Do knowledge workers disadvantage service workers?

Caught an interesting interview with Richard Florida on NPR this morning. The key message was that in cities that are growing through increasing reliance on knowledge workers, there is also a problem with rising income inequality. The interview then focused on what might be done in order to reduce the inequality arising between knowledge workers and service workers. Florida correctly identifies that it is as an issue of productivity. When the interviewer asked for examples of workable policy solutions, Florida really didn't have anything that hasn't already been tried and found wanting.

The crux for Florida is that

The full effects of talent clustering are even more insidious. Avent points to research by Rebecca Diamond, a graduate student in economics at Harvard, which shows that this sorting process involves migrations away from as well as to knowledge-based metros. As she puts it, "[t]he combination of desirable wage and amenity growth for all workers causes large amounts of in-migration, as college workers are particularly attracted by desirable amenities, while low skill workers are particularly attracted by desirable wages." But this leads directly to higher housing costs, which according to Diamond "disproportionately discourage low skill workers from living in these high wage, high amenity cities." This creates an additional level of inequality — inequality of well-being — where more skilled workers not only take home more money, but benefit from better neighborhoods, superior amenities, and better schools. This well-being inequality, Diamond explains, is an additional 20 percent higher than can be explained by the simple wage gap between college and high school grads.

The whole interview was a great example of the critical meandering that we do - sometimes to a productive end and sometimes not. The researchers are well intentioned and using useful research tools towards important ends, but because they are more hedgehogs (they know one small thing well) than foxes (they know a little about many things), I think they end up in the wrong place.

In this case there is a lot of meat for discussion. The first issue is whether income inequality is actually rising in these cities. Let's accept that it is for argument purposes.

The next issue is whether income inequality is a problem. If we are in a zero-sum situation it certainly could be, but what about the case where all parties are realizing increases in income and well-being, it is just that some are increasing at a faster rate than others? This was the case in the US in the 1990s and early 2000s. The consequence of this scenario is that we have both rising well-being and rising inequality. This might offend our sense of equity but does it have any real-world consequence? I have seen no empirical evidence that it does. And is there anything that can be done about it? That also is not clear. All OECD countries have seen increases in their income inequality distributions in the past three decades. Some have more equal distributions than others but all are becoming more unequal for reasons we do not understand. It almost seems as if rising income inequality is a consequence of income growth.

But let's go back one step further. What is the evidence that there is increasing inequality in cities that are growing by switching to a knowledge worker economy. This finding is discussed in Richard Florida's article More Losers Than Winners in America's New Economic Geography. The first thing to note is that, as is so often the case, the strong conclusions are based on relatively few data points demonstrating rather weak correlations. We want to know answers to our important questions but our data is insufficiently robust to provide reliable answers. So we go with the tentative indications the data does provide and in our conversation we quickly lose all the caveats and hedgings that we start out with and end up treating these weakly indicative results as if they were true results.

That's just the nature of these discussions - something to be guarded against but a frequent problem.

More fundamental is what appears to me to be a problem with temporal misalignment in the researchers study. It appears to me that the researchers are not dealing with apples to apples comparisons. Specifically they are looking at Income levels and Housing costs, failing to take into account that these are driven by quite different variables. It is like treating operating expenses and capital expenses as the same thing because they both are expressed in dollars. You can get away with that for a short while but any business that fails to account for the differences between operating expenses and capital expenses is soon out of business.

The key finding that the researchers arrive at is that in cities with high concentrations of knowledge workers, the price of houses goes up. New knowledge workers coming in can afford the inflated prices of housing. New service workers coming in, who may be paid more than they used to be before the conversion to the knowledge driven economy, will still not be paid enough more to cover the inflation in housing and therefore are worse off. To the researchers, this represents a problem of income inequality. Their proposed solutions are the tired old ones of minimum wage, better education and training, etc.

The first observation I have is that I think they have missed the boat because of temporal misalignment. They are looking at the prospective impact not on the actual impact. In other words, it is true that for new workers coming into the new economy, those that are lower compensated service workers will be less able to afford inflated housing than knowledge workers and will be worse off than the knowledge workers. They may be better off than they were and they may be better off than where they came from but they are comparatively worse off than those in the more remunerative fields. But this seems a little like bemoaning the fact that we can't eliminate the problem that there is always a bottom 20%. If everyone is better off, then what are we worried about?

The related observation is that the temporal misalignment misses out on other beneficiaries. There is one set of issues for newcomers as described above. There is another issue for those already here. Existing low-income homeowners benefit from the inflation of their capital stock. What was once a $50,000 home in a marginal neighborhood is now a $200,000 home in a rising neighborhood. The low-income worker can choose to stay with the rising tide, in which case they are unaffected by the issue that the researchers are concerned about (income adjusted for new housing costs), or they can realize a sizeable capital benefit by selling and moving to another neighborhood that has not experienced the increase.

In a dynamic economy of decision-making agents there are near infinite trade-offs to be made: consume now or save for later; realize capital gains or maintain investment, reduce capital costs by increasing commuting costs, etc. The question is what are the net benefits and costs and who are the winners and are there any losers. Do the costs experienced by new incoming service workers outweigh the benefits to the existing property owning service workers? I don't know the answer but I know that we need to know before we start getting too concerned about the new workers.

There is another temporal issue in the researchers analysis and that is one of time period. They are taking a snapshot in time rather than a long term perspective of before and after. Famously, the Chinese Prime Minister Chou En Lai, when asked what the net effect of the French Revolution had been, is supposed to have answered "It's too soon to tell."

Similarly with a dynamic economic system. Because of stickiness of wages, price of consumer goods, and capital costs there will always be smaller or larger gaps in the market. Housing prices rise and fall, out of synch with wages which rise and fall for given types of work, and which are out of synch with the cost of food and energy rising and falling. Where there is open competition in a free market not sullied by collusion, regulatory capture and rent seeking defects, those market imbalances between supply and demand will clear relatively quickly. It does little good to obsess about the near term imbalances. What is important is what are the long term implications and who are the long term winners and losers. In the scenario the researchers are looking at, they are expressing concern that new service workers coming in are at a disadvantage to new knowledge workers because of their differential in productivity and because of the increased costs of housing. But all that is evidence of is that the market participants are using the pricing mechanism to signal relative degrees of scarcity and abundance. Left to its own, the market will come to some equilibrium over some extent of time. Service workers wages will rise to cover the cost of housing, service work will be replaced by automation, public transportation will improve to allow low income service workers to live elsewhere more cheaply, an excess supply of knowledge workers will lead to a decline in knowledge worker income leading to a reduction in inflationary pressure on housing: there are an infinite number of combinations that are likely to arise.

And that is the beauty of the market - in a system characterized by change, uncertainty, internal feedback mechanisms, etc. it is not possible to accurately forecast long term outcomes. Instead innumerable autonomous decision-making agents individually and unconsciously evolve to a sophisticated answer to the incomprehensible problem.

It would seem to me that the researchers, absent a grounding in market mechanisms and economics, are potentially obsessing about what might be a non-problem.

But while I suspect that their putative concern is misplaced and that their research is flawed, it is still a good catalyst for critical thinking about important issues.

One other thought is prompted by their research and the radio interview. I am confident that Florida is absolutely correct that the crux of the issue is productivity, even if we are not clear or confident about how to create the circumstances for increased productivity.

The paucity and sheer exhaustion of their proposed solutions does usefully prompt the question, what is the essential distinction between the characteristics of a successful service worker (success being predicated and measured in terms of productivity) and a successful knowledge worker? What is involved in raising the productivity of either group? Here is my take on it.

Service workers productivity is much more dependent on non-cognitive skills (behaviors and values) than manufacturing workers and knowledge workers. Manufacturing worker's productivity to a material degree and Knowledge worker's productivity to a very significant degree are much more dependent on cognitive skills (knowledge, skills & experience, decision-making).

The courses of action necessary to improve knowledge, experience and decision-making are quite different than the courses of action necessary to improve values and behaviors. Not only are they different but they are easier to reach consensus on. We are likely to reach agreement on the necessary body of knowledge, the experience and the decision-making components of a manufacturing or knowledge worker job much more readily than we are to agree on the most critical behaviors and values necessary for improved service worker productivity.

My overall concluison is that Florida is propbably wrong, but in attempting to make his argument, he is usefully wrong.