Pavel Kohout makes a counterintuitive argument regarding pro-natalist tax policies:

What happens if the tax burden for families with children is cut at the expense of the childless? We can expect only a small increase in the fertility rate among families with children. If you already have two children, would you be likely to have another just because of a tax advantage? Maybe, but generally, tax stimuli are not a significant factor in family planning. The marginal increase in the birthrate as a result of tax advantages would be nearly negligible.

However, there would be a more significant impact on the fertility rate within the group of people who do not yet have children. An increased tax burden would make it more expensive to form a family. People normally marry and procreate only after they have achieved a certain level of financial strength and independence. This rule has been an ages-old norm for human behavior with the exception of welfare state-dependent classes for whom child benefits constitute a major part of their income. Imposing tax penalties for the unmarried might delay procreation by several years, thus cutting the marginal fertility rate. Some of these people would remain unmarried forever. The total impact of "pro-family" tax policies would certainly be negative.

Unfortunately, the Kahout's only gives one piece of empirical evidence: the failure of such policies in Italy under Mussolini before World War II. Therefore, I can't draw any firm conclusions. Speculating though, I would think that if the tax burden on singles causes them to delay marriage by years, then a similar tax burden on the married couples would similarly delay birth of their Nth child. Kahout seems to be confusing an average couple he supposes content with an average of two children with a marginal couple indifferent between whether or not to try to conceive another child now. For Kahout's claims to hold, I would think the an additional hypothesis is necessary: when measured in terms of percentage of income, the tax burden on singles would have to be higher than it is for married couples.