Unpleasant Facts and Other Musings

Net neutrality is not something people who prefer a free market economy would support apriori. It goes against the idea that businesses should be able to compete how they see fit. If a business wanted to lay fiber and give away cheap internet on the condition that their social network, streaming service or Amazon referral links are used, then in theory that should be okay. But we don’t have a competitive open market with many different players. The internet service providers have relationships with municipality governments that help them deploy fiber to their residents. Sometimes the municipality keeps the ISP’s competitors out due to relationships between local officials and ISP lobbyists, other times it is due to bureaucratic indifference or incompetence. It takes a concentrated effort to avoid the accumulation of cumbersome rules and regulations that cause the high costs and slow deployment times that will keep ISPs out of an area that already has a competing provider. Either way, ISPs often achieve monopoly power over their local markets. A 2013 report from the US Department of Commerce shows very oligopolistic competition at best for providers at speeds over 25 Mpbs, and almost no competition above 100 Mbps.

Given time and an incentive to increase their revenue quarter after quarter, companies will attempt to extract additional rents from their position as the gatekeeper consumers and online businesses. Net neutrality rules force ISPs to act as dumb pipes to the internet will prevent these local monopolies from using their government-backed power to pick winners and losers on the rest of the internet.  In our status quo of regional internet service provider monopolies and oligopolies, some form of net neutrality is a desirable compromise.

The 2015 Title II Order by the FCC which was an implementation of net neutrality in the United States is seen by many as a regulatory overreach. It wasn’t because their general framework was that far off, but it was quite a stretch to independently label ISPs as public utilities under authority used to regulate telephone monopolies in the 1930s. This type of significant rule change without congressional input is worrisome for those who want the responsibility of lawmaking to fall upon an elected legislative body. We shouldn’t look to the FCC to implement these rules on their own, the rules should be spelled out by a congress that is concerned with protecting interstate commerce from companies who are empowered by barriers to entry put up by local governments.

One thing that might get lost in this process is the implementation of internet service providers with unique business models. It would be best if room for experimentation was left open in areas where there is already significant competition between multiple providers at high speeds. Then we might see if consumers will be willing to forgo their dumb pipe options in return for a lower cost service. This is the last thing that major tech incumbents want. This approach might allow emerging tech companies buy their way into customer scale, or it might force to major tech companies to spend more to entrench their positions. Either way, the ISPs might be able to figure out ways to make a little more profit while their customers find additional consumer surplus. The key is that the ISP is trying out these innovative business plans in scenarios where the consumer has other reasonable options.

A recent New York Times article tells the tale of two janitors. One at Eastman Kodak in the early 80's had their education paid for and later worked her way up to CTO, later going on to be a senior executive at other companies. The other is at Apple today and doesn't even directly work for Apple. She has minimal benefits and no obvious path for career advancement. That's in large part because she works for a company that contracts out her labor and not for Apple directly*.

The narrative the Times advanced is that companies chose to focus on their core competency and have outsourced other work. Most modern tech companies have decided to focus on their most productive employees and are letting contractors deal with the supposedly interchangeable employees working on low skill problems.

What many people don't see is that this is not an economic relationship that has sprung out of thin air. A lot of it can be interpreted as a company's logical response to the regulatory environment. There are at least three different areas that incentivize companies to keep low skilled workers at arms-length, only hiring them through secondary employers. There are laws around unionization, legal risks from the Equal Employment Opportunity Commission and the Affordable Care Act has made the choice even more clear from a cost perspective.

At first glance, it's obvious why laws about unions would cause companies to choose to contract their labor. For companies that are innovating quickly, becoming unionized raises the spectre of what happened to Detroit auto companies. The long term accumulation of rules and resistance to automation that would cost jobs helped cause the Big Three to fall far behind their more nimble competition from Japan. As Steve Babson puts it in "Working Detroit: The Making of a Union Town"

"The absence of union factory rules gave Japanese management the flexibility to change workloads and reassign jobs without opposition but it also left workers with little protection against speed-up of favoritism."

Today's tech companies need to move fast and keep up with a quickly changing technology landscape, so their instinct is to avoid unionization. How do they do this? First, they spoil their workers more than any union ever did. Companies provide meals, social events, gym memberships and some tech companies even do laundry for employees. This is a multi-purpose policy, not only do these perks discourage talk of unionization, they are also meant to encourage employees to spend more time at work and make them less likely to leave the company even when they are being paid significantly under their market price.

But keeping the higher skilled employees is only part of the equation. The second is keeping employees that are likely to unionize outside of the company. Low skilled workers are most likely to form unions. Unions might possibly strike and upset the sympathetic high skilled employees. The laws empowering the National Labor Relations Board grants union employees specific protection. But those protections stop when they target employers other than their own. From the NLRB FAQ:

"A union cannot strike or picket an employer to force it to stop doing business with another employer who is the primary target of a labor dispute."

After looking at this rule, it should be obvious why companies would want to avoid giving employment status to a class of workers that are unionized or might decide to form a union in the future. By hiring these workers through another employer, they are shielding themselves from the more inconvenient aspects of labor law.

Next we have the Equal Employment Opportunity Commission. The EEOC does not only look for employers who are discriminating in hiring based on age, race, sex, etc. They will also punish treatment of employees based on these factors. If a corporation wants to treat some employees as first class employees with more benefits and another group as second class, they would have significant legal risk when the second group consists of employees that are more likely to belong to a protected class. (And the Times piece does mention how there are legal requirements that employees are offered the same health insurance and 401(k) benefits.) The EEOC will be collecting pay data by sex, ethnicity and race. Under this scenario, it will look a lot better for everyone involved to continue contracting out the low paid jobs.

On top of all of this we have the Affordable Care Act's impact on full-time employment. The ACA puts the burden of providing healthcare fully on full-time employers while leaving part time employers and contract workers almost completely off the hook. The question of who should be responsible for healthcare payments is a political question, but it should be applied equally to all types of employment relationships. Two part-time employees working 20 hours a week should cost an employer the same as a single full-time employee working 40 hours a week. That's not the case today, the 40-hour workweek employee is more expensive. And rather than be the bad guys who are hiring part time workers to avoid paying for health insurance, it is easier for corporations to pay another company to hire and manage workers in this manner.

When designing and implementing policy, it's hard to account for the long-term impact. In this case, the incentives created by multiple regulatory bodies have combined to raise the risk and cost of full-time low-skill employees. They are highly incentivized to hire another company to shield them from a direct relationship. Laws that were originally intended to protect people and keep society integrated might be serving to bifurcate it even further.

*On top of the low wage she pays inordinately high rent to live in the area. A significant part of the struggle of low wage workers in the Bay Area comes from the lack of affordable housing. Most people should realize by now that this is an artificial hardship imposed on the poor by voters who think they are preserving the character of their neighborhood, preventing sprawl or protecting the environment. In reality, the main concern of voters is boosting the value of their house after they have locked in the amount of taxes they have to pay thanks to California’s Proposition 13.

There is a social phenomena called Pulling up the ladder. The basic idea is that when a person or group has success in a certain way, they advance policies that prevent people from having the opportunity to succeed in a similar manner.

Pulling up the ladder is seen in NIMBYism. Established residents who built their houses in a neighborhood pull up the ladder when they decide that extra rules and permits are needed and that no one else should be allowed to build a home in their neighborhood as easily or as cheaply as they did. Pulling up the ladder is seen when practitioners promote occupational licensing that forces new entrants to go through thousands of hours of training and low paid apprenticeships, but grandfather in current practitioners who do not have to follow these burdensome rules. 

Bill Gates provides us with an egregious example of pulling up the ladder in a recent interview with QZ.com when he advocates for the taxation of any robots that replace human jobs.

Certainly there will be taxes that relate to automation. Right now, the human worker who does, say, $50,000 worth of work in a factory, that income is taxed and you get income tax, social security tax, all those things. If a robot comes in to do the same thing, you’d think that we’d tax the robot at a similar level.

...

There are many ways to take that extra productivity and generate more taxes. Exactly how you’d do it, measure it, you know, it’s interesting for people to start talking about now. Some of it can come on the profits that are generated by the labor-saving efficiency there. Some of it can come directly in some type of robot tax. I don’t think the robot companies are going to be outraged that there might be a tax. It’s OK.

First, anyway you cut it, this is just really bad policy. Companies utilizing robots should be taxed with the same rules as every other company. Tax theory is not something that can be covered in a short blog post, but there are two frameworks that make it easy to see why this is a bad idea.

One framework favored by some economists is that taxes should be designed to force people and organizations to internalize negative externalities. This is called a Pigouvian tax. If a certain behavior is not good for society, then raising its price will reduce the harms to society while generating revenue. This type of approach is exemplified by those pushing for a carbon tax. Carbon is identified as harmful, and raising the price of release carbon into the atmosphere will likely result in less carbon released into the atmosphere (improperly implemented, it also incentivizes more manufacturing in countries that do not have carbon taxes). Cigarette, alcohol and sugary drink taxes are often justified under a similar approach, as consumption of unhealthy goods can create significant costs for the healthcare system*. The flipside to this logic is obvious. Taxes should also encourage, not discourage, desirable behavior. A tax designed to explicitly raise the cost of research and investment relative to other activities would be a very bad idea.

Another framework is that taxes should avoid unnecessarily distorting economic behavior. Even economists who promote Pigouvian taxes would agree that a tax that changes behavior without purposefully targeting a desired externality is a poorly designed tax. Taxes that can be avoided with additional work from lawyers and accountants are particularly inefficient. Resources devoted to getting around the tax may make sense for individuals and companies, but are a deadweight loss to society. The value-added tax, other than encouraging exports to countries without value-added taxes, is both easy to enforce and does not skew behavioral incentives significantly. It is therefore widely used across the developed world outside of the United States.

So it’s pretty obvious why trying to apply an additional robot tax on companies would be a bad idea. Measuring whether a robot is taking a job is not a simple task, and companies will be incentivized to make the use of automation unrelated to any losses of jobs. The tax would cause significant amounts of new inefficiency as companies on the verge of automating significant tasks would spend lots of money trying to figure out how to minimize or avoid the robot tax. 

But even worse, taxing the implementation of robots in existing companies would slow down the adoption of new technology. If implemented only in the United States, the robot tax would also cause U.S. companies to fall behind places in the world where the implementation of robots was not discouraged by taxation. If the United States wanted to permanently relinquish its status as the country at the forefront of the production-possibility frontier, this tax would be a great way to start.

What makes Bill Gate's suggestion particularly egregious is that the source of his wealth came from the widespread distribution of labor saving technology, software! The following is from the BLS summary of Occupational changes during the 20th century.

The growing use of computers and other electronic devices, which simplified or eliminated many clerical activities, caused the post-1980 decline. Automated switching and voice messaging affected telephone operators; personal computers, word-processing software, optical scanners, electronic mail, and voice messaging, secretaries and typists; accounting and database software, bookkeepers; ATM’s and telephone and online banking, tellers; and computerized checkout terminals, cashiers.

Imagine what would have happened to Microsoft if every time it sold a spreadsheet tool it had to pay a tax for the number of bookkeepers it replaced. Or if all users of Microsoft Office were taxed for the secretaries and typists that were no longer needed. If the United States applied the policies Bill Gates now suggests now to software in the 1980's, the digital revolution would have been strangled in its infancy and Bill Gate would not be in the position he is in today. Specifically taxing companies that implement cutting edge labor saving technology is silly. It wouldn't have been a good idea then and it's not a good idea now. 

Bill Gates has led a unique life over the past few decades so he might not realize this, but our society could still be a whole lot richer. It is still far from wealthy enough to implement any reasonably sized universal basic income. Safety nets, retraining and regulatory reform have important roles to help workers displaced by our fast evolving economy. But the last thing we want to do is implement taxes that slow down innovation or encourage innovative companies to locate its business outside our borders.

So please ignore Bill Gates when he blithely suggests that anyone using robots to replace labor should have to pay extra taxes. Not only is he advocating for pulling up the ladder behind him, listening to him would impoverish our future. Our society cannot afford to treat labor saving innovation like a negative externality to be taxed.

*Somewhat morbidly, unhealthy behaviors that shorten life expectancies actually seem to reduce total costs on the healthcare system.

On January 30th, the White House issued a new executive order targeting regulatory reform. The rule has been promoted, by both the White House and the media, as the rule that forces regulatory agencies to get rid of two rules for every new rule they issue.

Both Canada and the UK have used one-in, one-out rules, with the UK switching towards implementing two for one. The statement below is from a policy paper by the Conservative and Liberal Democrat Coalition government in December 2012 as they switched one One-In, One-Out (OIOO) to One-In, Two-Out.

It is clear that the OIOO rule has delivered a profound culture change across government as demonstrated not only by the continuing increase in deregulatory measures but also in the high number of Departments in credit at the end of the OIOO period.

Building on this culture change, Government is now pressing Departments to deregulate further and faster to free up business from unnecessary red tape and deliver growth. From January 2013, our rule is doubled to One-in, Two-out (OITO). The Red Tape Challenge will continue to be an important vehicle for Departments to reduce regulatory burdens and identify OUTs.

Given the UK’s experience, it makes sense that Trump is starting with two for one. He is not a man of half measures.

But the number rules are not the most important aspect of this policy. A single rule could have a miniscule impact, and repealing two of them might not matter if the new rule imposed gigantic costs. The executive order does not overlook this:

“In furtherance of the requirement of subsection (a) of this section, any new incremental costs associated with new regulations shall, to the extent permitted by law, be offset by the elimination of existing costs associated with at least two prior regulations”

Trump essentially told the agencies that he is capping the costs they are imposing on the economy. They can reduce costs by being more efficient, or by prioritizing their rule making around the most important problems. He is giving every agency a regulatory budget.

The idea of a regulatory budget is simply applying the concept of a budget to regulations. If applied correctly, government agencies are only be allowed to impose a certain amount of cost on society regardless of their net benefit. As agencies are eager to remain relevant and fix the greatest problems of the day, they will be incentivized to remove or remake some of the costlier rules. The benefits of regulatory policy are not ignored, they are taken into account by the amount of cost each agency is allowed to impose on the economy. Trump’s executive order is designed so the regulatory budget of agencies will be increased or decreased, as needed.

During the Presidential budget process, the Director shall identify to agencies a total amount of incremental costs that will be allowed for each agency in issuing new regulations and repealing regulations for the next fiscal year.  No regulations exceeding the agency's total incremental cost allowance will be permitted in that fiscal year, unless required by law or approved in writing by the Director.  The total incremental cost allowance may allow an increase or require a reduction in total regulatory cost.

While the order looks sound, implementation will be key. First, there is the question of how these costs are measured. The direct cost of paperwork and compliance are not the only costs imposed by regulations, there are also indirect costs that are difficult to take into account but which are often larger than the direct costs.

Second, there is the question of prioritizing the right rules to cut. Right now, this order has implemented partial regulatory budgeting, probably the best that can be done without support from Congress. The costs that the agency can impose, along with the number of rules, has been capped. And while we know the total number of rules issued by each agency, we do not yet know the costs and benefits associated with each rule. Only some rules have been analyzed, and many of those analyses are out of date. Agencies enforcing many costly rules with few real benefits should be made to cut much more than two for one, while agencies providing significant benefits in excess to their costs might be given increase in their regulatory budgets. This is an approach favored by Cass R. Sunstein, one time head of Obama’s Office of Information and Regulatory Affairs.

It is important to remember that regulatory budgeting isn’t some new crazy idea. Canada and the United Kingdom have applied regulatory budgeting. In British Columbia, the Deregulation Office used a regulatory budget approach to cut the number of requirements to 55% of their 2001 level. This is made easier in Canada and UK because the ruling party of coalition in a parliamentary system can change the law. In the US, when an agency gets rid of an outdated and costly rule, they may run into legal obstacles if that rule was legislatively mandated.

It would be best if the agencies were operating under the explicit orders or Congress and the Executive branch to create real change. This isn’t just about rolling back inefficient regulations, it’s also about making the government work far more efficiently. It’s a good start.