Indeed. And similarly, as the paper mentions, there are many places in the US where someone earning just under $100k would only be counted as middle class, even

Indeed. And similarly, as the paper mentions, there are many places in the US where someone earning just under $100k would only be counted as middle class, even though they'd be well into the upper middle class. So the important question then is there a big bias in one direction or another? The paper concludes that the answer to this is no. So why would they use such an ostensibly broad metric in the first place? The paper also goes into this. The reason is that on a micro level there are often extreme quality of living differences that are not accurately reflected by any data. For instance the CPI inflation rate of San Francisco, according to California, is only 11% higher than the US average. By any metric you'll be able to point to various little bubbles throughout the nation and indicate, accurately, that the macro level data does not hold true. So you need to try to pick aggregate data that most likely balances out the biases so much as possible. Do you think this fails to do so?