Investigating Factors Affecting the U.S. National Nursery Trade: The Implications of a Gravity Model

Literature on the domestic trade of nursery crops is sparse. Based on national survey data collected in years 1999, 2004, and 2009, we used augmented gravity models to investigate the primary factors affecting the value of trade for both large and small nurseries. We found that the impact of distance on trade value was different between large nurseries and small nurseries; the impact of distance on national nursery trade has been decreasing over time; and the level of impact of distance on nursery trade differs across regions. Additionally, the value of nursery trade was affected by plant types the nurseries produced and other business characteristics. In the United States, nursery and greenhouse production represents one of the largest horticulture industry sectors and has become an important contributor to the nation’s agricultural economy. The Economic Research Service of the U.S. Department of Agriculture (2007) reports that, in 2006, U.S. nursery and greenhouse crops generated $16.9 billion in sales, which was an increase of nearly 18% since 2000. Nursery crops include woody ornamental crops such as trees and shrubs including deciduous shade and flowering trees, vines and groundcovers, and foliage plants. According to the latest survey conducted by the National Agricultural Statistics Service, growers of nursery crops in the 17 U.S. states surveyed gained $4.65 billion in gross wholesale value in 2006, which was a 17% increase from 2003. Another survey that is conducted independently of these government statistics is entitled: Trade Flows and Marketing Practices Within the United States Nursery Industry. The survey has been conducted by the Green Industry Research Consortium five times at 5-year intervals since 1989. Several studies have been conducted using the database from these surveys. Hinson et al. (1995) used the 1989 survey data to study choices of nursery transaction channels. Behe et al. (2008) and Hodges et al. (2008) investigated the regional differences in the production and marketing practices in the U.S. nursery industry. Velástegui Andrade and Hinson (2009) subsequently studied choices of nursery market channels. Few studies have been conducted that investigate the factors that affect the flow of products in nursery crop trade. The gravity model has been widely used for agricultural trade-related studies. Zahniser et al. (2002) explored the impact of regionalism on U.S. agricultural exports using a gravity model. Grant and Lambert (2008) investigated the effects of regional trade agreements on members’ agricultural trade. Both studies found that the gravity model performed well on estimating agricultural trade. Koo et al. (1994) used the gravity model for a single agricultural commodity. They found that distances had a significant effect on trade flows of meat. As another type of agricultural product, we hypothesize that factors such as distance and state gross domestic product would also affect the trade of nursery crops. As the name implies, the idea of the gravity model actually stems from Newton’s law for gravitational force between two objects. The gravitational force is inversely proportional to the distance between two objects and directly proportional to the masses of objects. In the basic gravity model of trade, the trade flow corresponds to the gravitational force; and the market sizes, which are often measured by gross domestic product (GDP), income, or population, are used to measure masses. Trade flows are expected to be inversely proportional to the distance between two regions and directly proportional to the market sizes. Anderson (1979) developed theoretical justifications for the gravity model. He stated that the gravity model was developed on the basis of constant elasticity of substitution. He assumed that preferences for traded goods are identical across countries and are homothetic and goods are considered differentiated by place of origin. Recently, Deardorff (1998) extended the assumption to all goods. For our study, we assumed preferences for nursery plants are identical across states and nursery plants are differentiated by state of origin. The gravity model can be augmented with other variables that are related to trade flows. When Tinbergen (1962) first introduced the gravity model to predict international trade flows, the key variables in the model included distance, a measure of market size for both exporting and importing countries, dummy variables for trade agreements, and a dummy variable as a common border in the gravity model. Pöyhönen (1963) used the square root of GDP to represent market size in the gravity model. Population and per-capita income were used by other researchers to measure market size (Carrillo and Li, 2002; Koo and Karamera, 1991; Linnemann, 1966). An augmented gravity model not only includes variables that measure distances and market sizes, but also other variables such as exchange rates, income differences, and dummy variables for infrastructure, common language, or adjacency (Bougheas et al., 1999; Deardorff, 1998; Inmaculanda and Felicitas, 2003). One of the most important factors that affect trade flows is the actual distance between two regions. A distance variable is used to capture the influence of transportation costs (both freight cost and time cost), which is one of the important parts of trade cost (Anderson and van Wincoop, 2004). Generally, trade increases as transportation cost decreases. Baier and Bergstrand (2001) showed that 8% to 9% of the trade flow growth could be explained by transportation cost reduction. The distance between two regions is often measured by air routes (Frankel, 1997) or by sea routes (Bikker, 1987). Most empirical studies show that distance between two regions negatively affects the trade (Egger, 2002; Leamer and Levinsohn, 1995). The distance between two regions does not change over time. However, transportation costs often change over time. The impact of distance on trade is expected to diminish in magnitude over time because of developments in transportation and storage technology. Several studies have confirmed this hypothesis Received for publication 22 Feb. 2011. Accepted for publication 10 May 2011. M.S. graduate student, Department of Applied Economics, University of Minnesota, 1970 Folwell Avenue, St. Paul, MN 55108. Assistant Professor, Departments of Horticultural Science and Applied Economics, Bachman Endowed Chair in Horticultural Marketing, University of Minnesota, 1970 Folwell Avenue, St. Paul, MN 55108. Professor, Department of Horticultural Sciences, Texas A & M University, 202 Horticulture Forest Science Building, College Station, TX 778432133. To whom reprint requests should be addressed; e-mail [email protected] 1518 HORTSCIENCE VOL. 46(11) NOVEMBER 2011 (Bikker, 1987; Brun et al., 2005; Frankel, 1997). A measure of market size is another factor included in gravity models. Generally, variables such as GDP, population, or income are used to measure the market size. Studies have shown that market size affects trade in a positive way: the larger the market size, the larger the trade volume (Llano et al., 2010). Most nurseries have both intraregional trade and interregional trade with the vast majority of nursery crops produced in the United States being sold to the domestic market. Brooker et al. (2005) reported that less than 10% of U.S. nurseries exported to foreign countries with those exports representing less than 2% of total annual firm sales in 2003. As a result of the lack of exports of nursery crops, we focus on the domestic nursery trade within the United States for this study. Previous studies showed that interregional trade tends to grow faster than intraregional and international trade (Hitomi et al., 2000; Munroe and Hewings, 1998). Interregional trade has been studied through various methodologies to estimate the effect of distance, transportation connectivity, and industry characteristics on commodity flows. Gravity models, Neuronal networks, and input–output models are among the well-known methods (Hitomi et al., 2000; Llano et al., 2010, Nijkamp et al., 2004). Llano et al. (2010) used the gravity model to study Spanish interregional trade. More importantly, they showed that the gravity model was a valid method to study interregional trade. Their gravity model estimation indicated that a 1% increase in the distance between two regions could cause a 1% decrease in the total bilateral trade in Spain. Similar to Llano et al. (2010), we used the gravity model to explore the major factors affecting domestic interstate nursery trade. Nursery trade flows are influenced by firm size. Velástegui Andrade and Hinson (2009) explored the impact of business characteristics on market channel by firm size. Their study showed that firm size affected the channel choice and differences exist between large and small nurseries. We expect factors that affect trade flow would differ between large nurseries. Therefore, we model large nurseries and small nurseries separately. To our knowledge, this is the first study to use the gravity model methodology to examine the U.S. domestic nursery trade. In addition, we extended the traditional gravity model specification with augmented variables associated with selected business characteristics. The following business characteristics are included: nursery location, years in business, contracted sales, market channel, percentage of sales to repeat customers, negotiated sales, and the level of exports. All of the major product types aforementioned were also included in the model. Another objective is to explore whether these variables affect the domestic trade conducted by large and small nurseries in different manners. Specifically, we tested the following hypotheses: 1) the impact of distance on large nurseries is smaller than it is on small nurseries; 2) the impact of distance on trade declines over time; 3) market sizes of the states where both buyers and sellers are located have positive impacts on the domestic nursery trade; and 4) other business characteristics affect the domestic nursery trade. Material and Methods Data. We used pooled cross-sectional survey data from 3 years to explore how distance, market size, and business characteristics affect the domestic trade in the nursery industry. The data used in this study were collected in 1999, 2004, and 2009 through mail-in surveys or Internet surveys. Twentytwo states participated in the survey in 1999 and 1718 nurseries (24% response rate) responded to the survey. Forty-four states participated in the survey in 2004 and 2485 nurseries (15.9% response rate) responded to the survey. In 2009, the survey was sent to nurseries in all 50 U.S. states and 3044 nurseries (18% response rate) completed the survey (Hall et al., 2011). We dropped selected observations from the data set if: 1) the destination was out of the United States because our focus was on the domestic nursery trade; 2) gross sales were equal to or less than $10,000, based on the USDA definition for commercial production (Velástegui Andrade and Hinson, 2009); 3) information about the year in which the firm was established was missing; 4) information about the nursery’s home state was missing; or 5) the percentage breakdown among plant categories was missing. After imposing these conditions, the number of firms in the sample was trimmed to 5140. The survey asked nursery producers to list the top five or six states (including their own state) that were destinations for their sales and indicate the percentage of sales to each of the five or six destination states. We used the sales to each destination as the dependent variable. Although many nurseries did not provide all five (or six) destinations, the total number of observations was 12,543. Model. The impact of distance and business characteristics on trade flow was estimated separately for small nurseries and large nurseries with nursery size measured by each nursery’s respective annual gross sales. Survey participants were asked to either write down their actual sales or choose from 11 sales ranges. For the observations that selected the sales ranges, the midpoints of each category were used to approximate the gross sales. Following Velástegui Andrade and Hinson (2009), we used $500,000 as the cutoff value to separate large and small nurseries, which resulted in 5,666 observations for large nurseries and 6,877 observations for small nurseries. Augmented gravity models were used to estimate how different factors affect the domestic trade of nursery crops. The augmented gravity model can be express as follows: ln X ij = b0 + b1 ln Y i + b2 ln Y j + b3 ln Dij + b4 t ln Dij + b5 R ln Dij