Robots vs. Human Labor: The Smackdown

There’s a war going on—you just don’t see it or feel it.

Yet.

Which war? We’re talking about Robots vs Human Labor: the smackdown.

Really? Again? Haven’t we been hearing about robots replacing Labor for half a century? Is this really an issue we have to worry about? It feels like we’ve been through this dance 100 times in the last 50 years, and it always ends up being a joke when the battery is pulled out of the back of the robot and someone raises a red flag as Labor wins.

For the most part, yeah—that pattern has in fact been the case. Robots have been deployed by Labor to make jobs better and more efficient. Even the American auto industry, which fought the adoption of machine-assisted-manufacturing until it almost killed itself, has given in. Today, robots are a key element inside the plants at GM and Ford. They can also serve cocktails and work as lab technicians.

Robots in one form or another are here to stay—but there really is something different in the waters today, something that wasn’t around 20 years ago or even 5 years ago: the tools, machinery, operating systems, battery power, diagnostics, competitive forces, and other controls which feed the productivity of robots have become seriously cheap to produce. And they’re being produced on a mass scale.

For the first time in history, there will be a rational financial discussion revolving around the capital efficiency of deploying a marginal robot instead of a marginal human. Sound cold?

You got it.

It’s All About the Benjamins

But here’s the thing: companies hire managers to make them money. It is the job of the CEO and everyone below her to add value to the bottom line—and if they don’t, the company fades into mediocrity and then oblivion in a globally competitive marketplace. See Europe for details.

So if Company X doesn’t take advantage of all that technology offers to profitably sell a lug nut for $1.28, then Company Y will take advantage of all that technology offers and sell that lug-nut for $1.02. Goodbye, Company X. It sounds harsh, but...so is Darwin.

The classic economic calculation that CEOs make goes like this: marginal product of capital over cost of capital equals marginal product of Labor over wage, or

MPK/K = MPL/L

Marginal product of capital (MPK) is the additional output from one additional unit of capital (K). Marginal product of Labor (MPL) is the additional output from one additional unit of Labor (L).

Basically, these CEOs look at the productivity of Labor over its wage rate versus the cost of capital to its efficiency. The cost of capital (i.e., machinery) has a big fixed cost—but little marginal cost. When the capital needed gets a whole lot smaller, it puts pressure on what management is able to pay Labor.

Consider this:

For producers, the end goal of producing goods is to make as much profit as they can. How do you calculate the profit? It’s the total revenue of all products produced minus the total cost of all products produced (what you made minus what you spent). If you want to break it down into averages, just divide the total profit by the total output.

So how do producers maximize their profits? Producers can increase the profits they make by reducing the cost of the inputs they use in production.

The problem is that most products on the market today are not made up of just one input. In fact, it is near impossible to think of anything that only takes one factor of production to make. So not only do producers have to worry about minimizing the cost of their inputs, they also need to consider the best combination of inputs that will help increase the total output but keep costs as low as possible, given the prices and production constraints they have.

If you've already learned about consumer utility maximization, you know how to figure out the best possible, lowest costing input combination. Here’s the equation:

MPL/PL = MPK/PK

Here’s an example:

McDonald’s is having a special: they’re selling their large McFlurry ice cream for $2.00 each. McDonald's needs both the McFlurry machine and the manpower to run the machine if it wants to produce these delectable treats. The goal is to sell 100 McFlurries per hour...at the least possible total cost. Each worker is paid $8.00/hour, and the McFlurry machine costs $20.00/hour to run and maintain. The marginal physical product, or additional output of the next marginal unit of input, of the worker is 24 McFlurries/hour; for the machine it’s 60/hour. So is McDonald’s minimizing their costs?

Would you like fries with that?

MPL/PL = MPK/PK
24/$8 = 60/$20
3/$1 = 3/$1

In this case, McDonald’s has done a good job in finding the best possible combination of inputs in order to minimize cost. 

But what if it hadn't?

Why does increasing one input decrease the marginal physical product? Production has diminishing marginal returns. After a certain point, adding more of a variable input within the constraints of a fixed input in the short run will actually hurt productivity and profits rather than increase it.

When producers are successful in finding the best combination of inputs in order to minimize costs in making goods and services, they're said to be at productive efficiency. This way, producers can maximize their output with the given constraints due to fixed inputs and, in the end, maximize their profits.

Everyone’s Favorite Robot: The ATM

Robots are not new; the only “new” element we’re beginning to encounter is that they’re cheaper to produce. Think about it: a computer with the processing power of your new iPhone (which cost about $400 today, give or take), in 1980 would have cost about $25,000. Same computations—just much less math per penny.

It’s also not new that society and business adopt and embrace robots at scale when they make sense.

We’d say that the greatest robot in history from a consumer perspective is The Versateller/Automated Teller Machine—you know it as an ATM. It counts money accurately, it’s usable 24/7, and it gives you 85% of everything you’d need from a bank. Plus, it’s never in a bad mood and it doesn’t ask you how your day is going as if it cares.

Faceless banking became the darling of the American banking experience in the '90s, and Americans clearly loved it. If they didn’t want a machine taking their deposits, checking their balances, and counting out their $20s, they would have waited for that “friendly” human face behind the counter to do it for them between the hours of 10:00 AM and noon and then 1:00 PM and 4:00 PM...plus three or four hours on a Saturday. If the world didn’t want Versatellers, there wouldn’t be millions of them installed.

But that’s not the case. We love our robots in banking.

How many bank teller jobs did those machines replace? Lots.

Then...the economy grew. And the scope and influence of banks grew. And with more liquidity in the system brought on by easy access to cash, you could argue that the Versateller robots actually helped the economy grow.

But what about those jobless tellers? What did they end up doing instead? Probably other semi-professional service jobs.

Versatellers crept in slowly starting in the late 1980s. And without a tidal wave of unemployed laborers from many industries, workers could find other venues in which to gain employment. The Versateller Transition was relatively painless, and the world of human workers survived that battle.

The McFlipper

Today we face a very different landscape.

Human economic vulnerability to robot labor changes dramatically when robots are cheap to produce and easy to scale. Duh. Commodity parts, standardized operating systems, a competitive market producing cheap servo motors and power systems, and simply better management of the processes has come together to create the much-discussed McFlipper.

The McFlipper is a robot that costs $100,000 to build and does a dozen things:

1. It pours drinks.
2. It flips burgers.
3. It senses via laser when burgers are done.
4. It cuts potatoes.
5. It fries fries.
6. It takes orders from windows speaking perfect English or Spanish.
7. It “markets” the new sandwich du jour by asking the buyer if they’d like to try the McSushi Platter for a dollar off.
8. It makes correct change from a $20 or debits correctly each time from an iPhone app.
9. It loads new buns as needed.
10. It senses when the grill needs to be cleaned.
11. It cleans the grill.
12. It calls for its own maintenance every 1,432 hours of operation as required by the people insuring Mr. Robot McFlipper.

The McFlipper does not do a few things:

1. Accidentally sweat or drop hair in the food
2. Be late to work
3. Get sick
4. Steal
5. Complain about work conditions
6. Need much human management
7. Need recruiting
8. File lawsuits for wide ranges of real or non-real abuses

Flipping burgers used to be the start of a blue-collar career that led to store management—maybe even regional management—and was one of the best ways for a high school graduate to earn a nice living and have a high quality life.

There are about 3.5 million similar jobs in the U.S. (and probably 6 times that in the world). Yeah, that’s a lot of jobs.

So let’s do some math.

A McFlipper costs $100,000 today—that’s for a robot that performs all of the above functions, at modest scale. (Obviously the very first McFlipper will have cost millions to produce with lots of research and development and failure having been a natural part of its construction.)

So at big scale—like if we are producing a million of them—we’re going to say that it costs $50,000 each. Uh, how can a relatively simple robot cost more than a very nice car? Well, it’s likely that the $50,000 per unit comes down dramatically over the next decade or so. See: car prices.

So this robot replaces…what?

Let’s see. The robot works 24 hours a day, 7 days a week, which are the hours of most fast food joints. That’s 168 hours. One full-time worker works 40 hours/week and earns $15/hour or $600/week or, say, $30,000/year. The company then pays pension and taxes and benefits on top of that $30,000 so that it costs their employer about $40,000 to pay that worker.

So if a McFlipper is going to cost $50,000 and costs, say, $5,000 a year in electricity, maintenance, insurance, and cleaning and has an expected lifespan of 10 years—at which point it is literally just thrown in the trash heap to be melted down into fishing hooks—we can “straight line” depreciate its value to the tune of about $5,000 a year. It costs the company $5,000 in depreciation and $5,000 in maintenance for a total of $10,000 a year.

Now there are other costs with the McFlipper, of course. It had to be procured (i.e. bought properly), installed, managed through that little earthquake, rebooted a few times during that lightning storm and, oh yeah, there was interest expense associated with the cost of that initial $50,000 in capital. And probably a bunch of other things we can’t predict like when the ghost of Jimmy Hoffa inhabits its operating system and makes it…dance.

But those costs map reasonably onto the additional costs to hiring—and firing—human beings to do the same jobs the robot is doing. There is “load” in real corporate America today in the form of lawyers, human resources, lots of exchanged paperwork, and a store manager who has to chide tardiness. Let’s just say that there’s a wash there in “load costs.”

So, in a given year where a McFlipper replaces 4 humans at $40,000 each for a total cost of $160,000 (and where the McFlipper costs $10,000 a year), that one robot is saving $150,000 a year. (How’d we come up with 4 bodies? Well, one body works 40 hours a week and a given store is open 168 hours. Using advanced Andreessen-approved calculus, that’s about 4 bodies that one robot replaces, assuming full utility of all 168 hours— and yes, at 2:00AM, the place likely doesn’t need to be fully staffed.) Multiply the number by a dozen employees per store and it’s stunning savings. It would let the fast food joint drop the price of burger meals 50 cents or more each—and still make the same profit.

Shareholders own the corporation, so they’re the ones making the tactical robot decisions. Shareholders (i.e., owners) hire management to make them money. And how do you make money? By spending less. Fast food is a highly competitive industry with loads of local players fighting the big national chains in a daily wrestling match in the real hunger games. Every penny counts, and if a McFlipper helps save even a nickel or a dime a burger, that’s big. In practice, per meal, it probably saves a lot more. Billions of dollars flow to the bottom line for shareholders and...they’re happy.

Would it be malfeasance/mismanagement if a company didn’t adopt the robots to do its bidding? Would you vote to hire management that made you less money than they could have? Do you think consumers will vote to pay 50 cents more for a meal made only by humans versus one with robots?

Probably not.

Other Industries

Of course, it’s not all about the McFlipper. What about other industries?

Perhaps the most inked arena of our time is the transportation industry: the driverless car, the driverless truck, the driverless ship…oh, and ever see the video of the Tesla factory? Or the humanless floor at the Amazon shipping department?

What about tollbooth operators, DMV workers, and moon walkers?

What we’re painting here is a tsunami—not a gentle rain like the one the world felt when Versatellers were unleashed around the world. And perhaps ironically, it’s worth noting that robots eat robots: it’s pretty likely that payment apps on cell phones, coupled with virtual currencies (hello, Bitcoin), will replace a very large percentage of Versatellers in the world. After all, the hard currency cash that is their primary value-add will no longer be needed at such full scale. That is, Versatellers will exist… but a third or fewer will remain in operation.

So what’s gonna eat the McFlipper?