Rather than have the government pay for your health care, the government will give you a voucher with which to buy private insurance. Initially, the voucher will be worth the same amount as the average cost of providing health care to people. But the insurance company will have higher administrative costs than Medicare, and it will have profit margins and such that Medicare doesn’t have, and it will pay more for services than Medicare does. So on day one you’ll lose your Medicare coverage and instead get a voucher that costs the government the same amount, but buys you much less in the way of health care services.
The way the government saves money over the long run, however, is that over time the voucher won’t keep up with the cost of health care. As the CBO explains in its analysis (PDF) of Ryan’s outline, the voucher will be “indexed to grow at a rate halfway between the general inflation rate, as measured by the consumer price index for all urban consumers (CPI-U), and the rate of price inflation for medical care, as measured by the consumer price index for medical care (CPI-M).” That means the value of the voucher “would increase at an average annual rate of 2.7 percent for the next 75 years, in comparison with the average annual growth rate of nearly 5 percent that CBO expects for per capita national spending for health care under current law.”
In other words, Ryan is proposing to ration care for seniors. He’ll take the baseline level of per capita medical costs for seniors in 2020 and then draw a curve representing 2.7 percent annual growth and say that any costs above that won’t be covered. If grandma’s got a bunch of money, then she can spend her money. If not, then the plug is pulled.
As Yglesias also notes, the plan wouldn't kick in until 2021, "which helps Ryan avoid needing to think about implementation details."