We’ve had a very long recent thread about the social costs and benefits of high frequency trading, where I’ve apparently managed to confuse a number of readers by switching back and forth, according to the convenience of the moment, between two different, but perfectly legitimate, social accounting systems.
To clarify matters, let’s forget for the moment about high frequency trading and look at something simpler — innovation in the IT industry, where it’s clear that profit-maximization can easily lead to too much innovation. I’ll do the accounting both ways to make it clear that both ways are right.
First, the assumptions:
Alice has developed a word processor, which she sells online. It costs her $5000 a year to maintain a server, where you can download a copy for $1000. She sells 100 copies a year, and therefore collects $100,000 in revenue. Most of the consuemrs who buy those copies value them at more than their price. In fact, the total value of those 100 copies to the consumers is $200,000.
Bob has an idea for a word processor that’s a little better than Alice’s, so that each consumer would be willing to pay $10 more for Bob’s than for Alice’s.
If Bob develops his word processor, how much can Alice charge for hers? Because her word processor is inferior to Bob’s, she’s got to undercut his price by $10 in order to maintain any customers at all. So if Bob charges $600, Alice charges $590. But then Bob can steal all of Alice’s customers by lowering his price to $599.99, whereupon Alice must lower her price to $589.99, whereupon Bob steals all her customers by lowering his price another penny….and the race to the bottom is on. But Alice’s price cannot fall below $50, because then she wouldn’t earn enough to cover her server costs. So Alice, who is smart enough to foresee all this, gives up and cedes the market to Bob.
Once Bob has the market to himself, he doesn’t have to worry about re-entry by Alice, because they both know perfectly well that the instant she renews her server contract, the race to the bottom will be back on and she’ll have spent $5000 for nothing.
Now if Bob sells his word processor for $1000, it’s he instead of Alice who earns $100,000 a year in revenue and therefore (after subtracting the server cost) $95,000 in profit. He weighs this against the $80,000 cost of developing his word processor and takes the plunge.
I claim that Bob’s decision is privately wise (i.e. wise from Bob’s point of view) and socially foolish (i.e. it reduces social welfare, defined as the total dollar value of all the gains to consumers and producers). We can calculate the costs and benefits of Bob’s decision in either of two equally legitimate ways. Because they are equally legitimate, they lead to the same bottom line: Bob’s private benefit exceeds his private cost by $15,000 (which is why he plunges ahead), while the social cost exceeds the social benefit by $79,000 (which is why we wish he wouldn’t).
Method I:
Private Cost: $80,000 (Bob’s cost of developing the software)
Social Cost: The same $80,000.
Private Benefit: $95,000 (Bob’s profit from developing the software)
Social Benefit: $1000 [The gain to 100 consumers, each of whom now has a word processor worth $10 more]
Method II:
Private Cost: $80,000 (Bob’s cost of developing the software)
Social Cost: $175,000 (Bob’s development cost plus Alice’s loss of $95,000 in profit)
Private Benefit: $95,000 (Bob’s profit)
Social Benefit: $96,000 (Bob’s profit plus the $1000 gain to consumers)
Bottom Line:
By either accounting method, the external cost (i.e. social cost minus private cost) exceeds the external benefit (i.e. the social benefit minus the private benefit) by $94,000. Therefore Bob makes the socially wasteful decision to develop his software.
Either way, Alice’s loss of $95,000 in profit is offset by Bob’s gain of $95,000 in profit. In Method I, we ignore this transfer since it washes out anyway. In Method II, we count it in the social caclulations, both as a loss to Alice and a gain to Bob. Again, both ways are correct.
Important Addition:
In this example, Bob faces a yes/no choice about whether to develop his software. In many examples, Bob makes a continuous choice about how much of an activity to engage in. In those cases, we expect that Bob engages in the activity until at the margin his private benefit is equal to his private cost. That considerably simplifies the calculation, by allowing us to ignore the private costs and benefits completely and focus entirely on the external costs and benefits (where “external” means “social minus private”).
One More Thing:
Bob’s innovation might well have social benefits we haven’t accounted for, primarily the inspiration it gives to other innovators who might face different numbers and whose innovations might be socially valuable. Those benefits aren’t accounted for here. Therefore, in IT, there might either be too much innovation or too little — but it would require an extraordinary coincidence for there to be just the right amount.
I have a couple of comments
1: This may be in violation of one of your assumptions but if the sales potential is big enough (ie if we take multiple years of sales rather than just one) then won’t the social benefits of Bob’s better software eventually outweigh the social costs of the resources used up by additional development costs ?
2. The model of one-off investment followed by low marginal costs sales would seem to be rare in the real world. A more likely scenario in the software world would be something like: Alice produces a product that is popular and can be sold for high margins. Seeing the success of Alice’s product competitors will produce similar products (some better some worse) that will cause Alice to both reduce price and improve quality. If she’s successful at this she will maintain good long term margins, if not she will go out of business.
If too much capital gets drawn into the software industry as a result of Alice’s initial profits so that average profits fall or go negative then this will be socially wasteful – but this would seem to be a result of bad entrepreneurship rather than an inherent problem with social accounting. And these losses would serve as a strong market signal for some dis-investment in this industry.
Rob Rawlings:
(ie if we take multiple years of sales rather than just one) then won’t the social benefits of Bob’s better software eventually outweigh the social costs of the resources used up by additional development costs ?
The simplest example assumes the market disappears after one year. If you want to make it a multi-year market, then we need to express all profits, costs, etc. in present-value terms, but the basic idea goes through.
If too much capital gets drawn into the software industry as a result of Alice’s initial profits so that average profits fall or go negative then this will be socially wasteful – but this would seem to be a result of bad entrepreneurship rather than an inherent problem with social accounting. And these losses would serve as a strong market signal for some dis-investment in this industry.
The result definitely depends on the assumptions, but I think you’ll have a hard time finding assumptions that imply we get exactly the right amount of investmnet.
May not be relevant, but couldn’t Alice offer to pay Bob $15,000.01 and then give him her company in one year? Alice will get 79,999.99 in profit for one more year and Bob will have no motivation to “waste” $80,000 of his money developing software.
Thomas Purycki: But then Bob’s cousins Carl, Dougie, Ethelbert, Frederika and Geordie all have an incentive to spend time thinking up ideas for software so they can also blackmail Alice (or her cousins who market various other software packages), and that time spent thinking becomes pure social waste.
Thinking is hard though, so if Carl, Dougie, etc are credibly capable of coming up with competing software, they would prefer to go right to the blackmailing and skip the wasteful thinking.
Also, wouldn’t the social waste in your scenario really hinge on what Bob’s next best use for $80,000 is? If prevented from developing software, Bob would choose to donate the $80,000 to the poor, the world would arguably be richer. If instead he decided to use the $80,000 to outfit his car with “The Club” anti-theft device, the world would be poorer than if he developed the software and added $1000 of value for his 100 customers.
I agree that this looks like a real issue. Is this way of looking at it correct ?
Suppose that there is a choice between investing in a new version of ProductA (and eliminating the incumbent) or in ProductB which is brand new.
Depending upon the choice made you either have a) Just an improved ProductA b) old productA and brand new productB
If you make the investment decision based purely on social utility then you will only choose a) only if the present value of the future stream of social utility exceeds that for b).
If you make the investment decision based purely on profitability then you will choose a) if the profits from investing in ProductA are greater than those from investing in ProductB even if the stream of social utility from the investment in the new ProductA is less than the combined social utility of the old ProductA and ProductB.
There are probably not many real-world “winner take all” industries where this would apply. I wonder if anyone will come up with a market-based solution to this scenario (still trying to myself).
Just realized, to the extent that the opportunity to blackmail with a mediocre idea entices Bob’s cousins to stop playing Titanfall and start thinking of ideas, isn’t this a good thing? Some of those cousins are bound to accidentally come up with ideas that are actually good in the sense of costing less to develop than the social benefits they provide. This might even be enough reason to make peace with the existence of “socially wasteful” investment, though I have no idea how to even begin measuring the social gains from brilliant innovations that would not have occurred without the opportunity to make winner take all profits from a mediocre innovation versus the cost of wasteful investment.
I suppose that one thing that will restrict entry into “winner-takes-all” industries is that fear of others entering (and wiping out your investment). This may restrict profit expectations and make other investments (that add more social value) seem more attractive in comparison.
This example seems bizarrely contrived compared to how actual competition in software products works and I can’t help think the differences are relevant to the comparison being made. You are missing a LOT of the social benefit to new development. In real life, each new program makes the software ecosystem more vibrant and robust, benefitting almost everyone.
Here’s how it works: Alice has a “first mover” advantage – because she was in the market first, everybody knows how to use her stuff. The name of her program shows up in job listings and resumes; there are books/websites/training programs for it and legacy documents created with it. Alice has a huge head-start on “mindshare” and development tools and the advantage of years of customer feedback helping her know what needs to be added or fixed.
For Bob to *take over* that market, his stuff would have to be *substantially* better or cheaper or both, which is very hard to do.
But the good news for Bob is that modern software is very complex, with multiple features and aspects. So to carve out a profitable market niche Bob doesn’t have to better than Alice at *everything*, he just has to be much better in one or two areas, while being worse or roughly on par in others. So let’s say Alice’s program is good for writers in general but lacks some special features that would be most useful for *screenwriters*. So Bob specializes in that. He adds substantial value – let’s say he doubles the consumer value – for the 20% of the market that switches to his new product, while 80% of her old customers stick with Alice. He also grows the market, bringing in additional customers that hadn’t even bought Alice’s program (because they are screenwriters and the program wasn’t worth getting without his enhancements).
Now Alice *could* try to copy his new features to get those customers back, but she also has the option to cede that market to him, work *less* on that feature area (now that the customers who care have moved on nobody’s complaining about it now!) and instead specialize on NON-screenwriters, simplifying and adding other features that grow HER market from new customers who weren’t ever interested in Bob’s stuff. The competition makes both products better over time at serving their own specialized niches. Alice had been resting on her laurels a bit but worrying about losing *more* of her market to Bob forces her to invest in keeping her remaining customers happy.
Software is not a “perfect competition” kind of good; Bob’s program is only *somewhat* substitutable for Alice’s. There’s overlap, but for Alice to entirely lose her market just isn’t going to happen unless the new product is SO much better a value that she SHOULD lose it. Because the goods aren’t perfect substitutes, they don’t have to compete on price. (they might, but they don’t have to).
The transferred customers would be a wash – Bob’s gain is Alice’s loss – except that these customers all are much happier with their software. But the *new* customers are all social gain as is the fact that many of the old customers like their software better post-competition. Since it’s not a winner-take-all market, Bob’s not spending enough on development to take the whole market, he’s only spending enough to capture his fraction of it, the ones who gain the most from his entry. Every new market entrant makes it more likely that whatever feature set you want as a customer, there’s probably a program out there for you. In general, for something like a word processor more market entrants is a very good thing for customers – at least until the market is *very* mature and no longer a source of enticingly large profit margins.
Steve: question about this scenario. What’s wrong with the argument that, the moment Bob tries to charge more than $60, Alice will jump back into the market and take all the customers again. Bob, knowing this, realizes he cannot charge more than $60 for the product, which isn’t enough to cover his cost of development, so he never invests in his new word processing software?
This new example seems fundamentally different from your previous argument. With the fiber example, your argument was that Alice’s product essentially becomes worthless once a slightly better product is available — the consumers aren’t willing to buy it if its offered for $10 less, they wouldn’t take it unless it were free.
Going back to the HFT example for a second, I have a question about the long-run situation.
Suppose we draw a circle around the HFT traders and their communications technology. Outside this circle are “investors”, and they pay costs to the market-makers inside the circle. The investors don’t care very much about extremely fast execution. Inside the circle, the market-makers and fiber companies are in a winner-takes-all battle to be the fastest.
In the long run, the total private benefits earned by everyone inside the circle must equal the total private costs incurred. The total private costs equal the total social costs (using Method I here). The total private benefits equal the revenue paid by the investors to the circle. This revenue must reflect the overall benefits to the outside world. Therefore the net social value of this circle cannot be negative — it may be lower than what it would have been if everyone agreed to set up a monopoly market-maker and cable company, but it shouldn’t be negative in the long run.
Bob deserved to take all of Alice’s customers, because Bob found a way to sell his software without needing to rent an expensive server.
“I claim that Bob’s decision is privately wise (i.e. wise from Bob’s point of view) and socially foolish (i.e. it reduces social welfare, defined as the total dollar value of all the gains to consumers and producers).”
How do you known it was Bob’s decision rather than Alice’s decision that reduced social welfare?
For that matter, how do you know that either of them reduced social welfare? Their actions only reduce welfare if there is some other arrangement such that after accounting for the costs of adopting that other arrangement, social welfare would be higher. Otherwise, the realized state of affairs is optimal.
Alice brought her word processor to market without the knowledge that Bob would create a better product. If there was a consultant who, for less than $95K, could have reliably predicted Bob’s product when Alice was initially deciding whether or not to begin work on her word processor, then Alice’s decision to create her word processor rather than hire the consultant was bad for social welfare. Now in the real world such a consultant is not plausible because limited knowledge is a binding constraint. However, once that constraint is taken into account, it is no longer obvious to say that their choices lead to a less optimal outcome.
Re: financial markets, why is HFT even worth talking about if we’re worried about wasted resources? The money spent by HFT firms on machinery is probably very small relative to the money that traditional active managers spend on data and that’s been going on for decades.
Doesn’t Alice have leverage? She can stay in the market down to $50. Her loss is capped at $5k, but Bob’s can be much higher. (Neither Carol nor Ted is in this position right now, so Bob does not necessarily create an incentive for blackmail from new entrants.) Alice can, today, pay 5k for the year in advance, executing an incredible threat. Now Steve’s race to the bottom fails it seems to me. So Bob and Alice will arrive at some split of the spoils, and I see no obvious unique value, though I haven’t tried very hard I admit. But unless Steve’s split is unique his argument has a flaw.
OK, I’ve been following the last 2 posts, and thank you, Scott, for this clarification. I has seemed obvious to me, that you were working varying angles on a point you REALLY wanted to make clear. Up til now, I was assuming that I just wasn’t getting the math behind your model of social good.
Now it is clear how your math works.
In method 1 and 2, you ignore Alice’s development costs, so this allows any incremental improvement less than new development costs to show as a loss. Anything less than 100% gain in the first year is a social loss.
Got it.
Tell me, has any investment ever, using this math, shown a net social gain? Using this math, we should all be happy with our abacus and spinning wheel? The telephone was a net gain over the telegraph, but I doubt it was so much better that in it’s first year it offset the total investment previously put into the telegraph…
Steve,
I’m not buying your race-to-the-bottom scenario. I suspected you were assuming this kind of thing in your HFT argument. It wouldn’t happen.
First of all, Bob’s floor is much higher than Alice’s since he has to recoup the $80,000 development cost. Alice can easily undercut him by more than $10 and fend off his attack. If he went down to $59, which is the point at which he’s guaranteed to put Alice out of business, it would take almost 90 years at that level of demand to make back the investment costs. Not gonna happen.
Second, it’s not valid to argue for a race to the bottom (presumably Bob and Alice are doing this calculation introspectively) without taking into account the demand curve. Both Alice and Bob know that reducing the price will significantly increase demand. Let’s say that going down to $500 doubles sales of that software type. Now the social benefits are huge because 2000 buyers are getting a $500 advantage over what existed previously (I’m ignoring small $10 quantities), for a total of $1,000,000 benefit. Bob ends up transferring $50,000 of Alice’s profit to himself, but provides a HUGE benefit to the consumer.
The bottom line, as far as I can see, is that it’s impossible to argue that the established product, whether HFT or software, will lose all its value via transfer to the challenger without giving a vast benefit to the consumer, thus providing a large social benefit.
To reinforce Ken P’s point, Landsburg has successfully shown that in a very small market the cost of one-off development may outstrip the marginal value of the innovation to consumers. In the real world where the market for a software product is going to be significantly larger than 100 customers and of significantly longer duration than one year, that conclusion is irrelevant. For example, if we expand the market to 1000 and keep Landsburg’s other assumptions the same, the new competition becomes socially beneficial.
“Social Benefit: $10,000 [The gain to 100 consumers, each of whom now has a word processor worth $10 more]”
Wouldn’t $10 for 100 people be $1000 in total?
How is the $80,000 development cost for Bob’s software a social cost and not a social benefit? Bob’s spending $80k on development was someone else’s $80k income. Method I should be:
$80k private cost for Bob to develops cide
$80k social benefit payment to code writers
$95k private benefit profit for Bob
$1k social benefit for 100 copies of $10 better code.
Ken P: Alice’s development costs are sunk, which makes them irrelevant.
Tell me, has any investment ever, using this math, shown a net social gain? Using this math, we should all be happy with our abacus and spinning wheel?
Is this a serious question? Would you prefer to go back to the world of the abacus and the spinning wheel?
Smylers: Thanks! Going back to fix this.
Capt J Parker: In a competitive market, the code writers are paid their opportunity cost. So the $80K social cost is a cost not because it comes out of Bob’s pocket, but because it represents the value of what the coders could have been doing if they weren’t working for Bob.
(If the coding market is not competitive, so that Bob pays the coders something other than their opportunity cost, then it’s the opportunity cost, not Bob’s out-of-pocket expense, that should go into the accounting.)
Nathan Ashby:
For example, if we expand the market to 1000 and keep Landsburg’s other assumptions the same, the new competition becomes socially beneficial.
Sure, and if you increase Bob’s development costs it becomes socially wasteful again. The whole point is that the competitition might or might not be socially beneficial, depending on the specific numbers.
Brian:
First of all, Bob’s floor is much higher than Alice’s since he has to recoup the $80,000 development cost.
This is what economists call the “sunk cost fallacy”. Once Bob has developed is software, the $80,000 development cost becomes irrelevant to any decisions he makes going forward.
Both Alice and Bob know that reducing the price will significantly increase demand.
They know that it will increase demand. I don’t where “significantly” comes from. To keep this example simple, I assumed the increase was small enough to ignore. If you want to assume a much bigger effect on demand, it’s still not hard to construct examples where you get the same result, but those examples would have been just slightly harder to follow.
The bottom line, as far as I can see, is that it’s impossible to argue that the established product, whether HFT or software, will lose all its value via transfer to the challenger without giving a vast benefit to the consumer, thus providing a large social benefit.
This is certainly wrong. I leave it as an exercise to figure out how big the effect on demand can be in this example while still preserving the result.
Glen Raphael:
This example seems bizarrely contrived compared to how actual competition in software products works and I can’t help think the differences are relevant to the comparison being made. You are missing a LOT of the social benefit to new development. In real life, each new program makes the software ecosystem more vibrant and robust, benefitting almost everyone.
Of *course* it’s contrived; that’s the nature of examples that are constructed to illustrate a point. The issue people were confused about was: “When Bob competes with Alice, shoul the lost value of Alice’s investment be counted as a cost in the social accounting?”. To understand basic conceptual issues like that, you always want to focus on the simplest available example, not the most realistic.
When I teach my students about supply and demand, I tell them that if cancer researchers discover that coffee prevents cancer, the demand curve for coffee will shift right and the price will rise. I’ve never yet had a student object that I’ve failed to account for the biochemical mechanisms that make it impossible for coffee to cure canceer. We could, of course, spend an extra class and a half talking about what is and what is not a plausible cure for cancer, but that would be a considerable disservice to the students who are trying to understand the underlying economics.
Edited to add:
I initially misread the commenter name and thought you were someone else. The following paragraph was intended for that someone else, not for you, but I’m leaving it here because it might still be helpful to others:
You appear (from some of your posts in the earlier thread) to be shaky on a lot of the underlying economics. Let me suggest (and this is meant in the friendliest possible way) that your insistence on “realism” is hampering, not furthering, your understanding. The way to learn this stuff, for almost everyone, is to work through a whole lot of artificially simple examples that have stripped away complications to focus on one issue at a time.
nivedita:
Steve: question about this scenario. What’s wrong with the argument that, the moment Bob tries to charge more than $60, Alice will jump back into the market and take all the customers again.
In order to jump back in, Alice has to commit herself to a $5000 server cost that she’ll never recoup.
Rob Rawlings:
I suppose that one thing that will restrict entry into “winner-takes-all” industries is that fear of others entering (and wiping out your investment). This may restrict profit expectations and make other investments (that add more social value) seem more attractive in comparison.
Yes, this is undoubtedly correct and important.
Rob Rawlings:
If you make the investment decision based purely on social utility then you will only choose a) only if the present value of the future stream of social utility exceeds that for b).
If you make the investment decision based purely on profitability then you will choose a) if the profits from investing in ProductA are greater than those from investing in ProductB even if the stream of social utility from the investment in the new ProductA is less than the combined social utility of the old ProductA and ProductB.
You got it!
nivedita (#11): I do not think I understand your example. Alice stands outside the circle and pays $4 to Bob, who stands inside the circle. To earn this $4, Bob incurred a $4 cost. Alice is willing to pay this $4 because it reflects some benefit to her.
If I understand you correctly, you’re saying that the world came out $4 ahead here because of the benefit that Alice must have incurred. But that overlooks the fact that after accounting for her payment to Bob, Alice had no net benefit.
The bottom line is that a) Bob incurred a $4 cost, b) Alice enjoyed a $4 benefit, and c) Alice made a $4 payment to Bob.
a) gives us a $4 social cost. b) gives us a $4 social benefit. With c) you have a choice: You can either say it’s a pure transfer, hence no cost or benefit at all, or you can say it’s a $4 cost to Alice coupled with a $4 benefit to Bob. Either way, c) contributes a net of zero.
So social gain in this example is zero no matter how you compute it.
Steve, your argument is to say that if one competitor can spend a bunch a money to intimidate the other competitor out of the market, then that private expense exceeds the social benefit.
Does this ever happen?
In your example, you assume that Alice will drop out just because of the possibility that Bob will sell for $59. But they both lose money if Bob sells for $59, and that is unlikely to happen. Alice loses $1 per sale, and Bob loses because he can never recoup his investment at that price. Bob’s investors will consider that operating at a loss. Either one could decide to try to stick out the losses in order to drive the other out of business, like a game of chicken. As long as the game of chicken persists, there is a huge social benefit of cheap software.
If Alice and Bob do not play chicken, then there is no reason for one to drop out of the market. Then there is the social benefit of competitively priced software.
Steve @30: I’m not saying the net benefit to the world is $4. I’m saying that the net benefit is not less than zero — i.e. in the long run, this circle might not create social value, but it doesn’t actually destroy social value.
Steve @27: yes, but this is an artificial feature of the problem it seems to me. This is based on assuming that the decisions are sequenced, so that Bob decides whether or not to invest, then Alice and Bob in some order decide whether to rent the server and what price to set if they do. Once Bob has sunk his development cost, suppose Alice decides, then Bob decides whether to rent the server. If Alice decides to rent, Bob will rent only if he can charge at least $50: but Alice now has a sunk cost and can cut her price all the way down to zero. Bob, realizing this, will not choose to rent the server if the most he can charge in that scenario is $10. Knowing that she can stop Bob from renting the server, Alice goes ahead and rents it. Knowing that Alice will do this, Bob never invests. You can make this come out differently by assuming the incremental value of Bob’s product is at least $50 so he’ll rent no matter what Alice does, and since he will rent, Alice doesn’t rent, and Bob invests.
If Bob went first on the server on the other hand, Alice won’t rent the server, since she knows Bob can now cut his cost all the way down to zero (or down to $10, since that’s enough to take away her sales). Knowing this, Bob rents the server, and knowing that he will rent the server, he invests.
But all this changes if we instead assume they’re playing the game simultaneously — i.e. Alice decides either not to rent, or rent and a selling price, and Bob at the same time chooses whether to invest, rent a selling price, or not to invest. Then you have to work out what the equilibrium strategies for the two are, and I’m pretty sure the expected value is going to come out much different.
Or if we just drop the artificial assumption that Alice has to make a decision to rent the server for a year once and for all — if she’s selling this product over multiple years, she can just sign up for 10 years of server space (six months’ revenues cover that), and then Bob has a very high hurdle to overcome.
Or if you assume that Alice doesn’t really need a server, she just contracts with a reseller who will take $50 out of each sale. In this scenario, Bob can’t sell above $60.
I guess what bothers me is that this $5000 server cost seems to be a crucial feature of the example.
I may be mistaken, but don’t you still need to correct the $85,000 net cost under the “Bottom Line” section to be $79,000? And I think the parenthetical social benefit under “Method II” should be corrected to $1,000.
Don’t mean to be picky, but these typos made it harder for me to follow.
The way I frame it is to say net value of economic resources used by Bob is $80,000. Net value of economic resources created is $1,000. But wait, isn’t the development cost fixed? Hence, the first group of consumers Bob sells don’t cover the social cost, but eventually Bob’s sales will cover it, right? So, as in your reply #2 I believe the single volume of 100 copies is an essential assumption to make your case. Yes, a multi-year analysis might include additional costs, but those would be exogenous to what you’ve outlined while the additional sales at a marginal cost of zero would seem to be implied otherwise.
Alice starts off making high monopoly profits so can use some of this yo build a war chest.
Suppose she makes a statement to the market “I guarantee you the best value word-processor or I will give it away for free”.
If a competitor enters the market she may have to spend $5000 a year making good on this commitment. But the very commitment will probably keep competition out and divert investment to other areas.
Steve Landsberg #23
Bob’s $80k spending on code development is income for someone else. If it isn’t covering the code developers opportunity cost then it is at least going somewhere and I don’t think you are accounting for the benefit to its recipient.
What if we were to think about consumption. If Bob’s activity increases total consumption then there is a net economic benefit. If not, then there is a net economic loss even if Bob gains. In Method #I Bob can increase his consumption by $15k (profit net of development expense.) His 1000 customers can increased their consumption by $1k ($1010 software at $1000 cost.) All pretty clear from your analysis. So now, what happens in the rest of the economy where Bob purchased software development services? In the software development services market $80k additional development services are produced so up to $80K is available for additional consumption from that market. Now you need to ask where do the resources come from to produce the additional $80k? They can’t come from any market where they were able to produce more than $80k of output otherwise their opportunity cost would be more than $80k. So consumption can’t decrease elsewhere more than $80k because of resource reallocation to the software development market.
So, Method #I should look like this:
$80k Private cost (decrease of consumption) for Bob for software development.
$80k Social benefit (increase in consumption) for software development suppliers.
$80k (at most) Social cost (decrease in consumption) from resource allocation to software development and away from anything any other type of production that then allows producers to consume. (i.e. the opportunity cost of software development resources)
$95k Private benefit (increase in consumption) to Bob in profit.
$1k Social benefit (increase in consumption) to consumers for $1010 software at $1000 cost.
Net increase in consumption $16k
Net social benefit $1000
Method II calculation:
“Private Benefit: $95,000 (Bob’s profit)
Social Benefit: $96,000 (Bob’s profit plus the $10,000 gain to consumers)”
I think that should read “…plus the $1,000 gain to consumers)”
Sorry. See my point has been mentioned already in the long thread. I assumed if the error were still visible it hadn’t been reported yet. Apparently market corrections are NOT instantaneous.
Re:my post #37 change Bob’s 1000 customers to Bob’s 100 customers. Just in case Ted Levy is watching.
nivedita:
I guess what bothers me is that this $5000 server cost seems to be a crucial feature of the example.
Yes, it is indeed a crucial feature of this particular example.
Fonzy Shazam (#35): Sorry for the typos. The correct number is $94,000, not $85,000 (as I had before).
I hope I finally have he arithmetic right. Let me know if I’ve screwed up again!
Capt J Parker (#37): You are right that the income to the software developers is a plus and the resource redirection to software development is a minus. The point you’re missing is that if the software development market is competitive, then these two amounts are equal at the margin so they cancel and can be left out of the accounting.
Steve Landsberg #47
Thank you for replying twice, Dr. Landsberg. I’m afraid I still think your analysis is incorrect. Bob pays for the code developers opportunity cost. The code developers opportunity cost and what I clumsily called the cost for resource redirection are one and the same cost. I quite fully agree that “if the software development market is competitive, then these two amounts [the (a)opportunity cost of software developers and (b)the amount Bob pays them] are equal at the margin so they cancel and can be left out of the accounting.” But you didn’t do this in Method I. You accounted for (a) but left out (b). To be clearer I should say you accounted for the social cost(a) but you left out the social benefit part of Bob’s payment (b) and only accounted for private cost to Bob in order to make that payment. The code developers could have been producing $80k of video games for Carol instead of code for Bob. But, that doesn’t mean that society would be is $80k better off in that instance (i.e. incur $80k less cost) because the $80k the code developers produce in that case is canceled by the $80k Carol must pay them. So, if the code developers ARE working for Bob and Bob is paying their opportunity cost where does the $80k social cost in Method I come from? I’ll ask the same question in a different way. Who must consume less (other then Bob) in order for the code developers to be working for Bob as opposed to working for someone else? If no one outside of Bob must consume less then no one is bearing a cost.
Sorry, the above post #44 should be referring to Dr. Landsberg’s reply #43
Capt. J Parker:
But, that doesn’t mean that society would be is $80k better off in that instance (i.e. incur $80k less cost) because the $80k the code developers produce in that case is canceled by the $80k Carol must pay them.
You’ll see this more clearly if you ignore all the payments, which cancel out in the end anyway (whatever X pays Y is a cost to X and a benefit — or, if you prefer, an negative cost to Y). Instead of looking at the payments, look at what the world’s got. If Bob develops his software, then some other project doesn’t get completed. That’s the social cost of development (which of course has to be weighed against the social benefits). The usual way we account for this is by counting Bob’s out-of-pocket payment as a social cost, because it represents the value of the alternative project.
Who must consume less (other then Bob) in order for the code developers to be working for Bob as opposed to working for someone else?
The people who would otherwise have consumed Carol’s project no longer have that option. They therefore choose to consume something they like slightly less, which is now no longer available to other people, who in turn choose to consume something they like slightly less, etc. All of this reduced value-of-consumption has to add up to the $80K.
Of course, offsetting this, we have all the people who are now made happier by Bob’s project. But by hypothesis, those people are gaining very little from Bob’s project, because it’s only a smidgeon better than Alice’s, which was available all along. So there’s a small benefit due to Bob’s project, weighed against a big loss from the absence of Carol’s.
Okay, I think I understand the disconnect I and others have with your position. I dare to say this is even what Don Beaudroux’s complaint would boil down to. First an example that I think best illustrates your many examples:
I sing in the shower every morning at no cost to me. For some reason all 250 of my neighbors can pay $1 each morning to hear this singing. The $1 per person or $250 total exactly equals their consumption benefit from hearing me sing. This continues for many mornings. Then my next door neighbor realizes that if he destroys $5 worth of resources, he can sing slightly better than me for that morning. Then all the other neighbors can and will begin paying him the new value they enjoy hearing him sing instead of me. We sing at the same time; so he fully replaces me in the shower singing business. The new value the listeners place on the singing is $1.01 or $252.50 cumulatively (notice he stopped paying but I started). Because he gets a huge net benefit of $247.50, he does this act of destroying $5 worth of resources (his personal cost) to get the revenue of $252.50 (his personal benefit). Keep in mind that to sing better than me he has to destroy $5 worth of resources each day (this gets us out if the confusion of sunk/fixed investment cost versus continual future production and hence social benefit). So the activity he does every morning is socially wasteful (destroys $5 worth of resources for 2.50 worth of NET ADDITIONAL social benefit), but it is individually beneficial since it reaps him such a big personal gain.
Where I have the problem with this and all of these examples is in this essential component “the activity destroys $5 worth of resources to create $2.50 of resources/benefits” or in other words “the individual incentive is to destroy more resources than is produced from a total society standpoint”. I contend that this is beyond fanciful. No matter how cleverly we craft an illustration to supposedly demonstrate this possibility, it is very, very unlikely to be true. Not impossible, but exceptionally remote. When applying this concept to a real-world problem like HFT, we are much, much more likely to be missing something essential that makes it not a case of destroying more than what is produced than it is we have properly diagnosed a case of socially destructive competition. We are beyond Occam’s Razor. Socially destructive competition in a free market is too unlikely and fleeting to worry about.
But maybe I’m wrong…
Fonzy, I think your example doesn’t work because the fixed cost of getting people to listen is actually essential. In your example, what would happen is that your neighbor would charge $0.01. If he tried to charge $0.02, you would offer to sing instead for $0.01.
Fonzy Shazam: Your first paragraph is exactly right as an illustration of what we’re talking about.
Your second paragraph relies on your gut instincts about what is and is not plausible, and it’s certainly wrong. The economics literature is full of examples. This phenomenon happens in the sports industry all the time, where the number one swimmer gets rewards that far exceed those to the number two swimmer, creating an incentive to invest huge resources in being number one, even though you’re adding only a small amount to the spectators’ experience. We see similar phenomena in the entertainment industry, especially since modern technology allows, say, the number one comedian to serve essentially the entire world. There are surely cases of this in corporations, where being CEO carries a much higher reward than being number two. I have no idea how important this phenomenon is in the IT industry, but given that a single software product, like a single comedian, can serve the entire world, it seems quite plausible that the phenomenon is widespread, and I have no idea why you’d say it’s “very very unlikely”. It seems like a statement this strong needs some kind of an argument, and I don’t see where you’ve offered one.
There are circumstances — thankfully quite common circumstances — in which competitive markets yield not only very good outcomes but the very best possible outcomes. I always stress to my students that we should view this as a miracle, both in the sense that it’s very good and in the sense that it’s completely the opposite of what we should expect. We see harmful competition every day. When something exciting happens at the ballpark, everyone stands up to see better and therefore nobody sees better. At parties, we all talk a little louder to be heard over everyone else, and as a result nobody is heard any better, but we all go home with sore throats.
The ubiquity of harmful competition is exactly, in my opinion, why we should cherish and nourish market conditions that render competition not only harmless but the world’s greatest force for good. THat makes it important for us to understand what those conditions are. Usually, they can be boiled down to the single word “freedom”. But not always.
nivedita: You are right that Fonzy needs something to prevent this sort of price competition. A fixed cost will do it, but I thought instead that he was assuming simply that the neighbors all want to throw dollar bills at the best singer, and that once they’ve heard the best singer, they lose all interest in anyone who’s not quite as good. That also works.
nivedita #48
You’re right. There is a flaw in my story to make it consistent with Landsburg’s original above as well as other examples including the Microsoft one he poses in a comment on Boudreaux’s blog post. To correct my story I need to say the $5 resource destruction is a one-time investment event *and* the advantage has a diminishing value to consumers such that the willingness to pay declines at a rate such that the cumulative value to society never quite reaches $5.
Yet, I believe my larger point is maintained–that these hypothetical examples are quite hard to apply to the world as we know it.
Thanks for noticing my error.
Landsburg #49 & #50
Great points, of course. And yes, I realized as I was making the point originally that I was speaking from my priors, but I think they are good priors (still not an argument, I know). I guess my argument would be made from a Hayekian respect for complexity. I think we are making a big leap from the talking louder at a party examples to HFT, et al. I think it is a bridge too far. But it is hard to prove the negative here. I’ll keep thinking…
Re: #15, #17, #21, #25. I think that KenB and Brian have essentially identified the problem with this scenario. Steve’s analysis assumes that the market for the software will last one more year (#2). Because Alice has already developed her software, her marginal cost for providing 1 more year worth of software (100 copies) is $5k. Bob, however, must incur 80k in development costs in addition to the 5k of server costs. The 80k is *not* a sunk cost because Bob incurs this cost only if he wants to produce the software this year. Alice’s past development costs (whatever they were) *are* sunk costs because she already incurred them, regardless of whether she sells software this year.
Because Alice has a 80k marginal cost advantage over Bob, she can reduce her price to anything below $850/copy (85k for 100 copies) to ward off Bob. The “social waste” here arises because Alice is not pricing to take advantage of her cost advantage, the fact that she incurs no marginal development costs to serve 1 more year of customers.
I am assuming here that Alice can anticipate the threat of Bob’s entry *before* Bob decides to develop his software. Maybe, Steve had in mind a scenario where Bob’s entry was a surprise to Alice, but his comment in #4 suggests that the potential of future competitors ought to be at least considered by Alice in her pricing. Regardless, the 79k of social waste clearly arises because of the market’s failure to account for Alice’s 80k cost advantage. Maybe it’s just semantics, but I don’t think it’s quite accurate to describe this waste as arising from Bob’s decision to compete. It seems that the cause is that the market price does not reflect Alice’s lower marginal cost. Finally, maybe a social benefit of Bob’s entry is that future incumbents in other markets will learn to better anticipate the threat of competitors and pre-emptively lower their prices in a way that Alice didn’t. That, in turn, could lead to less social waste of the type that Steve describes here.
“But Alice’s price cannot fall below $50, because then she wouldn’t earn enough to cover her server costs. So Alice, who is smart enough to foresee all this, gives up and cedes the market to Bob. ”
I think this is the key assumption. I think if Alice is really smart she will realize that if she communicate the fact that she won’t give up in this situation but take a loss rather than ceding the market then new entrants will be deterred and wasteful investment avoided.
Only competitors who can make a profit even if Alice gives her product away will enter. Fear of this will motivate Alice to continually improve her product even without actual competitors. I suspect that these conditions may result in maximum consumer benefits even with a winner-take-all monopoly.
Steve Landsberg #46
Thank you again for the reply. I have been sloppy in stating a clear objection to your Method I analysis. I try harder to do so taking into account you explanations. My objection is that the social cost you lit for Bob’s software development is offset by Bob’s payment for software development resources and that you did not account for that offset in Method I.
When Bob develops his software the world is out $80k of software development resource. The world gains $80k from Bob. You are not accounting for this $80k from Bob in Method I. The net cost to the world caused by Bob’s utilization of software development resources is zero.
You still need to account for the $80k from Bob. That $80k allows consumption that could not happen if Carol’s project proceeds and Bob makes no such payment. So, the $80k loss from not doing Carlo’s project is offset by Bob’s $80k payment.
I will also use your suggestion of “looking at what the world has” to answer my question of “Who must consume less (other than Bob) in order for the code developers to be working for Bob as opposed to working for someone else?”
If Bob doesn’t develop his software the world has produces $X for itself and can therefore consume $X. If Bob develops his software. The world produces $80k less because it loses the production that Bob demands. At the same time the world gains $80k form Bob so the world’s consumption does not decrease because of Bobs development.
There is Alice’s software available all along. Bob and Carol are each contemplating their projects which will each cost $80k and use identical kinds of development resources and have an end result that will compete with Alice. If there is an $80k loss from not doing Carols project and doing Bob’s instead there must also be an $80k loss from not doing Bob’s Project and doing Carol’s instead. Conversely, if there is an $80k benefit to doing Carol’s Project there is also an $80k Benefit from doing Bob’s. This Benefit from doing Bobs project does not show up in your Method I analysis. If it was in your Method I analysis it would offset the $80k Social Cost for Bob’s software development that you list.
It seems like there’s a hidden assumption that no one has mentioned yet: When Alice decided to develop her word processor, she had no idea anyone could ever make another word processor that could compete with hers. I suggest that that assumption is implausible, and therefore this example is not in equilibrium.
Ben:
When you build a house, you might foresee the possibility that it could be destroyed by an earthquake, and you might decide the risk is worth taking. It does not follow that earthquakes are not destructive.
Capt J Parker: I haven’t yet read all of your newest comment, but I got this far:
The world gains $80k from Bob.
No. Bob is part of the world. There is no gain when money leaves Bob’s pocket and enters someone else’s.
While I get the point of your post and that SB=SC for efficiency, the arithmetic is throwing me off.
This is what I’m calculating.
Alice’s already developed software:
PC: $5,000 (server maintenance)
SC: $5,000 (same as above)
PB: $100,000 ($1,000 x 100)
SB: $100,000 (($2,000 x 100)-($1,000 x 100))
Thus, PB-PC = SB-SC
Now onto Bob:
PC: $85,000 (developmental costs and server maintenance)
SC: $85,000 (same as above)
PB: $100,000 ($1,000 x 100)
SB: $101,000 (($2,010 x 100)-($1,000 x 100))
PB-PC < SB-SC by $1,000 ($15,000 < $16,000), which suggests underinvestment, so there should be more Bobs.
Where did the extra $1000 go that consumers valued the software package at? It was there for Alice, but then you said Bob developed one that they value $10 more than Alice's. So shouldn't that be $1,010 in consumer surplus?
I am probably missing the point of this exercise. Among the assumptions made to provide the data to complete the social accounting analysis is that Bob is omniscient. He knows before creating his product what its development costs will be and, more implausibly, that it will perform as he envisions it will perform and, further, that it will be valued by customers as he assumes. The social cost of a particular innovation can be calculated–in theory–after the fact. But, without the benefit of knowing the unknowable, cannot be ascertained before action is taken.
Steve Landsberg # 58
You said:
Yes, lets agree that Bob is part of the world and Carol is part of the world and by extension the same accounting rules apply identically to both. In your kind response #46 you said:
and you said further:
By this reasoning if Bob were to decide not to do his project and Carol did hers instead there is an $80k social benefit from doing Carol’s and not Bob’s. This is assumed to be true without saying anything about the nature of Carols project other than it used $80k of software development resource that Bob did not use. But, let’s suppose we come to find out that Carols project is the exact same thing as Bob’s. In that instance, again by your reasoning, we have a small social benefit from Carol’s software because it is still only a smidgeon better than Alice’s, which was available all along. We also have all the people that are now avoiding the $80k loss you attribute to not doing Carols project. So there’s a small benefit due to Carol’s software, and a big gain ($80k) from doing Carols project and not Bob’s. This erroneous result comes from the fact that there is an $80k benefit from Bob’s project just the same as there is an $80k benefit doing Carol’s project but you left out the $80k benefit of doing Bob’s project in your Method I analysis. Your analysis argues that there is an $80k cost when Bob develops a software product but three would be an $80k benefit (or at least the avoidance of an $80k cost)if Carol were to develop the exact same product.
Charles G. Phillips: So…if we make the alternative assumption that Bob is not omniscient, do you have a proof that he’ll undertake all and only those investment projects that are socially desirable?
Suppose people are wondering whether every square is also a fourth power. I point out that 36 is a square, but it’s not a fourth power. Someone replies that this example doesn’t count, because 36 is even. I think I could reasonably ask why that’s relevant, and whether this person has some reason to believe that if we excluded even numbers, we would somehow be able to get the result.
To belabor the point: People are wondering whether private investment decisions are socially optimal. I point out an example of a private investment decision that is not socially optimal. You point out that in that example, the investor is omniscient. I think it’s fair to ask whether you have some reason to believe that this is relevant to the question, and in particular whether you have a proof that without the omniscience, we’d somehow always manage to have social and private optima coinciding.
(PS: The answer is that you can’t have such a proof, because the result you want is not true — I can easily tweak this example to add some uncertainty without changing the bottom line. But if you *think* you have such a proof and care to share it, I’ll be glad to help you find the error.)
Capt J Parker:
By this reasoning if Bob were to decide not to do his project and Carol did hers instead there is an $80k social benefit from doing Carol’s and not Bob’s. This is assumed to be true without saying anything about the nature of Carols project other than it used $80k of software development resource that Bob did not use. But, let’s suppose we come to find out that Carols project is the exact same thing as Bob’s. In that instance, again by your reasoning, we have a small social benefit from Carol’s software because it is still only a smidgeon better than Alice’s, which was available all along.
Ah, yes, this is correct. If Bob is distracting people from doing something (nearly) socially worthless, then he’s not doing much harm.
You can also note that if Bob hires coders who would otherwise have been committing arson, he’s done the world a whole lot of good.
The usual assumption in this stuff is that for the most part, people’s wages reflect the social opportunity costs of their activities — that is, that most markets (in particular, all relevant markets other than the one we’re investigating) price things properly. With that assumption, it’s Bob who’s socially destructive; without it, it’s the entire software industry — Carol and all the rest of them — who are jointly destructive.
But yes, if your point is that Bob is hiring coders who are overpaid relative to their social value, then you’re absolutely right that it’s their social value, not their wage, that should go into the calculation.
Professor Landsburg,
As I said, I must be missing the point of this exercise. It is impossible to know before Bob or anyone else makes an investment whether such investment will be socially desirable. And, since it is not possible to know, why bother to worry about it? I can imagine statists who would like the responsibility for subjecting private investment to a “social desirability” test, but that is the road to Solyndra and Fiskar.
Charles G Phillips:
But, if you are observing that some private investments are not socially desirable, I am quite certain that you are correct. Again, so what?
Well, first, there were people to whom this wasn’t clear, so it was intended to help them.
Second, I had confused some of those people in an earlier thread by switching back and forth between two methods of accounting. This was intended to help those people past the difficulty I’d created.
Steve Landsberg #63
My point is that you claim that there is an $80k social benefit when Carol develops a specific software product but you also claim via Method 1 that there is no such $80k social benefit should it happen to be Bob instead of Carol who develops the exact same software product and this is an accounting error on your part.
Steve, solve for the equilibria. If Bob will enter the market with p=0, then Alice enters the market and there is $195,000 social gain. If Bob will enter the market with p=1, then Alice doesn’t enter the market and there is $121,000 social gain.
If Bob will enter the market with p=0.5, then Alice examines her options. If she enters the market, she makes $95,000 half of the time and loses $5,000 half of the time. If she doesn’t enter the market, she makes $0. Her risk profile probably prompts her to enter the market in this case, so there is a social gain of $195,000 half the time and a social gain of $116,000 half the time.
If Bob will enter the market with p=0.95-epsilon, then Alice again decides to enter the market (supposing she is risk-neutral), though doing so nets her only a very small amount of utility. There is a social gain of $195,000 5% of the time and a social gain of $116,000 95% of the time.
The way your scenario is phrased, Alice acted initially with the assumption that p=0, and then Bob still entered the market. That is not reasonable, and that’s why it runs against intuition.
These arguemnts about harmful competition are dubious. I agree with Fonzy that “Socially destructive competition in a free market is too unlikely and fleeting to worry about.”
Steve responds with an argument about how the no. 2 comedian is almost as good as the best. Tf true then NBC did not need to pay millions of dollars to Seinfeld when they could hire a low-cost substitute. Now Seinfeld is off the show, and where is that low-cost substitute?
Roger:
Steve responds with an argument about how the no. 2 comedian is almost as good as the best. Tf true then NBC did not need to pay millions of dollars to Seinfeld when they could hire a low-cost substitute.
Wow, have you missed the point! The entire point here is that private rewards are divorced from social rewards. What better illustration could there be than Seinfeld earning millions when the next-best guy would have been almost as good?
Ben: You told me in one paragraph that if p=.5 then Alice might enter the market; then you told me in another paragraph that because Alice entered the market we must have p=0. Please stop to think before posting.
Capt. J Parker:
My point is that you claim that there is an $80k social benefit when Carol develops a specific software product but you also claim via Method 1 that there is no such $80k social benefit should it happen to be Bob instead of Carol who develops the exact same software product and this is an accounting error on your part.
And my point is that there was an implicit assumption that Carol’s product is worth $80,000. Now that assumption is explicit.
Charles G. Phillips:
As I said, I must be missing the point of this exercise. It is impossible to know before Bob or anyone else makes an investment whether such investment will be socially desirable. And, since it is not possible to know, why bother to worry about it?
It is also impossible to predict earthquakes with anything close to certainty. Does it follow that we should stop trying to understand earthquakes?
Steve Landsberg #71
And my point is that there is also now an explicit assumption that Bob’s product must also be worth $80,000. But that assumption is missing from your Method I analysis.
Surely this argument applies, to some extent, to any industry with start-up costs. No?
Capt J Parker:
And my point is that there is also now an explicit assumption that Bob’s product must also be worth $80,000. But that assumption is missing from your Method I analysis.
I totally don’t get this. Bob’s project displaces Alice’s. Carol’s project is to upgrade the internal email at her company. She’s willing to pay $80,000 for it, it impinges on nobody but herself, so its social value must be $80,000. The “not impinging on anyone else” bit applies to Carol and not to Bob. Why is that impossible?
Steve Landsberg: #75
In #46 you said:
And your starting assumptions were:
and
Bob is not competing with Alice for software development resources he is competing with Carol. It is Carol’s development project this will not get done if Bob does his. And likewise it is Bob’s project that will not get done if Carol does hers. Bob and Carol are both “part of the world” so the same accounting rules must apply to both. They are symmetrical actors. If Bob imposes a cost because Carol’s project doesn’t get done then Carol imposes the same cost because Bob’s project doesn’t get done. (assuming they use an identical amount of resources) If Carol’s project creates a social value because she is willing to pay $80k for it then Bob’s project creates the same social value because he is willing to pay $80k for it.
Capt. J Parker:
Bob is not competing with Alice for software development resources he is competing with Carol.
This is correct.
Bob and Carol are both “part of the world” so the same accounting rules must apply to both. They are symmetrical actors.
This is incorrect. They are not symmetrical actors if one has a project that displaces Alice and the other has a project that displaces nothing.
Suppose Bob wants to hire Doug to burn Alice’s house down while Carol wants to hire Doug to build a house. Would you argue that because they’re competing for Doug’s services, Bob and Carol are “symmetrical actors” and that we have to treat their projects equally in the social accounting?
Steve, that’s not what I said at all. Please read before posting.
Ben: I did read the part where you wrote this:
The way your scenario is phrased, Alice acted initially with the assumption that p=0
I also read the early part where you said that Alice might have entered the market at some $p>0$.
Since the only think we know about Alice is that she entered the market, and since she might have entered the market assuming $p>0$, we can certainly have no grounds to infer that Alice assumed $p=0$.
You forgot to add in that now Alice is able to spend her time developing a different software. Just like the 2nd best comedian can go on to increase the social benefit by writing a different show.
bob: Alice could do that anyway. The supposition here was that the software is already written, and can be disseminated at zero marginal cost.
Steve Landsberg #77
This is a false analogy because in the software example it is Bob who does the torching not the software developers. If Bob wants to hire Ted (who’s Doug?) to build a napalm bomb and Carol wants to hire the very capable Ted to build a house and Ted charges the same amount for each activity then absolutely I would treat both projects symmetrically for cost accounting social or otherwise. Then if Bob sets the Bomb off in Alice’s house or alternatively Carol decides to rent her house at below market to cash strapped entrepreneurs then I fully agree there are very different costs and benefits to those two situations. But, in your Method I analysis you try to take into account the costs of building the bomb (incorrectly in my view) but you don’t seem to say much about the costs accrued from setting it off in Alice’s house.
In post #46 you said:
This implies the social cost of Bob developing his code (or building the bomb) has nothing to do with what the code (or bomb) will be used for only what he must pay in development costs.
Do you still believe this is true? I think you are correct in wanting to factor in the fact that Bob is going to torch Alice’s house because that is the reason why a profitable new enterprise might benefit the entrepreneur and his customers but hurt society as a whole. Putting Alice out of business (rendering her productive capital unproductive) is the externality that causes the social cost of Bob’s project to exceed Bob’s private cost and therefore prompts Bob to make a socially economically inefficient decision. Your Method I seems to say that Bob’s project’s social cost equals Bob’s private development cost but then there is some reason to double count the development cost.
One more thought
What is the social cost of destroying Alice’s capital? Is it the NPV of her expected future income stream? Is it her original cost? Is it the depreciated value of her capital. It is clearly not the same as Bob’s development cost because Bob could easily use an $80k napalm bomb to destroy either a $3 million dollar mansion or a $500 tarpaper shack.
Steve Landsberg,
I am very greatful for your willingness to have a conversation on this subject. If I’m slow in responding please understand that it’s because the work week is here not because Iv’e lost interest
Ben, though I think solving for the equilibrium wasn’t the point of the original post, it does seem to be interesting to try.
You also need to consider what price they will set if they do enter (where “enter” for Alice just means that she rents the server for the year, since she’s already developed her software). i.e. their choices span not enter, plus enter and set some price where price can plausibly range from $50 to $1000 for Alice and $850 to $1010 for Bob (though given that Bob must charge at least $850, Alice won’t actually charge anything less than $840).
I tried to work out what the equilibrium could be, and I seem to conclude that Alice will be in the market for sure (but set her price probabilistically), and Bob will enter with probability about 16% (again setting his price probabilistically), which already leads to social waste. Of course my math might be wrong.
Capt. J Parker:
My point is that you claim that there is an $80k social benefit when Carol develops a specific software product but you also claim via Method 1 that there is no such $80k social benefit should it happen to be Bob instead of Carol who develops the exact same software product and this is an accounting error on your part.
And my point is that there is no presumption that Bob and Carol are developing exactly the same software product. If they are, then I absolutely agree with you. But in this example, I’ve assumed that they’re not.
Capt J Parker:
What is the social cost of destroying Alice’s capital? Is it the NPV of her expected future income stream? Is it her original cost? Is it the depreciated value of her capital. It is clearly not the same as Bob’s development cost because Bob could easily use an $80k napalm bomb to destroy either a $3 million dollar mansion or a $500 tarpaper shack.
Wait, wait….I thought you wanted to do everything via Method I. If we use Method I, there is no social cost to destroying Alice’s capital. The social cost consists entirely of the forgone value of the alternative use of the resources (e.g. developers’ time) to which Bob lays claim.
If we assume, as is usually the case, that the alternative value of those resources is well measured by their price, then we count the social cost as $80K. If instead the alternative was to do something just as socially useless as Bob’s project, then we count the social cost as $0. If the alternative use was to commit arson, then we count the social cost as negative.
But as long as you’re focused on Method I — which I thought you were — Alice’s losses don’t go into the accounting at all, so I don’t understand the relevance of your question about how to count Alice’s losses.
I just recalled a famous economist (who is often mentioned on this blog) once writing about “the economics of QWERTY”, the upshot of which seemed to be a concern that *under*-investment in winner-take-all situations can leave us trapped with inferior technologies. Of course this would not rule out over-investment in every case in any event, and it also seems history has not been too kind to the QWERTY story (even in the case of typewriters). But it seems what may be bothering people (beyond the financial traders) is 1) that notions of “winner take all” and (particularly involving) “network effects” tend to suggest that larger benefits are attainable from even small improvements in a way that isn’t consistent with the limited (e.g. one-round) framework being used here; and 2) while the example of HFT may be an exception, people are presuming that the purpose of focusing on it is so we’re expected to generalize from it (although Landsburg clearly suggested we’re in case-by-case territory here)
P.S. — that this blog is the best entertainment value on the internet (for all things touching on economics at least) was solidified by the “Methinks’ wager” and its aftermath.
Steve Landsberg #86
You said:
and
Well, Method 1 includes a cost that is derived from Alice’s house being set on fire alright, it’s just a little hard to see this cost because of all the smoke. Here is why I conclude that: Bob consumes $85k of resources from the world and he delivers $101k of software to the world. We know this is true because all those resources are well valued by their price. So If Bob were really Carol who is competing with no one (in other words if Alice’s losses were truly not in the picture In Method I like you claim they are not) then the cost benefit analysis for Bob’s project would end there with the $85K social cost and the $101k social benefit and we would conclude Bob delivered a net social benefit of $16k to the world. But Alice and her forgone revenue are part of the picture apparently. Alice is the only factor that, in Method I, allows you to say “wait, wait… the world could have received $100k worth of essentially the same software from Alice and under that scenario the world would only have had to give up $5k of server resources so Bob is really only delivering $1k of software benefit to the world while consuming $80k of net resources. So, in Method 1 it is really the fact that Bob has shoved Alice out of the market (set her house on fire) that makes Bob’s project have negative social value. Buried in Method 1 is an assumption that Alice’s inability to maintain her $100k sales in a competitive market results in a $100k social cost that is to be charged against Bob. This assumption seems wrong to me.
Steve Landsburg:
I notice that you’ve not replied to my comments 17 & 18. I’d be honored if you would.
Following up on both, we know that all stock market gains and losses net to zero. Meanwhile millions get spent on financial research. But this can’t all be a social loss because at least some amount of financial research is needed for capital markets to price assets. You might argue that the total expenditure on financial research is too high (just look at the size of the financial sector!) or too low (decades of published research and value premium hasn’t been arbed away!) but is there a social loss? I would argue that we can’t answer unless we know the cost of implementing some corrective action. Determining the optimal amount of financial research and making sure that just the right amount got done must cost something. If that cost is more than the costs associated with doing nothing, then the status quo is the optimal outcome and I don’t see how you can claim that there is a social loss associated with the status quo.
I’m pretty sure I understand the point, but I don’t think this is the best example to illustrate it, because I think the logical conclusion to the scenario is not the one described.
If Alice can’t go below $50 or she won’t make any money, then I don’t see why Bob can make the decision to sell below $850, as he won’t make any money at that point either. If he DOES price down to $50 in order to drive Alice out, then when he attempts to raise the price Alice jumps back in and steals the market from him. Bob can’t make a dime in this scenario, because his product can’t compete at the lower end of the pricing scale, so he always loses the price war.
The ultimate result I see of this exchange is Alice selling her software at $839.99. $16,000 loss to Alice thanks to Bob’s idea, but $15000 gain to Alice’s customers thanks to Bob’s idea. The net gain is now $94,000, only $1000 in social loss thanks to Bob’s idea.
If, however, the customers are willing to buy the best at any price to a maximum of $110 (the value of Bob’s software), then the price war is meaningless, Alice never wins, and the $80,000 social cost is back.