Earnings and Disposable Income of Farmers in Sweden, 1997-2012

Earnings and Disposable Income of Farmers in Sweden, 1997-2012 Abstract This study presents a comprehensive analysis of farmers’ income in Sweden. The results indicate that farm households in Sweden do well from a standard-of-living perspective, but that farming is still a low-paid occupation from a return-on-skills perspective. Nevertheless, farm earnings increased faster over the study period than earnings in the general population, owing equally to higher farm earnings for operators and higher off-farm earnings for their spouse. Since few female spouses make farm earnings, evaluating farm household earnings from mainly a household perspective fails to acknowledge the individual careers of farmer and spouse. Farm income, off-farm income, disposable income, inequality, Gini, poverty Using longitudinal data for a period of 16 years (1997–2012), in this study we analyze changes in disposable income and earnings for farm households and changes in personal earnings for male and female farm operators, relative to those of households and men and women in the total population in Sweden. A fair standard of living for the agricultural community has been an objective in the European Union (EU) common agricultural policy (CAP) since the start.1 “Standard of living” is usually interpreted as the economic well-being of people, referring to their consumption possibilities measured by disposable income.2 In studies of the agricultural community’s standard of living, the unit of observation is usually the disposable income of farm households, based on the assumption that households comprise individuals who share income and decisions on expenditure. One example is the seminal paper by Gardner (1992), which concluded that the farm problem of low income has disappeared. Since then, evidence continues to show that farm households have similar economic well-being to non-farm households (Mishra et al. 2002; Organisation for Economic Co-operation and Development 2003; Katchova 2008; Hill and Bradley 2015). However, this is primarily because farm households, and particularly the farmer’s spouse, are devoting more time to off-farm work (Hanson and Spitze 1976; Ahearn, Johnson, and Strickland 1985; Findeis and Reddy 1987; Gardner 1992; Mishra et al. 2002; Ahearn, El-Osta, and Dewbre 2006; El-Ostra, Mishra, and Morehart 2008; Hill and Bradley 2015). Studies show that part-time farming is important for farm survival in Europe (Glauben et al. 2006; Breustedt and Glauben 2007).3 High off-farm income may also be efficient for risk spreading, and individuals may hold multiple jobs as a way to obtain new skills (Panos, Pouliakas, and Zangelidis 2014). However, studies that measure income at the household level may overlook problems at the individual level. Such studies implicitly presume that household incomes are shared equally between spouses, a presumption that is highly sensitive from a gender perspective and not necessarily true.4 When the off-farm labor supply of a female spouse increases, it may have both positive and negative effects on that woman’s family and labor market situation. Pluriactivity increases the economic autonomy of women (Blanc and MacKinnon 1990), but it is uncertain whether it increases women’s bargaining power and involvement in farm decision making (Gasson and Winter 1992; Evans and Ilbery 1996). In addition, pluriactivity may not necessarily decrease women’s farm work (Evans and Ilbery 1996). Thus, a double burden (farm and family) may become a triple burden (i.e., farm, family, and career; Blanc and MacKinnon 1990). In addition, applying a household perspective to single-person farm households is problematic; for this group of farmers, the relevant comparison is other single-person households. An alternative is to investigate individual earnings. The “fair standard of living” objective in Article 39 of the Treaty on the Functioning of the European Union is specified as: “ … in particular by increasing the individual earnings of persons engaged in agriculture.” Individual earnings are usually regarded as measuring the return to human capital (Becker 1964; Willis 1986). When comparing income between men and women, between races or ethnic groups, or between sectors, the level of comparison is typically the individual. However, the individual earnings of farmers are rarely investigated. A recent study by the European Parliament states that: “evidence points to farmers NOT being a particularly low-income sector of society in most Member States judged on the basis of their household disposable incomes,” (Hill and Bradley 2015), but this statement mainly refers to the standard of living of farm households, not the returns to farming (skills). Thus, the novel contribution of our study is to analyze the income of farm households from two dimensions: (a) the household-individual dimension and (b) the disposable income-earnings dimension. We use a broad definition of farm households (persons related by family ties sharing the same dwelling, where at least one member has some income from farming) since, with a narrow definition, a longitudinal analysis would eventually lose those individuals who are most successful in gaining off-farm income (Hill 2012).5 It is relevant to evaluate the returns to farming and not only the standard living of farmers, that is, to explore the disposable income-earnings dimension. Low return to farming may indicate inefficiency and provide an incentive to reallocate resources to sectors with higher levels of remuneration. However, it could be argued that farming is more involved in the production of public goods than other sectors (OECD 2001; Brunstad, Gaasland, and Vårdal 2005; Huylenbroeck et al. 2007). Due to non-rivalry in consumption, it is difficult to obtain compensation through the market for such goods, and therefore low earnings in farming could be the result of market failure rather than inefficiency. In that case, relying on market forces alone would result in non-optimal resource allocation from a societal perspective. Several of the subsidies in the EU Rural Development Programs in the second pillar of the CAP are motivated by environmental concerns in a broad sense.6 Article 39 in the Treaty also suggests other concerns, such as food security (the objective of the CAP is to “assure the availability of supplies and to ensure that supplies reach consumers at reasonable prices”). Accordingly, information on the returns on farming is of interest to policy makers. However, recent changes in farm household income in the EU are poorly documented because there has been no system in place for recording farm household income statistics in the EU since 2002 (Hill 2012; Hill and Bradley 2015). Existing income data in the EU are measured at the national or farm level (as farm family income), meaning that data on off-farm income are not collected. This means that the standard of living of farm households is underreported. Hence, in most Member States detailed longitudinal data are available for evaluating the “production objective” in the Treaty, but the “standard of living objective” is difficult to evaluate. The European Court of Auditors (2016) has pointed out that the “statistical data used to analyse farmers’ incomes has significant limitations.” This study is probably the most comprehensive analysis to date of farm household income for an EU Member State. We use longitudinal register data to evaluate changes in earnings of Swedish farm households in the period 1997–2012. Individual data allow us to divide household earnings into farm and non-farm earnings and, using the Multigenerational register, we allocate household earnings between family members. Previous Research and Measurement of Farm Household Income In the United States, 80% of total household income is off-farm income, more than 50% of farm operators work off-farm, and 60% of operators of small farms (<5 ha UAA) work less than one-quarter of their time on-farm (El-Ostra, Mishra, and Morehart 2008).7 Furthermore, a substantial proportion of household income is provided by farm operators’ family members, with farm spouses devoting more time to off-farm work over time (Hanson and Spitze 1976; Mishra et al. 2002). Mishra et al. (2002) report that, in the United States, almost half of farm spouses work off-farm, and that the proportion increased by 65% between 1969 and 1999. In the EU, around one-third of farm operators have off-farm income and about 50% of household income derives from off-farm sources. As in the United States, the off-farm proportion seems to be increasing over time (OECD 2003; Barthomeuf 2008; Hill and Bradley 2015).8 In Sweden, 50% of farm operators had off-farm work in 2005 (Barthomeuf 2008). A Swedish survey using firm-level data found that 30% of farms engaged in diversification, although income from paid employment outside the farm was not included (Swedish Board of Agriculture 2007). A Finnish study using tax register data for 1986 showed that household income per farm family unit amounted to only 71% of the household income per family unit of industrial workers, although this increased to 95% when disposable income per family unit was compared (Puurunen 1990). Measuring Earnings As noted, since 2002 there has been no system for measuring farm household income in the EU. The Eurostat’s Income of the Agricultural Households Sector (IAHAS) statistics started in the late 1980s and terminated in 2002, but the quality of the IAHS data was a problem much earlier (Hill 2012; Hill and Bradley 2015).9 The EU currently uses two systems to measure returns from farming: Economic Accounts for Agriculture (EAA) at the Member State level (see Eurostat 2002, 2010), and the Farm Accountancy Data Network (FADN) at the farm level. In FADN, farm family income is measured per farm or per family work unit on the farm. However, the income indicator in both the EAA and FADN excludes off-farm income.10 The 2005 Farm Structure Survey (Barthomeuf 2008) collects information on other gainful activities (OGA) for farmers in the EU, but spouses are not included. The survey asks if the farmer receives income from diversification activities such as tourism and refining/processing, and income from paid employment outside the farm. However, the Farm Structure Survey collects no actual data on incomes. Note also that, unlike our definition of farm income, which is based on the total returns from a farm’s human capital (whether it originates from work on one’s own farm or from work on other farms), the 2005 Farm Structure Survey classifies paid farm work outside the farm as off-farm income. The Swedish Board of Agriculture (2007) merges micro-level register data with a survey of off-farm income. From these data, they identify off-farm income at farm level (i.e., farm diversification), but exclude paid employment outside the farm. Instead, data on total individual returns from business, work, and capital are collected, but these total returns cannot be separated into remuneration from farm and off-farm activities. Moreover, the data are for a single year, 2005, and farm household earnings are not compared against population averages. To measure the standard of living of farm households, the preferred income measure is net disposable income, which was collected in the IAHS (Hill 2012). However, when examining differences in labor market returns between farm household members and the population at large, earnings are preferred. Inequality in disposable income is caused by labor market factors, but also by factors pertaining to the tax system, capital returns, and non-labor market transfers (i.e., child support, housing benefit, and social security benefits). Earnings include income from business and work, and labor market transfers. Labor market transfers are conditional on labor market participation and provide income support while the individual is absent from work, that is, they include parental leave payments, sickness payments, and unemployment benefits. For the present study, we generated data on all these variables by merging information from several Swedish registers. Thus, with unusually good data, this study presents a broad socioeconomic overview of Swedish farmers. Data and Descriptive Statistics For this study, we used the Longitudinal Integration Database for Health Insurance and Labour Market Studies (LISA), which includes a broad range of indicators on demographics, labor market status, income, and education for the entire Swedish population (aged 16 and older). From LISA, we obtained a full sample of (a) individuals receiving income from business or work in agriculture 1997–2012, and (b) their children and spouses.11 In this study we restrict the sample to farm households that contained the business operator, that is, farm employees and their family members were excluded. The analysis was carried out at the individual and the household levels, with changes followed over time. By using the household indicator in LISA and linking LISA to the Multigenerational Register, we acquired detailed information on household composition. We apply a “broad” definition of farm households as households where at least one member earns a farm income from business or work. The farm operator is defined as the household member with the highest mean farm earnings 1997–2012 even when some other household member, for example, a child or a spouse, is the listed farm operator. With the decoupling reform in 2005, a number of people who owned some agricultural land became eligible for single farm payments from Pillar I of the CAP and were thus registered as having agricultural earnings. In 2004 (when the first applications for the new payments were made), the sample of individuals with agricultural earnings increased by around 30%. However, from 2007, four hectares of land was set as the minimum required for receiving the single farm payment, which decreased the number of “farmers” substantially, again by around 30%. Households that primarily receive income from CAP subsidies, and which may be regarded as “pure subsidy farmers,” affect the results for the years around the decoupling reform, the main effect being that mean farm earnings became substantially lower in 2004–2006 and off-farm earnings became much higher. We therefore remove this group of farmers in the most efficient way available by restricting the sample to only include farmers receiving farm earnings for at least five years in the period 1997–2012 (when a three- or four-year restriction on farm earnings is applied, the “bias” in the results remain). Nevertheless, these “subsidy farmers” still affect the results for off-farm income (see figure 10). Earnings from forestry and extension services are included in farm earnings. For Sweden, where forestry and farming are closely linked, this is reasonable. As the data thus only indicate whether the main income is from farming or forestry, we were unable to separate income from these two sources unless the farmer had two different firms, one for farming and one for forestry, which is uncommon. However, since we restrict the sample to farm operator households, we know that some part of the income is from farming. About 14.7% of operators have earnings mainly from forestry and about 2.9% have earnings mainly from extension services. Most people in Sweden retire at the legal retirement age of 65. As the intention in this study is to compare earnings for the farming population and the total Swedish population, the relevant group for comparison is the working population (aged 20–64), although farmers, being self-employed, may choose to retire later. Thus, a household is removed from the sample when the operator reached the legal retirement age of 65 or when the household stopped farming, that is, household farm income became zero. If another operator took over the farm (e.g., a child), it is classified as a new household. We have 42,898 households and farm operators in our sample. The sample also contains 34,396 partners and 48,063 children (16 years and older), 91% of the operators are male and 68% have a partner.12 The sample decreases substantially over time: it is 37% smaller in 2012 than in 1997. The main reason is retirement: 67% (25% out of 37%) of the decrease is due to retirement. On average, the operators have farm earnings for 10.5 out of 16 years. Importantly, the results in this study are basically the same whether we use a balanced sample (farmers who report farm income for all years) or an unbalanced sample (where there is inflow and outflow of farmers). Table 1 presents descriptive statistics for male and female farm operators with/without a spouse. Table 1 Descriptive Statistics on the Sample of Farm Operators (Averages over All Years) Male operator with a partner Female operator with a partner Male operator without a partner Female operator without a partner N = 25,868 N = 4,208 N = 12,651 N = 1,785 Mean Standard deviation Mean Standard deviation Mean Standard deviation Mean Standard deviation Years with positive farm earnings 13.52 0.01 12.22 0.02 13.05 0.01 11.6 0.03 Individual earnings (thousands) 213.13 0.26 175.59 0.6 180.28 0.41 180.33 1.03 Earnings from farming (thousands) 170.95 0.24 127.58 0.53 147.74 0.39 123.46 0.93 Off-farm earnings (thousands) 42.18 0.19 48.01 0.47 32.54 0.24 56.87 0.88 Household total earnings (thousands) 439.46 0.44 426.68 1.3 223.4 0.55 215.57 1.46 Household disp. earnings (thousands) 423.97 0.91 434.44 2.62 256.76 1.52 236.63 3.53 Age 48.72 0.02 48.82 0.05 44.71 0.03 42.63 0.11 Children 1.33 0.00 1.12 0.01 0.13 0.00 0.4 0.01 Years of schooling 10.74 0.00 11.29 0.01 10.64 0.01 11.71 0.02 Northern Sweden 13.3% 17.3% 14.1% 17.9% Forrest districts in Sweden 33.4% 33, 0% 33.9% 30.2% Plains districts in Sweden 53.3% 49.7% 52, 0% 51.9% Proportion with off-farm earnings 50.8% 53.8% 36.2% 52.8% Proportion below the poverty rate 13.6% 7.1% 20.6% 25.4% Crop 19.8% 12.1% 19.3% 11.3% Milk 18, 0% 10.1% 18.2% 18.7% Meat 15.5% 12.3% 15.7% 14.3% Mix 29, 0% 47.8% 28.3% 32, 0% Forestry 14.7% 9.9% 15.2% 10.8% Extension services 2.9% 7.8% 3.3% 12.9% Small-sized farms 30.9% 34.6% 38.8% 32.8% Mid-sized farms 34.5% 30.1% 32.3% 23.1% Large-sized farms 34.6% 35.2% 28.9% 44.1% Male operator with a partner Female operator with a partner Male operator without a partner Female operator without a partner N = 25,868 N = 4,208 N = 12,651 N = 1,785 Mean Standard deviation Mean Standard deviation Mean Standard deviation Mean Standard deviation Years with positive farm earnings 13.52 0.01 12.22 0.02 13.05 0.01 11.6 0.03 Individual earnings (thousands) 213.13 0.26 175.59 0.6 180.28 0.41 180.33 1.03 Earnings from farming (thousands) 170.95 0.24 127.58 0.53 147.74 0.39 123.46 0.93 Off-farm earnings (thousands) 42.18 0.19 48.01 0.47 32.54 0.24 56.87 0.88 Household total earnings (thousands) 439.46 0.44 426.68 1.3 223.4 0.55 215.57 1.46 Household disp. earnings (thousands) 423.97 0.91 434.44 2.62 256.76 1.52 236.63 3.53 Age 48.72 0.02 48.82 0.05 44.71 0.03 42.63 0.11 Children 1.33 0.00 1.12 0.01 0.13 0.00 0.4 0.01 Years of schooling 10.74 0.00 11.29 0.01 10.64 0.01 11.71 0.02 Northern Sweden 13.3% 17.3% 14.1% 17.9% Forrest districts in Sweden 33.4% 33, 0% 33.9% 30.2% Plains districts in Sweden 53.3% 49.7% 52, 0% 51.9% Proportion with off-farm earnings 50.8% 53.8% 36.2% 52.8% Proportion below the poverty rate 13.6% 7.1% 20.6% 25.4% Crop 19.8% 12.1% 19.3% 11.3% Milk 18, 0% 10.1% 18.2% 18.7% Meat 15.5% 12.3% 15.7% 14.3% Mix 29, 0% 47.8% 28.3% 32, 0% Forestry 14.7% 9.9% 15.2% 10.8% Extension services 2.9% 7.8% 3.3% 12.9% Small-sized farms 30.9% 34.6% 38.8% 32.8% Mid-sized farms 34.5% 30.1% 32.3% 23.1% Large-sized farms 34.6% 35.2% 28.9% 44.1% Note: The sum of the different types of operators is smaller than the total number of operators because an individual can change civil status over time. Table 1 Descriptive Statistics on the Sample of Farm Operators (Averages over All Years) Male operator with a partner Female operator with a partner Male operator without a partner Female operator without a partner N = 25,868 N = 4,208 N = 12,651 N = 1,785 Mean Standard deviation Mean Standard deviation Mean Standard deviation Mean Standard deviation Years with positive farm earnings 13.52 0.01 12.22 0.02 13.05 0.01 11.6 0.03 Individual earnings (thousands) 213.13 0.26 175.59 0.6 180.28 0.41 180.33 1.03 Earnings from farming (thousands) 170.95 0.24 127.58 0.53 147.74 0.39 123.46 0.93 Off-farm earnings (thousands) 42.18 0.19 48.01 0.47 32.54 0.24 56.87 0.88 Household total earnings (thousands) 439.46 0.44 426.68 1.3 223.4 0.55 215.57 1.46 Household disp. earnings (thousands) 423.97 0.91 434.44 2.62 256.76 1.52 236.63 3.53 Age 48.72 0.02 48.82 0.05 44.71 0.03 42.63 0.11 Children 1.33 0.00 1.12 0.01 0.13 0.00 0.4 0.01 Years of schooling 10.74 0.00 11.29 0.01 10.64 0.01 11.71 0.02 Northern Sweden 13.3% 17.3% 14.1% 17.9% Forrest districts in Sweden 33.4% 33, 0% 33.9% 30.2% Plains districts in Sweden 53.3% 49.7% 52, 0% 51.9% Proportion with off-farm earnings 50.8% 53.8% 36.2% 52.8% Proportion below the poverty rate 13.6% 7.1% 20.6% 25.4% Crop 19.8% 12.1% 19.3% 11.3% Milk 18, 0% 10.1% 18.2% 18.7% Meat 15.5% 12.3% 15.7% 14.3% Mix 29, 0% 47.8% 28.3% 32, 0% Forestry 14.7% 9.9% 15.2% 10.8% Extension services 2.9% 7.8% 3.3% 12.9% Small-sized farms 30.9% 34.6% 38.8% 32.8% Mid-sized farms 34.5% 30.1% 32.3% 23.1% Large-sized farms 34.6% 35.2% 28.9% 44.1% Male operator with a partner Female operator with a partner Male operator without a partner Female operator without a partner N = 25,868 N = 4,208 N = 12,651 N = 1,785 Mean Standard deviation Mean Standard deviation Mean Standard deviation Mean Standard deviation Years with positive farm earnings 13.52 0.01 12.22 0.02 13.05 0.01 11.6 0.03 Individual earnings (thousands) 213.13 0.26 175.59 0.6 180.28 0.41 180.33 1.03 Earnings from farming (thousands) 170.95 0.24 127.58 0.53 147.74 0.39 123.46 0.93 Off-farm earnings (thousands) 42.18 0.19 48.01 0.47 32.54 0.24 56.87 0.88 Household total earnings (thousands) 439.46 0.44 426.68 1.3 223.4 0.55 215.57 1.46 Household disp. earnings (thousands) 423.97 0.91 434.44 2.62 256.76 1.52 236.63 3.53 Age 48.72 0.02 48.82 0.05 44.71 0.03 42.63 0.11 Children 1.33 0.00 1.12 0.01 0.13 0.00 0.4 0.01 Years of schooling 10.74 0.00 11.29 0.01 10.64 0.01 11.71 0.02 Northern Sweden 13.3% 17.3% 14.1% 17.9% Forrest districts in Sweden 33.4% 33, 0% 33.9% 30.2% Plains districts in Sweden 53.3% 49.7% 52, 0% 51.9% Proportion with off-farm earnings 50.8% 53.8% 36.2% 52.8% Proportion below the poverty rate 13.6% 7.1% 20.6% 25.4% Crop 19.8% 12.1% 19.3% 11.3% Milk 18, 0% 10.1% 18.2% 18.7% Meat 15.5% 12.3% 15.7% 14.3% Mix 29, 0% 47.8% 28.3% 32, 0% Forestry 14.7% 9.9% 15.2% 10.8% Extension services 2.9% 7.8% 3.3% 12.9% Small-sized farms 30.9% 34.6% 38.8% 32.8% Mid-sized farms 34.5% 30.1% 32.3% 23.1% Large-sized farms 34.6% 35.2% 28.9% 44.1% Note: The sum of the different types of operators is smaller than the total number of operators because an individual can change civil status over time. From Statistics Sweden (SCB), we also have individual and household earnings at the aggregate level for the Swedish working age population (20–64).13 In the analysis, we define households as families with at least two adults, that is, single adult households are not analyzed as households in this study, and this definition is used for both the farm population and the total population.14 Single farmers, that is, operators without a spouse, are analyzed separately. Farm households are not excluded from total household statistics but, because they constitute only around 2.4% of the total number of households, their contribution to the total population statistics is minor. Structure of the Analysis and Method for Comparing Earnings The difference in age composition between farmers and the total population has to be accounted for when comparing the groups. Our approach for calculating the mean individual and household earnings of farmers is, therefore, to estimate (with a pooled OLS) the earnings on 16 time dummies, βt, and 44 age dummies, δi: Earningsit=βt+δi. By excluding a constant and using the age 37 as the reference category, β shows the yearly average individual earnings for the age group 37. The reference age is set at the average age of individuals in the total working-age population. However, when comparing households (and individuals living in households with at least two adults), the reference age is set at 47, that is, the average age of the oldest working-age adult in farm households is higher than the average age in the total working-age population.15 This model removes an age bias in the results. All earnings variables are measured in 2012 prices. The comparison of earnings— βt to the average earnings in the population—has an individual, or household, perspective, and not a farm perspective per se. This means that the main findings were split according to type of household, and not according to type of farm business or farm size. Thus, we divide the households into those with a male and those with a female operator, and households where the operator has a partner and households where the operator is single.16 In statistics on the total population, the households are not separated according to gender.17 Our particular interests are on the comparison of household earnings and disposable income and household off-farm income. For 70% of the sample, we are able to match the data on income and earnings to farm-level data, which allows us to explore the impact of farm size on changes in earnings. The next step is to use the individuals’ Standard Industrial Classification (SNI) codes to classify the households into different types of households: (a) crop, (b) milk, (c) beef, (d) mix, (e) forestry, and (f) extension services. Inequality is analyzed by calculating the Gini coefficient and the at-risk-of-poverty rate. As this study is mainly descriptive, the findings may be affected by selection bias, that is, that farmers possess different skills than the general population. Nordin, Blomquist, and Waldo (2016) found that there is positive selection into farming in Sweden. In that study, a sibling approach was used to compare the earnings of farm siblings (sharing the same farm background) following different career paths, that is, one sibling chooses farming and the other chooses a non-farming occupation. A sibling fixed effect model that removes the common background characteristics of siblings was found to give a larger penalty to farming than an OLS model, indicating that positive selection into farming is likely. Based on those findings, farmers’ earnings are underestimated rather than overestimated in a returns to farming perspective. It is also important to point out that in this study we intentionally used a simple model and chose not to control for other factors (than age). From a standard of living perspective this is reasonable (a low standard of living is still low even if it can be explained by, e.g., education), but from a return to farming perspective one could consider controlling for education, for example. However, since education is an endogenous variable (choosing farming and education level is a joint decision), it can be questioned whether controlling for education is the correct choice. For example, a low return to farming may be caused by low education level of the farmer, or the education level may be low because of the choice to become a farmer. Besides, we also prefer having the same specification in all comparisons. Results Household Earnings and Disposable Income Average household earnings for farm households and for the total household population in the period 1998–2012 are shown in figure 1a) and b). The left y-axis shows household earnings and the right y-axis shows the gap in household earnings between farm households and the total household population. For the total household population, earnings increased by 33% between 1998 and 2012, while for farm households earnings increased by 56% and 53% for households with a male and a female operator, respectively. Owing to the larger increase in earnings for farm households, the gap between these and the total household population decreased by around 12 percentage points between 1998 and 2012. Though difficult to deduce from the figures, female operator households had around 3% higher earnings than male operator households; later we show that this is due to high off-farm earnings for the female operator’s spouse. Figure 1 View largeDownload slide Household earnings of farm households with a male operator or a female operator, compared with earnings for the total population of households, 1998–2012 Figure 1 View largeDownload slide Household earnings of farm households with a male operator or a female operator, compared with earnings for the total population of households, 1998–2012 Figure 2a) and b) show the corresponding values for disposable income. Disposable incomes were similar for farm households and other households in the study period, with the gap changing from somewhat negative (around 2–5 percentage points lower in farming) to somewhat positive (around 2–5 percentage points higher). Thus, while there was a substantial gap in earnings between farm households and other households, even in 2012, the gap in household disposable income was small and in fact disposable income was higher for farm households in 2006–2012. In other words, for the total population disposable income was about 20% lower than earnings, but for farm households it was about 5% lower in the beginning of the study period and similar by the end of the period. Figure 2 View largeDownload slide Household disposable income of farm households with a male operator or a female operator compared with that of the total population of households, 1997–2012 Figure 2 View largeDownload slide Household disposable income of farm households with a male operator or a female operator compared with that of the total population of households, 1997–2012 What could explain the relatively high disposable income of farm households? Compared to earnings, disposable income includes capital returns and non-work related transfers, and deducts taxes. Since farm households possess significant wealth (Organisation for Economic Co-operation and Development 2003), high capital returns might be an explanation. We are able to compare differences in capital returns for the year 2002.18 Farm household net capital returns in that year were around 40,000 Swedish Krona (SEK), which was about SEK 5,000 higher than net capital returns for the total population.19 However, the gap in household earnings between farm households and the total household population in that year was around SEK 140,000 (see figure 1a) and b)). Hence, higher capital returns for farm households can only explain about 3.5% (SEK 5,000 of SEK 140,000) of the relatively high disposable income of farm households (in relation to earnings for farm households). Moreover, non-work related transfers (mainly child allowance, but also social benefits and housing benefits) are similar for farm households and other households, making up about 4.9% (Swedish Board of Agriculture 2007) and 4.5% (Government Proposition 2004) of disposable income for farm households and the total population, respectively.20 Our data contains no information on taxes, so we were unable to investigate the taxes paid by farm households. However, the only remaining explanation for the relatively high disposable income of farm households is substantially lower taxes. Farmers can be expected to pay lower taxes, as they can take out part of the farm income as income from business activities, which is taxed at a lower rate than income from work, but it is surprising that this had such a large impact. Thus, we conclude that the taxation and tax deduction system gives farmers a standard of living that is comparable to that of the rest of the population. Does Farm Size Matter? For 70% of the farmers in our sample, we were able to match the individual data with farm-level data from the Swedish Board of Agriculture.21 The farm level data are available for the years 1999, 2003, 2005, 2007, and 2010. Based on hectares of arable land and number of animal units, we divide the farms into large-, mid- and small-sized farms, with each group containing one-third of all farms. Figure 3 shows that large- and mid-sized farms had, respectively, about 17% and 4% higher household earnings than the small-sized farms. The absolute increase in household earnings between 1999 and 2010 was almost the same for large- and small-sized farms, but the relative increase was larger for small farms. For mid-sized farms, both the absolute and the relative increase in household earnings were smaller than for large-sized farms. All these findings are for households with a male operator, but the results were similar for the total sample of farmers. Moreover, separate analyses for different production areas in Sweden (northern Sweden, woodlands and plains) show similar results to the aggregate sample (not reported). Figure 3 View largeDownload slide Farms household earnings of small-, middle- and large-sized farms. 1999–2010 Figure 3 View largeDownload slide Farms household earnings of small-, middle- and large-sized farms. 1999–2010 Individual Earnings We divide the sample into male and female operators, with or without a spouse (i.e., whether the operator lived in a household including two adults or was single). For the total population, the sample is not divided conditional on having a spouse. To gain a complete picture of total household earnings, children’s earnings should also have been reported separately. However, to focus the analysis we merely report that they earned around 8% of the total household earnings and that most of these earnings were from off-farm sources. Figure 4a) shows that earnings for male operators with a partner increased from 68% to 85% of the male population average. For male operators without a spouse, the relative earnings increased from 61% to 77% of the male population average. The earnings of single male operators were around SEK 27,000 lower than the earnings of male operators with a spouse; in absolute terms the difference increased over time, but in relative terms it was almost the same in 1997 as in 2012. Higher individual earnings for operators with a spouse may be because of selection (if the same factors determine earnings and success on the marriage market) or labor division within households (if the wife allocates more time to household work, the husband may allocate more time to farm or off-farm work). Figure 4 View largeDownload slide Individual earnings of the total population and of farms with a male operator or a female operator, with and without a partner, 1997–2012 Figure 4 View largeDownload slide Individual earnings of the total population and of farms with a male operator or a female operator, with and without a partner, 1997–2012 For female operators (figure 4b)), earnings increased from about 77% to 93% of the female population average. Thus, the gap to the population average was smaller for women and there was no major difference in earnings between female operators with or without a partner. So far, we have not divided the earnings into farm and non-farm earnings. Figure 5a) and b) show household earnings separated into the operator’s farm and non-farm earnings, and the spouse’s farm and non-farm earnings for households with a male or female operator with a spouse. Figure 5 View largeDownload slide Earnings from the farm and off-farm earnings of farms with a male operator or a female operator and their spouse, 1997–2012 Figure 5 View largeDownload slide Earnings from the farm and off-farm earnings of farms with a male operator or a female operator and their spouse, 1997–2012 First, the relatively large increase in earnings for farm households in the study period (see figure 1a) and b)) can be explained by a large increase in the operator’s farm earnings and the partner’s off-farm earnings. The annual increase in farm earnings was 3.1% and 2.85% for male and female operators, respectively. There was an even higher annual increase in the spouse’s off-farm earnings, of around 4%. For the total population, the annual increase in earnings was 1.74% and 2.20% for men and women, respectively. Moreover, in absolute terms, male operators increased their farm earnings by the same amount as spouses increased their off-farm earnings, by around SEK 80,000 (figure 5a)). This differs from female operator households: figure 5b) shows a much larger absolute increase in male spouses’ off-farm earnings (around SEK 110,000) than female operators’ farm earnings (around SEK 55,000). The most important findings from figure 5a) and b) are the low off-farm earnings for operators and the low farm earnings for spouses. This indicates a clear division of labor within households, where operators work mainly on the farm and spouses mainly off-farm. This was particularly clear in male operator households, where spouses’ farm engagement was very low. Moreover, while operators’ off-farm earnings increased (although from a low level) in the study period, spouses’ farm engagement hardly increased at all. These latter findings are central for understanding the household earnings of Swedish farmers, that is, the low off-farm earnings for operators and the low farm earnings for spouses, indicating that the division of labor in Swedish farm households is substantial. We also examine whether the results varied with farm size (supplementary online appendix figures A2–A4) and found that the division of labor was similar for small-, med- and large-sized farms.22 The main difference was that the operator’s and the spouse’s farm earnings increased and their off-farm earnings decreased with farm size. This meant that for small-sized farms the female spouse’s off-farm earnings were higher than the male operator’s farm earnings—actually, up until 2003 she had higher total earnings than her husband, as well. Operator’s Off-farm Labor Market Supply Next, we examine the variation in off-farm earnings and the relationship between off-farm earnings and total earnings for male operators (results for female operators are available upon request). Figure 6 shows household and individual earnings for male operators (with partner) with different degrees of engagement in farming. The x-axis shows farm earnings as a share of total earnings, that is, 100% means that the operator lacks off-farm earnings. The bars (measured at the right-hand y-axis) show the share of farmers in each “engagement group.” Figure 6 View largeDownload slide Relationship between male operator (with a partner) households and individual earnings and their degree of engagement in farming Figure 6 View largeDownload slide Relationship between male operator (with a partner) households and individual earnings and their degree of engagement in farming Almost 50% of the operators had zero off-farm earnings and 20% had less than 10% off-farm earnings. Relatively balanced engagement (i.e., 30% to 70% earnings from farming) was rare (only 7% of the operators). For male operators without a partner (not shown), 64% lacked off-farm earnings. Moreover, as figure 6 shows, both household and operator earnings decreased with increasing engagement in farming, so operators with 100% farm earnings had around two-thirds of the earnings of operators with 1% to 30% farm earnings.23 Over time, the difference decreased somewhat (not reported), but in 2012 operators with 100% farm earnings still had 30% lower earnings than those with a small engagement in farming. At the household level, earnings were around 20% to 25% lower for households with operators with 100% farm earnings than for operators with low farm earnings (1% to 30%). Spouses’ Labor Market Supply Next, we investigate the female spouses’ labor market supply. We find that around 65% of spouses had only off-farm earnings and that the share increased by more than 7 percentage points from 1997 to 2012 (figure 7). Another 5% to 10% were outside the labor force and had zero earnings; about half of these had retired or retired early. Figure 7 View largeDownload slide Labor supply of female farm spouses, 1997–2012 Figure 7 View largeDownload slide Labor supply of female farm spouses, 1997–2012 The likely explanation for the decreasing farm income for spouses is investment in higher education. Figure 8a) shows that the female spouses were more likely to invest in higher education than male operators, and that the within-household gender gap in higher education increased substantially between 1998 and 2012: the gap was 13 percentage points in 1997 and has almost doubled, to 22 percentage points, in 2012. In comparison, for the total population the gender gap in higher education was about 10 percentage points in 2012. In addition, female spouses had 1.5 more years of schooling than male operators. Since investment in education generally precedes matching in the marriage market, higher-educated spouses probably chose a career path that made dual work less likely, as higher education increased the probability of having zero farm income by 25% (not reported). Regarding the female spouses’ career choice, more than 50% of them worked as teachers or in healthcare. Figure 8b) shows the corresponding education statistics for households with a female operator. Here, the gender gap in higher education was small, only 2.5 percentage points in 2012. Figure 8 View largeDownload slide Proportion of operators and spouses with higher education (>12 years of schooling) on farms with a male operator or a female operator, 1997–2012 Figure 8 View largeDownload slide Proportion of operators and spouses with higher education (>12 years of schooling) on farms with a male operator or a female operator, 1997–2012 Figure 9 View largeDownload slide Farm earning of operators in different farming branches. 1997-2012 Figure 9 View largeDownload slide Farm earning of operators in different farming branches. 1997-2012 Income in Different Branches Farm and off-farm earnings from different types of farming are shown in figures 9 and 10.24 As can be seen from figure 9, farmers with income mainly originating from forestry had the highest farm earnings. The lowest farm earnings were found for meat farmers. Farmers with crop and mixed farms also had relatively low farm earnings. The largest variation in earnings was found for milk farmers; between 1997–2006 their annual increase in earnings was 5.0%, but in 2006–2012 it was only 0.5%. Farmers engaged in extension services showed the lowest annual increase in earnings (1.4%). The high earnings of milk farmers had a negative impact on off-farm earnings, with milk farmers having particularly low off-farm earnings (figure 10). The highest off-farm earnings were found for farmers engaged in crop production or forestry. Note that in 2004–2006, there were still some “subsidy farmers” in the data, and these increased off-farm earnings for farmers in general in 2004–2006. Figure 10 View largeDownload slide Off-farm earnings of operators in different farming branches. 1997-2012 Figure 10 View largeDownload slide Off-farm earnings of operators in different farming branches. 1997-2012 Inequality in Farming Income inequality in farming has fallen in the United States (Gardner 1992). By 1987, the incidence of poverty was for the first time lower among farm households than among non-farm households (U.S. Department of Commerce 1990). Corresponding inequality and poverty data are not available for the EU (Hill 2012). Thus, we examine inequality in two dimensions: the Gini coefficient and the at-risk-of-poverty rate. Usually, the Gini coefficient is calculated as household equivalized disposable income, which is obtained by dividing household disposable income by the number of “equivalent adults,” using a standard (equivalence) scale.25 Figure 11 shows the Gini coefficient, in household equivalized disposable income, for farm households and the total population. For farm households, it also shows the Gini coefficient in household equivalized earnings. As in most other European countries (Roine and Waldenström 2015), inequality in the total population has increased in Sweden: in 2006–2012, the Gini coefficient was about 7% higher than around the turn of the millennium.26 An increase in the earned income tax credit in 2006 is the main explanation for this change. For farm households, there was no visible trend in the Gini coefficient for equivalized disposable income, which fluctuated around 0.33. Thus, in 2007–2011 overall inequality in Sweden was around the same as inequality within the farm household population. Figure 11 View largeDownload slide Gini coefficient for farm households and the total population,1997-2012 Figure 11 View largeDownload slide Gini coefficient for farm households and the total population,1997-2012 For household equivalized earnings, the Gini coefficient decreased for farm households. We had no data on the Gini coefficient in household equivalized earnings for the total population, but Bengtsson, Edin, and Holmlund (2014) report that it was stable over the period. The at-risk-of-poverty rate is the proportion of people with equivalized disposable income below a standard threshold, which is set at 60% of the national median equivalized disposable income. Since we use the yearly national median, we calculate the relative at-risk-of-poverty rate, not the absolute rate, which uses the national median for a base year. Figure 12 shows the at-risk-of-poverty rate for the total and the farm population. Before 2004, the at-risk-of-poverty rate for the total population is only available for all ages, but, as figure 12 shows, the rate was similar for the working population (here 18–64) and all ages after 2004, particularly for the period 2004–2007.27 Figure 12 View largeDownload slide At-risk-of-poverty rate for farm households and the total population,1997-2012 Figure 12 View largeDownload slide At-risk-of-poverty rate for farm households and the total population,1997-2012 There was also a clear difference in trends for the farm population and the total population; whereas poverty increased in Sweden in general, it decreased in the farm population (figure 12). These different trends resulted in the rate being almost the same in the total population as in the farm population in 2010–2012, around 13–14%. To test whether the decreasing poverty rate in farming was due to retirement of low-income farmers, that is, due to a cohort effect, we calculated the at-risk-of-poverty rate for the age group 20–55. We found that the decrease in the at-risk-of-poverty rate for this group was almost the same as in figure 12. Conclusions This study analyzes farmers’ income from many different dimensions. Previous studies have analyzed farm households’ disposable income, but by applying an individual perspective and comparing disposable incomes to earnings, we present novel information on a known topic. The difference in earnings between the farm population and the total household population decreased by around 12 percentage points in the study period (1998–2012), to around 15% in 2012. For farmers without a partner, earnings were particularly low. We have not controlled for number of hours worked because of a lack of data. However, existing survey data do not suggest that the number of hours worked in farming is smaller than in other sectors.28 However, farm households’ disposable income was similar to that of the total household population. This was not due to high capital returns or high non-work related transfers in farming, but rather to a favorable tax system. Thus, from a standard of living perspective farm households do well, but from a returns to farming perspective farming is still a low-paid occupation, which may reduce incentives to allocate resources to the sector. In the long run, this could mean that EU environmental objectives and food security objectives are not met. Nevertheless, the returns to farming increased substantially from the end of the 1990s (measured as operators’ farm earnings), but the large increase in farm household earnings was equally caused by higher off-farm earnings for their partners. The partners seemed to choose a career path that made dual work less likely and their relatively high level of education reduced the probability of them having farm income. Hence, there was a clear division of labor within households; the operators worked mainly on the farm and the spouses mainly off-farm. When examining farm earnings from a household perspective, we implicitly assumed that the farm is a “family farm.” Garner and de la O Campos (2014), who have reviewed 36 definitions of the term “family farm,” claim that the most defining character of a family farm is the reliance on family labor—both women’s and men’s. However, we found that few farms in Sweden rely on both spouses’ labor (at least 70% of the female spouses have no farm earnings) and that applying mainly a family perspective is not justified. To evaluate the broader economic situation of farm households, the individual careers of the operator and the spouse would need to be considered. If the ambition is to attract more women to the sector, more knowledge is needed about women’s situations, and this is not feasible without an individual perspective. Finally, whereas inequality in Sweden increased over the period, within-farming inequality either decreased (household equivalized earnings) or remained stable (household equivalized disposable income). The at-risk-of-poverty rate also decreased in the farming population. Supplementary Material Supplementary material is available online at Applied Economic Perspectives and Policy online. Footnotes 1 Article 39, paragraph 1b, Treaty on the Functioning of the European Union (former Treaty of Rome). Available at: http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:02016ME/TXT-20160901&from=EN (Treaty on the Functioning of the European Union) and: http://ec.europa.eu/archives/emu_history/documents/treaties/rometreaty2.pdf (Treaty of Rome). 2 Disposable income includes earnings from labor, land, and capital (regardless of the sector in which these earnings originate), plus subsidies, minus taxes. However, the value of own consumption of farm products is not accounted for (see FAO 2011 and Hill 2012). 3 This is rarely the case in other occupations: in the United States, only about 5% of the population have multiple job holdings and the number is declining (Hirsch, Husain, and Winters 2016). Kimmel and Conway (2001) claim that multiple job holdings are generally undesirable and that the main motive for taking a second job is economic hardship. 4 This relates to the ambiguities regarding the definition of a “household”. If it is defined as “all persons sharing the same dwelling”, it is unlikely that incomes are pooled and decisions on their use taken in agreement. The problem remains even if households are defined as “persons with family ties sharing the same dwelling,” (see FAO 2011; Hill 2012, and references therein). 5 A narrow definition would include only persons who share the same dwelling and for which the main source of income is farming (FAO 2011; Hill 2012). 6 See Regulation (EU) No. 1305/2013. Available at: http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32013R1305&from=SV. 7 However, two-thirds of U.S. farms are small farms (sales less than $40,000) and contribute only 10% of the total farm production, but earn more than 75% of the off-farm income (Ahearn, Johnson, and Strickland 1985). 8 There are differences in the definition of a farm between the United States and the EU that could presumably affect comparisons. However, our aim was not to compare the significance of off-farm income for the standard of living of farmers in the United States and the EU, but to determine whether off-farm income is significant in both regions, and whether failure to account for this when analyzing the standard of living of farmers leads to questionable results. 9 Hill and Bradley (2015) list a number of reasons why adequate farm household income statistics are lacking for the EU; a consistent feature is that such statistics may be politically sensitive and may show that farmers are in a relatively favorable income position. 10 Another potential source is the Income, Social Inclusion and Living Conditions survey (EU-SILC). available at: http://ec.europa.eu/eurostat/web/income-and-living-conditions. However, according to Hill (2012), there are problems with coverage when disaggregating according to sector of occupation, as there are too few observations on farm households for reliable analysis of income distribution issues. 11 The Swedish Standard Industrial Classification (SNI) code is used for classifying firms as agricultural businesses. 12 However, the total number of individuals in the sample is smaller than the sum of operators, partners, children, and other family members because an individual can start out as a child and later become an operator, or a partner in another household. 13 Data on average earnings at the individual level are downloadable from Statistics Sweden’s homepage, but data on earnings at the household level were ordered separately. 14 Non-married cohabiting couples without children in common are also defined as singles in Swedish register data and are therefore included as singles in our analysis. 15 Using the reference age of 37 for households also has no major impact on the results, but using a much older (than 47) or younger (than 37) reference age affects the results somewhat. 16 Article 3, paragraph 2, in the Treaty of Amsterdam states that in all its activities “the Community shall aim to eliminate inequalities and to promote the equality of women and men”, and analyzes male and female farmers’ earnings separately. Available at: http://www.europarl.europa.eu/topics/treaty/pdf/amst-en.pdf. 17 To divide the households on gender, we have to designate a head of family. However, in a gender-equal society such as Sweden with family tax splitting, designating a head of family is both difficult and outdated. 18 Data on individual capital returns for the farming population are available in LISA for the entire period, but for the total population aggregate capital returns data are only available for 2002. The capital returns data are collected from an attachment to Government Proposition (2004). To confirm that the capital returns for 2002 are representative for the entire period, we compare total capital taxes, data on which are available for the entire period, with the returns to capital for farm households. These show a similar pattern over the period (see figure A1 which shows, in comparison with 1997, the percentage change in total capital taxes for the total population and the returns to capital for farm households). 19 Net capital returns is the difference between the returns and losses from mainly interest, dividends and capital gains from stocks and real estate. Importantly, land rents are included in income from business, and not in capital returns. 20 Note: CAP subsidies are recorded as income. 21 To be able to match the individual data with the farm data, the individual”s main income has to be from agriculture. For the match sampled, the farm income was 18% higher than for the unmatched sample (i.e. the matched farmers had a somewhat higher farming commitment), but, over time, the matched and unmatched farmers earnings developed similarly. 22 This is only possible for male operators with a partner. There were too few female operators with a partner to divide those data on farm size. 23 These results may seem to contradict each other. In Figure 3 we showed that small farms had the lowest household earnings and here we show that farmers with low engagement had the highest (household) earnings. However, farm size and engagement level are only marginally correlated. Low engagement is mainly an indication of “hobby farming” and most small farms have relatively high engagement. 24 A few farms change branch over time (mainly from mixed to a specific branch), we use the mode branch. 25 SCB gives the following weights to the members of the household: 1.0 to the first adult; 0.51 to the partner; 0.6 to each subsequent person aged 19 and over; 0.52 to the first child aged under 19, and 0.42 to each subsequent child aged under 19. 26 The Gini coefficient for the total population is without an age restriction, but Bengtsson, Edin, and Holmlund (2014) show that the Gini coefficient is similar for the entire population and the working population. 27 The at-risk-of-poverty rate is taken from the Eurostat database. 28 Statistics Sweden’s data on number of hours worked in different sectors are generated from surveys where the respondent is asked how many hours she worked during the previous week. However, for agriculture, where activity varies with season, results may not be representative for average number of hours worked per week. Also, the data does not allow separating the agricultural sector from the forestry and fishing sectors. References Ahearn M. , El-Osta H. , Dewbre J . 2006 . The Impact of Coupled and Decoupled Government Subsidies on off-Farm Labor Participation of U.S. Farm Operators . American Journal of Agricultural Economics 88 : 393 – 408 . Google Scholar CrossRef Search ADS Ahearn M. , Johnson J. , Strickland R . 1985 . The Distribution of Income and Wealth of Farm Operator Households . American Journal of Agricultural Economics 67 : 1087 – 94 . Google Scholar CrossRef Search ADS Barthomeuf L.-T. 2008 . Other Gainful Activities, Pluriactivity and Farm Diversification in EU-27. Brussels: European Commission, Directorate General for Agriculture and Rural Development, G.2. Economic analysis of EU agriculture. Bengtsson N. , Edin P.-A. , Holmlund B . 2014 . Wages, employment and incomes – do the gaps increase in Sweden? Fiscal Policy Council Report 2014/1. Blanc M. , MacKinnon N . 1990 . Gender Relations and the Family Farm in Western Europe . Journal of Rural Studies 6 : 401 – 5 . Breustedt G. , Glauben T . 2007 . Driving Forces behind Exiting from Farming in Western Europe . Journal of Agricultural Economics 58 ( 1 ): 115 – 27 . Brunstad R. , Gaasland I. , Vårdal E . 2005 . Multifunctionality of Agriculture: An Inquiry into the Complementarity between Landscape Preservation and Food Security . European Review of Agricultural Economics 32 : 469 – 88 . Google Scholar CrossRef Search ADS El-Ostra H. , Mishra A. , Morehart M . 2008 . Off-Farm Labor Participation Decisions of Married Farm Couples and the Role of Government Payments . Review of Agricultural Economics 30 : 311 – 32 . Google Scholar CrossRef Search ADS European Court of Auditors . 2016 . Is the Commission’s System for Performance Measurement in Relation to Farmers’ Incomes Well Designed and Based on Sound Data? Special Report No 1/2016. Eurostat . 2002 . Income of the Agricultural Households Sector—2001 Report. Theme 5, Luxembourg. Eurostat . 2010 . Draft Version of European System of Accounts: ESA 2010 . Luxembourg . Evans N. , Ilbery B . 1996 . Exploring the Influence of Farm- Based Pluriactivity on Gender Relations in Capitalist Agriculture . Sociologia Ruralis 36 : 74 – 92 . Google Scholar CrossRef Search ADS FAO . 2011 . The Wye Group Handbook. Statistics on Rural Development and Agricultural Household Income , ed. Pizzoli E. United Nations . Findeis J. , Reddy V . 1987 . Decomposition of the Income Distribution of Farm Families . Northeastern Journal of Agricultural Economics 16 : 165 – 73 . Gardner B. 1992 . Changing Economic Perspective on the Farm Problem . Journal of Economic Literature 30 : 62 – 101 . Garner E. , de la O. Campos A.P . 2014 . Identifying the “Family Farm”. An Informal Discussion of the Concepts and Definitions. ESA WP No. 14-10. Rome: Agricultural Development Economics Division, Food and Agriculture Organization of the United Nations (FOA). Gasson R. , Winter M . 1992 . Gender Relations and Farm Household Pluriactivity . Journal of Rural Studies 8 : 387 – 97 . Google Scholar CrossRef Search ADS Glauben T. , Tietje H. , Weiss C . 2006 . Agriculture on the move: Exploring regional differences in farm exit rates in Western Germany . Review of Regional Research 26 : 103 – 18 . Google Scholar CrossRef Search ADS Government Proposition . 2004 . Fördelningspolitisk redogörelse (Bilaga 3). Proposition 2004/05: 1. Hanson R. , Spitze R . 1976 . An Economic Analysis of Off-farm Income in the Improvement of Illinois Farm Family Income. Agricultural Economic Research Report, Department of Agricultural Economics, University of Illinois (139): 43. Hill B. 2012 . Farm Incomes, Wealth and Agricultural Policy—Filling the CAP’s Core Information Gap . 4th edition . Oxfordshire : CAB International . Google Scholar CrossRef Search ADS Hill B. , Bradley D . 2015 . Comparisons of Farmers’ Incomes in the EU Member States. Directorate-General for Internal Policies, IP/B/AGRI/IC/2014-68. Policy Department B: Structural and Cohesion Policies Agriculture and Rural Development. Hirsch B. , Husain M. , Winters J . 2016 . The Puzzling Fixity of Multiple Job Holding across Regions and Labor Markets. IZA DP No. 9631. Huylenbroeck G. , Vandermeulen V. , Mettepenningen E. , Verspecht A . 2007 . Multifunctionality of Agriculture: A Review of Definitions, Evidence, and Instruments . Living Reviews in Landscape Research 1: 1 – 43 . Katchova A. 2008 . A Comparison of the Economic Well-Being of Farm and Nonfarm Households . American Journal of Agricultural Economics 90 : 733 – 47 . Google Scholar CrossRef Search ADS Mishra A. , El-Osta H. , Morehart M. , Johnson J. , Hopkins J . 2002 . Income, Wealth, and the Economic Well-Being of Farm Households. Washington DC: U.S. Department of Agriculture, Economic Research Service, Agricultural Economic Report No. 812. Nordin M. , Blomquist J. , Waldo S . 2016 . The Income Penalty of Farming and Fishing: Results from a Sibling Approach . European Review of Agricultural Economics 43 ( 3 ): 383 – 400 . Google Scholar CrossRef Search ADS Organisation for Economic Co-operation and Development . 2001 . Multifunctionality: Towards an Analytical Framework . Paris . Organisation for Economic Co-operation and Development . 2003 . Farm Household Income—Issues and Policy Responses . Paris . Panos G. , Pouliakas K. , Zangelidis A . 2014 . Multiple Job Holding, Skill Diversification, and Mobility . Industrial Relations 53 : 223 – 72 . Puurunen M. 1990 . A Comparative Study on Farmers' Income . Finland : Research Publications 62, Agricultural Economics Research Institute . Roine J. , Waldenström D . 2015 . Long-Run Trends in the Distribution of Income and Wealth, Chapter 7. In Handbook of Income Distribution, Volume 2A , eds Atkinson A. , Fourguignon F. , 469 – 592 . Amsterdam : Elsevier B.V . Swedish Board of Agriculture . 2007 . Other Gainful Activities on the Agricultural Holding and Income of the Agricultural Household. Report 2007: 3. © The Author(s) 2018. Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Applied Economic Perspectives and Policy Oxford University Press

Earnings and Disposable Income of Farmers in Sweden, 1997-2012

Loading next page...
 
/lp/ou_press/earnings-and-disposable-income-of-farmers-in-sweden-1997-2012-DXMxlbOuGM
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
2040-5790
eISSN
2040-5804
D.O.I.
10.1093/aepp/ppy005
Publisher site
See Article on Publisher Site

Abstract

Abstract This study presents a comprehensive analysis of farmers’ income in Sweden. The results indicate that farm households in Sweden do well from a standard-of-living perspective, but that farming is still a low-paid occupation from a return-on-skills perspective. Nevertheless, farm earnings increased faster over the study period than earnings in the general population, owing equally to higher farm earnings for operators and higher off-farm earnings for their spouse. Since few female spouses make farm earnings, evaluating farm household earnings from mainly a household perspective fails to acknowledge the individual careers of farmer and spouse. Farm income, off-farm income, disposable income, inequality, Gini, poverty Using longitudinal data for a period of 16 years (1997–2012), in this study we analyze changes in disposable income and earnings for farm households and changes in personal earnings for male and female farm operators, relative to those of households and men and women in the total population in Sweden. A fair standard of living for the agricultural community has been an objective in the European Union (EU) common agricultural policy (CAP) since the start.1 “Standard of living” is usually interpreted as the economic well-being of people, referring to their consumption possibilities measured by disposable income.2 In studies of the agricultural community’s standard of living, the unit of observation is usually the disposable income of farm households, based on the assumption that households comprise individuals who share income and decisions on expenditure. One example is the seminal paper by Gardner (1992), which concluded that the farm problem of low income has disappeared. Since then, evidence continues to show that farm households have similar economic well-being to non-farm households (Mishra et al. 2002; Organisation for Economic Co-operation and Development 2003; Katchova 2008; Hill and Bradley 2015). However, this is primarily because farm households, and particularly the farmer’s spouse, are devoting more time to off-farm work (Hanson and Spitze 1976; Ahearn, Johnson, and Strickland 1985; Findeis and Reddy 1987; Gardner 1992; Mishra et al. 2002; Ahearn, El-Osta, and Dewbre 2006; El-Ostra, Mishra, and Morehart 2008; Hill and Bradley 2015). Studies show that part-time farming is important for farm survival in Europe (Glauben et al. 2006; Breustedt and Glauben 2007).3 High off-farm income may also be efficient for risk spreading, and individuals may hold multiple jobs as a way to obtain new skills (Panos, Pouliakas, and Zangelidis 2014). However, studies that measure income at the household level may overlook problems at the individual level. Such studies implicitly presume that household incomes are shared equally between spouses, a presumption that is highly sensitive from a gender perspective and not necessarily true.4 When the off-farm labor supply of a female spouse increases, it may have both positive and negative effects on that woman’s family and labor market situation. Pluriactivity increases the economic autonomy of women (Blanc and MacKinnon 1990), but it is uncertain whether it increases women’s bargaining power and involvement in farm decision making (Gasson and Winter 1992; Evans and Ilbery 1996). In addition, pluriactivity may not necessarily decrease women’s farm work (Evans and Ilbery 1996). Thus, a double burden (farm and family) may become a triple burden (i.e., farm, family, and career; Blanc and MacKinnon 1990). In addition, applying a household perspective to single-person farm households is problematic; for this group of farmers, the relevant comparison is other single-person households. An alternative is to investigate individual earnings. The “fair standard of living” objective in Article 39 of the Treaty on the Functioning of the European Union is specified as: “ … in particular by increasing the individual earnings of persons engaged in agriculture.” Individual earnings are usually regarded as measuring the return to human capital (Becker 1964; Willis 1986). When comparing income between men and women, between races or ethnic groups, or between sectors, the level of comparison is typically the individual. However, the individual earnings of farmers are rarely investigated. A recent study by the European Parliament states that: “evidence points to farmers NOT being a particularly low-income sector of society in most Member States judged on the basis of their household disposable incomes,” (Hill and Bradley 2015), but this statement mainly refers to the standard of living of farm households, not the returns to farming (skills). Thus, the novel contribution of our study is to analyze the income of farm households from two dimensions: (a) the household-individual dimension and (b) the disposable income-earnings dimension. We use a broad definition of farm households (persons related by family ties sharing the same dwelling, where at least one member has some income from farming) since, with a narrow definition, a longitudinal analysis would eventually lose those individuals who are most successful in gaining off-farm income (Hill 2012).5 It is relevant to evaluate the returns to farming and not only the standard living of farmers, that is, to explore the disposable income-earnings dimension. Low return to farming may indicate inefficiency and provide an incentive to reallocate resources to sectors with higher levels of remuneration. However, it could be argued that farming is more involved in the production of public goods than other sectors (OECD 2001; Brunstad, Gaasland, and Vårdal 2005; Huylenbroeck et al. 2007). Due to non-rivalry in consumption, it is difficult to obtain compensation through the market for such goods, and therefore low earnings in farming could be the result of market failure rather than inefficiency. In that case, relying on market forces alone would result in non-optimal resource allocation from a societal perspective. Several of the subsidies in the EU Rural Development Programs in the second pillar of the CAP are motivated by environmental concerns in a broad sense.6 Article 39 in the Treaty also suggests other concerns, such as food security (the objective of the CAP is to “assure the availability of supplies and to ensure that supplies reach consumers at reasonable prices”). Accordingly, information on the returns on farming is of interest to policy makers. However, recent changes in farm household income in the EU are poorly documented because there has been no system in place for recording farm household income statistics in the EU since 2002 (Hill 2012; Hill and Bradley 2015). Existing income data in the EU are measured at the national or farm level (as farm family income), meaning that data on off-farm income are not collected. This means that the standard of living of farm households is underreported. Hence, in most Member States detailed longitudinal data are available for evaluating the “production objective” in the Treaty, but the “standard of living objective” is difficult to evaluate. The European Court of Auditors (2016) has pointed out that the “statistical data used to analyse farmers’ incomes has significant limitations.” This study is probably the most comprehensive analysis to date of farm household income for an EU Member State. We use longitudinal register data to evaluate changes in earnings of Swedish farm households in the period 1997–2012. Individual data allow us to divide household earnings into farm and non-farm earnings and, using the Multigenerational register, we allocate household earnings between family members. Previous Research and Measurement of Farm Household Income In the United States, 80% of total household income is off-farm income, more than 50% of farm operators work off-farm, and 60% of operators of small farms (<5 ha UAA) work less than one-quarter of their time on-farm (El-Ostra, Mishra, and Morehart 2008).7 Furthermore, a substantial proportion of household income is provided by farm operators’ family members, with farm spouses devoting more time to off-farm work over time (Hanson and Spitze 1976; Mishra et al. 2002). Mishra et al. (2002) report that, in the United States, almost half of farm spouses work off-farm, and that the proportion increased by 65% between 1969 and 1999. In the EU, around one-third of farm operators have off-farm income and about 50% of household income derives from off-farm sources. As in the United States, the off-farm proportion seems to be increasing over time (OECD 2003; Barthomeuf 2008; Hill and Bradley 2015).8 In Sweden, 50% of farm operators had off-farm work in 2005 (Barthomeuf 2008). A Swedish survey using firm-level data found that 30% of farms engaged in diversification, although income from paid employment outside the farm was not included (Swedish Board of Agriculture 2007). A Finnish study using tax register data for 1986 showed that household income per farm family unit amounted to only 71% of the household income per family unit of industrial workers, although this increased to 95% when disposable income per family unit was compared (Puurunen 1990). Measuring Earnings As noted, since 2002 there has been no system for measuring farm household income in the EU. The Eurostat’s Income of the Agricultural Households Sector (IAHAS) statistics started in the late 1980s and terminated in 2002, but the quality of the IAHS data was a problem much earlier (Hill 2012; Hill and Bradley 2015).9 The EU currently uses two systems to measure returns from farming: Economic Accounts for Agriculture (EAA) at the Member State level (see Eurostat 2002, 2010), and the Farm Accountancy Data Network (FADN) at the farm level. In FADN, farm family income is measured per farm or per family work unit on the farm. However, the income indicator in both the EAA and FADN excludes off-farm income.10 The 2005 Farm Structure Survey (Barthomeuf 2008) collects information on other gainful activities (OGA) for farmers in the EU, but spouses are not included. The survey asks if the farmer receives income from diversification activities such as tourism and refining/processing, and income from paid employment outside the farm. However, the Farm Structure Survey collects no actual data on incomes. Note also that, unlike our definition of farm income, which is based on the total returns from a farm’s human capital (whether it originates from work on one’s own farm or from work on other farms), the 2005 Farm Structure Survey classifies paid farm work outside the farm as off-farm income. The Swedish Board of Agriculture (2007) merges micro-level register data with a survey of off-farm income. From these data, they identify off-farm income at farm level (i.e., farm diversification), but exclude paid employment outside the farm. Instead, data on total individual returns from business, work, and capital are collected, but these total returns cannot be separated into remuneration from farm and off-farm activities. Moreover, the data are for a single year, 2005, and farm household earnings are not compared against population averages. To measure the standard of living of farm households, the preferred income measure is net disposable income, which was collected in the IAHS (Hill 2012). However, when examining differences in labor market returns between farm household members and the population at large, earnings are preferred. Inequality in disposable income is caused by labor market factors, but also by factors pertaining to the tax system, capital returns, and non-labor market transfers (i.e., child support, housing benefit, and social security benefits). Earnings include income from business and work, and labor market transfers. Labor market transfers are conditional on labor market participation and provide income support while the individual is absent from work, that is, they include parental leave payments, sickness payments, and unemployment benefits. For the present study, we generated data on all these variables by merging information from several Swedish registers. Thus, with unusually good data, this study presents a broad socioeconomic overview of Swedish farmers. Data and Descriptive Statistics For this study, we used the Longitudinal Integration Database for Health Insurance and Labour Market Studies (LISA), which includes a broad range of indicators on demographics, labor market status, income, and education for the entire Swedish population (aged 16 and older). From LISA, we obtained a full sample of (a) individuals receiving income from business or work in agriculture 1997–2012, and (b) their children and spouses.11 In this study we restrict the sample to farm households that contained the business operator, that is, farm employees and their family members were excluded. The analysis was carried out at the individual and the household levels, with changes followed over time. By using the household indicator in LISA and linking LISA to the Multigenerational Register, we acquired detailed information on household composition. We apply a “broad” definition of farm households as households where at least one member earns a farm income from business or work. The farm operator is defined as the household member with the highest mean farm earnings 1997–2012 even when some other household member, for example, a child or a spouse, is the listed farm operator. With the decoupling reform in 2005, a number of people who owned some agricultural land became eligible for single farm payments from Pillar I of the CAP and were thus registered as having agricultural earnings. In 2004 (when the first applications for the new payments were made), the sample of individuals with agricultural earnings increased by around 30%. However, from 2007, four hectares of land was set as the minimum required for receiving the single farm payment, which decreased the number of “farmers” substantially, again by around 30%. Households that primarily receive income from CAP subsidies, and which may be regarded as “pure subsidy farmers,” affect the results for the years around the decoupling reform, the main effect being that mean farm earnings became substantially lower in 2004–2006 and off-farm earnings became much higher. We therefore remove this group of farmers in the most efficient way available by restricting the sample to only include farmers receiving farm earnings for at least five years in the period 1997–2012 (when a three- or four-year restriction on farm earnings is applied, the “bias” in the results remain). Nevertheless, these “subsidy farmers” still affect the results for off-farm income (see figure 10). Earnings from forestry and extension services are included in farm earnings. For Sweden, where forestry and farming are closely linked, this is reasonable. As the data thus only indicate whether the main income is from farming or forestry, we were unable to separate income from these two sources unless the farmer had two different firms, one for farming and one for forestry, which is uncommon. However, since we restrict the sample to farm operator households, we know that some part of the income is from farming. About 14.7% of operators have earnings mainly from forestry and about 2.9% have earnings mainly from extension services. Most people in Sweden retire at the legal retirement age of 65. As the intention in this study is to compare earnings for the farming population and the total Swedish population, the relevant group for comparison is the working population (aged 20–64), although farmers, being self-employed, may choose to retire later. Thus, a household is removed from the sample when the operator reached the legal retirement age of 65 or when the household stopped farming, that is, household farm income became zero. If another operator took over the farm (e.g., a child), it is classified as a new household. We have 42,898 households and farm operators in our sample. The sample also contains 34,396 partners and 48,063 children (16 years and older), 91% of the operators are male and 68% have a partner.12 The sample decreases substantially over time: it is 37% smaller in 2012 than in 1997. The main reason is retirement: 67% (25% out of 37%) of the decrease is due to retirement. On average, the operators have farm earnings for 10.5 out of 16 years. Importantly, the results in this study are basically the same whether we use a balanced sample (farmers who report farm income for all years) or an unbalanced sample (where there is inflow and outflow of farmers). Table 1 presents descriptive statistics for male and female farm operators with/without a spouse. Table 1 Descriptive Statistics on the Sample of Farm Operators (Averages over All Years) Male operator with a partner Female operator with a partner Male operator without a partner Female operator without a partner N = 25,868 N = 4,208 N = 12,651 N = 1,785 Mean Standard deviation Mean Standard deviation Mean Standard deviation Mean Standard deviation Years with positive farm earnings 13.52 0.01 12.22 0.02 13.05 0.01 11.6 0.03 Individual earnings (thousands) 213.13 0.26 175.59 0.6 180.28 0.41 180.33 1.03 Earnings from farming (thousands) 170.95 0.24 127.58 0.53 147.74 0.39 123.46 0.93 Off-farm earnings (thousands) 42.18 0.19 48.01 0.47 32.54 0.24 56.87 0.88 Household total earnings (thousands) 439.46 0.44 426.68 1.3 223.4 0.55 215.57 1.46 Household disp. earnings (thousands) 423.97 0.91 434.44 2.62 256.76 1.52 236.63 3.53 Age 48.72 0.02 48.82 0.05 44.71 0.03 42.63 0.11 Children 1.33 0.00 1.12 0.01 0.13 0.00 0.4 0.01 Years of schooling 10.74 0.00 11.29 0.01 10.64 0.01 11.71 0.02 Northern Sweden 13.3% 17.3% 14.1% 17.9% Forrest districts in Sweden 33.4% 33, 0% 33.9% 30.2% Plains districts in Sweden 53.3% 49.7% 52, 0% 51.9% Proportion with off-farm earnings 50.8% 53.8% 36.2% 52.8% Proportion below the poverty rate 13.6% 7.1% 20.6% 25.4% Crop 19.8% 12.1% 19.3% 11.3% Milk 18, 0% 10.1% 18.2% 18.7% Meat 15.5% 12.3% 15.7% 14.3% Mix 29, 0% 47.8% 28.3% 32, 0% Forestry 14.7% 9.9% 15.2% 10.8% Extension services 2.9% 7.8% 3.3% 12.9% Small-sized farms 30.9% 34.6% 38.8% 32.8% Mid-sized farms 34.5% 30.1% 32.3% 23.1% Large-sized farms 34.6% 35.2% 28.9% 44.1% Male operator with a partner Female operator with a partner Male operator without a partner Female operator without a partner N = 25,868 N = 4,208 N = 12,651 N = 1,785 Mean Standard deviation Mean Standard deviation Mean Standard deviation Mean Standard deviation Years with positive farm earnings 13.52 0.01 12.22 0.02 13.05 0.01 11.6 0.03 Individual earnings (thousands) 213.13 0.26 175.59 0.6 180.28 0.41 180.33 1.03 Earnings from farming (thousands) 170.95 0.24 127.58 0.53 147.74 0.39 123.46 0.93 Off-farm earnings (thousands) 42.18 0.19 48.01 0.47 32.54 0.24 56.87 0.88 Household total earnings (thousands) 439.46 0.44 426.68 1.3 223.4 0.55 215.57 1.46 Household disp. earnings (thousands) 423.97 0.91 434.44 2.62 256.76 1.52 236.63 3.53 Age 48.72 0.02 48.82 0.05 44.71 0.03 42.63 0.11 Children 1.33 0.00 1.12 0.01 0.13 0.00 0.4 0.01 Years of schooling 10.74 0.00 11.29 0.01 10.64 0.01 11.71 0.02 Northern Sweden 13.3% 17.3% 14.1% 17.9% Forrest districts in Sweden 33.4% 33, 0% 33.9% 30.2% Plains districts in Sweden 53.3% 49.7% 52, 0% 51.9% Proportion with off-farm earnings 50.8% 53.8% 36.2% 52.8% Proportion below the poverty rate 13.6% 7.1% 20.6% 25.4% Crop 19.8% 12.1% 19.3% 11.3% Milk 18, 0% 10.1% 18.2% 18.7% Meat 15.5% 12.3% 15.7% 14.3% Mix 29, 0% 47.8% 28.3% 32, 0% Forestry 14.7% 9.9% 15.2% 10.8% Extension services 2.9% 7.8% 3.3% 12.9% Small-sized farms 30.9% 34.6% 38.8% 32.8% Mid-sized farms 34.5% 30.1% 32.3% 23.1% Large-sized farms 34.6% 35.2% 28.9% 44.1% Note: The sum of the different types of operators is smaller than the total number of operators because an individual can change civil status over time. Table 1 Descriptive Statistics on the Sample of Farm Operators (Averages over All Years) Male operator with a partner Female operator with a partner Male operator without a partner Female operator without a partner N = 25,868 N = 4,208 N = 12,651 N = 1,785 Mean Standard deviation Mean Standard deviation Mean Standard deviation Mean Standard deviation Years with positive farm earnings 13.52 0.01 12.22 0.02 13.05 0.01 11.6 0.03 Individual earnings (thousands) 213.13 0.26 175.59 0.6 180.28 0.41 180.33 1.03 Earnings from farming (thousands) 170.95 0.24 127.58 0.53 147.74 0.39 123.46 0.93 Off-farm earnings (thousands) 42.18 0.19 48.01 0.47 32.54 0.24 56.87 0.88 Household total earnings (thousands) 439.46 0.44 426.68 1.3 223.4 0.55 215.57 1.46 Household disp. earnings (thousands) 423.97 0.91 434.44 2.62 256.76 1.52 236.63 3.53 Age 48.72 0.02 48.82 0.05 44.71 0.03 42.63 0.11 Children 1.33 0.00 1.12 0.01 0.13 0.00 0.4 0.01 Years of schooling 10.74 0.00 11.29 0.01 10.64 0.01 11.71 0.02 Northern Sweden 13.3% 17.3% 14.1% 17.9% Forrest districts in Sweden 33.4% 33, 0% 33.9% 30.2% Plains districts in Sweden 53.3% 49.7% 52, 0% 51.9% Proportion with off-farm earnings 50.8% 53.8% 36.2% 52.8% Proportion below the poverty rate 13.6% 7.1% 20.6% 25.4% Crop 19.8% 12.1% 19.3% 11.3% Milk 18, 0% 10.1% 18.2% 18.7% Meat 15.5% 12.3% 15.7% 14.3% Mix 29, 0% 47.8% 28.3% 32, 0% Forestry 14.7% 9.9% 15.2% 10.8% Extension services 2.9% 7.8% 3.3% 12.9% Small-sized farms 30.9% 34.6% 38.8% 32.8% Mid-sized farms 34.5% 30.1% 32.3% 23.1% Large-sized farms 34.6% 35.2% 28.9% 44.1% Male operator with a partner Female operator with a partner Male operator without a partner Female operator without a partner N = 25,868 N = 4,208 N = 12,651 N = 1,785 Mean Standard deviation Mean Standard deviation Mean Standard deviation Mean Standard deviation Years with positive farm earnings 13.52 0.01 12.22 0.02 13.05 0.01 11.6 0.03 Individual earnings (thousands) 213.13 0.26 175.59 0.6 180.28 0.41 180.33 1.03 Earnings from farming (thousands) 170.95 0.24 127.58 0.53 147.74 0.39 123.46 0.93 Off-farm earnings (thousands) 42.18 0.19 48.01 0.47 32.54 0.24 56.87 0.88 Household total earnings (thousands) 439.46 0.44 426.68 1.3 223.4 0.55 215.57 1.46 Household disp. earnings (thousands) 423.97 0.91 434.44 2.62 256.76 1.52 236.63 3.53 Age 48.72 0.02 48.82 0.05 44.71 0.03 42.63 0.11 Children 1.33 0.00 1.12 0.01 0.13 0.00 0.4 0.01 Years of schooling 10.74 0.00 11.29 0.01 10.64 0.01 11.71 0.02 Northern Sweden 13.3% 17.3% 14.1% 17.9% Forrest districts in Sweden 33.4% 33, 0% 33.9% 30.2% Plains districts in Sweden 53.3% 49.7% 52, 0% 51.9% Proportion with off-farm earnings 50.8% 53.8% 36.2% 52.8% Proportion below the poverty rate 13.6% 7.1% 20.6% 25.4% Crop 19.8% 12.1% 19.3% 11.3% Milk 18, 0% 10.1% 18.2% 18.7% Meat 15.5% 12.3% 15.7% 14.3% Mix 29, 0% 47.8% 28.3% 32, 0% Forestry 14.7% 9.9% 15.2% 10.8% Extension services 2.9% 7.8% 3.3% 12.9% Small-sized farms 30.9% 34.6% 38.8% 32.8% Mid-sized farms 34.5% 30.1% 32.3% 23.1% Large-sized farms 34.6% 35.2% 28.9% 44.1% Note: The sum of the different types of operators is smaller than the total number of operators because an individual can change civil status over time. From Statistics Sweden (SCB), we also have individual and household earnings at the aggregate level for the Swedish working age population (20–64).13 In the analysis, we define households as families with at least two adults, that is, single adult households are not analyzed as households in this study, and this definition is used for both the farm population and the total population.14 Single farmers, that is, operators without a spouse, are analyzed separately. Farm households are not excluded from total household statistics but, because they constitute only around 2.4% of the total number of households, their contribution to the total population statistics is minor. Structure of the Analysis and Method for Comparing Earnings The difference in age composition between farmers and the total population has to be accounted for when comparing the groups. Our approach for calculating the mean individual and household earnings of farmers is, therefore, to estimate (with a pooled OLS) the earnings on 16 time dummies, βt, and 44 age dummies, δi: Earningsit=βt+δi. By excluding a constant and using the age 37 as the reference category, β shows the yearly average individual earnings for the age group 37. The reference age is set at the average age of individuals in the total working-age population. However, when comparing households (and individuals living in households with at least two adults), the reference age is set at 47, that is, the average age of the oldest working-age adult in farm households is higher than the average age in the total working-age population.15 This model removes an age bias in the results. All earnings variables are measured in 2012 prices. The comparison of earnings— βt to the average earnings in the population—has an individual, or household, perspective, and not a farm perspective per se. This means that the main findings were split according to type of household, and not according to type of farm business or farm size. Thus, we divide the households into those with a male and those with a female operator, and households where the operator has a partner and households where the operator is single.16 In statistics on the total population, the households are not separated according to gender.17 Our particular interests are on the comparison of household earnings and disposable income and household off-farm income. For 70% of the sample, we are able to match the data on income and earnings to farm-level data, which allows us to explore the impact of farm size on changes in earnings. The next step is to use the individuals’ Standard Industrial Classification (SNI) codes to classify the households into different types of households: (a) crop, (b) milk, (c) beef, (d) mix, (e) forestry, and (f) extension services. Inequality is analyzed by calculating the Gini coefficient and the at-risk-of-poverty rate. As this study is mainly descriptive, the findings may be affected by selection bias, that is, that farmers possess different skills than the general population. Nordin, Blomquist, and Waldo (2016) found that there is positive selection into farming in Sweden. In that study, a sibling approach was used to compare the earnings of farm siblings (sharing the same farm background) following different career paths, that is, one sibling chooses farming and the other chooses a non-farming occupation. A sibling fixed effect model that removes the common background characteristics of siblings was found to give a larger penalty to farming than an OLS model, indicating that positive selection into farming is likely. Based on those findings, farmers’ earnings are underestimated rather than overestimated in a returns to farming perspective. It is also important to point out that in this study we intentionally used a simple model and chose not to control for other factors (than age). From a standard of living perspective this is reasonable (a low standard of living is still low even if it can be explained by, e.g., education), but from a return to farming perspective one could consider controlling for education, for example. However, since education is an endogenous variable (choosing farming and education level is a joint decision), it can be questioned whether controlling for education is the correct choice. For example, a low return to farming may be caused by low education level of the farmer, or the education level may be low because of the choice to become a farmer. Besides, we also prefer having the same specification in all comparisons. Results Household Earnings and Disposable Income Average household earnings for farm households and for the total household population in the period 1998–2012 are shown in figure 1a) and b). The left y-axis shows household earnings and the right y-axis shows the gap in household earnings between farm households and the total household population. For the total household population, earnings increased by 33% between 1998 and 2012, while for farm households earnings increased by 56% and 53% for households with a male and a female operator, respectively. Owing to the larger increase in earnings for farm households, the gap between these and the total household population decreased by around 12 percentage points between 1998 and 2012. Though difficult to deduce from the figures, female operator households had around 3% higher earnings than male operator households; later we show that this is due to high off-farm earnings for the female operator’s spouse. Figure 1 View largeDownload slide Household earnings of farm households with a male operator or a female operator, compared with earnings for the total population of households, 1998–2012 Figure 1 View largeDownload slide Household earnings of farm households with a male operator or a female operator, compared with earnings for the total population of households, 1998–2012 Figure 2a) and b) show the corresponding values for disposable income. Disposable incomes were similar for farm households and other households in the study period, with the gap changing from somewhat negative (around 2–5 percentage points lower in farming) to somewhat positive (around 2–5 percentage points higher). Thus, while there was a substantial gap in earnings between farm households and other households, even in 2012, the gap in household disposable income was small and in fact disposable income was higher for farm households in 2006–2012. In other words, for the total population disposable income was about 20% lower than earnings, but for farm households it was about 5% lower in the beginning of the study period and similar by the end of the period. Figure 2 View largeDownload slide Household disposable income of farm households with a male operator or a female operator compared with that of the total population of households, 1997–2012 Figure 2 View largeDownload slide Household disposable income of farm households with a male operator or a female operator compared with that of the total population of households, 1997–2012 What could explain the relatively high disposable income of farm households? Compared to earnings, disposable income includes capital returns and non-work related transfers, and deducts taxes. Since farm households possess significant wealth (Organisation for Economic Co-operation and Development 2003), high capital returns might be an explanation. We are able to compare differences in capital returns for the year 2002.18 Farm household net capital returns in that year were around 40,000 Swedish Krona (SEK), which was about SEK 5,000 higher than net capital returns for the total population.19 However, the gap in household earnings between farm households and the total household population in that year was around SEK 140,000 (see figure 1a) and b)). Hence, higher capital returns for farm households can only explain about 3.5% (SEK 5,000 of SEK 140,000) of the relatively high disposable income of farm households (in relation to earnings for farm households). Moreover, non-work related transfers (mainly child allowance, but also social benefits and housing benefits) are similar for farm households and other households, making up about 4.9% (Swedish Board of Agriculture 2007) and 4.5% (Government Proposition 2004) of disposable income for farm households and the total population, respectively.20 Our data contains no information on taxes, so we were unable to investigate the taxes paid by farm households. However, the only remaining explanation for the relatively high disposable income of farm households is substantially lower taxes. Farmers can be expected to pay lower taxes, as they can take out part of the farm income as income from business activities, which is taxed at a lower rate than income from work, but it is surprising that this had such a large impact. Thus, we conclude that the taxation and tax deduction system gives farmers a standard of living that is comparable to that of the rest of the population. Does Farm Size Matter? For 70% of the farmers in our sample, we were able to match the individual data with farm-level data from the Swedish Board of Agriculture.21 The farm level data are available for the years 1999, 2003, 2005, 2007, and 2010. Based on hectares of arable land and number of animal units, we divide the farms into large-, mid- and small-sized farms, with each group containing one-third of all farms. Figure 3 shows that large- and mid-sized farms had, respectively, about 17% and 4% higher household earnings than the small-sized farms. The absolute increase in household earnings between 1999 and 2010 was almost the same for large- and small-sized farms, but the relative increase was larger for small farms. For mid-sized farms, both the absolute and the relative increase in household earnings were smaller than for large-sized farms. All these findings are for households with a male operator, but the results were similar for the total sample of farmers. Moreover, separate analyses for different production areas in Sweden (northern Sweden, woodlands and plains) show similar results to the aggregate sample (not reported). Figure 3 View largeDownload slide Farms household earnings of small-, middle- and large-sized farms. 1999–2010 Figure 3 View largeDownload slide Farms household earnings of small-, middle- and large-sized farms. 1999–2010 Individual Earnings We divide the sample into male and female operators, with or without a spouse (i.e., whether the operator lived in a household including two adults or was single). For the total population, the sample is not divided conditional on having a spouse. To gain a complete picture of total household earnings, children’s earnings should also have been reported separately. However, to focus the analysis we merely report that they earned around 8% of the total household earnings and that most of these earnings were from off-farm sources. Figure 4a) shows that earnings for male operators with a partner increased from 68% to 85% of the male population average. For male operators without a spouse, the relative earnings increased from 61% to 77% of the male population average. The earnings of single male operators were around SEK 27,000 lower than the earnings of male operators with a spouse; in absolute terms the difference increased over time, but in relative terms it was almost the same in 1997 as in 2012. Higher individual earnings for operators with a spouse may be because of selection (if the same factors determine earnings and success on the marriage market) or labor division within households (if the wife allocates more time to household work, the husband may allocate more time to farm or off-farm work). Figure 4 View largeDownload slide Individual earnings of the total population and of farms with a male operator or a female operator, with and without a partner, 1997–2012 Figure 4 View largeDownload slide Individual earnings of the total population and of farms with a male operator or a female operator, with and without a partner, 1997–2012 For female operators (figure 4b)), earnings increased from about 77% to 93% of the female population average. Thus, the gap to the population average was smaller for women and there was no major difference in earnings between female operators with or without a partner. So far, we have not divided the earnings into farm and non-farm earnings. Figure 5a) and b) show household earnings separated into the operator’s farm and non-farm earnings, and the spouse’s farm and non-farm earnings for households with a male or female operator with a spouse. Figure 5 View largeDownload slide Earnings from the farm and off-farm earnings of farms with a male operator or a female operator and their spouse, 1997–2012 Figure 5 View largeDownload slide Earnings from the farm and off-farm earnings of farms with a male operator or a female operator and their spouse, 1997–2012 First, the relatively large increase in earnings for farm households in the study period (see figure 1a) and b)) can be explained by a large increase in the operator’s farm earnings and the partner’s off-farm earnings. The annual increase in farm earnings was 3.1% and 2.85% for male and female operators, respectively. There was an even higher annual increase in the spouse’s off-farm earnings, of around 4%. For the total population, the annual increase in earnings was 1.74% and 2.20% for men and women, respectively. Moreover, in absolute terms, male operators increased their farm earnings by the same amount as spouses increased their off-farm earnings, by around SEK 80,000 (figure 5a)). This differs from female operator households: figure 5b) shows a much larger absolute increase in male spouses’ off-farm earnings (around SEK 110,000) than female operators’ farm earnings (around SEK 55,000). The most important findings from figure 5a) and b) are the low off-farm earnings for operators and the low farm earnings for spouses. This indicates a clear division of labor within households, where operators work mainly on the farm and spouses mainly off-farm. This was particularly clear in male operator households, where spouses’ farm engagement was very low. Moreover, while operators’ off-farm earnings increased (although from a low level) in the study period, spouses’ farm engagement hardly increased at all. These latter findings are central for understanding the household earnings of Swedish farmers, that is, the low off-farm earnings for operators and the low farm earnings for spouses, indicating that the division of labor in Swedish farm households is substantial. We also examine whether the results varied with farm size (supplementary online appendix figures A2–A4) and found that the division of labor was similar for small-, med- and large-sized farms.22 The main difference was that the operator’s and the spouse’s farm earnings increased and their off-farm earnings decreased with farm size. This meant that for small-sized farms the female spouse’s off-farm earnings were higher than the male operator’s farm earnings—actually, up until 2003 she had higher total earnings than her husband, as well. Operator’s Off-farm Labor Market Supply Next, we examine the variation in off-farm earnings and the relationship between off-farm earnings and total earnings for male operators (results for female operators are available upon request). Figure 6 shows household and individual earnings for male operators (with partner) with different degrees of engagement in farming. The x-axis shows farm earnings as a share of total earnings, that is, 100% means that the operator lacks off-farm earnings. The bars (measured at the right-hand y-axis) show the share of farmers in each “engagement group.” Figure 6 View largeDownload slide Relationship between male operator (with a partner) households and individual earnings and their degree of engagement in farming Figure 6 View largeDownload slide Relationship between male operator (with a partner) households and individual earnings and their degree of engagement in farming Almost 50% of the operators had zero off-farm earnings and 20% had less than 10% off-farm earnings. Relatively balanced engagement (i.e., 30% to 70% earnings from farming) was rare (only 7% of the operators). For male operators without a partner (not shown), 64% lacked off-farm earnings. Moreover, as figure 6 shows, both household and operator earnings decreased with increasing engagement in farming, so operators with 100% farm earnings had around two-thirds of the earnings of operators with 1% to 30% farm earnings.23 Over time, the difference decreased somewhat (not reported), but in 2012 operators with 100% farm earnings still had 30% lower earnings than those with a small engagement in farming. At the household level, earnings were around 20% to 25% lower for households with operators with 100% farm earnings than for operators with low farm earnings (1% to 30%). Spouses’ Labor Market Supply Next, we investigate the female spouses’ labor market supply. We find that around 65% of spouses had only off-farm earnings and that the share increased by more than 7 percentage points from 1997 to 2012 (figure 7). Another 5% to 10% were outside the labor force and had zero earnings; about half of these had retired or retired early. Figure 7 View largeDownload slide Labor supply of female farm spouses, 1997–2012 Figure 7 View largeDownload slide Labor supply of female farm spouses, 1997–2012 The likely explanation for the decreasing farm income for spouses is investment in higher education. Figure 8a) shows that the female spouses were more likely to invest in higher education than male operators, and that the within-household gender gap in higher education increased substantially between 1998 and 2012: the gap was 13 percentage points in 1997 and has almost doubled, to 22 percentage points, in 2012. In comparison, for the total population the gender gap in higher education was about 10 percentage points in 2012. In addition, female spouses had 1.5 more years of schooling than male operators. Since investment in education generally precedes matching in the marriage market, higher-educated spouses probably chose a career path that made dual work less likely, as higher education increased the probability of having zero farm income by 25% (not reported). Regarding the female spouses’ career choice, more than 50% of them worked as teachers or in healthcare. Figure 8b) shows the corresponding education statistics for households with a female operator. Here, the gender gap in higher education was small, only 2.5 percentage points in 2012. Figure 8 View largeDownload slide Proportion of operators and spouses with higher education (>12 years of schooling) on farms with a male operator or a female operator, 1997–2012 Figure 8 View largeDownload slide Proportion of operators and spouses with higher education (>12 years of schooling) on farms with a male operator or a female operator, 1997–2012 Figure 9 View largeDownload slide Farm earning of operators in different farming branches. 1997-2012 Figure 9 View largeDownload slide Farm earning of operators in different farming branches. 1997-2012 Income in Different Branches Farm and off-farm earnings from different types of farming are shown in figures 9 and 10.24 As can be seen from figure 9, farmers with income mainly originating from forestry had the highest farm earnings. The lowest farm earnings were found for meat farmers. Farmers with crop and mixed farms also had relatively low farm earnings. The largest variation in earnings was found for milk farmers; between 1997–2006 their annual increase in earnings was 5.0%, but in 2006–2012 it was only 0.5%. Farmers engaged in extension services showed the lowest annual increase in earnings (1.4%). The high earnings of milk farmers had a negative impact on off-farm earnings, with milk farmers having particularly low off-farm earnings (figure 10). The highest off-farm earnings were found for farmers engaged in crop production or forestry. Note that in 2004–2006, there were still some “subsidy farmers” in the data, and these increased off-farm earnings for farmers in general in 2004–2006. Figure 10 View largeDownload slide Off-farm earnings of operators in different farming branches. 1997-2012 Figure 10 View largeDownload slide Off-farm earnings of operators in different farming branches. 1997-2012 Inequality in Farming Income inequality in farming has fallen in the United States (Gardner 1992). By 1987, the incidence of poverty was for the first time lower among farm households than among non-farm households (U.S. Department of Commerce 1990). Corresponding inequality and poverty data are not available for the EU (Hill 2012). Thus, we examine inequality in two dimensions: the Gini coefficient and the at-risk-of-poverty rate. Usually, the Gini coefficient is calculated as household equivalized disposable income, which is obtained by dividing household disposable income by the number of “equivalent adults,” using a standard (equivalence) scale.25 Figure 11 shows the Gini coefficient, in household equivalized disposable income, for farm households and the total population. For farm households, it also shows the Gini coefficient in household equivalized earnings. As in most other European countries (Roine and Waldenström 2015), inequality in the total population has increased in Sweden: in 2006–2012, the Gini coefficient was about 7% higher than around the turn of the millennium.26 An increase in the earned income tax credit in 2006 is the main explanation for this change. For farm households, there was no visible trend in the Gini coefficient for equivalized disposable income, which fluctuated around 0.33. Thus, in 2007–2011 overall inequality in Sweden was around the same as inequality within the farm household population. Figure 11 View largeDownload slide Gini coefficient for farm households and the total population,1997-2012 Figure 11 View largeDownload slide Gini coefficient for farm households and the total population,1997-2012 For household equivalized earnings, the Gini coefficient decreased for farm households. We had no data on the Gini coefficient in household equivalized earnings for the total population, but Bengtsson, Edin, and Holmlund (2014) report that it was stable over the period. The at-risk-of-poverty rate is the proportion of people with equivalized disposable income below a standard threshold, which is set at 60% of the national median equivalized disposable income. Since we use the yearly national median, we calculate the relative at-risk-of-poverty rate, not the absolute rate, which uses the national median for a base year. Figure 12 shows the at-risk-of-poverty rate for the total and the farm population. Before 2004, the at-risk-of-poverty rate for the total population is only available for all ages, but, as figure 12 shows, the rate was similar for the working population (here 18–64) and all ages after 2004, particularly for the period 2004–2007.27 Figure 12 View largeDownload slide At-risk-of-poverty rate for farm households and the total population,1997-2012 Figure 12 View largeDownload slide At-risk-of-poverty rate for farm households and the total population,1997-2012 There was also a clear difference in trends for the farm population and the total population; whereas poverty increased in Sweden in general, it decreased in the farm population (figure 12). These different trends resulted in the rate being almost the same in the total population as in the farm population in 2010–2012, around 13–14%. To test whether the decreasing poverty rate in farming was due to retirement of low-income farmers, that is, due to a cohort effect, we calculated the at-risk-of-poverty rate for the age group 20–55. We found that the decrease in the at-risk-of-poverty rate for this group was almost the same as in figure 12. Conclusions This study analyzes farmers’ income from many different dimensions. Previous studies have analyzed farm households’ disposable income, but by applying an individual perspective and comparing disposable incomes to earnings, we present novel information on a known topic. The difference in earnings between the farm population and the total household population decreased by around 12 percentage points in the study period (1998–2012), to around 15% in 2012. For farmers without a partner, earnings were particularly low. We have not controlled for number of hours worked because of a lack of data. However, existing survey data do not suggest that the number of hours worked in farming is smaller than in other sectors.28 However, farm households’ disposable income was similar to that of the total household population. This was not due to high capital returns or high non-work related transfers in farming, but rather to a favorable tax system. Thus, from a standard of living perspective farm households do well, but from a returns to farming perspective farming is still a low-paid occupation, which may reduce incentives to allocate resources to the sector. In the long run, this could mean that EU environmental objectives and food security objectives are not met. Nevertheless, the returns to farming increased substantially from the end of the 1990s (measured as operators’ farm earnings), but the large increase in farm household earnings was equally caused by higher off-farm earnings for their partners. The partners seemed to choose a career path that made dual work less likely and their relatively high level of education reduced the probability of them having farm income. Hence, there was a clear division of labor within households; the operators worked mainly on the farm and the spouses mainly off-farm. When examining farm earnings from a household perspective, we implicitly assumed that the farm is a “family farm.” Garner and de la O Campos (2014), who have reviewed 36 definitions of the term “family farm,” claim that the most defining character of a family farm is the reliance on family labor—both women’s and men’s. However, we found that few farms in Sweden rely on both spouses’ labor (at least 70% of the female spouses have no farm earnings) and that applying mainly a family perspective is not justified. To evaluate the broader economic situation of farm households, the individual careers of the operator and the spouse would need to be considered. If the ambition is to attract more women to the sector, more knowledge is needed about women’s situations, and this is not feasible without an individual perspective. Finally, whereas inequality in Sweden increased over the period, within-farming inequality either decreased (household equivalized earnings) or remained stable (household equivalized disposable income). The at-risk-of-poverty rate also decreased in the farming population. Supplementary Material Supplementary material is available online at Applied Economic Perspectives and Policy online. Footnotes 1 Article 39, paragraph 1b, Treaty on the Functioning of the European Union (former Treaty of Rome). Available at: http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:02016ME/TXT-20160901&from=EN (Treaty on the Functioning of the European Union) and: http://ec.europa.eu/archives/emu_history/documents/treaties/rometreaty2.pdf (Treaty of Rome). 2 Disposable income includes earnings from labor, land, and capital (regardless of the sector in which these earnings originate), plus subsidies, minus taxes. However, the value of own consumption of farm products is not accounted for (see FAO 2011 and Hill 2012). 3 This is rarely the case in other occupations: in the United States, only about 5% of the population have multiple job holdings and the number is declining (Hirsch, Husain, and Winters 2016). Kimmel and Conway (2001) claim that multiple job holdings are generally undesirable and that the main motive for taking a second job is economic hardship. 4 This relates to the ambiguities regarding the definition of a “household”. If it is defined as “all persons sharing the same dwelling”, it is unlikely that incomes are pooled and decisions on their use taken in agreement. The problem remains even if households are defined as “persons with family ties sharing the same dwelling,” (see FAO 2011; Hill 2012, and references therein). 5 A narrow definition would include only persons who share the same dwelling and for which the main source of income is farming (FAO 2011; Hill 2012). 6 See Regulation (EU) No. 1305/2013. Available at: http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32013R1305&from=SV. 7 However, two-thirds of U.S. farms are small farms (sales less than $40,000) and contribute only 10% of the total farm production, but earn more than 75% of the off-farm income (Ahearn, Johnson, and Strickland 1985). 8 There are differences in the definition of a farm between the United States and the EU that could presumably affect comparisons. However, our aim was not to compare the significance of off-farm income for the standard of living of farmers in the United States and the EU, but to determine whether off-farm income is significant in both regions, and whether failure to account for this when analyzing the standard of living of farmers leads to questionable results. 9 Hill and Bradley (2015) list a number of reasons why adequate farm household income statistics are lacking for the EU; a consistent feature is that such statistics may be politically sensitive and may show that farmers are in a relatively favorable income position. 10 Another potential source is the Income, Social Inclusion and Living Conditions survey (EU-SILC). available at: http://ec.europa.eu/eurostat/web/income-and-living-conditions. However, according to Hill (2012), there are problems with coverage when disaggregating according to sector of occupation, as there are too few observations on farm households for reliable analysis of income distribution issues. 11 The Swedish Standard Industrial Classification (SNI) code is used for classifying firms as agricultural businesses. 12 However, the total number of individuals in the sample is smaller than the sum of operators, partners, children, and other family members because an individual can start out as a child and later become an operator, or a partner in another household. 13 Data on average earnings at the individual level are downloadable from Statistics Sweden’s homepage, but data on earnings at the household level were ordered separately. 14 Non-married cohabiting couples without children in common are also defined as singles in Swedish register data and are therefore included as singles in our analysis. 15 Using the reference age of 37 for households also has no major impact on the results, but using a much older (than 47) or younger (than 37) reference age affects the results somewhat. 16 Article 3, paragraph 2, in the Treaty of Amsterdam states that in all its activities “the Community shall aim to eliminate inequalities and to promote the equality of women and men”, and analyzes male and female farmers’ earnings separately. Available at: http://www.europarl.europa.eu/topics/treaty/pdf/amst-en.pdf. 17 To divide the households on gender, we have to designate a head of family. However, in a gender-equal society such as Sweden with family tax splitting, designating a head of family is both difficult and outdated. 18 Data on individual capital returns for the farming population are available in LISA for the entire period, but for the total population aggregate capital returns data are only available for 2002. The capital returns data are collected from an attachment to Government Proposition (2004). To confirm that the capital returns for 2002 are representative for the entire period, we compare total capital taxes, data on which are available for the entire period, with the returns to capital for farm households. These show a similar pattern over the period (see figure A1 which shows, in comparison with 1997, the percentage change in total capital taxes for the total population and the returns to capital for farm households). 19 Net capital returns is the difference between the returns and losses from mainly interest, dividends and capital gains from stocks and real estate. Importantly, land rents are included in income from business, and not in capital returns. 20 Note: CAP subsidies are recorded as income. 21 To be able to match the individual data with the farm data, the individual”s main income has to be from agriculture. For the match sampled, the farm income was 18% higher than for the unmatched sample (i.e. the matched farmers had a somewhat higher farming commitment), but, over time, the matched and unmatched farmers earnings developed similarly. 22 This is only possible for male operators with a partner. There were too few female operators with a partner to divide those data on farm size. 23 These results may seem to contradict each other. In Figure 3 we showed that small farms had the lowest household earnings and here we show that farmers with low engagement had the highest (household) earnings. However, farm size and engagement level are only marginally correlated. Low engagement is mainly an indication of “hobby farming” and most small farms have relatively high engagement. 24 A few farms change branch over time (mainly from mixed to a specific branch), we use the mode branch. 25 SCB gives the following weights to the members of the household: 1.0 to the first adult; 0.51 to the partner; 0.6 to each subsequent person aged 19 and over; 0.52 to the first child aged under 19, and 0.42 to each subsequent child aged under 19. 26 The Gini coefficient for the total population is without an age restriction, but Bengtsson, Edin, and Holmlund (2014) show that the Gini coefficient is similar for the entire population and the working population. 27 The at-risk-of-poverty rate is taken from the Eurostat database. 28 Statistics Sweden’s data on number of hours worked in different sectors are generated from surveys where the respondent is asked how many hours she worked during the previous week. However, for agriculture, where activity varies with season, results may not be representative for average number of hours worked per week. Also, the data does not allow separating the agricultural sector from the forestry and fishing sectors. References Ahearn M. , El-Osta H. , Dewbre J . 2006 . The Impact of Coupled and Decoupled Government Subsidies on off-Farm Labor Participation of U.S. Farm Operators . American Journal of Agricultural Economics 88 : 393 – 408 . Google Scholar CrossRef Search ADS Ahearn M. , Johnson J. , Strickland R . 1985 . The Distribution of Income and Wealth of Farm Operator Households . American Journal of Agricultural Economics 67 : 1087 – 94 . Google Scholar CrossRef Search ADS Barthomeuf L.-T. 2008 . Other Gainful Activities, Pluriactivity and Farm Diversification in EU-27. Brussels: European Commission, Directorate General for Agriculture and Rural Development, G.2. Economic analysis of EU agriculture. Bengtsson N. , Edin P.-A. , Holmlund B . 2014 . Wages, employment and incomes – do the gaps increase in Sweden? Fiscal Policy Council Report 2014/1. Blanc M. , MacKinnon N . 1990 . Gender Relations and the Family Farm in Western Europe . Journal of Rural Studies 6 : 401 – 5 . Breustedt G. , Glauben T . 2007 . Driving Forces behind Exiting from Farming in Western Europe . Journal of Agricultural Economics 58 ( 1 ): 115 – 27 . Brunstad R. , Gaasland I. , Vårdal E . 2005 . Multifunctionality of Agriculture: An Inquiry into the Complementarity between Landscape Preservation and Food Security . European Review of Agricultural Economics 32 : 469 – 88 . Google Scholar CrossRef Search ADS El-Ostra H. , Mishra A. , Morehart M . 2008 . Off-Farm Labor Participation Decisions of Married Farm Couples and the Role of Government Payments . Review of Agricultural Economics 30 : 311 – 32 . Google Scholar CrossRef Search ADS European Court of Auditors . 2016 . Is the Commission’s System for Performance Measurement in Relation to Farmers’ Incomes Well Designed and Based on Sound Data? Special Report No 1/2016. Eurostat . 2002 . Income of the Agricultural Households Sector—2001 Report. Theme 5, Luxembourg. Eurostat . 2010 . Draft Version of European System of Accounts: ESA 2010 . Luxembourg . Evans N. , Ilbery B . 1996 . Exploring the Influence of Farm- Based Pluriactivity on Gender Relations in Capitalist Agriculture . Sociologia Ruralis 36 : 74 – 92 . Google Scholar CrossRef Search ADS FAO . 2011 . The Wye Group Handbook. Statistics on Rural Development and Agricultural Household Income , ed. Pizzoli E. United Nations . Findeis J. , Reddy V . 1987 . Decomposition of the Income Distribution of Farm Families . Northeastern Journal of Agricultural Economics 16 : 165 – 73 . Gardner B. 1992 . Changing Economic Perspective on the Farm Problem . Journal of Economic Literature 30 : 62 – 101 . Garner E. , de la O. Campos A.P . 2014 . Identifying the “Family Farm”. An Informal Discussion of the Concepts and Definitions. ESA WP No. 14-10. Rome: Agricultural Development Economics Division, Food and Agriculture Organization of the United Nations (FOA). Gasson R. , Winter M . 1992 . Gender Relations and Farm Household Pluriactivity . Journal of Rural Studies 8 : 387 – 97 . Google Scholar CrossRef Search ADS Glauben T. , Tietje H. , Weiss C . 2006 . Agriculture on the move: Exploring regional differences in farm exit rates in Western Germany . Review of Regional Research 26 : 103 – 18 . Google Scholar CrossRef Search ADS Government Proposition . 2004 . Fördelningspolitisk redogörelse (Bilaga 3). Proposition 2004/05: 1. Hanson R. , Spitze R . 1976 . An Economic Analysis of Off-farm Income in the Improvement of Illinois Farm Family Income. Agricultural Economic Research Report, Department of Agricultural Economics, University of Illinois (139): 43. Hill B. 2012 . Farm Incomes, Wealth and Agricultural Policy—Filling the CAP’s Core Information Gap . 4th edition . Oxfordshire : CAB International . Google Scholar CrossRef Search ADS Hill B. , Bradley D . 2015 . Comparisons of Farmers’ Incomes in the EU Member States. Directorate-General for Internal Policies, IP/B/AGRI/IC/2014-68. Policy Department B: Structural and Cohesion Policies Agriculture and Rural Development. Hirsch B. , Husain M. , Winters J . 2016 . The Puzzling Fixity of Multiple Job Holding across Regions and Labor Markets. IZA DP No. 9631. Huylenbroeck G. , Vandermeulen V. , Mettepenningen E. , Verspecht A . 2007 . Multifunctionality of Agriculture: A Review of Definitions, Evidence, and Instruments . Living Reviews in Landscape Research 1: 1 – 43 . Katchova A. 2008 . A Comparison of the Economic Well-Being of Farm and Nonfarm Households . American Journal of Agricultural Economics 90 : 733 – 47 . Google Scholar CrossRef Search ADS Mishra A. , El-Osta H. , Morehart M. , Johnson J. , Hopkins J . 2002 . Income, Wealth, and the Economic Well-Being of Farm Households. Washington DC: U.S. Department of Agriculture, Economic Research Service, Agricultural Economic Report No. 812. Nordin M. , Blomquist J. , Waldo S . 2016 . The Income Penalty of Farming and Fishing: Results from a Sibling Approach . European Review of Agricultural Economics 43 ( 3 ): 383 – 400 . Google Scholar CrossRef Search ADS Organisation for Economic Co-operation and Development . 2001 . Multifunctionality: Towards an Analytical Framework . Paris . Organisation for Economic Co-operation and Development . 2003 . Farm Household Income—Issues and Policy Responses . Paris . Panos G. , Pouliakas K. , Zangelidis A . 2014 . Multiple Job Holding, Skill Diversification, and Mobility . Industrial Relations 53 : 223 – 72 . Puurunen M. 1990 . A Comparative Study on Farmers' Income . Finland : Research Publications 62, Agricultural Economics Research Institute . Roine J. , Waldenström D . 2015 . Long-Run Trends in the Distribution of Income and Wealth, Chapter 7. In Handbook of Income Distribution, Volume 2A , eds Atkinson A. , Fourguignon F. , 469 – 592 . Amsterdam : Elsevier B.V . Swedish Board of Agriculture . 2007 . Other Gainful Activities on the Agricultural Holding and Income of the Agricultural Household. Report 2007: 3. © The Author(s) 2018. Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Applied Economic Perspectives and PolicyOxford University Press

Published: Apr 23, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off