|When Good Models Happen to Bad People|
No single model received more attention in 2014 than one produced by the Center for Retirement Research at Boston College titled "Long-Term Care: How Big a Risk?" (November 2014, Number 14 - 18, Leora Friedberg, et. al.). At the same time, no other model has been more widely misinterpreted, wrongly extrapolated or gleefully co-opted by LTCI detractors. Since the New Year is considered a time for looking forward, let's begin by clearing this foggy hangover from 2014, then speak no more of it.
MEDICAID 1, LTCI 0
This is the way the CRR model's conclusions are most often presented. By using monthly instead of yearly data, it was found that average nursing home stays are 30% shorter than previously believed (on average a man stays less than 12 months, a woman 17 months). In fact, 45% of patient stays do not exceed three months. Even worse, LTCI is duplicative since Medicare will cover these short stays—the study assumes the first three months of "all episodes of care are covered by Medicare."
Finally, CRR corrects a previous model that understated the probability of ever needing care by 32% to 63%. The conclusion? Since long-term care is a relatively high-probability event, but less catastrophic than previously understood, it makes less economic sense to insure against.
The media jumped all over it:
• "Maybe You Don't Need Long-Term Care Insurance After All" (Bloomberg)
• "Here's a New Reason to Think Twice Before Buying Long-Term Care Insurance" (Time/Money)
• "'Spending Down' for Medicaid is the Most Practical LTC Financing Plan for Most Americans, Researchers Assert" (McKnight's)
• "Is Long-Term Care Insurance for You?" (Wall Street Journal)
• "Boston College Finds Rip-Off in Long Term Care Insurance Costs When Compared to Other Options, Opines UltraTrust.com" (Estate Street Partners)
Readers who dove into these articles seeking sound advice were met with takeaways such as this: "Forgoing long-term care insurance and relying on Medicaid is the smartest financial planning decision for the majority of unmarried Americans." Lacking were any qualifications concerning Medicaid's notoriously low reimbursement rates, institutional bias, record of poor quality or inability to access care.
This was our first sign of trouble. CRR assumes all "rational, far-sighted, well-informed" individuals make decisions entirely on the basis of money. We do not. As economists, they'd have been better served with a model in which rational individuals make decisions which maximize our utility. Had they done so, their buyers would've valued higher-quality care and the ability to remain at home with family, tilting the scales in favor of LTCI.
Meanwhile, the Bloomberg piece acknowledges that the biggest threat to a retiree's nest egg "isn't a stock market crash. It's a long illness requiring round-the-clock care." Unfortunately, thanks to the new CRR model, not only should most people "just skip [LTC insurance]," but the majority of Americans (all but the richest 20% to 30% of singles) should "[spend] down their assets and then [let] Medicaid pick up the tab."
Lest we dismiss this study for its preoccupation with singles, we are warned that "forthcoming research will show long-term care insurance makes even less sense for married couples."
And why did the researchers focus on singles anyway, when 82% of all LTCI policies are purchased by members of couples? They argue that since 75%+ of nursing home residents are over age 65 and single, their limitation to singles is "not significant." Once again, our economists have set out on the wrong foot: they are not modeling nursing home residents -- they are modeling buyers. Oh, dear.
The Average Family Has 2.5 Children
I've been careful in my choice of words: What the Center for Retirement Research produced was an economic model. Framing it otherwise (a study or research report) suggests a methodology or outcome that we shouldn't reinforce. Models exist in the abstract, not reality. This one invented hypothetical buyers in a controlled environment.
One particularly unfortunate problem with CRR—overlooked in the entire hubbub—is that it sought to answer a question of its own making, not one that anybody had been asking. Namely, why do only a certain percentage of single individuals (an assumption of their own creation that disagrees with other contemporary sources ) buy LTCI, differing from the percentage predicted by the Brown & Finkelstein Model (i.e., the famous "Medicaid Crowd-Out Effect")? This model was an attempt to reconcile the two numbers.
Now, models can serve a purpose, but they are inherently limited. In the case of CRR, even its "new" data remain archaic (10 years old) and don't square with reality. After all, insurance is built primarily around the remote but catastrophic risk—not the occasional shopping cart dinging your car door. This is why buyers and sellers have played a tug of war between unlimited benefit periods and short-term care. One is hard to offer profitably, while the other is hard to make desirable.
Worst of all, the model presumes that buyers care only about nursing facilities, when the exact opposite is true. Most of our clients are motivated to purchase LTCI for its ability to do the one thing Medicaid is worst-equipped to do: keep them out of the nursing home.
Then, in a final Hail Mary, they assume Medicare pays for most short stays—which one nursing home worker laughs off: "[I] can count on two hands out of the thousands of patients I've served how many have actually received 100 days of Medicare coverage."
Ultimately, the economists got the results they hoped for (had they not, would this study have seen the light of day?) and were able to achieve agreement between the Brown & Finklestein model and their own:
Singles aged 65+ who "make optimal saving and insurance decisions" (how many real people do you know like that?) are substantially less willing to buy an option to purchase LTC insurance at market premiums, based on a more-accurate transition matrix updated to 2004 based on monthly probabilities instead of annual transition events.
Go back and read that sentence again.
Not much of a headline-grabber, huh? We should be asking ourselves what all the hoopla was about—particularly since a landmark study was released almost simultaneously as CRR, which contained some of the most newsworthy, compelling and positive research about LTCI in over a decade. Do you remember the financial media covering this report with the same enthusiasm as the Boston College model? Do you recall seeing any of the above publications covering it at all?
Since the article above first appeared in the author’s private newsletter, even more national newspapers and magazines have picked up the Boston College findings and catapulted the damage. Well-meaning individuals of great intellect have every right to disagree over how to best fund long-term care in the United States, but we mustn’t admit only those facts that agree with our preconceived narrative. After all, our clients deserve nothing less.