In describing how human knowledge advances, Donald Campbell devised the BVSR model to explain the evolutionary way in which knowledge accumulates. BVSR stands for Blind Variation and Selective Retention and is more or less natural selection applied to creativity and epistemology. Others have explored Campbell's with great insight and how it applies to fields such as organization science.
Evolutionary economics moves in and out of fashion but for the most part remains on the fringes of economic thought. Prompted in part by Paul's post earlier today, I'm wondering today about startup firms in a BVSR model of economic dynamics. It's no surprise (and, I don't think, controversial) that free market competition works through an evolutionary process of selection. Individuals and companies try various things (new products, services, processes) and their fate hinges on whether or not the market "selects" them vis-a-vis competing ideas. These new products and services are the variations in BVSR, whether radically new innovations or incremental improvements on something that already exists. We throw lots of spaghetti against the wall to see what sticks. Economic variation, of course, is not completely random since firms often seek to satisfy a previously identified market need or solve a heretofore unsolved problem. Variation is, however, blind in Campbell's sense, in that we do not know beforehand what will or will not succeed. "Blind" refers to our knowledge (more specifically, our lack of knowledge) about the consequences of any particular variation.
So we continuously introduce new variations into the market. The manner in which they are winnowed and propagated is "selective retention." Selection is clear: a new market variation either succeeds or it doesn't. Or, more precisely, it either survives (is selected) or doesn't survive (is not selected or selected against), with survival not always a proxy for subjective superiority. Here, it seems like the BVSR model advances somewhat beyond the standard view of the free-market-as-natural-selection (in addition to the explicit recognition that variations are blind). Once a variation is selected, what happens to it? It is retained, meaning it somehow propagates through the population. A variation doesn't merely survive--it grows and spreads.
How would you increase economic growth or increase innovation through the BVSR model? Let's say each element is multiplied by the others, so growth or innovation is the result of V x S x R, all divided by B. Increasing any one element in the numerator should theoretically increase the product. (It's not clear that this should be true, by the way. I haven't convinced myself that multiplication is correct here. I suppose it's conceivable that it could or should be, for example, "+" R. This isn't a research paper and I'm just following an idea here, so please feel free to take issue with this. At this point, it's of de minimis concern.) According to this, then, you could increase growth or innovation by increasing variation (the introduction of new products and services), improving the selection process (so only "good" or "right" variations are selected), and improving retention, to the properly selected variations quickly achieve rapid scale and propagation, disseminating their benefits to everyone.
How would we approach startups here? Obviously, startups are one type of variation (at the population-level of the economy). They are selected for and against: roughly half of new firms (48%) survive to age five, with the survival curve slowly falling after that. (I've seen some data indicating that one-third of new firms survive to either age 7 or age 10.) Survival, of course, tells you little about success or quality, and selection in the case of firms is an ongoing process. Some research suggests that only the tiniest handful of companies survive for more than a few decades (suggesting that groups such as those companies on the Fortune 500 are both an incredibly small minority and, perhaps, a secondary story--maybe failure is the dominant story of economic selection). Retention could be either the organic growth of young companies or, I suppose, acquisition by larger, more established companies. (Revenue or employment growth or both would here serve as a proxy for another way of looking at retention: spread of a company's product through the market.) With the denominator, "Blind," we might want to reduce the blindness of our variations--that is, wouldn't we want to have a better idea of the prospective fate of new firms? Or, should B be another multiple here? Would increasing blindness be better because it would throw more of the burden onto the competitive process? Perhaps B is neutral here? That is, perhaps it cannot be increased or decreased?
In any case, let's assume, with good reason, that startups are absolutely essential to innovation and economic growth. To increase these, what would we do with startups in the BVSR model? To begin with, could we reduce the blindness of a particular startup's fate, or the fate of all startups together? (It's not clear, moreover, what Blind should refer to here: mere survival to a certain age? Success, however nebulously defined?) Let's say Blind refers to the state of knowledge facing the founders of a new company: they do not know, and can't know, whether or not they will "succeed," however they define that term. (Although, if your objective is to simply work for yourself or the ability to wear flip-flops to work, then merely starting your own company can be considered success.) From a macro perspective, it's not clear that we can either reduce or increase blindness at the startup stage--the figure given above for survival to age five (48%) has been remarkably consistent across time. The survival rate for new firms, that is, doesn't appear to vary much founding year. There is some variance by sector but again, the survival lines within each sector are rather consistent, so it's not clear how we could either raise or lower the survival line. Of course, if survival rates have been consistent for thirty years, then perhaps the Blind element is neutralized: there is a 50-50 chance your firm will still be around five years from now.
Now, to Variation. Here, you would simply want to increase the number of new firms that are founded each year, with the expectation that trying more new variations would lead to more "good" things being selected and retained in the economy. An increase here means an increase there. I am sympathetic to this strategy, but I'm unclear as to how we do it. Firm formation in the United States has been pretty steady for two or three decades and the question of how we increase it is not so simple. Things like awareness and lowering barriers to entry are probably our best bets but these will vary by sector and geography and characteristics of the founder. All of this assumes, of course, that a basic increase in startups (more variation) will result in an increase in innovation and growth. And this assumes that we would see a similar proportion of innovative ideas and potentially successful firms in a larger pool of variations--is that a valid assumption?
Selection of startups occurs through a variety of mechanisms. One, of course, is the market itself: you start a new firm and quickly find out whether or not it will selected in or out. If we increased the number of startups (increased V), then conceivably our selection mechanisms could remain the same, simply working on a larger population of firms. If, however, V remained constant, we might want to intensify S by looking more carefully (somehow) for promising ideas or "sure-thing" innovations. I suppose this is the role played by venture capital firms and accelerators such as Y Combinator. They have many entrepreneurs approach them and apply to them, but only select a few for funding or developing. Those that they do not select still go on to start new companies, so the claim of these particular selection mechanisms is that they are intensifying S so as to increase R: selecting the "best" of V, those with the most potential for retention and propagation by the market. As an overall strategy to enhance Selection, we would look to increase market competition in general, removing protective barriers around certain industries and firms, for example. A perennial complaint about some European economies (and conclusion of numerous McKinsey reports) is that they are not as competitive as the United States--stultified selection suppresses innovation and economic growth. It's also conceivable that simply increasing V would naturally increase S (they are, after all, multiplied by each other)--more firms overall would mean more firms competing.
These notions of Selection, however, presume that any selection process, whether in biological evolution or in free market competition, is an optimizing one--that the "best" organism or population or idea or company always beats out inferior competitors. Adaptive fit, after all, implies such optimization. Yet we know that natural selection is only one way in which genes and organisms are winnowed and propagated. The role of accident and randomness in evolution cannot be forgotten: we like to think of dinosaurs as having met their proper evolutionary fate because they were all slow and dim-witted. But many evolutionary theorists have pointed out that, if a meteor or volcanic explosion did the dinosaurs in, then that isn't necessarily "natural selection" in terms of adaptation and competition. It was an exogenous event and the emerging population of smaller mammals happened to be better suited for survival. So, yes, an exogenous event "selected" out dinosaurs, but it wasn't necessarily the "optimizing" path we like to imagine. The same applies to the selection process in terms of firms and ideas: Silicon Valley bested Boston and Dallas because it had a better ecosystem, the selection process was more optimizing. But what if William Shockley's mother lived in Dallas, not Palo Alto? What if Frederick Terman had not contracted tuberculosis and had ended up returning to MIT? I realize these are entirely speculative and that the obvious riposte is, 'Well, that's just how things happened.' My point is simply that we shouldn't assume that Selection always implies optimizing; human behavior and group dynamics are such that the economic process contains plenty of satisficing behavior. (We also shouldn't forget that an entity such as a firm can become over-adapted to certain circumstances and thus too sessile to change; selection could work too well.)
Finally, how would we enhance Retention? What would this mean? It could entail somehow promoting those firms that are selected--say, those that survive to age six--via certain rewards such as a retroactive tax credit or something like that. Or, it could mean favoring those firms that grow rapidly in their first few years and have rapidly passed through V and S and we want to enhance their retention and propagation. It could mean reducing barriers to "scale," whether in the form of regulations, corporate taxes, whatever. (I continue to maintain, however, that the effect of certain taxes on entrepreneurship is not nearly as straightforward as some might like to believe.) Maybe we think of Retention as going public, in which case perhaps changes need to be made to Sarbanes-Oxley. Over the past few years, around 80% of VC exits have been acquisitions, not IPOs, and many of the companies coming out of accelerators are acquired by companies such as Google.
In any case, I don't know if this has added anything to the general discussion of entrepreneurship and economic growth or not. If we think about startups through the BVSR model, there are probably certain areas in which we can make changes to enhance the economic impact of startups, although some of them steer quite close to "picking winners." In general, as I said, I favor increasing V, but I'm as yet unclear as to how we go about doing this. Importantly, increased Variation should apply to firms of all types, not simply in favored sectors or those sectors we think (or would like to think) represent promising areas of innovation. Firm formation and innovation are compounding phenomena, so increasing the overall level of startups should (emphasize, should) raise our rates of innovation and growth, and this should be true whatever sector of the economy we are discussing. Information technology has been hugely important for American productivity, both in its production and its use in sectors like retail. We needed IT, but we also needed all those new retail outlets and supply chain innovations for the IT to become productive. It's possible that a BVSR-type of analysis in, say, 1995 might have concluded: we need more IT startups and fewer new retail stores. The economy might not have performed as well as it did.