Valuing product innovation: genetically engineered varieties in US corn and soybeans

• Author: Edward D. Perry,GianCarlo Moschini,Federico Ciliberto
• DOI: http://doi.org/10.1111/1756-2171.12290
• Date: 01 September 2019
• Published date: 01 September 2019

RAND Journal of Economics

Vol.50, No. 3, Fall 2019

pp. 615–644

Valuing product innovation: genetically

engineered varieties in US corn and soybeans

Federico Ciliberto∗

GianCarlo Moschini∗∗

and

Edward D. Perry∗∗∗

We develop a discrete-choice model of differentiated products for US corn and soybean seed

demand to study the welfare impact of genetically engineered(GE) crop varieties. Using a unique

data set spanning the period 1996–2011, we find that the welfare impact of the GE innovation is

significant. In the last five years of the period analyzed, our preferredcounterfactual indicates that

total surplus due to GE traits was $5.18 billion per year, with seed manufacturers appropriating

56% of this surplus. The seed industry obtained more surplus from GE corn, whereas farmers

received more surplus from GE soybeans.

1. Introduction

Innovation, in the form of new and improved crop varieties, has long played a critical role in

the quest to ensure sufficient food supply for a rapidly growing world population. Conventional

breeding activities have led to remarkable successes, such as hybrid maize (Griliches, 1957) and

the green revolution (Evenson and Gollin, 2003). Genetically engineered (GE) crop varieties

build on this tradition by exploiting the recombinant DNA tools of modern biotechnology. First

introduced commercially in 1996, by most standards, GE varieties have been very successful

(Moschini, 2008). Despite being essentially limited to four main crops (maize, soybean, cotton,

and canola), as of 2017, GE varieties were grown on more than 469 million acres worldwide.The

United States has been at the forefront of these developments: in 2017, GE varieties wereplanted

on more than 183 million acres of US farmland, nearly 91% of which was maize and soybeans

(ISAAA, 2017).

∗University of Virginia,CEPR, London, and DIW, Berlin; fc3p@virginia.edu.

∗∗Iowa State University; moschini@iastate.edu.

∗∗∗Kansas State University; edperry@ksu.edu.

We thank twoanonymous referees and Editor Chad Syverson for helpful comments and suggestions. This research was

supported by the National Institute of Foodand Ag riculture, US Department of Agriculture, grant no. 2015-67023-22954.

Ciliberto also thanks the Bankard Fund for Political Economy at the University of Virginia for financial support, and

Moschini gratefully acknowledges the support of the Pioneer Endowed Chair in Science and Technology Policy at Iowa

State University.

C2019, The RAND Corporation. 615

616 / THE RAND JOURNAL OF ECONOMICS

Notwithstanding their productivity-enhancing potential, GE crops have been highly contro-

versial. Concerns raised include the fear that GE products are harmful to human health and/or the

environment, and ethical objections related to human manipulation of the DNA of living plants

and animals. Many of these concerns have been allayed (Bennett et al., 2013). In particular, the

environmental impacts of GE varieties appear to be generally positive (NRC, 2010; Bar rows,

Sexton, and Zilberman, 2014; Perry, Moschini, and Hennessy, 2016). However, a separate per-

sistent source of public mistrust relates to the ownership interests of multinational corporations

that commercialize GE products. Unlike innovations underpinning the green revolution, which

were largely the result of publiclysponsored research and development (R&D) activities (Wright,

2012), GE crop varieties havebeen primarily developed by private firms, with US seed companies

(Monsanto in primis) at the forefront. The proprietary nature of GE technologies, and an ongoing

consolidation of the seed and agrochemical industry, has heightened concerns about the pricing

of these new products, their contribution to welfare, and the actual beneficiaries of the innovation

(Clancy and Moschini, 2017).

In this article, we provide noveleconometric evidence on the welfare effects of the introduc-

tion of GE crop varieties. We draw on a large, proprietary data set of plot-level seed choices by

a representative sample of US farmers for the two most important GE crops, corn and soybeans.

The data span the period from 1996 (the year GE corn and soybean varieties werefirst introduced)

to 2011 (by which time the average adoption rate of GE varieties exceeded 90%), and contain

information on the specific seed products that farmers buy—brand, amount bought, area planted,

price paid, and which (if any) GE traits are included in the seed. The richness of the data allows

us to estimate an explicit structural model of farmers’ demand for seed varieties rooted in the

theory of discrete choice in a differentiated product setting (Anderson, De Palma, and Thisse,

1992). Although this model pertains to a production input (seeds) used by competitive firms,

rather than consumer products, it is nonetheless in the tradition of the empirical industrial orga-

nization (IO) literature on demand estimation in industries with differentiated products (Berry,

1994; Goldberg, 1995; Berry, Levinsohn, and Pakes, 1995; Nevo, 2001). This demand model

provides the structural foundation for evaluating the welfare impacts of the introduction of new

characteristics—GE traits—into seed products, along the lines of the seminal contributions of

Trajtenberg (1989) and Petrin (2002).

The discrete-choice model of seed demand that we specify and estimate presumes individual

profit-maximizing choices, with farmers modelled as choosing between all corn and soybean

varieties (in addition to the outside option). Specifically, we model the demand for corn and

soybean seed products using a two-levelnested logit specification (Verboven, 1996; Bj¨

ornerstedt

and Verboven, 2016). The upper level consists of the outside option (planting a crop other than

corn or soybeans, or not planting at all) and the set of inside options, the latter encompassing all

corn and soybean seed products. The inside options are partitioned into two subgroups, one for

soybean seed products and the other for corn seed products. This two-level nested specification

is particularly suited to the institutional realities of US corn and soybean production, including

the role played by the widespread practice of crop rotation.

Estimates from this demand model allow us to infer the willingness-to-pay(WTP) of far mers

for seed products over time, and,more specifically, for the GE traits progressively embedded into

seed varieties. The total WTP provides a first-order approximationto the ex post total sur plus cre-

ated by the innovation.We find that the introduction of GE traits in corn and soybeans, overthe pe-

riod 1996–2011, increased total surplus by $30.6 billion. Using observed price premia commanded

by GE varieties, we estimate that seed companies’ revenue increased by about $24.3 billion, sug-

gesting that innovating firms captured the larger share of the surplus created by the innovation.

Next, we implement an alternative, more structural, procedure that can accounts for two ad-

ditional crucial effects: the contribution of GE varieties to increased seed product differentiation

in the industry (which, ceteris paribus, is valuable to users and a potential source of additional

revenues for sellers), and the competitiveprice effects caused by the innovation itself. Specifically,

we use the structure of the estimated demand model to construct and simulate counterfactual

C

The RAND Corporation 2019.