Economic complexity

25 March 2010

The network effect of mobile interconnect as it pertains to the development of countries

All things in the world comprise a system, and every system in the world has a network effect. The variable is the strength of the bonds. The stronger the bonds, the harder it is to breach the network, and vice versa.

R Haussmann and C Hidalgo have recently shown that income inequality between countries is correlated to the economic complexity of the respective countries. The idea is that if you export wine, then your industries will be wine-related (farming, bottling). If you export textiles then your economy must have textile-related industries. Industries that support textiles happen to support a wider range of complementary industries than the industries that support wine. So it is easier for the textile producing country to jump to the production of other goods.

The final point is that complexity is self-reinforcing. The more complex your economy, the more complex it will become (barring external influences). The inference is that complexity will continue ad infinitum. Ironically, it has been shown that the more complex the system the more prone that system is to failing (the financial crisis is a vivid recent example.) The most appropriate comparison (in my mind) is software programming.

Programming is an iterative process whereby you construct a solution based on a series of “if this, then this, otherwise that”. The longer the code the more likely you are to make mistakes. If you wait until you’ve finished the entire job before debugging it takes you much longer to work out all the mistakes in logic and spelling (it’s interesting to note that in software programming both substance and form are important).

If you make a complex program live before you debug then it will fail 99% of the time, and depending on how it interacts with its surrounding environment, it can shut down the entire system. So debugging is important! Unfortunately the real world is not like the world of software programming where you can debug in a non-live environment. We have to debug in the real world, and sometimes the bugs only manifest themselves after several years.

Here’s a philosophical question: If everything is a function of a system, and every system must be designed, who designed the first system?

The Three Mile Island accident was caused by human error. The operators manually overrode the injection of emergency cooling water into the reactor. Who is to blame for the disaster? The operator that manually overrode the system, or the person that designed a system that allowed a manual override?

It all comes down to the question of “what would a reasonable person have done in that situation?” Punishment can only be meted to those that can be shown to have been malicious or grossly negligent (as measured against the values of the reasonable person). Who decides what the values of the reasonable person are? In some countries the death penalty is considered acceptable, in other countries it is considered unacceptable. There is only one shared value for all human beings. Survival. All other values are supportive. Some nations have evolved in such a way that the best means of survival was ensuring that lawbreakers were killed. Others have evolved in such a way that the best means of survival was not killing lawbreakers but rather imprisoning them for life.

So the trick to understanding what the supportive values are in any context is to understand that context.

Back to the real world.

The regulation of mobile phone operators can be used as a case study of how strong network bonds can be broken and as an illustrative example for reducing the economic inequality between countries.

First, some definitions:

  • On-net: Calls within the same network (Vodacom customer to Vodacom customer)
  • On-net rate: The rate charged to customers for On-net calls
  • Off-net: Calls between networks (Vodacom customer to MTN customer)
  • Off-net rate: The rate charged to customers for Off-net calls
  • Interconnect rate: The rate charged between networks for cross-network calls

Most countries followed a standard process when issuing licenses to GSM spectrum, i.e.: 2 or 3 licenses were issued.

Once licensed, the operators start building networks and acquiring customers. For the purposes of this illustrative example, let’s assume that the networks are Vodacom and MTN.

In order for customers of Vodacom to speak to customers of MTN, the networks must enter into an interconnect agreement which regulates what the networks charge each other for cross-network calls. If the interconnect rate is higher than the on-net rate, then it is impossible for operators to charge customers the same for on-net and off-net calls without running at a loss on the off-net calls.

Example: MTN

Assume the following:

On-net rate (MTN to MTN)                  R2/min (MTN to MTN)

Interconnect rate                           R3/min

If MTN were to set the off-net rate as equal to the on-net rate (R2/min) it would make a loss of R1/min (R2/min less R3/min = -R1/min). Therefore it must set the Off-net rate at a minimum of R3/min.

The above example shows that a high Interconnect rate allows the networks to give their customers a greater advantage for on-net calls by setting significantly lower on-net rates that are still profitable. This implies that the more customers a network has, the more attractive that network is to potential new customers (all things being equal).

Example: Vodacom vs. MTN

Vodacom                  MTN

Customers                           10,000,000                  5,000,000

On-net calls                  R2/min                           R2/min

Off-net calls                  R3/min                           R3/min

When a potential new customer looks at the above information, he/she realizes that it is 33% cheaper for her to phone most of her friends if she is on Vodacom’s network, because (on average) 66% of her friends will be on Vodacom’s network. Therefore her monthly bill is likely to 33% cheaper with Vodacom than with MTN. Therefore she is more likely to choose Vodacom (assuming price is the most important factor in her decision).

The scenario above gives the operator with the largest market share a self-reinforcing advantage in the market place (all other things being equal). The bigger Vodacom’s market share, the more customers they will attract, the more their market share will grow, the more customers they will attract, etc.

Let’s call this the “Interconnect Bond”. It is very weak in the early stages of the market, but as the market grows it becomes exponentially stronger[1]. The Interconnect Bond is very powerful and virtually impossible to break through normal market forces. The only means of breaking this bond is for a smaller operator to willingly take losses on cross-network calls in order to erase the price disadvantage for consumers.

Most developed countries have recognized the existence of the Interconnect Bond and regulators have stepped in and forced operators to reduce the interconnect rate to equal or below the retail rate for on-net calls, thereby allowing the smaller operators to charge the same for off-net calls as for on-net calls.

However, it is not quite so simple! The symmetrical dropping of Interconnect Rates resulted in a significant drop in revenues for the smaller operator.

Example:

Assume each customer makes one call per month for 1 minute

  Vodacom   MTN   Total
           
Interconnect Rate 3.00   3.00    
Customers 15,000,000 75% 5,000,000 25% 20,000,000
Total minutes 15,000,000   5,000,000   20,000,000
Outgoing minutes 5,000,000   3,750,000   8,750,000
Incoming minutes 3,750,000   5,000,000   8,750,000
Net incoming (outgoing) minutes (1,250,000)   1,250,000  
Net Interconnect Revenue (Cost) (3,750,000)   3,750,000  

Because MTN has fewer customers than Vodacom, it must, by definition, originate fewer calls and receive more calls. Therefore it is a nett receiver of minutes and Interconnect revenue. In the example above MTN would normally receive R3,750,000 per month. If the Interconnect Rate was reduced to R2 then they would receive R2,500,000, 33% less than previously.

This is obviously detrimental to MTN. It is already earning significantly less than Vodacom, and now Vodacom has had an immediate boost to its financial position (because its interconnect costs have been reduced by 33%). Of course, the advantage for MTN is that potential new customers do not choose their operator by virtue of market size, and MTN can now compete through the other available means (distribution, branding, product, etc).

Unfortunately, by the time Interconnect Rates are forced down by the regulator, the market is nearing saturation and the only potential customers for MTN are existing Vodacom customers. Naturally it is much more difficult and costly to convert someone that is already a Vodacom customer as opposed to someone with no existing mobile phone. (Mobile service providers are extremely sticky, mainly due to the non-portability of your phone number, hence the introduction of number-portability regulation in many countries).

Therefore, in order to alleviate the seemingly unfair financial advantage that has been conferred upon Vodacom, the regulator introduces asymmetrical Interconnect Rates, whereby the largest operator is forced to drop its interconnect rate to equal or less than its on-net rate, whilst the smaller operator may retain the higher interconnect rate, thereby leaving it’s interconnect revenue unaffected, and also allowing it to charge lower off-net rates than the larger operator. The asymmetrical rate regime remains in place until both operators are deemed to be equal in market share at which point a symmetrical rate regime is imposed.

What happens if Vodacom once again regain their dominant market share position through better marketing, distribution, products etc? Do you re-impose asymmetrical interconnect? I would think that gives the wrong incentive to the companies, i.e.: rather don’t work hard because if you succeed then the government will force you to concede your lead.

For now, let’s ignore the question as to whether the above process is fair. This scenario is a microcosm of the advantages that developed countries have over developing countries. The Complexity Bond of the country’s economy replaces the Interconnect Bond in the case of the mobile networks, and without regulation (asymmetrical regulation in favour of the developing country) the gap cannot be closed.

Of course, once the gap is closed then the Complexity advantage is removed and the countries will compete on a level playing field.

Jared Diamond postulates in Guns, Germs and Steel that the emergence of Europeans as the most successful race can be traced to geographic advantages (he won the Pulitzer and Aventis prizes for his efforts so one must assume he did his homework). That implies that the subsequent power they have acquired was a result of a lucky break at the start of the race. The argument for closing the inequality gap can therefore be the righting of a previous random coincidence that inferred a self-perpetuating unfair advantage.

What happens if the gap appears again purely because the inhabitants of the respective countries are more productive? Who knows, but at least you’ll have a fair starting point.


[1] Metcalfe’s Law