Pro-Poor Land Reforms, Education and Fertility: International Evidence, 1820-2010��

​

Jakob Madsen (UWA), Lan Anh Tong (Deakin U), Mehmet Ulubasoglu (Deakin U)

​

THE VIETNAM ECONOMIST ANNUAL MEETING 2023 (VEAM 2023)

26-27 Jul 2023, Da Nang, Vietnam

​

This paper...

• We investigate the effects of pro-poor land reforms on education and fertility using panel data from 104 countries over the period 1820-2010.
• Banerjee, 2000; Proto, 2007; Galor et al., 2009; Basu et al., 2010; Gersbach and Siemers, 2010; Besley et al., 2012; Yuki, 2016)
• We use data on land reforms from Bhattacharya, Mitra and Ulubasoglu (2019), who coded 372 major land reforms enacted in 165 countries during the period 1900 to 2010, including their motives and implementation status.
• We examine the effects of pro-poor land reforms on:
• school enrollment rates at primary, secondary and tertiary levels of education
• fertility; if pro-poor land reforms promote education, we would expect fertility to decline (QQ-tradeoff): change in the returns to education or easier access to credit enabled by land reforms will simultaneously change the gradient of the QQ-tradeoff that will favor quality (education) over quantity (fertility)

​

• We focus on pro-poor land reforms because:
• many land reforms are not intended to benefit the poor and are undertaken for other reasons, such as regulatory reforms aimed at improving land administration, e.g. reforms that formalize informal and customary land systems

Land reforms and economic development

• Land reforms have been crucial for human capital development and industrialization (Galor et al. 2009 and Baten and Hippe 2018).
• Large landowners have an interest in keeping the agricultural labor force uneducated
• Pro-poor land reforms have the potential to reduce the adverse influence of large landowners on education

​

• Pro-poor land reforms enable the new landowners to gain access to credit to finance education and to counteract the foregone income from child labor (Galor and Zeira, 1993; Checchi and Garcı́a-Peñalosa, 2004; Garcı́a-Peñalosa, 2010; Madsen et al., 2018)

​

• Proto (2007) and Gersbach and Siemers (2010) suggest that income from tenurial contracts or land ownership promotes education of the offspring of poor previously credit-constrained dynasties, who eventually find work in the formal manufacturing sector

​

• Survey evidence for Taiwan indicates that the mid-1950s land reform was influential for educational aspirations (Wu, 1977)
• 1952, for example, parents had to be reminded, swayed, and threatened to obey compulsory school attendance laws
• By 1959, after the land reforms had been implemented, education was considered economically feasible and considered by parents to be necessary as preparation for better job opportunities of their children in the urban sectors

​

Literature on land reforms, education, and fertility

• Besley and Burgess (2000):
• use panel data for 16 Indian states over the period 1958-1992
• assess the impact of land reforms on wages and per capita income
• find that while land reforms impact agricultural wages positively, but there is no robust relationship between per capita income and land reforms

​

• de la Croix and Doepke (2003): show theoretically that a reduced income inequality simultaneously reduces fertility and promotes education because more weight gets placed on families that value high education and low fertility

​

• Galor et al. (2009):
• show theoretically that the distribution of landownership is a barrier for a growth-promoting educational policy, provided that landowners have sufficient political power
• Using US state-level data over the period 1880-1940, they find that land inequality adversely affects public investment in education

Land Reform Data

• Major land reforms are significant amendments to key laws and directives that changed the course of the fundamental rights of land ownership, land tenure and land use (Bhattacharya, Mitra and Ulubasoglu 2019).
• All land reforms are categorised into 12 different motives: expropriation, landholding ceiling, redistribution, distribution, tenure security, consolidation, nationalisation, privatisation, collectivisation, restitution, CICRT, other.
• An alternative categorisation is pro-poor vs non-pro-poor. (140/372). A land reform is pro-poor if the stated law or directive contains the following words regarding the targets and beneficiaries of the initiative: Landless; poor; landless agricultural labor; bonded labor; marginal farmers; reduce poverty; peasants; subsistence peasants; and subsistence farmers
• Land reforms are also categorised with respect to their implementation status. Implemented and not implemented.
• Land reform is coded one in the year of enactment/implementation

Land Reform Enactments Around the World, 1900-2010

Econometric Framework�

Data

• Gross enrollment rates are obtained from Lee and Lee (2016) and are available in five-year intervals over the period 1820-2010 for 104 of our land reform-country sample
• Fertility data from Delventhal et al. (2021)
• Land reforms data:
• extend land reforms data of Bhattacharya, Mitra and Ulubasoglu (2019) to include the period 1820-1899:
• We identify 388 land reforms in our extended sample, of which 141 are identified as being pro-poor

Identification strategy�

• Endogeneity problems:
• the land reforms and education policies are not random
• education may promote land reforms because an educated population is more likely to push for land reforms than an uneducated population
• the omission of unobservables, such as policies that simultaneously implement land and educational reforms

​

• To address endogeneity, we:
• use US influence, or CIA supported initiatives, as an instrument for pro-poor land reforms
• undertake falsification tests in which non-pro-poor land reforms are used as principal explanatory variables

​

• IV:
• US influence is measured by CIA supported initiatives to successfully intervene in a foreign country to either install a new leader or supported an existing leader to help maintain the power of the regime
• the data (compiled by Berger et al., 2013), span the period 1947-1989 for 97 countries in our sample for which overlapping data are available
• a country-year in which the CIA was successful in its intervention takes the value of 1, and zero otherwise
• we accumulate US interventions over time (Like for the measurement of pro-poor land reforms)
• lag the US influence stock by one year to allow for the time-lag in the legislative process of land reforms

US influence as an IV

• The selection of countries where the CIA undertook the covert operations was not for the purposes of influencing land reform per se.

​

• It is for the purpose of installing or supporting CIA-preferred governments that would help thwart perceived communist threats (the National Security Council Directive No. 4, 1947):
• land reforms were not the primary purpose of US foreign policy makers
• countries selected by the CIA were countries that, in the CIA’s judgement, faced the biggest threat of communist revolts
• pro-poor and redistributive reforms were cost-effective preemptive inventions against communist revolutions in the Cold War era
• Such preemptive measures were commonly undertaken during the Cold War as a counter to a communist threat or social unrest (Acemoglu and Robinson 2000; Weyland 2010; Aidt and Jensen 2014; Madsen et al., 2018).

​

• Over the period 1947-1989, the CIA successfully exerted its influence in 43 countries, leaving 61 countries in our sample without its influence in the entire period

​

• IV methods may result in biased treatment effects if subjects are preselected based on their received treatments:
• there are no obvious reasons why US pre-selects countries with higher land inequality (more likely to undertake land reforms, Alesina and Rodrik, 1994 and Persson and Tabellini, 1994) to exert influence
• the correlation between US influence stock and land inequality is negligible, 0.05

US influence as an IV��Figure 1: Average accumulated pro-poor land reforms�

Data

Descriptive statistics

Event Study

• To test the parallel trend assumptions (i.e., enrolment rate between treatment and control should not be different before the enactment of pro-poor reforms)
• To track the dynamics of the effects of pro-poor land reforms.
• We focus on the first pro-poor land reform of each country only, because the second or the third reforms may follow the previous reform after a short period where the effect of the previous reform is still ongoing.

​

Event Study Plots

Regression results�FE-OLS regressions�

Regression results�IV regressions

Regression results: OLS vs IV estimates

• The coefficients of LR in the IV regressions are larger than their OLS

​

• Reasons that OLS regressions are downwardly biased:
• feedback effects from education to land reforms: An increasing educational standard that is associated with increasing industrialization reduces the need for land reforms because poor farmers have alternative work opportunities in cities where earnings are traditionally higher than in the countryside (Kuznets, 1955).
• omission of technological progress from the estimates. In the Unified Growth Theory framework:
• the tradeoff between education and fertility is predominantly driven by the returns to education, which in turn is predominantly driven by skill-biased technological progress in manufacturing (Galor and Weil, 2000; Galor and Moav, 2002; Galor et al., 2009).
• skill-biased technological progress simultaneously increases income and school enrollment and, consequently, reduces the need for land reforms.

​

Falsification tests �

• We undertake 2SLS regressions in which land reforms are subdivided into 12 main motivational categories.
• expropriation,
• redistribution,
• distribution,
• Imposition of landholding ceiling,
• restitution,
• consolidation,
• improving tenure security,
• privatization,
• collectivization,
• nationalization,
• CICRT (recognizing customary, indigenous, community, religious and traditional owners) land rights, and other

​

• The first three land reforms are pro-poor land reforms in that the reforms are mostly undertaken to transfer the land from rich to poor

Falsification tests

Family farms as a mediator �

• The land reform literature suggests that parents’ decisions about their children’s schooling is heavily influenced by their wealth, which is likely to increase substantially among the poor in the wake of pro-poor land reforms (Galor et al., 2009).

​

• To check for wealth distribution as a mediator of land reforms to education, we carry out Sobel-tests (Sobel, 1982), where the fraction of agricultural land owned by family farms is used as the mediating variable.
• we test whether family farmland significantly channels the influence of pro-poor land reforms to GERs, where the family farm data are from Vanhanen (2003).

​

• The family-farm land share is probably the best available historical proxy for the wealth distribution of the rural population.
• credit is essential for education, and access to credit is often conditional on land wealth as a collateral (Galor and Zeira 1993)

​

• If the family farmland share of agricultural land is mediating the effects of pro-poor land reforms on school enrollments, then we would expect
• family-farmland shares to be positively affected by pro-poor land reforms;
• pro-poor reforms to lose their significance as determinants of school enrollment when family-owned farmland is included in the regression and is significant.
• the Sobel-test meets these two conditions.

​

Family farms as a mediator �

Fertility and pro-poor land reforms

Check for exclusion restrictions and inclusion of confounders�

• We check whether the exclusion restriction is violated, or the coefficients of land reforms are subject to endogeneity biases due to omitted variables by controlling for:

​

• per capita income,
• democracy,
• KGB/Soviet influence as confounders.

​

• Estimates of the effect of land reforms on outcomes are largely the same
• We run the zero-first-stage test to check exclusion restriction:
• in a subsample for which the first stage (that is, the effect of the IV on the treatment variable) is zero, the reduced form (that is, the effect of the IV on outcomes) should be zero too if the exclusion restriction is satisfied.

​

• the zero-first-stage test passes for all education enrolment rate and fertility, for enacted vs implemented pro-poor LR, for contemporary and 10 years lag effects.

Discussion

• Pro-poor land reforms result in:
• Higher education enrolment (primary, secondary, tertiary)
• Lower fertility rate
• Implemented land reforms have larger effects on outcomes than enacted (almost double in magnitude for all outcomes)
• Family farm share is the mediating factor translating the effect of land reforms to education and fertility outcomes.

​