Leah Bevis is a PhD candidate at Cornell’s Dyson School.
Recent findings by The Equality of Opportunity Project made big press in the US: intergenerational economic mobility in this country is far lower than that of most other developed nations, though it varies widely across states. A couple decades of studies have found similar results: among rich nations, intergenerational mobility is particularly low in the United States and England, and particularly high in northern European countries.
We know far less, however, about intergenerational mobility in developing nations. This is primarily because collecting income data is difficult in poor countries, and collecting income data first for parents and then for their children is very, very difficult. According to work by Gary Salon, a leading expert on intergenerational income mobility, the scant evidence available points towards even higher intergenerational income inequality (i.e., lower mobility) in poor countries than in rich countries.
But even if this is so, we know next to nothing about why intergenerational income mobility exists, or does not exist, in any particular developing context. (In fact, we don’t know much about why it exists in any context.) Is it that richer parents simply pass cash or assets on to their children, in the form of land, animals, or dowry money? Do wealthy, educated parents tend to educate their children more, or keep their children healthier, in order to later earn higher levels of income? Or do families share some sort of invisible productivity level across generations – perhaps via inherited social networks or a taught work ethic? Additionally, most studies examine income transmission from fathers to sons; we know almost (but not quite) nothing about mobility differences for daughters vs. sons.
In my paper with Chris Barrett, “Decomposing Intergenerational Income Elasticity: The gender-differentiated contribution of capital transmission in rural Philippines,” we tackle these research gaps in the context of Bukidnon, a rural, landlocked region of the Philippines. We decompose intergenerational income transmission (measured via elasticity) into five distinct pathways: the intergenerational transmissions of health, education, land, and spouse education, plus a residual correlation in productivity. What’s more, we do this for both sons and daughters. The data we use are a rare example of the elusive, parent-child income panel. In 1984 almost all parents interviewed were farmers. By 2003 some of the wealthier kids had salaried jobs such as government employees or teachers, but most were still farming.
We present two types of analysis. First, we show results for intergenerational capital transmission, for land, education, health, and spouse education. Second, we calculate intergenerational income transmission, and decompose it into capital and productivity transmissions. All of this is done separately for sons and daughters. (Note for economists: The associations between capital transmission and income transmission are non-causal. Nevertheless, they are suggestive of the pathways most likely to underpin income transmission in this context.)
We find strong parent-to-child capital transmission. More educated parents have more educated children, healthier parents have healthier children, and wealthier, more educated parents procure more educated spouses for their children. Of the various types of parent capital, mother’s education is by far the most influential; it is positively associated with child education, spouse education, son landholdings, and later life income.
We also find strong intergenerational income transmission: almost twice as strong (or half as mobile) as that of most developed countries, and approximately equal for sons and daughters. The fascinating bit, however, is how the pathways associated with this transmission vary across child gender. Capital transmission explains pretty much all of the association between parent income and son income; parent income is irrelevant once you know the other parent characteristics.
For daughters, this is not the case. Parents do transmit human and productive assets to daughters, but it doesn’t explain the correlation between parent and daughter income. In fact, even knowing parent capital levels and adult daughter capital levels won’t fully explain daughter’s income – parent income will still help to predict it. We view this residual income correlation as reflecting a “shared productivity level” between parents and daughters. But we don’t know what causes it.
Possibly, this productivity transmission reflects parent intervention in the marriage market: richer parents use social connections or other means to secure “high-value spouses” for their daughters – value that we the economists do not observe. (While we do control for spouse education, there is reason to think that education may not capture “husband value” for daughters as well as it captures “wife value” for sons.)
Our paper offers a few key takeaways. First, it provides more evidence that intergenerational income mobility may be very low in the developing world, more so than in richer countries. Second, it shows that in poor, agrarian contexts the intergenerational transmission of human and productive capital are highly associated with, and likely underpinning, the correlation between parent and child income. Third, and most importantly, it illustrates that even when parent-to-child income transmission appears similar for daughters and sons, the pathways behind this transmission may differ starkly by gender. Understanding these pathways, and their gender-specific nature, is crucial for policymakers who wish to provide all children with reasonably equal economic opportunities in life.
