Professor Murmann's Research Blog: The Stanley Reiter Lecture 2005

The Stanley Reiter Lecture 2005

On January 26, 2005 I delivered the Stanley Reiter Award Lecture. The Reiter award is named for Stanley Reiter, Charles E. and Emma H. Morrison Professor of Managerial Economics and Decision Sciences at Kellogg. It is presented to a Kellogg faculty member whose paper is judged by a panel of Kellogg professors across disciplines to be the best paper written in the preceding four calendar years. I received the award for my book Knowledge and Competitive Advantage: The Coevolution of Firms, Technology and National Institutions. You can also read the text of the lecture by clicking on “More” button or by downloading it as a Word file.  Alternatively,  you can watch a video (58 minutes) of the lecture with Real Player here: Lecture Video. If you watch the video, you should download the Slides that I presented during the lecture but which are not visible in the video.


In 1958 Stan Reiter and Jonathan Hughes published an article with the unassuming title, “The First 1,945 British Steamships.” The purpose of the article was to show how statistical tools and theories developed during the first half of the 20th century would make it possible to generate evidence about what had happened in the past that formerly had been beyond the reach of the traditional toolkit of historians. What was radical about this paper, in my view, was not that Stan and his conspirators went against existing views of professional historians—after all, for thousands of years historians have attacked each other’s writings and called for revised accounts of past events as well as of the causal processes that brought about these episodes in human affairs. Stan & Co.‘s “The First 1,945 British Steamships” was radical in advocating the use of statistical methods to generate new data about what had actually occurred in history.  For anyone trained in the social sciences today, the benefits of using statistical theory and econometric tools such as multivariate regressions are obvious. Stan & Co. were the first (or at least among the first) to realize that multivariate regressions could be an extremely powerful tool in the hands of economic historians.

The mainstream of the discipline at the time, however, seems not to have embraced Stan & Co.‘s methodological vision of adding statistical methods to the toolkit of economic history. Given the contribution of the paper, the obvious outlet for “The First 1,945 British Steamships” would have been the Journal of Economic History.  Instead, it appeared in the Journal of the American Statistical Association, even though the techniques Stan & Co. were using must have been basic for any professional academic statistician at the time. Reading the paper today, this placement appears quite ironic because the paper is so deeply steeped in a historical methodology. Stan & Co. made use of all of the traditional methods of historians to get as accurate a picture as possible about how and why the British merchant fleet developed as it did.  The theoretical reason why this empirical exercise was important is this: Knowing how quickly British shipping grew would make it possible to estimate the growth of British overseas trade, which in turn would provide clues about the growth and sectoral changes in what was at the time the leading economy in world.  A theory of economic growth that claims to be any good should certainly be able to explain the path of the British case!

The statistical analysis Stan & Co carried out in “The First 1,945 British Steamships” is merely the tip of the iceberg of other pieces of information and methods of inference they used to arrive at their conclusions. The way Stan and Jonathan conducted their empirical research reminds one of Sherlock Holmes and Dr. Watson: Every lead is pursued, evidence of all shapes and forms that might help solve the puzzle is put in relationship to one another to construct an overall picture, and every possible objection to their interpretation of the facts is taken into account to prevent an embarrassing fall in the court of scholarly opinion. The regression analyses Stan & Co. conducted in the paper were carried out with deep knowledge of the context. When the conclusions they reached about the development of the first 1,945 British steamships differed from the prevailing notions among professional historians, they went back to individual eyewitness accounts to corroborate the historical validity of their conclusions.

Stan & Co did not seek to overthrow the traditional tools of historians; they merely wanted to spread the use of quantification and statistical methods as an additional powerful tool potentially capable of generating a large number of new insights. 



Despite this strong potential, there seems to have been equally strong resistance among the mainstream economic historians who lacked Stan & Co.‘s statistical skills.  Two years after publishing “The First 1945 British Steamships,” Stan, Jonathan Hughes, and their Purdue colleague Lance Davis were given the opportunity to explain their new quantitative methods to the profession at large in the Journal of Economic History. Their paper, “Aspects of Quantitative Research in Economic History,” provides many clues about their vision for research and how it was perceived at the time by the mainstream. Stan & Co wrote:

An examination of the literature of economic history indicates that, while the qualitative stream in the discipline has usually been larger, there has been from the earliest times a significant and respectable flow of quantitative work [...] None the less, the total amount of work in the field is small. Why? Is it because quantitative work is unrewarding? We think not. (pp. 541-2)

They also wrote:

In brief, the logical structure necessary to make historical reconstructions from the surviving debris of past economic life essentially involves ideas of history, economics and statistics. The offspring of such an act of interdisciplinary miscegenation calls for a name worthy of it; at Purdue the resulting discipline has been labeled “Cliometrics.” (p. 540)

[Clio, by the way, is the name of the muse of history.]

We are not suggesting in this paper that there is to be “new” economic history which will render non-quantitative economic historians technologically unemployed. It should be obvious that we regard ideas from statistics and data-processing as natural aspects of problems of historical study. It should also be obvious that the historian’s special knowledge and viewpoint is essential to the useful employment of quantitative methods. Our main point is that modern statistical techniques and computing equipment make possible the intensive exploitation of a vein of historical materials that was perforce only little worked in the past; and that if even a few economic historians would take the time to learn even a little of these new techniques, the 1960s could easily prove the most productive years in the history of the discipline. (p. 546)

It is not difficult to understand and share Stan & Co.‘s excitement about the scientific progress the development of computers and multivariate statistics might make possible. After all, big advances in science always came about not because a new generation of scientists was smarter than the previous one, but because new instruments such as the telescope and microscope made it possible to obtain data that previously had been unavailable.  Using computers to run multivariate statistical analyses promised to bring the power of the experimental design to non-experimental settings. Now it was possible to determine the incremental effect of one variable by holding all other causal variables constant—not through experimental manipulations but through mathematical ones.  Seeing this possibility, who would not be wildly enthusiastic about the breakthroughs that computers and statistical theory seemed to offer the non-experimental sciences?

In the 1958 and 1960 papers, Stan & Co. refrained from any polemic against the mainstream of the profession. Instead, they tried to make their case by doing empirical research that demonstrates the value of adding quantitative analyses to historical scholarship and by explaining the power of using statistical techniques as an additional tool in the historian’s toolkit. Only at the end of their paper “Aspects of Quantitative Research in Economic History” did they take off their gloves to express a stark judgment about the status quo: 

On the other hand, if the discipline chooses to remain completely in the literary tradition, we can see small hope for anything but a continual rehashing of the already existing sources and a continuation of the century-long cleavage between economics and economic history—a cleavage that should soon disappear if the economic historian is able to provide the economists with new data and new interpretations of the process of economic life.  (p. 546)

Most of the time when young scholars advocate an overthrow of the status quo, the revolution never comes to pass. Stan and Co., however, successfully instigated a full-scale revolution in American economic history. They were able to inspire a new generation of doctoral students to adopt their quantitative methods.  Once doctoral students had shifted from the narrative tradition to the quantitative methods, it was only a matter of demographic time until quantitative economic history became the mainstream.

In fact, the revolution that Stan & Co. initiated in economic history appears to have devoured the very vision they had in mind: Cliometrics as a combination of history, economic theory, and statistics. By the third generation, Cliometrics seems to have become entirely about statistical techniques and economic theory. The deep contextual understanding that was so central in being able to construct compelling statistical analyses in Stan & Co.‘s work on “The First 1,945 British Steamships” has been lost in the process of importing ever more advanced statistical methods into the discipline.  A parallel development occurred in organizational sociology and macro organizational behavior, two fields with which I am intimately familiar.

30 Years Later

When I arrived in graduate school 30 years later in 1991 to study for a Ph.D. in the social sciences, historical scholarship had almost completely disappeared from the research agenda.  We were socialized to believe that what historians do is both unsystematic and completely atheoretical: in short, the exact opposite of what a good social scientist would aspire to engage in. I spent my first couple of years like anyone else running econometric analyses on large data sets whose underlying empirical reality I knew little about, learning all the reasons why what I was doing was so much more sophisticated than what those storytelling historians were engaged in. We were seeking the Newtonian laws of the social universe while the intellectually feeble Ph.D. students in history would at best learn how to become journalists of particular long-gone times and places, work totally useless for managing the affairs of today and tomorrow.

My book Knowledge and Competitive Advantage: The Coevolution of Firms, Technology and National Institutions is testament to how fundamentally I changed my view about the value of history for the social sciences in general and the study of how industries and firms develop in particular. So let me briefly tell you how this change of heart came about and why I think the historical evidence I painstakingly assembled is the chief scientific contribution of my book.

Working on the question of how different types of technological innovations would affect the development of industries, I stumbled on Hugh Aitken’s books on the history of radio; Thomas Hughes’ history of the development electric power networks in Chicago, Berlin, and London; and Walter Vincenti’s work on the development of airplanes and the discipline of aeronautical engineering.  It was simply not true that historians were merely telling one damn fact after another. The best historians don’t shy away from abstractions and theory. Aitken, for example, in his history draws heavily on role theory from sociology, Hughes on general system theory, and Vincenti on evolutionary theory that my book builds on and tries to develop a little further.  Joel Mokyr, who is among us today, develops in his recent book The Gifts of Athena an abstract theory of different kinds of knowledge and then uses this theory to explain why and where the industrial revolution occurred.  I also found that the thick descriptions—to use a term coined by the anthropologist Cliffort Geertz—historians were using would, unlike a regression table, make it much easier to think up and try alternative theoretical explanations for the phenomenon at hand.  I also found that the field of history includes an institutional feature that is important for any good empirical science.  Historians in their quest for professional recognition compete over who comes up with the more accurate description of what actually happened in the world.  Among the contributions of Stan & Co.‘s “The First 1,945 British Steamships” is precisely that it provided a more accurate description of the growth of British overseas shipping.

One of the most prominent questions of the management discipline already in the late 1980s and early 90s concerned how firms could acquire a sustainable competitive advantage. I thought that this was a fundamental question worthy of serious attention.  When you reflect a moment on this question you realize it has a clear temporal component.  So ideally you want to trace firms over long periods of time. You also want to trace all the firms that enter a particular industry to have a systematic view of the contest that eventually will produce winners and losers.  Given my reading of the history of electrification, I had a strong suspicion that institutional differences of countries would play an important role in which firms would gain and which would lose competitive advantages in global industries.  In my view there was no obviously compelling model of how national institutions would influence the competitive fortunes of firms that started up within their borders.  I realized that the historian’s method of carefully studying and organizing the empirical reality of the phenomenon would probably be the right way to go to come up with a good proposal for the key variables of a model.

But this was not enough to make me want to use a historical methodology that was so uncommon in my field. Theoretical concerns were what really pushed me towards a historical method. Taking my doctorate in a management department gave me the opportunity to read widely. The two writers who impressed me the most and who pushed me into a historical direction were Herbert Simon and Donald Campbell. Simon’s The Sciences of the Artificial and three of Campbell’s articles on evolutionary theory are my all-time favorite readings.  What appealed to me in Simon and Campbell was their vision that the different fields of science had to be consistent and build on one another.  Both also had a broad view of what science was about. Simon, for example, tells us on page 1 of The Sciences of the Artificial,

The central task of natural science is to make the wonderful commonplace: to show that complexity when properly viewed is only a mask for simplicity; to find pattern in apparent chaos.

There were six intriguing ideas in Simon and Campbell that I deal with in my book. [Note: From here on the lecture is not based a text that I read but rather an extemporaneous speech that I later described from a video and edited a bit so it would flow better in written form.] I want to briefly mention those six ideas and then I’m going to show you what I do with these six ideas in my book.  So here are the two readings that have influenced me most deeply in my work. 



The first idea—and this comes out of Donald Campbell—is that evolution is a general process.  You all know a story like this one from high school biology: grizzly bears have brown color when they live in British Columbia, Canada, and then when they move up north, perhaps trying to find new food, they have white fur. Why?  Evolutionary biologists give the following kind of answer. Let’s say the grizzly bears living in British Columbia are all brown. However, there are always random mutations in the genes that code for fur color in grizzly bear babies.  This means that bears with all kinds of different colors are born: one grizzly bear is green, one is grey, one is white, etc. If these different colored bears move up to the North Pole, which bear is going to survive?  The grizzly bears, which by chance are going to have white color, and will not be as visible to predators.  As a consequence, over time, the population of brown grizzly bears moving to the North Pole would all become white grizzly bears. Not because they design their own clothes to fit the environment, but simply because natural selection is changing their population.



Now the genius of Campbell was to say, My god, this process is a general process.  The biological case is merely a special case. That was Campbell’s great contribution.  Campbell not only theorized about this.  He came up with super examples—which he wrote up in a chapter honoring Karl Popper’s evolutionary epistemology.  Campbell had better examples than anyone else.  Campbell articulated very clearly that the theory of evolution is an abstract theory.  All you need is three processes—a mechanism for introducing variation, consistent selection pressures, and a mechanism for preserving particular variations.  In the case of the population of grizzly bears moving up north, random mutation is the source of variation, selection brings about the differential survival of white bears, and the DNA of the animals which remembers white fur from one generation to the next—ensures that the children are also going to be white.



The second and third key idea that I deal with in my book comes out of Simon’s The Sciences of the Artificial. What a person cannot do, he or she will not do no matter how strong the urge to do it.  The third idea is, in the face of real world complexity, the business firm turns to procedures that find good enough answers whose best answers are unknowable.



For the purposes of my book I’ve transformed these ideas a little bit—completely consistent with the writings of Simon, March and Cyert—to apply to firms: What a firm cannot do, it will not do—no matter how strong the incentives for doing.  Also, in the face of real world complexity, the business firm develops standard operating procedures to deal with most decision making situations.



The fourth idea in The Sciences of the Artificial is that the evolution of firms and of economies does not lead to any easily predictable equilibrium, much less of an optimum, but is a complex process probably continuing indefinitely, and that process is probably best understood by an examination of its history.  This means that if you are an evolutionary theorist, the data by necessity is historical.



The fifth idea—found in both Campbell and in Simon—is that many phenomena display an hierarchical organization.  As you look at a simple example, please focus on the two left component dots. 



The basic idea is that when you look at the interactions on the component level, components only interact with a few others and not with all components.  Look at the two component dots on the far left of the slide above. These two components interact with one another, and then interact with all the other components not as individuals, but as aggregates.  I don’t have time to go into the details of this idea.  I have put together a simple representation to show you what I did with this idea in my book.



I said to myself, If I want to understand how products and services develop, how do I need to conceptualize the process?

If you look at the plans for products and services, they are themselves nested within firms, which are deciding what to make and what not to make.  These firms in turn are nested in an industry. Competition in the industry decides what firm will survive and what firm will fail.  Such industries are nested in national economies, and the country national economies in turn are nested in global economies.



Here is a crucial idea out of Campbell spelling out the concept of a hierarchy of selection processes. “It is important to recognize what are selection criteria at one level are but trials of the criteria at the next higher, more fundamental, more encompassing, less frequently invoked level.” When I read this the first time, it was not clear to me that this was the most important idea in Campbell. But after years of reflection on evolutionary theory, I realized the idea of a hierarchy of selection processes is crucial to making an evolutionary account of firms, industries and economies work.



Let me give a short overview what this hierarchical selection model means.  Picture a product.  A product initially is launched by a manager.  A manager decides which product to launch and which product not to launch.  (I would have really liked if the above picture showed many, many products and a manager picking out only one. I simply don’t have space for many products dots, so please imagine there are many.)  Which product that will be produced in a plant is determined by selection criteria of the most immediate environment of the product—the firm that is making it. (Remember the grizzly bear was selected to be white by the environmental conditions of the North Pole.) So the most immediate environment is the firm, and the next larger environment is the industry. You can conceptually make the environment ever larger and more remote until you reach the global economy, where again a product either flourishes or does not.

I realized from Campbell’s idea of a selection hierarchy that you can get the entire system to self-organize and become increasingly adaptive as long as the high frequency events are in the center and the lower frequency events are outside.  In other words, the environment must change less quickly than the focal population, which is trying to adapt to the environment.

The sixth idea I examine in my book comes from Campbell’s work on creative thought. As many of you know, Campbell was a psychology professor for many years here at Northwestern University. “The variation in the selection retention model unequivocally implies the greater the heterogeneity in volume of trials, the greater the chance of a productive elevation.  Unconventionality and no doubt numerosity are a necessary, if not sufficient, condition of creativity.”  The conventional wisdom about genius is that a genius has deep intuition and just sees the right answer.  Campbell fundamentally rejects this romantic view. For him, a large number of different ideas are generated on a subconscious level of the mind, some of which flow into the consciousness of the person.  A “Campbellian genius” selects from all the ideas in his or her brain those ideas which later on turn out to be the great ideas. Campbell argues that on the level on individual psychology it is numerosity that underlies genius: The person who works harder, the person who generates more ideas, is the person who is going to be the genius.



I realized in my book that this is an idea I could transpose from the level of the individual brain to the level of the individual national industry.  Those national environments where you have more unconventional trials will in the end come up with better products and flourish.

I thought these ideas were really exciting and that’s why I wanted to introduce you to them. You may say: Anyone can come with ideas, the crucial question is, are they true?   Do they really explain something about the empirical world?  What I’m now going to do now is walk you through some empirical evidence that I have in my book and let you be the judge of whether there is anything to the six ideas that I’ve just presented. 



Before I do that, I want tell you about two big lucky breaks that I had doing my work.  One lucky break was that Richard Nelson was at Columbia. Together with Sidney Winter he had created an evolutionary theory about economic change. Any evolutionary theorist needs historical data to test the theory.  So when I was trying to do a historical dissertation, Richard Nelson was very happy to support this venture and act as my advisor. He and I have had between 60 to 80 conversations concerning the material I present in Knowledge and Competitive Advantage. He deserves of lot of credit for the final product. 



I could have not done my study without a second person, Ernst Homburg. Altogether 60 people helped me with the book, but without these two people, this book would simply not exist. 

Let me tell you why the synthetic dye industry is great industry to study.  As I indicated before, I was interested in how national institutions shape the competitive position of firms over long periods of time.  So I needed an industry that starts at the same time in different countries, to be able to compare how national institutions have an impact. I began by looking at about eight different industries as possible candidates for this comparison.  For example, machine tools did not work because the firms were all small private companies, which meant there was no good public data there.  By contrast, chemicals had a lot of big public firms in the 1880s, and when you are a big public firm that becomes successful, you leave a long paper trail.  This means that it would be much easier to do research on the chemical industry.

By coincidence, I read a little footnote to a database on the dye industry.  Recall that doctoral students and assistant professors need good datasets above everything else.  So I said to myself, This person may have useful data and I should visit him.

What Ernst Homburg had was a long file drawer of little paper cards that he had put together starting in 1979 as part of a project on the history of dye making technology. For five years his group of four people collected any piece of information that they could find anywhere on the synthetic dye industry.  They organized their source material on index cards. Each individual firm had cards on which any piece of information related to the firm was recorded.  A card for a firm might have a note that the company exhibited dyes in 1860, another note that the firm appeared in a trade directory in 1870, and then still another that the firm showed products at an 1885 exhibition.  As you can see in this example, there is data missing from the years between the dates. But I realized that using my computer skills, we could create a firm’s life history from this raw data. And so together with Ernst Homburg, I pieced together the life histories of 379 firms that left any kind of trail from 1857 to 1914. This data, as I will show you later, is extremely powerful.

I should also say that the reason I selected synthetic dyes for my study rather than all chemicals is that the chemical industry has too many products and I wanted to keep the technology as constant as possible.  The first synthetic dye—and this is important for the idea of selection—was created in 1857 by a young student by the name of William Henry Perkin enrolled in the Royal College of chemistry in London.  Perkin did not want to create a dye, he wanted to synthesize quinine, a drug used for malaria.  He had huge incentives for finding a synthetic route to quinine because the vast British Empire needed quinine to protect traders and troops from malaria that was endemic in the tropics.  Perkin recognized that he had not made quinine in his test tube, but a potential coloring material to dye textiles. He entered the industry, and by 1862, the global market share looked like this.  British firms had 50% market share, French firms had 40% market share, German firms between 2-3%, and Swiss between 1-3%.



Here is a common expert prediction of what was going to happen to production shares in this industry.



“England will be the greatest color producing country in the world.”  Remember that until 1870, most dyes were natural dyes that came from India, Mexico, France, and Germany but not Britain. Given the initial lead of British firms in the synthetic dye industry and its perceived ability to replace natural dyes with synthetic counterparts, the experts predicted the Britain was going to be the greatest nation in the world in terms of dye production.



Here’s what happens by 1873.  German firms collectively have a 50% market share. By 1880 the market share looks like this. 



Germany has between 75% and 90% of the world market share depending upon whether you count German-owned plants in foreign countries.



Let’s look at some of the leading firms in those countries.  By 1913 the three largest German companies, BASF, Bayer and Hoechst, each had a domestic production share of 22% and a global market share of 20%, adding up to a global market share of 60%. (You are familiar with Bayer from taking aspirin.)

Now let’s look at the British leading firms.  Levinstein and Read Holliday each have a 30% domestic production share.  Notice that concentration happens in all these countries.  Together the two firms only have 4% global market share. Now we are up to 64% global market share if we sum across the top five firms. Finally, Schoellkopf and Heller Mertz, the biggest companies in the United States, have respectively 50% and 21% domestic market share, but as you can see on the slide, they have a very, very small global market share.

A puzzle arises when you are trying to explain the German dominance in the industry.  You might say that when one country dominates an industry, it’s obvious that all the demand must be in this one particular country.  What I’ve done in the book is say okay, if 95% of all the dyes go into textile coloring, let’s look at the size of the textile industry and estimate, by looking at the spindle capacity, the demand for dyes in the different countries.  In 1852 Britain has a spindle capacity that is four times as large as that of the United States, five times as large as that of France, 23 times as large as that of Switzerland, and 23 times as large as that of Germany.



By 1913 Germany dominates the synthetic dye industry, and spindle capacity looks like this.  In Britain capacity is still twice as large as in the United States, seven times as large as in France, five times as large as in Germany, and forty times as large as in Switzerland.  So clearly, concentration of demand will not explain German dominance in the industry because if that were the case, Britain should have run away with this prize.  One may then say it is not quantity of demand, but what matters is where the high quality demand is coming from, stimulating innovation. But when you look where high quality demand is coming from, you find it mainly in the silk industry of Paris and Lyon.  This is the high-tech part of the dyeing industry.  But again, France is not the country that is running away with the prize.

If you look at supply for the explanation, you expect that the key raw materials for making synthetic dyes would come from Germany.  But it turns out that until 1887, raw materials are shipped from Britain to Germany and to Switzerland.  Again, the explanation does not work.  What I do in my book is to argue that Germany overtook Britain and France in the synthetic dye industry because of the differences in national institutions of these countries. 

So let me give you an overview of what happens over time to the number of companies that participate in this industrial contest.



As you can see, although the British have the first firm, the number of British firms never goes up to more than 15.  In the French case, anyone who does any work in industrial organization economics might see a nice little shake out here.  There is a simple model that predicts that a shake out will happen if one or more firms get ahead of their competitors in terms of output and thereby lower their unit cost.  When I first looked at this that is, of course, what I thought myself, but that’s not always what’s creating the shakeout here.  In this case it is a patent ruling, giving one firm the monopoly on one dye—but this one dye is a precursor of most other synthetic dyes at the time—so the other firms are closed down by the police!  That is how you get a shake out in France. As you can see, knowing the details is essential to interpreting industrial dynamics.

Now in the U.S., we have a nice little shake out around 1883. Firms go up and then boom, half of the firms go out of existence.  Why is that shaking out occurring?  Because the United States abolishes tariffs on dyes, so now the Swiss and German dyes can come in and just kill half the population.  Now look at Germany.  In Germany, the number of firms increases until the 1890s.

Coming back to six ideas that I discussed earlier, here is a piece of data which is really essential, either confirming there is something to this evolution explanation or that this kind of explanation makes no sense whatsoever in this context. 



The number of entries in Germany is 116 firms, 63 in Britain, 16 in France, 47 Britain, 35 in the U.S. and 23 in Switzerland.  We see that the largest number of entries is in Germany.  These are new entries over the period of 1857 to 1914.  In the second column you see the firm exit figures for the period.  Not only does Germany have more entries, it also has more exits!  A key conclusion that I reach in my book is that Germany achieves global domination in the synthetic dye industry not only by having more firm entries, but also by having many more failures.  Think back to Campbell, numerosity and variety is what creates success.  If you look at the failure rate, the failure rate is above 71% in every country.



Now I want to zoom in and focus for a moment on an individual firm laboratory.  Theoretically, after academics figured out the chemistry in this, they realized that there are billions of possible dye molecules.  R&D scientists in these labs really have to think about, which one of the billions of possible ones should I try to synthesize?  Here is what the Bayer firm did in 1906. Its researchers synthesized 2,656 new dye molecules.  Of those 2,656, they only tested 60 on a larger scale. Of those 60, they introduced 36 into the market, and that’s actually over counting because some of these were just re-formulations of last year’s dyes.  So Bayer probably introduced 20 new dyes into the market that year. What we see here is an enormous weeding out process from the brains of the research scientists all the way to the market. If you have a firm and you make a product that nobody wants, you and your firm are going to be selected out.

Why did Bayer create so many synthetic dyes?  Although chemical theory allows you to narrow down the search process—at this time Bayer had about 350 chemists—you still need still need to synthesize a lot of dyes. The only two other companies that can compete with Bayer at the turn of the 20th century in terms of their R&D capabilities are BASF and Hoechst.  You may ask how did this competitive strength come about?  Did firms in 1865 tell themselves, if we have a big R&D lab, then by 1900 we’re going to clobber everyone?  No, people did not foresee large R&D labs as standard function of dye firms.  Let me tell you briefly how the Bayer R&D lab came about because this history provides some support for the evolutionary ideas I discussed earlier.

BASF and Hoechst are the firms that created the first R&D laboratories in history in the 1870s. Carl Rumpf, the son-in-law of Mr. Bayer, realized that these two firms were coming up with new dyes in ways never seen before.  Before this time, every firm in Germany was copying the dye innovation of French and British firms.  So Rumpf goes to the owner of the firm, his father-in-law Mr. Bayer, and says, “Look, let us hire some research chemists and let us try R&D out on a small scale.”  The owner says no.  Rumpf then says, “Okay, I’m going to hire two research chemists with my own money. I’m going to send them to the University of Strasbourg on a trial program, so that they can figure out how to synthesize synthetic indigo, the queen of all dyes, representing the biggest market of all natural dyes.”  He tells two recruits to spend six months at the University of Strasbourg, to solve the synthetic indigo synthesis, and then to bring the process back to Bayer so that Bayer can produce it.  Well, it took another 30 years to create synthetic indigo. But Rumpf brought them back into the company and the two young chemists created some novel dyes, and when the company owners realized my God, we can create better dyes than our competition, then Bayer hired even more chemists who focused on synthesizing new dyes.  So over time, gradually and incrementally, Bayer experimented with how to organize innovation processes in the firm.  To sum up this development, the large R&D laboratory at Bayer and at other firms in the industry came about by means of a trial and error, and trial and success process.

Now I want to show you a little how the national institutional context of science mattered for the development of corporate R&D labs. I cannot present all the information I provide in the book today.  I’m just going to give you a sample of it.  Let’s look at global share of organic chemistry publications. 



In 1852 Germany has 29% of the global share and France has 35%, so organic chemistry, in fact, is stronger in France five years before the birth of the industry.  Let’s look at 1862.  I’m going to show you France and Germany because it’s easier to see the pattern. In 1852, Germany has 38%, and France has 23%.  By 1877, Germany has 50% to 67% and France 15.2%.  (I put the German figure to lie between 50 and 67 % because the person who did the accounting was an American professor of chemistry who did not realize that a German language publication could be not only in Germany, but also in Switzerland and Austria. I’ve corrected for the fact that 67% German share is simply too high.)  The pattern you see here is France in fact is a little stronger in organic chemistry in the beginning of the industry and over time, just when the French dye industry is becoming weak in France, organic chemistry becomes weaker.  This parallel development comes about through processes of co-evolution that I spell out in great detail in the book.  The basic intuition is that as German and Swiss firms get strong, they lobby to get more professorships in organic chemistry at the national universities.  As the Germans become stronger, they organize a trade organization and become more effective at lobbying for changes in the chemistry discipline in Germany.  The French industry, severely decimated after the reduction of producers, did not have the ability to lobby the state to maintain strength in organic chemistry.

I want to quickly talk about what happens in the United States. In 1907, only 3.6 %  of all chemical publications are concerned with organic chemistry in the U.S.  It’s the complete reverse of the German situation.  In Germany it’s all organic, in America it’s all inorganic chemistry.  Why?  Because Mr. Rockefeller lobbied so that the universities in America would train people who could analyze and refine oil, and because the farmers of America made sure that they got a chemical discipline which would help them to grow food and raise animals.  The idea here is that over time the strong industries mold their environment.  They shape the environment, and in the book I detail how this process of shaping the environment happens.  In fact, it’s not just that they’re trying to perform well given the particular competitive environment. They are also trying to change the selection environment in their presence.

Now I want to show you a little more detailed data, which shows the transition of leadership from France to Germany.  This is out of a French publication.  It is a count of the number of aromatic organic chemistry papers.  Aromatic organic chemistry is the chemistry most closely aligned with dyes.  It is a subfield of organic chemistry, but something you have to master in order to create new dyes.


In 1864, just seven years after the start of the industry, 14% of abstracted papers in France are in aromatic organic chemistry, 35% of which are German.  In 1867, 38% are aromatic. As the industry becomes more powerful, a lot of other chemists say, I’m going to work on this hot subject.  We see the share of aromatic goes up to 40% percent of all chemical publications. But look at what happens to the German share.  It goes up to 96-97%.  In fact, there is a mistake in these figures because the French guy who put these figures together also did not realize that German language publications do not necessarily have to be from Germany.  They could be from Switzerland or Austria. But even if you take away 20% from the German figure, the pattern remains clear. German based authors are completely dominating aromatic organic chemistry.

In last five minutes of my lecture, I want to tell you a little bit more about what I do in the different chapters of book, after giving an introduction to all the material in Chapter 1.



In Chapter 2 I analyze how patent law and other institutions besides the national university systems impacted the performance of the national dye industries.  In Chapter 3 I examine how differences in a country’s institutions impacted how firms originating from a specific country operated.  This comes from Campbell, the psychologist who many of you know as a methodologist and the designer of new experimental designs.  I realized, after examining the life histories of 357 firms, that I would like to zoom in and observe the micro-causal processes on the level of particular firms.  I already showed you a little bit of information about the Bayer case.



What I’ve done here is to select two firms from each of these three countries:  Germany, Britain and the United States. I selected a firm that became very successful by 1914 and another firm that turns out to be failure. I have a winner and a loser from each country. Ideally you want to have firms that are almost identical in terms of their backgrounds, because then you see how individual agency matters.  In Germany I was able to select two firms in the same city—one is Bayer and one is Jaeger (a firm you’ve never heard of). Tracing them over time, Bayer becomes a global firm, and Jaeger doesn’t go anywhere.  Then I ask myself what differentiates the winners from the losers in these three countries?  It turns out all the winners have access to what I call the organic chemistry knowledge network.

You have to get access to a university laboratory, or you could hire students who know how to do organic chemistry well.  The British winner Levinstein,—who came from Germany originally—hires German chemists.  The American winner sends his son to Munich to study under Professor Bayer, and to bring organic chemistry knowledge back to the firm in the U.S. The American failure—the American Aniline Company—lacked knowledge of advanced organic chemistry.  They merely read German books on how to make dyes.  They realize the American market is getting bigger and bigger and that they should get into the action. But what happens is they cannot overcome the fact that they cannot acquire the capabilities to compete from merely reading books. After eight years they’re out of business.



Finally in Chapter 4, I highlight that lobbying is going in each country and that these activities change the institutional environment in each country. In fact, when I look at lobbying in detail, I find much evidence that industrial players bring about changes in patent laws and changes in university appointments.  In Britain what happens is the British textile industries get all the textile chemists and professors of textile chemistry they want, but it’s very, very hard to convince a university to create new professorships in organic chemistry because the dye industry is so small.



I want to end by saying there’s a lot more in the book. I have not gone at all into the technical details of the co-evolutionary theory that I articulate in Chapter 5. The book is in the library, so if you are interested, you can pick it up.  I also have not talked at all about the implications I spell out in Chapter 5 concerning public policy and management. An evolutionary point of view makes you think very differently about how to manage a firm.  You also think very differently about how to structure an economy.

As you will have noticed, I have not presented any regression table that formed the tip of the iceberg of the empirical analysis in Stan’s 1954 British steamships, but I do think that what I’ve done in my book actually is pretty close to the vision that Stan & Jonathan Hughes had for Cliometrics. I am glad Stan 47 years later is still with us here today to make that judgment himself.

Q&A Session:

Q: Dipak Jain: Can you tell us to what extent the lessons learned from your study still apply today?

A: I think there are many parallels in contemporary high-tech industries. If you think, for example, about how the biotech industry evolved and why biotech is strong in the United States and in Britain, but not nearly as strong in Germany and Japan, the same co-evolutionary processes seem to be at work. I have not researched biotechnology as much as the synthetic dye industry. Based on what I already know about biotech, I can say that the exchange of personnel from university to firms, but also from firms back into the university where people can actually be appointed professors after having worked in industry, is very important for the competitive success of a national biotech industry.  One of things I want to do in the future is to show with detailed data how these co-evolutionary processes play out in biotech as well.  My point is this co-evolution of academic disciplines and industrial sectors is a generalizable process. I’m actually working on a paper right now that tries to articulate the specific mechanisms of this coevolutionary process.

Q: Joel Mokyr: I was wondering if you could say something about the importance of intellectual property rights and the differences in this institution from one country to the next. How important were these differences for the development of the synthetic dye industry?
A: In the book I discuss in detail the differences in intellectual property rights across the dye producing countries.  Let me highlight one important effect of these differences. Going back to the statistics I presented earlier on firm entry and exit, you recognize that Germany had many more firm entries than the other countries.  There were two reasons why Germany had more entries. First, you better get one chemist who knows a little bit about this in order to start a firm. The risk set of entrepreneurs is in fact determined by the number of chemists in the environment. Because Germany produced more chemists than Britain, more entry was possible in Germany.

Second, because Germany was split into 39 different states, it did not have an effective patent law until 1877.  So you had free entry into the industry there.  British and French firms restricted entry into the market by filing for patent monopolies. What the German firms did for the first 20 years is simply to copy the patented dye innovations that came out of Britain and France.  The differences in intellectual property rights were very important for the long-term success of each national industry. Free entry in Germany proved to be an important advantage in the long run. Having more firms in the industry makes success more likely. Think about it this way: If you’re trying a hundred times, you are more likely to hit upon a good firm strategy than if you’re only trying five times. Remember Campbell’s idea about the importance of numerosity of trails for success. What the patent laws do is they allow more trials in Germany and very few trials in Britain and in France.

Q: Ravi Jagannathan: How did the patent laws work at that time? Could German firms enforce patents in Britain?  If they could not, they could not have gained an advantage in Britain. According to my understanding of history the British tried very hard and just could not succeed, and that the British looked upon by the German chemists as if they were God.  So did the British simply lack the technical expertise at the university level to replicate what was going on in Germany?
A: No. That’s not true. From 1845 until 1865, there’s a German Professor, Hofmann, at the Royal College of Chemistry. He is the teacher of Perkin, the inventor of the first synthetic dye.  Nearly all the people who initially started British synthetic dye firms were students at this college.  In 1865, Hofmann leaves for Germany to become a Professor in Berlin. This clearly was a blow to the British industry because now the organic chemical expertise was reduced in Britain, and because he started to train students primarily for German industry. But Britain initially had substantial strength in organic chemistry. The problems set in more as the synthetic industry was declining in Britain and it became hard to convince the government or private supporters to fund synthetic organic chemistry.  The textile industry wanted schools to train textile chemists.  The brewing industry wanted chemists to help them with brewing.  The brewing and textile industries in Britain individually were many times larger than the British dye industry. So this is all about power and lobbying on one level.  Academic fields change in part because of this lobbying, or because of the government intervention.

You can have government intervention without firm lobbying.  This is precisely what happens in Britain, France and the U.S. after 1914. This is why 1914 was a natural stopping point for my book.  As soon as World War I began, governments discovered that the organic chemicals that go into making synthetic dyes were immensely important for military purposes.  The British wanted ammunitions and that required much more capacity in organic chemicals production. So the British, the French, and the Americans during World War I say, never ever in the world will we allow the Germans to dominate this industry because that creates huge military problems for us.  As a result, there is massive state intervention in these three countries. The British did create organic chemical capabilities at huge costs. 

So let me speak a little more about the creation of strong academic disciplines in a country.  I want to be very clear.  I don’t really care about the dye industry at all, although I spent nine years on it. What I tried to do was consistent with what Stan and his conspirators wanted to do, that is, I tried to develop an empirically grounded theory. In my view, before you come up with general principles, you’ve got to make sure that your concepts have analytic power in at least in one context. Once you have concepts that work, it’s always easy to abstract away from the context—to push it to high levels of abstraction.  The problem is if you start with an abstraction, and the abstraction doesn’t help you to illuminate any context because it is too abstract, then you are never going to end up with a good theory. This is why I studied the dye industry in great detail and developed concepts at a level of abstraction where they have analytical bite but remain general enough that they can be applied to other industries. 

Based on my dye industry study, I have recently formulated a general hypothesis about the co-evolution of industries and academic disciplines.  To give you the intuition, let’s engage in a thought experiment.



Let’s rank all the industrial sectors in the economy from weak to strong. You can use GDP, or you can use export shares, you can use any other measure you want to rank order all the industries.  Rank order also all academic disciplines within a country from weak to strong. My claim, based on the study of the synthetic dye industry, is that because of competition among players for favorable tax and tariffs rates and for other favorable treatments by governments, and because of competition of economic disciplines for resources, a co-specialization will occur in a country among industries and academic disciplines. My general hypothesis is that the strong/weak, and weak/strong cells, the off diagonal cells, are unstable. 





What am I’m predicting is competitive forces over time will make the world look like this:



The configurations that are stable within a particular country are when you have a strong discipline and you have a strong industry, or when you have a weak discipline and a weak industry. But it’s not sustainable to be strong and weak, or weak and strong.  The dynamic competitive forces will push observations into the cells on the diagonal.

Once I show this general hypothesis to you, you may say, yeah, this is obvious. Of course it must be true. But how many of you thought about the relationship of academic disciplines and industries in this way before I presented you with these ideas?  What remains to be done, of course, is an empirical test of these ideas with historical data from many different countries. If you look at all industrializing countries over time and can show that these predictions are true, you have gained from my point of view a much deeper understanding of the relationship between industries and academic disciplines.  The results from this research will be very relevant for policy makers and independent of whether the data shows the predictions to be true and false. 

Q: Dipak Jain: I know Campbell’s work on experimental design and measurement, but can you tell us where Campbell wrote about evolutionary theory? 

A: There are three key papers on this topic by Campbell. One is on creative thought (1960). The second paper is socio-cultural evolution (1969) and the third one is on evolutionary epistemology (1974).




Aitken, H. G. J. 1976. Syntony and Spark: The Origins of Radio. New York: John Wiley & Sons.

Aitken, H. G. J. 1985. The Continuous Wave: Technology and American Radio, 1900 - 1932. Princeton, New Jersey: Princeton University Press.

Campbell, D. T. 1960. Blind Variation and Selective Retention in Creative Thought as in Other Thought Processes. Psychological Review, 67: 380-400.

Campbell, D. T. 1969. Variation and Selective Retention in Socio-Cultural Evolution. General Systems, 14: 69-85.

Campbell, D. T. 1974. Evolutionary Epistemology. In P. A. Schilpp (Ed.), The Philosophy of Karl Popper., Vol. 14: 413-463. La Salle, Ill.: Open Court.

Cyert, R. M. & March, J. G. 1963. A Behavioral Theory of the Firm. Englewood Cliffs, N.J.: Prentice-Hall.

Davis, Lance E., Jonathan R. T. Hughes, and Stanley Reiter. 1960.  Aspects of Quantitative Research in Economic History. The Journal of Economic History 20(4): 539-547.

Hughes, Jonathan R. T., and Stanley Reiter (1958). The First 1,945 British Steamships. Journal of the American Statistical Association 53(282): 360-381.

Hughes, T. P. 1983. Networks of Power. Baltimore: The Johns Hopkins University Press.

Mokyr, J. 2002. The Gifts of Athena: Historical Origins of the Knowledge Economy. Princeton: Princeton University Press.

Murmann, J. P. 2003. Knowledge and Competitive Advantage: The Coevolution of Firms, Technology, and National Institutions. New York: Cambridge University Press.

Nelson, R. R. & Winter, S. G. 1982. An Evolutionary Theory of Economic Change. Cambridge, MA: The Belknap Press of Harvard University Press.

Simon, H. A. 1981. The Sciences of the Artificial (Second ed.). Cambridge, Massachusetts: MIT Press.

Vincenti, W. 1990. What Engineers Know and How They Know It. Baltimore: Johns Hopkins Press.