utilization & engineering

How Clustering Dynamics Influence Lumber Utilization Patterns in the Amish-Based Industry in Ohio

Matthew S. Bumgardner, Gary W. Graham, P. Charles Goebel, and Robert L. Romig

Preliminary studies have suggested that the Amish-based furniture and related products manu- ticularly for the middle and higher (i.e., ap- facturing cluster located in and around Holmes County, Ohio, uses sizeable quantities of pearance) grades (Luppold and Bumgardner lumber. The number of firms within the cluster has grown even as the broader domestic furniture 2008, Buehlmann et al. 2009). manufacturing sector has contracted. The present study was undertaken in 2008 (spring/summer) This downturn in US furniture manu- to develop lumber use estimates specific to Amish manufacturing and provide more detail facturing has been precipitated largely by in- regarding the impacts of clustering on lumber consumption patterns. Results, based on 196 firms creasing imports from low-cost sources such responding to a survey, suggested that lumber use ratios (bd ft per employee) differed among as China, Vietnam, and other locations in firms of different sizes and type of product manufactured, but in aggregate was similar to the Southeast Asia (Figure 1). A recent report broader US furniture industry. Red oak was the most commonly used species. Local suppliers of from the High Point (North Carolina) In- hardwood lumber and components were used extensively by most firms. The study confirmed that ternational Home Furnishings Market indi- ABSTRACT the Holmes County furniture cluster was important to regional hardwood demand, facilitated by cated that 79% of the new offerings shown well-developed supply chains that enable high specialization and enhance aggregate productivity at this major biannual furniture trade show among the numerous small manufacturers. were imported (Appalachian Hardwood Manufacturers, Inc., 2006). Although not a Keywords: furniture, Amish, hardwood lumber, clustering, manufacturing measure of actual imported volume, such data are important because design and prod- uct trends originate at this and other large steep decline in the manufacture of consumption by the furniture industry has furniture markets (e.g., Las Vegas, Nevada). household furniture in the declined from 34% of total domestic ap- Based on actual volumes of domestic prod- A United States has had a substantial pearance-grade production (excluding ma- uct shipments and imports, it is estimated impact on domestic employment and mar- terial used to produce pallets and rail ties) in that nearly 60% of the nonupholstered kets for hardwood lumber. From 1999 to 1999 to just 15% in 2008 (Hardwood Mar- wood household furniture sold in the 2008, production employment in the US ket Report 2009). In the absence of a viable United States is imported (Cochran 2008). nonupholstered wood household furniture domestic furniture industry, US hardwood Given these overall trends, there have been industry declined by over 62%, or nearly lumber demand becomes increasingly reli- calls in recent years for a “paradigm shift” in 70,000 employees (Bureau of Labor Statis- ant on housing and remodeling markets the US wood household furniture industry tics n.d.). Nationally, hardwood lumber (e.g., cabinets, flooring, and millwork), par- to regain manufacturing competitiveness

Received September 23, 2009; accepted February 16, 2010. Matthew Bumgardner ([email protected]) is forest products technologist, Northern Research Station, US Forest Service, 359 Main Road, Delaware, OH 43015. Gary W. Graham ([email protected]) is extension specialist, Ohio State University Extension, Ohio Agricultural Research and Development Center, Wooster, OH. P. Charles Goebel ([email protected]) is associate professor, School of Environment and Natural Resources, The Ohio State University, Wooster, OH. Robert L. Romig ([email protected]) is emeritus, School of Environmental and Natural Resources, Ohio State University, Wooster, OH. This work was supported by the US Forest Service, Northern Research Station, Ohio State University Extension, the Ohio Agricultural Research and Development Center (OARDC), and The Ohio State University. Use of trade names in this article does not constitute endorsement of any product or service.

74 Journal of Forestry • March 2011 (Schuler and Buehlmann 2003). Key ele- ments of the paradigm shift include manu- facture of customized products and strategic supply chain alliances, which can be facili- tated by economic clustering (Schuler and Buehlmann 2003). Similarly, Dugan (2009) offers several “new rules” for the US furni- ture industry, including a focus on agility, niche marketing, lean production practices, and supply chain development. One sector of the domestic furniture in- dustry that has performed well during this period of globalization is the Amish-based manufacturing cluster located in Holmes and surrounding counties in northeastern Ohio [1]. The cluster has grown as the broader domestic furniture manufacturing industry has contracted. For example, over one-quarter of the firms operating in the cluster in 2005 had formed since 2000, even as wood furniture imports rose rapidly rela- Figure 1. Major sources (top five for 2008) of nonupholstered wood household furniture tive to domestic production (Bumgardner et imported by the United States, 2000–2008. (Source: International Trade Administration al. 2007). Firms within this cluster appear to n.d..) be using many practices associated with a new paradigm for the industry. For example, tors and added others, including the impor- whole. The cooperation came in a variety of consumers of Amish-made products often tance of stakeholder cooperation, entrepre- forms, ranging from joint marketing and are given options related to wood species, neurial thinking among the clustered firms product development to sharing machinery. finish, and even hardware for their furniture and associated organizations, leadership, The authors concluded that the success of pieces. Specialized supply chains have devel- and, often times, adequate funding sources the local economy was dependent on the lo- oped to facilitate this customization. (Aguilar et al. 2009). Cluster theory has cal industry as a whole, not individual firms In this article, we review some of the been proposed as a framework for promot- or products (Mottiar and Ingle 2007). Bres- competitive advantages associated with clus- ing economic development within commu- nahan et al. (2001) add to this notion with ters and relate these advantages to the dy- nities adjacent to or embedded by US na- their findings from multiple international namics found in the Holmes County furni- tional forests (Rojas 2007) and development case studies that, once established, clusters ture cluster. We then assess how clustering of bioenergy opportunities (Bratkovich et al. enabled opportunities to take advantage of affects lumber use patterns among the 2009). In Europe, examples of successful regional economies of scale rather than at mostly small and specialized Amish furni- furniture-related manufacturing clusters are the level of individual firms. ture manufacturers. found in northern Italy and Denmark Porter (1998) describes the coopera- (Schuler and Buehlmann 2003). tion and competition in clusters in terms of Competitive Advantages of Mottiar and Ingle (2007), studying a interfirm vertical integration. Although the Economic Clustering wood furniture manufacturing district in competition lies with rivals competing for Clusters, or geographic concentrations Ireland, developed the concept of interpre- customers (such competition is critical to of interconnected companies in a given field, neurship, which merges elements of entre- cluster success), the cooperation is vertical. can promote competitive advantage to man- preneurship (focused on the individual), Clusters offer a specialized supplier base that ufacturers through increased productivity, and intrapreneurship (focused on the firm). can lower transaction and inventory costs. rapid innovation, and new business forma- Interpreneurship was used by these authors Furthermore, the proximity of suppliers and tion (Porter 1998). Clusters are not uncom- to explain the success of the cluster in terms manufacturers enhances communications mon in forest-based industries in the United of strong interfirm relations among the nu- among firms and fosters closer, more infor- States. In the Pacific Northwest, analysis of merous small firms embedded in the local mal relationships. For example, success in three different clusters revealed several fac- community, as well as a social milieu, which boat building clusters in Australia was found tors important to success, including proxim- was defined as “… a close link between soci- to be associated with manufacturers that in- ity to regional markets, availability of skilled ety and firms; the relationships between the volved their suppliers in helping solve prob- labor, a plentiful raw material supply, and actors in the economy are not purely eco- lems and discussing design and production formation of new complementary businesses nomic” (Mottiar and Ingle 2007, p. 669). requirements (Jones 1996). In the Holmes through spinoff ventures (Braden et al. This setting resulted in cooperation and in- County furniture cluster, an example of 1998). Other case studies of wood products formation exchange among firms, creating a such collaboration between manufacturers clusters throughout the United States and social network that allowed individual firms and suppliers is associated with Ohio Certi- Europe confirmed many of these success fac- to benefit from the growth of the cluster as a fied Stains, a program where finish suppliers

Journal of Forestry • March 2011 75 work with manufacturers to develop and ad- firms, possibly even including some large Amish furniture manufacturing was exerting here to a set of color standards that enables firms, but absent a dominate one.” In such substantial influence on regional hardwood consistency among products made by mul- districts, firms have been found to be more lumber demand, but that further research tiple manufacturers and enables placement specialized than nondistrict firms, attribut- was warranted; the initial estimates were of semicustomized orders by consumers in able to being part of a “system of network based on application of an input productiv- retail showrooms (Terreri 2008). production” (Brookfield 2008). This spe- ity ratio derived from secondary data for the cialization arises in two ways: (1) firms out- broader US furniture industry (17,433 bd The Holmes County Furniture source inputs to a greater degree and (2) they ft/employee) and not Amish manufacturing Cluster have fewer product lines. Specialization also specifically. is evident in the Holmes County furniture A second objective was to develop a bet- Background cluster. As one Amish manufacturer there has ter understanding of lumber use patterns Economic clusters have been defined as, stated, “What the cluster does is it spreads out within the cluster, including analysis of in- “… critical masses—in one particular the investment risk. Many of the shops, there- put productivity, species use, and distribu- place—of unusual competitive success in fore, specialize in a relatively narrow field of tion channels for hardwood lumber. Porter particular fields” (Porter 1998, p. 78). This production” (Terreri 2008, p. 21). (1998) claims that productivity, driven in definition seems to be an accurate character- In Holmes County, firm specialization part by the presence of deep and specialized ization of the Holmes County furniture also is consistent with Mottiar and Ingle’s supply chains, is a major competitive advan- cluster. For example, a preliminary assess- (2007) view of embeddedness within an in- tage arising from clusters. The interfirm dy- ment found that this cluster, comprising, dustrial district, where important motiva- namics evident in the Holmes County fur- roughly, a two-county area [2], consumed tions for firms are to remain viable and con- niture cluster might be enhancing the 11% of the volume of hardwood lumber tinue living in the area. For example, very cluster’s overall productivity even though produced in the state of Ohio, or 19% of the small Amish firms may reach a point where most firms are very small, and smaller firms lumber used in appearance-based applica- they do not wish to grow beyond a customer are known to have lower productivity levels tions (i.e., excluding pallets and rail ties). base that can be served by the family mem- than larger firms (Tambunan 2005), specif- The corresponding volume was approxi- bers already employed, because furniture ically in terms of input productivity mately 43 mmbf, aggregated across more making has become a way to pursue an at- (Punches et al. 1995). Furthermore, a reli- than 400 mostly small shops. The mean home occupation as farming becomes in- ance on outsourcing of specialized manufac- number of employees for firms in the cluster creasingly unviable (Kreps et al. 1994, Low- turing inputs has been associated with mul- was 7.2, the median was 4.0, and the mode ery and Noble 2000). These firms might ticentered clusters (Brookfield 2008). Thus, was 2 (Bumgardner et al. 2007). Conversely, choose to work with a larger firm to manu- it might be expected that although most of the typical size of a furniture firm in the facture specific products for the larger firm’s the furniture firms in Holmes County broader US industry is approximately 27 line. In this way, they rely on the larger firm would be consuming limited quantities of employees, derived by dividing the number for product development, marketing, and hardwood lumber directly, component of paid employees by the number of estab- distribution as the larger firm expands its manufacturers in the cluster would be rela- lishments (US Census Bureau 2008a). sales (Terreri 2008). Similarly, Bresnahan et tively larger lumber consumers and impor- Associations have formed to promote al. (2001) found that the growth of at least tant to supplying the furniture firms with the interests of firms in Holmes County, as some of the startup firms within clusters was value-added products that require less pro- have trade shows that serve as opportunities an indication of cluster success, and that the cessing before assembly. However, to date, for individual manufacturers to present new larger firms eventually helped form the ver- primary information on wood use by Amish products and meet with existing and poten- tical linkages that enhanced continued clus- furniture manufacturers has been unavail- tial retail customers. For example, in its 2nd ter growth; however, many other firms able to evaluate this premise. year, the Ohio Hardwood Furniture Mar- within the clusters investigated preferred to Such information also provides valu- ket, a furniture trade show held in Holmes remain small. able cues as to the functioning of the cluster County and coordinated by The Hardwood Mottiar and Ingle’s (2007) notion of a in the broader furniture market. Anecdotal Furniture Builder’s Guild (a committee professional milieu in clusters also seems ev- evidence suggests that red oak (Quercus spp., within the Holmes County Chamber of Com- ident from a quote from an Amish manufac- mostly rubra L.) is the primary species used merce) attracted 120 exhibitors. Amish furni- turer in Holmes County, “We sell parts to by cluster firms (Terreri 2008), even though ture firms from surrounding states (e.g., Indi- and buy from our competitors. You either demand has shifted from red oak to more ana, ) also have started exhibiting know them or know of them” (Terreri 2008, diffuse-porous species such as cherry at the show [3]. p. 21). Unique interfirms linkages are the (Prunus serotina Ehrh.) and (Acer result. According to Porter (1998), repeated spp.) in the US marketplace for appearance- Interfirm Dynamics market exchanges are common among prox- based products such as furniture (Luppold Within the Holmes County furniture imal companies within clusters. and Bumgardner 2007). For example, only cluster, which consists of numerous small 3% of the bedroom and dining room show- firms and a few relatively large firms, the Study Objectives ings at the 2008 High Point Furniture Mar- structure resembles a multicentered indus- The first objective of this study was to ket were in red oak, while cherry, maple, trial district, which is defined by Brookfield estimate total lumber use by the Holmes rubberwood (Hevea brasiliensis Muell. Arg.), (2008, p. 408) as, “… an industrial district County furniture cluster. The results from and white oak (Quercus alba L.) were 12, 9, 9, made up of a number of [locally-owned] preliminary research have suggested that and 7%, respectively (Appalachian Hardwood

76 Journal of Forestry • March 2011 Manufacturers, Inc., 2008). Ash (Fraxinus spp.), walnut (Juglans nigra L.), and birch (Be- tula spp.) also were higher than red oak. Methods A questionnaire was developed with in- put from several manufacturers and suppli- ers working in the Holmes County furniture cluster. Although not formally pretested, the questionnaire was discussed line by line in two separate group meetings with these rep- resentatives. For the present study, the “Holmes County region” was defined as Holmes County and portions of five sur- rounding counties in northeastern Ohio, representing an area of approximately 1,000 mi2. A map was provided on the question- naire to make clear to respondents the geo- graphic definition of the cluster for the pur- Figure 2. Breakdown of the sample by sales category for 2007. poses of the study. A packet containing the seven-page spondents because it seemed apparent from Table 1. Product types manufactured in questionnaire, a cover letter, and postage- other data provided that a correctable error the cluster. paid return envelope was sent in mid-May of had occurred. 2008 to 569 firms. The sampling frame was Percent of The Furniture Book: A Complete Guide to the Assessment of Potential Nonresponse Product type Respondentsa Furniture Manufacturers and Wholesalers in Bias Ohio’s Amish Country (Anonymous 2005). A Potential nonresponse bias in the sur- Household furniture 80.6 Office furniture 35.7 reminder postcard was sent to nonrespon- vey was assessed in two ways. First, sample Cabinets 29.1 dents in mid-June. Last, a second packet statistics were compared with known popu- Components and dimension 20.5 (containing a duplicate questionnaire, post- lation parameters developed from The Fur- Moldings/millwork 7.1 Institutional/contract furniture 5.6 age-paid return envelope, and updated cover niture Book (Anonymous 2005), as discussed Outdoor furnitureb 2.6 letter) was sent to all nonrespondents in late by Bumgardner et al. (2007). The mean em- a June. Dennis (2003) recommends that mail Because respondents checked all categories that applied, col- ployment and establishment year in the umn totals to more than 100%. surveys of small business owners include at sample was 7.6 (median ϭ 4.0) and 1994 b Firms in which their production was strictly outdoor furniture least three contacts to improve the response (median ϭ 1996), respectively, which com- were not included in the study. Outdoor furniture listed here was from firms also producing interior furniture. rate. All mailings originated from (and were pared favorably with mean values of 7.2 for ϭ returned to) the Ohio Agricultural Research employees (median 4.0) and 1994 (me- In addition, early respondents (first ϭ and Development Center in Wooster, Ohio. dian 1996) for establishment year for the round and reminder card) were compared A total of 196 usable questionnaires population. Furthermore, Bumgardner et al. with late respondents (second round) on were returned for an adjusted response rate (2007) reported a total of 2,723 manufac- several demographic variables. No signifi- of 43.4% after removing undeliverable ad- turing employees in the Holmes County cant differences were found for the follow- dresses and those respondents that were not furniture cluster; the sample included 1,433 ing variables: gross sales for 2007 (P ϭ 0.15, manufacturers (e.g., finishers, suppliers, and employees or 52.6% of that total, which based on a chi-square test), proportion pro- distributors). Over 96% of respondents in- compared favorably with the survey re- ducing household furniture (P ϭ 0.36, dicated that they were the firm owner or co- sponse rate of 43.4% (for developing lumber based on a z-test for proportions), number owner; the remainder indicated they were use estimates, the effective response rate of of furniture pieces produced per year (P ϭ shop managers, with the exception of one 52.6% is used in the remainder of the article 0.38, based on a t-test), establishment year who indicated being a worker. since it likely is a better reflection of the pro- (P ϭ 0.92, based on a t-test), number of On inspection of the data, it was appar- duction capacity of the sample). Finally, the employees (P ϭ 0.41, based on a t-test), and ent that a small number of the lumber use proportion of non-Amish owned firms was ratio of lumber use per employee (P ϭ 0.86, figures provided by respondents were unre- reported to be approximately 15% for the based on a t-test). Thus, nonresponse bias alistic given the number of employees em- overall Holmes County furniture cluster was assumed not to be a major factor when ployed at the firm and/or the number and (Bumgardner et al. 2007); this figure was interpreting the results. type of furniture pieces manufactured. Some 14% for the sample (based on those respon- appeared overly high (e.g., 65,000,000 bd dents indicating they powered their shops Results ft) and others seemed too low (e.g., 100 bd with either single-phase or three-phase con- ft). Such values were removed from the data nections to the electric grid), suggesting Background Characteristics set for four respondents. Furthermore, lum- close agreement between the sample and The distribution of respondents by sales ber use responses were altered for two re- population. category for 2007 showed that respondents

Journal of Forestry • March 2011 77 Table 2. Wood use estimates for responding firms. Lumber Use Volume A variety of wood material types were Wood use measure Quantity (bf) used by respondents to make products. Nearly all firms used some hardwood lum- Lumber consumed by responding firms 16,987,153 ber, and a majority also used hardwood di- Dimension consumed by responding firms 3,437,529 Dimension consumed, adjusted up for associated lumber consumptiona 5,288,506 mension (defined on the questionnaire as Total lumber consumption by responding firms (sum of rows 1 and 3) 22,275,659b squares, blocks, and edge-glued products) and plywood. Most plywood is used for a Assuming 65% yield of dimension from hardwood lumber (Buehlmann et al. 1998). b Based on the effective response rate of 52.6%, total lumber use for the cluster was estimated to be 42.3 mmbf. drawer bottoms and the backs of larger pieces (dressers, hutches, entertainment cen- ters and more). Dimension has had some Table 3. Number of firms and employees, lumber use per employee (in bd ft), and lumber consumption by firm category. value added but requires further processing at the furniture shop. About 40% of respon- dents also reported using components (e.g., Small furniture firms Large furniture firms chair parts, drawer fronts, and so on, that are (1–5 employees) (6 or more employees) Components firms ready for assembly into complete pieces). Number of firmsa 83 40 24 Thus, most firms desire some preprocessing Number of employees 312 744 377 of materials by upstream suppliers. Lumber use per employee (mean)b,c 7,986 10,846 31,786 90% Confidence interval for the mean (6,168; 9,803) (8,172; 13,521) (22,723; 40,850) For the present study, the primary in- Total lumber consumption 2,491,632 8,069,424 11,983,322 terest was in use of hardwood lumber and dimension, given that most Amish furniture a Does not total to 196 because of some missing values for lumber use and/or firm size used to calculate ratios. b The associated medians were 5,712; 7,035; and 26,044, respectively, suggesting the means were reasonable measures of central is solid wood construction and these materi- tendency for each group. als typically are measured in bd ft [4]. The c Weighted average (by number of employees) equal to 15,732 bd ft/employee. volume of dimension consumption by firm was adjusted upward (by 35%) to form a were fairly evenly distributed across catego- (54%) of respondents indicated they lost lumber equivalent, because associated lum- ries up to the $500 thousand mark (Figure sales volume in 2007 compared with 2006; a ber consumption would be higher because 2). Sixty-seven percent had gross sales of plurality of the sample (25%) indicated sales of production losses related to cutting-to- $500,000 or less and about one-quarter had volume was off by about 10%. The timing of size, defecting, and more (Buehlmann et al. sales of less than $100,000, suggesting the the study is noteworthy, with these answers 1998). Wood use in components was ac- small nature of most firms. Slightly more being based on 2007, just before (or including) counted for by the hardwood lumber used than 5% of the sample had sales of $3 mil- the recession beginning in late 2007–2008; by component manufacturers in the sample. lion or more. These figures confirm the the study results should be interpreted with Thus, all wood use estimates are based on bd “multicentered” nature of the cluster. These this caution. ft of hardwood lumber consumed. As shown data also were used to develop an estimate of Although the majority of respondents in Table 2, lumber use for responding firms total sales for the cluster. By assigning each were furniture manufacturers, several prod- was 22.3 mmbf. Based on the effective re- respondent a sales figure representing the uct types were represented. As shown in Ta- sponse rate of 52.6%, total lumber use for midpoint of the category they selected, the ble 1, over 80% of respondents indicated the cluster was estimated to be 42.3 mmbf. sample accounted for approximately that they produced household furniture. $148,606,000 in sales in 2007; extrapolat- Lumber Input Productivity Additionally, 36% produced office furniture ing by the effective response rate of 52.6% To assess the input productivity charac- and 6% produced institutional/contract fur- gave an estimated total gross sales figure of teristics of the cluster, three categories were niture. Nearly 21% produced components $282.5 million. An estimate of $280.7 mil- established: small furniture firms (1–5 em- (defined as ready to assemble) or dimension lion was provided by Bumgardner et al. ployees, n ϭ 109), large furniture firms (6 or (defined as squares, blocks, and edge-glued (2007) using secondary data sources, sug- more employees, n ϭ 52), and components gesting close agreement. This represents ap- products) to support manufacturing both firms (n ϭ 28) [5], which also were a mix of proximately 3% of total US production of within and outside the cluster. Thus, the re- smaller and larger firms, but not numerous nonupholstered wood household furniture gion supports a variety of related wood prod- enough to separate based on size. Lumber (Cochran 2008). ucts production, although furniture is use and input productivity measures by firm The average firm operated 42.7 hours/ clearly the primary final product. Approxi- size and type of product manufactured are week; the most common response to this mately 69% of furniture production, on av- shown in Table 3. Modest increases in input question was 45 hours/week. For 66% of erage, was sold in stores dedicated to Amish- productivity were noted when moving from respondents, wood products manufacturing made products, showing the importance of the small to large furniture firm categories. was their sole occupation. Among those this channel to the Holmes County furni- Lumber use per employee increased from with multiple occupations, 46% counted ture cluster. As described previously, many 7,986 to 10,846 bd ft (although the confi- wood products manufacturing as their Amish-dedicated stores allow for semicus- dence intervals for the two means over- “full-time” occupation, and 29% indicated tomized orders of Amish products (choice of lapped, thus any differences are negligible; a that farming or agribusiness was their full- species, finish, and hardware), a key compo- t-test also suggested that the difference was time occupation. Slightly over one-half nent of the Amish model. not significant, P ϭ 0.14). Although these

78 Journal of Forestry • March 2011 Table 4. Species of lumber consumed in Table 5. Suppliers of hardwood lumber for the cluster, furniture, and component firms. the cluster.

Furniture firms Component firms Species category Percent of total volumea Furniture firms (percent of total Component firms (percent of total Source (average) volume) (average) volume) Red oak 45.5 Cherry 17.0 ...... (%) ...... Yellow-poplar 11.2 From local distributors and 90.7 85.7 71.9 38.9 Soft maple 8.2 sawmills White oak 6.2 From other distributors and 6.7 10.7 23.1 58.8 Hard maple 3.8 sawmills Hickory 3.0 Other 2.6 3.6 5.0 2.3 Walnut 1.4 1.2 Other 2.5 a These figures were very similar in terms of percent of total exposed parts such as drawer sides, whereas Discussion and Conclusion volume in the cluster (shown) and average percentages calcu- lated across firms for each species (ignoring volume), suggesting red oak and cherry are used as visible sur- The Holmes County furniture cluster is that smaller firms and larger firms were similar in their species faces. Red oak was the dominant species characterized by many small, Amish-owned use. used, accounting for nearly one-half of total and -operated firms producing household consumption (45%). Nationally, produc- furniture and related wood products. This tion of red oak lumber (including graded study indicated that the cluster consumes a figures were well lower than the broader fur- lumber for appearance-based uses and in- significant volume of hardwood lumber, ap- niture industry ratio of 17,433 bd ft/em- dustrial lumber for use as pallets and railway proximately 42 mmbf annually. Although ployee (Bumgardner et al. 2007), compo- ties) was about a one-third of total production this result was consistent with previous find- nent firms had a ratio three to four times in 2007 (US Census Bureau 2008b); there- ings as to the cluster’s importance to regional higher than furniture firms. Over one-half fore, it seems that Amish manufacturing is ac- demand for hardwood lumber (Bumgardner (53%) of total lumber use was by compo- counting for a disproportionate volume of red et al. 2007), the present study provided ad- nent firms, which accounted for only 26% oak consumption and is therefore an impor- ditional information by detailing lumber use of the total employees and just 14% of the tant regional source of appearance-grade de- by company type and size, which sheds light number of firms in the sample. The mand for this species. on the input productivity associated with weighted average lumber use ratio across cat- the cluster. egories was 15,732 bd ft/employee. Thus, Lumber Distribution Channels A final consideration was the channels In aggregate, ratios of lumber use per although the small size of even “large” firms employee were similar to the broader furni- in the cluster seemed to reduce lumber input by which lumber was procured by manufac- ture industry. However, ratios were much productivity compared with the broader in- turers in the cluster. As shown in Table 5, higher among the component firms in the dustry, inclusion of the component produc- the vast majority of lumber received by fur- cluster compared with the furniture firms. ers, many supplying local furniture produc- niture manufacturers was locally oriented Thus, although few in relative number, it ers, works to drive up ratios comparable with (within the cluster), whether measured as to- seems that more than one-half of hardwood the overall industry. tal volume or as average percent of volume. lumber use by the overall cluster is ac- There was evidence of clustering effects This indicates that furniture firms, regard- counted for by component manufacturers, for the location of component firms near the less of size, were quite similar in their local and this material then goes to furniture furniture producers. For the average compo- lumber sourcing patterns. However, the sit- shops to be processed further and assembled, nent firm, 53% of their product sales stayed uation appeared somewhat different for or is exported outside the cluster. This likely within the Holmes County region, and 73% component firms. Although the majority of is a reflection, in part, on the small size and stayed within Ohio. Moreover, furniture lumber purchases were local on average, producers within the cluster sourced over nonlocal sources became the majority based specialized nature of most of the furniture 90% of their components, on average, from on total volume. This finding suggests that firms—there is a division of labor and well- local shops. smaller component firms tended to source defined supply chains within the cluster and locally while the largest of these firms (the local supply sources are important. It thus Wood Species Use largest users in the cluster) sought lumber seems evident that clustering enables these A variety of hardwood species were used from suppliers over a wider geographic small firms to reach aggregate lumber input in production, although only a few were range. Overall, local suppliers of hardwood productivity levels comparable with the commonplace. Most notably, red oak, lumber play a major role in the cluster, al- broader industry. As stated by Porter (1998, cherry, and yellow-poplar (Liriodendron tu- though these suppliers do not necessarily p. 80), “A cluster allows each member to lipifera L.) each accounted for at least 10% procure all their logs (in the case of sawmills) benefit as if it had greater scale or as if it had of total consumption (Table 4), measured or lumber/dimension (in the case of distrib- joined with others without sacrificing its by summing total consumption for each spe- utors) locally. Such reliance on local busi- flexibility.” This aggregate productivity af- cies across firms (firms provided information ness relationships between lumber suppliers fords the numerous small firms the opportu- on the proportion of each species they used and manufacturers is consistent with Por- nity to remain flexible and maintain a de- in addition to their total consumption). ter’s (1998) notion that deep and specialized sired lifestyle, e.g., family-based at-home Much of the yellow-poplar is used for non- supply chains are common within clusters. employment in the Holmes County area. As

Journal of Forestry • March 2011 79 one respondent commented when asked on ufacturing cluster helps sustain regional for- BRATKOVICH, S., K. FERNHOLZ,J.BOWYER, AND the questionnaire why the cluster had est-based economies and provides more A. LINDBURG. 2009. Forest-based economic clus- grown, “Farms weren’t available … by diversification for domestic appearance- ters: Models for sustainable economic develop- ment. Dovetail Partners, Inc., Minneapolis, working together [the Holmes County re- grade hardwood lumber markets beyond MN.9p. gion] grew to be a great source for furni- those directly related to housing construc- BRESNAHAN, T., A. GAMBARDELLA, AND A. SAX- ture.” One of the “new rules” for competi- tion and remodeling. ENIAN. 2001. ”Old economy” inputs for “new tiveness in the US furniture industry is to economy” outcomes: Cluster formation in the stay small while mastering a specific area of Endnotes new Silicon valleys. Ind. Corp. Change 10(4): expertise (Dugan 2009), which clustering in [1] A variety of wood products are manufactured 835–860. by the Amish community in the Holmes BROOKFIELD, J. 2008. Firm clustering and spe- a multicentered district helps the Amish cialization: A study of Taiwan’s machine tool firms to achieve through specialization County region, but household furniture is the principal product. For consistency, the industry. Small Bus. Econ. 30:405–422. (Brookfield 2008). term “Holmes County furniture cluster” is BUEHLMANN, U., J.K. WIEDENBECK, AND D.E. KLINE. 1998. Character-marked furniture: Po- Nearly one-half of total production for used throughout this article. tential for lumber yield increase in rip-first the cluster was in red oak in 2007, suggest- [2] Although most of the firms are concentrated rough mills. For. Prod. J. 48(4):43–50. ing that Amish manufacturing is an impor- in a two-county area (Holmes and Wayne; BUEHLMANN, U., M. BUMGARDNER,A.SCHULER, Ohio has 88 total counties), portions of the tant regional source of demand for red oak AND J. CRISSEY. 2009. Surviving the deepening given recent declines in popularity for this cluster extend into some surrounding coun- downturn. Mod. Woodwork. 23(4):20–25. ties, for a total area covering approximately species. Interestingly, anecdotal evidence 2 BUMGARDNER, M., R. ROMIG, AND W. LUPPOLD. 1,000 mi (Lowery and Noble 2000). 2007. Wood use by Ohio’s Amish furniture from discussions with manufacturers within [3] This information was sourced from the pro- the cluster suggests that red oak use likely cluster. For. Prod. J. 57(12):6–12. gram of the 2009 Ohio Hardwood Furniture BUREAU OF LABOR STATISTICS. n.d. Employment, was proportionally even higher within re- Market. hours, and earnings—National. Available online cent years, indicating that firms are moving [4] Respondents reported that, on average, 89% at www.bls.gov/data/; last accessed May 29, toward designs more consistent with the of their total wood material costs were for 2009. broader marketplace and diversifying as the solid wood materials (i.e., excluding plywood COCHRAN, M. 2008. Bulletin of hardwood market cluster grows. As one respondent indicated, and composite panels). Luppold and Bum- statistics: 2007. US For. Serv. Res. Note NRS- gardner (2009) reported that solid wood ma- a limitation to future growth of the cluster 18. 24 p. terials comprised about 57% of total wood DENNIS, W.J. 2003. Raising response rates in was “building oak country style furniture.” material costs for the broader household fur- mail surveys of small business owners: results Porter (1998) discusses the “collective iner- niture industry. of an experiment. J. Small Bus. Manag. 41(3): tia” that can form within clusters if compa- [5] Although mostly components firms, five 278–295. nies become too inward looking and thus were producers of dimension products. DUGAN, M.K. 2009. The furniture wars: How unable to perceive the need for innovation. Lumber use by these five firms was reduced America lost a fifty billion dollar industry. However, it also can be said that the cluster to account for the proportion of their pro- Goosepen Press, Conover, NC. 450 p. HARDWOOD MARKET REPORT. 2009. 2008: The has developed and grown to date by focusing duction that was consumed locally, which was counted as dimension use by the furni- year at a glance—12th Annual statistical anal- on this niche. Going forward, exposure to ture firms. ysis of the North American hardwood market- broader markets might be reflected in place. Hardwood Market Rep., Memphis, changes in the species mix used. TN. 78 p. Similar to previous work based on dif- Literature Cited INTERNATIONAL TRADE ADMINISTRATION. n.d. Household and office furniture index page. ferent methods, a total value of shipments AGUILAR, F.X., S.M. BRATKOVICH,K.FERNHOLZ, Available online at www.ita.doc.gov/td/ocg/ from the Holmes County furniture cluster A. GARRARD,R.GRALA,L.LEIGHTLEY,W. MARTIN, AND I.A. MUNN. 2009. The status of furniture.htm; last accessed July 27, 2009. of approximately $280 million was derived and opportunities for business clustering within JONES, R. 1996. Small-firm success and supplier in the present study, which represents about the forest products sector in the U.S. Full Rep., relations in the Australian boat-building in- 3% of total US production, this among ap- US Endowment for Forestry and Communi- dustry: A contrast of two regions. J. Small Bus. Manag. 34(2):71–78. proximately 400 mostly small firms operat- ties, Inc., Greenville, SC. 27 p. KREPS, G.M., J.F. DONNERMEYER, AND M.W. ing within roughly a two-county area. Al- ANONYMOUS. 2005. The furniture book: A complete KREPS. 1994. The changing occupational guide to the furniture manufacturers and wholesal- though a relatively small portion of the structure of Amish males. Rural Sociol. 59(4): overall US wood furniture industry, the ers in Ohio’s Amish country, 2005–2006 Ed. 708–719. Overland Publishing, Inc., Orrvile, OH. 168 p. model used in the cluster is consistent with a LOWERY, S., AND A.G. NOBLE. 2000. The chang- APPALACHIAN HARDWOOD MANUFACTURERS,INC. ing occupational structure of the Amish of the “paradigm shift” (Schuler and Buehlmann 2006. Wood species grow internationally at Mar- 2003) and thus an example of what can work Holmes County, Ohio, settlement. Great ket. Appalachian Hardwood Manufacturers, Lakes Geogr. 7(1):26–37. in the US-based industry. In the past, a large Inc., High Point, NC. 7 p. LUPPOLD, W., AND M. BUMGARDNER. 2007. Exam- US furniture plant would complete the en- APPALACHIAN HARDWOOD MANUFACTURERS,INC. ination of lumber price trends for major hard- tire production process under one roof, from 2008. Furniture makers utilize more solids, veneers. wood species. Wood Fiber Sci. 39(3):404–413. lumber procurement and drying, to manu- Appalachian Hardwood Manufacturers, Inc., LUPPOLD, W., AND M. BUMGARDNER. 2008. Forty facture of components in the rough mill, to High Point, NC. 7 p. years of hardwood lumber consumption: 1963 to BRADEN, R., H. FOSSUM,I.EASTIN,J.DIRKS, AND final product assembly and finishing. In the 2002. For. Prod. J. 58(5):6–12. E. LOWELL. 1998. The role of manufacturing LUPPOLD, W., AND M. BUMGARDNER. 2009. The Holmes County furniture cluster, firms are clusters in the Pacific Northwest forest products wood household furniture and kitchen cabinet smaller and more specialized but intercon- industry. Working Pap. 66, CINTRAFOR, industries: A contrast in fortune. For. Prod. J. nected. The presence of this successful man- Seattle, WA. 42 p. 59(11/12):93–99.

80 Journal of Forestry • March 2011 MOTTIAR, Z., AND S. INGLE. 2007. Broadening A new model for assessing national-forest-based nat- TERRERI, A. 2008. Amish furniture makers suc- the entrepreneurial perspective: Interpreneur- ural resources products and services. US For. Serv. ceed in cluster. TimberLine 14(5):16–23. ship in an Irish furniture region. Int. Small Bus. Gen. Tech. Rep. PNW-GTR-703. 33 p. US CENSUS BUREAU. 2008a. Industry statistics J. 25(6):667–680. SCHULER, A., AND U. BUEHLMANN. 2003. Identi- sampler: NAICS 337122 nonupholstered wood PORTER, M.E. 1998. Clusters and the new eco- fying future competitive business strategies for the household furniture manufacturing. Available nomics of competition. Harvard Bus. Rev. No- U.S. furniture industry: Benchmarking and par- online at www.census.gov/econ/census02/ vember–December, 77–90. adigm shifts. US For. Serv. Gen. Tech. Rep. data/industry/E337122.HTM; last accessed PUNCHES, J.W., E.N. HANSEN, AND R.J. BUSH. NE-304. 15 p. Nov. 18, 2009. 1995. Productivity characteristics of the U.S. TAMBUNAN, T. 2005. Promoting small and me- US CENSUS BUREAU. 2008b. Lumber production wood cabinet industry. For. Prod. J. 45(10): dium enterprises with a clustering approach: A and mill stocks—2007. Current Industrial Rep. 33–38. policy experience from Indonesia. J. Small Bus. MA321T(07), US Census Bureau, Washing- ROJAS, T.D. 2007. National forest economic clusters: Manag. 43(2):138–154. ton, DC. 8 p.

Journal of Forestry • March 2011 81