This extended essay is an edited version of some posts I submitted to a private email list. I thought I'd share them with you as the questions raised are important, and it's a debate worth having.
There's a gaping hole in our understanding of telecom. A yawning, jagged, iceberg-sized rip in the fabric of our knowledge. We've no idea how the world of bit-shifting interacts with the world of atom-arranging. How do telecom and transport substitute or complement one another?
This is kind of important. Trillions of dollars of immobile capital infrastructure are at stake if we misunderstand the relationship between physical and virtual transport. Would uncontended gigabit connectivity everywhere make cities' populations explode or implode? What if all those freeways to the suburbs become roads to nowhere (ever more than they are now)?
Substitution and elasticity, an introduction
Many have claimed that slightly counter-intuitively, telecom and transport are complements, not substitutes. For example, cyberguru Andrew Odlyzko writes (p10):
While telecommuting and videoconferencing are likely to grow, that will not reduce road congestion. There is this strangely persistent myth (which I have traced back to the 1830s, and its origins are surely even earlier) that telecommunications and transportation are substitutes for each other. They are not, and are in fact positively correlated. Hence we should expect growth of travel at the same time as telecom usage is booming.
I'm not sure if this is always true. The exceptions to the above claim might be important for how our human geography changes.
Three definitions that enable us to have a meaningful debate. Income elasticity is how the fast consumption of a good varies with total income. Price elasticity is likewise, but for price. Price cross-elasticity is how the demand varies with the price of some other good. It is the direction of the cross-elasticity that defines whether a good is a complement (guns and bullets), or a substitute (guns and crossbows).
Reasons to be suspicious of the claim
So why should we doubt this claim about telecom and transport being complements?
Firstly, correlation isn't causation. After all, the stork population of Denmark apparently tracked the birth rate rather well for a long time.
There is also a misunderstanding what substitute or complementary goods really are, in the technical sense. The positive income elasticities of travel and communications may swamp any substitution cross-elasticities in a period of rapidly rising incomes. This is to say, the effects of getting richer outweigh the effects of getting relatively cheaper. Trending over time isn't enough to separate out the cross-elasticity from the income elasticity.
Indeed, it is not uncommon for goods to have opposite price and income elasticities, and in the extreme this can lead to the paradox of Giffen goods where consumption actually rises with price at the same time as real incomes fall. Telecom might possibly be such a good; it would be very exciting if it were demonstrated to be so, since real examples are rare to non-existent. It would be interesting to track the price and usage of telecom in places like Iraq where 20%+ income falls are the norm.
Secondly, we're only just beginning to have telecom of a grade where the sense of "being there" resembles in any way a substitute for actual travel. There may be a point in the graph where past relationships break down. We could sub-segment our market into "travel for information exchange" vs. "travel for sense of presence", and might find substitution occuring in the former already, and the latter impending.
Thirdly, there may be a hysteresis effect. Telecom and travel have large fixed capital assets, but telecom has low operational costs. Rising incomes stimulate telecom demand in a virtuous circle. But falling real incomes don't make the infrastructure go away, and the low operational costs of telecom may indeed cause substitution. This scenario would be in play following a large oil price shock.
Fourthly, we can also see limits to growth in travel that don't exist in telecom. If there are complements, travel is going to run out of runway:
Meaningful debate needs meaningful terms
Lastly, and most importantly, "travel" and "telecom" are not necessarily single monolithic things that can be meaningfully reasoned about side-by-side. Travel comes in three broad forms, each of which has different price, income and cross elasticity. The three forms are "travel as producer transport", where the travel is incidental to your producing some other economic good; "travel as consumer transport", where it is incidental to consuming other goods and services (e.g. shopping or restaurant visits); and "travel as consumer application", where it is an end in itself (e.g. going to Greece on vacation).
These last two are clear telecom complements via e-commerce and social computing. Travel incidental to consumption is only substitutable to the extent that the desired end good or service can be replaced by an information good. But Zagat restaurant guide isn't much of a replacement for a real dinner, so this isn't normally the case. Information goods are normally deficient substitutes for physical consumer experiences. Plus there's no real telecom substitute for actually standing in the Parthenon. The income effects of better information and network technologies swamp the price substitution effects anyway. The more you earn the more you travel to consume, regardless of marginal price substitution of travel and telecom.
The exception case is "travel as producer transport". This appears to have unusual and interesting price and income elasticity curves, which may contradict the claim that travel and telecom are complements.
At the very low end the ability to travel at all is a status symbol. But for the worker classes in developed countries, people have to be paid to commute or stay in business hotels because it is boring and unpleasant. Every management and IT consultant out on the road whinges endlessly about the travails of travel. The inconveniences of being an oil explorer or truck driver generate premium wages compared to comparatively skilled occupations. You only seek to spend time with work acquaintances in inverse proportion to having a real life. But go a bit higher up the income scale and the boss will stay at the head office while the salesmen run around the client sites. At the very high end, you want to show off the executive jet, and travel becomes desirable again. This makes for a strange, wiggly income elasticity curve. Substitution may occur in patches.
To make a higher personal income you need to specialise, and that means your contacts come from a smaller global pool that you're more likely to have to travel to meet. So as incomes rise people are obliged to do more of it despite its increasing unattractiveness with respect to leisure alternatives. It's a hard call on as to whether commuting and business travel is an "inferior" good (negative income elasticity -- willing to pay less as income rises).
Does substition exist?
Thought experiment/sanity check: if we had universally accessible and perfectly immersive VR meeting and presence systems available (i.e. a huge drop in telecom prices), would business people, consultants and workers travel as much? Probably not, unless there was a free lunch involved. So these appear to be price substitutes, not complements, when you factor out income effects.
So we can at least dream of a case of substitution occuring, as an existence proof. In my mind I already have several more concrete existence proofs of substitution:
Much business travel is very mundane and once you've established an in-person rapport the only reason for continued physical presence is because today it's too darn hard to mimic a whiteboard and manic arm waving using NetMeeting. Even eye-to-eye sales can see substitution: you get telemarketing calls trying to sell you double glazing as well as door-to-door calls. Changes in the prices of oil and phone calls would change their relative demand. These may be complements and substitutes in different situations. But when? I don't know.
Price shocks to test cross-elasticity
Suppose we anticipate a two price shocks. The first is a massive universal decrease in the cost of connectivity because of cheap long-haul fiber and user-built municipal meshes. Everyone gets a zillion-dollar OC-192 into their home for ten bucks a month. This would probably greatly erode the desire to commute, reducing the attractiveness of central business districts and increasing the wish to live in smaller towns near to recreation and shopping facilities. So your billion-dollar fixed public transport network with a hub-and-spoke architecture might be a waste of money. Or in other words, y'all move to the Red States.
The second possible price shock is a calamitous increase in the cost of travel because of the hypothesised Peak Oil effect and lack of scaleable energy substitutes. (The jury's still out, you make up your own mind.) This will have a large negative income effect on society. This might greatly increase the demand for business travel as people strive to reclaim the former income levels. The effect on telecom depends on whether it turns out to be a Giffen good -- to me, an open question. The price effect of a substitute good (business travel) rising in cost might boost the demand for telecom. Who knows - it's hard to tell which effect wins. But the Blue States look like nicer places to live in this scenario. (If you're broke and homeless, at least you can sleep with the beautiful people.)
Conclusions
The separation of the "tele" from the "com" suggests we should be looking separately at the layers of the architecture. That means considering the information services separately from the bit haulage. What's the price cross-elasticity of going to NASCAR races with car-racing TV shows and video games? And will that have the same effect on connectivity demand? I suspect the answer is we know even less at this level of detail than we do at the aggregate level.
As Andrew Odlyzko also observed (p8), nobody's ever properly studied and documented the history of price discrimination in network industries.
There is no definitive treatment of the history of pricing in various network industries. Even some basic questions, such as the nature of the "just price" doctrine of medieval scholars, are still being debated.
Unfortunately, I've also yet to see a good model for how telecom and travel are related, which might cost us dearly in the future. This is despite it being the core motivator for (at no exaggeration) trillions of dollars of annual economic activity. It would be no surprise if we were equally in the dark about telecom and travel cross-elasticity.
If someone isn't working on a PhD thesis in this area, they should be.
In conclusion, (i) blanket statements about how bit movement and atom movement relate are unlikely to be valid, (ii) even given our low level of knowledge, there are likely to be significant (read: "billions of dollars") local substitutions as incomes and relative prices vary, even if globally aggregated data hides those (iii) the interesting problem is finding under what circumstances information goods substitute or augment the consumption of physcial transport, and, (iv) there's a lot of money riding on this.
Posted by Martin Geddes at 11:14 PMTrackBack URL for this entry:
http://www.telepocalypse.net/cgi-sys/cgiwrap/mgeddes/MT/mt-tb.cgi/373
Listed below are links to weblogs that reference Arranging atoms:
»
Digital Convergence from Blogcritics
Telepresence, timeshifting and the digital divide
[Read more]