Basic Idea (See GoogleDoc for more)

Demonstrate competence/excellence in each of 7 areas:

Logic Model # Flow Chart # Decision Trees # Difference Equations/Causal Loops/Stock & Flow # Linear Programing # Discounting # Cost Benefit

with an eye to these skills:

Default Case

Childcare subsidy for student parents eligible for program like CalWorks: A state agency is considering a childcare subsidy that would facilitate single parents’ attainment of college degrees. The benefit would cost $10k per recipient per year for four years. The expectation is that individuals with a college degree will earn more than individuals without a college degree. This means that they generate more revenue in the form of income tax. They are also less likely to require government assistance of various kinds — call this amount A. Assume current rules limit us to a ten year time horizon. Assume the average salary difference between non-college grads and college grads is D (but get the real info here) and that the marginal tax rate can be found here. Assume a 5% discount rate. At first, ignore inflation.

You may also use an example from your MPR or other work to demonstrate competence/excellence.

Logic Model

Why do we think this will work? What other "wrap around" services might we want to build in to the program and why?

Flow Chart

Help a student navigate the program

Decision Trees

Help an individual choose whether to enroll in college, seek the next degree, etc.
Model an imperfect test for something like likelihood of degree completion
What is the value of information? For a student wondering if s/he should pursue a given program. For program on whether to enroll a given student.

Diff Eqns/Causal Loops/Stock & Flow & Linear Programing


We are tasked with analyzing some proposals for providing childcare subsidies for college students with the goal of increasing timely degree completion.

We propose a subsidy program that would give eligible students $10,000 per year to pay for child care to make class attendance and school work easier to manage in the belief that this will increase academic achievement and make finishing the degree more likely.

We have ascertained that there are different completion rates based on current GPA with A students more likely to complete than B etc. But we also know that there are more single parents in the lower grade ranges (related, we suspect to the challenges of going to school while trying to raise children).
We also know that the current grad rate by GPA varies from a high of 75% for A students to a low of 20% for D students. One issue we anticipate is that we cannot know whether we are giving a grant to a student who would already graduate on time without the assistance. We will treat the existing graduation rate for each GPA level as a measure of this phenomenon (that is, we expect that 75% of the A students who get the subsidy would have graduated anyway - we recognize that this likely UNDERCOUNTS the number who would benefit since we think parenting is one of the things that influences the 25% who do not graduate).

We estimate that even with the support of the subsidy, the likelihood that it will result in a given student graduating on time is related to the student's current academic performance. We think the success rate from the subsidy alone will be about 90% for A students and about 40% for D students.

Finally, we think we need to keep a positive grade incentive in the picture so we can't only offer the program to students with lower grades. In fact, we think we have to say that the number of slots reserved for A students has to be greater than or equal to the number of Bs which has to be greater than or equal to the number of Cs etc.

What numbers of slots should we aim for if our goal is to maximize the number of degree completions. Each subsidy costs 10,000 and we have 750,000 to fund the project.


What would be an optimal mix of slots for A, B, C, and D students?
If this were managed by lottery, how would you sell/explain the idea that an A student might not get a grant while a B or C student did?


Create a causal loop diagram related to this project. Be as inventive as you wish.

  • Student stress * Parenting stress * Number of kids * Academic preparation * Family economic background * Academic performance (e.g., GPA) * Initial enrollment in junior college * Persistence * Completion of AA * Transfer to 4 year * Persistence * Degree completion

Other things we should consider?


Some of our subsidies won't be used - students drop out at mid-semester or from year to year - and some might have to be extended slightly to help students finish in, say, just one more semester. And some students will "age out" as their kids exceed the eligibility age before they complete their degree. Further, some students enter the program at year one, others year two, etc. Build a stock and flow model that will give us a sense of how many students we can support at a time for a given budget.

Let's assume that our first time recipients are 40,30,20,10% 1st, 2nd, 3rd, and 4th year students. And that the drop out rate from year to year is 25%. And that each year we grant 20% of the seniors an extra semester of funding (half a year's amount).

How many new grants can we make each year if we want to keep the number of grants active at any one time under 200?

Discounting and Cost Benefit

Child Care Subsidy: Five CBAs

Formulate five versions of CBA of child care subsidy for "under-resourced" college student parents:

  1. Go/No go:
  2. Project A vs. Project B:
  3. How much of Project A:
  4. Mix of fixed-size programs:
  5. Mix of scalable programs:
Gather Research Data
  1. Actual cost of child care around here
  2. Overhead to set up a child care center
  3. California discount rate
  4. Time horizon for payback
  5. What might a child care facility based on student schedules have to look like time-wise?
  6. What does it cost to staff a cc facility?
  7. What are the numbers of college eligible single parents or students who are single parents? Or just parents?
  8. What are some numbers about grad likelihood depending on grades, age, income, etc.
  9. What are some data about the distribution of grades, age, income, etc.
  10. What's the data on earnings for HS grad, some college, BA look like?
Multiple Project Possibilities
  1. Cash subsidy for parents receiving assistance under other program.
  2. Reimbursement program for parents receiving assistance under other program.
  3. Subsidized, sliding scale childcare on campus
  4. Childcare coop on campus
  5. Etc.
How to Model the Benefits

The direct benefit is increased likelihood of degree completion. We can express this as degrees completed that would not otherwise have been completed.

Simplest version: what is current graduation rate among this group? what do we think the graduation rate will be with the program?
More complicated version: program might work better/worse for different types of students. By grades. Age of kids. Number of kids.

Bringing Pareto and Kaldor-Hicks into the picture

How would we do that?


  1. Is college worth it? Too many degrees are a waste of money. The return on higher education would be much better if college were cheaper. The Economist Apr 5th 2014 (17,500 on average).
  2. Pew Research Center 2014 The Rising Cost of Not Going to College
  3. Hillary Clinton's Focus on Students Who Are Single Parents Inside Higher Education, May 27, 2016 by nEllen Wexler