The Economic Determinants of the Number of Minority Farmers in the Southeast Region of the United States, 1969-1997

Previous migration research dealt mainly with forces that affect migration. In what follows, we review the literature on off-farm migration among minorities in the United States, emphasizing the economic determinants of occupational migration out of agriculture. We highlight the lack of previous research that examines the economic motivations for moving out of agriculture, a gap that this research seeks to address. Greenwood (1975) defined migration as farmers moving from one occupation to another for better opportunities, or economic motivations. Stark (2003) described migration as an individual response to a wage differential. Here, we define migration as when a person leaves farming to migrate to another career. The focus of this research is on minority farm opertors, as defined by the U.S. Census Bureau (U.S. Department of Commerce, Bureau of the Census). While our primary interest is on African-American farmers, data definitions and availability resulted in our study of farm operators of all minority groups. The Census defines the category, “Black and other races” to include Blacks, American Indians (Native Americans), Asian or Pacific Islanders, and all other racial groups other than White. We will further explore this definition in the data section below.

Salamon (1976) provided additional evidence that African Americans who owned land were losing their ability to maintain ownership, due to an inability to generate adequate income. Schulman (1989) discovered that both the average gross income and average total acres operated for non-white farmers were less than half that of white farmers, leaving non-white farmers at a higher risk of migration. Dawra (1990) asked why a farmer would want to explore other opportunities to maintain a decent level of living. One reason might be that a majority of small-scale farmers had been adversely affected by a decline in prices, which has caused an increase in debt. A second possible reason was the steady downward trend in prices, as well as a lack of resources. A third possible reason could be because farming is among the least cost-effective occupations in their region, which gives them the option to look into other career fields (Dawra 1990).

Greenwood (1975), stated, “a finding common to a number of gross migration studies is that income (and job) opportunities provide a better explanation of in-migration than they do of out-migration” (p. 400). Household responsibilities are the general effects that would cause an individual to migrate for newer and better opportunities such as employment (Lee and Roseman 1999). According to Bass and Alexander (1972), the choice of where to work due to the environment and climate may be as significant as to work and for whom to work for. Their research indicated that Whites were more attracted to better climatic and economic conditions relative to Nonwhites, who were more attracted to better economic conditions alone. This result suggests a potential divergence between migration patterns between white and minority farm operators, a hypothesis which is thoroughly analyzed and reported in this study below. Brown, Christy, and Gebremldhin (1994) studied the influence of technical and institutional forces that affects the population increase of African American farmers. They argued that the changes in the structure of agriculture had a significant impact on small-scale farmers by constraining the strategies available to farmers to increasing their farm size.

Grim (1995) stated that between the 1950’s and the 1970’s, in spite of the farm programs such as loan increases to Black Farmers, there were still a large amount of farmers who left the field to search for better jobs, valuable educations, better housings, and more recreation. Reynolds (2003) stated that increases in land ownership after the early 1900’s were partly due to a significant rise in cotton prices that lasted until the outbreak of World War I in 1914. Reynolds’ research was consistent with Gilbert, Sharp, and Felin (2001). The authors stated that the difference in the Census reports from the 1997 to 2002 is that it shows a significant increase in the South, but particularly with black tenant farmers and sharecroppers. According to Ponder (1971), land ownership was of prime consideration to remaining in farming because the tenant had to give up his land when the owner wishes it and because of this the probability of minority farmers staying in the agricultural field would be low.

Browne (1973) studied the effects of agricultural technology, farm subsidy programs, and general tendency for farm youth to gravitate toward urbanized areas. This has been an issue for some time according to Gilbert, Sharp, and Wood, (2002) who discussed how out of all private agricultural land, Whites accounted for 96% of the owners, 97% of the value, and 98% of the acre; while 25 million acres of land is owned by minorities. Their paper discussed the social, economic, cultural, and political consequences that are a result of land ownership. Molnar, Thompson, and Beauford (1988) identified another cause of this decrease as the advent of machinery that encouraged large farms and eliminated the need for small-scale tenant farmers. They also believed that African Americans faced great structural barriers such as discriminatory attitudes that often blocked their advancement in agriculture.

Wood and Gilbert (1998) asserted that farming may be less attractive to the younger generation due to the fact that it is looked at as a memory of slavery and sharecropping. Their beliefs are that we are wasting our time on trying to convince others to enter farming but to encourage the improvement of poor rural communities through education, training, and economic development. They believe that if agriculture would be a more viable business and a way of life by encouraging land retention and recovery efforts from the past, then the decline of African-American farmers and landowners could be reversed. Wood and Gilbert (2000) stated that the primary reason for decline of African-American farmers was due to the twin engines of increased mechanization and the dismantling of the sharecropping system. The research primarily targeted African-American farmers in the Mississippi Delta. However, these previous studies showed that a significant amount of African-American farmers still owned their land and would like to return but due to public policies, economic pressures, and racial oppression, many minority farmers find it impossible to return.

Several studies have attempted to investigate the relationships between the flow of labor out of agriculture and economic variables. Barkley (1990) analyzed a migration model, and found that when farm income increased relative to nonfarm income, the level of agricultural employment increased. Mundlak’s (2000) research was similar to Barkley’s, and he also hypothesized that if nonagricultural jobs were more attractive than agricultural jobs, then a decrease in farm labor is likely; if agriculture was more attractive than nonagriculture, an increase in farm labor is expected. In the research reported here, we use the economic approach of Barkley (1990) to explain minority migration of out of agriculture, as modeled formally in the next section. This approach emphasizes the economic differences between agriculture and the nonfarm sector. Institutions and structural changes have been shown to be important determinants of minority migration out of agriculture, but here influence labor movements only through their impact on economic variables, primarily the returns to labor in each sector.

Following Barkley (1990), a migration equation model was developed to examine the determinants of the number of minority farm owners in the Southeast Region of the United States. The Census of Agriculture defines a minority farm operator as an individual who farms the land; note that this category does not include hired farm workers. The category, “Black and other races” includes Blacks, American Indians (Native Americans), Asian or Pacific Islanders, and all other racial groups other than White (U.S. Department of Commerce, Bureau of the Census). The Census definition is for race and ethnicity; the data in these categories include both men and women. Thus, white females are not included in the minority data reported here.

Historically, the decline in the share of agriculture in the farm labor force has occurred over centuries (Barkley 1990). The term migration is an approximation to actual occupational migration out of production agriculture, and it considers only changes in the number of jobs in the farm sector. There are two ways of measuring changes in the number of workers in a given occupation such as agriculture: (1) the level of labor (L), and (2) changes in this level, or the rate of migration (M). All abbreviations in what follows are listed in Table 1. To determine the total amount of Labor in agriculture, we defined of L_ag to be a function of wages in agriculture (W_ag), nonagricultural wages (W_nonag), and agricultural output prices (P_c, P_r). Eq. (1) through (5) were used to determine the total amount of labor employed in agriculture. where and

The variable P_c is the cotton price and P_r is the rice price. Eq. (4) is the definition of D, the returns to labor in nonagriculture relative to the returns to labor in agriculture. The variable D is defined as the total non agriculture GDP per person divided by the total agricultural GDP per person.

Following Mundlak (2000) and Barkley (1990), the relative size of the labor force (g) is introduced into the migration equation by defining the variable g to be equal to the total amount of labor in non agriculture divided by the total amount of labor in agriculture, as in Eq. (5).

The size of the agricultural labor force relative to the nonfarm labor force is included in the model to capture the ability of the nonfarm sector to absorb farmers exiting the agricultural sector. A large nonfarm economy provides much better prospects for successful occupational change out of farming. Thus, the size of agriculture relative to the nonfarm economy is captured in the variable g, and included in the empirical model.

In Eq. (6), we also used the real price indexes (P_c and P_r) of two different crops, cotton and rice, to determine the level of migration. For an example, if the price of cotton (P_c) increases; than the number of minority farmers migrating into agriculture would be expected to increase as well. The number of workers in agriculture (L_ag) is expected to depend on the relative size of the labor force in each sector, reflecting the probability of obtaining employment in each sector. Where D = non-agriculture GDP per person/agriculture GDP per person, g = labor in non-agriculture/labor in agriculture, P_c is the real price of cotton, and P_r is the real price of rice.

A regression was estimated with the dependent variable equal to the total number of minority farmers in each state, and the independent (explanatory) variables include D, g, and cotton and rice prices. We estimated alternative regressions with using dummy variables of the twelve southern states and years of 1969–1997. Also we tested several other potential explanatory variables including the prices of corn, soybeans, and sorghum.

However, each of these variables was statistically insignificant in regression trials. Four out of the six crop price variables were eliminated due to potential multicollinearity (Kennedy 2008). Collinearity exists across the commodity price time series, since commodity prices are highly correlated due to market forces. This approach has been used extensively in models that include commodity prices, including Blank and Ayer (1987), Bailey and Womack (1985), and Chambers and Just (1979). Out of the six crop prices that were originally included in the regression, two were found to have statistical significance in preliminary regressions: cotton and rice. The model is specified in (7).

Where:

Note that this model is specified without race included; the specification is general enough to accommodate occupational migration for any racial group. We estimated the model for minorities in American agriculture, to extend the results of Barkley (1990), who estimated the same model for all farmers in the United States. Additionally, we compare migration rates out of agriculture in the Southeast Region across white farmers and minority farmers. These populations changes are further discussed in the next section, where differences between minority and white farm owner migration is highlighted. In the model estimated here, the expected signs of the coefficients are (all other variables remaining constant):

We hypothesize that if farming in the Southeast were to become more economically attractive, more minority farmers are likely to remain in agriculture in the Southeast region. As farming revenues and the prices for crop variables increase, we expect for the total number of minority farmers to increase, or, since the overall trend in the number of farmers is downward, decrease at a decreasing rate.

There is an interesting alternative hypothesis regarding the estimated coefficients on the commodity prices. Higher commodity prices could provide the necessary level of income to cover the cost of migration, in which case the coefficients β3 and β4 would have a negative sign. In this case, the supply of labor to agriculture would be “backward bending,” or the income effect would outweigh the effect of supply response to higher earnings (Hanoch 1965). The regression model estimated here would also allow for this possibility, and the empirical results will demonstrate which hypothesis is supported by the data.

The data were taken from the Bureau of the Census (US Department of Commerce) for the number of white and minority farmers in the Southeast. We selected the Southeast due to the fact that the Southeast Region is historically more prominent in minority farming than other regions in the United States. According to the Census, in 1964 there were a total of 199,952 nonwhite farm operators in the United States, and of this total, 92% were African Americans and 8% were classified as other nonwhite. Ninety-two percent of all nonwhite operators were in the South, and of these 98% were black and so being that the South had a larger percentage of African Americans (Ponder 1971). Therefore, our targeted areas were the 12 states in the Southeast Region, including Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Virginia, and West Virginia. Texas was excluded from the original data set, as it was considered to have more in common with the semi-arid Great Plains agriculture, rather than the Southeastern region examined here. In an alternative specification of the model, Texas was included to find the similarities and differences in minority migration out of agriculture between Texas and the Southeast region.

Data on the number of farmers in all twelve Southern states were collected from the United States Department of Commerce, Census Bureau. Real per capita gross domestic product (GDP) data were collected from the United States Department of Commerce Bureau of Economic Analysis (BEA). The data included the years of 1969 to 1997 for the twelve Southern states. We originally desired to start the data with the year of 1920, but due to lack of data availability for Gross Domestic Product in agriculture and nonagriculture and an accurate account for the total of minority farmers, we were constrained to start the data in 1969. The ending data was 1997, since the 2002 Census data had been redefined, and were no longer consistent with the data from the earlier time period. Specifically, beginning in 2002, multiple operators per farm were counted, leading to significantly larger numbers of minority farmers included, compared to previous years, as evident in Table 2 and Fig. 1.

Also, beginning with the 1997 Census of Agriculture, the National Agricultural Statistical Service (NASS) directed the Census, instead of the Department of Commerce. This change led to special efforts to more accurately measure women and minority farmers, particularly in the 2002 Census (USDA/NASS Quick Stats). These activities included, but were not limited to, obtaining mail lists from organizations likely to contain names and addresses of minority farm operators and conducting pre-census promotion activities that targeted women, American Indian and Alaska Native, Black and African American, and Spanish, Hispanic, or Latino origin farm operators (USDA/NASS Quick Stats). Gilbert, Sharp, and Felin (2001) provided an outstanding review and summary of these data issues, and concluded, “Despite the drawbacks of the Census of Agriculture, it is the best source of data on farmers. It is the most comprehensive compiling national data down to the county level. It recurs every 5 years, is accessible, and easy to use.” Thus, we use the Census data in this research, but limit our time period to before 2002, due to the change in statistical methodology.

The GDP data were taken from the United States Department of Commerce/BEA website located under the “Regional” section for Gross Domestic Product by State. The GDP in agriculture data were also taken from the same section but under the section named agriculture, forestry, fishing, and hunting. The GDP in Non Agriculture data were calculated by subtracting the total GDP for each state from the total amount of GDP in agriculture for each state. The total amount of Employment was taken from the U.S. Department of Commerce BEA website located under Regional Economic Accounts and State Annual Personal Income.

The second set of data required inflation adjusted prices of rice and cotton. This information was taken from the annual publications of the (USDA/NASS Agricultural Prices). To adjust for inflation, the Consumer Price Index (CPI) was used, taken from the Bureau of Labor Statistics (United States Department of Labor, Bureau of Labor Statistics 2008). Dollars were adjusted to 100 in 2007.

The first included crop price was cotton. Cotton in the South is a very dominant cash crop and generates three-fourths of the world’s cotton supply. “Cotton is the single most important textile fiber in the world, accounting for nearly 40% of the total world fiber production. While some 80 countries from around the globe produce cotton, the United States, China, and India together provide over half the world’s cotton. The U.S. cotton industry typically generates over 400,000 jobs in the industry sectors from farm to textile mill” (USDA/ERS Briefing Room Cotton, 2008). Rice was first planted in the USA in South Carolina and found its place in society mainly in the southern states such as Arkansas, Louisiana, and east Texas since the 1800 s. “Rice is produced worldwide and is the primary staple for more than half the world’s population. In the United States, rice farming is a high-cost, high-yielding, large-scale production sector that depends on the global market for almost half its annual sales.” (USDA/ERS Briefing Room Rice 2008).

Table 2 shows the total number of minority farmers from the years of 1964–2002 (United States Department of Commerce, Bureau of the Census). Table 3 is the total number of white farmers for the time period 1964–2002 (United States Department of Commerce, Bureau of the Census). Table 4 reports the total number of white and non white farmer’s form the years of 1964–2002. This Census data were also collected from the United States Department of Commerce, Bureau of the Census. Although much effort was expended making the census mail list (CML) as complete as possible by National Agricultural Statistics Service (NASS), the total coverage of farms was considered inadequate. Gilbert, Sharp, and Felin (2001) reported definitional changes in the census instrument resulted in large changes in the number of minorities in agriculture. Because of these major changes, we excluded the 2002 census numbers due to lack of comparability, as detailed below.

Therefore, we investigated how the number of minority farmers changed over the period 1969 to 1997, and compare the rate of change to that of all farmers during the same period (Fig. 2). This allows us to see if minority occupational migration out of agriculture is similar to or different from all farmer migration. Table 5 presents data indicating the percent change in the total number of all farmers as well as the total amount of minority farmers. We took the total number of white and non-white farmers during the years of 1969 and 1997 to give us an overview of the structural trends that have taken place over that time period. The percentage of minority farmers had a higher decrease in change than the total number of all white farmers (Table 5). These results indicate that the changes in the number of minorities farming were greater than the changes in the number of white farmers over time, demonstrating that there is a difference between minority farmers and white farmers entering and exiting agriculture. For example, in Alabama, the percent change for all farmers equaled −42% between the years 1969 to 1997. The percent change for minority farmers for the years 1969 to 1997 was −77%. There was a significantly larger percentage of minority farmers entering and exiting the field of agriculture than all farmers. However, if the all farmer percentages were similar to minority farmers, then we could conclude that labor migration was similar for both minority and whites. Since the changes differ significantly, minority migration levels differ, resulting in the motivation for this study of minority farmers, to measure the impact of economic variables on minority migration out of agriculture separately from all farm operators, which was studied by Barkley (1990).

The results for the regression that includes Texas in Table 5 demonstrate why it was excluded from the original data. During the period 1969 to 1997, the number of minority farm operators in Texas increased from 5375 to 7862. This pattern is opposite of the twelve Southeastern states excluding Texas, where large decreases of minority farmers occurred. Texas agriculture is distinct and separate from agriculture in the Southeast, as the arid Great Plains receive a much lower level of precipitation than the Southeastern states.

This study has provided empirical evidence that (a) minority farmers’ response to economic conditions in the Southeastern Region of the United States is statistically significant and (b) the returns to farming, relative to nonfarm occupations returns to labor is associated with a direct correlation with a minority farmer’s decision to migrate into or out of the field of agriculture and (c) globalization and international trade are likely to cause a response to minority farmers migration. In today’s society, farming is neither an option nor a necessity unless there is an economic benefit for farmers. For this reason, it has caused fewer minority farmers to continue into the field of agriculture. The responsiveness of labor to migrate in or out of agriculture based on labor returns is one primary determinant of minority farmer’s occupational choice. Our research has shown that economic determinants have a direct effect on the number of minority farmers migrating in and out of the field of agriculture. Our results have also demonstrated that if income was to increase in the nonagricultural sector, then the total number of minority farmers would decrease due to better income opportunities. Our results also show that if more people were employed the nonagricultural sector, than minority farmers would also decrease because of the assumption of better job opportunities. The final conclusion was the impact of price of cotton, which was found to have a statistically significant impact on the migration of minority labor out of agriculture in the Southeastern region.

Cotton is produced globally, and one of the most important textile fibers in the world, including in China and India. Our results demonstrated that if the export price of cotton were to increase, then the number of minority farmers in the American Southeast would increase. This analysis was a good indicator that the impact of international trade could have a strong effect on the determinants of minority farmers migration out of agriculture. Historically, the United States Department of Agriculture and other Agriculture Extension Agencies have made several attempts to make the field of agriculture as attractive as possible to minority farmers, along with a few trials and errors. Agriculture Secretary Ann M. Veneman (2003) stated on September 3, 2003 in Washington, D.C.:

We are committed to helping the nation’s minority and disadvantaged farmers... The grants will help many farmers and ranchers to successfully acquire, own, operate and retain farms and ranches by delivering a wide range of outreach and assistance activities including farm management, financial management and marketing.

To the extent that trade agreements increase commodity export prices, globalization and trade are consistent with this stated policy. Therefore, our research results indicate that globalization and international trade agreements may be complementary to other public policies intended to support minority farmers.