The little-known microeconomics textbook Zowning and Bupan does an excellent job of presenting the case against medicare vouchers:
Consider the figure below. The pre-voucher budget line is AZ, and the consumer's choice is shown by point W.
The medicare voucher will affect the recipient in one of two ways. Figure a shows one possibility. In this case, the consumer chooses market basket W' when the budget line is AA'Z' under the medicare voucher program. If the consumer receives instead a cash grant of equal value, the budget line is A"Z', and the same market basket W' would be chosen. Consumption of medical care increases, but only because medical care is a normal good, and more of a normal good is consumed at a higher income. Note, however, that medical care consumption rises by less than the amount of the subsidy. For the preferences shown in Figure a, medical care consumption increases by 5 units. The consumer uses the remainder of the subsidy to increase purchases of other goods. (Because other goods, taken as a group, are certainly normal goods, too, the purchase of these goods will also increase).
Figure b, where the indifference curves differ from those of Figure a, shows another possible outcome of the medicare voucher. In this case, with a direct cash subsidy, the consumer prefers to consume at point W', where U3 is tangent to the budget line A"A'Z'. The medicare voucher prohibits such an outcome, however; the consumer must choose amount the options shown by the AA'Z' budget line. Faced with these alternatives, the best the consumer can do is choose point A', because the highest indifference curve attainable is U2, which passes through the market basket at point A'.
When the situation shown in Figure b occurs, the voucher increases the consumer's purchases of both medical and non-medical items. Indeed, regardless of whether Figure a or Figure b is the relevant case, the medicare voucher cannot in reality avoid being used in part to finance increased consumption of non-medical items. This result is particularly interesting because many proponents of subsidies of this sort emphasized that the subsidy should not be used to finance consumption of "unnecessary" goods (such as vodka or junk food). In practice, it is difficult to design a subsidy that will increase consumption of only the subsidized good and not increase consumption of other goods at the same time.
Note also that the consumer in Figure b will be better off if given $50 to spend as he or she wishes instead of $50 in medicare vouchers. This observation illustrates the general proposition that recipients of a subsidy will be better off if the subsidy is given as cash. The situation in Figure a illustrates why there is a qualification ot this proposition: in some cases the consumer is equally well off under either subsidy. THere is no case, however, when the consumer is better off with a subsidy to particular good rather than an equivalent cash subsidy.
There is no Zowning and Bupan textbook. I just changed "food" to "medical care" and "food stamps" to "medicare vouchers" in the analysis of food stamps in Browning and Zupan's intermediate microeconomics textbook. Actually, Browning and Zupan provide a spirited defence of vouchers in the 11th edition of their text, arguing that they would be more efficient than President Obama's health care reforms.
It's a pattern repeated in some other economics textbooks (e.g. Harvey Rosen's widely used public finance text). Where a good is presently publicly financed or provided, as is the case for medicare or elementary education, the text argues for the superiority of vouchers over public provision or finance. Where a good is provided through vouchers, as is the case for food stamps, the text argues for the superiority of cash transfers over subsidies. Where cash transfers exist, the text shows that the high effective marginal tax rates implicit in tax/transfer programs reduce work incentives.
Not every economics text argues for minimal government, and not every economist believes in it. But those who do are unlikely to stop arguing for smaller government just because a few government provided services have been replaced with vouchers.