CHILDREN AND FAMILIES
EDUCATION AND THE ARTS
ENERGY AND ENVIRONMENT
HEALTH AND HEALTH CARE
INFRASTRUCTURE AND
TRANSPORTATION
INTERNATIONAL AFFAIRS
LAW AND BUSINESS
NATIONAL SECURITY
POPULATION AND AGING
PUBLIC SAFETY
SCIENCE AND TECHNOLOGY
TERRORISM AND
HOMELAND SECURITY
The RAND Corporation is a nonprofit institution that helps improve policy and decisionmaking through research and analysis.
This electronic document was made available from www.rand.org
as a public service of the RAND Corporation.
Skip all front matter: Jump to Page 1
Browse Reports & Bookstore
6
Make a charitable contribution
Visit RAND at www.rand.org
Explore the Pardee RAND Graduate School
View document details
Limited Electronic Distribution Rights
This document and trademark(s) contained herein are protected by law as indicated in a notice appearing later in this work. This electronic representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of RAND electronic documents to a non-RAND website is prohibited. RAND electronic documents are protected under copyright law. Permission is required from
RAND to reproduce, or reuse in another form, any of our research documents for commercial use. For information on reprint and linking permissions, please see RAND Permissions .

This product is part of the Pardee RAND Graduate School (PRGS) dissertation series.
PRGS dissertations are produced by graduate fellows of the Pardee RAND Graduate
School, the world’s leading producer of Ph.D.’s in policy analysis. The dissertation has been supervised, reviewed, and approved by the graduate fellow’s faculty committee.

This document was submitted as a dissertation in April 2013 in partial fulfillment of the requirements of the doctoral degree in public policy analysis at the Pardee
RAND Graduate School. The faculty committee that supervised and approved the dissertation consisted of Siddhartha Dalal (Chair), Nicholas Burger, and
Robert Lempert.
PARDEE RAND GRADUATE SCHOOL

The Pardee RAND Graduate School dissertation series reproduces dissertations that have been approved by the student’s dissertation committee.
The RAND Corporation is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND’s publications do not necessarily reflect the opinions of its research clients and sponsors.
R ® is a registered trademark.
Permission is given to duplicate this document for personal use only, as long as it is unaltered and complete. Copies may not be duplicated for commercial purposes. Unauthorized posting of RAND documents to a non-RAND website is prohibited. RAND documents are protected under copyright law. For information on reprint and linking permissions, please visit the RAND permissions page (http://www.rand.org/publications/permissions.html).
Published 2013 by the RAND Corporation
1776 Main Street, P.O. Box 2138, Santa Monica, CA 90407-2138
1200 South Hayes Street, Arlington, VA 22202-5050
4570 Fifth Avenue, Suite 600, Pittsburgh, PA 15213-2665
RAND URL: http://www.rand.org
To order RAND documents or to obtain additional information, contact
Distribution Services: Telephone: (310) 451-7002;
Fax: (310) 451-6915; Email: order@rand.org

Summary
The   extent   to   which   technological   change   will   help   mitigate   climate   change   is   a   subject   of   some   debate   among   climate   policy   researchers   and   stakeholders.
In   particular,   models   predicting   the   costs   and   effectiveness   of   climate   policies   rarely   include   the   effects   of   technological   change   because   it   is   difficult   to   predict   how   much   change   will   occur.
This   study   takes   an   empirical   approach   to   filling   this   gap   in   the   case   of   solar   and   wind   energy   research.
In   order   to   do   so,   a   new   set   of   data   is   collected   to   serve   as   a   proxy   for   technological   change,   focusing   on   research   production:   the   number   of   technical   journal   articles   on   solar   or   wind   energy   published   each   month.
A   combination   of   hand ‐ sorting   and   Bayesian   logistic   modeling   is   used   to   identify   these   articles,   which   number   in   the   tens   of   thousands.
The   resulting   monthly   article   counts   are   used   to   assess   how   solar   and   wind   research   production   may   have   responded   to   major   U.S.
renewable   energy   policies.
The   first   essay   focuses   on   these   empirical   findings,   the   second   on   methodological   comparisons,   and   the   third   on   subcategories   of   solar   energy   and   their   potential   relationships   with   policy.
Both   direct   and   indirect   subsidies   are   found   to   be   associated   with   increases   in   research   production,   as   described   in   the   first   essay.
Indirect   subsidies   are   represented   by   the   largest
U.S.
federal   support   for   renewable   energy,   the   Production   Tax   Credit,   considered   in   combination   with   the   renewable   energy   Investment   Tax   Credit.
Most   of   the   value   of   these   credits   is   used   for   wind,   although   they   are   also   available   for   solar.
These   credits   are   widely   considered   to   drive   installations   of   wind   turbines,   but   their   effects   on   research   were   previously   unmeasured.
Increasing   the   tax   credits   by   $20   million   per   year   (1%   of   their   2008   value)   is   found   to   be   associated   with   a   1%   increase   in   solar   article   counts   and   positive   but   statistically   insignificant   effects   on   wind   article   counts.
To   the   author’s   knowledge,   this   is   the   first   study   to   successfully   quantify   these   tax   credits’   relationship   with   innovation.
Also   investigated   is   the   impact   of   federal   solar   or   wind   research   funding,   i.e.
direct   subsidies   for   research.
For   every   $1   million   dollars   spent   on   solar   or   wind   research,   solar   or   wind   article   counts   increase   by   1 ‐ 2%.
Although   the   impact   of   research   funding   on   research   is   intuitive,   this   result   contradicts   the   hypothesis   that   public   funds   might   simply   displace   private   research   funding   with   no   net   ix

increase   in   research.
These   findings   can   be   used   as   inputs   to   climate   policy   models   in   order   to   calibrate   the   effects   of   renewable   energy   production   subsidies   and   research   subsidies   on   solar   and   wind   energy   research.
The   second   essay   focuses   on   methods   of   measuring   innovation   and   how   they   affect   the   results.
Three   measures   of   innovation—article   counts   from   the   first   essay,   article   counts   based   on   keyword   selection,   and   patents—are   constructed   and   subjected   to   the   same   analysis   as   in   the   first   essay.
Patents,   selected   using   either   patent   classes   or   by   searching   for   keywords,   are   the   leading   measure   of   innovation   for   economic   and   policy   analyses.
Similarly,   keyword   selection   is   the   usual   method   of   finding   and   counting   articles,   which   has   been   done   in   many   other   contexts   including   examination   of   topical   trends   within   wind   and   solar   energy   research.
Testing   against   a   hand ‐ sorted   sample   of   articles   indicates   that   Bayesian   regression   and   keyword   selection   are   roughly   similarly   successful   at   identifying   relevant   articles,   although   assessment   methods   used   favor   keyword   selection.
Bayesian   article   selection   identifies   more   articles   and   may   be   more   reliable,   but   keyword   selection   is   faster   and   in   this   case,   both   types   of   article   counts   give   substantively   analogous   results   when   regressed   with   tax   credits   and   subsidies.
These   findings   suggest   that   it   is   at   least   sometimes   possible   to   choose   keywords   which   work   approximately   as   well   as   the   Bayesian   regression   models   used.
Thus,   reasons   for   selecting   one   method   over   the   other   in   the   future   may   be   case ‐ specific.
For   both   wind   and   solar   energy,   patents’   associations   with   research   funding   are   very   similar   to   the   associations   for   articles,   while   their   associations   with   the   tax   credits   are   not   statistically   significantly   different   from   zero   or   from   article   count   results.
Patents’   failure   to   find   an   association   with   tax   credits   when   article   counts   do   find   such   a   result   may   be   caused   in   part   by   patents   being   far   fewer   in   number   than   articles.
These   findings   can   be   interpreted   to   suggest   that   article   counts   are   as   valid   a   measure   of   innovation   impacts   as   patent   counts,   and   are   more   effective   at   measuring   small   effects.
Further   suggestions   for   future   research   methods
are   discussed.
Again   using   Bayesian   regression   and   keyword   approaches,   the   third   essay   investigates   trends   over   time   in   the   two   main   subcategories   of   solar   energy   research:   monocrystalline   silicon   and   thin   film.
While   demonstrating   the   methods   with   more   challenging   article   topics,   x

this   study   also   offers   preliminary   findings   on   whether   public   policies   have   affected   the   distribution   of   research   effort   between   monocrystalline   silicon   and   thin   film.
Monocrystalline   silicon,   known   also   as   the   “first   generation”   of   solar   cells,   is   the   technology   used   in   most   commercially   available   silicon   cells   to   date.
Thin   films   are   a   competing   technology   which,   since   they   use   thinner   materials,   are   often   less   efficient   but   potentially   cheaper.
They   are   known   as   the   “second   generation”   of   solar   technologies.
Predictions   that   thin   films   imminently   will   dominate   the   market   date   to   at   least   1985.
This   study   finds   that   thin   film   articles   collected   by   either   method   have   far   outnumbered   monocrystalline   articles   consistently   throughout   1985 ‐
2010,   the   entire   time   period   considered.
U.S.
research   subsidies   appear   to   have   potentially   favored   monocrystalline   silicon   over   thin   film   research   in   the   more   applied   database,   demonstrating   that   research   subsidies   can   be   used   to   steer   net   solar   research.
The   ratio   of   monocrystalline   silicon   to   thin   film   research   appears   to   have   been   otherwise   unaffected   by   the
U.S.
tax   credits   and   research   subsidies,   consistent   with   the   philosophy   that   policies   should
avoid   being   technology ‐ specific.
Taken   together,   these   essays   provide   both   substantive   and   methodological   results.
Firstly,   they   demonstrate   the   substantial   impacts   that   public   policies   can   have   on   the   volume   of   solar   and   wind   innovation,   even   when   innovation   is   not   the   policy’s   direct   target.
They   provide   detailed   results   on   the   relative   sizes   of   these   impacts,   which   can   be   used   as   inputs   to   subsequent   analyses.
Simultaneously,   they   show   that   hand ‐ sorting   followed   by   Bayesian   regression   is   an   effective   way   to   count   large   numbers   of   articles   for   the   purposes   of   social   science   research,   although   carefully   chosen   keywords   may   sometimes   perform   similarly   well.
The   solar   and   wind   article   counts   thus   created   are   also   suitable   for   many   more   studies   than   could   be   conducted   here.
This   dissertation   provides   new   solar   and   wind   article   count   data,   the
methods   used   to   create   it,   and   information   on   how   direct   and   indirect   subsidies   can   drive   solar   and   wind   research.
xi