Splitters and lumpers

After reading this quote from Darwin, Matt asks whether there are too few lumpers in economics:

“It is good,” opined Charles Darwin in an 1857 letter to the botanist J. D. Hooker, “to have hair-splitters and lumpers.” He was talking about how best to classify varieties of flora; being Darwin, he managed to establish an enduring intellectual distinction in a parenthetical aside. A century and a half later, his observation still holds. Splitters, focusing on difference, make sense of the world by dividing it into many small categories. Lumpers, focusing on likeness, sort it into a few big groups.

The distinction seems to relate very closely to McCloskey’s idea of economic rhetoric, where narratives of events are interpreted through the lens of a particular metaphor. The splitters are the people who explore the details of the narrative, trying to understand the minutiae of a situation. The lumpers follow in their tracks, surveying the assembled work of the splitters and constructing new metaphors that draw together seemingly disparate strands of work. Often, the work of the lumpers is considered the true genius but, as Darwin observed, they are complementary and neither could exhibit their talent without the other.

In the field of economics today, the splitters are often the technicians: mathematical savants adding incrementally more depth to existing models, or econometric whizzes who use novel techniques and data to test the technicians’ models. Often the theoretical splitters work outside the limelight, while statistical talents like Levitt, Angrist, and Wolfers are now finding their work rather trendy!

The lumpers of economics are the Nobel prize-winners who developed new metaphors for understanding the technicians’ work. Ricardo’s model of comparative advantage that allowed economists an initial understanding of trade. Becker’s model of human capital that has spread like wildfire through economics. Stiglitz’ unification of the field of asymmetric information that is now central to so much of micro and macro theory. These developments created new metaphors for interpreting the data that allowed the technicians to open up new fields of research. In each case they drew together large quantities of prior work and their talent was to see the common strand that united them.

So would it be better to have more of these lumpers? Not necessarily. There are obviously complementarities between the splitters who do the digging and explore different narratives, and the lumpers who generate new metaphors to interpret those narratives in a coherent way. To have ‘too many’ of one there would need to be problems with the implicit prices in the rewards for each, otherwise they would provide a sufficient equilibration mechanism to ensure that we have ‘enough’ people doing each. Perhaps the real problem is that we provide stardom to one group, while passing over the importance of all the splitting that came before.

4 replies
  1. Matt Nolan
    Matt Nolan says:

    “Perhaps the real problem is that we provide stardom to one group, while
    passing over the importance of all the splitting that came before.”

    This would lead to too many lumpers right, interesting.

    • jamesz
      jamesz says:

      I don’t know much about superstar effects, but doesn’t it depend on the risk preferences of the economist population? Assuming that only the few superstars achieve more than the average economist.

      • Matt Nolan
        Matt Nolan says:

        Sorry I assumed your comment was stating that we only provide, ex-ante, a positive probability of superstardom to lumpers. As a result, this potential benefit only exists if you decide to take the lumper path rather than the splitting path.

        • jamesz
          jamesz says:

          I was saying that. But it could still be that the expected income of each type is very close and people choose between them largely based on risk preference.

Comments are closed.