ag Census shows some interesting trends in the US and Western region. First, it may be important to note what the USDA currently includes in its measures. The USDA National Agricultural Statistics Service asks producers to estimate, the value of agricultural products sold directly to consumers for human consumption, thereby 1

Authors are Professor, Graduate Research Assistant and Graduate Research Assistant in the Department of Agricultural and Resource Economics at Colorado State University.

1

Western Economics Forum, Fall 2012 excluding some popular, directed marketed products such as cut flowers and nursery products. It should also be noted that these estimates exclude farms that report less than $1000 in gross commercial sales. Given Low and Vogel’s (2011) conclusions that direct markets are more important for small farmers, this censoring may be important. Moreover, if farmers and ranchers sell their products through vertically integrated cooperatives, value chains, or regional hubs, those sales are also absent from the numbers presented here. Still, we can learn from what data is available on the sample collected for direct marketing, including roadside stands, farmers markets, pick-your-own sites, and other consumer-focused markets. For the US, the value of agricultural products directly sold by producers under the above definition increased from $812,204,000 to $1,211,270,000 between 2002 and 2007, an increase of 49%. The number of farms direct marketing also increased from 116,733 to 136,817 (a 6.2% growth in farms), and in comparison, total farm numbers grew more slowly from 2.12 million to 2.2 million (3.4% increase). The revenue from direct sales on the average farm is very small in absolute terms, though it did increase significantly in relative terms from $6,958 to $8,853 (27% higher than five years earlier). As previously noted by Diamond and Soto (2009), the increase in direct marketing among most Western states is even more dramatic than US trends (Table 1). Between 2002 and 2007, 3,572 farms began direct marketing in this region (29,721 farms up from 26,149) so that, on average, 8.5% of all farms now do some direct marketing (compared to 6.2% for the US as a whole). In this region, the greatest share of farms direct marketing are in Oregon (16.3%) and Washington (13.8%), which could at least partially be explained by their longer growing seasons for fresh produce. But, this growth may also be fueled by the number of marketing support programs and infrastructure investments made (according to information available in the USDA Food compass, http://www.usda.gov/maps/maps/kyfcompassmap.htm). The greatest growth in number of farms direct marketing was in New Mexico and Utah (both with over 40% growth in numbers of farms participating since 2002). This increase in activity resulted in a 50% increase in direct sales revenues for the region as sales jumped from $219.8 to $330.2 million in 2007. Similarly, the average sales per farm increased from $6,612 to $7,884 (which translates to 19.2% growth). These revenues include channels outside of farmers markets as well, including roadside stands, Community Supported Agriculture programs (where members buy shares early in the season in exchange for weekly box deliveries) and pick-your-own farms, illustrating the significant shift to more diverse, customer-driven marketing strategies by Western producers. Among states, California accounts for 49% of the direct marketing revenues. As of 2007, Oregon, Washington, Colorado, New Mexico and Utah all exceeded $10 million dollars in direct sales as well. The greatest growth in direct marketing revenues originated with farms in Oregon (163%), New Mexico (70%), Utah (44%) and California (42%). Still, the average sales per farm grew by over 160% in Oregon, 30% in California, 19% in New Mexico and 2% in Utah, while the average sales per farm increased by an average of 20% when one considers the whole region. Still, Diamond and Soto (2009) show that in terms of relative importance, the Western region trails other parts of the U.S. (the Northeast and Mid-Atlantic) significantly, perhaps because of the extreme rurality of some areas of the West, or the climatic challenges of running markets in areas where the growing seasons are cut short by late Spring and early fall frosts.

2

Western Economics Forum, Fall 2012 Table 1- Direct Marketing Trends in Western US States, 2002-07 State

Arizona California Colorado Idaho Montana Nevada New Mexico Oregon Utah Washington Wyoming Western Region USA

Direct Mktg (Farms, 2007)

Direct Mktg (Farms, 2002)

Direct Mktg (% of Farms, 2007)

Direct Market Value 2002 ($,000)

Direct Market Value, 2007 ($,000)

Share of Value through Direct Mkt (2007)

Direct Market Sales Growth (20022007)

Direct Market Value/ Farm (2002)

Direct Market Value/ Farm (2007)

863 7,068 2,777 2,076 1,287 200 1,529 6,274 1,584 5,418 645 29,721

711 6,436 2,343 1,632 1,164 246 1,071 6,383 1,115 4,527 521 26,149

5.5% 8.7% 7.5% 8.2% 4.4% 6.4% 7.3% 16.3% 9.5% 13.8% 5.8% 8.5%

$3,911 $114,356 $17,406 $5,889 $4,523 $1,606 $6,582 $21,411 $6,983 $34,753 $2,381 $219,800

$5,247 $162,896 $22,584 $7,840 $6,321 $1,074 $11,193 $56,362 $10,098 $43,537 $3,025 $330,200

0.16% 0.48% 0.37% 0.14% 0.23% 0.21% 0.51% 1.29% 0.71% 0.64% 0.26% 0.61%

34.16% 42.45% 29.75% 33.13% 39.75% -33.13% 70.05% 163.24% 44.61% 25.28% 27.05% 50.2%

$5,501 $17,768 $7,429 $3,609 $3,886 $6,528 $6,146 $3,354 $6,262 $7,677 $4,570 $6,612

$6,080 $23,047 $8,133 $3,776 $4,911 $5,372 $7,320 $8,983 $6,375 $8,036 $4,690 $7,884

136,817

116,733

6.2%

$812,204

$1,211,270

0.41%

49.13%

$6,958

$8,853

3

Western Economics Forum, Fall 2012 Table 2: County in each State with Greatest Revenues from Direct Marketing and Greatest Growth in Direct Marketing between 2002 and 2007 Direct Growth in Marketing Direct Marketing State/County Revenues State/County Revenues (2007, (2002-2007, %) $,000s) Wyoming/Park 520 Wyoming/Sweetwater 316 Montana/Ravalli 576 California/Colusa 344 Nevada/Churchill 652 Nevada/Churchill 359 New Mexico/Bernalillo 1393 Idaho/Bonneville 497 Arizona/Maricopa 1549 Arizona/Graham 723 Idaho/Canyon 1829 Colorado/Archuleta 890 Utah/Utah 2824 Montana/Judith Basin 1113 Washington/Franklin 4086 Oregon/Jackson 1291 Colorado/Mesa 4729 Utah/Garfield 1675 Oregon/Jackson 13920 Washington/Franklin 1684 California/Fresno 17170 New Mexico/Union 3500 All of the top five counties with respect to total direct marketing revenues are in California, which is not surprising given its longer seasons and climates that can produce a wide variety of higher value, consumer-ready products (fruits, vegetables, nuts). Still, the highest growth in direct marketing activities is in states that have seen significant population growth (U.S. Census) and show a large number of marketing innovations in the USDA Food compass cited above (such as regional food hubs, value-added enterprises). These include New Mexico, Washington, Utah and Oregon, (all reporting upwards of 1000% growth). Table 2 shows the largest direct marketing counties in each state (ranked by total direct marketing revenues) and the county with the highest growth in direct marketing revenues in each state from 2002 to 2007. Since these are the counties where direct marketing farms are located, rather than where those products are sold, it may be interesting to look at common characteristics of these counties. Two of the counties with the largest amount of direct sales, Jackson County, Oregon, and Fresno County, California, are heavily urban-influenced as defined by the USDA-ERS urban influence codes. It is likely that some of this supply may “spillover” to other adjacent or nearby counties, especially since many consumer-dense areas have little production remaining in their counties. More specifically, in the case of Mesa County in Colorado, there is direct evidence from market reports that a number of fruit farms in that county ship as far as 350 miles to reach the Front Range markets of Colorado and Southern Wyoming where most of the consumer buying power is located but colder winters inhibit fruit tree production. When one turns to the counties with high growth in direct marketing, some counties near urban areas also stand out (Franklin and Jackson).Others are in areas where there have been projects to support the development of cooperatives and farmers markets over the past decade (as mapped in the USDA Know your Farmer, Know your Food website: accessed March 2013). Similar to the states themselves, there is a great degree of difference in the direct marketing revenues among the counties, with the Pacific Coast showing the greatest direct marketing activity. States that reported high population growth during the mid-2000’s, such as Utah, Washington and Colorado, also appear to have significant direct marketing activity and growth.

4

Western Economics Forum, Fall 2012 For example, Utah and Colorado both have seen sustained population growth, concentrated in corridors along the Rocky Mountains, have been identified as best places to live by magazines such as Money, and also have some of the highest growth direct sales counties.

Farmers Market Trends The presence, growth and new creation of farmers markets is one of the most apparent signals of consumer and producer interest in developing direct markets. Still, since there are some organizational costs in establishing such markets, such markets will only continue or develop in the presence of sufficient consumer demand (and word of mouth about traffic and sales by participating vendors). The USDA’s Agricultural Marketing Service recent analysis of a 2009 survey of market managers led them to conclude, “…growth in farmers market numbers, although still increasing, is continuing at a slower pace. The reduction in the growth rate of farmers markets may indicate that farmers markets are approaching a saturation point. . . raising questions about whether current levels of growth are sustainable (p. 78).” (Ragland and Tropp, 2009) Figure 1: National Count of Farmers Market Directory Self-Reported Listings 9,000 Number of Farmers Market

8,000 7,864

7,000 6,000 5,000 4,685

4,000 3,706

3,000 2,000

2,410

2,863

1,000 -

1996 2000 2004 2008 2012 Source: USDA-AMS-Marketing Services Division, 2012 Note: These are self-reported, and because responses are voluntary, may not be an accurate census of farmers markets. Still, they are one of the best cited measures of general trends in this sector

They continue to say that market managers should therefore be careful to understand their visitors and meet their expectations for a variety of key criteria that may affect their frequency of trips and purchasing behavior. Onozaka et al (2010) reported that a significant share of consumers value and purchase local food.Those choosing to shop in direct markets report relatively higher concern about the importance of protecting local farmland and supporting the local economy; therefore, this could be seen as a signal to policy makers of the relationship between the interest in local foods and emerging public policy issues. The number of farmers markets in the United States has grown dramatically, increasing 226 percent from 1996 to 2012, with over 7,800 farmers markets operating in the United States (Figure 1). Farmers markets also serve as a key direct marketing channel for small and mid-

5

Western Economics Forum, Fall 2012 size US producers (Diamond and Soto, 2009), so with recent growth in their numbers, one could perceive that the prevalence of direct sales may boost farm income. Table 3: Number of Markets listed in USDA’s National Farmers Market Directory State 2004 2006 2008 2010 2012 Arizona California Colorado Idaho Montana Nevada New Mexico Oregon Utah Washington

82

40 459

52 496

52 505

75 580

62

77

95

94

26 28

25 34

36 40

61 46

67

16 42

25 43

19 45

32 58

74 20

86 21

89 27

120 32

38 69 164

87

97

97

136

827 166 65

37 148

Wyoming 43 13 18 29 31 Table 3 depicts the number of markets listed in USDA’s National Farmers Market Directory. In 2012, the lion’s share of these markets is located in California followed by Colorado and Oregon. The Western States considered here comprise about 22% of the total count of farmers markets in the US in 2012. Since 2004, number of farmers markets has grown dramatically, representing 97% growth from 2004 to 2012. The increasing number of farmers markets may indicate growing importance and significance of direct marketing to producers in Western states.

Farmers Markets’ Role in the Health of Communities There are a number of studies that have attempted to evaluate and quantify the contribution of farmers markets and other direct marketing channels to a myriad of community development issues including public health, economic development and resiliency, as well as compilations that summarize key points of these studies (Martinez et al, 2009; Brown and Miller, 2008). Here we summarize highlights of these studies to suggest community implications that may be result from continued growth of such markets in the Western region. Story et al (2009) note that the American Public Health Association (APHA) and American Medical Association (AMA) have both passed resolutions concerning the linkage of a sustainable agriculture and food system to the public health of our nation, “….(where) a sustainable food system has been defined as one that . . . encourages local production and distribution infrastructures; makes nutritious food available, accessible, and affordable to all; is humane and just—protecting farmers and other workers, consumers, and communities (p. 223)”. Beyond this national group, numerous state and local based wellness organizations perceive benefits of having a viable farmers markets and direct marketing (Kaiser Permanente in California, Livewell Colorado, Oregon Public Health Institute). Still, others see these conclusions as conjecture, so we are beginning to see more in-depth research on the role of farmers markets on dietary and health outcomes. Thilmany McFadden, and Low (2012) used secondary data to begin exploring whether there may be some interdependence between direct markets and the extent to which households in the US incorporate USDA’s dietary guidelines in their daily diets. Specifically, the study used 2007

6

Western Economics Forum, Fall 2012 Census of Agriculture data and calculated the correlation between measures of local food marketing and health outcomes. They found out that direct sales, number of farmers markets and fruit and vegetable sales are negatively correlated with obesity and cardiovascular disease mortality rate; implying that the public health community’s conjectures about the role of more fresh produce-laden direct markets on helping to lower adult obesity rate and the cardiovascular death rate are not without merit. However to understand the broader relationships between local food marketing and health outcomes, the authors of the study have pointed out that more research should be done by incorporating more variables like income, education and population. Other studies have attempted to assess the positive health outcomes of increased local food consumption. By moving toward a more community tied food systems, high rates of unemployment rate and obesity can be reduced significantly (Conner and Levine, 2006). Low income children benefited more according to Chomitz, et al. (2010) when there is increased availability to healthy food and physical activity leading to reduced obesity. Lea (2005) argued that food that is locally obtained may be healthier because “they retain more nutrients than less fresh food.” Brown and Miller (2008) provided a fairly in-depth summary of the role those farmers markets may play in impacting farmers economic viability, the economies of the communities where markets exist, and in anchoring other local food system development activities taking place throughout the U.S. Hilchey, Lyson and Gillespie (2004) suggest that farmers markets enhance producers’ business opportunities, foster business skills and have positive effects on producervendor families. For the broader community, they concluded that markets may have spillover and multiplier effects to other adjacent businesses on market day, support entrepreneurial startups in agriculture and food industries while supporting food nutrition, security and educational goals. To examine the potential economic benefits of farmers markets and other producer-focused regional food project on communities, several studies report using regional input-output models (Gunter, Thilmany, and Sullins 2012). Examples include Hughes et al (2008) study that found farmers markets have a positive impact on the economy of West Virginia by creating more than 40 full time jobs and $1.07 million in net economic impact to the state. Otto and Varner (2005) estimated Iowa benefited up to $31.5 million from farmers markets. Cummings, et al. (1999) study on Ontario determined that farmers markets generated $1 billion in secondary effects. However, many caution that the conclusions drawn from these studies must be used carefully, since assumptions about region impacted, net benefits of new marketing channels (which may draw activity from competing sectors) and potential to strengthen community economies are all very place-based questions. There are also potential environmental benefits that are gained through local food systems, due to the shorter distance that food travels and the energy saved and reduced greenhouse emission. But, many researchers (including Martinez et al, 2010) argue that these results are inconclusive and transportation based full supply chain studies should be perused more in the future. Conclusions Martinez et al (2009) concluded that, “. . . findings are mixed on the impact of local food systems on local economic development and better nutrition levels among consumers, and sparse literature is so far inconclusive about whether localization reduces energy use or greenhouse gas emissions (page 3).” Generally, current trends suggest that farmers markets continue to grow in popularity as producers seek to personally connect with consumers. But, what seems

7

Western Economics Forum, Fall 2012 to be a more recent promotional tool for farmers markets are efforts made by partners in the economic development and public health communities. These partners’ interest in using farmers market to address broader community issues, and evaluate outcomes that address a community’s quality of life (jobs, health indicators), have encouraged them to bring new ideas and resources to farmers markets. But, if the sustainability of farmers markets is based on resources from these partners, it motivates the need to further explore the relationship between direct marketing, local economic development, and public health in data analysis, case study and evaluation activities related to direct markets. This paper updates the 2004 study that documented the notable expansion of the farmers market sector in the Western U.S. (Thilmany and Watson, 2004). In addition to more reported direct marketing activity, there appears to be an expanding policy agenda that implicates farmers markets and other food system relocalization efforts as a key ingredient of community health. However, it bears repeating that one should be careful in interpreting these numbers for two reasons: 1) some of these increases are from very small starting base numbers for revenues; and, 2) some believe these numbers are still lower than what actually occurs because the types of farms that direct market are less likely to be reporting to the Ag Census.

References: Brown, C. and S. Miller. 2008. “The Impacts of Local Markets: A Review of Research on Farmers Markets and Community Supported Agriculture (CSA).” Am. J. Agr. Econ. 90(5): 1298-1302. Chomitz, V. A., McGowan, R. J., Wendel, J. M., Williams, S. A., Cabral, H. J., King, S. E., et al. (2010). Healthy Living Cambridge Kids: A Community-based Participatory Effort to Promote Healthy Weight and Fitness. Obesity, 18(1s), S45-S53. Conner, D. S., & Levine, R. (2006). Circles of association: the connections of community-based food systems. Journal of Hunger & Environmental Nutrition, 5-25. Cummings, H., G. Kora, and D. Murray. 1999. “Farmers Markets in Ontario and Their Economic Impact 1998.” School of Rural Planning and Development, University of Guelph. Diamond, A and R. Soto. Facts on Direct-to-Consumer Food Marketing: Incorporating Data from the 2007 Census of Agriculture. USDA-Ag Marketing Service. May 2009. Gunter, A., D. Thilmany and M. Sullins. 2012. What is the New Version of Scale Efficient: A Values-Based Supply Chain Approach. Proceedings of the Journal of Food Distribution Research. 43(1): 27-34. Hilchey, D., T. Lyson and G. Gillespie. 2004. “Farmers Markets and Rural Economic Development.” Community and Economic Development Toolbox. Hughes, D.W., C. Brown, S. Miller, and T. McConnell. 2008. “Evaluating the economic impact of farmers markets in an opportunity cost framework: A West Virginia example.” Journal of Agricultural and Applied Economics 40(1):253-265. Lea, E. (2005). Food, health, the environment and consumers’ dietary choices. Nutrition & Dietetics, 62(1), 21-25. Martinez, S., M. Hand, M. Da Pra, S. Pollack, K. Ralston, T. Smith, S. Vogel, S. Clark, L. Lohr, S. Low, and C. Newman. 2010. Local Food Systems: Concepts, Impacts, and Issues, ERR 97, U.S. Department of Agriculture, Economic Research Service, May 2010. Onozaka,Y. G. Nurse, and D. Thilmany McFadden. 2010. “Local Food Consumers: How Motivations and Perceptions Translate to Buying Behavior.” CHOICES. 1st Quarter 2010 | 25(1).

8

Western Economics Forum, Fall 2012 Otto, D. and T. Varner. 2005. “Consumers, vendors, and the Economic Importance of Iowa Farmers Markets: An Economic Impact Survey Analysis.” Iowa State University, Leopold Center for Sustainable Agriculture. March 2005. Payne, T. 2002. US Farmers Markets—2000: A Study of Emerging Trend. Agricultural Marketing Service, USDA, Washington, DC (http://www.ams.usda.gov/directmarketing/) Ragland, E. and. D. Tropp. 2009. 2006 National Farmers Market Manager Survey. USDAAgricultural Marketing Service. http://www.ams.usda.gov/AMSv1.0/getfile?dDocName=STELPRDC5077203&acct=wdm geninfo Story, M., Hamm, M.W, Wallinga, D. 2009. Food Systems and Public Health: Linkages to Achieve Healthier Diets and Healthier Communities. Journal of Hunger and Environmental Nutrition 4(3-4): 219-224. Thilmany McFadden, D. and S.A.Low. 2012.”Will Local Foods Influence American Diets?” CHOICES: The Magazine of Food, Farm and Resource Issues. Thilmany, D. and P. Watson. 2004. “The Increasing Role of Direct Marketing and Farmers Markets for Western US Producers.” Western Economics Forum 3(December): 19-25. USDA Agricultural Census. County Level Data. Volume 1. http://www.agcensus.usda.gov/Publications/2007/index.php (Accessed September 2012) USDA Agricultural Marketing Service. Farmers Market Directory. http://www.ams.usda.gov/farmersmarkets/map.htm (Accessed September 2012) USDA Know your Farmer, Know your Food Compass. http://www.usda.gov/wps/portal/usda/usdahome?navid=KYF_COMPASS (Accessed March 2013

9


Wheat Variety Selection: An Application of Portfolio Theory in Colorado1 Ryan Mortenson2, Jay Parsons3, Dustin L. Pendell4 and Scott D. Haley5 Introduction and Background Each year prior to the growing season, wheat growers are faced with choices when it comes to selecting which wheat varieties to plant. Several Land Grant Universities annually publish results of wheat variety performance trials where both private and public wheat varieties are tested. From these outreach publications, wheat growers can get a reliable sense of the expected performance of the trial varieties for their location. Intuitively, growers select wheat varieties based on previous experiences and the published trial results of the previous year. The correlation between yield performances of the different varieties is largely ignored and a more thorough investigation could lead to increased yield stability. As expected, any agricultural activity involves risk from diverse sources such as weather variation or disease. Barkley, Peterson, and Shroyer (2010) identified three major strategies to reduce risk in wheat production. The first strategy to reduce risk involves the development of new breeds with agronomic characteristics appropriate to the growing region. The traits of multiple varieties can be combined to create new cultivars that will potentially reduce the variation of yields. The second strategy is to create mixtures or blends of the seed of a few different varieties prior to planting in order to increase the genetic diversity. The third strategy is to create a portfolio by selecting multiple wheat varieties and planting them in different fields. The number of planted acres of wheat has stayed consistent over the past 10 years in Colorado; therefore, one way to maintain and possibly increase wheat yields is through better risk management strategies. According to Bosley (2010), Colorado growers tend to plant two or three different varieties of wheat in a given year. The selection of varieties is made primarily by a combination of previous experiences, gut feelings, suggestions made by friends, family or seed distributors and an examination of the test plot yields from the previous year. Through the examination of the year-to-year variance of a given cultivar (variety), and comparing that with the variance and covariance of other cultivars, “portfolios” of wheat varieties can be developed. The portfolios lie graphically on a single line and represent points where variation is minimized for a given level of yield. This line represents the mean-variance 1

Address correspondence to Jay Parsons, Department of Agricultural and Resource Economics, Clark B320, Colorado State University, Fort Collins, CO 80523. Email: [email protected] 2

Former Graduate Student, Department of Agricultural and Resource Economics, Colorado State University 3

Special Assistant Professor, Department of Agricultural and Resource Economics, Colorado State University

4

Associate Professor, Department of Agricultural and Resource Economics, Colorado State University

5

Professor, Department of Soil and Crop Science, Colorado State University

10

Western Economics Forum, Fall 2012 efficiency frontier. Portfolios can be developed based on the producers’ risk preferences, whether it is to maximize yield given a target variance or minimize variance given a target yield. The term portfolio originates from finance and refers to a group of financial instruments such as investments, holdings, and funds that are used to stabilize or reduce exposure to the risks of the financial market. The term is appropriate for wheat variety analysis in the sense that creating a portfolio of wheat varieties helps reduce wheat producers’ exposure to yield risk. There are a several recent studies that have used portfolio theory on grain crops including Nalley et al. (2009) on rice varieties grown in Arkansas; Nalley and Barkley (2010) on wheat varietal selection in Yaqui Valley of Northwestern Mexico; Barkley, Peterson, and Shroyer (2010) in Kansas wheat varietal selection; and Park et al. (2012) on wheat selection for dryland wheat producers in the Texas High Plains. This paper applies existing portfolio theory methods to wheat varietal selection to help Colorado wheat producers make more informed planting decisions. Portfolios are created for northeast and southeast Colorado. The estimated standard deviation is used as a proxy for measuring the “risk” or variation of a given wheat variety portfolio. Although applying portfolio methods to wheat production has been done in Kansas and Texas, this is the first known study to evaluate wheat varieties in Colorado in this manner. This is important because producers in Colorado generally grow different varieties than producers in those states (USDA/NASS Kansas Field Office 2012; USDA/NASS Texas Field Office, 2012). The timing of this present study is especially important given that it includes several popular varieties recently released by the Colorado State University Wheat Breeding and Genetics Program with different trait characteristics designed to address specific Colorado growing conditions.

Methodology and Data The model used in this study to estimate the efficiency frontier for Colorado wheat varieties is based on research by Markowitz (1952). In this research, the method of minimizing the expected variation, as measured by standard deviation, subject to a given level of expected (mean) yield, is used. The frontier is estimated by solving a sequence of quadratic programming problems. It is assumed that a wheat producer has a given number of acres (X) and wishes to produce on the efficiency frontier of mean-variance (MV) by allocating X acres to a combination of varieties. The variable xi represents the percentage of total acres planted of variety i where i = 1, …, n and Σixi = X or 100% of the producer’s land dedicated to wheat production. This frontier is the maximization of the mean yields given a target level of variation or the minimization of variation given a target mean yield. By defining y i as the mean yield of variety i, the total wheat yield will be the weighted mean yield, equal to: Σixiyi. The MV efficiency frontier is estimated by minimizing total farm variation (V) for each possible level of mean yields (yi) as given in equation (1): (1)

min V = ΣiΣjxixjσij, for a given level of λ subject to xi ≥ 0 for all i.

The total wheat variety yield variation (V) is defined as: (2)

V = ΣiΣjxixjσij

11


where xi is the percentage of total acres planted to variety i and xj is the percentage of total acres planted to variety j, σij is the covariance of yields for varieties i and j and σij is the variance when i = j. Hazell and Norton (1986) explain that the intuition behind equation (2) is that by combining varieties that have negatively related covariates, a more stable yield will likely occur. Also, a variety that may appear to be risky or have a large variance can still be an option when combined with a variety that shares a negative covariate. The constraint ensures non-negative returns after the quadratic (i.e., it is not possible to plant a negative percentage of wheat variety i). The sum of the mean yields for varieties x and y are set equal to λ, where λ is the target yield for a given portfolio: (3)

λ = Σ ix iy i .

By varying the target yield (λ) over the feasible range, the MV efficiency frontier can be estimated. The same process described above can be performed using a target variation (standard deviation) instead of a target yield. This allows a producer to maximize yield for a given target level of variation. Data on wheat yields are obtained from the Colorado Wheat Variety Database (Colorado State University Wheat Breeding and Genetics Program). Yields from 2000 – 2011 for dryland trial locations throughout Colorado are used to carry out the analysis. The varieties selected are based on three sets of criteria: 1) the variety is tested in the CSU trials; 2) the variety appears within the National Agricultural Statistics Service (NASS) annual publication “Winter Wheat Varieties” for Colorado for the years 2009, 2010, and 2011; and 3) there are at least three years of comparable mean yields between each variety used to estimate the covariates. A total of 13 wheat varieties met the above criteria and are selected for the analysis. The resulting varietal selection can be seen in Table1. Table 1. Selected Colorado Wheat Varieties Source, Year of Release, and Percent Planted Acres, 2009-2011 Variety Source Year 2009 2010 2011 Above CSU 2001 3.2% 3.2% 2.8% Akron/Ankor CSU 1994/2002 2.8% 2.6% 1.3% Bill Brown CSU 2007 0.0% 2.5% 5.1% Bond CL CSU 2004 4.8% 4.9% 3.9% Danby KSU 2005 1.2% 0.0% 0.0% Hatcher CSU 2004 32.9% 26.5% 34.5% Jagalene Agripro 2001 8.4% 6.8% 1.6% Jagger KSU 1994 4.0% 3.2% 1.9% Prairie Red CSU 1998 5.6% 5.6% 1.5% Prowers 99 CSU 1999 2.0% 1.6% 0.0% Ripper CSU 2006 6.8% 12.5% 12.1% TAM 111 TAMU 2002 8.0% 7.5% 9.5% Yuma CSU 1991 2.7% 1.1% 0.0% Total Wheat 2,630,000 2,478,000 2,345,000 Acres Planted Source: USDA/NASS Colorado Agricultural Statistics Service.

12


Summary statistics and the coefficients of variation are reported for the Northeast region and Southeast region of Colorado in Table 2 and Table 3, respectively. Because there are distinct differences in production levels between Northeast and Southeast Colorado, this study divides the data to develop separate wheat portfolios that are appropriate for the given region. Table 2. Selected Variety Summary Statistics: Northeast Colorado Region, 2000-2011 Mean Standard Coefficient Min Max Observations Variety Annual Yield Deviation of Variation Ripper 50.48 11.84 0.24 4.76 87.65 48 Bill Brown 49.61 11.93 0.24 12.34 84.31 40 Bond CL 49.38 13.33 0.27 10.91 97.26 51 Hatcher 49.09 13.37 0.27 2.17 97.61 56 TAM 111 47.83 15.48 0.32 4.17 101.27 47 Above 47.66 12.52 0.26 5.31 93.06 61 Jagger 46.60 10.85 0.23 13.57 93.17 61 Danby 46.24 14.26 0.31 3.83 83.45 40 Prairie Red 46.02 11.17 0.24 6.02 88.47 61 Jagalene 44.88 12.26 0.27 4.34 90.57 42 Yuma 44.58 12.48 0.28 6.42 93.36 52 Akron/Ankor 41.93 11.78 0.28 3.94 89.39 47 Prowers 99 40.07 10.09 0.25 6.71 83.31 47 Source: USDA/NASS Colorado Agricultural Statistics Service. In the Northeast region of Colorado, Ripper had the highest average yield at 50.48 bu./ac. followed by Bill Brown (49.61 bu./ac.) and Bond CL (49.38 bu./ac.). Prowers 99 had the lowest yield and the lowest variation (Table 2). In the Southeast region, mean yields are slightly lower than in the Northeast region. Ripper had the highest average yield with 44.86 bu./ac. followed by Bill Brown (44.64 bu./ac.) and Hatcher (44.21 bu./ac.). Similar to the Northeastern region, Prowers99 had the lowest yield (34.04 bu./ac.). However, Akron/Ankor had the lowest variation in the Southeast region (Table 3). Table 3. Selected Variety Summary Statistics: Southeast Colorado Region, 2000-2011 Mean Standard Coefficient Min Max Observations Variety Annual Yield Deviation of Variation Ripper 44.86 9.30 0.21 15.03 75.59 26 Bill Brown 44.64 12.12 0.27 14.65 70.50 23 Hatcher 44.21 11.31 0.26 13.42 76.71 29 Bond CL 42.23 9.43 0.22 15.41 68.09 26 Danby 41.77 10.81 0.26 13.13 68.30 23 Above 40.91 8.44 0.21 13.51 62.80 32 TAM 111 40.53 12.67 0.31 11.70 77.38 24 Prairie Red 39.23 9.10 0.23 10.37 59.48 32 Akron/Ankor 37.55 8.41 0.22 15.37 69.18 23 Jagger 37.01 10.18 0.28 9.99 68.80 32 Yuma 36.79 9.27 0.25 16.56 71.27 26 Jagalene 36.43 11.61 0.32 14.18 74.68 20 Prowers 99 34.04 8.90 0.26 12.56 58.11 25 Source: USDA/NASS Colorado Agricultural Statistics Service.

13

Western Economics Forum, Fall 2012 Through the application of portfolio theory to Colorado varietal selection, wheat producers can potentially increase yield and reduce variability by combining wheat varieties that respond differently to growing environments. Through the calculation of means, standard deviations and covariates, it can be estimated as to how each variety’s yield responds to different environmental factors relative to each of the other varieties. Ideally, varieties that have a negative covariate would be integrated into the planting plans to reduce risk.

Estimation Procedures and Results Complete data on wheat variety yield means, standard deviations and covariances are used to estimate wheat portfolios along the efficiency frontier. Standard deviations are estimated across years and pairwise covariates of the selected wheat varieties are estimated. By varying the target yield, while minimizing the standard deviation for the given target yield, the optimal portfolios are established and efficiency frontiers are constructed. A Variance/Covariance matrix for the Northeast and Southeast regions can be found in Tables A1 and A2 of the Appendix, respectively. 2011 Actual Portfolio vs. 2011 Potential Portfolio The following wheat varieties: Above, Akron/Ankor, Bill Brown, Bond CL, Hatcher, Jagalene, Jagger, Prairie Red, Ripper, and TAM 111 were listed in NASS’s “Winter Wheat Varieties – 2011 Crop” and accounted for 75.2% of total acres planted statewide (USDA/NASS Colorado Field Office, 2012). The survey also provides the planted acres percentages for the Northeast and Southeast regions. By proportioning the varieties’ percentage planted to equal 100%, it allows the estimation of the variation (V) and mean yield (E) for the actual portfolio in 2011 (2011 Actual Portfolio). The variation is then held constant at the 2011 Actual Portfolio level for each region and quadratic programming is used to maximize the mean yield providing an estimate of the 2011 Potential Portfolio for each region. 6 Northeast Region Efficiency Frontier Portfolio Results The standard deviation of the 2011 Actual Portfolio (12.91 bu./ac.) was estimated, and the expected yield was maximized using quadratic programing, allowing for the estimation of the 2011 Potential Portfolio for the Northeast region. The estimated yield difference between the two portfolios was nearly 0.5 bu./ac. Ripper was the highest yielding variety at 50.5 bu./acre (Table 2) and constitutes the highest point on the efficiency frontier (Figure 1). Prowers 99 was the variety with the lowest variation with a standard deviation of 10.09 bu./ac. (Table 2) and is the left most and lowest point on the efficiency frontier (Figure 1). Using these two points as the extremes, an efficiency frontier was drawn between the two points by varying the target mean yield and then minimizing the portfolio variance for the given varied yield. Several portfolios were developed representing the points along the efficiency frontier between the two extremes (Table 4). The portfolios contain the percentage of each variety to be planted in order to obtain certain levels of yield and variation.

6

A statewide analysis was also conducted for Colorado. The statewide analysis results are not reported to here to conserve space and because of the similarities between the statewide and regional analyses (specifically Southeast region). The results for the statewide analysis are available from the authors upon request.

14

Western Economics Forum, Fall 2012 Three portfolios offer the lowest coefficient of variation (CV). A portfolio of 89% Jagger and 11% Ripper (CV = 0.23), a portfolio of 59.1% Jagger and 40.9% Ripper and a portfolio of 29.2% Jagger and 70.8% Ripper. These three portfolios could be suggested to those farmers looking to minimize risk while keeping expected yields relatively high. Choosing the latter portfolio would increase yield by 0.25 bushels per acre when compared to the “Actual Portfolio resulting in a $650,100 increase in production for the Northeast Region.7 Figure 1 shows the steepness of the efficiency frontier drawn by the portfolios found in Table 4. Table 4. Portfolio Analysis of Northeast Region Wheat Varieties, 2000-2011 Target Standard Mean Yield Deviation Coefficient of Variation Portfolio (Bu./Acre) (Bu./Acre) 100% Prowers 99

40.07

10.09

0.25

17.8% Jagger 82.2% Prowers 99

41.23

10.27

0.25


42.39

10.44

0.25


43.55

10.58

0.24

71% Jagger 29% Prowers 99

44.71

10.69

0.24

87.1% Jagger 1.8% Prairie Red 11.1% Prowers 99

45.87

10.79

0.24

89% Jagger 11% Ripper

47.03

10.94

0.23

59.1% Jagger 40.9 Ripper

48.19

11.22

0.23

29.2% Jagger 70.8% Ripper

49.35

11.52

0.23

100%Ripper

50.48

11.84

0.24

2011 Actual Portfolio of Planted Varieties in Northeast Coloradoa

49.10

12.91

0.26

2011Potential Portfoliob 82% Bond CL 17.7% Ripper

49.58

12.91

0.26

a

The “2011 Actual Portfolio” defined here is based on the percentage planted from the NASS 2011 publication and those varieties found in the CSU Trials, proportioned to equal 100%. b The “2011 Potential Portfolio” is estimated by maximizing the target yield while holding the variance at the 2011 Actual Portfolio variance. 7

Estimated by multiplying 0.25 with 394,000 acres of wheat planted (NASS) and a wheat price received of $6.60/bu for 2011 (NASS).

15

Western Economics Forum, Fall 2012 By moving left from the 2011 Potential Portfolio for the Northeast region towards a portfolio that lies on the efficiency frontier an estimated 11% reduction in risk, as measured by the standard deviation can be achieved without giving up potential yield. In fact, some of the estimated portfolios on the efficiency frontier would both increase expected yield and reduce the variation when compared with the 2011 Potential Portfolio for the Northeast region.

Figure 1. Northeast Colorado Region Wheat Efficiency Frontier, 2011 Southeast Region Efficiency Frontier Portfolio Results By holding the standard deviation of the Actual Portfolio (11.24 bu./ac.) constant and maximizing the expected yield an estimate of the 2011 Potential Portfolio for the Southeast region can be calculated. The estimated yield difference between the Actual and the Potential portfolio for the Southeast region was nearly one bu./ac. (Table 5). The Southeast region analysis offers some very interesting results. A single variety did not have the lowest variation, but rather a portfolio produced the lowest variation. This provides empirical evidence towards Hazell and Norton’s (1986) discussion that creating a portfolio of varieties that have negatively related covariates can produce a more stable yielding result. A portfolio of 43.4% Akron/Ankor, 23.9% Prairie Red and 32.9% Prowers 99 would result in a minimized standard deviation of 8.08 bu./ac. in the Southeast region (Table 4), whereas the best any one variety could do is a standard deviation of 8.41 bu./ac..

16

Western Economics Forum, Fall 2012 Table 5. Portfolio Analysis of Southeast Region Wheat Varieties, 2000-2011 Standard Target Mean Yield Coefficient of Portfolio Deviation (Bu./Acre) Variation (Bu./Acre) 43.4% Akron/Ankor 23.9% Prairie Red 36.80 8.08 0.22 32.9% Prowers 99 22.5% Above 32.4% Akron/Ankor 18.8 % Prairie Red 26.3% Prowers 99

37.70

8.11

0.22

38.60

8.17

0.21

39.50

8.25

0.21

92.6% Above 7.4% Prowers 99

40.40

8.36

0.21

90.1% Above 9.9% Ripper

41.30

8.55

0.21

42.20

8.77

0.21

43.10

8.97

0.21

0.2% Above 32.4% Bond CL 67.4% Ripper

44.00

9.11

0.21

100% Ripper

44.86

9.30

0.21

2011 Actual Portfolio of Planted Varieties in Southeast Coloradoa

43.84

11.24

0.26

2011 Potential Portfoliob 28.1% Bill Brown 71.9% Ripper

44.80

11.24

0.26

48.3% Above 18.8% Akron/Ankor 11.2% Prairie Red 21.7% Prowers 99 74.2% Above 5.1% Akron/Ankor 3.5% Prairie Red 17.2% Prowers 99

67.4% Above 32.6% Ripper 44.6% Above 55.4% Ripper

a

The “2011 Actual Portfolio” defined here is based on the percentage planted from the NASS 2011 publication and those varieties found in the CSU Trials, proportioned to equal 100%. b The “2011 Potential Portfolio” is estimated by maximizing the target yield while holding the variance at the 2011 Actual Portfolio variance.

17

Western Economics Forum, Fall 2012 Using the portfolio with the smallest variation and the variety with the highest yield, a frontier was constructed for the Southeast region that resulted in the portfolios found in Table 5 and depicted graphically in Figure 2. Three of the portfolios offer equal coefficients of variation and could be good recommendations to growers. Portfolios made up of 92.6% Above and 7.4% Prowers 99, 90.1% Above and 9.9% Ripper, or 0.2% Above, 32.4% Bond CL, and 67.4% Ripper all have the smallest CV of 0.21 for the Southeast region. The latter portfolio, when compared to the “Actual Portfolio” offers the potential of an additional 0.16 bushels per acre resulting in an additional value of $396,000 to wheat producers in the Southwest Region.8 A move from the 2011 Actual Portfolio for the Southeast region to the 2011 Potential Portfolio provides a small 2% increase in expected yield while maintaining the same level of variation. However, a leftward movement from the 2011 Actual Portfolio to an estimated portfolio that lies on the efficiency frontier has the potential of reducing risk by 19% as measured by the standard deviation without reducing yield. Furthermore, there are portfolios on the efficiency frontier that offer slight increases in expected wheat yield along with a significant decrease in variation (see Figure 2).

Figure 2. Southeast Colorado Region Wheat Efficiency Frontier, 2011

8

Estimated by multiplying 0.16 with 375,000 acres of wheat planted (NASS) and a wheat price received of $6.60/bu for 2011 (NASS).

18


Conclusions and Implications As an addition to the many tools already available to wheat growers in Colorado, the creation of variety portfolios offers a statistical method to help minimize risk and stabilize yields. This application of portfolio theory to Colorado wheat offers a quantitative look at the relationship among wheat varieties. By analyzing the covariates of wheat varieties, growers can take advantage of the ways in which the varieties react to different growing conditions. This analysis found that double-digit percentage decreases in risk as measured by the standard deviation can be achieved by Colorado wheat producers without sacrificing potential yield. According to our analysis, this potential reduction in risk is greater in the Southeast quadrant of the state than it is in the Northeast (19% versus 11%). Furthermore, it was found that portfolios exist on the risk-return efficiency frontier in both the Northeast and the Southeast growing regions of Colorado whereby wheat producers have the potential to slightly increase expected yield and significantly decrease yield variation. All varieties included in this study had at least three years of trial data but many had more than three years. Therefore, a couple of acknowledge limitations of this study are that some varieties may look artificially good or bad depending upon the growing conditions for the years they were included in the trial data and the very latest varieties with less than three years of data are not included in our analysis. However, the results of this analysis seem to fit with anecdotal grower experiences over the last several years. This suggests that this study and the model it contains could provide a powerful tool for helping producers make effective wheat variety planting decisions from a risk management perspective.

References Barkley, A., H.H. Peterson, and J. Shroyer. 2010. “Wheat Variety Selection to Maximize Returns and Minimize Risk: An Application of Portfolio Theory.” Journal of Agricultural and Applied Economics 42(1):39-55. Bosley, B. 2010. “2010 Colorado Wheat Improvement Work Team Survey of Wheat Growers.” Colorado State University. Colorado State University Wheat Breeding and Genetics Program. 2012. http://wheat.colostate.edu/CSUWheatBreeding/Database.html. Last accessed August 2012. Hazell, P.B.R., R.D. Norton. 1986. Mathematical Programming for Economic Analysis in Agriculture. New York: MacMillan Publishing Company. Markowitz, H. 1952. “Portfolio Selection.” The Journal of Finance 7(1):77-91. Nalley, L.L. and A. Barkley. 2010. “Using Portfolio Theory to Enhance Wheat Yield Stability in Low-Income Nations: An Application in the Yaqui Valley of Northwestern Mexico.” Journal of Agricultural and Resource Economics 35(2):334-347. Nalley, L.L., A. Barkley, B. Watkins, and J. Hignight. 2009. “Enhancing Farm Profitability through Portfolio Analysis: the Case of Spatial Rice Variety Selection.” Journal of Agricultural and Applied Economics 41(3):641-652.

19

Western Economics Forum, Fall 2012 USDA National Agricultural Statistics Service (NASS). Various Years. “Winter Wheat Seedings by Variety Survey.” USDA/NASS Colorado Field Office. Park, S.C., J. Cho, S.J. Bevers, S. Amosson, J.C. Rudd. 2012 Paper prepared for presentation at the Southern Agricultural Economics Association Annual Meeting, Birmingham, Alabama, 4-7 February. U.S. Department of Agriculture National Agricultural Statistics Service (USDA/NASS) Colorado Field Office. “Winter Wheat Varieties.” Retrieved from http://www.nass.usda.gov/Statistics_by_State/Colorado/Publications/Special_Interest_R eports/index.asp November 8, 2012. U.S. Department of Agriculture National Agricultural Statistics Service (USDA/NASS) Kansas Field Office. “Wheat Varieties.” Retrieved from http://www.nass.usda.gov/Statistics_by_State/Kansas/Publications/Crops/Whtvar/index. asp November 8, 2012. U.S. Department of Agriculture National Agricultural Statistics Service (USDA/NASS) Texas Field Office. “Wheat Variety Results.” Retrieved from http://www.nass.usda.gov/Statistics_by_State/Texas/Publications/Crop_Reports/Wheat/t wheat_var.htm November 8, 2012.

20

Appendix Table A1. Northeast Variance/Covariance Matrix

Above Akron/Ankor Bill Brown Bond CL Danby Hatcher Jagalene Jagger Prairie Red Prowers 99 Ripper TAM 111 Yuma

Above Akron/Ankor Bill Brown Bond CL Danby Hatcher Jagalene Jagger Prairie Red Prowers 99 Ripper TAM 111 Yuma

Above

Akron/Ankor

Bill Brown

Bond CL

Danby

Hatcher

Jagalene

Jagger

Prairie Red

Prowers 99

Ripper

TAM 111

Yuma

156.7337 145.3474 181.2197 174.2980 200.2007 170.4131 168.9258 130.2716 139.0155 133.0614 155.7298 155.7298 154.3097

145.3474 138.6889 166.1552 176.0958 200.8963 167.5007 156.3714 125.8534 131.3519 120.7634 169.9424 198.1911 154.3097

181.2197 166.1552 142.3727 154.8137 166.8779 162.4318 163.1433 136.9370 149.3487 169.4237 147.8465 147.8465 161.1132

174.2980 176.0958 154.8137 177.8010 183.4218 182.3091 164.0084 151.7798 156.0470 163.6424 143.8880 209.6938 189.3869

200.2007 200.8963 166.8779 183.4218 203.3953 198.5493 192.0854 167.5954 177.6258 195.6767 176.5349 256.0343 193.6006

170.4131 167.5007 162.4318 182.3091 198.5493 178.7384 170.9361 143.8908 151.3601 149.8616 155.6225 155.6225 178.0601

168.9258 156.3714 163.1433 164.0084 192.0854 170.9361 150.2280 144.1234 152.0757 143.6160 157.5391 180.3579 163.5015

130.2716 125.8534 136.9370 151.7798 167.5954 143.8908 144.1234 117.6381 115.8604 112.9358 126.9766 184.1481 135.2790

139.0155 131.3519 149.3487 156.0470 177.6258 151.3601 152.0757 115.8604 124.6576 121.2769 138.9878 191.7038 139.3252

133.0614 120.7634 169.4237 163.6424 195.6767 149.8616 143.6160 112.9358 121.2769 101.7585 156.9093 178.3648 129.7346

155.7298 169.9424 147.8465 143.8880 176.5349 155.6225 157.5391 126.9766 138.9878 156.9093 140.2153 192.5013 170.5154

155.7298 198.1911 147.8465 209.6938 256.0343 155.6225 180.3579 184.1481 191.7038 178.3648 192.5013 239.7455 205.1560

154.3097 154.3097 161.1132 189.3869 193.6006 178.0601 163.5015 135.2790 139.3252 129.7346 170.5154 205.1560 155.7875

Above

Akron/Ankor

Bill Brown

Bond CL

Danby

Hatcher

Jagalene

Jagger

Prairie Red

Prowers 99

Ripper

TAM 111

Yuma

71.3174 67.7452 112.1091 80.2096 103.3449 92.7332 96.5551 79.5357 70.1884 60.3631 80.3758 115.6296 71.2686

67.7452 70.7528 119.4692 79.2067 114.9311 91.6882 100.8925 82.9447 62.8130 59.9136 76.8059 116.3176 72.4105

112.1091 119.4692 146.9181 123.1223 123.4512 153.3947 142.0586 125.2335 119.3786 143.4522 109.9223 169.0942 142.5819

80.2096 79.2067 123.1223 88.8406 108.2071 107.7546 89.6794 92.5135 85.9545 90.0164 77.8585 115.9512 91.2421

103.3449 114.9311 123.4512 108.2071 116.7728 135.5307 127.7653 110.8919 111.2004 129.9894 90.4854 155.0817 131.9668

92.7332 91.6882 153.3947 107.7546 135.5307 127.8755 125.2267 108.5331 93.8531 93.8531 97.2137 148.6502 110.1024

96.5551 100.8925 142.0586 89.6794 127.7653 125.2267 134.7978 121.9989 100.6282 82.8992 86.9002 143.2513 108.5611

79.5357 82.9447 125.2335 92.5135 110.8919 108.5331 121.9989 103.6899 73.8580 84.4621 95.8871 140.9424 94.2968

70.1884 62.8130 119.3786 85.9545 111.2004 93.8531 100.6282 73.8580 82.7742 55.8694 76.4002 122.3268 63.9256

60.3631 59.9136 143.4522 90.0164 129.9894 93.8531 82.8992 84.4621 55.8694 79.2211 73.9364 107.6801 72.3909

80.3758 76.8059 109.9223 77.8585 90.4854 97.2137 86.9002 95.8871 76.4002 73.9364 86.4532 108.1636 84.3044

115.6296 116.3176 169.0942 115.9512 155.0817 148.6502 143.2513 140.9424 122.3268 107.6801 108.1636 160.4715 128.8454

71.2686 72.4105 142.5819 91.2421 131.9668 110.1024 108.5611 94.2968 63.9256 72.3909 84.3044 128.8454 85.9360


What can a comprehensive plan really tell us about a region?: A cluster analysis of county comprehensive plans in Idaho Nicholas S. Brown1 and Phil Watson2 Introduction Comprehensive plans are used by communities and regions to set goals and guidelines for future growth and development. Comprehensive plans are developed by regions to “advance the welfare of the people” and foster the creation of “better social, economic, and physical environments” (Bammi and Bammi 1979). These plans are intended to represent the priorities of the community and direct how future development should occur. They generally require public participation in the decision-making process and are often required by funding agencies. While comprehensive plans are not regulatory themselves, policy makers and managers generally cite how a proposed management action corresponds to the region’s comprehensive plan and comprehensive plans are increasingly being used to provide legitimacy to management actions (Sullivan 2004). Comprehensive plans can be done by any size community but are found regularly at the city and county level. As each community produces its own comprehensive plan, there is variation of plan emphasis, quality, and elements of uniqueness reflective of each community. Some attempts at standardization have been done either through state mandates or model sections created by advocacy groups, however, differences remain which makes comparing and contrasting plans difficult. Questions remain, then, as to how much heterogeneity exists in regional comprehensive plans and whether comprehensive plans really do reflect variation across regions. This study attempts to deal with differences in plans in order to see if the content of the plans are reflective of independent traits of the county. It does this by looking solely at the content of county comprehensive plan objectives in order to determine the priorities of the counties. It is hypothesized that the objectives enumerated in the comprehensive plans will indicate broader county goals, priorities, and desired outcomes for the locality and the region, which will be seen in the independent traits of the community. This is done by rating objectives in eight required sections of comprehensive plans in the state of Idaho. A cluster analysis is then performed to create four groups of counties in order to compare the comprehensive plan objectives of these counties to other independent traits. This method, using Idaho as an example, shows that different types of counties do cluster with comprehensive plan traits and while no causality can be established here, this correlative information could possibly be used for improving future planning efforts.

1

Fiscal Analyst, Legislative Fiscal Division, State of Montana Assistant Professor, Department of Agricultural Economics and Rural Sociology and the Bioregional Planning Graduate Program, University of Idaho 2

22


Review of Prior Literature Previous research has attempted to create comparison criteria for comprehensive plans for the purposes of improved evaluation. For example, Baer (1997) created a list of criteria to evaluate plans along with a categorization for types and timing of evaluation. Grimes, in an evaluation for the Virginia Transportation Research Council, identified objectives by evaluation inventory, assessments, and recommendations in plans in Virginia (Grimes 2006). Berke and Manta (2009) used a process of looking at objectives of plans in terms of recommended versus required for 30 comprehensive plans by grouping words in the policies such as encourage, consider, or should, versus words such as shall, will, or require (Berke and Manta 2000). Evans-Cowley and Gough took this and incorporated a third rating in an evaluation of New Urbanist plans in post-Katrina Mississippi. Here they used a 0, 1, and 2 rating, an element received a 0 if it was not present, a 1 if it was present but only recommended, and a 2 if it was required (Evans-Cowley and Gough 2009). A combination of these methods was used, detailed in the methodology section, in order to place numerical values to comprehensive plan objectives for use in a cluster analysis. Cluster analysis is a common technique in data analysis. It was first described in land use analysis in the 1970’s, although not recommended as the computing power at that time was not adequate (Hopkins 1977). More recently, cluster analyses have been used for varied projects such as finding areas for reinvestment in Philadelphia (Schamess 2006), and finding clusters of different groups of people based on health in two metropolitan areas (Rovniak et al. 2010). One of the most applicable studies using cluster analysis for this evaluation is from the Brookings Institution. Here they used a set of twelve content areas with differing criteria to create clusters based on land use regulations of 50 metropolitan areas around the country (Pendall, Puentes, and Martin 2006).

Methodology Endogenous Variables Comprehensive plans are most commonly organized into broadly defined sections dealing with topics such as land use, property rights, and economic development. This commonality creates a good starting point when comparing multiple plans. In Idaho’s case, Chapter 65 of the Local Land Use Act within Title 67 of the Idaho code specifies 14 required sections in comprehensive plans (Table 1). These required elements provide both a simple and less subjective way to categorize objectives. However, for the sake of brevity and because some sections do not lend themselves well to the scoring process, only eight of the mandated 14 sections were used in this study (Table 1). For example, because regulating population dynamics can easily create legal problems, population sections tended to be more uniform across counties.

23


Table 1: Comprehensive plan sections mandated for inclusion by Idaho State Code Required Element

Evaluated in this Study

Property Rights

Yes

Population

No

School Facilities and Transportation

No

Economic Development

Yes

Land Use

Yes

Natural Resources

Yes

Hazardous Areas

No

Public Services, Facilities, and Utilities

Yes

Transportation

Yes

Recreation

No

Special Areas or Sites

No

Housing

Yes

Community Design

Yes

Implementation

No

Once the sections from the state requirements were chosen, a dichotomous metric within each of these sections was established so that each objective within the section would fall into one of two categories. A more detailed description of this process is provided in Appendix 1. The comprehensive plan sections and their dichotomous categories evaluated in this study are Private Property (“explicit protection” versus “placing restrictions”), Economic Development (“actively pursuing” versus “responding to”), Land Use (“protection of existing” versus “preparing for growth”), Natural Resources (“promote traditional economic use” versus “conservation”), Public Services (“county responsibility” versus “other’s responsibility”), Transportation (“nonroad” versus “road”), Housing (“county responsibility” versus “other’s responsibility”), and Community Design (“county responsibility” versus “other’s responsibility”).

24

Western Economics Forum, Fall 2012 Once the dichotomous categories for each section were established, each objective was evaluated accordingly. If the objective was soft in language using words like should, encourage, or suggest, the objective was given a score of 1 in the given category. If the language in the objective was firm such as ‘requiring’, ‘will’, or ‘must’, then the objective was given a score of 2, this methodology is supported by previous work (Grimes 2006). The final scoring for both categories in each of the eight sections was first calculated by taking the mean of all the scored objectives in a category, then multiplying that by the percentage of objectives in that category versus the opposite category in the same section, creating a weighted mean. Thus a given category could receive a score between zero, if it was not addressed at all, and 2 if it was addressed with firm language and the county gave no emphasis to the competing category. The weighted mean solved the problem of a county appearing to have similar emphasis in the two categories of a given section when one of the two categories had many more objectives devoted to it. There are notably several counties missing from the analysis. They were omitted either because they lack a comprehensive plan, their comprehensive plan was not available at the time of analysis, or the comprehensive plan was produced before the creation of state guidelines. Of the 44 counties in Idaho, nine were omitted from this analysis due to the criteria above. Clustering Once the scores for both categories in the eight sections were established, a k-means cluster analysis was performed to group similar counties based solely on the elements in their comprehensive plan. Clusters are a mathematical way of organizing data based on an algorithm that creates groups in which data most closely resemble other data in that group and establishes differences from other groups. Cluster analysis organizes observations into clusters so that the similarity of characteristics within each cluster is maximized, while maximizing differences between clusters (Xu and Wunsch 2009, Aldenderfer and Blashfield 1984). Cluster analysis uses multiple variables to determine the characteristics that differentiate those clusters and which observations belong in which cluster. In k-means cluster analysis, the similarity or dissimilarity of sets of data is determined by Euclidean distance from the cluster centroid (Xu and Wunsch 2009). Clusters were created based on the scored elements of the comprehensive plans (endogenous variables) as described above. A common “rule of thumb” (Aldenderfer and Blashfield 1984) for determining how many clusters to include in a k-means cluster analysis is:

𝑘𝑘 =

𝑛𝑛 2

Where k is the number of clusters in the data and n is the number of observations. Under this criterion our data contain four unique clusters. Exogenous Variables In addition to these endogenous characteristics of the comprehensive plans themselves, demographic, geographic, and economic data were also collected and analyzed in relation to the comprehensive plan based clusters. This combination of information provides potential

25

Western Economics Forum, Fall 2012 insights into the relationship between what the counties prioritize in their comprehensive plans and the characteristics of the respective county clusters. The exogenous variables represent given characteristics of the counties that are not directly related to the comprehensive plans. Thirteen exogenous variables were chosen to produce a description about the four clusters. These variables were chosen based on the availability and reliability of the data sources as well as their combined ability to produce a picture of the type of county they collectively describe. Exogenous variables used in this study are presented in Table 2. Table 2: Average values of descriptive variables for counties in their respective clusters. These variables were not used in the cluster analysis.

Population, 2000 Census Median Household Income, 2008 Age of Comp. Plan, Years Population Change, 20002009 Home Ownership Rate, 2000 Urban/Rural Continuum Code (Higher= More Rural)

Cluster 1 Mean

% of Total Mean

Cluster 2 Mean

% of Total Mean

Cluster 3 Mean

% of Total Mean

Cluster 4 Mean

% of Total Mean

Mean Across All Analyzed Counties

19,942

61.98%

60,046

186.63%

13,582

42.21%

29,753

92.48%

32,174

$42,516

96.17%

$47,572

107.61%

$41,019

92.79%

$45,200

102.25%

$44,207

6.4

112.28%

6.3

110.53%

2.2

38.60%

5.8

101.75%

5.7

6.20%

60.78%

19.50%

191.18%

5.90%

57.84%

7.50%

73.53%

10.20%

76.40%

102.25%

73.80%

98.77%

74.10%

99.17%

72.80%

97.43%

74.72%

6.14

112.45%

4.70

86.08%

5.60

102.56%

5.00

91.58%

5.46

4.29

98.85%

4.20

96.77%

4.40

101.38%

4.67

107.60%

4.34

Natural Amenity Index (Higher= More Natural Amenities) Creative Class Score % Voted for Obama, 2008

0.154

89.53%

0.189

109.88%

0.175

101.74%

0.182

105.81%

0.172

26.70%

86.69%

32.70%

106.17%

32.80%

106.49%

35.50%

115.26%

30.80%

Unemployment Rate, March 2010

11.64%

103.65%

10%

89.05%

11.30%

100.62%

12.20%

108.64%

11.23%

8.40%

49.12%

33%

192.98%

8.50%

49.71%

18%

105.26%

17.10%

17.20%

91.49%

19.10%

101.60%

22.40%

119.15%

19.30%

102.66%

18.80%

0.42

100.72%

0.41

98.09%

0.43

102.63%

0.41

99.04%

0.42

% Change in Private Nonfarm Employment, 2000-2008 % with Bachelor's Degree or Higher Gini Coefficient (Measure of Income Inequality)

26

Western Economics Forum, Fall 2012 Basic demographic data, including population and population change between 2000 and 2009, were obtained for each respective county. A variety of economic indicators were used including homeownership rate, median household income, the unemployment rate, and the percent change in private non-farm employment between 2000 and 2008. The 2000-2008employment change was included to help define the counties without the 2008 market collapse. The age of the plan was included to explore the possibility of newer plans containing objectives that represent changing approaches in the planning field. The education rate of a county, defined as those with a bachelor’s degree or higher, as well as the percentage of people in the county who voted for the Democratic presidential candidate in the 2008 election as a representation of the political climate, were also used. Several metrics were included that were based on outside research, these were the UrbanRural Continuum, the Amenity Index, the Creative Class Index, and the Gini coefficient. The Urban-Rural Continuum is a measure of the rural-ness or urban-ness of a county from the Economic Research Service (USDA ERS 2003). The Amenity Index is a measure of the level of natural amenities within a county on a scale of 1 to 7 (USDA ERS 1999). This in many ways represents the natural beauty of an area, as a higher natural amenity score is awarded for features such as mountains, coastline, and number of days of sunshine. The Creative Class Index is a measure of how much of the working population is involved in creative industries; it is based on work by Richard Florida, who asserts that the creative class is important to the growth of an area (Wojan, Lambert, and McGranahan 2007). The last variable is the Gini coefficient which gives a measure of income inequality (Ray 1998). The clusters created from the objectives in the comprehensive plans were compared against each of these exogenous variables to create a story about the counties’ characteristics and priorities.

Results and Discussion A graphical depiction of the results of the cluster analysis is presented in Figure 1 and tables of the exogenous and endogenous variables are summarized in Tables 2 and 3, respectively. The composition of the clusters and the resulting traits can tell an interesting story. From the exogenous variables we can get almost a personality of the clusters; the rural nature of a cluster, the general education level how much employment has grown. Then, comparing that against the trends in the cluster’s comprehensive plans we can see what types of priorities different groups of counties have in comparison to their traits. In many ways the objectives line up well with conventional thinking about the different types of counties, there are, however, exceptions.

27


Figure 1: Map of Idaho with comprehensive plan based cluster groups

28


Table 3 - Means of comprehensive plan scores used in cluster analysis Cluster 4 Mean

% of Total Mean

Mean Across All Counties

Traditional Economic Use

0.61

161.58%

0.288

75.79%

0.10

27.11%

0.22

57.11%

0.380

Conservation

0.63

80.43%

0.885

112.45%

0.73

92.50%

1.03

131.39%

0.787

Explicit Protection Placing Restrictions

1.12

200.36%

0.16

28.14%

0.00

0.00%

0.39

69.00%

0.56

0.05

21.50%

0.04

20.09%

0.00

0.00%

1.07

500.47%

0.214

Protection of Existing

0.74

140.42%

0.52

98.10%

0.19

36.43%

0.33

62.24%

0.527

Preparing for Growth

0.51

85.62%

0.66

110.87%

0.45

75.92%

0.81

136.12%

0.598

Actively Pursuing

0.85

122.35%

0.63

89.83%

0.20

28.65%

0.87

124.21%

0.698

Responding to

0.33

85.23%

0.33

85.23%

0.65

168.39%

0.29

75.91%

0.386

0.68

105.14%

0.81

125.70%

0.20

30.37%

0.66

102.80%

0.642

0.50

79.46%

0.38

60.35%

1.42

224.80%

0.70

110.27%

0.633

0.28

95.49%

0.30

104.51%

0.33

114.24%

0.26

91.32%

0.288

0.85

125.67%

0.41

60.24%

0.63

93.18%

0.76

112.46%

0.674

County Responsibility

0.81

100.00%

0.89

109.62%

0.65

80.02%

0.81

100.25%

0.811

Other's Responsibility

0.32

92.53%

0.25

70.40%

0.56

162.07%

0.40

114.94%

0.348

Non-road

0.17

40.92%

0.63

152.78%

0.65

156.17%

0.43

103.15%

0.413

Road

1.22

131.61%

0.67

72.06%

0.72

77.89%

0.85

91.37%

0.927

Transportation

Housing

Community Design

Public Services

Economic Development

Natural Resources

Cluster 3 Mean

% of Total Mean

Property Rights

Cluster 2 Mean

% of Total Mean

Land Use

Cluster 1 Mean

% of Total Mean

County Responsibility Other's Responsibility County Responsibility Other's Responsibility

29

Western Economics Forum, Fall 2012 Cluster 1 (The Traditionalists) The comprehensive plans from counties in Cluster 1 placed a large emphasis on protecting the traditional economic uses of natural resources. Cluster 1’s emphasis on protecting traditional economic uses of land was 62% above the full sample mean; the only cluster well above the sample mean on this measure. In contrast, they were also the least likely (20% below sample mean) to place emphasis on conservation of natural resources. Cluster 1 was also by far the most likely to explicitly protect property rights and among the least likely to place restrictions on property rights. This cluster was twice as likely to explicitly protect property rights, and had almost no objectives targeted at placing restrictions on property rights. Similarly in their land use sections, they were 40% above the sample mean in their emphasis on protecting existing land use rights while being 14% below the state mean on using land use objectives to prepare for growth. This was the only cluster that was above the state mean for protecting existing land uses. This cluster’s counties took the position of actively pursuing economic development, 22% above the sample mean, as opposed to preparing to respond to it, scoring 15% below the mean. On transportation networks, this cluster is clearly set on road emphasis. They have almost nothing to say on non-road transportation objectives, being 59% below the mean, while being 32% above the sample mean on road objective’s emphasis, not surprising in a rural area where low population densities make transit difficult. In the category of community design, cluster 1 counties place 26% more emphasis on objectives for third party responsibility and place 5% less on county responsibility. While the county responsibility difference is fairly small, the third party emphasis is the highest among all clusters. Despite not being part of the cluster analysis, the exogenous variable show marked differences across the clusters. Cluster 1 counties are primarily the rural counties in the state without the recreational emphasis of having large ski resorts. They have low populations, being 38% below the mean of all the counties in the analysis. The median household income is relatively close to the sample mean at 4% under. In addition to low population totals, Cluster 1 counties are also not growing quickly. The 2000 to 2009 population change was 6.2% which is 39% below the sample mean growth rate. These counties are 12% more rural (higher score on the urban/rural continuum) and as such they exhibit some of the other fairly typical traits of rural areas. The creative class index, for example, is lowest in these counties at 10% below the mean. Growth in employment in the nonrecession years of the last decade, 2000-2008, was 51% below the sample mean. Cluster 1 is the only cluster below the sample mean, 8% below the mean for this variable. These counties were 13% less likely to have voted for Obama in 2008, the only cluster less likely, than the sample mean. The “traditionalists” of Cluster 1 are the rural counties in the state, and share many of the characteristics of rural counties across the nation. Their objectives focus on improving roads, protecting existing land uses, protecting property rights, and actively pursuing economic development opportunities. They balance objectives regarding county and third party responsibility fairly well with the exception of community development; in that section they placed the majority of the emphasis on third parties.

30

Western Economics Forum, Fall 2012 Cluster 2 (Proactive Counties) Cluster 2’s comprehensive plans have very few objectives that address actions on property rights, whether explicit protection, or placing restrictions. They are below the sample mean on both property right elements. Cluster 2 is also below the mean, yet fairly balanced, on economic development elements in their comprehensive plans. However, cluster 2 comprehensive plans are significantly above the mean in terms of county responsibility in providing public services, and in their emphasis on non-road transportation. This cluster was also above the mean in emphasizing community responsibility in both community design and housing, and is the only cluster that is consistently above the mean on making these objectives the county’s responsibility and below the mean on making third parties responsible. In terms of the exogenous variables, this cluster of counties are the growing urban counties that include Ada County, which is by far the most populated county in the state and is home to Idaho’s largest city, Boise, and other highly populated counties such as Kootenai County (includes the city of Coeur d’Alene) and Twin Falls County (includes the city of Twin Falls). This cluster also includes some of the high recreation counties including Teton and Shoshone. Population of these counties is far and away the highest of any cluster at 87% above the mean. This is also represented in the urban/rural continuum with Cluster 2 being the most urban on the continuum. They also have the highest median household income at 8% higher than the sample mean. Given the differences in counties that make up these clusters, specifically in this cluster the high recreation counties with the urban counties, an additional regression analysis would be a nice additional step, but that is beyond the scope of this paper. Cluster 2 counties also showed the highest level of population growth, with population growth from 2000 to 2009 being 91% above the sample mean. The urbanite counties also had the highest growth in private nonfarm employment over the last decade, 93% above the state mean. This coincides with their current lowest unemployment, 11% below the sample mean, the only cluster with a below average unemployment rate. In keeping with what would be expected for more urban regions, this cluster had the highest creative class index in the state, 10% above the mean. Interestingly, this cluster had the lowest natural amenity index, although that is only 3% below the sample mean. These counties are the highest income counties, lead the state in population growth, and have had the best growth in employment over the last decade. All of this is fairly consistent with urban areas. While the urbanite counties had fairly balanced approaches to most of their sections, it is interesting to note that in the sections looking at county responsibility and third party responsibility for given objectives, these counties consistently pushed towards county responsibility. This emphasis on county responsibility suggests that these counties may be more willing to invest in themselves. Cluster 3 (Passive Counties) In many ways, cluster 3 is the inverse of cluster 2. Cluster 3 counties also did not address property rights, neither placing restrictions on them nor giving them explicit protection. This cluster was also below the mean in statements regarding either conservation or traditional economic use of natural resources. Land use is another area where these counties are much below the mean, 74% and 24% below state averages respectively on objectives that protect existing land use and land use objectives that actively prepare for economic growth. These counties are also below the mean on objectives for active pursuit of economic development. However, they score above average (68% above the sample the mean) in elements which

31

Western Economics Forum, Fall 2012 respond to economic development. In transportation Cluster 3 is very similar to Cluster 2 being 56% above the mean on non-road objectives and 22% below the mean on strictly road objectives. This cluster takes a fairly laissez-faire approach to the provision of public services. Cluster 3 is 70% below the mean on elements which make the provision of public services a county responsibility and are 125% above the mean on elements which enumerate third party responsibility for public services. A similar situation is seen in the housing section where cluster 3 is 20% below the mean for county responsibility and 62% above the mean for third party responsibility. However, Cluster 3 had the largest emphasis of any cluster on county responsibility for community design at 14% above the mean and was 7% below the mean on third party responsibility for community design. The counties in Cluster 3 tended to be the smallest counties in the sample, having the lowest average population of all the clusters. Cluster 3 is 58% below the state mean for population. It is also slightly below average in income, at 7% below the sample mean, and is a low population growth cluster at 42% below the mean. Similar to cluster 1, cluster 3 experienced much below average job growth from 2000-2008, at 50% below the sample mean. Cluster 3 had, on average, the most current comprehensive plans in the sample. The comprehensive plans across the entire sample (all clusters) range from one to ten years since updating, and the age of the plans is fairly scattered throughout the clusters. The exception is cluster 3 where all the comprehensive plans were only one, two, or three years old representing a cluster mean 61% below the mean across the entire sample. This indicates that the counties with the smallest population in this study also had the newest comprehensive plans. What distinguishes the “passive counties” of Cluster 3 is their relative lack of required elements in their comprehensive plans. When they do enumerate a requirement it is most often a requirement that a party other than the county be responsible for that respective element. Cluster 3 counties are low population, low population growth, primarily low income, and low employment growth. Additionally, their plans say very little about economic use of natural resources, property rights, or protection of existing land uses. These counties are also pushing for more multi-modal transportation options, being more passive in economic development, and placing responsibility for community development with the counties. This all suggests a desire to adapt to changes in land use and economic environments. Cluster 4 (The Conservationists) Counties in Cluster 4 place high emphasis on conservation of natural resources. They are 43% below the mean on objectives relating to traditional economic use of natural resources, while being 31% above on conservation of natural resources, the highest of any cluster. They were also low on explicit protection of property rights, 31% below the mean, although they did address this issue. Conversely, they were 400% above the mean on placing restrictions on property rights, by far the highest. Similarly to the property rights section these counties placed less emphasis on protection of existing land use in their objectives, 38% below the mean, and were the highest with using their objectives to prepare for growth, 36% above the mean. In the economic development section, Cluster 4 counties were heavily tilted toward actively pursuing economic development at 24% above the mean, over responding to it, where they were 24% below the mean.

32

Western Economics Forum, Fall 2012 For county responsibility and third party responsibility in public services, community design and housing, Cluster 4 is just about average in every category. These counties are also quite average on transportation, being closer to the mean on both road and non-road objectives than any other Cluster. In terms of exogenous variables, the counties in cluster 4 are the closest to the mean in terms of population size, income, home ownership, educational attainment, and income inequality. They were also the closest to the sample mean for population growth even though they are 26% below the state mean growth rate, indicating there was a great deal of variation in this variable. The differences for Cluster 4 arise in three areas. First they have the highest Amenity Index, 8% above the state mean, and while there is not a great amount of variability here, it is twice as far away from the mean as the next furthest cluster. Second, these counties voted for Mr. Obama, a proxy for political leaning, at a greater rate (15% above the sample mean) than any of the other clusters. Finally, these clusters have the highest unemployment rate (9% above the sample mean), due in part to the housing boom and bust in high amenity areas. “The Conservationists” are average in income, population, and population growth. But they are also attractive (high on the Amenity Index), voted more for Mr. Obama in the 2008 presidential election, and have a higher than average creative class. This gives us a profile of these counties that are average in many ways but have some reliance on tourism. Based on this there are certain features within the comprehensive plans that one would expect to see, and that play out well. Emphasis on conservation of natural resources, high emphasis on restrictions on property rights, and on preparing for growth in their land use are all expected in areas that have likely been experiencing high development pressure.

Conclusions This analysis shows that there are some distinctions in the way counties with differing characteristics place emphasis within their comprehensive plans. It establishes a relationship between counties with separate clustered priorities, as defined by their objectives, and those clusters differing exogenous characteristics. It does not prove any sort of causation, but these correlations could be used as a good starting point to begin looking for causality. For example, “does investment in community design contribute to the differences in per capita income within a county?”; or “does higher per capita income result in more demand for community design?”. While causality cannot be established between the similarities in the comprehensive plans and the demographic and economic characteristics of the counties, a crucial purpose of the comprehensive plan is for a community to identify where it wants to go. Whether it be a proactive vision of the future or simply identify the status quo as a priority, a community identifies its identity and goals though the comprehensive plan. As can be seen by our results, counties with similar interests (objectives in their comprehensive plans) have many other similarities; this is at least suggestive that comprehensive plans are fulfilling their objective of representing their specific communities. While these results might suggest this linkage, comprehensive plans and in a larger context local government, is only one of a myriad of factors that produce the characteristics of a county. Nonetheless, the suggestive results here could offer insights to counties looking for model comprehensive plans when their rewrite cycle comes around. Offering a list of counties that may seem very different, but have some underlying similarities and similarities in comprehensive plans, could allow a rewriting county to choose among other model compressive

33

Western Economics Forum, Fall 2012 plans. This would allow them to find example counties that may have similar strengths, but be in a more desirable position economically and/or demographically. Beyond using this information to create a hypothesis for future research, much of this information could be useful for individual counties. As counties grow and develop, the assumptions upon which their comprehensive plans also change. Examining plans of counties which are already facing situations that your respective county is anticipating gives insights into how to amend your comprehensive plan for your future community needs. It is all too often the case that communities necessarily develop comprehensive plans based on current conditions rather than developing plans for desired conditions. It is often the case that communities eschew changing planning efforts until the absolutely and obviously necessary. Unfortunately, when the need is obvious, the opportunity to implement effective change has already passed. It is hoped that, by looking at the comprehensive plans of regions with characteristics similar to the desired future conditions of their own community, communities can use this information to help establish a unique vision while benefiting from the experiences of other communities Likewise, if a county likes the outcomes of a different cluster in regard to a certain section, they could look at the ways these counties write their objectives and adjust theirs accordingly. This could also be scaled up to rewrite whole plans. If one type of plan consistently is associated with more positive county attributes this could even be used to start giving a metric to what actually makes a good plan, which has been a somewhat elusive goal in the field of planning.

References Ada, County. 2007. “Ada County Comprehensive Plan.” Benewah, County. 2003. “Benewah County Comprehensive Plan.” Berke, Philip, and Maria Manta. 2000. “Are We Planning for Sustainable Development? An Evaluation of 30 Comprehensive Plans.” Journal of the American Planning Association 66 (1). Bingham, County. 2005. “Bingham County Comprehensive Plan.” Bureau of Labor Statistics. 2011. “Tables and Maps Created by BLS.” Accessed May 10. http://www.bls.gov/lau/tables.htm. Camas, County. 2006. “Camas County Comprehensive Plan.” Caribou, County. 2006. “2006 Comprehensive Plan Caribou County Idaho.” Elmore, County. 2004. “2004 Comprehensive Growth and Development Plan.” Evans-Cowley, Jennifer, and Meghan Zimmerman Gough. 2009. “Evaluation New Urbanist Plans in Post-Katrina Mississippi.” Journal of Urban Design 14 (4) (November): 439– 461. Grimes, Matthew. “Virgina Transportation Research Council Research Report: An Evaluation of County Comprehensive Plans in Virginia.” http:/www.virginiadot.org/vtrc/main/online_reports/pdf/07-r6.pdf. Hopkins, Lewis D. 1977. “Methods for Generating Land Suitability Maps: A Comparative Evaluation.” Journal of the American Institute of Planners 43 (4): 386–400. doi:10.1080/01944367708977903. Jefferson, County. 2005. “Jefferson County Comprehensive Plan.”

34

Western Economics Forum, Fall 2012 Jerome, County. 2006. “Jerome County Comprehensive Plan.” Karl Gudmunds, and USDA Economic Research Service. 2011. “ERS/USDA Data Sets Natural Amenity Index.” Accessed May 10. http://www.ers.usda.gov/data/. Kootenai, County. 2010. “Kootenai County Comprehensive Plan.” Nez Perce, County. 1998. “Nez Perce County Comprehensive Plan.” Pendall, Rolf, Robert Puentes, and Jonathan Martin. 2006. “From Traditional to Reformed: A Review of the Land Use Regulations in the Nation’s 50 Largest Metropolitan Areas”. The Brookings Institiution. Power, County. 2009. “Power County Comprehensive Plan.” Rovniak, Liza S., James F. Sallis, Brian E. Saelens, Lawrence D. Frank, Simon J. Marshall, Gregory J. Norman, Terry L. Conway, Kelli L. Cain, and Melbourne F. Hovell. 2010. “Adults’ Physical Activity Patterns Across Life Domains: Cluster Analysis with Replication.” Health Psychology  : Official Journal of the Division of Health Psychology, American Psychological Association 29 (5) (September): 496–505. doi:10.1037/a0020428. Schamess, Lisa. 2006. “Everything Old Is New Again.” American Planning Associaton, December. Tim Parker, and USDA Economic Research Service. 2011. “ERS/USDA Briefing Room Measuring Rurality: Rural-Urban Continuum Codes.” Accessed May 10. http://www.ers.usda.gov/briefing/rurality/ruralurbcon/. Washington, County. 2000. “Washington County Comprehensive Plan.” Wojan, T. R., D. M. Lambert, and D. A. McGranahan. 2007. “Emoting with Their Feet: Bohemian Attraction to Creative Milieu.” Journal of Economic Geography 7 (6) (May): 711–736. doi:10.1093/jeg/lbm029.

35


Appendix 1 Detailed Methods for Scoring Comprehensive Plans The rating metric for each of the eight selected state required components broke down into the following criteria. Private Property objectives were either classified as explicit protection of private property, or placing restrictions on private property. Explicit protection would include any objective that had a main focus of protecting private property rights, whereas placing restrictions would include any objective that placed or suggested any type of limitation on private property rights. An example of an objective that explicitly protects private property is, “encourage the protection of the property rights of landowners to the extent possible” (Bingham 2005). This is in contrast to the view that there should be limitations placed on private property rights, “Property owners must recognize they are only temporary stewards of the land, and shall preserve and maintain their property for the benefit of future generations” (Elmore 2004, 13). Natural resource objectives were categorized as either promoting conservation or promoting traditional economic uses of natural resources. Conservation objectives included anything that preserved natural resources simply for the sake of the resource, with the primary regard not being its monetary value. Traditional economic use objectives included any objective that had a monetary focus but consisted primarily of timber, mining, and agricultural uses. An example of a conservation objective would be, “Protect and preserve the natural beauty and habitat of the Boise River and the black cottonwood forest and land abutting the river” (Ada 2007, 6–12). Whereas an example of an objective focusing on traditional economic uses of natural resources would be, “Work cooperatively with relevant agencies to, identify and protect productive resource farm, timber, and mining lands”(Kootenai 2010, 11–10). There were many instances of objectives addressing tourism in the Natural Resources section, which could fit into either category. In this case, these objectives were classified under traditional economic use as tourism objectives primarily focused on economic impacts. The land use section is often the largest of any section in comprehensive plans. Objectives in this section were broken down between protection of existing use, and preparing for growth. Protection of existing use includes any objective that specifically addresses an existing use and gives it some form of deference. Preparing for growth encapsulated the majority of objectives not dealing with existing uses, but dealing with issues such as directing growth, placement of new development, and desired outcomes. For example, an objective addressing the protection of existing use is, “Prevent the loss of range and agricultural lands” (Jerome 2006, 79). Protection of existing agricultural land is a common theme in Idaho. An objective preparing for growth, on the other hand, would be exemplified as, “Encourage rural residential cluster developments outside incorporated city's limits that will encourage self-sustaining communities”(Caribou 2006, 2). This example objective is targeted at dealing with development growth as opposed to protecting existing uses. While both approaches may have the effect of mitigating effects in the county, the focus and approach is distinguishable. The two strategies identified to separate the economic development objectives are actively pursuing economic development or responding to it. The former represents a proactive and directional stance toward economic development and the later represents a more reactionary or facilitatory approach. All counties hope for more economic development, so this difference in actively pursuing versus preparing an area for additional economic development is a way to distinguish more nuanced differences in strategies. An example of actively pursuing would be, “to attract new and retain current business in Camas County with a focus on light manufacturing and production and the tourism and recreational industry, creating a better wage base for

36

Western Economics Forum, Fall 2012 county residents”(Camas 2006, 12). This is clearly an active stance the county is taking. In contrast, an example of responding to economic development would be, “Provide the best method of transition between those areas in the county that are agricultural and those areas that may be suitable for other types of development”(Power 2009, 15). This represents an objective that more passively facilitates economic development. As opposed to actively pursuing the specific industry, this strategy responds to a potential new industry through more generalized assistance to revealed needs. These are just two examples in a wide variety of strategies that communities might employ, and of course, additional objectives that might be found elsewhere could contribute to economic development in practice. A trend that has been emerging in the transportation field is a push toward more multi-modal transportation. This includes not only biking and walking but also use of different forms of public transportation. However, in many counties this has not become an important issue. Thus the transportation section is separated into objectives that promote some sort of multi-modal policy, and those that deal only with traditional road transportation. An example of a non-road objective is, “Expand pedestrian, bicycle and transit facilities to provide transportation alternatives and promote an environment that is inviting for pedestrians, bicyclists and transit riders”(Jefferson 2005, 75). This is contrasted with an example of a road-centric objective, “Recognize the value of the roads systems to the agricultural community and work to reduce conflicts between transportation needs of agriculture and non-agricultural pursuits”(Washington 2000, 21). The final three sections, Public Services, Housing, and Community Development were all broken down based on who was responsible for carrying out the objective. The objectives could be primarily categorized as either the county being responsible for the actions in the objective, or a third party being responsible, often a company doing business in the area. For example, if a county wants to “encourage the designation of open space in new developments,”(Benewah 2003, 42) even though the county is encouraging it, the main responsibility will be with the developer creating a new development. As such, this would be an objective where the primary responsibility is with a third party. If the primary responsibility for the objective is on the county such as, “Nez Perce County should develop design standards to be included in county land use ordinances”(Nez Perce 1998, 2–2), it would be categorized as a county responsibility. Holding third parties to a set of standards is an important for all communities; this categorical dichotomy attempts more to address the participatory level of involvement of the county in these objectives.

37


A Case Study in Habitat Equivalency Analysis: The Pacific Connector Gas Pipeline David T. Taylor, Thomas Foulke and Archie Reeve1 Introduction When wildlife habitat is impacted through an adverse event, such as an oil spill, or a construction project, replacing that habitat’s ecological services becomes an important issue. Just how replacement is accomplished and to what extent is often a process fraught with complex issues, legal challenges and much acrimony. This process can drag on for years, sometimes decades. A method where all parties can find common ground and that reduces value judgments would reduce costs to both taxpayers and businesses and aid in the swifter recovery of impacted habitats and species. Habitat Equivalency Analysis (HEA) is a relatively new approach that helps answer the fundamental question of how much habitat replacement is enough. In this paper, we showcase a real world use of HEA on a natural gas pipeline project that crosses threatened and endangered species habitat. HEA was developed by the National Oceanographic and Atmospheric Administration (NOAA) and the Minerals Management Service (MMS) as a way to scale compensation for habitat damage (e.g. Penn and Tomasi 2002, Roach and Wade 2006). In effect, the method provides complete in-kind (i.e. of similar quality and quantity) replacement of lost ecological services between the time of impact (e.g., an oil spill) and the time services are restored or created to their full replacement value. HEA uses the concept of discounting to value the stream of ecological services over time. HEA makes the assumption that society values the flow of ecological services over time. Discounting theory assumes that people place a greater value on services today than those put off for future use. Every year in a discounting series thus has a specific value in terms of services provided until full restoration is complete. Typically, a standard discount rate of three percent is assumed (Peacock, 1995). So each year it takes to replace or restore services, a habitat capable of producing three percent of the remaining lost services must also be provided. This approach of equating ecosystem service flows has significant advantages over trying to place dollar values alone on complex ecosystem services. This is one reason that HEA has helped bring parties with disparate views to the table. HEA tries to move beyond value judgments. Until recently, habitat compensation simply involved replacing impacted habit acres. HEA recognizes that in the time between impact and full restoration or compensation, ecological services need to be provided. HEA provides a framework for both compensation and restoration and can address the issue of “no net loss” of ecological function. The goal of HEA is to answer the fundamental question of how much is enough, by providing a framework for analysis.

1

Dr. Taylor and Mr. Foulke are a Professor and Senior Research Scientist, respectively, in the University of Wyoming, Department of Agricultural & Applied Economics. Dr. Reeve is a Principal for Edge Environmental Inc.

38

Western Economics Forum, Fall 2012 HEA has been used in at least one case where it has been accepted as a basis for settlement in federal court (United States of America vs. Melvin A. Fisher et al. 1997). Other application examples include such diverse areas as freshwater streams, sea grass beds and coral reefs (Chapman et al. 1998, Fonseca eat al. 2000, Milton and Dodge 2001). And because it is a generic method it can be adapted to a variety of situations, including those involving both habitats and individual species. Three pieces of information are required to conduct HEA: 1) the type of ecological services that have been damaged, 2) the extent of damage, and 3) the rate at which recovery will occur. Furthermore, determining which service is most appropriate to replace and the degree to which the study area provided this service prior to impact are probably the most important and potentially the most controversial steps in the HEA process. Figure 1. Generic representation of the HEA process.

Habitats provide multiple services and opinions may vary widely concerning which service should be the focus of the restoration efforts. HEA is not capable of, nor is it designed to resolve these issues. These issues must be negotiated by the interested parties. In addition, estimating the degree of services supplied by a specific parcel of habitat prior to damages and the extent to which it has been damaged can be difficult. Again, this is an issue for negotiation between the interested parties and is not a function of HEA.

39


Choosing a metric, some measure or indicator of ecological services provided, is another important step. Ecological services come in many different forms. Different species and habits both provide and require different ecological services, so it is difficult to give a broad definition. For example, Old growth timber in the Pacific Northwest provides areas for species such as the spotted owl with nesting locations. Spotted owls will only nest in Late Successional or Old Growth (LSOG) timber. So nesting is an example of an ecological service provided by LSOG timber. A metric is necessary in order to monitor the degree to which restoration efforts are meeting expectations. The metric should represent the qualities and quantities of the service provided by both the impacted and restored habitat, should be easy to measure and should have transparent characteristics. Metrics which represent multiple services have obvious advantages in that a more comprehensive assessment can be obtained. Finally, it is essential that the amount of service to be restored is small compared to the total available, so that no change occurs in the underlying value per unit of service. In order to apply HEA, replacement of a portion of the resource should not be so large as to influence the overall value of the resource; otherwise the appropriate amount of habitat would change. The structure of HEA is relatively simple (Figure 1). Calculations of how much habitat to restore or replace are based on estimates of the total loss in services from the damaged or lost habitat. Total loss is estimated from the degree and time of initial damage and the loss in services that occur during the time between the initial damage and when the restored or replaced habitat becomes fully functional. The discounting process adjusts how these services are supplied over time until full restoration has occurred. The details of this process will be discussed in the example presented below.

Project Overview The Pacific Connector Gas Pipeline (PCGP) project2 is part of the larger Jordan Cove Energy Project LP3. Located in Coos Bay, Oregon, the Jordan Cove project proposes to build a liquefied natural gas (LNG) re-gasification terminal to import natural gas into the western United States. The PCGP will transport imported natural gas to Malin, Oregon, some 234 miles away over the Southern Oregon Cascades; a route which intersects ecologically sensitive areas. The route contains several threatened and endangered species, including the northern spotted owl (Strix occidentalis caurina), marbled murrelet (Brachyramphus marmoratus), and coho salmon (Oncorhynchus kisutch). A natural gas hub, the confluence of several pipelines, is located in Malin and allows imported natural gas to enter the Western US natural gas grid. Permitting work on the project began in 2007.

HEA Application Methodology HEA is a service-to-service or resource-to-resource approach to natural resource valuation that can account for changes in the baseline while estimating interim losses. The fundamental concept is that compensation for lost ecological services can be provided by restoration/compensation projects that provide comparable services (i.e., compensatory mitigation). Compensatory services can include substitute service, out-of-kind services, and both on and off-site services. For the PCGP, the question is what services would the affected 2 3

(http://www.pacificconnectorgp.com/index.php) (http://www.jordancoveenergy.com/index.htm)

40

Western Economics Forum, Fall 2012 habitats have provided in the absence of the PCGP Project? With HEA, the replacement services are quantified in physical, in our case, acre-years.4 The selected project is then scaled so that the quantity of replacement services equals the quantity of lost services in present value terms. In our analysis, the quantity of acre-years is based on the age of the forest land being disturbed or provided as compensation. We assume older forest land provides more services for old-growth/mature forest-dependent species than younger forests, with the relationship assumed to be linear (non-linear comparisons are possible, but add a degree of complexity). HEA involves three basic steps: 1) Assess the present value of lost services relative to the base line. In this analysis the loss (“debit”) is measured in acre-years. 2) Select the appropriate compensatory restoration/mitigation measures. The “relative productivity” of a proposed restoration measure compared to what was injured is evaluated in the number of acre-years restored for every acre included in the measure. 3) Identify the area for the restoration/mitigation measure (scaling). This is the measure that equates the total discounted quantity of lost services to the total discounted quantity of replacement services to compensate for the lost habitat. The basic framework for the PCGP project analysis was originally developed as the Oregon Pipeline HEA spreadsheet model by Dr. Kristin Skrabis, Resource Economist, Office of Policy Analysis, U.S. Department of Interior with input from Doug Young, U.S. Fish and Wildlife Service. The parameters for the model were modified by Taylor and Foulke based on estimates from Edge Environmental, Inc. regarding the physical quantities of forest habitat injured by the PCGP project. The models went through multiple iterations following a review of the various versions and discussions with the agencies involved. Two models were eventually selected by Edge Environmental to be used in the Compensatory Mitigation Plan (CMP) prepared for the PCGP project. In the HEA modeling application developed by Taylor and Foulke for the CMP, there are two areas, identified as Area 1 and Area 2, which will be affected differently by the PCGP project (Figure 2). Area 1 is a corridor extending to 15 feet of each side of the proposed pipeline centerline – the 30-foot maintenance corridor - which will remain in an herbaceous/shrub state for the life of the project, estimated at 50 years. In forested habitats, conifer trees will be replanted within the construction right-of-way and other cleared areas outside of the 30-foot maintenance corridor and allowed to return to its pre-construction state. Those areas are identified as Area 2.

4

An acre-year refers to all natural resource services provided by one acre for one year.

41

Western Economics Forum, Fall 2012 Figure 2. PCGP typical right-of-way cross section.

30ft Area 1 Area 2

Late Successional-Old Growth Forest HEA Model We combine late successional and old growth (LSOG) forests into a single category to reflect landscape conditions in the region. Since the age mid-point of late successional forest is 127.5 years old and the age mid-point for old growth is estimated by biologists to be 250 years old (related to 325-year old stands), the mid-point of the combined age classes weighted by disturbed acres is 203.4 years (Table 1). This is an important parameter that is used to compute a linear slope for the production function of replanted forest. For effects to LSOG forest, ecological services associated with those seral stages are assumed to begin when replanted trees within Area 2 are 80 years old. Because Area 1 is not replanted, losses of ecological services are assumed to last in perpetuity. The current LSOG HEA model that has been reviewed by an economist at the Department of Interior and assumes that in-kind replacement compensatory habitat will be obtained three years after construction on the PCGP Project begins. That is, all compensatory habitat obtained is assumed to provide the same ecological services as the habitat lost. Consequently, discounting (at a rate of 3 percent per year) applies only to the 3-year interim between clearing and acquisition so that the amount of compensatory habitat obtained will be about 9 percent more than the amount of habitat removed. In addition to the area of habitat within Area 1 that will be removed “in perpetuity”, conifers that are planted following construction within former late

42

Western Economics Forum, Fall 2012 successional-old growth habitat in Area 2 will not start providing equivalent ecological services for 80 years and are assumed to provide full ecological services at the end of the next 123.4 years when the habitat is 203.4 years old. Table 1. Forest growth stage categories.

Mid-Seral Late Successional Old Growth Late Successional plus Old Growth*

From To 40 80 80 175 175 325 80 325

Mid-point* 60 127.5 250 203.4

years years years years

. *Weighted mid-point based on number of acres in late successional reserves in each age category. The HEA model only computes the amount of late successional-old growth habitat required for compensation based on the standard three percent discount rate. Three percent is the discount rate used by the USFS and the Department of Interior based on analysis done by Peacock (1995). All compensatory habitat obtained is assumed to be equivalent to the habitat affected (in this case, in-kind replacement of LSOG). An agreement between the PCGP developers and the agencies involved assumes that the compensatory habitat obtained will be held in perpetuity. This feature has been incorporated into the model to evaluate the amounts of compensatory habitat required to offset impacts to late successional-old growth habitats described for several listed species and to lands allocated under the Northwest Forest Plan, or NWFP (Forest Service and BLM, 1994), applicable to the PCGP project. Mid-Seral Forest HEA Model The authors also developed a second HEA model to evaluate effects to mid-seral (M-S) forests and to compute the amounts of compensatory habitat necessary to offset lost ecological services to forests between 40 and 80 years old. The assumptions applied to the M-S HEA model are similar to those described for the LSOG model. Loss of ecological services in Area 1 will continue in perpetuity, but ecological services associated with mid-seral forests begin when replanted trees within Area 2 are 40 years old. The mid-point of M-S forests is 60 years, which is used to compute a linear slope for the production function of replanted forest. This is accomplished by calculating the amount of annual recovery. For example, if it takes 20 years to reach full recovery (0 to 100 percent), then each year represents an additional 5 percent of recovery or a 5 percent slope. M-S forests provide nesting, roosting and foraging (NRF) habitat for northern spotted, and some nesting habitat and recruitment habitat for marbled murrelets (reference?). In addition, mid-seral forests within riparian zones provide shade that affects stream water temperatures, and large woody debris, both of which are important to coho salmon (reference?). Loss of M-S forest by construction of the PCPG in Area 2 will last for 40 years. At 40yearsold, restored M-S forest begins to provide lost services and provides 100 percent of ecological services when it reaches 60 years old.

43

Western Economics Forum, Fall 2012 Similar to the LSOG model, the M-S HEA model computes the amount of mid-seral habitat required for compensation based on the standard three percent discount rate. All compensatory habitat obtained is assumed to be equivalent to the habitat affected (in-kind replacement). Because conifers will be planted within Area 2 following construction, they will begin providing services as mid-seral forest after 40 years and those services reduce the amount of compensatory habitat computed by the HEA model. The assumption that compensatory habitat obtained will be held in perpetuity and that once obtained, it will remain as mid-seral habitat in perpetuity is contained within the model. Of course, that assumption is not realistic since, with the passage of time, mid-seral forest would eventually become late successional-old growth forest and contribute to offsetting impact to late seral stage forest in addition to the compensatory habitat obtained for that purpose. The assumption of being held as mid-seral forest in perpetuity is retained for simplicity, and to facilitate understanding during implementation of the HEA model. The ecological services that would be provided by mid-seral forests that eventually become late successional-old growth forests within 40 years of acquisition are not accounted for in either of the HEA models.

Example HEA Application Table 2 lists the parameters for both the LSOG and MS models. Recall that both models are based on in-kind replacement of lost ecological services, with replacement beginning three years after disturbance. Area 1 of the right-of-way is lost in perpetuity, and Area 2 is replanted and allowed to regenerate. Table 3 illustrates the results for the combined LSOG and the M-S models by habitat type including NWFP land use allocations and listed species impacted. The first two columns are impacted acres for Areas 1 and 2. Column three lists total impacted acres. Column 4 lists the model results of mitigation acres required. The results from the LSOG model are presented at the top of Table 3. Zero acres are shown for Marbled Murrelet, Southern Oregon/Northern California Coast Coho Salmon (SO/NCC), and Oregon Coast Coho Salmon (OC) because all their LSOG habitat overlaps with the Late Successional Reserves, Riparian Reserves, and Northern Spotted Owl acres of habitat. Thus their acres of affected habitat have already been accounted for in the other resource categories.

44

Western Economics Forum, Fall 2012 Table 2. Model Parameters.

Date of Analysis Start Date of Injury State Date of Recovery Years Until Full Recovery Starting Loss of Services Ending Loss of Services Is Loss in Perpetuity Annual Recovery Rate Start Date of Acquisition Discount Rate Unit of Injury

LSOG Area 1 Permanent

LSOG Area 2 Temporary

M-S Area 1 Permanent

2008 2010 2010 0 100% 100% Yes 0.00% 2013 3.00% Acre

2008 2010 2092 123.4 100% 0% No 0.81% 2013 3.00% Acre

2008 2010 2010 0 100% 100% Yes 0.00% 2013 3.00% Acre

M-S Area 2 Temporary 2008 2010 2052 20 100% 0% No 5.00% 2013 3.00% Acre

The results indicate that 1.09 acres of in-kind replacement are required for each acre of LSOG habitat that is permanently disturbed. This represents the three-year lag between disturbance and in-kind replacement. The results also indicate that 1.07 acres of in-kind replacement are required for each acre of LSOG habitat that is temporarily disturbed. The latter in-kind replacement acreage is smaller since the loss is not in perpetuity. HEA results in an overall average (between losses in perpetuity and temporary losses) of 1.074 acres as compensatory mitigation that would be appropriate for every acre of LSOG habitat affected by the project. The results from the M-S model are presented in the lower portion of Table 3. Unlike the LSOG results, acres are shown for all categories of habitat. M-S acres for Marbled Murrelet, and the salmon species are included because not all of these types of affected habitat overlap with Successional Reserves, Riparian Reserves, and Northern Spotted Owl acres of affected habitat. The results indicate that 1.09 acres of in-kind mitigation are required for each acre of M-S habitat that is permanently disturbed. Again, this represents the three-year lag between disturbance and in-kind replacement for permanently disturbed lands. The results also indicate that 0.86 acres of in-kind replacement are required for each acre of M-S habitat that is temporarily disturbed. The latter in-kind replacement acreage is smaller since the loss is not in perpetuity. HEA results in an overall average (between losses in perpetuity and temporary losses) of 0.947 acres as compensatory mitigation that would be appropriate for every acre of M-S habitat affected by the project.

45


Summary and Status Habitat Equivalency Analysis is still a relatively new tool in the analysts’ tool box, but shows promise in being able to help solve some of the more intractable issues. In this project, HEA was used to provide a framework for discussion so that all parties could come to the table with their issues in a ‘common currency’. This allowed for some of the more qualitative elements of ecological services to be resolved. HEA still requires negotiation. Choosing indicator species and metrics for recovery requires all parties do due diligence before they come to the table. But it does help remove some of the value judgments and inject more science-based arguments into the decision-making process. The original model worked well for the initial calculations, but changes by agencies dictated additional modeling to refine the functional form of the production function. This highlights the fact that the “devil is in the detail”. Agency changes (part of the negotiation process since more cards are on the table) became a reoccurring theme. The models, though static, are easily changed in their spreadsheet form and are more easily understood by agency personnel. In the end, project developers, and state and federal government agencies were able to come together and approve the project. The fact that the PCGP goes through some of the most contentious habitat zones in the country (northern spotted owl and old-growth timber) and is still the only LNG project so far approved on the US West Coast, proves that HEA is a viable contribution.

46


Table 3. Areas of LSOG Forested Habitat Impacted (Acres) Late-Successional - Old Growth Habitat LSOG LSOG Area 1 Area 2 Impacted Impacted (Acres) (Acres)

LSOG Total Impacted (Acres)

LSOG Compensation Mitigation (Acres)

Late Successional Reserves Riparian Reserves Northern Spotted Owl Marbled Murrelet SO/NCC Coho Salmon OC Coho Salmon

41.83 6.00 117.96 0.00 0.00 0.00

104.90 17.85 351.90 0.00 0.00 0.00

146.73 23.85 469.86 0.00 0.00 0.00

157.66 25.60 504.48 0.00 0.00 0.00

Total LSOG In-kind replacement ratio

165.79 1.09

474.65 1.07

640.44

687.74 1.074

Mid-Seral Habitat M-S Area 1 Impacted (Acres)

M-S Area 2 Impacted (Acres)

M-S Total Impacted (Acres)

M-S Compensation Mitigation (Acres)

Late Successional Reserves Riparian Reserves Northern Spotted Owl Marbled Murrelet SO/NCC Coho Salmon OC Coho Salmon

7.81 4.72 164.29 30.31 3.15 1.16

18.01 17.30 185.71 96.22 15.19 10.82

25.82 22.02 350.00 126.53 18.34 11.98

23.98 19.99 338.84 115.67 16.47 10.55

Total Mid-Seral In-kind replacement ratio

211.44 1.09

343.25 0.86

554.69

525.50 0.947

47


References Chapman, D., N. Iadanza, and T. Penn. 1998. Calculating resource compensation: An application of the services-to-services approach to the Blackbird Mine Hazardous Waste Site. NOAA Damage Assessment and Restoration Program Technical Report 97-1. Available online at http://www.darrp.noaa.gov/pacific/black/pdf/blackfnl.pdf. Fonseca, M.S., B.E. Julius, and W.J. Kenworthy. 2000. Integrating biology and economics in sea grass restoration: How much is enough and why? Ecological Engineering 15:227237. Milton, J. W. and R. E. Dodge. 2001. Applying Habitat Equivalency Analysis for Coral Reef Damage Assessment and Restoration. Bulletin of Marine Science 69:975-988. Peacock, Bruce. 1995. The Appropriate Discount Rate for Social Policy Analysis: Discussion and Estimation. U.S. Department of Interior, Office of Policy Analysis. 1849 C Street NW (Mail Stop 4426) Washington D.C. Issue Paper 11/22/95. Penn, T., and T. Tomasi. 2002. Calculating resource restoration for an oil discharge in Lake Barre, Louisiana, USA. Environmental Management 29: 691-702. Ray, Gary L.. 2008. Habitat Equivalency Analysis: A Potential Tool for Estimating Environmental Benefits. ERDC TN-EMRRP-EI-02, January 2008. Roach, B., and W.W. Wade. 2006. Policy evaluation of natural resource injuries using habitat equivalency analysis. Ecological Economics 58:421-433. United States of America, Department of Agriculture, Forest Service. Northwest Forest Plan Record of Decision. http://www.reo.gov/general/aboutnwfp.htm 13 April, 1994. United State of America vs. Melvin A. Fisher et al. 1997. 92-10027-CIV-DAVIS.

48

The Western Economics Forum is an electronic, peerreviewed publication of the Western Agricultural Economics Association.

Past issues of the Western Economics Forum and submission guidelines can be found at: http://agecon.uwyo.edu/WAEA/waeawef.htm

Information about the Western Agricultural Economics Association can be found at: http://agecon.uwyo.edu/WAEA/default.htm WAEA membership is open to all those interested in agricultural and resource management issues in Western North America.