PTWhat makes a health plan operator take on debt?
Health plan operators collect monthly fees and pay for consultations, tests and hospital stays. To finance everything else, each one chooses between using its own money and borrowing. The thesis followed about 200 operators in Brazil from 2018 to 2023 to see what weighs on that choice. Explained in 12 screens.
For every R$ 100 of assets, the average operator in the sample had R$ 25 of technical reserves, R$ 16 of long-term liabilities and R$ 4 of debt. Averages from Table 38; these are different measures and do not add up.
Of every R$ 100 in monthly fees, about R$ 75 goes back out in consultations, tests and hospital stays
This is the claims ratio (in Brazil, sinistralidade): the share of what the operator receives from its plans that goes to the health expenses of its beneficiaries. What is left has to pay for everything else.
In the thesis sample, the average claims ratio was 75%. But some operators spent more than they took in. Pick a case.
Claims ratio = healthcare expenses ÷ revenue from health plans. Sample values for 2018 to 2023 (Table 38: median 77%, a quarter above 83%, maximum 121%). The 2023 figure for medical-hospital operators (87%) is from ANS, Brazil's regulator, as cited in the thesis.
More than 80 million plan memberships, and an agency watching the books
Brazil's private health plan market has more than 1,200 companies of eight types. All answer to the National Supplementary Health Agency (ANS), which monitors their figures and requires technical reserves: money set aside to honor the care already promised.
An operator has to satisfy ANS, which wants solvency, and investors, who want returns. Tap a type.
Beneficiaries in December of each year (2024: September), Table 24 (SIB/ANS). One person can hold both kinds of plan, so we speak of memberships; the sum of the two (83 million in 2023) is our calculation. The types of operator are described in the thesis Introduction.
Two theories give opposite advice about profit
Borrow or use your own money? Finance has two classic explanations, and the thesis tests both on health plan operators.
Illustrative diagram, no data. Each theory's predictions come from sections 1.1 and 1.7 of the thesis (Myers, 1984; Myers and Majluf, 1984). For the claims ratio and technical reserves, the thesis allows for either sign (Table 37).
Of 1,292 operators, about 200 with comparable accounts remained
The author used the financial statements operators file with ANS, year by year, from 2018 to 2023, and compared each operator with itself over time (panel data analysis).
COVID-19 pandemic years in the period
Tables 35 and 40. The sample table lists two filters (722 and 54 operators); the remaining cuts down to 200 are not detailed (1,292 − 722 − 54 = 516, our calculation). Extreme values were trimmed (winsorized at 1% and 99%) without dropping records.
Three yardsticks for debt, and on two of them there is almost none
"How much an operator owes" depends on what you count. The thesis used three measures of leverage, all as a share of assets. Pick a yardstick and see what it picks up on the balance sheet.
Illustrative balance sheet: block sizes are not data. The formulas are from Table 36 and the panel figures from Table 38 (1,138 to 1,145 records).
Operators that differ widely
Besides debt, the thesis measured four features of each operator in each year. They vary a lot: some post losses and some earn half their assets; some have almost no buildings or equipment and others hold two thirds of their assets in them.
Table 38 (descriptive statistics, 2018–2023, values trimmed at 1% and 99%). All measures as a share of assets, except the claims ratio, which is a share of plan revenue.
Two factors point the same way on every yardstick
The author estimated six models: each debt yardstick, with and without the type of plan the operator sells. Profit and fixed assets had the same sign in all six; the others change from one yardstick to the next. Tap a block.
See the coefficients as a table
Table 40. Fixed effects (each operator compared with itself) for the net and long-term yardsticks; random effects for total debt, as the Hausman test indicated. R² from 11% to 27%. Association, not causation.
Those who earn more owe less; those with more buildings and equipment owe more
Both theories show up together. Profit follows the queue (pecking order): operators that earn more use their own money and owe less. Fixed assets follow the balance (trade-off): they serve as collateral and come with more debt.
Pick a factor and raise it, starting from the average operator.
Our calculation. It starts from the averages in Table 38 (total debt 4% and long-term 16% of assets) and applies the coefficients of the basic models in Table 40 (3 and 5), all else equal. The net debt yardstick has a different scale and is left out, but it points the same way. It shows association, not causation.
Claims ratio and technical reserves: it depends on the yardstick
The two factors specific to operators did not give a single sign. On the total debt yardstick, neither had an effect. On the other two, the claims ratio came with more debt, and technical reserves switched sign.
Tap a box to read the result and the author's explanation.
Table 40. Arrows: ↑ more debt, ↓ less debt, — no statistical effect. Stars: significance at 10% (*), 5% (**) and 1% (***). The explanations are the author's (sections 5.3 and 6.2).
Not everything that seemed obvious held up
Place your bet and flip the card. Answers from Table 40, in the three models with plan type.
Debt, profit and regulation go together in private health care
Operator managers
The author recommends close management of the claims ratio and technical reserves, balancing capital structure with financial sustainability.
ANS and regulators
For the author, reserve rules ensure continuity of care but weigh more on smaller operators; she suggests rules that take differences between operators into account.
Investors
Our reading: operating profit and fixed assets are the steadiest signals of an operator's debt: they had the same sign in all six models.
Researchers
The author suggests interviews with managers and regulators, longer periods, comparisons by size and the effect of technologies such as telemedicine.
What this research does not say
- Only balance sheet figures: secondary data from ANS, without the managers' reasons behind their decisions. · according to the author
- Six years (2018–2023), with the pandemic in the middle: too short to see long-term effects. · according to the author
- Left out were operators with assets below R$ 1 million and those with negative equity, which may borrow differently. · according to the author
- Association, not causation, and much of the variation in debt (73% to 89%) is left unexplained by the models. · our note
Credits
PEREIRA, Carmem Campos. Determinantes de estrutura de capital no setor de saúde do Brasil. Tese (Doutorado em Administração de Empresas) — Universidade Presbiteriana Mackenzie, São Paulo, 2025. Orientador: Prof. Dr. Wilson Toshiro Nakamura.
Data from Brazil's National Supplementary Health Agency (ANS). This version simplifies the work for a general audience; the formulas and full tables are in the original text, in Portuguese.