What this covers

Susan Mayer discusses her research on why direct cash transfers to poor families produce disappointingly small improvements in children's outcomes, despite widespread expectation that money solves poverty. Speaking with Russ Roberts on EconTalk, Mayer walks through the methodological problems that have led economists and sociologists to overstate income's causal power: standard regressions fail to isolate the effect of money itself because parental income correlates with dozens of unmeasured traits—stability, planning horizons, patience—that independently shape child outcomes. She demonstrates this Selection Bias by showing that income measured after a child's test scores or teenage pregnancy predicts the outcome nearly as well as income before it, proving the real causal agent is something stable about the parent, not the dollars.

The conversation spans three main territories. First, Mayer charts what actually predicts child outcomes: not what money buys (better housing, groceries, cars) but parental behavior—how often a parent reads to a child, takes them on outings, enforces routines. These behaviors cost little and barely shift when income rises, explaining the weak income effect. Second, she examines why behavior change is so difficult: present bias, the same orientation that makes people eat donuts despite knowing better, makes parents skip bedtime reading even when they intend otherwise. Third, she proposes that poverty is not monolithic—the transiently poor need different help than the persistently poor, demanding a triage approach rather than universal cash. She sketches low-cost behavioral solutions, including a field experiment that doubled reading time through reminders and goal-setting, and argues that schools alone cannot carry the load since children spend only 13–15% of their waking hours there. Throughout, Mayer challenges the Streetlight Effect: society assumes money is the problem because it is measurable, not because it is causal.

Sharpest takeaway

Mayer argues that once basic material needs are met, transferring additional money to poor families has little measurable effect on children's outcomes, because the parental characteristics and behaviors that drive those outcomes—not income itself—are the real causal factors, so genuinely helping vulnerable children requires changing parenting behavior rather than just raising income.

  • Standard regressions overstated income effects by failing to control for stable parental characteristics correlated with income
  • Income after an outcome appears to affect it just as much as income before, revealing the effect is an artifact of unmeasured parental traits
  • Behaviors that predict children's outcomes (reading, outings) cost little and don't change much when income rises

The argument · threads6 threads · 31 claims
0.79

Income itself has little causal effect on child outcomes because stable parental traits, not dollars, explain the apparent income correlation.

4 pointscentrality 5/5
  • Standard OLS regressions of child outcomes on income are inadequate for isolating the effect of income because parental income is correlated with a host of other characteristics that also influence children's outcomes, and it is essentially impossible to control for all of them since we don't even know the full list.

    “this is sort of inadequate for isolating the effect of income because a parent's income is correlated with a whole bunch of other characteristics that might also influence children's outcomes. And, it's essentially impossible to control all those possible things. We're not even sure what that list of possible things is.”

  • Comparing income measured before versus after an outcome (e.g., a child's reading scores or a teenage pregnancy) isolates bias: only income before an outcome should be able to affect it, yet income measured after the outcome also appeared to have a substantial effect, proving that stable parental characteristics—not the dollars—were driving the apparent income effect.

    “income after those events had--appeared to have--a substantial effect on those outcomes. Meaning, that the stable parental characteristics were likely to be--were likely to bias the estimates”

  • If income itself matters, the source of a dollar should not matter, yet 'other income' (alimony, lawsuit settlements) that is only weakly correlated with parental characteristics has a much weaker effect on child outcomes than earnings or welfare income, which are strongly correlated with parental characteristics—again indicating the apparent income effect reflects parental traits rather than purchasing power.

    “if income matters, then the source of income doesn't matter. A dollar is a dollar... It turns out that some sources of income are more correlated with parents' characteristics than other sources”

  • As parents' income rises they buy more and better goods and services—better housing, more groceries, eating out, better cars—but these are not the things that predict children's outcomes; what predicts outcomes is behavior like how much a parent reads to a child or takes them on outings, which cost little and don't change much with income, explaining why there is no large income effect.

    “as parents' income goes up, they do buy more goods and services... But those are not the things that predict kids' outcomes. Things that predict kids' outcomes are things like how much the parent reads to the child, whether the parent takes the child on outings. And it turns out those kinds of things don't cost that much money”

0.73

Schools cannot be the primary lever because children spend only 15% of waking hours there; family behavior during the other 85% matters far more.

3 pointscentrality 5/5
  • Children spend only about 13 to 15% of their waking hours between ages zero and 18 in school, so even though improving schools helps, the silver bullet cannot be in schools—we have to do something about families, where children spend the other 85% of their time.

    “What percent of waking hours between the ages of zero and 18 do you think children spend in schools?... it's between about 13 and 15%. It's not much... the solution, if you're looking for a silver bullet... is not going to be in schools. We've got to do something about families.”

  • There is much less variance in the quality of schools than in the quality of families, and society latches onto schools because they are institutions we know how to change, measure, and work with—but the sustainable, less costly solution is getting families to do the things they already know they ought to do.

    “there's much less variance in the quality of schools already than there is in the variance of the quality of families. So I think we latch on to schools because there's a--it's an institution. We know how to change institutions.”

  • There are three families of theory for why poverty harms children: the investment theory (rich parents invest in what children need and get a return), the good-parent/behavioral theory (low income depresses and impairs parenting), and the structural theory (bad neighborhoods and low-income peers in schools harm children's chances)—all of which imply money can purchase better child well-being.

    “among economists, it generally sort of follows from Becker's work--the idea that rich parents have money to invest in the things that children need to succeed; they make that investment, and they get a return on that investment”

0.71

Changing parental behavior like reading to children is hard because of present bias and positional competition, but low-cost behavioral nudges can work.

5 pointscentrality 4/5
  • Every parent—rich or poor, educated or not, immigrant or not—says they should read to their child and play educational games, so the gap in child-rearing is not an information problem; the real mystery is why advantaged parents do these things while disadvantaged parents who want to do them don't.

    “every parent, rich, poor, educated, not educated, immigrant, not immigrant, every parent says, 'I should read to my kid every day.'... It's not that parents don't know what to do. It's that they just don't. And then the big mystery is why?”

  • The norm that parents should read to their children has become extremely dominant—virtually every low-income Head Start parent says they should and expects a cognitive return—thanks to Head Start and reading campaigns, but the norm shift has not translated into actual reading behavior; the harder remaining step is getting non-reading parents to actually read.

    “virtually every parent says they should read to their kid... It means that the norm of reading is very dominant... I think we can thank Head Start and all the reading programs that have been tried for raising the consciousness about reading.”

  • By reminding parents to read, telling them exactly how much they actually read (countering their optimistic self-estimates), and getting them to set and stick to goals, Mayer's experiment doubled the amount of time preschool parents read to their children in a non-invasive way that parents liked.

    “By telling them exactly the truth about that and reminding them not to--to set a goal and stick to it, you can change behavior. We doubled the amount of time that preschool parents read to their kids using those techniques.”

  • Behavior with long-term rather than short-term payoffs is the hardest to change because of present bias—the same orientation that makes people reach for the donut despite knowing it harms their health is what makes parents fail to read to their kids, enforce tooth-brushing, or serve vegetables, even though they intend to.

    “that kind of, we'll call that present bias, present orientation, is one of the most important things that makes parents not do the things they think they should do for their kids. So, not read to their kids each night, not make their kids brush their teeth each night, throw pizza on the table rather than vegetables.”

  • Parenting is in part a positional competition: if no parent read to their kids it wouldn't matter, but because advantaged parents read to theirs, it becomes important for disadvantaged parents to do so too—so norms keep rising and disadvantaged parents are always somehow behind the curve, even with diminishing returns.

    “parenting is in some sense like competition, right? If no parent ever read to their kid, it wouldn't matter if parents read to their kid. But when advantaged parents read to their kids, it becomes really important for disadvantaged parents to read to their kids.”

0.71

No single solution exists because income, neighborhoods, and inequality each proved insufficient, requiring fundamentally different multi-pronged approaches.

5 pointscentrality 4/5
  • Across her career Mayer hypothesized neighborhood effects, then income, then relative income/inequality as the cause of poor children's outcomes, and each turned out not to be the silver bullet—leading to the conclusion that there is no single solution, which should not discourage efforts but means helping vulnerable kids requires a fundamentally different, multi-pronged approach.

    “I thought it was neighborhood effects; and that turned out maybe not to be the silver bullet. Then I thought it was income and that turned out not to be the silver bullet... Well, that turned out not to be the silver bullet. And so I'm now thinking, well, there's not a silver bullet”

  • In the early 1990s the academic consensus was that income—literally dollars—had a big effect on child outcomes, implying that simply transferring money (the metaphor of a plane dropping money over a neighborhood) would dramatically change children's lives and make the poverty problem cheaply solvable.

    “the mental experiment here is, what if a plane flew over a neighborhood and just dropped money down? Now, how much would the lives of the children in that neighborhood change?... I think the consensus at that point in time was that it would change a lot”

  • Historically, policymakers treated income itself as the problem and money transfer as the obvious solution, but refused to do it for fear of disincentive effects—that giving money to adults would reduce their inclination to work and marry.

    “people didn't want to do that because of the potential disincentive effects or the possibility that transferring money to adults would reduce their inclination to work and to marry and to do the things that we expect of people.”

  • Transferring money, although it sounds difficult, is actually one of the easiest and most efficient things governments can do, and is cheap compared to alternatives like improving schools or neighborhoods—so if income were the cause, the solution would have been simple and inexpensive.

    “it's one of the easiest and most efficient things that governments can do.”

  • A great deal of poverty is in suburban areas, and most poor people—even most long-term poor—do not live in concentrated-poverty neighborhoods, which makes geographically-focused programs like the Harlem Children's Zone hard to apply to the dispersed long-term poor.

    “a great deal of poverty is in suburban areas, for example... most poor people don't live in an area like that. And most even long-term poor don't live in an area like that.”

0.65

Money transfers alone are ineffective for most poor families; targeted help must fit what specific families actually need.

4 pointscentrality 4/5
  • Giving the very poorest, most deprived parents money for basic material needs can help their children, but in general simple money transfers are not the most effective way to solve the problems associated with growing up in poor families.

    “Giving the very, very, poorest, most deprived, parents money to feed their children and provide basic material well-being can help their children. But, in general, simple transfers of money is probably not the most effective way to solve the problems that we associate with growing up in poor families.”

  • If you want to help poor children you have to actually know what help they need rather than assuming the solution is money; finding that income matters little does not mean we should reduce transfers, but if the goal is to improve test scores, schooling, and future earnings, we should look at factors other than raising income.

    “if you want to help poor children, you have to really know what help they need. You can't just think you have the solution--that would be money. You have to really have the solution, which might be something else.”

  • Because people need different things depending on which kind of poor they are, the poor should be triaged: for those who fall into temporary poverty the best help is probably just giving money, whereas a family poor for five years needs a much more intensive social-worker model to get out of poverty.

    “this idea of sort of triaging the poor is a good one, where, when people fall into poverty, probably the best thing you can do is just give money... But that's really different from what a family who's been poor for five years needs. They may need much more social worker kind of model”

  • Simply reminding parents to get their kids to school and giving them simple tips—such as putting the child's backpack by the door the night before—reduces chronic absenteeism by something like 20% at almost no cost, illustrating that we should be more open-minded about low-cost behavioral solutions.

    “just reminding parents to get their kids to school and giving them some tips, simple things, like, 'Put your kid's backpack by the door the day before, the night before school,' reduces chronic absenteeism by something like 20%. It's literally almost costless.”

0.52

Poverty is mostly temporary and dispersed, not concentrated or permanent, making class-based assumptions and geographically-targeted programs misaligned with reality.

4 pointscentrality 3/5
  • The view of America as a fixed class society is wrong; only about 5-6% of people who begin a spell of poverty are still poor three years later, and roughly 60% of the adult population has been poor at some point, so any single year's income measure is very unrepresentative and conflates the transiently poor with the persistently poor.

    “now I think the number of people who begin a spell of poverty who were poor three years later, is only something like 5% or 6%.”

  • Most people who are poor are not poor for very long and manage to be perfectly good parents through periods of low income, which shows that some low-income parents can be very good parents and some high-income parents can be quite bad parents.

    “most people who are poor aren't poor for very long. And they manage to be perfectly good parents through periods of low-income. And so we know that some low-income parents can be very good parents and we know that some high-income parents can be really bad parents.”

  • Changing behavior permanently is very different from changing it during an active intervention, and almost nothing scales—both the durability of behavior change after a program ends and the failure of interventions to scale are enormous, recurring problems in the field.

    “changing behavior permanently is really different from changing behavior for the period in which an intervention is actually active. So that's a big deal. Scaling--almost nothing scales. That is a huge, huge issue.”

  • If everyone had a college education we would still have great variation in earnings and wages, because other characteristics related to productivity would become the new triage criteria—so increasing the number of people going to college does not by itself reduce inequality; the marginal college-goer differs from the inframarginal one.

    “Imagine everyone has a college education. We will still have great variation in earnings and wages because other things will become more important. Because the last person to get an education is going to be really different from that first person who got an education”