For a very interesting and insightful analysis of demographic trends in the developed world, see
Rite of Spring: Russia's Fertility Trends
By Anatoly Karlin
available at http://www.russiablog.org/2009/04/russias_fertility_future.php
One of the more interesting observations by Mr. Karlin was the following:
A Womanly Demographic History of Russia
An even more meaningful measure is the net female reproduction coefficient (NFRC). It takes into account two things that the TFR doesn’t, at least not explicitly - a) the male-female ratio at birth and b) the female death rate, pre- and during childbearing age. Although the replacement level TFR is usually quoted as being 2.1, as mentioned above it varies in practice. Although that is indeed the case in most modern industrial countries, in underdeveloped and/or traditional societies with high female mortality rates in early years and/or high male to female ratios, the TFR needs to be as high as 2.5, 3.0 or more for generation reproduction. This is because a lot of females die before they can give birth to more girls. Although China has a nominally respectable TFR of 1.7-1.8, it is effectively considerably lower due to societal preference for males and the resulting skewed demographic profile.
The net female reproduction coefficient explicitly takes the two factors above into account - any value greater than 1 ensures long-term population growth, while a value of less than 1 implies impending decline. In the graph below you can see a graph of Russia’s NFRC from 1960 to 2005.
Today all the world’s major industrial nations are not producing enough girls to maintain their current population levels in the long-term. The US as a whole just about makes an exception, although only thanks to the help of highly fertile Hispanics. In Russia, the NFRC increased since 2005 to 0.67, which puts it above most east-central European countries but still significantly below France, Scandinavia and the Anglosphere.
So, even were China to raise its TFR to 2.1 (which it apparently has no intention of doing), if it retained its current ratio of boys to girls at 119 to 100, it's NFRC would still be less than 1, roughly 0.96, and it would still be on track for falling natural growth rates (i.e., it would shrink over time absent immigration). Thus, in countries with unnaturally skewed birth rates favoring boys over girls (such as China and India), a TFR of 2.1 is actually below replacement level. For example, in China, to achieve a replacement level NFRC of 1 (let's use 1.05 or exactly 1/2 of the normally cited 2.1), the TFR would have to be about 2.3 if the distribution of boy babies compared to girl babies in that nation remains unchanged and assuming that 2.1 would be the replacement TFR but for the maldistribution of boys and girls in China.