Innovation, adoption, and diffusion have long been studied as important determinants of
economic growth. One category of innovation is the process innovation—the implementation of
improved production methods that may be embodied in new technologies or procedures. Process
innovations, while not necessarily impacting attributes of the final product, generate benefits in
the production process such as increasing productivity and decreasing costs. David (1990) argues
that diffusion of process innovation evolves within and across sectors and emphasizes the
importance of historical studies in understanding them. He demonstrates his approach comparing
the diffusion of the computer to that of the dynamo 100 years earlier. Olmstead and Rhode
(1995; 2001) also take a historical perspective on the diffusion of two mechanical process
innovations—the reapers and the tractor in the U.S. during the 19th and 20th centuries—and show
how the technologies, as well as markets and institutions, evolved to facilitate diffusion. This
article aims to expand this literature by focusing on the diffusion of drip irrigation—a process
innovation which enhances water use efficiency compared to traditional flood and furrow
irrigation systems. Concentrating on drip irrigation in California, we examine how diffusion of a
technology undergoes multiple waves of improvements and coevolution with other agronomical
practices. We highlight the role of the private and public sector in adapting process innovations
to local needs and show the necessity of historical analysis and perspective in assessing a
technology’s impacts. Lastly, given that the adoption of process innovations is often motivated
by growth in productivity, we empirically estimate the impact of drip irrigation on crop yields
and net farm income in California.
Drip irrigation conveys water to plants through a network of pipes and emitters, and
allows slow and controlled application of water. It is more capital intensive than traditional
irrigation technologies, but allocates a smaller volume of water per unit of time with higher
2
precision. Modern drip irrigation was first introduced in California in the late 1960s. By 1988—
a little over 20 years later—it was adopted on only 5% of irrigated land. Yet fast-forward another
twenty years and by 2010 over 40% of the irrigated land in California is irrigated using drip
systems (Tindula, Orang, and Snyder 2013). Having spread across diverse crops and regions in a
non-linear trajectory, drip irrigation is a prime example of a process innovation whose diffusion,
as we will show, is explained by the theory laid out in the threshold and the real option value
models.
We focus on drip irrigation for three reasons: First, drip was selected as one of the major
success stories over the last 100 years by an informal survey of California County Directors.
Second, it is timely to study a technology whose adoption has been shaped by droughts during a
period with a severe California drought. Third, it is first study of this kind for a technology that
aims to conserve a natural resource (water) and the lessons of its diffusion may apply to similar
technologies.
This article integrates multiple data sources in order to trace the dynamics of the diffusion
of drip irrigation over time. We begin by summarizing the conceptual background on technology
adoption and diffusion. In particular, David (1966) introduced the threshold model of technology
diffusion, which—in sharp contrast with imitation models of adoption—argued that adoption is
an explicit economic choice. However, a shortcoming of much of the conceptual literature on the
threshold model is the lack of explicit consideration of the coevolution of the technology and
complementary production processes and the contribution of private and public sector to this
coevolution (Olmstead and Rhode 2008). For a process innovation to be widely adopted among
potential users the technology may require adaptation to the needs of heterogeneous clientele.
Furthermore, the other components of production (fertilizing, seeding) may require adaptation to
3
the new technology. In agriculture, this coevolution often involves joint efforts of the private and
public sector. Given the vital, and often neglected, role of public sector R&D in the diffusion
process, we pay particular attention to institutions (i.e. Cooperative Extension) and the
complementarity of public and private efforts that affect the direction of diffusion.
Second, we outline the evolution of drip irrigation in California and highlight the linkages
between empirical evidence and theory. To construct the narrative we rely both on existing
literature and personal interviews with Irrigation Specialists and Farm Advisors who were
influential in drip’s diffusion in California. With these first-hand accounts, this narrative offers a
unique compilation of published works and previously undocumented oral histories.
Third, we juxtapose our historical analysis of drip’s diffusion in California with an
empirical analysis of the effects of drip irrigation on crop productivity (namely, the yield effect)
and farm income. One major concern about process innovations has been quantifying their
impact on productivity. For instance, David (1990) pointed out that the popular phrase,
“computers are everywhere but in the productivity statistics,” could have been said also for the
dynamo 100 years prior. Since we also find that drip irrigation appears to be everywhere in
Californian agriculture today—with broad adoption across locations and crops—we ask whether
the value of drip irrigation adoption can be quantified in the productivity statistics. In particular,
what has been the effect of drip irrigation adoption on crop yields? We focus on quantifying the
yield effect because it is one of the most discussed and often unexpected advantages of adopting
drip irrigation. Our empirical research design endeavors to answer whether the benefit of a
positive yield effect can be quantified (1) by surveying the published agronomic literature on the
yield effects of drip irrigation, and (2) by employing two distinct panel datasets on drip adoption
in California to estimate correlations between adoption and yield. We then use our estimated
4
yield effects to perform back-of-the-envelope calculations of drip’s effect on California net farm
income from crop production. Specifically, we estimate a yield effect of drip ranging from 16-
48%, depending on the crop and location, and an increase in the earnings of crop production in
California between 2.6-7.4% annually.
This article analyzes the history of drip irrigation in California to provide 1) a conceptual
framework for understanding the evolving diffusion of a process innovation and 2) estimates of
the value of process innovation adoption and diffusion. We show that the diffusion of drip was
gradual, Cooperative Extension was essential, and the gains for California agriculture were
substantial.
