There is no easy fix to the dropout problem


This page in:

This page in: Espanol
Students at secondary school. Photo: © Charlotte Kesl / World Bank

When I joined the Mexican ministry of education in 2008, one of the first challenges I had was to identify effective policies to reduce dropout rates in upper secondary (grades 10, 11 and 12). Eight years, two randomized control trials, numerous workshops, and several diagnostics later, I still don’t have a precise answer. 

No easy fix
In 2008, less than six out of every ten students enrolled in 10th grade in Mexico graduated. With the expansion of the conditional cash transfers (CCT) program in Mexico  showing a positive effect on retention rates in lower secondary, and surveys indicating “economic constraints” as the most important reason behind dropping out an obvious strategy was therefore to set up a scholarship program for upper secondary.
We did so along with an evaluation strategy to identify if the program was indeed reducing dropouts. In an ongoing paper with Orazio Attanasio and Costas Meghir, exploiting a randomized control trial, we show that, between 2009 and 2012, the upper secondary scholarship program did not have a significant effect on graduation rates nor did it raise math or Spanish learning outcomes.
While I was still working with the Ministry of Education, I came across a well-known paper by R. Jensen showing that providing secondary school kids in the Dominican Republic with information about the returns to secondary schooling was enough to reduce dropout rates.
In 2009, we invited Jensen to Mexico City to help us set up the intervention and the evaluation strategy (which relied on a randomized control trial). Again, this intervention did not work as expected. Using the census-based standardized test Enlace, in a recent working paper, Ciro Avitabile and I show that providing students with information on the economic benefits associated with upper secondary was not enough to reduce dropouts in Mexico. As a consolation, the information intervention had a positive, significant and economically large effect on learning outcomes, especially among girls.
What is going on?
Why aren’t these two policies which, ex-ante, seem to be sensible, working to reduce dropouts? I think that part of the explanation is related to our limited understanding of the causes behind dropouts in upper secondary. Although dropout rates in Mexico and most Latin American countries manifest in upper secondary, the root of the problem can be found earlier in the life and education trajectories of those that fail to graduate.
Inequality and poverty in Mexico and the rest of Latin America are unacceptably high. Young children from poor households in Latin America are not receiving the early stimulation needed to accumulate cognitive and socioemotional skills throughout life. In extreme cases where pregnant women are not well-nourished, the poorest babies are at a disadvantage from birth. When these kids enter the formal education system  they already have a significant skills deficit compared with children from non-poor households. To make things worse, children from the poorest households in Latin America tend to go to the worse performing schools with deficient infrastructure and learning materials and, perhaps more important, with unqualified teachers. Inequalities at birth are thus exacerbated as a child grows, instead of being reduced or leveled.
When poverty and low cognitive skills are combined with lack of adult guidance and supervision, the chances of upper secondary graduation are slim. When the dropout problem is viewed as the outcome of a more structural, deeper problem, it is not surprising that a monthly transfer of $40 or information regarding the economic benefits of schooling is not enough to fix the dropout problem.
What can be done?    
The most effective policy is prevention- this attacks the root of the problem with high quality early childhood interventions. In addition, improving the quality of teaching in public schools would also reduce the upper secondary dropout problem, especially among student from poor households.
But that should not rule out interventions that can be implemented at the upper secondary education level to help reduce dropouts in the current cohort. Is there now new evidence about what can help kids who are now 15-years-old and have deficient education trajectories and lack of adult guidance? And, an even bigger challenge: is there a way of keeping marginalized kids in school and improving their skills?
Recent evidence from a randomized control trial in Chicago shows that successful interventions exist. Disadvantaged high school kids in marginalized areas in Chicago went through a one hour a day math two-on-one tutoring program and one hour per week of cognitive behavioral therapy (CBT). The results are impressive. Math test scores improved by half a standard deviation  and graduation rates are expected to improve 46 percent (14 percentage points) after one year.
The experience in Chicago shows that there are synergies from dealing with cognitive and socio-emotional deficiencies simultaneously, but each require a separate, well-designed intervention, lots of leadership and human and financial resources.    
Although dropout rates in Mexico have reduced marginally since 2008, the situation is not very different right now and it is a common challenge from Tijuana to Tierra del Fuego. The starting point to address the dropout problem in upper secondary is to realize that there is no easy fix. The solution demands a great amount of effort and resources, but if we manage to provide opportunities to our disadvantaged youth, it will be worth it.  
Follow Rafael de Hoyos on Twitter.
Find out more about the World Bank Group’s work on education on Twitter and Flipboard.



Join the Conversation

April 26, 2016

Rafael, Thank you for the excellent blog post. A partner of ours, Escalera, (an educational non-profit based out of Chiapas, Mexico) tested a cross-cut combination of scholarships and information in a 4-arm RCT. Early evidence is coming in and starting to show similarly mixed results--working in some places but not others--in short: there's no easy fix!
So, what can be done? I'm somewhat skeptical of your proposed solution--not because I'm against CBT (I'm quite familiar with the evidence, find it very promising and most certainly worth testing), but mostly because it sounds like the same story as before--back then, there was good evidence that economic constrains were binding and so a scholarship program sounds like an obvious, easy answer. In trying to find the easy fix, are we not falling into the same folly as before?
Added to this complexity is that from Tijuana to Tierra del Fuego (let alone from Chicago to Chiapas), the binding constraints to improving school graduation are likely to be different.
Is there a better way of thinking about educational financing to more systematically test and adapt interventions to the complex environments we live in? A smarter way to ensure that educational financing achieves educational outcomes?
There may be one way--paying for educational outcomes directly. Our work with Escalera and the State Government of Chiapas was the first (to our knowledge) of a state government paying providers for educational outcomes. It builds on the experiences of other national governments and donors (CIFF, DFID, WB's REACH program) that have also paid providers based on educational outcomes--putting educational policymaking closer to those to whom it affects.
Paying for outcomes is not easy, and itself requires a lot of tweaking to get right. But what these early examples are showing is that it is becoming an increasingly popular tool within a broader toolkit that policymakers can use to achieve their educational objectives.
All the best,
Partner & RBF Lead,

April 29, 2016

Hi Michael,
Thanks for your comments. The proposed solution is conceptually very different from the previous ones since the diagnostic identifies the problem starting since birth (in many cases) as oppose to being the outcome of a current constraint (income or other). Therefore, the real solution is prevention through ECD and compensatory basic education. But that would only benefit future generations.
What about the current adolescents? For them the most effective intervention (that I know of) is the one implemented in Chicago. It is not only CBT, it is also one hour of additional math instruction per day and personalized mentoring. Implementing that is far from easy and quite costly indeed. In a way is like trying to fix a lifetime of deficiencies in one year.
Regarding paying for outcomes, the evidence regarding its effectiveness is not conclusive. Have a look at the following blog post by colleagues here at the Bank.……
What the evidence suggests is that you have to learn more about the "production function", change some of the inputs and help schools use them in an effective way, rather than simply pay for outcomes.
Happy to discuss further.

May 11, 2016

Hi Rafa,
Thanks for this.
Of potential interest is the recently completed USAID-funded School Dropout Prevention Pilot (SDPP) Program that tested the effectiveness of dropout prevention interventions in four target countries: Cambodia, India, Tajikistan and Timor-Leste.
More here: