Professor Murmann's Blog: Evolutionary Economics—The State of the Science

Evolutionary Economics—The State of the Science

This is a talk I gave at a conference New Perspectives on Telecommunications and Pharmaceuticals in Europe and the United States: Conference on Evolutionary Economics:
Conference Program

Good morning. Let me give you a quick road map of my presentation. First, I will discuss where we are in terms of evolutionary economics, beginning with Nelson and Winter, 1982, the key book in this literature. Then I’ll provide a quick review of the ideas behind evolutionary accounts, laying out the requirements for a valid evolutionary explanation. I’ll follow this with a discussion of recent trends in the literature over the last six or seven years, addressing what I believe to be some of the key outstanding issues that should be addressed by the evolutionary perspective. Finally, time-permitting I’ll speculate a little bit about how one can make economics more an evolutionary science, and about what can be done to make evolutionary ideas more accepted.

First, I will talk about the position of Nelson and Winter’s book, An Evolutionary Theory of Economic Change, 1982, in the literature on evolutionary economics. One might say that all of evolutionary economics is just a footnote to Nelson and Winter. Let me try to establish for you how one might arrive at this conclusion by conducting a comparative citation analysis of key contributions in evolutionary economics.


If you look at how many citations each article in Volume 5 of the The American Economic Review (1982) has received on average since 1982, you will find that each article has been cited by 30 other articles. Nelson and Winter’s landmark (1982) book,  by contrast,  has received 2,020 citations in the same period (The source for this information is ISI Web of Citation Database). Cohen and Levinthal’s (1990) article on Absorptive Capacity places second in this ranking with 484 citations, while Tushman and Anderson’s (1986) piece on Competency Enhancing and Competency Destroying Innovations has received 425 cites. The famous article by Brian Arthur (1989) Competing Technologies, Increasing Returns, and Lock-in by Historically Small Events has received 420 cites, Tushman and Romanelli (1985), 371; Cohen and Levinthal (1989), 247. These comparative figures give you an indication of how central the work of Nelson and Winter is, not only in the evolution of economics, but in terms of social sciences in general. What Nelson and Winter did was to combine and recombine existing ideas, and formulate them into a very robust framework that sustained the case for evolutionary economics for twenty years.


Next I’d like to lay out some of the key requirements for providing an evolutionary explanation of how the telecommunication or pharmaceutical industries have developed. First, you need a unit of transmission. In evolutionary economics the unit of transmission typically is seen as routines or standard operating procedures. Routines play a role analogous to genes in biological evolution.


Second, you need sources of variation. For an evolutionary account to be persuasive, you want to hunt for the different ideas, practices, and products that the firm tried over time, including those that did not succeed. These can be innovations in technology; they can be innovations in procedures. So, as far as variations are innovations, you may have a new CEO of a company who establishes a new vision, and then promotes people from within the organization to try out new things.


The third element necessary is a retention system, i.e., a mechanism that retains those variants that work better. Of course, in biology it is information in the form of genes. Genes are the retention mechanism that ensures that the blueprint for how to mold a human being gets passed on from one generation to the next.


You also need a mechanism of transmission. And the mechanism of transmission in economic evolution is some kind of social interaction. Often it is imitation. A firm sees another firm doing something and tries to imitate it. Similarly, individuals see somebody else coming up with a new idea, and try to imitate it. This is how you can get routines or practices transmitted from one person to the next, from one human generation to the next.


Fifth, you need a process of transformation. What you need next is a mechanism whereby some trials, or some variants are more successful than others. Such a selection process allows some variants to grow or multiply while others contract and sometimes even go fully out of existence. This process of transformation in evolutionary economics is often measured by the different rates of adoption of routines in the relevant population of entities.


Finally, if you’re trying to explain, for instance, why the German dye industry looked very different over time from the British dye industry, you need to specify some source of isolation. If the individual firm and the individual entities are able to intermingle freely, you won’t see clear differences. If you’re trying to account for why a firm or a given national industry looks so different, you somehow have to establish that there is social boundary, for example, insurmountable social distance, between the two populations. The measure of a social distance can be many different things, such as a national boundaries that fosters different legal systems.


There is one thing I really want to leave you with. In evolutionary economics what matters is not intentions, but consequences. Saying that the explanation of why Microsoft is great is that they had Bill Gates does not work as an evolutionary explanation. The reason why things stay, why things get promulgated into the future is because they work. Things that work draw more resources or more people who want to adopt these ideas and so the community grows over time. Intentions by themselves don’t explain why we see social structures continue over time.


It is useful to think of selection as a filter. On the most abstract level it is an information filter. In evolutionary economics the selection criterion typically is profits. Those firms that make a lot of profits have more money to invest in new plants and new R&D projects. They will prosper. Those firms that are not profitable or make losses have less to invest. If they are unprofitable for a long time, they will simply be wiped out.


The figure above of firms with different commitments to R&D investment shows that the firms which did R&D have become much, much, bigger over time. They had successful products and were able to attract and accumulate capital resources. Their success created a positive feedback loop.


For an evolutionary argument to work, you always need to have at least two levels of analysis—a population and some individual entities that form that population. Simply describing growth stages in the development of a firm is not an example of evolution. This kind of development or change process does not constitute an evolutionary explanation because it fails to specify two levels that are required for a successful evolutionary explanation.


Now let me show you how you can construct an evolutionary argument of how an industry develops over time. What you need is a group of firms at time T1. In this chart, I use yellow, green and blue symbols to represent three different categories of firms and related activity. They could be, for instance, activities of a firm that focus on microelectronics, those practices that focus on software, and those practices that focus on, let’s say, making fruit products. At time T-1 you take stock of all the different practices that go on in the firm and you classify them into three categories. And let’s say in this case that 85 percent of the practices fall into the microelectronics category, 10 percent fall into the software category, and 1 percent in food. Now you have to specify some mechanism by which the firm internally selects whether or not a particular practice is expanded or contracted over time. Oberserving the firm again at time T-2, you can construct an evolutionary account of how the frequency of different characteristic practices changes over time within a certain population of firms.


Let me summarize the key points so far. Populations, not individuals, evolve. But a firm can be modeled as a population of individual entities. Strategic initiatives within a firm, for example, could be seen as individual entities. Robert Burgelman (1991) has written a wonderful paper on inter-organization ecology, in which he depicts a firm as a population of entities. He looked at strategic initiatives within Intel. Some of you may know the story:  Intel used to be a DRAM memory company. Over time, it became focused on logic circuits, the main processor chips that are now powering your PCs. Bob Burgelman showed that for the longest time the people in the company believed they were a memory company. However, if you actually look at the allocation of resource and R&D spending, the firm over time became more and more focused on making Intel microprocessors and not DRAM chips. After a while, Intel’s top management decided to actually go out of the memory business. But this was a strategic decision that came after the firm, in terms of its actual practices, had already become a microprocessor company. What Burgelman uses as the selection device here is the competition for resources and senior management’s attention among Intel’s different businesses and initiatives. What happened was that the people who ran Intel’s microprocessor business were able to convince top management to give them more and more resources over time. Without having made a conscious decision to transform themselves from a memory chip to a microprocessor firm, Intel became a company that essentially left behind its original business.


In models of industry evolution need to have some stochastic element because they should account for the fact that technology development typically is not predictable in terms its consequences. And so a lot of what creates the variation in evolutionary economics is the stochastic process such as technological innovations, which generates the variants from which some variants can then be selected.


The final point with which I want to leave you is that of course there is a big distinction between industrial evolution and biological evolution. In biological evolution the processes that generate variation, namely random mutations of genes, are very distinct from the selection processes in biological evolution. That is, of course, not the case in social evolution in general and industrial evolution in particular. The variants in industrial evolution are not generated in completely random fashion. People think about making a better microprocessor. They want to develop a technology which could potentially do “X.”  Often it turns out that when you’re trying to develop something it turns into something very different than what you anticipated, and that is, of course, the random element.


Let me stop here. Thank you very much.