I think his point was that it's more likely taking the total income and total property costs and computing the average that way. But if you instead compute the same per-city and then look at the middle of those it can give you a drastically different value.
No, the ratio is lower in cities [1], only 16.9 in Beijing, which has the highest city ratio. I'm guessing because of the very large poor population, it drags the average income down, whereas people in cities aren't making poor farmer wages, so the individual city ratios are lower.
If the majority of population lives in rural areas, the prices of big cities have a major impact on the average, skewing it.
But actually people living in the big cities can afford housing more than a person with "rural area" salary can afford the average price, even though they are most likely able to afford the much lower prices in their area.
[0] https://en.wikipedia.org/wiki/Simpson's_paradox