It’s obvious to anyone reading this blog: good education is important. But wait? Why exactly? Of course, we want children to learn to read, write, and do arithmetic, and hopefully much more. But today I also want to talk about something that is sometimes a more uncomfortable topic in education: the economic argument. Those who learn more at school also earn more later in life. But how certain is that?
We have known for some time that people who stay in school longer earn more on average. Research also shows a link between adults’ cognitive skills and their income. But can we also demonstrate that what children learn in primary school, for example, correlates with what they earn fifteen or twenty years later? Das and colleagues investigated this in a new NBER working paper. Their results are interesting, although, as is often the case, they need some nuance. After all, this is still my blog post.
From primary school to the labour market
The researchers used two large longitudinal studies from four distinct countries: Ethiopia, India, Pakistan, and Peru. By combining these datasets, they obtained data on 2,852 people, with the first measurements taken when the children were approximately twelve years old. In Pakistan, the children were on average slightly younger at this first measurement.
The researchers looked at their learning performance. In Ethiopia, India, and Peru, this specifically concerned mathematics, while in Pakistan it involved a combination of mathematics and language. Subsequently, in follow-up measurements, they examined how these children fared later as young adults. In doing so, they asked how long they had attended school and whether or not they participated in higher education. Were they working in the meantime? How much did they earn?
The participants were approximately 25 to 29 years old at the last measurements. We are therefore talking about a fairly long period between the first measurement and the final survey of their income. A period in which a lot can happen, of course. But still…
This is immediately one of the strong points of this research. We have quite a few studies that link learning performance at a young age to later school results, but research that follows the same children until they actually enter the labour market is much scarcer. Especially in low- or middle-income countries, such as some in this study.
Perform better, stay in school longer, and earn more?
The initial results will likely not come as a real surprise. Children who performed better around the age of twelve turned out to have pursued a longer period of education on average later in life. A one-standard-deviation difference in learning performance was associated with approximately 1.1 to 1.5 years of extra education. Also, the likelihood of someone having pursued higher education or vocational training was 9.3 to 13.5 percentage points higher. These associations were found in all four countries.
Now what about earnings, or rather, income? Here, in all but one of the four nations, there is a definite positive relationship. An individual showing one standard deviation greater learning in childhood earns about 8% more as an adult in Pakistan, and between 13% and 14% in India and Peru.
It is also remarkable that the positive relationship did not seem to hold only for kids at the upper end of the learning scale. Across the range of learning scores in the three nations, this positive relationship appears linear. It means that good earnings were not exclusive to the brightest kids.
But you already noticed an exception in the datasets: Ethiopia. Here, no statistical relationship was found between learning performance at a young age and later income. The explanation may lie in the civil war in the country at the time of the latest study. But that remains a possible explanation, not a proven one.
Moreover, there is something worth paying attention to. Better learning performance did not predict whether a person would be employed. Rather, the relationship was with how much people earned, provided they had earnings in the first place. I will have to get back to this matter, as it has some nuances and is not a minor issue…
Because how does that actually happen?
Up to this point, you might think that we are mainly getting a confirmation of what we already suspected. Those who study better and attend school longer may obtain a higher degree and therefore earn more later on.
However, the story turns out not to be so simple. The researchers investigated the extent to which the correlation between learning performance at age twelve and later income could be explained by two factors: the number of years of education a person ultimately completed and how well that person performed on cognitive tests at a later age. You would expect those two factors to explain much of the relationship. And they do, but they don’t explain everything.
In Pakistan, the researchers were able to explain about 50% of that correlation; in India, 56%; and in Peru, only 18%. This means that learning performance around age twelve was still correlated with later income, even after accounting for years of education and how well that person scored on cognitive tests later.
But does this mean that what you know at age twelve has lasting economic value, even if you later receive the same amount of education and perform comparably on cognitive tests? Perhaps. We cannot say this with certainty based on this research.
Why is that? The researchers see several possible explanations. It is possible that tests at a later age do not measure all the skills that matter. Another possibility is that learning performance at a younger age is related to other characteristics that prove important later in the labour market, but which were not included in the study. Moreover, another complication is that in three of the four countries, cognitive skills were measured around age 19, while income data were not collected until almost ten years later. Of course, a great deal could have happened in the meantime.
The researchers themselves acknowledge these limitations as well. Their research shows that years of education and later cognitive skills cannot explain the full correlation. Why is that? We do not yet know that with this.
And now for the nuances
Up until now, I have consistently spoken about correlations. That is, of course, no coincidence. After all, it is not an experimental study. I don’t think that would be possible either. Imagine if the researchers randomly assigned children to learn more or less to see later how much they would earn… No, they have investigated how existing differences in learning performance relate to later outcomes.
In doing so, the researchers take into account various background characteristics, such as the educational level of the parents and the wealth of the family. This strengthens the analyses, but it would be an illusion to think that all possible explanations have been ruled out.
For example, students who do well in school might be different from their peers in other respects. This could be in terms of their home environment, other abilities, or even just the opportunities available to them. Hence, we cannot simply say that if we increase a child’s performance by one standard deviation, that child will earn 8 to 14% more in the future.
There are still some limitations. The samples are not representative of the entire population of the four countries. Moreover, the study concerns relatively young adults, whose further income development has yet to take place. As I indicated earlier in the text, because the income analyses only include people who actually had an income, we must be particularly cautious with conclusions, for example regarding women. In some of the countries studied, their labour participation was particularly low.
The authors themselves believe that the estimates may even be conservative, since the economic payoff of skills can continue to rise further in a career. This is certainly possible, but I would be careful. After all, unobserved heterogeneity among children can also exaggerate the apparent effect.