My thought process is that sports and music are a bit different.

Music is simpler in that I think it can be explained by supply and demand. As worldwide wealth has grown, more children will have access to musical instruments at an early age, and will find it easier to get into top training programs to learn classical music. Meanwhile, people have more options for musical styles so demand in classical music has relatively decreased. I recall Jimmy Page in the documentary "It Might Get Loud" remarking on how boring music used to be before the 60s. Rock and Roll was a worldwide phenomenon because nobody had heard anything like it before. I think another major growth occurred at the turn of the century beginning with Napster when the internet gave people a lot more options for finding music.

I'm more inclined to believe the sports story because sports competition yields constant feedback. The teams that don't perform are spotted quite quickly and forced to reform. You can get a reasonable approximation of a player's ability by estimating points scored per minute vs. points given up per minute, and that information can be updated constantly. The best trainers will produce the best competitors, and so will indirectly also receive feedback on their performance.

I don't place much trust in most international comparisons on education because most of them don't make demographics adjustments. Finland and Japan both have fairly low immigrant populations. It's a lot easier to educate a population when they all start out speaking the same language. The article mentions Canada once, which does have a large immigrant population, but doesn't go into any detail about their system. Some important considerations:

Some studies have found the gains of the Flynn effect to be particularly concentrated at the lower end of the distribution. Teasdale and Owen (1989), for example, found the effect primarily reduced the number of low-end scores, resulting in an increased number of moderately high scores, with no increase in very high scores. https://en.wikipedia.org/wiki/Flynn_effect

In the 2002 book IQ and the Wealth of Nations, and IQ and Global Inequality in 2006, Richard Lynn and Tatu Vanhanen created estimates of average IQs for 113 nations. They estimated IQs of 79 other nations based on neighboring nations or other via other manners. They also created an estimate of "quality of human conditions" for each nation based on gross national product per capita, adult literacy rate, fraction of the population to enroll in secondary education, life expectancy, and rate of democratization. Lynn and Vanhanen found a substantial correlation between the national IQ scores they created and these various socioeconomic factors. They conclude that national IQ influences these measures of well-being, and that national differences in IQ are heavily influenced by genetics, although they also allow for some environmental contributions to it. They regard nutrition as the most important environmental factor, and education a secondary factor. https://en.wikipedia.org/wiki/Nations_and_intelligence

My guess is a lot of undeveloped countries, especially the poor, aren't reaching a minimum education level; whatever that minimum might be although causality could run in a number of directions. I don't agree with Lynn and Vanhanen's proposed explanation. I'm not sure how effective better training actually would be without a feedback mechanism. How can you assess the quality of training without one? I've been meaning to read Elizabeth Green's book so may get around to that soon enough.