Guns, Germs, and Steel: The Fates of Human Societies

by Jared Diamond

  These questions arise at many different levels and for many types of groups. They apply to the organization of entire countries: remember the perennial arguments about whether the best form of government is a benign dictatorship, a federal system, or an anarchical free-for-all. The same questions arise about the organization of different companies within the same industry. How can we account for the fact that Microsoft has been so successful recently, while IBM, which was formerly successful, fell behind but then drastically changed its organization and improved its success? How can we explain the different successes of different industrial belts? When I was a boy growing up in Boston, Route 128, the industrial belt around Boston, led the world in scientific creativity and imagination. But Route 128 has fallen behind, and now Silicon Valley is the center of innovation. The relations of businesses to one another in Silicon Valley and on Route 128 are very different, possibly resulting in those different outcomes.

  Of course, there are also the famous differences between the productivities of the economies of whole countries, such as Japan, the United States, France, and Germany. Actually, though, there are big differences between the productivity and wealth of different business sectors even within the same country. For example, the Korean steel industry is equal in efficiency to ours, but all other Korean industries lag behind their American counterparts. What is it about the different organization of these various Korean industries that accounts for their differences in productivity within the same country?

  Obviously, answers to these questions about differences in organizational success depend partly on the idiosyncrasies of individuals. For example, the success of Microsoft has surely had something to do with the personal talents of Bill Gates. Even with a superior corporate organization, Microsoft would not be successful with an ineffectual leader. Nevertheless, one can still ask: all other things being equal, or else in the long run, or else on the average, what form of organization of human groups is best?

  My comparison of the histories of China, the Indian subcontinent, and Europe in the epilogue of GGS suggested an answer to this question as applied to technological innovation in whole countries. As explained in the preceding section, I inferred that competition between different political entities spurred innovation in geographically fragmented Europe, and that the lack of such competition held innovation back in unified China. Would that mean that a higher degree of political fragmentation than Europe’s would be even better? Probably not: India was geographically even more fragmented than Europe, but less innovative technologically. This suggested to me the Optimal Fragmentation Principle: innovation proceeds most rapidly in a society with some optimal intermediate degree of fragmentation: a too-unified society is at a disadvantage, and so is a too-fragmented society.

  This inference rang a bell with Bill Lewis and other executives of McKinsey Global Institute, a leading consulting firm based in Washington, D.C., which carries out comparative studies of the economies of countries and industries all over the world. The executives were so struck by the parallels between their business experience and my historical inferences that they presented a copy of GGS to each of the firm’s several hundred partners, and they presented me with copies of their reports on the economies of the United States, France, Germany, Korea, Japan, Brazil, and other countries. They, too, detected a key role of competition and group size in spurring innovation. Here are some of the conclusions that I gleaned from conversations with McKinsey executives and from their reports:

  We Americans often fantasize that German and Japanese industries are super-efficient, exceeding American industries in productivity. In reality, that’s not true: on the average across all industries, America’s industrial productivity is higher than that in either Japan or Germany. But those average figures conceal big differences among the industries of each country, related to differences in organization—and those differences are very instructive. Let me give you two examples from McKinsey case studies on the German beer industry and the Japanese food-processing industry.

  Germans make wonderful beer. Every time that my wife and I fly to Germany for a visit, we carry with us an empty suitcase, so we can fill it with bottles of German beer to bring back to the United States and enjoy over the following year. Yet the productivity of the German beer industry is only 43 percent that of the U.S. beer industry. Meanwhile, the German metalworking and steel industries are equal in productivity to their American counterparts. Since the Germans are evidently perfectly capable of organizing industries well, why can’t they do so when it comes to beer?

  It turns out that the German beer industry suffers from small-scale production. There are a thousand tiny beer companies in Germany, shielded from competition with one another because each German brewery has virtually a local monopoly, and they are also shielded from competition with imports. The United States has 67 major beer breweries, producing 23 billion liters of beer per year. All of Germany’s 1,000 breweries combined produce only half as much. Thus the average U.S. brewery produces 31 times more beer than the average German brewery.

  This fact results from local tastes and German government policies. German beer drinkers are fiercely loyal to their local brand, so there are no national brands in Germany analogous to our Budweiser, Miller, or Coors. Instead, most German beer is consumed within 30 miles of the factory where it is brewed. Therefore, the German beer industry cannot profit from economies of scale. In the beer business, as in other businesses, production costs decrease greatly with scale. The bigger the refrigerating unit for making beer, and the longer the assembly line for filling bottles with beer, the lower the cost of manufacturing beer. Those tiny German beer companies are relatively inefficient. There’s no competition; there are just a thousand local monopolies.

  The local beer loyalties of individual German drinkers are reinforced by German laws that make it hard for foreign beers to compete in the German market. The German government has so-called beer purity laws that specify exactly what can go into beer. Not surprisingly, those government purity specifications are based on what German breweries put into beer, and not on what American, French, and Swedish breweries like to put into beer. Because of those laws, not much foreign beer gets exported to Germany, and because of inefficiency and high prices much less of that wonderful German beer than you would otherwise expect gets sold abroad. (Before you object that German Löwenbräu beer is widely available in the United States, please read the label on the next bottle of Löwenbräu that you drink here: it’s not produced in Germany but in North America, under license, in big factories with North American productivity and efficiencies of scale.)

  The German soap industry and consumer electronics industry are similarly inefficient; their companies are not exposed to competition with one another, nor are they exposed to foreign competition, and so they do not acquire the best practices of international industry. (When is the last time that you bought an imported TV set made in Germany?) But those disadvantages are not shared by the German metal and steel industries, in which big German companies have to compete with one another and internationally, and thus are forced to acquire the best international practices.

  My other favorite example from the McKinsey reports concerns the Japanese food-processing industry. We Americans tend to be paranoid about Japanese efficiency, and it is indeed formidable in some industries—but not in food-processing. The efficiency of the Japanese food-processing industry is a miserable 32 percent that of ours. There are 67,000 food-processing companies in Japan, compared to only 21,000 in the United States, which has twice Japan’s population—so the average U.S. food-processing company is six times bigger than its Japanese counterpart. Why does the Japanese food-processing industry, like the German beer industry, consist of small companies with local monopolies? Basically, the answer is the same two reasons: local taste and government policies.

  The Japanese are fanatics for fresh food. A container of milk in a U.S. supermarket bears only one date: the expiration date. When my wi
fe and I visited a Tokyo supermarket with one of my wife’s Japanese cousins, we were surprised to discover that in Japan a milk container bears three dates: the date the milk was manufactured, the date it arrived at the supermarket, and the expiration date. Milk production in Japan always starts at one minute past midnight, so that the milk that goes to market in the morning can be labeled as today’s milk. If the milk were produced at 11:59 P.M., the date on the container would have to indicate that the milk was made yesterday, and no Japanese consumer would buy it.

  As a result, Japanese food-processing companies enjoy local monopolies. A milk producer in northern Japan cannot hope to compete in southern Japan, because transporting milk there would take an extra day or two, a fatal disadvantage in the eyes of consumers. These local monopolies are reinforced by the Japanese government, which obstructs the import of foreign processed food by imposing a 10-day quarantine, among other restrictions. (Imagine how Japanese consumers who shun food labeled as only one day old feel about food 10 days old.) Hence Japanese food-producing companies are not exposed to either domestic or foreign competition, and they don’t learn the best international methods for producing food. Partly as a result, food prices in Japan are very high: the best beef costs $200 a pound, while chicken costs $25 a pound.

  Some other Japanese industries are organized very differently from the food processors. For instance, Japanese steel, metal, car, car parts, camera, and consumer electronic companies compete fiercely and have higher productivities than their U.S. counterparts. But the Japanese soap, beer, and computer industries, like the Japanese food-processing industry, are not exposed to competition, do not apply the best practices, and thus have lower productivities than the corresponding industries in the United States. (If you look around your house, you are likely to find that your TV set and camera, and possibly also your car, are Japanese, but that your computer and soap are not.)

  Finally, let’s apply these lessons to comparing different industrial belts or businesses within the United States. Since the publication of GGS, I’ve spent much time talking with people from Silicon Valley and from Route 128, and they tell me that these two industrial belts are quite different in terms of corporate ethos. Silicon Valley consists of lots of companies that are fiercely competitive with one another. Nevertheless, there is much collaboration—a free flow of ideas, people, and information among companies. In contrast, I’m told, the businesses of Route 128 are much more secretive and insulated from one another, like Japanese milk-producing companies.

  What about the contrast between Microsoft and IBM? Since GGS was published, I’ve acquired friends at Microsoft and have learned about that corporation’s distinctive organization. Microsoft has lots of units, each comprised of 5 to 10 people, with free communication among units, and the units are not micromanaged; they are allowed a great deal of freedom in pursuing their own ideas. That unusual organization at Microsoft—which in essence is broken into many competing semi-independent units—contrasts with the organization at IBM, which until some years ago consisted of much more insulated groups and resulted in IBM’s loss of competitive ability. Then IBM acquired a new chief executive officer who changed things drastically: IBM now has a more Microsoft-like organization, and I’m told that IBM’s innovativeness has improved as a result.

  All of this suggests that we may be able to extract a general principle about group organization. If your goal is innovation and competitive ability, you don’t want either excessive unity or excessive fragmentation. Instead, you want your country, industry, industrial belt, or company to be broken up into groups that compete with one another while maintaining relatively free communication—like the U.S. federal government system, with its built-in competition between our 50 states.

  THE REMAINING EXTENSION of GGS has been into one of the central questions of world economics: why are some countries (like the United States and Switzerland) rich, while other countries (like Paraguay and Mali) are poor? Per-capita gross national products (GNP) of the world’s richest countries are more than 100 times those of the poorest countries. This is not just a challenging theoretical question giving employment to economics professors, but also one with important policy implications. If we could identify the answers, then poor countries could concentrate on changing the things that keep them poor and on adopting the things that make other countries rich.

  Obviously, part of the answer depends on differences in human institutions. The clearest evidence for this view comes from pairs of countries that divide essentially the same environment but have very different institutions and, associated with those institutions, different per-capita GNPs. Four flagrant examples are the comparison of South Korea with North Korea, the former West Germany with the former East Germany, the Dominican Republic with Haiti, and Israel with its Arab neighbors. Among the many “good institutions” often invoked to explain the greater wealth of the first-named country of each of these pairs are effective rule of law, enforcement of contracts, protection of private property rights, lack of corruption, low frequency of assassinations, openness to trade and to flow of capital, incentives for investment, and so on.

  Undoubtedly, good institutions are indeed part of the answer to the different wealths of nations. Many, perhaps most, economists go further and believe that good institutions are overwhelmingly the most important explanation. Many governments, agencies, and foundations base their policies, foreign aid, and loans on this explanation, by making the development of good institutions in poor countries their top priority.

  But there is increasing recognition that this good-institutions view is incomplete—not wrong, just incomplete—and that other important factors need addressing if poor countries are to become rich. This recognition has its own policy implications. One cannot just introduce good institutions to poor countries like Paraguay and Mali and expect those countries to adopt the institutions and achieve the per-capita GNPs of the United States and Switzerland. The criticisms of the good-institutions view are of two main types. One type recognizes the importance of other proximate variables besides good institutions, such as public health, soil- and climate-imposed limits on agricultural productivity, and environmental fragility. The other type concerns the origin of good institutions.

  According to the latter criticism, it is not enough to consider good institutions as a proximate influence whose origins are of no further practical interest. Good institutions are not a random variable that could have popped up anywhere around the globe, in Denmark or in Somalia, with equal probability. Instead, it seems to me that, in the past, good institutions always arose because of a long chain of historical connections from ultimate causes rooted in geography to the proximate dependent variables of the institutions. We must understand that chain if we hope, now, to produce good institutions quickly in countries lacking them.

  At the time that I wrote GGS, I commented, “The nations rising to new power [today] are still ones that were incorporated thousands of years ago into the old centers of dominance based on food production, or that have been repopulated by peoples from those centers…. The hand of history’s course at 8,000 B.C. lies heavily on us.” Two new papers by economists (Olsson and Hibbs, and Bockstette, Chanda, and Putterman) have subjected this postulated heavy hand of history to detailed tests. It turns out that countries in regions with long histories of state societies or agriculture have higher per-capita GNP than countries with short histories, even after other variables have been controlled. The effect explains a large fraction of the variance in GNP. Even just among countries with still-low or recently low GNPs, countries in regions with long histories of state societies or agriculture, like South Korea, Japan, and China, have higher growth rates than countries with short histories, such as New Guinea and the Philippines, even though some of the countries with short histories are much richer in natural resources.

  There are many obvious reasons for these effects of history, such as that long experience of state societies and agriculture implies experience
d administrators, experience with market economies, and so on. Statistically, part of that ultimate effect of history proves to be mediated by the familiar proximate causes of good institutions. But there is still a large effect of history remaining after one controls for the usual measures of good institutions. Hence there must be other mediating proximate mechanisms as well. Thus a key problem will be to understand the detailed chain of causation from a long history of state societies and agriculture to modern economic growth, in order to help developing countries advance up that chain more quickly.

  In short, the themes of GGS seem to me to be not only a driving force in the ancient world but also a ripe area for study in the modern world.

  ACKNOWLEDGMENTS

  IT IS A PLEASURE FOR ME TO ACKNOWLEDGE THE CONTRIBUTIONS of many people to this book. My teachers at Roxbury Latin School introduced me to the fascination of history. My great debt to my many New Guinea friends will be obvious from the frequency with which I cite their experiences. I owe an equally great debt (and absolution from responsibility for my errors) to my many scientist friends and professional colleagues, who patiently explained the subtleties of their subjects and read my drafts. In particular, Peter Bellwood, Kent Flannery, Patrick Kirch, and my wife, Marie Cohen, read the whole manuscript, and Charles Heiser, Jr., David Keightley, Bruce Smith, Richard Yarnell, and Daniel Zohary each read several chapters. Earlier versions of several of the chapters appeared as articles in Discover magazine and in Natural History magazine. The National Geographic Society, World Wildlife Fund, and University of California at Los Angeles supported my fieldwork on Pacific islands. I have been fortunate to have John Brockman and Katinka Matson as my agents, Lori Iversen and Lori Rosen as my research assistants and secretaries, Ellen Modecki as my illustrator, and as my editors Donald Lamm at W. W. Norton, Neil Belton and Will Sulkin at Jonathan Cape, Willi Köhler at Fischer, Marc Zabludoff and Mark Wheeler and Polly Shulman at Discover, and Ellen Goldensohn and Alan Ternes at Natural History.