Changes in relative numbers of high-intelligent and low-intelligent people in the world should never lead to a change in global IQ.
IQ is not a measure of amount of intelligence, it is a measure of amount of intelligence relative to peers.
For example, see this source:
And it is very important for people not to confuse between
intelligence and relative intelligence. Intelligence can never be
measured and IQ is not the measure of intelligence. IQ is simply the
measure of relative intelligence derived by a single or set of
standardized tests.
For example, a young child can have a higher IQ than an adult, even though the adult has plenty more intellectual capacity, because, the peers of a child are not the same as the peers of an adult:
Modern IQ tests use a "deviation IQ" rather than a ratio IQ. With
this method, test takers are referenced to other people of their own
age. The average IQ is still 100, but deviations from the average are
assigned a number which corresponds to a percentile rank.
Esentially, IQ is a measure of ranking, not of amount of ability. The rankings are expressed in IQ points, which are usually equal to 1/15th of a standard deviation of all possible test-takers.
There are standards for assigning IQs to people who get a certain number of questions correctly. These standards change over time, so that the average is always 100:
Because populations experience IQ gains over time, IQ tests must be
constantly restandardized so that subjects are not scored against
inaccurate norms.
That being said, the standards are not updated all the time, and certainly not in all countries. What Lynn and Vanhanen do, in the study that inspired your question, is to give a test that has been standardized in the Western world, to nationals of many countries across the globe.
Norms: Norm groups included in the manual are: British children
between the ages of 6 and 16; Irish children between the ages of 6 and
12; military and civilian subjects between the ages of 20 and 65. A
supplement includes norms from Canada, the United States, and Germany.
This is the standard that all other people are compared to, when assessing their ranking on the IQ scale.
If the tests were always standardized correctly, that is to (a representative sample of) all potential test-takers, the average global IQ would be 100 every time it is measured. So, no, global IQ should never drop.
But perhaps it does drop? It is, after all, a highly imperfect set of numbers. If global IQ were to drop, we could speculate on the reasons. One of the reasons could be that a putative gap in intelligence between more and less developed countries is widening. This means that the Western standards for assigning IQ points get tougher and tougher for poor, more populous countries. Note that this still doesn't need to mean that global levels of intelligence are dropping! All countries could do better nowadays, but the gap might still widen. Following the pattern of IQ scores over time simply doesn't tell us what is happening with the actual level of intelligence.