Literature Review

View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

Archived

This page has been archived on the Web.

Determinants of off-farm income

The extensive literature on off-farm labour supply and off-farm income provides many insights on farmers that are more likely to have off-farm income. This literature reports on the relationship between the characteristics of farms (e.g. type, size, business organization) and farmers (e.g. age, education, family size) and off-farm labour allocation. In terms of farmers' characteristics, the literature suggests that age would have an inverted U-shape relationship with the likelihood of off-farm work; higher education would increase the likelihood of working off-farm; and farming experience would reduce the likelihood of off-farm work (Furtan, Van Kooten, and Thompson, 1985; Mishra and Goodwin, 1997; Howard and Swidinsky, 2000; Alasia et al., 2007; El-Osta, Mishra, and Morehart, 2008).

Regarding farm characteristics, dairy farmers and to a lesser extent hog and vegetable farmers would be identified as being less likely to work off the farm, while the reverse would be true for grain and oilseed farmers (Howard and Swidinsky ,2000; Alasia et al.,2007). Most studies also report that farm size, as would be expected, would have a negative impact on the likelihood of off-farm work by the operator. This result appears to be similar regardless of the indicator used to measure farm size (e.g. gross sales, farm capital, acreage) (Mishra and Goodwin, 1997; Mishra and Holthausen, 2002; Howard and Swidinsky, 2000; Alasia et al., 2007; El-Osta, Mishra, and Morehart, 2008).

The impact of farm location and regional characteristics would have also been investigated in recent studies. Results are, however, not as robust and are sometimes unexpected. Intuition would suggest that population density is positively linked with a more dynamic labour market, thus increasing the likelihood of off-farm work. However, Howard and Swidinsky (2000) and Alasia et al. (2007) provide evidence that population density is negatively related to the likelihood of off-farm work. Similarly, distance to town or metropolitan areas has been found to be insignificant or to affect positively the likelihood of off-farm work, which is somewhat counter-intuitive (Mishra and Goodwin, 1997; Alasia et al., 2007; El-Osta, Mishra, and Morehart, 2008). Howard and Swidinsky (2000) found population density would increase the number of hours worked off the farm.

Government program payments would decrease the likelihood of off-farm work (Mishra and Goodwin, 1997; Howard and Swidinsky, 2000). To the extent that most payments are countercyclical and meant to stabilize farm income, the negative relationship with off-farm income may suggest that off-farm income is used as a substitute for program payments in an effort to manage farm income risk.

Farm income risk and off-farm labour supply

While many authors refer to farm income risk as a key motivator leading farmers to work off-farm, the literature providing empirical assessment of the relationship between farm-income risk and off-farm labour allocation is limited. Data availability is likely the key factor explaining the limited number of empirical studies. In order to study farm income risk, farm level longitudinal data are more suitable; however, such data sets remain scarce. In fact, given the paucity of farm level data most studies had to rely on aggregated data, despite the limitations imposed by aggregation biases in risk measures (OECD, 2009). Mishra and Goodwin (1997) is the only study found which uses farm-level data. Moreover, their study is based on a small sample which reduces the confidence with which these results can be generalized to the entire farm population.

Kyle (1993) was among the first to study the impact of farm income risk on off-farm income. Using state-level data from 1960 to 1986 and a standard linear regression, the study found that the share of off-farm income as a proportion of total income was higher in American states with higher relative variability of net farm income. These early results were supported by the work of Mishra and Holthausen (2002). This later study used county-level data and a logit model to estimate the impact of farm and farmer characteristics such as age, farm size, off-farm wage, and income variability on the likelihood of off-farm work. Results suggest that higher variability in farm income would be associated with higher off-farm income.

The role of off-farm income in reducing total farm household income variability was also studied by Mishra and Sandretto (2002). They examined the evolution of aggregate U.S. farm income and farm income variability between 1967 and 1999. Aggregated data at the national level were used to perform an analysis based on the variance, covariance of income components over time, including farm income, and off-farm income. The authors concluded that off-farm income has played an important role in reducing total income variability.

In terms of farm-level study, Mishra and Goodwin (1997) investigated the determinants of off-farm income for 300 Kansas farms. Farmers and their spouses were asked to report 10 years of on- and off-farm income (1981 to 1991) as well as various demographics (e.g. education, experience, distance to town, and family size) and farm characteristics (e.g. size based on acreage, leverage, program payments). Given that farms without off-farm income represented a significant share of the sample, a Tobit model was used to address data censoring issues. Results indicate that higher farm income variability would increase the likelihood of having off-farm income. To our knowledge, their study is the only one estimating the relationship between farm income risk and off-farm work based on operator-level data.

The Tobit model used by Mishra and Goodwin (1997) implicitly assumes that farm income variability would have the same impact on deciding whether or not to work off-farm and choosing the amount of off-farm labour. This assumption may not be appropriate. In fact, in their study of off-farm labour supply Howard and Swidinsky (2000) rejected the Tobit specification in favour of a more general two-part model. They also found that diverse explanatory variables such as age, spouse's income, and population density could have inverse effects on the operator's off-farm labour market participation and the number of hours supplied by the operator.

This study takes advantage of a farm operator-level longitudinal taxation data set developed by Statistics Canada and investigates the impact of farm income risk as an explanatory factor for off-farm labour allocation by farm operator. The data set also allows us to explore the robustness of this relationship across farm typologies and size, which has not been explored by previous studies. While farm income risk may be of greater significance for operators of larger farms as it tends to represent a higher proportion of their total income, these operators also face greater labour constraints which may prevent them from taking advantage of off-farm opportunities. This question is addressed in this study by comparing the results for five different farm typologies including operators of hobby/pension farms and operators of commercial farms of different sizes. A two-part model is developed to address data censoring issues and to assess the relationship between farm income risk and both the decision to participate in the off-farm labour market and the quantity of labour supplied.