Traditional economics has extensive theories that purport to explain how companies grow. One drawback of these theories is their limited explanation of how companies interact and influence one another in the marketplace. Traditional economics tend to model marketplace interaction in three ways.
One, it assumes perfect competition between firms. In this model, companies have roughly equal influence in the marketplace, and can independently set its own prices. On the other extreme, traditional economics assumes the existence of a monopoly, where only one firm exists in the market. The third model that traditional economics offers is that of oligopoly, where a few firms dominate the market.
These models, while useful have their weaknesses. In the real marketplace, perfect competition and monopolies are relatively rare. While oligopoly is not uncommon, it is nonetheless true that in many real instances, markets are characterized by many different sized firms, influencing each other in a myriad of different ways, a situation that the traditional economic theories do not generally treat. They also fail to explain how firms in the real marketplace come to be of different sizes in the first place, in other words, they fail to explain the varying company growth rates.
There are economic thinkers who believe that there are fundamental laws that govern the way companies grow. Also academic research has suggested that irrespective of industry or sector, the growth of firms takes a precise mathematical form, which tells us what kind of distribution of small, medium and large companies to expect.
For instance, in 1996, a physicist by the name of Gene Stanley, working with an economist by the name of Michael Salinger and other colleagues at Boston University, Massachusetts, USA looked at the growth rates of all publicly traded manufacturing firms in the US between 1975 and 1991, a dataset encompassing about 8,000 companies. They found that the probability of a company having a certain growth rate decreases, as the growth rate increases, and these changes follow a mathematical relationship known as a power law, which tend to appear often in physics experiments that study phase transitions (the point where matter changes from one state to another).
A power law is a mathematical relationship between two variables such that a change in one variable leads to a large change in the other variable because the change in the other variable is raised to a power (Like 2 raised to power 4, which equals 16. If 2 is changed to 3, 3 raised to power 4 becomes 81, a big jump from 16).
The work of Stanley et al was backed up by independent studies carried out by Robert Axtell of the Brookings Institute, a highly influential American think-tank. He built an agent interaction model (such models are very popular in statistical physics), where the agents represented workers trying to improve their lot, like real workers in society.
In the model, workers are free to come together to form firms or to go solo, though there were incentives built into the model for workers to come together to form companies. When the model is run, the distribution of sizes of the companies that are formed follow a power law, just like what Stanley et al found in real life publicly traded manufacturing firms in the US. Robert Axtell would further validate his own work by looking at the statistics of 20 million US companies in 1997.
The distribution of firm sizes in this dataset also follow a power law. While no one is claiming that this is a perfect model of company growth, it does seem to be able to make realistic predictions about the statistical properties of firms. The model also captures another feature of the real business world that traditional economic theories do not acknowledge and that is the high turnover of firms (i.e. firms coming into and going out of existence).
Physics models are not only capable of realistically modeling individual company growth, they are also capable of modeling how best companies in an industry could go about forming business alliances. Companies often have need of forming alliances, especially in technology-based markets.
A typical scenario is when an industry has evolved competing technical standards, usually two, and where for some appreciable time, there isn’t any clarity about which standard will win out. Companies will have to balance a number of competing interests in order to decide which standard to back, for example, like which standard is likely to win in the long run, not being in the same alliance with a major rival, etc.
The US mobile telecoms market of the late 1990s, early 2000s was a striking example of the scenario described above. At the time, two rival mobile technology standards, the Code Division Multiple Access (CDMA) standard and the Global System for Mobile Communications (GSM), which was based on the Time Division Multiple Access (TDMA) technology were battling for supremacy.
Players in that telecom market backed one standard or another. Some major players that backed the CDMA standard included Verizon and Sprint. Other major players that backed GSM included American Telephone & Telegraph (AT&T) and T-Mobile. So if you were a T-Mobile subscriber, you couldn’t talk to Verizon and Sprint subscribers. This mess came about because the US, more than any other country, tries to allow market forces determine such outcomes.
Sometimes it works well like in the computer operating systems market where Microsoft Windows emerge as the de facto standard, sometimes it doesn’t, like in the mobile telecoms market of the time. By contrast, Europe essentially legislated GSM to be their standard. This probably a played a crucial role in helping GSM win the ultimate battle for the global “Second-Generation” (2G) mobile phone market.
For the US’ major players and other smaller operators, picking which standard to back would have been no easy feat. Had they had access to the physics models, we shall be describing next, perhaps they would have had an easier time of it.
One such model, devised by political scientist Robert Axelrod of the University of Michigan in the United States, is inspired by the condensation of gas particles. The condensation of particles into discrete clusters is a common phenomenon in physics. When companies replace gas particles, they are condensing into discrete alliances.
Each firm is like a particle with an individually tailored force towards every other particle. In condensing into various alliances, the companies are looking for the most stable configuration, which is something statistical physicists have observed gas particles do. Axelrod and his colleagues assigned different energy levels, with the most stable configuration being the one with the lowest energy level. The most stable configuration being the one with the lowest energy state is known to occur in physics as well.
So does the model work? To test it, Axelrod and co modeled a situation that occurred in the early 1980s concerning the standardization battles around the Unix operating system. At the time, there had been hundreds of different Unix flavors but eventually two major standards emerged and nine major companies were deciding which alliance to belong to.
There are 256 ways to split nine companies into two groups. The configuration that the model decided had the lowest energy level, therefore the most stable came very close to the groupings that had actually emerged back then. The model only assigned one company to the wrong group. The probability of getting that result by pure chance is 1 in 15, strongly suggesting that the model works.
The model doesn’t only work for assigning companies. It can also work for countries taking sides in a war situation. To test this situation, Axelrod and co tried to see if the model could correctly predict which alliance the European nations that took part in World War 2 decided to join.
There were two alliances; the Allied forces headed by Britain and the Axis powers headed by Germany. Seventeen European nations took part in World War 2. There are 65,536 ways to divide 17 countries into two groups. To help the model decide, the researchers fed the model data on the interactions between each pair of countries on the basis of six factors, each derived from the political, economic, and demographic situation in 1936. The factors are ethnicity, religion, territorial disputes, ideology, economy, and past history.
The model predicted the correct alliance for all but two nations, Portugal and Poland, both wrongly placed in the Axis group. The probability of getting this kind of result by chance is 1 in 200. Even the two nations it got wrong are understandable once you are familiar with the historical circumstances. Portugal didn’t really pick a side, it was formally neutral like Switzerland and Sweden, though it had a defense pact with Britain, making it less neutral than either Switzerland or Sweden.
In 1936, Poland was equally antagonistic towards Germany and the Soviet Union, with both countries in different groups. Ideally, it wouldn’t have wanted to be in a group with either of these two but since there were only two groups, and each of them was in a different group, it had to be in a group with one of them and the model chose the wrong one.
The error is even more forgivable when you consider that the model was using data from 1936. You might recall that World War 2 started in 1939/40. By then, Poland was much more antagonistic towards Germany than towards the Soviet Union. When the model was run using 1939 data, it got all nations correctly. The probability of getting this result by chance is 1 in 3,000.
In this series, we have been looking at how physics-based models have shown their usefulness in modeling social phenomena. These examples of how physics models can tell us something of how workers come together to form companies, companies coming together to form business alliances and nations forming war alliances is further proof that physics has the potential to illuminate our understanding of social relationships or in the case of war, antisocial relationships.
Bibliography
- Ball, Philip. 2005 Critical Mass: How One Thing Leads to Another. London: Arrow Books
