If you buy insurance on the exchanges, your premiums are probably going up. How much will depend on what state you live in. New data released by the government indicate that the lucky citizens of four states will see the price of their two cheapest options fall (by significant amounts, in Indiana and Mississippi). On the other hand, citizens of 19 states, almost 3 million people, will see increases of more than 10 percent, and four states will see increases of more than 25 percent. On average, the premiums of the benchmark plans (the second-lowest silver plan offered in the market) are set to rise about 7.5 percent.

What you think about this probably depends on what you think of the law. If you're a supporter, 7.5 percent may compare favorably to premium increases in the years leading up to the law's passage. If you opposed it, well, 7.5 percent sounds like a pretty large increase. And two states actually hit the 35 percent mark predicted by Donald Trump. Those big numbers will mean sticker shock for the folks buying without subsidies -- and a higher bill for the taxpayers who pay subsidies.

What can we learn from the government's figures? -- There was a lot of mispricing in the early years. Supporters of the law got very excited about low premiums in the first couple years of the program's operation. Finally, it seemed, some government program had gotten health care costs under control! But those premiums were not based on the actual costs of providing insurance to people buying insurance on the exchanges. They were guesses. We now know that the insurers guessed wrong; many more lost money than made it in the first two years of the program's operation. Those guesses are now being adjusted based on actual experience, which means that premiums will probably rise both this year and next.

CartoonDavies' latest cartoon: Nassau's got mailCommentSubmit your letterReader essaysGet published in Newsday

-- Bigger states are doing better. The government's numbers are a weighted average of all the states offering policies through a federal exchange; that average is 7.5 percent. But the unweighted average -- what you get by just averaging all the rate increases together and dividing by the number of states -- is 11.5 percent. A weighted average is the correct way to look at national premiums. Alaska's 21,000 people did see 30 percent rate hikes, but they are a small fraction of the overall insured population. But comparing the weighted average to the unweighted does tell us something: the outlying large increases are concentrated in smaller markets. In the weighted average, they are swamped by Florida's dainty 1.2 percent increase for its 1.6 million customers. In many ways, it's not surprising that smaller markets are where we're seeing the most extreme changes. (In both directions: Mississippi premiums declined 8.2 percent.) Smaller markets have less competition for either insurers or the providers they have to pay -- if Cody, Wyoming, only has one hospital, the insurer can't threaten to take their business to another provider. And the smaller the insurance pool, the harder it is for an insurer to get an accurate actuarial estimate of their expenses.

-- The failure of the co-ops is probably affecting prices, at least a little bit. Nine co-ops have failed so far. Four of them were in states that run their own exchanges, and five were in states with federal exchanges: Nevada, Iowa, Nebraska, Tennessee, South Carolina and Louisiana. (One co-op served Iowa and Nebraska.) All of the states that have had co-op failures had increases above the national average, which I'd guess represents the disappearance of badly underpriced plans from the marketplaces.

-- These numbers aren't a very accurate picture of what people are actually paying. Remember, this is not an estimate of the premiums everyone is paying; they're a look at the cost of a specific plan, the second-cheapest silver plan in a given market. In 2014, about two-thirds of customers bought either the cheapest or second-cheapest silver plan. But that leaves a lot of the market that isn't buying the benchmark plan, and the other plans' cost may have gone up or down at a different rate.

advertisement | advertise on newsday

For that reason, ignore anyone who tries to compare these figures to the average premium increases before Obamacare kicked in. Those figures represent the average increase in what everyone was paying for insurance, not the increase in the price of a specific plan. The numbers aren't comparable.

-- Reports of the death of health-care-cost growth were greatly exaggerated. Health-care costs have been moderating for about a decade, and when premiums stayed low in 2015, people dared to hope that this had something to do with Obamacare. (I've always been skeptical.) But health-care costs ticked up at a brisk clip last year, and as sure as night follows day, premiums will rise to match.

-- States with big price hikes are going to have a hard time attracting unsubsidized customers to the exchanges. I don't have to explain this, do I? If they're not buying it now, they're not going to buy it when it costs 30 percent more. Does this matter? Some, at least. Small exchanges are going to run big operating losses. And the unsubsidized folks who are declining to buy coverage are probably healthier than the unsubsidized folks who are ponying up for the insurance, so keeping them out of the pool means a higher average cost to cover the folks who remain. Some people who are buying now may get sticker shock and drop it.

Sign up for The Point

Go inside New York politics.

-- The government is going to be spending more on subsidies this year. Most people buying on the exchange will be insulated from price increases by the subsidies. The taxpayer is providing the insulation.

Is this a disaster? Not yet. Insurance was underpriced in the first few years, and now it's rising toward the market level. What matters is what happens in the future. Is this a one-year blip, as insurers adjust? Or is a 7.5 percent annual hike something we can expect for years to come? If it's the former, then this is, well, a blip. If it's the "new normal," then it presents some issues. Three or four years of high- single-digit rate increases will probably make unsubsidized folks rethink their decision to purchase insurance, raising the specter of adverse selection.

And even subsidized folks will start feeling the pinch if premiums rise too much, because once the government's spending on subsidies exceeds 0.504 percent of GDP, the subsidy formula is supposed to change to cap total expenditures -- meaning that some folks currently getting subsidies will no longer be able to count on having their premiums capped at a certain percentage of their income. We're nowhere near that amount yet, and may not get there if enrollment in exchange policies has truly stalled. On the other hand, it wouldn't take all that long to get there with premium costs growing at two or three times the rate of GDP.

If I had to guess, I'd say that this year's premium increase represents a lot of one-time factors that will taper off after 2016. But I'm not very confident in that guess -- and if health-care costs keep growing as fast as they did last year, that guess will be wrong.

The upshot is that it's still early innings for the health-care program. We don't know what the mature program will look like: how many people it will enroll, at what cost. We'll have to see what happens.

advertisement | advertise on newsday

Megan McArdle is a Bloomberg View columnist who writes on economics, business and public policy.