The marginal productivity of geniuses: why Apple's workers earn modest wages

Apple has recently come under fire for the low wages paid to its retail employees, the sales "specialists" and technical support "geniuses." According to a recent New York Times article, Apple is doing something "unique in the annals of retailing". It pays "a modest hourly wage, and no commission, to employees who typically have college degrees and who at the highest performing levels can move as much as $3 million in goods a year." 

Undergraduate students learn that, in a competitive market, wages are equal to the marginal value product of labour, that is, the dollar value of what the marginal worker produces. Formally, w=p*MP_L where w=the wage rate, p=the price of output and MP_L is the marginal product of labour.

The proof of this proposition is simple: if an additional worker produces goods and services worth, say, $12, and the wage rate is, say, $10, hiring more workers adds to the firm's profits. Thus, if the marginal value product of labour is greater than the wage rate, the firm will hire more workers. Eventually, however, as more workers are hired, their productivity will start to fall. For example, the additional workers will take on lower and lower value tasks. Once the marginal productivity has fallen to the point where w=p*MP_L, the firm stops hiring.

By the same reasoning, if the wage rate is more than the value of what a worker produces, the firm will fire workers. The only possible equilibrium is one where wages are determined by productivity.

Note that I'm assuming the market is competitive, so no one firm's hiring decisions has any impact on the prevailing wage rate, and each firm is too small to have any impact on the market price of its output. 

Apple's policy of no commissions and moderate wages might appear to violate the "workers are paid the marginal value product of labour" rule. After all, according to the New York Times, "Divide revenue by total number of employees and you find that last year, each Apple store employee — that includes non-sales staff like technicians and people stocking shelves — brought in $473,000." If one tenth of the value of Apple store sales can be attributed to their retail employees, that would imply an average annual salary of $47,300, and a typical employee makes closer to $25,000.

"The marginal product of labour" is an abstraction. The idea is simple enough - what an additional worker brings to the production enterprise - but the implementation is not.

For example, think about the simplest possible production technology: one worker + one shovel digging a hole. Without the shovel, the worker produces nothing. Without the worker, the shovel produces nothing. Together, they produce a hole in the ground. What is the "marginal productivity" of the worker? Of the shovel? It's like the sound of one hand clapping - it's impossible to tell what each member of the production process contributes.

[Update] Another way of putting this is that the very idea of a marginal product only makes sense with certain types of production functions, ones for which it is possible to make small adjustments in the quantity of each input used. For example, if I am making a cheese sauce from scratch, I can work out the 'marginal productivity' of cheese, the amount of extra deliciousness added by each additional tablespoon of grated cheese. When input ratios are fixed, however, the marginal product calculation makes less sense. For example, making a white sauce from scratch requires a fixed ratio of flour and milk (2 tbsp flour to each cup of milk); any less it's too runny, any more it's too thick. The marginal productivity of flour is at first very high (without it the sauce couldn't be made) and then abruptly drops to zero. The sudden drop in the productivity of, say, flour means that the marginal product is - at the point where the marginal product function is discontinuous - undefined.

With fixed-input-ratio production technologies, like one shovel/one worker, allocating the profits is like dividing a pie. When two parties are bargaining over who gets a pie, no one will ever accept less than their outside option -  the amount that she could get if she walked away. If the worker's outside option is $10 per hour, he will never accept any wage less than $10 per hour; if the outside option is $2, a wage of $7 starts sounding pretty good. 

Outside options explain why Apple can pay its workers a fairly low wage. If a typical Apple specialist or genius walked away from their job, their next best alternative might be an even worse retail position.  

Could Apple just walk away from its employees? At an individual employee level, it can - there are more people looking to work at Apple stores than positions available. The picture to the right suggests that Apple might be able to walk away from its specialists at a collective level, too.

That is the basic problem with the argument that Apple workers should earn high wages because they sell products with thousands of dollars. Ultimately, it doesn't matter how much you sell, if you can be replaced by a vending machine, you are in a weak position to bargain for a higher salary. 

Comments

[Min]

"Apple's policy of no commissions and moderate wages might appear to violate the "workers are paid the marginal value product of labour" rule."

What about the dictum that "incentives matter"? IMX, commissions are normal for salespeople. Those on a straight salary are derisively called "order takers". What is behind Apple's policy of no commissions?

[Frances Woolley]

Min, according to the NY Times article quoted above: "At Apple, the decision not to offer commissions was made, Ms. Bruno said, before a store had opened. The idea was that such incentives would work against the company’s primary goals — finding customers the right products, rather than the most expensive ones, and establishing long-term rapport with the brand. Commissions, it was also thought, would foster employee competition, which would undermine camaraderie."

In terms of incentives: "Though commissions are not offered, many managers keep close tabs on sales of warranties, known as Apple Care, and One to One, which is personal tutoring for a fee. Employees often had goals for “attachments” as these add-ons are called — 40 percent of certain products should include One to One, and 65 percent should include Apple Care. For a sales employee who wanted to climb Apple’s in-store ladder — to technician or manager, for instance — those numbers were important."

[Nathan Tankus]

"ndergraduate students learn that, in a competitive market, wages are equal to the marginal value product of labour, that is, the dollar value of what the marginal worker produces. Formally, w=p*MPL where w=the wage rate, p=the price of output and MPL is the marginal product of labour.

The proof of this proposition is simple: if an additional worker produces goods and services worth, say, $12, and the wage rate is, say, $10, hiring more workers adds to the firm's profits. Thus, if the marginal value product of labour is greater than the wage rate, the firm will hire more workers. Eventually, however, as more workers are hired, their productivity will start to fall. For example, the additional workers will take on lower and lower value tasks. Once the marginal productivity has fallen to the point where w=p*MPL, the firm stops hiring.

By the same reasoning, if the wage rate is more than the value of what a worker produces, the firm will fire workers. The only possible equilibrium is one where wages are determined by productivity"

do understand that this is circular reasoning? what you're essentially saying is "assume a solely supply constrained economy, ergo worker's wages are determined by what they can supply".

It seems quite clear from the empirical evidence (see for example Blinder's book http://www.amazon.com/Asking-about-Prices-Understanding-Stickiness/dp/0871541211) that most firms behave very differently from how your positing: Namely they are generally demand constrained and subject to increasing returns to scale.

[rsj]

Once the store is fully staffed with cashiers, hiring additional cashiers means that all the cashiers are spending some of their time standing around, waiting for customers. It doesn't bring in more customers.

In that case, it wouldn't be possible for Apple to pay cashiers their MVP, even if it tried. As long as Apple does a competent job and is not understaffed, then MVP is always zero.

That is another way of saying that retail staffing, for apple, is a cost center, not a profit center.

[Nathan Tankus]

@RSJ: are you assuming there isn't any unused fixed capital? I'm not sure if you've been to an apple store, but a large part of it's appeal is the free technical support and ability to try all devices for free. Depending on the foot traffic (which in New york seems to be always packed all the time) more technical support could be provided per hour and potentially more items sold (since some people will only purchase items once they know what's wrong with what they have/ have been told what they need).

[DavidN]

@Nathan: Frances is assuming a competitive market so by definition supply is effectively infinite for each individual firm, so in fact wages are determined by demand, firm demand for labour.

Wrt increasing returns, rsj is right, retail stores are physical space constrained, you can only fit so many people in a finite amount of space. Eventually you will reach decreasing returns to scale by adding more employees taking up space for products and space for customers to roam.

Also, sure 'most firms ... are generally demand constrained' but Apple isn’t most firms. From public statements they’ve repeatedly stated they selling iPhones and iPads as soon as they produce them (hence 2-4 week waiting time for shipping). You can infer they’re supply constrained, in manufacturing at least.

[DavidN]

Supply constrained in the short run. They would be expanding capacity for obvious reasons.

[Michael]

"Ultimately, it doesn't matter how much you sell, if you can be replaced by a vending machine, you are in a weak position to bargain for a higher salary."

Economists need to be careful with statements like this. ATM machines haven't replaced bank tellers; Televisions haven't replaced the lectures of college teachers;... Deskilling of labor by machines does happen, but that is often only one side of the coin. "Re-skilling" also happens. A vending machine is fine if, ahead of the purchase, the customer knows what he or she wants. However, sales require a sales effort ("finding customers the right products") which a vending machine cannot provide.

I would think that an excess supply of labor is more significant to the workers' weak bargaining position.

[jt]

The real lesson for economics/business students:
- people will work for less if they receive some psychic benefit; the brand can work magic on the sales and cost side
- holding out unattainable future carrots in lieue of reduced salary (now) works! ...from Apple genuises to VPs jockeying for the SVP slots!

[Jim Sentance]

"Apple's policy of no commissions and moderate wages might appear to violate the "workers are paid the marginal value product of labour" rule. After all, according to the New York Times, "Divide revenue by total number of employees and you find that last year, each Apple store employee — that includes non-sales staff like technicians and people stocking shelves — brought in $473,000." If one tenth of the value of Apple store sales can be attributed to their retail employees, that would imply an average annual salary of $47,300, and a typical employee makes closer to $25,000."

I don't see anything there that precludes Apple actually paying the marginal value product of labour.

[Stephen Gordon]

Yeah, that part I didn't get, either.

[DavidN]

'For example, think about the simplest possible production technology: one worker + one shovel digging a hole. Without the shovel, the worker produces nothing. Without the worker, the shovel produces nothing. Together, they produce a hole in the ground. What is the "marginal productivity" of the worker? Of the shovel? It's like the sound of one hand clapping - it's impossible to tell what each member of the production process contributes.

...

Outside options explain why Apple can pay its workers a fairly low wage. If a typical Apple specialist or genius walked away from their job, their next best alternative might be an even worse retail position.’

The way I interpret this is you’re saying the labour market is competitive so Apple pays the market wage which might be lower than the value of marginal product of labour for Apple store employees because Apples 'capital' (e.g. branding, product design, operations) is better than other firms i.e. 'capital' in vmpl function is higher than for other firms.

[DavidN]

So why doesn’t Apple keep adding employees till vmpl (assuming diminishing margins) equals horizontal labour supply? The only reason I can think of is capacity (i.e. demand) constraints, e.g. manufacturing (i.e. supply of inventory for stores) and no. of retail stores.

It takes time to expand manufacturing capacity and the same goes for retail stores e.g. Apple stated last year they would have 25 stores in China by this time this year. They still only have 7-8.

[DavidN]

Another explanation could be vmpl is a flat finite line (finite because there are fixed no. of retail stores).

[UnlearningEcon]

I can never resist quoting Joan Robinson on the subject, who really nailed the problem with MPT:

"There is the problem of the relative levels of different types of earned income. Here we have the famous marginal productivity theory... The real wage of each type of labour is supposed to measure its marginal product to society. The salary of a professor of economics measures his contribution to society and the wage of a garbage collector measures his contribution. Of course this is a very comforting doctrine for professors of economics but I fear that once more the argument is circular. There is not any measure of marginal products except the wages themselves. In short, we have not got a theory of distribution. We have nothing to say on the subject which above all others occupies the minds of the people whom economics is supposed to enlighten."

[Nick Rowe]

Unlearning: remember. If economists could measure things like marginal products easily and accurately, we wouldn't need a market economy. We would just make economists central planners, like in the North Korea that Joan Robinson thought would do so well.

[rsj]

If economists could measure things like marginal products easily and accurately, we wouldn't need a market economy.

But then how are firms supposed to measure it?

[Nick Rowe]

rsj: short answer: with difficulty. But with less difficulty than some economist who is not working full-time in that same business trying to measure things like that. Who do you think would have the best estimate of the marginal productivity of fertiliser on my brother's farm, and how to get the best estimate? Me, or my brother?

[rsj]

If you talk the owner of a restaurant and ask him why they don't hire more waitresses or cooks, they will tell you that they have enough for the customers that they already have. Hiring another cook wont bring in a single extra customer.

The cook is paid the going wage. A friend of mine worked for a famous restaurant as a pastry cook, and was paid the same wage -- $15/hr -- as if she was working for a five and diner where you can buy a dinner for $9.

That was the going wage, but the revenue created was much higher in one case than in another.

It was really only when she got a unionized job at a hotel that she started making a lot of money -- $30-40/hr.

But note that this does not mean that the restaurant can keep hiring pastry chefs, because only so much pastry is ordered by their customers. Nor does it mean that it is earning excess profits, as the value is captured due to power relations by other members of the restaurant -- the head "celebrity" chef, the manager, the owner, etc.

[Nick Rowe]

rsj: restaurants are a good example of an imperfectly competitive firm. Farms are a good example of a perfectly competitive firm. With imperfect competition, you have to multiply P.MP by 1/(1-E) where E is elasticity of demand.

And, by the way, the (misnamed) "marginal productivity theory of distribution" does not say that MP determines wages (and other input prices). It says that wages and MP are co-determined in equilibrium by supply and demand curves, and that the MP curve is what creates the demand curve. You need the supply curve too. (And the demand curve for the product).

[Michael]

"If economists could measure things like marginal products easily and accurately, we wouldn't need a market economy."

Such an argument doesn't solve the circularity problem. Nor does the argumentum ad hominem.

[Nathan Tankus]

@Nick: that attack on Joan Robinson was very cheap and had nothing to do with the point.

I think your assertion that a farm is a perfectly competitive firm is spurious. I'd like to see some more evidence on that.

[Nick Rowe]

Michael: The Black Death would be one example. The effect on wages and land rents of an exogenous fall in labour/land ratios is what the MP theory would predict.

Nathan: next time I see him, I will ask my brother how big an impact he thinks a doubling his wheat output would have on the market price of wheat. But I suspect he would be surprised by the question.

[rsj]

Nick,

What I was trying to point out, is that in the absence of a walrassian auctioneer, you have no mechanism to that ensures that the prices and quantities that are actually obtained are those that would lie on the intersection of the notional curves.

Lacking the auctioneer, you need to appeal to some convergence result, that says the results we see are "close" to the ideal results in some manner.

I'm not aware of any such convergence result. In which case, why would anyone expect Apple to pay the ideal wage? Luck?

[rsj]

next time I see him, I will ask my brother how big an impact he thinks a doubling his wheat output would have on the market price of wheat. But I suspect he would be surprised by the question.

I would ask him whether he sells his goods for the marginal cost of production. I think he would be surprised at the question. And also ask him whether his average cost is declining or increasing with quantity sold.

[Determinant]

Frances, you do know I'm an expert on this subject? I used to work for an (outsourced) call centre staffing Apple's toll-free support line. I was a Tier 1 agent (first contact). "Product Specialists" and the Apple Store Geniuses are equivalent at Tier 2.

I will admit my wage was $25,000 and I worked shifts. I did have a health plan. This was a Canadian call centre, but there were more than a few of those. I was there for a year and a half. My happiest day was when I said "I quit!" I did it at 9AM when I got there.

Apple contracts out most of its support line staffing to third-party BPO companies. It does have two call centres which are staffed by Apple employees in Austin and Sacramento. Internal employees are paid twice what contractors make. All centres are measured against each other for call statistics. It's a stacked rat race. The contractors have to be better than the internal centres and the internal centres have to keep up to keep from being outsourced. Apple manages its operations to ensure that management has a favourable position. They are a big, big outsourcer.

BTW The Apple Store in Toronto would not hire Tier 2 techs from my call centre, even though with their training they could go straight to the store floor. Nope, they systematically disbarred that migration because it was a drain on their outsourced operations.

Apple is a excellent case study in the exercise of company economic power.

[Frances Woolley]

Currently have very limited internet access, will respond to comments in a couple of days, sorry for the delay, Frances

[Nick Rowe]

rsj: "I would ask him whether he sells his goods for the marginal cost of production."

That's the wrong way to ask the question. It should be asked the other way around. Does he adjust his inputs and output so that the marginal cost of production equals the price at which he sells his goods?

And from my conversations with my family and other farmers over the years, I have a rough sense of the answer. They are often thinking out loud: if I buy/rent an extra x, and the extra output is y, and the price of wheat is P, will x pay for itself? Now it's true, x is often a bundle of inputs, rather than one input alone. But the MP theory can easily be modified to handle the cases of joint inputs, or of envelope theorems.

[rsj]

I don't know the world of the farmer. All my life, I have never worked for a firm whose marginal costs of production was anywhere the price of the good. The lowest gross margin for any firm I've worked for was 60%, and in all cases, the marginal cost of production was either zero (software) or very small (hardware, but with the software being the primary component of value added). The single largest expense was always research and development, not operations or production.

The farmer is an example of controlling quantity, but not the price. Hence the stickler about "and the price of wheat is P".

If you could predict the future price of wheat over the lifetime of a typical capital good, you would be a very rich man.

So even in the case of the approximation to the perfectly competitive firm, your brother, when buying a capital good, will impose a large discount before pulling the trigger. He will require that the capital good does much more than just pay for itself at the going lending rate, because there is a risk that the price of wheat will fall and he will not be able to repay the loan.

The bank will impose one risk premium when lending to him, and he will impose an additional risk premium when agreeing to the loan, and as a result, ex-post, the marginal price will never be the cost of production. Only in the case of a cataclysmic failure of prices will the MP be at or below the cost of production (taking into account capital rental costs).

The other extreme example is the restaurant, in which you control your selling price, but do not quantity sold.

In that case, the restaurant wont hire an additional cook because doing so wont bring in more customers. MVP falls off a cliff. The only way to bring in more customers is to lower the selling price. Hiring an additional cook will not bring more diners to your restaurant, and the cook will have nothing to do but stand around. MVP is zero for the extra cook.

So both the farm and the restaurant will charge a price above marginal cost ex-post in all states of the world that are reasonably expected to be obtained. In one case because of risk and in the other case because MVP falls off a cliff.

And all firms between these two extremes will also charge a price higher than MC, because the typical firm will add both a risk premium and it faces the MVP cliff problem.

[TACJ]

"@Nick: that attack on Joan Robinson was very cheap and had nothing to do with the point."

+1

Also:

http://crookedtimber.org/2004/09/29/i-didnt-go-to-evil-medical-school-for-seven-years-to-be-called-mr-evil/

Also:

Don't farming conglomerates make profits? And if so, doesn't that mean farming isn't a perfectly competitive industry?

[Nick Rowe]

rsj: "The other extreme example is the restaurant, in which you control your selling price, but do not [control NR] quantity sold."

Here's another way of looking at it: restaurants face a downward-sloping demand curve. We can think of restaurants as choosing a point on that demand curve. We can think of restaurants as choosing either price or quantity demanded, but not both. In the limit, as the demand curve becomes perfectly elastic, we approach perfect competition, where the firm chooses quantity.

If there's uncertainty, and the firm is risk-averse, there's a risk premium. Do we add that to include it in MC? Or subtract it from MR? Does it matter which we do?

[Nick Rowe]

TACJ: "Don't farming conglomerates make profits? And if so, doesn't that mean farming isn't a perfectly competitive industry?"

There are two main assumptions for the model of perfect competition:

The first is the assumption that individual firms face horizontal (perfectly elastic) demand and supply curves, and so take prices as given. That is the assumption relevant to the marginal product theory.

The second assumption is free entry and exit. That is the assumption that is necessary for zero profit long run equilibrium. It is also necessary for zero profit in monopolistically competitive firms. That assumption is not needed for marginal productivity theory. So it's not relevant here.

C'mon guys. This is micro 1000. And a century old. Textbook stuff. I spend the rest of my life teaching this stuff. I don't want to repeat it all here. I've got other stuff I want to do.

[BSF]

Frances: The fact that the Average Revenue Product of Labour is $473,000 doesn't mean that the Marginal Revenue Product of Labour can't be $25000. The MRPL depends on how much additional revenue a firm will earn if it hires one more worker, without having to change the quantity of any other inputs it uses. In the case of retail that could well depend in part on customers' value of time: if customers are going to be quick to leave a store where they aren't being served, the MRPL of another worker would include the revenue that doesn't walk out the door. If customers are very loyal to a brand, and willing to stand around a relatively long time to buy one of that outfit's products, the part of a retail worker's MRPL which comes from keeping them in the store might actually be small, making the MRPL smaller than it would otherwise be. Plus, of course, you might have a lot of kids who really want to work for Apple, at least for a while.

[Determinant]

You don't have to either. Trust me, I was in the system, Apple makes sure that Apple, not labour, has the market power. Apple freely and frequently uses its market power to its own advantage.

It's not a perfectly competitive system and its not about supply/demand for labour because the assumption that the two parties are equal and contracting freely is false.

I think this moves us into second-year micro.

[rsj]

Here's another way of looking at it: restaurants face a downward-sloping demand curve. We can think of restaurants as choosing a point on that demand curve.

I don't think that's the reasoning. Or at least, you are describing a long run equilibrium, but the words "choosing" suggest that there is a degree of freedom in the short run that is not there. This is a conversation I had with a coffee shop owner (a real conversation, not one of those Thomas Friedman conversations :P)

"How's business?"
"It's too quiet"
"Have you ever thought of lowering your price?"
"No, you never lower the price. You only keep it steady or raise it"
"But you don't think you would get more customers if you lowered your price?"
"No, I think it's about how many are walking the streets. You really think they will say "Coffee at Hank's is 10 cents cheaper so let's go out today?""
"So you don't think you sell would any more coffee at all if you lowered the price by 10 cents? You would sell *exactly the same amount*?"
"Maybe after a really long time, one or two new people would come in, but all the existing customers who are already here would pay 10 cents less, so it wouldn't be worth it".
"By that logic, why don't you raise the price by 10 cents?"
"Because I might lose some customers"

And here is another conversation I had, with a convenience store owner

"How do you decide how much to charge?"
"I double the cost. Sometimes I use MSRP"
"When do you double the cost and when do you use MSRP?"
"For cigarettes and wine, I use MSRP. And for beer. For other things I double"
"And how has business been?"
"It's been slow"
"Do you ever think of lowering prices? To get more customers?"
"No, you can't lower prices"
"Why not?"
"Because you make less money"
"How can you be sure? Maybe you would make more"
"Well, I've been doing this for 30 years, and the business really depends on how many people walk into the store. It's not about the prices. All the stores in the area do the same thing"

And here is another conversation I had with the same store owner, (a wonderful man, German immigrant) who was telling a customer that they had no brand of a certain beer left.
"It's very popular. They always run out"
I interrupted "So why don't you raise the price?"
At this point, both the customer and Chuck (the owner) looked at me in a shocked way
"Well, I guess they could."
"No, why don't you raise the price?"
"Well, it's up to the maker to raise the price"
"OK, why don't they?"
There was a moment of silence as everyone thought of it. The customer came up with an idea, to which the owner agreed:
"Well, what if everyone did that?"
(Chuck) "That's right. What if they all started raising prices?"
(Me) "If they did, then they would sell less, no? Then they would have to lower prices again"
(Chuck) "How do you know they would? But its a good question."

[rsj]

BSF,
The MRPL depends on how much additional revenue a firm will earn if it hires one more worker, without having to change the quantity of any other inputs it uses.

Right, so a large firm such as Apple may have well over 100,000 inputs. And it is supposed to hold 99,999 fixed, and change one. Then repeat.

Of course, it needs to do this more than 100,000 times, it needs to do it 100,000! times. Or roughly 10^10^6 times. After each time, it is supposed to wait and see how much revenue it makes, right? So if it waits for One Christmas shopping season, then the universe will experience heat death long before Apple determines a single price. That assumes there is no problem with determining whether the revenue increase is due to hiring a particularly good worker, or bad worker, or just randomness.

Moreover, as it is doing this price engineering, supposedly it is not rolling out new products, business conditions are not changing, and it is not changing its production and distribution techniques.

[Nick Rowe]

rsj: your first conversation. Interesting. I built a macro model with a demand curve like that once. To explain sticky prices. There's a dynamic kink in the demand curve at the existing price. Because all existing customers hear about a price rise, and some leave, but it takes a long time for potential new customers to hear about a price cut. So it's more elastic on the upside than on the downside. Which means a discontinuity in the MR curve. There's a (small) literature on this. Can't remember where to find it now. Can't even remember who started it. Okun? Stiglitz?

gated link here

Now I'm going to stop hijacking Frances' post, because we have all wandered waaaay off-topic. And Frances is unable to bring us into line, being off the internet.

[Frances Woolley]

Jim and Stephen - ""I don't see anything there that precludes Apple actually paying the marginal value product of labour."

True, which is why I said "might appear to violate" rather than "violates" - it's entirely possible that retail employees produce little of value.

Nick: "Now I'm going to stop hijacking Frances' post"

I'm glad you did.

[Determinant]

Jim and Stephen - ""I don't see anything there that precludes Apple actually paying the marginal value product of labour."

True, which is why I said "might appear to violate" rather than "violates" - it's entirely possible that retail employees produce little of value.

Possible, but there is a better possibility. Apple is aware of its own market power and position. It knows that if it uses its economic power, it can demand lower wages from its employees and higher profits for itself. Given that incentive, they take it. They consciously take actions to drive down wages to contain costs. It's not about vending machines, it's about winning a game of chicken.

Or how about their contacting arrangements with phone staff to keep call centres bidding against one another?

ISTM that your post Frances really looks at a firm as a distribution mechanism for wages. But Apple doesn't see itself that way; it sees wages as a pure cost, one to be contained at all costs. I've been there and seen it in action.

[Frances Woolley]

Michael: "I would think that an excess supply of labor is more significant to the workers' weak bargaining position."

An excess supply of labour allows Apple to walk away from any one individual worker; vending machines allow Apple to walk away from entire categories of employees. Both are important.

DavidN: "So why doesn’t Apple keep adding employees till vmpl (assuming diminishing margins) equals horizontal labour supply?"

You're assuming the possibility of calculating a vmpl. Suppose there's a Leontief production technology, e.g. one worker needs exactly one shovel, and additional workers or shovels are useless. The production function isn't nicely differentiable, so vmpl isn't defined. As others have noted, the Apple store is somewhat like this - there are limited number of workers that can be hired per store, otherwise the store would get too overcrowded. Why doesn't Apple just duplicate its existing operations? It can, but that doesn't solve the problem of non-differentiable production functions.

When production functions are differentiable, e.g. for Nick's farming example, when a farmer can work out the increase in expected yield by seeding a field more densely, and output/input markets are competitive, the marginal product theory makes more sense.

Determinant - interesting story about working at Apple.

[Frances Woolley]

As always, people's comments have forced me to sharpen my thinking and clarify the logic underlying my argument. Please see the update above where I expand upon a point that's crucial to the argument that I'm making here: with certain types of production technologies, the marginal product of labour may be undefined.

[DavidN]

Yes, I was one of those people who pointed out there’s a constraint on how many employees you can have in one store.

I think there is an argument to be made that VMPL is low relative to VMPK (where K stands for 'capital' which in this case includes products, store itself, management practices, marketing etc) for Apple retail stores. If look at this post by Horace Dediu: http://www.asymco.com/2012/04/18/apple-stores-have-seventeen-times-better-performance-than-the-average-retailer/

Apple generates the most revenue per square foot of US retailers and beats the next best (Tiffany & Co) by a fair margin. I doubt you can put all that down to superior productivity of Apple store employees.

It could be that VMPL is a lot higher for Apple store employees than other retailers but I think you can put a lot of that down to the products, management practices, and marketing in addition to the quality of employees. As a thought experiment suppose you swapped the Macs for PCs, iPods for Zunes (if they still exist), and iPhones and iPads with their Android equivalent, would the Apple store with Apple employees generate the same amount of revenue? I doubt it.

If you swapped an Apple retail employee with a generic retailer, I bet VMPL for Apple will fall and the VMPL for the generic rise but VMPL Apple store still > VMPL generic store.

[Ryan]

Prof. Woolley - Good post, and I would add that there is probably some intangible compensation going on at Apple, i.e. there is intrinsic value in working for Apple above and beyond the monetary compensation + benefits.

[Steve Roth]

What Ryan said.

What this analysis ignores -- as do the limited quantities of others that I've read -- is the value of job satisfaction, and its effect on the elasticity of labor supply.

You could think of Apple as providing "satisfying work," which workers pay for with their labor.

Or you could imagine the outside option of being a rep for a medical device manufacturer at a higher wage, from which one must deduct the "cost" of lower job satisfaction/utility for the worker.

I pondered this briefly in two posts so will just ref them, and more importantly recommend the paper by Ian Steedman which prompted the posts.

http://www.asymptosis.com/job-satisfaction-and-elasticity-of-labor-supply.html

Considered this way, it seems like you arrive at an analysis is which there are (at least) two complexly interacting S/D diagrams, at which point the S/D model becomes very problematic, maybe intractable, as a tool for thought experiments, and perhaps for more rigorous modeling as well.

[Determinant]

What this analysis ignores -- as do the limited quantities of others that I've read -- is the value of job satisfaction, and its effect on the elasticity of labor supply.

You could think of Apple as providing "satisfying work," which workers pay for with their labor.

Apple doesn't really do "satisfying work". Heck, when Apple famously gave all its employees a free iPhone, the offer didn't include BPO contractors on the support line (several thousand of us). I had to lie to customers that day about possessing an iPhone.

What Apple does do is structure competitive systems in favour of itself. Apple actually has several retail channels, of which the Apple Store is only one. There is also apple.com, a toll-free sales line and Apple-Authorized Resellers. The Apple Store is not the only sales channel, by design. So if Apple Store employee pay gets too high, Apple can always increase reseller sales or online orders. Apple has several sales channels all of which are in a permanent race to the bottom on terms of costs, to Apple's profit.

Sometimes the use of economic power is apparent, real and very pertinent in the analysis. Apple is one of those cases. Supply/Demand doesn't cut it.

[Kate Jackson]

This is why people need to at least read Marx's labor theory of value in beginning econ. Though I'm no expert, I understand him to say that wages will eventually drop to the minimum needed to sustain the worker -- food, clothes, lodging -- and the rest, the "surplus" value, goes to the employer, er, capitalist. This is because of two forces (1) gains in technology, which requires less from input from the worker -- whether it's division of labor or the advent of robots and vending machines; and (2) increasing number of population subject to capital markets. So, certainly it holds true that a company will not hire a worker if the pay is greater than the marginal productivity of labor. But that is only half the story -- and not the half that we usually see become relevant.

It's another one of these demand vs supply side mix ups. There are always going to be more people than there are jobs. With the marginal productivity of labor theory, we assume both are equal -- it being a perfectly competitive market. Which isn't true in real life, of course!

[Ryan]

Seriously, Kate? Labor theory of value?

[Frances Woolley]

Kate: "There are always going to be more people than there are jobs."

I don't know if I agree with that. People always find something to do - the question is whether or not their work pays them a living wage.

[Bob Smith]

DavidN: "Apple generates the most revenue per square foot of US retailers and beats the next best (Tiffany & Co) by a fair margin. I doubt you can put all that down to superior productivity of Apple store employees."

No, indeed one fairly plausible explanation is that Apple (not unlike Tiffany & Co.) tends to set up relatively small shops in "premium" locations (in Toronto, for example, think the Eaton Center, Sherway Garden, Fairview Mall and the Yorkdale Mall). Compare that with the practices of other retailers of Ipads and Iphone (say, Best Buy or Walmart) who establish stores in less "premium" locations. No doubt that different strategies result in higher real estate costs (per square foot certainly, and possibly overall) for Apple, which are, in part, reflected in smaller store sizes, and in part in higher revenue for square foot (those locations are "premium" for a reason - namely they attract shoppers willing to spend a lot of money - let's face it, there are certain segment's of Toronto's population who would kill themselves before they wandered into the,say, Dufferin Mall Walmart to buy an Ipad). Of course, in that example, the higher revenue per square foot owes nothing to the higher productivity of Apple workers and everything to the old axiom of real estate, "location, location, location".

[Determinant]

There is the mistaken assumption here that Apple employees are the exclusive retail agents of Apple. They are not. Apple-Authorized Resellers are independent third-parties. Further, I did not work for Apple. I worked for a Canadian company which was a Business-Process Outsourcing company, a call-centre manager and contractor. I had an employment relationship with them, they had a service contract with Apple.

Regardless of the fact that my labour was in furtherance of Apple's business and to Apple's profit, the fact that I sold Apple products (repeatedly), took Apple training and followed Apple policies, I was not, under any circumstances, an employee of Apple Inc. or Apple Canada Ltd. Apple does not follow the manufacturer model of a large number internal employees producing value-added work. Much of Apple's work is contracted out. Apple is a designer and marketer.

[DavidN]

Determinant,

The figures quoted in the NYT and the link I provided are for Apple retail stores, does not include other distribution channels.

[DavidN]

Bob,

You might be surprised some Apple stores are huge: http://www.google.com.au/search?q=biggest+apple+store&hl=en&safe=off&client=safari&rls=en&prmd=imvnsu&tbm=isch&tbo=u&source=univ&sa=X&ei=rLYYUKqqIsiZiQfZ3oHYBQ&ved=0CHAQsAQ&biw=1254&bih=1284

Not as big as department stores but huge for a store that carries only a handful of different products.

[JakeS]

Apple pays its retail employees shit wages because Apple prefers to operate in jurisdictions where strikes are broken with tear gas and rubber bullets... and produce its goods in jurisdictions where strikes are broken with machine guns.

Marginal productivity my arse.

- Jake

[BSF]

Frances: From the August 1 Wall Street Journal,

Article titled "High Marx For China's Wages", by Tom Orlik:
-------------------------------------------------------------------
"There's no problem with China's labor productivity. The World Bank estimates that output per worker has grown more than 8% a year since the mid 1990s. But a glut of workers means wages have stayed low. Most of the benefits of rising productivity accrued to owners and investors in the form of higher profits.

"Now, though, a demographic shift means a change from excess supply of workers to excess demand in China's labor markets. Data for the second quarter show that despite a sharp downturn in growth, there are still more vacancies than there are workers to fill them.

"The impact of that shift on the distribution of income is already starting to appear. In the first half of 2012, average wages rose 13.1% on-year; profits were down 2.2%."
---------------------------------------------------------------------

[Peter]

Wonderful stuff - and I don't want to get off topic - but Nick, "the price of wheat" - in Canada? Isn't the price fixed? (Agreed, in a perfectly competitive world the producer is a price-taker so from the quantity setting perspective it shouldn't make a difference, but if the price is fixed of course he isn't going to think about how his output affects the price. He should be thinking about how his output effects his revenue less costs.)

@rsj - I am still convinced that most competitive industries including convenience stores, painters, etc. are not profit maximizers. Certainly the ones I talk to don't consider the cost of their staff vs their staff's contribution. If you can pay $40k or $25k to fill the position, they will choose the $25k regardless of the benefit (and where I work, the business is losing far more than $15k by not having truly effective staff).

[Determinant]

DavidN:

I wasn't trying to rebut your arguments, just trying to illustrate that market power and the use thereof is not a negligible concern when analyzing how Apple behaves. I was in the system and so I am trying to pass along my experience and knowledge of that system to the thread.

My point was that Apple is not like General Motors or even a grocery store. Frances' was tending toward that idea, and I wanted to draw her, and the rest of the thread away from it.

[pgl]

Two additions. In a way what you are saying is that marginal is not the same as average. Yes, Apple's value-added per employee may be high but that's an average concept. The other addition - wages for a lot of Silicon Valley companies are a modest component of total compensation. Do these employees receive stock based compensation such as stock options?

[Determinant]

pgl:

No. That's reserved for executives. If you can be outsourced (see my previous posts), you don't get stock options, as a rule.

[Bob Smith]

Jake: "Apple pays its retail employees shit wages because Apple prefers to operate in jurisdictions where strikes are broken with tear gas and rubber bullets"
Jurisdictions like Toronto, New York, Montreal? Really? Its manufacturing workers work in locations where strikes are broken with real bullets and visits by the local "security" service in the middle of the night (notably a certain commie dictatorship that shall remained unamed), but that doesn't really tell us much about its retail workers in, say, North America (which was the subject of the OP).

PGL, I think the average vs. marginal distinction is crucial. Frances' initial observation was the Apple's average revenue per worker at each apple store was something like 20 times their average wage ($470,000 vs. $25,000) and she wondered if this didn't raise questions about whether apple paid its workers the marginal product of labour. But if you step back and think about the cost structure facing the apple store, that result makes sense. Indeed, one would be shocked if the average revenue per labour wasn't 20 times the average wage per worker.

Think about, let's say the Apple store sells, on average 100 ipads per worker at a retail price of $500 (this isn't the reality, but a simplication to illustrate the point. Gee, the average revenue per worker is 20 times the average wage ($500,000 vs. $25,000). Ok, but Apple Stores have to pay for their inventory (I don't know if Apple Stores are organized as separate legal entities from Apple, but even if they aren't, they would be charged the wholesale price internally). Let's say they're retail price reflects a 100% mark-up (probably not unrealistic), so that their average cost per worker for products is $250,000. So apple workers have to bring in revenue that's 11 times their average wage just to break-even, ignoring all other costs. Toss in overhead, rent, cost of capital, allowances for theft or breakage, training, yada, yada, yada, it isn't at all unreasonable to think that Apple just breaks even on the last worker hired.

DavidN, I'll defer to you on the point (what can I say, I'm a blackberry man). The main point, which I think isn't disputable, is that the Apple Store does set itself up in premium locations.

[Bob Smith]

Also, I wonder about Frances' title "the marginal productivity of geniuses: why Apple's workers earn modest wages"? Are Apple Store workers "geniuses"? No doubt the technicians have a skill set that appears wonderous to a certain generation (i.e., people who weren't born and raise immersed in electronic gadgets and computers - and I hate to admit it, but I'm probably one of those people. Did I mention I was a blackberry man?) and, in that sense they're "geniuses".

On the other hand, my (now) unimpressive ability to resolve (minor) computer problems was once upon a time a source of wonder and pride to my parents - I still get calls from my father to help him back-up documents on word (my parents also think the Apple Store workers are gods). To them (and their generation), I was a genius with computers. Yet, by any objective measure, I wasn't a genius, I had some very rudimentary skills that just happened to be beyond their ken.

Also, on a last point, the $25,000 figure is a bit misleading, since the article also mentions that Apple has relatively generous benefits including health benefits (not an irrelevant consideration in the US).

[iPhoned]

As much as I'd LOVE to see commissions or $20/hr starting wages at Apple Stores, in comparison to other retail, the wages are pretty good and the bennies are spectacular, but people are sworn to secrecy about that so...