SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Literature Review
  5. 3. Specific Factors and the Incentive for Political Control of Labor Mobility
  6. 4. Landowner Power
  7. 5. Empirical Analysis
  8. 6. Methods and Results
  9. 7. Conclusion
  10. Acknowledgments
  11. References

Political actors today seek to influence labor mobility via education just as they have for centuries. Landowners in countries with large industrial sectors attempt to suppress education levels to maintain their labor supply, as educated workers are able to move into industrial work more easily than uneducated workers. However, the relationship between large landowners and education is more complex than has been previously theorized. Using a specific-factors model, I show that large landowners in countries with little economic development actually have an incentive to increase education levels. They realize the returns of an educated workforce without fearing their mobility because competing industrial opportunities for the workers do not exist. In either case, the ability of landowners to achieve their political goals is a function of their ability to overcome the collective action problem and effectively influence the state's provision of education. Powerful landowners successfully deny education in industrialized countries and provide it in agricultural countries. An analysis of panel data covering 77 countries from 1975 to 2000 confirms the conditional nature of the relationship.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Literature Review
  5. 3. Specific Factors and the Incentive for Political Control of Labor Mobility
  6. 4. Landowner Power
  7. 5. Empirical Analysis
  8. 6. Methods and Results
  9. 7. Conclusion
  10. Acknowledgments
  11. References

For the first 350 years of European colonization, New World governments divided domestic labor markets into two groups, slaves and freemen. Freemen responded to market pressures for their labors and received market wages while slaves worked as unpaid, forced laborers bound to their owner's land. Slave-owning landowners reaped the benefits of this system by undercutting agricultural competitors around the world who had to pay their laborers wages. Gradually, violent and non-violent events in the New and Old Worlds brought an end to slavery. Slaves became freemen, joined national labor forces, and began competing for jobs off the farm as well as on it. The abolition of slavery created a national labor pool and a functioning labor market free from such explicit political intervention.

The abolition of slavery did not, however, end the economic incentives for landowners to keep their agricultural workers from leaving their farms. Since landowners lost the battle over legalized slavery at the national level they turned to other tactics to capture rural labor, keep wages low, and maintain their profits. One prominent method of achieving this was to deny rural workers education.1 If landowners could prevent their workers from becoming educated they could also keep them on the farm because uneducated workers had a difficult time leaving farms and finding employment elsewhere. The battle over laborers’ ability to move from one type of employment to another therefore shifted from battles over slavery in the national political arena to local and state battles over labor mobility. Here, powerful landowners could still successfully use their political influence. Occasionally, these new battles took place on the national stage but for the most part they were waged where landowners were still able to be powerfully influential.2

This article argues that actors continue to influence labor mobility by influencing education. Education is a key determinant of labor mobility and political actors recognize this. In countries with industrial labor markets, agricultural landowners attempt to restrict education, resulting in limited labor mobility, a higher supply of rural workers, and depressed agricultural wages. Conversely, in countries without competing industrial labor markets agricultural landowners need not worry about the mobility of their laborers. These landowners have an incentive to provide some education to improve the productivity of their workers. In either case, when agricultural actors are powerful enough, they are dramatically successful in altering educational outcomes to achieve their ends.

I establish this political dynamic using a specific-factors model to show the economic motivations of landowners with respect to education in the national economy. I then show that land inequality is an indicator of the political power of landowners and their ability to achieve their political ends. I present evidence demonstrating that the relationship between land inequality and educational outcomes is conditional on the existence of competing industrial labor markets and is consistent with the predictions of the specific-factors model. I conclude by reflecting on how these findings affect our broader understanding of domestic and international politics.

2. Literature Review

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Literature Review
  5. 3. Specific Factors and the Incentive for Political Control of Labor Mobility
  6. 4. Landowner Power
  7. 5. Empirical Analysis
  8. 6. Methods and Results
  9. 7. Conclusion
  10. Acknowledgments
  11. References

Large landowners have long been an economically and politically powerful group in society seeking political arrangements that benefit their bottom line. In agricultural societies, landowners had the potential to be particularly powerful as they dominated the most important sector in the economy and earned most of its income. When foreign economic policies or democratization threatened their economic interest landowners played an influential role in the national political battles of their time. Prussian Junkers famously drove German trade policy before World War Two (Gerschenkron, 1943). Slave owners in the United States capped decades of sectional conflict over the economic orientation of the nation with a civil war in an effort to protect their interests (Bensel, 2000). European and Latin American landowners resisted democratization in order to avoid having their income or land redistributed (Acemoglu and Robinson, 2000, 2006a,b; Huber and Safford, 1995; Rueschemeyer et al., 1992).

Landowners impact local policies even more powerfully than they do national ones when it affects their interests. One of the primary policies of interest is education. Multiple studies have found a negative, linear relationship between the presence of large landowners and education spending and outcomes (e.g., Wegenast, 2009). This relationship existed in the American south where planation-style agriculture and educational suppression went hand in hand (Margo, 1990, 1991). Similar subnational work found that states and regions with large landowners consistently have lower levels of education than those with more equal landowners. This pattern holds among the federal states that comprise India (Banerjee and Iyer, 2005), the United States (Galor et al., 2009), and Brazil (Wegenast, 2010).

This negative relationship reflects incentives that landowners face with respect to education. First, public education is costly. In agricultural districts dominated by a few large landowners capturing all of the income, the taxes to pay for education would be taken the pockets of the landowners themselves. Rather than pay these costs, landowners prevent or suppress taxation for public education and create separate education systems for their own children (Engerman et al., 2009; Mariscal and Sokoloff, 2000). Second, if the masses remain uneducated then they are less able to politically organize and participate politically (Brady et al., 1995).3 Third, time spent in the classroom is time children do not spend working in the fields, which reduces profits. Lastly, landowners may believe that mass education is not appropriate for the masses from different ethnic or racial groups (Habyarimana et al., 2007). Diversity decreases support for redistribution (Klor and Shayo, 2010; Lind, 2007), decreases spending in American cities and states (Alesina et al., 1999; Luttmer, 2001), and has a negative effect on public goods provision cross-nationally (Alesina and Glaeser, 2004).

These factors exert a constant, negative pull on education levels but landowners may also suppress education to inhibit the ability of their workers to migrate to urban areas.4 Economists have long linked increased education with an increased likelihood of labor migration in both developed and developing countries (e.g., Greenwood, 1971, 1975; Molloy et al., 2011; Sjaastad, 1962; Zhao, 1997). Workers with higher levels of education are more mobile. They are able to work in higher paying non-agricultural jobs, more likely to learn about those jobs, and their educations reduce their attachments to local cultures making them more likely to migrate to cities for work. Agricultural elites therefore have an incentive keep education levels low in order to keep laborers in their labor pool. However, without the presence of an industrial labor market to lure agricultural workers away, agricultural elites may actually favor educating their workforce to have more productive workers. The relationship between landowners and education may therefore be conditional and this dynamic is examined below.

3. Specific Factors and the Incentive for Political Control of Labor Mobility

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Literature Review
  5. 3. Specific Factors and the Incentive for Political Control of Labor Mobility
  6. 4. Landowner Power
  7. 5. Empirical Analysis
  8. 6. Methods and Results
  9. 7. Conclusion
  10. Acknowledgments
  11. References

Landowners are concerned about their workers’ education levels because it affects worker productivity and worker mobility. Educated workers are more productive and more mobile. A specific-factors model shows why groups in society have an incentive to influence labor mobility.5 Each country is assumed to be a closed labor market. There are two sectors in the country, agriculture and manufacturing, both of which require the input of two factors to produce their products. Agriculture requires inputs of land and of labor. Manufacturing requires inputs of capital and of labor. Land and capital are fixed factors, meaning that they can only be used to produce goods in agriculture and industry, respectively. Labor is the mobile factor, meaning that laborers are assumed to move from one sector to the other based on demand. Wages are determined by the interaction of labor demand between the manufacturing and agricultural sector. The specific-factors model assumes diminishing returns to labor in both sectors. Because each additional worker per sector is slightly less productive than the previous one, there is a point at which employers cease to hire new employees because they would produce less value than the wage they would be paid. In the model, the wages paid by each sector are the same because an imbalance in wages causes labor to migrate from the low paying sector to the high paying sector.

Now assume a productivity shock significantly improves the marginal productivity of labor in manufacturing. This shock throws the system out of equilibrium. The total demand for labor in manufacturing increases and manufacturers now try to employ additional laborers. Wages in the manufacturing sector increase because there is the same supply of manufacturing workers and a higher demand for them. Manufacturers compete among themselves to employ a fixed number of industrial laborers. Because the supply of industrial workers is fixed, the workers’ wages rise. The workers, not the manufacturers, profit from the increase in productivity as their wages rise. Meanwhile the agricultural sector is unaffected. The supply and demand for agricultural laborers remains constant so wages remain constant as well.

In the long-run, however, the productivity shock in manufacturing alters the labor supply and the wages paid in the agricultural sector as well. Labor shifts from agricultural employment to industrial employment in response to higher wages offered in the manufacturing sector. This causes the number of laborers in the agriculture to shrink. A smaller supply of labor and constant demand for it results in higher wages. Wages for both factors equilibrate and labor stops migrating. Industrial wages are now lower than they were immediately after the shock, but higher than they were before the shock occurred. Conversely, agricultural wages are now higher. Assuming that the international market sets prices for the goods produced by both sectors, then the following distributional consequences occur: the manufacturers have higher profits; laborers in both manufacturing and agriculture have higher wages. Only the landowners lose. Their profits decline because they must now pay higher wages to their laborers while selling their goods at the same price as before.

Politics enters the specific-factors model when the degree of labor mobility can be influenced by political actions. In the example above, the productivity shock decreased the number of workers in agriculture. Any decrease in the supply of rural labor will result in rising wages paid by agricultural landowners. Unless landowners can prevent labor from migrating and wages from equilibrating they are trapped between the new price of their labor and the fixed prices their goods fetch on the international market. Therefore, in a specific-factor framework, factor owners have a strong incentive to create policies that restrict or enhance labor mobility.6 Agricultural landowners do so using their political influence to prevent or slow labor migration to the cities. Though explicit labor restrictions are rare in modern societies, education is a less explicit but still potent means of influencing labor mobility.7

The specific-factors model leads us to a different expectation in undeveloped countries as landowners face a different calculus. Education increases productivity allowing more goods to be produced with the same quantities of labor and land. In countries without an industrial sector, educated workers cannot migrate to better paying industrial jobs because they do not exist. The agricultural labor pool remains the same size and wages remain constant. Higher productivity and constant wages result in higher profits for landowners. In this scenario, the productivity gains of education accrue entirely to landowners. Education becomes a private good, solely benefiting the landowners, instead of a public good benefiting society. Therefore in undeveloped societies, landowners prefer more education because the productivity gains of education enrich the landowners.

4. Landowner Power

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Literature Review
  5. 3. Specific Factors and the Incentive for Political Control of Labor Mobility
  6. 4. Landowner Power
  7. 5. Empirical Analysis
  8. 6. Methods and Results
  9. 7. Conclusion
  10. Acknowledgments
  11. References

Economic interest does not translate perfectly into political influence. The more politically powerful landowners are as a group, the greater their ability to affect education levels. Like any group, landowners are better able to realize their interest when they are able to overcome the collective action problem effectively for their preferences (Olson, 1965). Agriculture elites should therefore be successful when land is concentrated in the hands of the few. A relatively small group of landowners will accrue the benefits of reduced labor mobility instead of the benefits being spread over thousands or hundreds of thousands of landowners.8 The concentrated benefits of lower wages incentivize the large landowners to take political action. Their small numbers increase their ability to sanction free riders that shirk the costly lobbying efforts required to keep education levels low. Thus, concentrated landowning enables landowners to overcome the collective action problem and decrease educational provision.9 This reduces labor mobility and keeps income inequality high.

5. Empirical Analysis

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Literature Review
  5. 3. Specific Factors and the Incentive for Political Control of Labor Mobility
  6. 4. Landowner Power
  7. 5. Empirical Analysis
  8. 6. Methods and Results
  9. 7. Conclusion
  10. Acknowledgments
  11. References

Having established the theoretical reasons to expect a conditional relationship between landownership patterns and education, I test for landowner influence on education using a large-N analysis with educational attainment as my dependent variable. Because the dependent variables are sampled at 5-year intervals, all of the measures are collapsed into 5-year averages

The dependent variable is education, which provides workers with skills and increases their ability to migrate from rural to urban jobs. Barro and Lee (1996, 2001) provide educational attainment figures for the over-15 population in 142 countries at 5-year intervals from 1960 to 2000. They measure the percentage of the population achieving various levels of education. I run the model using four different measures of educational attainment: percentage of the population that has graduated from secondary school, percentage of the population that has graduated from primary school, percentage of the population that has attended at least one year of school, and the average level of education. Unfortunately, these measures do not discern between urban and rural education levels. This restriction makes it impossible to isolate the effect of landlords specifically in rural agricultural areas. Therefore, finding a significant and substantive effect on the national level will probably under represent the true effect of landlords on education in the areas they control.

Land inequality is my measure of landowner political influence. The measurement is a gini score that ranges between zero (every landowner has an equal amount of land) and one (a single landowner possesses all the land). Land inequality data are drawn from the Food and Agriculture Organization's (FAO) World Census of Agriculture. Frankema (2009) combines FAO data with data from the International Institute of Agriculture to produce the most complete, publicly available, land inequality data set, which covers 105 countries.10 His land inequality ginis are constructed from decile measures of landholding and agricultural land data. Landholding data measure the land per farm and therefore the ability of the farmer to produce agricultural income. Frankema samples land inequality at various points across the 20th century with the earliest measurement in 1907 and the latest in 1999. Most countries have several measurements with at least one falling around 1960. For those countries with only one measurement, I hold land inequality constant at that value over time. For countries with more than one measurement, I assume a linear movement between the two. For time periods after the last recorded value, I assume the value remains constant. Generally, land inequality does not vary widely within a country except in times of land reform.11

Figures 1 and 2 maps land inequality data for three different time periods across the globe. The size of each circle represents its land inequality gini. For example, Mozambique has a small dot representing a land inequality gini of 36.7 and its easterly neighbor, Madagascar, has a larger dot representing a land inequality gini of 80.4. The colors represent the different time periods. Yellow represents observations between 1907 and 1940. Blue represents observations from 1941 to 1960. Red represents observations from 1961 to 2000. When multiple observations are available for a country then the circles overlap and shade each other. For example, Spain has a small red ring around a blue circle because land inequality slightly increased between its two measures in 1960 and 1989.

image

Figure 1. Map of Land Inequality Gini Scores Note: Yellow dots are observations between 1907 and 1940. Blue dots are observations from 1941 to 1960. Red dots are observations from 1961 to 2000.

Download figure to PowerPoint

image

Figure 2. Land Inequality By Region in 2000 Notes: Data from Frankema (2009). Regions defined by World Bank's World Development Indicators: EAP = East Asian & Pacific, ECA = East Europe & Central Asia, LAC = Latin American & the Caribbean, MNA = Middle East & North Africa, OECD = Orgnization for Economic Co-operation and Development, SAS = South Asia, SSA = Sub-Saharan Africa.

Download figure to PowerPoint

Several patterns emerge from the data. Latin America is the most unequal region of the world and remains so over time. It stands in greatest contrast to inequality in Asia. Second, Africa has large regional differences. West Africa has low levels of land inequality, while many of the countries in East Africa have fairly high levels. Europe has the greatest concentration of observations, including some for geographic units that are currently subsumed in larger nation states (e.g., Wales in the United Kingdom).

This measure of land inequality is the lower bound measure of actual agriculture income inequality. The FAO's land inequality data, which is the baseline for most land inequality indices, measure the distribution of land farmed by particular farmers, not the distribution among the actual landowners. Since landowners may rent their properties to multiple farmers and accrue profits from each of them, the ideal measure of landowning inequality would be the property distribution among owners, not the distribution among those renting the land (landholders). These land inequality measures, therefore, systematically underestimate land inequality and bias my results toward a finding of no effect. Unfortunately, the ideal landownership data do not exist. In the absence of an alternative landowning data set, I use Frankema's data.

Countries with high levels of economic development display a negative, linear relationship between land inequality and education as shown in Figure 3. Figure 3 plots the relationship between education (represented here as the percentage of the population over age 15 that has some secondary schooling) and land inequality for countries with GDP per capita incomes above the median in 1980. The higher the land inequality score, the lower the percentage of the over-15 population that has some secondary education. This relationship supports the theory that landowners fear urban migration and reduce education when they are politically able to do so.

image

Figure 3. Land Inequality and Education in Countries with Per Capita Incomes Above the Median in 1980 Notes: Land inequality data from Frankema (2009). Education (percentage of adults with some secondary education) is from Barro and Lee (2001).

Download figure to PowerPoint

High landowning inequality in poor countries empowers landowners to achieve their goal of a more educated and productive workforce as shown in Figure 4. Figure 4 plots the relationship between education and land inequality in countries with GDP per Capita below the median in 1980. As predicted, there is a positive relationship between land inequality and education.

image

Figure 4. Land Inequality and Education in Countries with Per Capita Incomes Below the Median in 1980 Notes: Land inequality data from Frankema (2009). Education (percentage of adults with some secondary education) is from Barro and Lee (2001).

Download figure to PowerPoint

Land inequality is positively correlated with the educational outcomes that agricultural elites pursue in both rich and poor countries. The same high levels of land inequality politically empower landowners to provide education in undeveloped countries and deny education in developed ones. Developed societies with high levels of land inequality have lower levels of education than those with low levels of land inequality. Less developed societies with high levels of land inequality have higher levels of education than those with low levels of land inequality. The theory and evidence lead to the expectation of a conditional relationship between education and land inequality.

5.1 Control Variables

Land inequality creates economic incentives and political opportunity for landed elites to influence education but other forces affect education as well. To test for the effect of landed elites on education those forces must be controlled for. The strongest predictor of educational attainment is economic development, measured here as Ln GDP per Capita. Mass education is, in part, an outgrowth of the process of industrialization and the demand for skills by both employers and workers. This variable is included to control for the demand for education produced by industrialization. A visual inspection of the relationship between GDP and educational levels shows a strong positive relationship. The GDP data are from the World Bank's World Development Indicators (2008).

While the demand for education is present in all societies, GDP per capita is not a perfect predictor of educational outcomes nor has it been so historically. In the 19th century, many countries in the Caribbean were richer than the United States and Canada and yet had much lower schooling rates (Mariscal and Sokoloff, 2000). Mariscal and Sokoloff suggest that ethnic diversity and income inequality play key roles in education decisions. In societies that are highly unequal or ethnically diverse, the elites controlling the government will create private academies instead of funding universal public schooling. Inequality negatively influences educational spending because it is less costly for the rich to funding private school serving only their children than it is to be taxed to fund public schools serving every child.

Ethnic diversity also reduces redistribution because people are less willing to redistribute to those they see as unlike themselves. To control for this influence, ethnic diversity is measured using the ethnicity fractionalization scores provided by Alesina et al. (2003). Fractionalization measures use the Herfindahl concentration index to create a score that measures the chance that any two randomly selected individuals in a country are a part of different groups. The formula is as follows:

  • display math

where si is the percentage of the population each ethnic group has within a country with n groups. This formula has been applied to ethnic, linguistic, cultural, and religious divisions within countries across the world. I test each model with different measures of diversity including Alesina et al.'s ethnic, linguistic, and religious measures as well as Fearon's (2003) ethnic and cultural diversity measures. The results do not differ significantly from measure to measure.

In societies that are high levels of income inequality the elites controlling the government will create private academies instead of funding universal public schooling (Engerman et al., 2009). Inequality negatively influences educational spending because it is less costly for the rich to funding private school serving only their children attend than it is to be taxed to fund public schools. Income inequality measure used here is a pre-tax gini with scores ranging from zero (perfect income equality in society) to one (all the income is earned by one person). Solt (2009) recently released a new income inequality data set, Standardizing the World Income Inequality Database (SWIID). SWIID is an update of the commonly used WIID (UNU-WIDER, 2008), a collection of hundreds of national-level inequality studies. WIID has been widely criticized for its use of apples and oranges measures and spotty coverage (Brune and Garrett, 2005). Solt addresses these concerns by making a series of standardizing assumptions, sorting through the WIID data, standardized the scores, and drew inferences from the primary data to fill missing measurements when appropriate.

Democracy is expected to be positively related to educational attainment as it is a primary predictor of social redistribution. In most societies, education is provided by the state and therefore political process as well as economic pressures influence educational outcomes. The state supply of education, like other public expenditures, is a function of the ability and desire of the state to tax and spend on this (possibly) redistributive service. If a government is fully democratic, then it may serve as a conduit of the median voter's preferences for educational provision (Meltzer and Richard, 1981). However, the median voter is only predictive in societies where democratic control is perfectly represented and where politics are organized around one issue. Limited enfranchisement results in particular groups being able to exercise the power of the state to realize their preferences and also prevents other groups from using the state as a tool for redistribution. Greater enfranchisement usually results in increased redistribution (e.g., Huber et al., 2008; Lindert, 2004). I use Marshall and Jagger's Polity IV data as my measure of democracy. They score each country on a scale of −10 to 10, with −10 as the most authoritarian and as 10 the most democratic (Marshall and Jaggers, 2008).

6. Methods and Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Literature Review
  5. 3. Specific Factors and the Incentive for Political Control of Labor Mobility
  6. 4. Landowner Power
  7. 5. Empirical Analysis
  8. 6. Methods and Results
  9. 7. Conclusion
  10. Acknowledgments
  11. References

The panel data include 77 countries measured at 5-year intervals between 1975 and 2000. Because the measure of ethnic diversity does not vary over time a fixed-effects model is inappropriate and so I use a random effects model. I include a lagged dependent variable as the dependent variable at time T is largely comprised the same population present in the previous period. Since the dependent variable measures the over-15 population with a given level of education, I lag the independent variables and the dependent variable 10 years so that all of the children receiving education at T-2 will be measured by the dependent variable at time T.

The results are displayed in Table 1. Each model includes an interaction term between Land Inequality and Ln GDP per Capita. Because the interaction term is between the two continuous variables, it is inappropriate to look at either the raw magnitudes or significance levels of the first three variables (Brambor et al., 2006). To understand the interactive effects between Land Inequality and Ln GDP per Capita, Figures 5-8 are provided. These figures correspond to Models 1, 2, 3, and 4 in Table 1.

Table 1. The Interactive Effect of Land Inequality and GDP Per Capita on Various Schooling Measures
 Model 1Model 2Model 3Model 4
Dependent variableSecondary completion (% of population)Primary completion (% of population)Any schooling (% of population)Average years of education
Notes:
  1. Estimates from a random effects models; Standard errors in parentheses.

  2. a

    Significant at 0.1.

  3. b

    Significant at 0.05.

  4. c

    Significant at 0.01.

Land inequality Gini44.81c42.64a30.65a4.24b
 (16.13)(24.26)(18.11(1.84)
Ln GDP per capita4.579c6.465c3.636c0.538c
 (1.150)(1.537)(1.175)(0.134)
(Land inequality Gini)a (Ln GDP per capita)−5.203b−5.849b−4.558b−0.555b
 (2.037)(2.680)(2.033)(0.233)
Ethnicity−0.0503−0.0411−0.0784−0.00865
 (0.0639)−0.0780)(0.0573)(0.00671)
Income inequality−10.61c−9.404a−3.185−0.630
 (4.070)(5.032)(3.580)(0.432)
Democracy−2.060−3.851−3.019a−0.250
 (1.733)(2.355)(1.825)(0.207)
Lagged dependent variable0.881c0.702c0.752c0.848c
Constant−20.69b−23.44b−7.276−2.176b
 (9.025)(11.88)(9.064)(1.037)
Overall R20.8810.9200.9520.954
Observations347347347347
image

Figure 5. Marginal Effect of Land Inequality on the percentage of Population that are Secondary School Graduates as GDP Per Capita Changes.

Download figure to PowerPoint

image

Figure 6. Marginal Effect of Land Inequality on percentage of Population that are Primary School Graduates as GDP Per Capita Changes.

Download figure to PowerPoint

image

Figure 7. Marginal Effect of Land Inequality on Any Schooling as GDP Per Capita Changes.

Download figure to PowerPoint

image

Figure 8. Marginal Effect of Land Inequality on Average Years of Schooling as GDP Per Capita Changes.

Download figure to PowerPoint

Figure 5 corresponds to Model 1 and shows the marginal effect of Land Inequality on the percentage of the over-15 population with high school degrees (Secondary Completion) as GDP per capita changes. The x-axis is Ln GDP per Capita and begins at four ($55 per capita) because no country in the sample scores below this point. The y-axis is the marginal effect of Land Inequality on Secondary Completion. The solid line is the marginal effect of Land Inequality on Secondary Completion as Ln GDP per Capita varies. The dotted lines above and below the solid linear line are the 90% confidence intervals. When per capita incomes exceed $4,000 (Ln GDP per Capita ˜ 8.3), Land Inequality is associated with a statistically significant decrease in Secondary Completion education. The impact is substantively important. In a country with a per capita income of $22,000 (Ln GDP per Capita = 10) having Land Inequality be 0.769 (one standard deviation above the mean) instead of 0.473 (one standard deviation below the mean) results in a 4% drop in high school graduates in the population. In a country of 30,000,000 people the two standard deviation increase results in nearly 830,000 fewer high school graduates.12 Between $4,000 and $270 Land Inequality does not have a statistically significant effect on Secondary Completion. When per capita incomes fall below $270 (Ln GDP per Capita ˜ 5.6), Land Inequality is associated with a statistically significant increase in Secondary Completion. For example, when per capita income is $550 (Ln GDP per Capita = 5) a two standard deviation increase in Land Inequality results in a 3.7% increase in the percentage of the population with from high school, a difference of nearly 790,000 additional graduates.

Figure 6, where the dependent variable is Primary Completion, shows a similar picture. The marginal effect of Land Inequality on Primary Completion is statistically significant at the 90% confidence level in both the most developed and least developed countries, again with opposite effects. The only real difference from Figure 5 is that the negative impact is now only statistically significant when per capita income exceeds $10,000 per year (Ln GDP per Capita ∼ 9.2) instead of $4,000. Above $10,000, there is a statistically significant and increasingly negative relationship between Land Inequality and Primary Completion. For a country with a GDP per capita of $36,000 (Ln GDP per Capita ~ 10.5) having Land Inequality be 0.769 instead of 0.473 results in 4.7% fewer primary school graduates. This is difference of 990,000 people in a country of 30,000,000. In the least developed countries, the effect of Land Inequality on Primary Completion is positive. When per capita incomes are $550 (Ln GDP per Capita = 5) having Land Inequality be 0.769 instead of 0.473 results in a 4.8% increase in primary school graduates, a difference of 1,000,000 people.

Figure 7 is very similar to Figures 5 and 6 but the effect shifts upward. The positive impact of Land Inequality on the percentage of the population that has attended at least 1 year of school (Any Schooling) in the least developed countries is higher than it is for the other dependent variables (5.6% compared to 4.8% for primary graduates and 3.7% for secondary graduates). The positive effect is also statistically significant for more countries. Land Inequality increases Any Schooling for countries with per capita incomes below $5,400 dollars (Ln GDP per Capita = 8.6). In contrast to the previous figures, Land Inequality does not have statistically significant negative impact on Any Schooling in wealthy countries. Finally, Figure 8 is in line with the previous figures. Land Inequality negatively affects Average Years of Education in high-income countries and positively affects it in low-income countries.

Taken as a group, Figures 5–8 present a clear message supporting the hypothesized conditional relationship. For Primary Completion, Secondary Completion, and Average Years of Schooling when Ln GDP per Capita is high, Land Inequality has a statistically significant and substantively important negative effect on education. The second side of the conditional relationship is also supported: when Ln GDP per Capita is low, Land Inequality has a positive effect on each measure of education.

The alternate explanations for education received mixed support in the models. Ethnic Diversity is not significant in any of the models. Alternate measures of ethnic diversity and other measures diversity (linguistic and religious) provide the same results. Income inequality is negatively associated with education in Models 1 and 2. The substantive effect is quite large, though less than half the magnitude of Land Inequality in very rich and poor countries. Moving from one standard deviation below the mean to one standard deviation above reduces the number of secondary graduates by 2.3% points or nearly 485,000 people. Democracy is consistently negative, though it only achieves statistically significance in Model 3. The results from all of the models are robust to alternate measures of diversity (cultural, religious, or ethnic), decade dummies, and regional dummies.

Overall, the results show the conditional effect of land inequality on education as GDP per capita varies. The consistent results strongly support the predicted conditional relationship. This is somewhat surprising given that the data systematically underestimate the true effect of land inequality in three important ways. First, the variables are measured at the national level and probably systematically under estimate the effect in particular subnational regions. The effect should be the strongest at the state and local levels where most education policy is set and funding decisions are made. That the effect still appears in national level data speaks to its impact in particular subnational regions. Second, as discussed previously, the land inequality data systematically underestimate the power of landowners because it measures landholding concentration. To the extent that farmers are renting their farms from larger landowners then the results are biased toward the null finding and understate land inequality's true effect on education. Third, the panel data are drawn from the three most recent decades because data are not available before these periods. One expects the effect of land inequality to be higher earlier in history when there was more variation in the dependent variables. Literacy rates in the Americas varied much more widely in the 19th century and the first half of the 20th century than they do today (Mariscal and Sokoloff, 2000; Engerman et al., 2009). The strong effect of land inequality in recent decades is suggestive of a stronger effect before 1975.

7. Conclusion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Literature Review
  5. 3. Specific Factors and the Incentive for Political Control of Labor Mobility
  6. 4. Landowner Power
  7. 5. Empirical Analysis
  8. 6. Methods and Results
  9. 7. Conclusion
  10. Acknowledgments
  11. References

The ability of wage laborers to work in free labor markets has been contested for centuries. The mobility of labor is, in part, the outcome of the political competition over education. As I argued above, agricultural landowners in different countries face different incentives with respect to education. Those that compete with an industrial sector work to keep laborers on the farm by reducing education levels. Landowners in societies without industrial competition have an incentive to educate their workers to create a more productive workforce and can do so with less fear of losing their workforce. In either case, landownership patterns explain when landowners are able to realize their political goals with respect to education.

The empirical evidence broadly supports the theory. Land inequality has a significant and substantive effect on all levels of education with the strongest effect on secondary education. The direction of the effect is conditional on the level of economic development. When landownership is concentrated, landowners in industrialized societies successfully suppress education and landowners in rural societies successfully promote it.

These political dynamics have other profound outcomes for societies aside from the labor market effects intended by landowners. The pro-education or antieducation bias of landowners influences the course of development and democratization for these societies. The quality of governance, political participation, and economic growth all vary profoundly based on the powerful influence of landlords on education. Ironically, this suggests that powerful landowners in the poorest societies may actually produce better outcomes compared to their smaller, less influential counterparts in other poor societies as they have economic rationale and the political power to increase education. Comparative case-studies of the political economy of education in these societies would a welcome next step in this line of research.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Literature Review
  5. 3. Specific Factors and the Incentive for Political Control of Labor Mobility
  6. 4. Landowner Power
  7. 5. Empirical Analysis
  8. 6. Methods and Results
  9. 7. Conclusion
  10. Acknowledgments
  11. References

The author thanks the anonymous reviewers, the editor of Economics and Politics, and Thomas Oatley for comments that improved this article significantly. The author also thanks Mark Creczenzi, Timothy McKeown, John Stephens, Erik Godwin, and Christine Carpino for their helpful feedback.

Notes
  1. 1

    Cohen (1991) and Blackmon (2008) discuss other methods landowners in the United States used to keep labor from migrating, including sharecropping, intimidation, and mass imprisonment.

  2. 2

    For example, the Civil Rights movement involved the federal government in the battle over state and local school desegregation. After this brief period of national exposure the politics of education and labor mobility largely disappeared from the national political debate.

  3. 3

    This is especially true if literacy requirements for voting are in place.

  4. 4

    Labor mobility plays a key role determining how political conflict is organized in society. For example, trade politics are alternately characterized as a conflict between the factors of production (capital, labor, and land) when labor mobility is high or a conflict between sectors (i.e., specific industries) when labor mobility is low. Because labor mobility predicts when either model is appropriate, scholars have spent considerable time and effort measuring labor mobility (Hiscox, 2001, 2002; Ladewig, 2006; Mukherjee et al., 2009). These studies strengthened our empirical knowledge of how labor mobility has changed over time and, in turn, improved our understanding of why political coalitions form in response to economic conflict. While a great deal of attention has been paid to the political influences of labor mobility very little attention has been paid to the political influences on labor mobility, a goal of this article.

  5. 5

    The specific-factors model makes a number of assumptions which if violated will impinge on the predictive abilities of the model. If the owners of land are also the owners of capital then society is best described by a factor model, not a specific-factors model. In such a society, owners are not incentivized to control the flow of labor between factors because they set wages in both of them.

  6. 6

    Both fixed factors, agriculture and industry, have an incentive to influence labor mobility. Urban industries have a difficult time influencing rural education because education policy is primarily made at the state and local level. Similarly, agricultural and industrial laborers have competing goals with respect to labor mobility. Agricultural labors want more mobility so they can move to jobs with higher wages. Industrial workers use their power to restrict labor mobility of rural workers by creating insider–outsider politics (Rueda, 2005, 2007).

  7. 7

    China's Hukou system, which creates a permit system for rural migrants, is a noted exception (Wang and Zuo, 1999; Wang, 2005)

  8. 8

    This article treats land inequality as an exogenous variable, but there is an interesting literature examining the origins of land inequality. Sokoloff and Engerman (2000); Engerman and Sokoloff (2002) argue that modern land inequality in the Americas originates from factor endowments of particular soils and climates. For example, specific combinations of hot climates, rich soils and high rainfalls enabled large and very profitable sugar plantations in the Caribbean. Because the returns to scale are very high in sugar production, highly unequal landholding patterns emerged. See Easterly (2007) for an empirically test of this hypothesis and van de Walle (2009) for an extension to the African context.

  9. 9

    Bates (1981) shows this same dynamic in the African context by studying the effectiveness of farmers in lobbying against price controls.

  10. 10

    Other prominent indices of land inequality include Taylor and Hudson's data set of 54 countries with data points circa 1960 (Taylor and Hudson, 1972). Deininger and Squire (1998) have a land inequality data set with 261 observations of 103 countries; however, only 60 observations have been published (Deininger and Olinto, 2000).

  11. 11

    Examples include Taiwanese and Korean land reforms in the 1950s (Frankema and Smits, 2005; Jeon and Kim, 2000).

  12. 12

    Thirty million is the average population of a country in 2000. On average, 30% of the population is under 15 years old (World Bank, 2008). The remaining 21 million people will score 5% higher on Any Education.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Literature Review
  5. 3. Specific Factors and the Incentive for Political Control of Labor Mobility
  6. 4. Landowner Power
  7. 5. Empirical Analysis
  8. 6. Methods and Results
  9. 7. Conclusion
  10. Acknowledgments
  11. References
  • Acemoglu, D. and Robinson, J. A., 2000, Why did the west extend the Franchise? Democracy, inequality, and growth in historical perspective. The Quarterly Journal of Economics 115, 11671199.
  • Acemoglu, D. and Robinson, J. A., 2006a, De Facto political power and institutional persistence. American Economic Review 96, 325220.
  • Acemoglu, D. and Robinson, J. A., 2006b, Economic backwardness in political perspective. American Political Science Review 100, 115131.
  • Alesina, A. and Glaeser, E. L., 2004, Fighting Poverty in the US and Europe: A World of Difference, Rodolfo Debenedetti Lectures. (Oxford University Press, Oxford).
  • Alesina, A., Baqir, R. and Easterly, W., 1999, Public goods and ethnic divisions. The Quarterly Journal of Economics 114, 12431284.
  • Alesina, A., Devleeschauwer, A., Easterly, W., Kurlat, S. and Wacziarg, R., 2003, Fractionalization. Journal of Economic Growth 8, 155194.
  • Banerjee, A. and Iyer, L., 2005, History, institutions, and economic performance: the legacy of colonial land tenure systems in India. American Economic Review 95, 11901213.
  • Barro, R. J. and Lee, J.-W., 1996, International measures of schooling years and schooling quality. The American Economic Review 86, 218223. doi:10.2307/2118126.
  • Barro, R. J. and Lee, J.-W., 2001, International data on educational attainment: updates and implications. Oxford Economic Paper 53, 541563.
  • Bates, R. H., 1981, Markets and States in Tropical Africa: The Political Basis of Agricultural Policies. (University of California Press, Berkeley).
  • Bensel, R. F. 2000, The Political Economy of American Industrialization, 1877–1900. (Cambridge University Press, Cambridge [England]).
  • Blackmon, D. A., 2008, Slavery by Another Name: The Re-Enslavement of Black People in America From the Civil War to World War II. (Doubleday, New York).
  • Brady, H. E., Verba, S. and Schlozman, K. L., 1995, Beyond SES: a resource model of political participation. The American Political Science Review 89, 271294. doi:10.2307/2082425.
  • Brambor, T., Clark, W. R. and Golder, M., 2006, Understanding interaction models: improving empirical analyses. Political Analysis 14, 6382.
  • Brune, N. and Garrett, G., 2005, The globalization Rorschach test: international economic integration, inequality, and the role of government. Annual Reviews in Political Science 8, 399423.
  • Cohen, W., 1991, At Freedom's Edge: Black Mobility and the Southern White Quest for Racial Control, 1861–1915. (Louisiana State University Press, Baton Rouge).
  • Deininger, K. W., OlintoP. 2000, Asset distribution, inequality, and growth. World Bank Policy Research Working Paper, 2375.
  • Deininger, K. and Squire, L., 1998, New ways of looking at old issues: inequality and growth. Journal of Development Economics 57, 259287.
  • Easterly, W., 2007, Inequality does cause underdevelopment: insights from a new instrument. Journal of Development Economics 84, 755776.
  • Engerman, S. L. and Sokoloff, K. L., 2002, Factor endowments, inequality, and paths of development among new world economies. Economía 3, 41109.
  • Engerman, S. L., Mariscal, E., Sokoloff, K. L., 2009, The evolution of schooling in the Americas, 1800–1925, in: D. Eltis, F. D. Lewis and K. L. Sokoloff, eds., Human Capital and Institutions: A Long-Run View, 1st edn. (Cambridge University Press, New York, NY) pp. 93142.
  • Fearon, J. D., 2003, Ethnic and cultural diversity by country. Journal of Economic Growth 8, 195222.
  • Frankema, E. and Smits, J.-P., 2005, Exploring the historical roots of Eastern Asia's post-war catch-up growth: a trade perspective, 1906–1999. Journal of the Asia Pacific Economy 10, 178194.
  • Frankema, E. 2009, The colonial origins of inequality: exploring the causes and consequences of land distribution, in: S. Klasen and F. Nowak-Lehmann, eds., Poverty, Inequality, and Policy in Latin America, CESifo Seminar Series (MIT Press, Cambridge, Mass) pp. 1946.
  • Galor, O., Moav, O. and Vollrath, D., 2009, Inequality in landownership, the emergence of human-capital promoting institutions, and the great divergence. Review of Economic Studies 76, 143179.
    Direct Link:
  • Gerschenkron, A., 1943, Bread and Democracy in Germany. (Cornell University Press, Ithaca).
  • Greenwood, M. J., 1971, An analysis of the determinants of internal labor mobility in India. The Annals of Regional Science 5, 137151.
  • Greenwood, M. J., 1975, Research on internal migration in the United States: a survey. Journal of Economic Literature 13, 397433.
  • Habyarimana, J., Humpherys, M., Posner, D. N. and Weinstein, J. M., 2007, Why does ethnic diversity undermine public goods provision? American Political Science Review 101, 709725. doi:10.1017/S0003055407070499.
  • Hiscox, M. J., 2001, Class versus industry cleavages: inter-industry factor mobility and the politics of trade. International Organization 55, 146. doi:10.1162/002081801551405.
  • Hiscox, M. J., 2002, International Trade and Political Conflict: Commerce, Coalitions, and Mobility. (Princeton University Press, Princeton, NJ).
  • Huber, E. and Safford, F., eds., 1995, Agrarian Structure & Political Power: Landlord & Peasantin the Making of Latin America. (University of Pittsburgh Press, Pittsburgh).
  • Huber, E., Mustillo, T. and Stephens, J. D., 2008, Politics and social spending in latin America. Journal of Politics 70, 420436.
  • Jeon, Y. and Kim, Y., 2000, Land reform, income redistribution, and agricultural production in Korea. Economic Development and Cultural Change 48, 253268.
  • Klor, E. F. and Shayo, M., 2010, Social identity and preferences over redistribution. Journal of Public Economics 94, 269278.
  • Ladewig, J. W., 2006, Domestic influences on international trade policy: factor mobility in the United States, 1963 to 1992. International Organization 60, 69103.
  • Lind, J. T., 2007, Fractionalization and the size of government. Journal of Public Economics 91, 5176. doi:10.1016/j.jpubeco.2006.09.006.
  • Lindert, P. H., 2004, Growing Public: Social Spending and Economic Growth Since the Eighteenth Century. (Cambridge University Press, Cambridge, UK).
  • Luttmer, E. F. P., 2001, Group loyalty and the taste for redistribution. Journal of Political Economy 109, 500528.
  • Margo, R. A., 1990, Race and Schooling in the South, 1880–1950: An Economic History. (University of Chicago Press, Chicago).
  • Margo, R. A., 1991, Segregated schools and the mobility hypothesis: a model of local government discrimination. The Quarterly Journal of Economics 106, 6173. doi:10.2307/2937906.
  • Mariscal, E., Sokoloff, K. L. 2000, Schooling, suffrage, and the persistence of inequality in the Americas, 1800–1945, in: S. H. Haber, ed., Political Institutions and Economic Growth in Latin America: Essays in Policy, History, and Political Economy (Hoover Institution Press, Stanford University, Stanford, California) pp. 294.
  • Marshall, M. G., Jaggers K. 2008, Polity IV Dataset. (University of Maryland, Center for International Development and Conflict Management, College Park). Available at: http://www.systemicpeace.org/polity/polity4.htm (April 24, 2009).
  • Meltzer, A. H. and Richard, S. F., 1981, A rational theory of the size of government. The Journal of Political Economy 89, 914927.
  • Molloy, R., Smith, C. L. and Wozniak, A., 2011, Internal migration in the United States. Journal of Economic Perspectives 25, 173196. doi:10.1257/jep.25.3.173.
  • Mukherjee, B., Smith, D. L. and Li, Q., 2009, Labor (Im)mobility and the politics of trade protection in majoritarian democracies. The Journal of Politics 71, 291308.
  • Olson, M., 1965, The Logic of Collective Action; Public Goods and the Theory of Groups. (Harvard University Press, Cambridge, Mass).
  • Rueda, D., 2005, Insider–outsider politics in industrialized democracies: the challenge to social democratic parties. American Political Science Review 99, 6174.
  • Rueda, D., 2007, Social Democracy Inside Out: Partisanship and Labor Market Policy in Industrialized Democracies. (Oxford University Press, Oxford).
  • Rueschemeyer, D., Stephens, E. and Stephens, J. D., 1992, Capitalist Development and Democracy. (University of Chicago Press, Chicago).
  • Sjaastad, L. A., 1962, The costs and returns of human migration. The Journal of Political Economy 70, 8093.
  • Sokoloff, K. L. and Engerman, S. L., 2000, History lessons: institutions, factors endowments, and paths of development in the new world. The Journal of Economic Perspectives 14, 217232.
  • Solt, F., 2009, Standardizing the world income inequality database. Social Science Quarterly 90, 231242.
  • Taylor, C. L. and Hudson, M. C., 1972, World Handbook of Political and Social Indicators, 2d edn. (Yale University Press, New Haven).
  • UNU-WIDER. 2008, May. World income inequality database, Version 2.0c. Available at: http://www.wider.unu.edu/research/Database/en_GB/database/ (accessed May 22, 2009).
  • van de Walle, N., 2009, The institutional origins of inequality in Sub-Saharan Africa. Annual Review of Political Science 12, 307327.
  • Wang, F.-L. 2005, Organizing through Division and Exclusion: China's Hukou System, 1st edn. (Stanford University Press, Stanford, CA).
  • Wang, F. and Zuo, X., 1999, Inside China's cities: institutional barriers and opportunities for urban migrants. The American Economic Review 89, 276280.
  • Wegenast, T., 2009, The legacy of landlords: educational distribution and development in a comparative perspective. Comparative Governance and Politics 3, 81107.
  • Wegenast, T., 2010, Cana, café, cacao: Agrarian structure and education in Brazil. Journal of Iberian and Latin American Economic History 28, 103137.
  • World Bank. 2008, World Development Indicators (WDI). (The World Bank, Washington, DC).
  • Zhao, Y., 1997, Labor migration and returns to rural education in China. American Journal of Agricultural Economics 79, 12781287.