Mainstream economists (looking at you, Paul Krugman) tend to say “It’s possible that extreme inequality hurts the economy, but we can’t really find any evidence that it does.”
Well, here's some. Using “Big Data” research techniques, these economists have realized something fundamental (and forgive me if I’m not summarizing it well): businesses live and die by their efforts at growing their customer base. If all income gains go to the top 1%, everybody else is left treading water—the potential new customers are a tiny fraction of the population, and they only buy exclusive, rich-oriented products. Their conclusion:
[T]here are two distortions that hurt growth when income grows so unequally. First, if income grows equally, then the 127 million American consumer units (households and families that buy things) all become potential new customers. Firms would then chase them, and the competitive dynamics of the market would create new opportunities to grow or create businesses. But, when only 1% have rising incomes, that is a growth of 1.2 million potential new customers. That is a vastly smaller set of opportunities for firms to grow.
Second, it is a limited set of tastes and preferences to go after; it is a market that lacks the scale for creating large numbers of jobs and production efficiencies that come from a mass market of 127 million new customers. This hurts productivity growth, as more jobs are created and aimed at smaller scale production.
So, rather than ask individual firms, "What would a $15-an-hour wage mean in paying their workers?" firms should be asked, "What would a 100-fold increase in their customer base mean?" Most firms are more concerned about the latter, without an understanding of ways to make that happen. But, if the economy is to grow, be dynamic and benefit workers and companies both, companies need to think about what policies make growth more equal.
I've written before about the many reasons we should tax the rich more heavily (spoiler: it’s not about making them pay “their fair share”). This is yet another one.