Incrementalism Beats YOLO

Are incremental policy change and its handmaiden, technocratic social science, doomed to failure? In these hyperpolarized times, there’s a growing view that the quality of life can’t be improved by modest policy interventions that are limited with respect to scope and scale. In this view, interventions need to be bold and broad, or the status quo will inevitably reassert itself. 

Of course, the evidence base for “bold and broad” interventions is often nonexistent, so what is really being advocated is a giant leap into the unknown — what one might call, for lack of a better term, the “you only live once” (YOLO) approach to policy. 

We disagree.

For starters, there is no shortage of important, well-documented examples in which narrow, incremental social policies have made the world better. For example, over the 20th century, most states incrementally raised their compulsory schooling laws (the minimum age to leave school). This is no kitchen-sink policy: It doesn’t change who’s a teacher, how schools are organized, which students attend which schools or a child’s family and neighborhood environments. Nor is this a YOLO policy: It was promulgated piecemeal without disrupting existing school systems.

The result? Did the world somehow reshape itself to counteract the intended effects, rendering the policy futile? No. The best available social science shows that mandates do cause young people to stay in school, and each extra year of schooling boosts earnings by 10-14%, increases wealth by 15%, improves health, increases voting and reduces incarceration rates by 0.1% for white people and by nearly four times as much (0.37%) for Black people. 

In other words, it is not true that only YOLO policies can help. Modest reforms, though narrow in scope and scale, can be effective, and, importantly, can be evaluated even if implemented at the population level. Indeed, a major development in social science is the ability to isolate causal relationships from “natural experiments” — what Joshua Angrist and Jörn-Steffen Pischke referred to as the “credibility revolution.”

In the case of compulsory schooling laws, the research lagged the policy, but there are plenty of examples of research leading policy as well. Consider, for example, the Mexican conditional cash-transfer program known as Oportunidades (now Progresa), another narrow intervention. It was rolled out as a randomized controlled trial (RCT) in 1998-1999. The program made payments to mothers based on children’s school attendance, preventive health care and use of nutritional supplements. The RCT evidence indicated positive results in all three domains. Longer-term results have been estimated using quasi-experimental methods and remain positive years later. The program not only had encouraging impacts but was also subsequently adopted nationwide in Mexico and then by over 50 other countries.

A coda to this story highlights how “negative knowledge” is also valuable for policy. When New York City launched a conditional-cash RCT, the results were spottier than in Mexico. That may be due to differences across contexts in the starting point; New Yorkers already had relatively high rates of school attendance and preventive care, for example. Learning that the setting matters is not a “failure” or a shortcoming of the incrementalist approach, but rather is key information for policy design — just as is true for many circumstances outside the social realm: For example, electric vehicle batteries don’t work well in very cold weather, but that doesn’t mean we should scrap them.

We offer a final example of how incremental change with high-quality evaluation can lead to greater effectiveness. A behavioral science-informed program by Chicago nonprofit Youth Guidance, Becoming a Man (BAM), works with teens in school to help them better recognize when their automatic responses may be unhelpful in difficult, high-stakes situations (“metacognition”). Two separate RCTs showed BAM can reduce violent crime arrests by 45-50%. However, those impacts did not persist for long after the program ended, and as BAM increased in scale, its effectiveness may have declined (although as Monica Bhatt and her co-authors 2021 note, the statistical uncertainty band here is quite wide). 

In response to these findings, the City of Chicago learned the active ingredient (behavioral science) can help, but some policy adjustments might also help. BAM reaches teens in school, but the data showed that the youth involved with gun violence are disproportionately disconnected from school. So the city realized a slightly more intensive program may be needed to generate more lasting effects. The result was Choose 2 Change (C2C), run by YAP and Children’s Home & Aid (now called Brightpoint). This does not throw the kitchen sink at people, since that leads costs to skyrocket and hence the number of people helped to plummet. C2C instead expands its program scope incrementally by adding a mentor to the behavioral science. 

An RCT shows that, 24 months out, violent-crime arrests declined by 39%; there were impacts even 36 months out. Partly informed by these results, the city in turn mounted a similarly intense program for adults at high risk for gun violence involvement, Heartland Alliance’s READI Chicago, which RCT evidence suggests is a promising policy. Similar interventions to these have even been found to be effective when delivered at scale within detention settings.

As an approach, incrementalism guided by policy research has yielded dividends. The world has gotten better. Since 1960, life expectancy increased by 10 years and child poverty plummeted while schooling completion soared. Since 1990, the murder rate has declined, and property crime rates have fallen to a fraction of former highs. Since 2010, overall imprisonment rates have declined, as did Black-white disparities in imprisonment. Incremental policies have been key to this progress, aided and abetted by technocratic scientific input. 

Moreover, what’s the alternative? Incrementalism is about “try it and see what happens.” When those policies fail, the emphasis on feedback lets us know and policy change is easy because policies are changing in small steps. But when YOLO policies fail, they fail spectacularly. HUD eventually spent billions of dollars to tear down what it had a few decades before spent billions to construct: high-rise public housing, dreamed up originally by progressive housing reformers. China’s one-child policy, designed to solve a demographic problem, may lead to an economic disaster. The collectivization of farms (as in Russia) and movement of urban populations to rural areas (as in Cambodia) started with high-minded intentions. In a world full of unintended consequences, what are the effects of defunding the police or ending capitalism or deporting millions of immigrants or ending NATO? The honest answer is “who knows?” Things can always get worse — far worse.

There is a time to embrace radical action, such as when a large asteroid is discovered about to crash into Earth. But in our quest for a happier, healthier, fairer society, radical systemic change is unlikely to provide the improvement we desire. It is a leap into the unknown. Good luck. YOLO, but for how long?