A big part of what economists do is work with data to answer questions. I've told you that economics is everywhere, and economists have researched questions from how discrimination affects people's job market prospects, to how peer pressure impacts people's willingness to give to charity, to even how drinking in college impacts grades. Regardless of your major, you should take at least an introductory statistics or econometrics course. Journalists and policymakers take advantage of the fact that most people aren't even aware of basic statistical methods to push some pretty lousy ideas.
We're not going to tackle anything too in-depth in this lecture, just an overview of how economists think about problems. Let's start with a question that might be meaningful to you. Does education increase income?
We know that on average in the data People with more education earn more income. That's a correlation. The two variables are related. They tend to move together.
More education is associated with higher income. But is it causal? Does more education cause higher income?
Economists are very concerned about questions of more. causality? And those are much tougher questions.
The relationship between education and income could be very straightforward. More education causes you to earn more income. That's human capital.
Education makes you more productive, which increases your wages. And it would suggest that policies that encourage more education could have big income effects. But that's not the whole story.
What if there's something else, another factor, that leads to both more education and higher income? These are sometimes called omitted or confounding variables. Omitted because they're not included in our education income model. or confounding because they can lead to misinterpretation. Suppose, reasonably, that people with higher levels of ability get higher levels of education.
That could be raw brain power or conscientiousness and perseverance. Maybe there's no direct relationship between education and income. It's just that high ability people are both likely to get a lot of education because it's easy and maybe even fun for them, and they're the sort of productive workers who get rewarded with higher incomes.
Ability as an omitted variable could make us think that education causes income, when really it's ability that leads to a spurious correlation between education and income in the data, but it's all due to ability. There's no causal relationship. Economists think about the counterfactual.
What would have happened in some other scenario? This is what causality is all about. What do we think would have happened to someone's income if they earned a college degree? compared to their income without a college degree. A good economist, or just any smart person, always asks, compared to what?
Sure, people with college degrees tend to earn more than people without college degrees, but that doesn't mean that taking people without college degrees and having them spend some time in school will have the same effect. We don't have the counterfactual. And much of modern economics is about finding ways to estimate that counterfactual. This is especially important in policy analysis. Someone might say, we introduced this expensive literacy program and test scores went up by 5%.
Compared to what? It's not enough to say things got better than they were last year. What would have happened this year in the absence of the program? Would test scores have gone up anyway?
What could that money have been spent on? Are there other programs that would have done a better job? Notice that this is all related to some of those key economics concepts like opportunity cost and marginal analysis. Let's take what might be an even more complicated question with a lot of policy importance. Because income...
Affect health. We can observe in the data that higher income people have better health outcomes. That's a correlation. The two variables are related. But is it causal?
Does higher income cause better health? Like with our first example, the relationship between income and health could be straightforward. Having more money makes you healthier. That would suggest that policies that transfer income to people could have big health impacts.
But we could have reverse causality. Maybe being healthier causes you to have higher income. Not missing work, being able to save more because you have lower health expenses, and so on.
Health causes income. In this case, policies that transfer income may not do much good for health. And omitted or confounding variables could cause problems here too. Suppose that having parents who have more resources, parental resources, leads people to both be healthier and have better health habits and also to earn more income because their parents enabled them to access more opportunities.
But maybe there's no direct relationship between income and health. We'd still see a positive correlation between income and health in the data, but it's all due to the impact of parents. When we think about these kinds of questions, we want to avoid a common pitfall. Anecdotal evidence or small samples.
We can come up with examples of the low-income chain smoker who lived to be 105 or the rich fitness nut who dropped dead at 40. But economists recognize the need for data with a large number of observations to be able to draw any meaningful conclusions. We're not denying that those counter-examples exist, but it'd be foolish to use them to conclude that being rich and in shape causes high mortality. And this shows us the importance of the counterfactual.
what would have happened in some other scenario? What do we think would happen to someone's health if they earned an additional $1,000 per year compared to their health with their actual earnings? If income improves health, we'd expect their health to be better in this counterfactual world with more money. If causality runs in the other direction, or if there are a lot of omitted or confounding variables, we wouldn't expect to see an impact on health. Economists are at the cutting edge of answering questions with data, tying those questions closely to basic economic principles, and importantly, avoiding pitfalls of mistaking correlation for causation or ignoring counterfactuals.