Any good idea is found independently, and there are always early works that seem to unambiguously imply the ultimate result even if they never really focused on it. So it is with low volatility investing.
Bob Haugen has been touting unconventional investment tactics for decades, starting with The Incredible January Effect published in 1987. Alas, the January effect was one of those anomalies centered on low-priced, small-sized, firms, that usually don't scale well. Do you remember the 'low price' effect? It was popular in the 1980s, along with the size effect, as small size and low priced stocks were highly correlated and both seemed correlated with very high returns. The low price funds are now something anyone associated with them at the time conveniently forgets, because trading them is very costly, like trading options instead of stocks.
Anyway, I saw him being interviewed at a Dutch conference, with the talk 'The Low Volatility Anomaly' on the screen, where he was arguing investors should move their equity exposures to low volatility oriented equity funds. At one point the interviewer asks if 'will you be known as one of the big heroes of the twenty first century?' and Bob gives a sort of 'awe shucks' type response.
Now, Haugen did, in 1996, argue that low-risk stocks tend to outperform high-risk stocks. But the emphasis on this paper was this was a consequence of market inefficiency. He created a model with many factors, grouped into 1) risk (eg, beta, volatility, 2) liquidity (eg, volume, price), 3) value (eg, P/E), 4) growth potential (eg,ROA), 5) past returns (eg,past 6-month return), and lastly 6) sector variables. Each class had ten or so highly correlated metrics within them. A more recent paper he did (again, with co-author Nardin Baker), basically applied the same idea. While he noted the highest returning portfolio constructed via this backfitting were of lower volatility and beta than the lower returning portfolios.
The emphasis was clearly on a model that predicted returns, and even now he touts 70 factors, many of which are highly correlated. The factor risk premia are calculated on a rolling basis, so signs on these factors bounce around. It's a classic kitchen sink regression. I don't see the emphasis as being 'low volatility', rather, that is a characteristic of his 'highest expected return' portfolios, which is his focus. For example, his 1996 Journal of Portfolio Management article on this riff was titled 'The Effects of Intrigue, Liquidity, Imprecision, and Bias on the Crosss-Section of Expected Returns.' Lately he discusses three types of volatility--event-driven, error-driven, and price-driven--which he tries to separate and use differently. I think he's slicing this too thin, as you get pretty similar results using total or idiosyncratic volatility. That's a lot going on, but clearly volatility and risk are not emphasized as prime movers, he just found that risk was inversely correlated with these more interesting factors.
I do think he's a bit disingenuous with his performance. His page showing the cumulative returns for his various models hows a rather incredible upward growth since 1999. However, I remember some using his factors in the latter half of 2003, and the model generated a significant draw down, well-outside the scope of his backtests. Interestingly, there's no evidence of that in his charts. I presume that was an earlier version that got dropped down the memory hole.
In sum, I think he didn't prioritize volatility until the low-volatility movement get really going to get credit for 'low volatility' investing per se. While I wrote my dissertation on this back in 1994, but it went over like the Hindenburg with academia, and I never got refereed publication on the volatility result, so I wouldn't say I'm a founding father (I didn't have an equilbrium story then, and this was before Freakonomics and the popularity of behavioral finance, so back then it just didn't make sense). I would say that the two main academic pieces I relied upon for my finding then were Bruce Lehman's 1990 Residual Risk Revisited, and Ed Miller's 1977 Risk, Uncertaint, and Divergence of Opinion. Both focused on volatility, and implicitly noted that there appeared an attractive investment strategy implied by the poor average returns to highly volatile stocks. Indeed, in 2001, Miller mentions this strategy in the Journal of Portfolio Management. It seems, one needs a theory to see something, and because Miller had this theory (winner's curse) he had thought relevant to equities, he saw the investing implication before others. That is, one could deduce it as a corollary of Fama and French's 1992 finding that beta was uncorrelated with average returns, but as Fama and French merely extended the standard model to other risks, they missed the low-volatility/beta bandwagon. For example, Haugen's big theory is that markets are inefficient, and so I think he wasn't focused on volatility for the reason that this insight is incomplete: inefficient in what way?
By the mid aughts, several academics had discovered this in various guises. There's Clarke, de Silva, and Thorley (2006) and Blitz and vav Vliet (2007), which focused on low volatility portfolios. Then there's Ang, Hodrick, Xing and Zhang (2006) highlighting the poor cross-sectional returns to higher volatility stocks, and that really was seminal for academics. Firms like Analytic Investors, Robeco, Arcadian, and Unigestion all started low volatility funds around this time, so they had all seen this years earlier. While all this was becoming common knowledge, I was fighting a lawsuit by a former employer trying to stop me from using volatility as an investment factor, and I remember thinking how crazy it was that I had to fight to use something I had been touting since 1993 that then had a pretty large publication thread.