One common way to measure the benefits of rural to urban migration is to compare the wages of migrants to urban areas to the wages of people who stay in rural areas. Even accounting for cost of living differences (e.g. the fact that rents and consumer prices are more expensive in cities), this comparison finds large wage differences, suggesting that there are important benefits of migration. Why might this approach not lead to accurate estimates of the benefits of migration? Explain, using the concepts of "selection" or "threats to identification" in your answer.