By Jian Chen
If the large turnout at the Princeton Public Schools community forum was any indication, the community seems anxious to learn about how future enrollment will look like from the professionals at Milone & MacBroom that the taxpayers paid $144,000 for. As someone who has closely followed this issue and crunched the numbers over the past two years, I’d like to take this opportunity to share with you some of my thoughts. But I’d like to preface this letter by stating that “writing to newspaper” is not my preferred way to communicate with our school district leadership on this subject. That was why I wrote to the superintendent and the board multiple times in the summer of 2018 to alert them about the flaws in the original demographic study and was also why I joined the district’s future enrollment committee last year along with several other community volunteers to work with demographer George Sundell directly to better understand the major drivers of the recent enrollment growth and to build a more robust enrollment forecast.
While the future enrollment committee made progress in areas such as forecast methodology, the committee was unable to complete its work and ceased its activities after the board hired M&M last fall to conduct a long-term planning review. At our last meeting, I spoke on behalf of the committee and clearly outlined to M&M’s lead consultant Mike Zuba the major issues that need to be addressed in their enrollment study. To my utter dismay, when M&M presented their “findings,” I noticed that none of these major issues, issues that the committee had largely reached consensus on, were addressed in M&M’s enrollment forecast. Instead, M&M’s enrollment model reverted back to the flawed methodology and doubled down on the dubious assumptions used in the initial enrollment study. As a result, the projected future enrollment growth has most likely been inflated. Continuing with the current set of enrollment projections will undoubtedly lead the community to wrong and costly decisions. I’m laying out the key issues here as to why M&M’s enrollment forecast is flawed. Since the board president declined my request to release M&M’s models in advance of the community forum, my analysis relies primarily on M&M’s presentation slides and publicly available data. As you will see, my numbers and argument are quite straightforward. You won’t need more than middle-school math to follow along.
To understand enrollment forecasts, it is important to introduce the cohort survival method, a simple statistical model widely used by demographers. The model uses historical student grade matriculation trends to predict future enrollment. The most important inputs of the model are survival rates, which are calculated by dividing the number of students of a particular grade level by the number of students of a grade lower in the prior year. In the case of “birth-to-K” survival rate, it is the number of kindergarteners divided by the number of births six years prior. Generally, higher survival rates lead to stronger forecasted enrollment growth and inputs of lower grade levels tend to have a disproportional impact on the results.
Inputs of survival rates are typically calculated using average survival rates of the past six years, a standard practice M&M acknowledges in its presentation (slide 24). However, in the three projection models presented – low, medium and high, M&M used five-year, weighted three-year and three-year average survival rates instead. You don’t have to be a demographer to question the wisdom of using a three-year trend to predict the future ten years out.
Overriding the standard six-year averages with higher and shorter-duration ones was one of the problems extensively discussed by the future enrollment committee. Committee members were concerned that the original demographer failed to sufficiently justify the use of three-year averages for the birth-to-K survival rates and managed to convince him to change them back to six-year averages. In my conversation with M&M consultant Rebecca Augur after their presentation, she couldn’t give a clear answer to why they chose to deviate from the standard practice. The effect of these overrides is difficult to assess without seeing the actual models and is further complicated by the fact that there are some sizable discrepancies between historical enrollment and birth data used by M&M and those used by the original demographer.
Flipping through M&M’s 94-page slide deck, you will be excused for not being able to pinpoint what exactly are the underlying drivers that will sustain the recent enrollment growth well into the next decade. The number of births has been on a decade-long declining trend. Housing transactions are cyclical and will always ebb and flow. Importantly, the underlying data seems at odds with the assertion that Princeton has been on a path of strong and sustained enrollment growth. According to PPS’ Comprehensive Annual Financial Report (CAFR), enrollment grew a total of 461 students in a ten-year period between 2009 and 2018. Yet, enrollment growth in 2015 (141) and 2017 (198) alone accounted for 74% of the overall growth. Does this astonishingly concentrated growth suggest sustained growth or anomaly instead? On slide 13 of the M&M presentation, we can see that the enrollment spikes in 2015 and 2017 coincided with the large number of tenants taking occupancy in the newly opened multi-family housing developments. Based on M&M’s numbers (slide 28), it is estimated that students generated by Avalon Princeton, Copperwood, Merwick and Stanworth since 2015 amounted close to half of the total enrollment increase of that ten-year period. Why does this matter?
First, lumping in enrollment originated from these multi-family housing units overstates survival rates and obfuscates the fact that “organic” enrollment growth driven by local demographics and housing transactions has been quite modest. M&M acknowledges this fact in its presentation (slide 28) by stating “There [multi-familiy] developments have contributed to elevated cohort survival rates over the last four years…” Second, since the students from the anticipated multi-family units to be constructed as a result of the affordable housing settlement are added on top of the projections of the cohort survival model, not excluding students from multi-family housing units while calculating historical survival rates inflates future enrollment growth by double-counting the effect of multi-family housing developments. Lastly, cohort survival method is not suited to forecast enrollment growth of multi-family housing units where the total number of housing units is fixed. With the occupancy rate at some of these developments already exceeding 90%, enrollment in multi-family developments cannot grow indefinitely. Not surprisingly, M&M’s enrollment forecast looks outlandish. For example, M&M’s medium-level model predicts that enrollment will grow by about 400 students in the next five years. Since the anticipated multi-family housing developments resulting from the recent affordable housing settlement agreement aren’t expected to produce meaningful number of new students until 2023 at the earliest, it seems highly unlikely enrollment for the next five years will grow at almost double the speed of the past ten years during which close to 900 multi-family units were created and occupied.
The school district leadership is fully aware of my long-held argument that a robust enrollment forecast requires modeling the “organic” growth and multi-family developments separately. It was clearly communicated with M&M as well. The previous demographer had agreed to run an alternative model using this method, although the Enrollment Committee didn’t get a chance to fully evaluate his preliminary results. So why did M&M knowingly choose to use the flawed methodology?
I can go on with more. But I guess you have got the gist of my argument. I didn’t write this letter to name and shame. I only did this because my many attempts to persuade the school district leadership in a more discrete manner have yielded nothing but frustration and disappointment. Public opinion seems to be the only thing that stands between our school district leadership and a problematic path they try to lead us down.
I again urge the school board to release M&M’s forecast models and key supporting documents so they can be thoroughly reviewed by the public. It is easy to pay lip service to transparency and making evidence-based decisions. It is only when the evidence contradicts our preconception that our honesty and integrity are truly put to the test. In a time when the school district leadership’s credibility is at a dismal low, taxpayers need to be convinced that consultants paid with taxpayer dollars are indeed working for us.
Mr. Chen works in finance and is a Princeton resident.