Economists typically think about two very different explanations for why people remain in the same insurance plan year after year. One possibility is that switching plans involves a set of hurdles or frictions: people dislike navigating new networks, they worry about losing trusted doctors, they find it stressful to compare complex pricing structures, or they simply lack the time and patience to reevaluate their options. These are often called “switching costs” or “state dependence,” terms that describe how the mere fact of having chosen a plan in the past makes someone more likely to stick with it.

Another possibility is that people are staying put because they truly value the plan they have. They may trust the insurer, they may have had good experiences with the customer service, they might be attached to a particular hospital or physician group included in their plan’s network, or they may simply find the plan a good fit for their needs. Economists call this “unobserved preference heterogeneity.” It means that even if we can’t see the reason a plan is attractive to someone, the loyalty may be sincere.

The difficulty is that these two explanations look almost identical in the data. If someone stays in a plan when the price goes up, is it because they couldn’t be bothered to switch, or because they chose to pay more in order to keep something they genuinely like? Untangling the two is one of the hardest and most important challenges in studying consumer choice in health insurance.

Traditionally, economists have tried to separate switching costs from preferences by imposing strong assumptions on consumer behavior. Many earlier models assume that consumers make choices according to a specific mathematical formula, or that the first decision a consumer made in the marketplace was somehow free of switching costs and based only on their genuine preferences. These assumptions are often unrealistic, and when they are wrong, they can seriously distort the conclusions researchers draw.

The new paper does something different. Instead of assuming that consumers behave in a particular way, the authors make use of simple, almost intuitive logic: if someone switches plans when prices change, that switch reveals something about their relative preferences. If, in a later year, the price gap between those same plans shifts in the opposite direction, the way the same consumer reacts again reveals something further. These patterns of switching and not switching provide clues about how much people value their old plan versus how much they are influenced by the cost of moving to a new one.

The authors formalize this idea using “moment inequalities,” a mathematical tool that lets them identify the range of switching costs that is consistent with observed behavior, without needing to assume anything rigid about the reasons people have for liking one plan over another. In other words, instead of trying to pin down a single precise number, their method allows them to derive upper and lower bounds on what switching costs could plausibly be.

The ideal testing ground for this method is the Massachusetts CommCare program, in which low-income adults had to choose from a set of standardized plans offered by private insurers. Standardized benefits meant that the plans differed mostly in their provider networks rather than in their cost-sharing designs, which makes it easier to isolate the effects of price and preferences.

The data follow individuals from one year to the next. The researchers can see what plan they had, what the prices were in each year, and whether they switched. These patterns, taken together, provide a treasure trove of information about how much switching costs might matter.

Earlier studies using traditional models estimated that switching costs in this market were enormous, often around $1,000 to $1,500 per year, a number large enough to significantly dampen competition. If switching is that costly, insurers might feel emboldened to raise prices, confident that most people will remain where they are.

But the new approach reveals a very different picture. Using the moment inequalities method, the researchers find that the upper bound on switching costs is closer to $100 to $200 per year. Even when they mimic the stronger assumptions used in prior research but allow for unobserved heterogeneity, they find switching costs around $600, still far below what previous studies suggested. It turns out that once you properly account for the fact that people genuinely like some plans more than others, the amount of friction in switching is much smaller than researchers had believed.

The implication is striking: much of the loyalty observed in health insurance markets is not due to people being stuck or inattentive but rather reflects stable, legitimate preferences. People tend to stick with plans that include their doctors, or whose networks match their medical needs, or that they trust for reasons that might not be fully visible to researchers.

The key reason earlier studies overstated switching costs is that they did not adequately allow for the existence of individual-specific preferences for each plan. Without accounting for these person-by-plan attachments, any persistent choice looked like inertia. People who stayed with a plan despite price changes seemed, to earlier models, to be held in place by some invisible force of inattention or hassle rather than by affection or loyalty.

The new model used in this study changes that. It recognizes that each person might have a unique preference for each plan that remains stable over time. Once you take this into account, much of what looked like switching reluctance in the old models turns out to be genuine preference.

The policy implications are substantial. If consumers are failing to switch because of large frictions, policymakers might want to simplify choices, increase reminders, or mandate clearer comparison tools to push consumers toward more active shopping. But if consumers stay because they genuinely prefer what they already have, then such interventions are less likely to work and might even be counterproductive. Instead, regulators might focus on maintaining diverse options, encouraging competition through innovation in provider networks, or ensuring that plans compete on characteristics consumers actually value.

The findings suggest that while there is still some cost to switching, it is not the dominant force. What matters more are the stable, meaningful differences across plans, differences that consumers themselves appear to understand quite well.