13 Pages
English
Gain access to the library to view online
Learn more

Auctioning green investment grants as a means of accelerating environmental innovation - article ; n°1 ; vol.83, pg 99-110

-

Gain access to the library to view online
Learn more
13 Pages
English

Description

Revue d'économie industrielle - Année 1998 - Volume 83 - Numéro 1 - Pages 99-110
Outre les incertitudes relatives aux bénéfices privés, l'abandon volontaire d'une technologie polluante peut être ultérieurement retardé par la perspective d'une réduction des coûts d'investissement liée à la diffusion du nouveau mode de production. Dans cet article, seront analysées les implications de ces sources d'inertie sur la conception d'aides publiques à l'investissement ayant pour but d'accélérer l'adoption de technologies vertes. Bien que la perspective d'une réduction des coûts d'investissement tende à ralentir le changement technologique spontané, il sera montré que l'existence d'effets externes positifs, liés à la diffusion du nouveau mode de production, offre au régulateur la possibilité de réduire les dépenses publiques en octroyant les subventions aux entreprises ayant des coûts inférieurs d'innovation, au lieu de subventionner, sans discrimination, tout le secteur concerné.
Besicles uncertainty about on-going private benefits, voluntary abandonment of a polluting technology may be (further) delayed if switching costs are expected to decline over time, because of positive extemalities stemming from the diffusion of the new (« green ») production process. In this paper we examine the implications of these sources of inertia on the design of investment grants aimed at accelerating environmental innovation. We show that although « network externalities » tend to decelerate spontaneous (decentralized) technological change, they provide the regulator with the opportunity of saving public funds, by targeting grants to agent(s) with lower switching costs, instead of subsidizing the entire industry indiscriminately.
12 pages
Source : Persée ; Ministère de la jeunesse, de l’éducation nationale et de la recherche, Direction de l’enseignement supérieur, Sous-direction des bibliothèques et de la documentation.

Subjects

Informations

Published by
Published 01 January 1998
Reads 45
Language English

Exrait

Cesare Dosi
Michele Moretto
Auctioning green investment grants as a means of accelerating
environmental innovation
In: Revue d'économie industrielle. Vol. 83. 1er trimestre 1998. pp. 99-110.
Résumé
Outre les incertitudes relatives aux bénéfices privés, l'abandon volontaire d'une technologie polluante peut être ultérieurement
retardé par la perspective d'une réduction des coûts d'investissement liée à la diffusion du nouveau mode de production. Dans
cet article, seront analysées les implications de ces sources d'inertie sur la conception d'aides publiques à l'investissement ayant
pour but d'accélérer l'adoption de technologies vertes. Bien que la perspective d'une réduction des coûts d'investissement tende
à ralentir le changement technologique spontané, il sera montré que l'existence d'effets externes positifs, liés à la diffusion du
nouveau mode de production, offre au régulateur la possibilité de réduire les dépenses publiques en octroyant les subventions
aux entreprises ayant des coûts inférieurs d'innovation, au lieu de subventionner, sans discrimination, tout le secteur concerné.
Abstract
Besicles uncertainty about on-going private benefits, voluntary abandonment of a polluting technology may be (further) delayed if
switching costs are expected to decline over time, because of positive extemalities stemming from the diffusion of the new («
green ») production process. In this paper we examine the implications of these sources of inertia on the design of investment
grants aimed at accelerating environmental innovation. We show that although « network externalities » tend to decelerate
spontaneous (decentralized) technological change, they provide the regulator with the opportunity of saving public funds, by
targeting grants to agent(s) with lower switching costs, instead of subsidizing the entire industry indiscriminately.
Citer ce document / Cite this document :
Dosi Cesare, Moretto Michele. Auctioning green investment grants as a means of accelerating environmental innovation. In:
Revue d'économie industrielle. Vol. 83. 1er trimestre 1998. pp. 99-110.
doi : 10.3406/rei.1998.1702
http://www.persee.fr/web/revues/home/prescript/article/rei_0154-3229_1998_num_83_1_1702_ NORMES ET STANDARDS II.
ENVIRONNEMENTAUX :
INCITATIONS À L'INNOVATION ? Cesare DOSI
Michèle MORETTO
University of Padova
and Fondazione ENI
Enrico Mattei
AUCTIONING GREEN INVESTMENT
GRANTS AS A MEANS
OF ACCELERATING ENVIRONMENTAL
INNOVATION*
Mots clés : Volonté d'innovation environnementale « Green investment timing »
Subventions d'investissement, enchères bayesiennes.
Key words : Voluntary Environmental Innovation, Green Investment Timing, Investment
Grants, Bayesian Auction.
I. — INTRODUCTION AND THE SCOPE OF THE PAPER
The increasing number of firms trying to gain a reputation as environmentall
y friendly companies - by voluntarily abandoning polluting technologies or by
overmeeting environmental standards - has somewhat undermined the convent
ional wisdom about the inevitable trade-off between the social benefits and
private costs of pollution abatement (Porter and Van der Linde, 1995).
However, whether or not available empirical evidence of spontaneous environ
mental innovation justifies the increasing optimism about the capacity of pro
fit-expectations to activate self-regulatory mechanisms, is still controversial.
Here we do not intend to question the so called «win-win» paradigm which
is becoming increasingly popular not only among industrialists and their lob
bies, but among policy-makers and economists as well. Rather, we wish to
draw attention to the timing of spontaneous environmental innovation. For ins
tance, even when firms have discovered theoretically profitable opportunities
(*) We thank Carlo Carraro, Gerard Mondello, and participants at the Workshops
«Dynamique Industrielle & Constraintes Environnementales», Sophia- Antipolis, May 12-
13, 1997 and «2nd Toulouse Conference on Environment and Resource Economics»,
Toulouse, May 14-16, 1997, for valuable comments. The usual disclaimer applies.
REVUE D'ÉCONOMIE INDUSTRIELLE — n° 83, 1er trimestre 1998 99 from «green» investments, various sources of inertia may slow down volunta
ry abandonment of polluting technologies, so that public intervention - such
as the provision of innovation grants - is still required to avoid undesirable
levels of pollutant accumulation (Dosi and Moretto, 1996a, 1997).
Besides uncertainty about the innovation's appropriable benefits (whether,
and to what extent, consumers will actually be willing to reward a «green»
company ? How environmental regulation will evolve and to what extent it
will be affected by spontaneous innovations, etc.), an irreversible technologic
al change may be further delayed if switching costs are expected to decline
over time, because of a learning effect or other positive «network» externalit
ies stemming from the diffusion of the new (green) production process.
In this paper we examine the impact of these sources of inertia on the privat
e timing of switching to a green technology, and their implications on the
design of investment grants aimed at accelerating the technological change.
We show that although the expectation of a declining switching cost tends to
decelerate spontaneous (decentralized) innovation, it provides the regulator
with the opportunity of saving public funds, by targeting grants to agent(s)
with lower switching costs, instead of subsidizing the entire industry indiscri
minately. However, this requires knowledge about private switching costs.
Otherwise, appropriate incentive mechanisms are required to minimize agents'
information rents. To find a cost-effective «screaming» mechanism, we will
consider a Bayesian auction, where firms are required to declare their optimal
private switching time (or, equivalently, their optimal «trigger value»), and the
subsidy is granted to the firm which announces the lowest one.
II. — THE PRIVATE INVESTMENT TIMING
1. Basic assumptions
For the sake of simplicity we consider only two firms (i = 1,2), belonging to
the same industry, which have to decide whether to keep their present (identi
cal) polluting technology, or to switch to a green one. We assume the two tech
nologies are incompatible and the decision to switch is irreversible.
Besides improving environmental quality (a public good), the technological
change is expected to provide appropriable benefits (higher revenues and/or
lower variable costs). However, firms face uncertainty about the actual green
technology's (relative) operating benefit (hereafter «the benefit») per unit of
time.
100 REVUE D'ÉCONOMIE INDUSTRIELLE — n° 83, 1" trimestre 1998 • Assumption 1. The benefit per unit of time is described by a geometric
Brownian motion (1) :
dxt = ax,dt + 0XtdBt with a, a> 0 and x0 = x. (1)
where dB, is the increment of a standard Wiener process.
Firms have the option to switch by affording a sunk adoption cost, C(6, k).
Such a cost, which accounts for the pure capital expenditure, k, as well as the
opportunity cost of forgone investment opportunities, 6, depends on the numb
er of firms that have adopted.
• Assumption 2. Ct(6,n) = d^n), i =1,2 and n = 1,2 with k(\) > k(2).
•3. Whilst k(2) and k(l) are common knowledge, 0,, reflecting
agent i's perception of the technological change's opportunity cost, is private
information and takes on values in Q = [0,6] œ R+ with cumulative distribu
tion G(6J, and density g( 6J, which are public knowledge. «Types» are inde
pendent between firms, so they do not convey information about the other
agent's private valuation parameter.
2. The waiting game
Because of the sunk nature of switching costs, uncertainty about xt tends to
slow down abandonment of the polluting technology: i.e., it is worthwhile to
wait before exercising the option to switch alive in the hope of getting more
information about the technological change's actual rentability {irreversibility
effect).
Moreover, as shown hereafter, the technological change may be further
delayed by the expectation of a fall in switching costs, due to the diffusion of
the environmental innovation, as long as the valuation parameter 6 is private
information. In fact, while the «network benefit» [k(l) - k(2)] would make
coordination advantageous, uncertainty about the other agent's investment
opportunity cost may lead to a war of attrition: i.e. each firm will find it wor
thwhile to wait longer before exercising its option to switch, in the hope that
the rival firm will innovate first, so as to gain the network benefit by following
suit. If this does not happen and it becomes clear that the rival is reluctant to
adopt the green technology, the agent may eventually decide to go first.
(1) For simplicity, we assume that the evolutionary pattern of the green technology's benefit
is independent of the number of agents who have adopted the technology.
REVUE D'ÉCONOMIE INDUSTRIELLE — n° 83, V trimestre 1998 J Q J It is important to point out two features of the model. First, at each instant
the players observe the realization of the state variable xt, and conditional on
their private valuation parameter 6 decide whether to adopt the green techno
logy. Second, there is a Bayesian learning process where agents learn via expe
rimentation. A Nash equilibrium will then be the solution of a pair of linked
stopping time problems, where each agent solves his switching problem by
taking account of his rival's possible actions and learning about the other's pri
vate valuation parameter from the fact that the latter has not switched up to that
moment.
As will be argued below, the Markovian nature of (1) (the shocks are serial
ly uncorrelated) and the Bayesian learning process ensure that the optimal stra
tegies need only be conditioned on the current state of xt and not on past
values. That is, each firm will optimally select, conditional on current info
rmation about the state variable xt and the distribution G, a stationary upper
trigger level x\ e X = [x1, X] cz R+. Thus, if at time t, xt > x*( and the other agent
has not yet switched, he unilaterally switches. On the contrary, if the rival has
switched at xt < x*,, agent i learns that he can go second by paying k(2).
However, although the operating decision requires instantaneous responses,
the certainty of being second does not imply switching immediately after the
first has switched. As the opportunity cost of depends on 6 and xt is
exogenous to the agents' actions there always exists a lower x**¡ < x*¡ , below
which the only dominant strategy for the firm is to keep the option to abandon
the polluting technology alive and wait longer before exercising it. Only when
xt crosses x\ the agent considers the possibility of switching second. As a result
we may have a situation where jc", < xt <x\ for ¿=1,2, and t > 0, so both players
wait for the other to change technology first. That is, a period of excess iner
tia where both users are willing to adopt the new technology if the other does
so first, but neither wants to be the leader (Farrell and Saloner, 1985). Yet,
during this period the player faces an opportunity cost of being stuck with the
polluting technology far longer than he wishes, a cost which increases the
option value of waiting.
Defining with the mappings a, = {a¿(t), t > 0}e Ah where a,{t) : {xt,G} — >
{(a), (b), (c)} are the strategies of agent / at time t, the countervailing interest
of the agents may be represented in continuous time by a policy rule of the fo
llowing type:
0<xt<x** (2)
Ua) if
x**<xt<x* for ¿ = 1,2 i>0. al(xt,G) = \(b) if
xt>x*
(c) if
At every time / > 0, each agent is induced to follow a bandwagon strategy by
choosing one of the following three actions: (a) never switch, regardless of the
1 02 REVUE D'ÉCONOMIE INDUSTRIELLE — n° 83, 1" trimestre 1998 other agent's behavior, (b) switch if the other has already done so, i.e. «jum
ping on the bandwagon»; (c) unilaterally switch, «initiating the bandwagon».
As the optimal choice by agent i of x, (and jc*, ) must take account of the fact
that as xt moves randomly he updates his conditional distribution of the other
firm's valuation parameter 0, and hence of the trigger values jc*; , / ïj, making
use of the distributional strategy approach proposed by Milgrom and Weber
(1985), we are able to define a continuous distribution function for agent /s
trigger level conditionalon the information available to agent / at time zero,
F(Xj ; x) (with density /fxy ; x)), and then place assumptions on the joint distr
ibution of (6, x*) on 0 x X so that the marginal distribution with respect to 0 is
the one specified by the prior beliefs G.
Making use of F(x*j ; x), the agent i's option value at time zero of investing
at time T, if agent j is still using the polluting technology and agent /s strate
gy is Tj is given by :
- 6lk(2)e~rT' \ }|
+ Pr(7; < TJ)E0{lÇxte-rtdt-eik(X)e-
In other words, the investment option is given by the option value if the
agent does not innovate till time Tj and then switches second at cost 6¿k(2),
plus the option value of staying out till time T{ and then going first, where Tj =
inf(i > 0 I xt =x\), i = 1,2, are the stochastic switching times at which the two
players decide unilaterally to change technology.
Although the above expression is evaluated at the normalized time zero
when x0 = x, at each time t > 0 agent / observes the realization of the state
variable xt, updates his conjectures on the other agent's threshold Ft(Xj ; xt) and
considers switching, if the rival has not switched up to that moment, maximiz
ing (3). However, with xt as the only information pattern, F, as the updated
conjectural distribution and the Markovian nature of (1), it will be argued
below (proposition 1) that the only credible Sub-game Perfect Nash
Equilibrium is constituted by a stationary policy rule a¡(F) of type (2).
Proposition 1 (i) When agents have access to continuous information pat
terns, the symmetric Sub-game Perfect Nash equilibrium involves each agent
planning to play the following stationary bandwagon strategy :
0<x,<x**
(a) if
x** <xt< x* for ¿ = 1,2 t > 0. (b) if
xt>x*
(c) if
D'ÉCONOMIE INDUSTRIELLE — n° 83, 1er trimestre 1998 103 REVUE Where the optimal stationary trigger levels x\ and x*\ e [xl, (ii^
0 < x*¡ < x*¡ are given by :
x**
ß
-í- O = -Ü- ß1 (r-a)k(2),
(5)
ß-i
Proof: See Dosi and Moretto (1996b, appendix B).
when It is agents worth face noting attrition that the is greater trigger level than the for one adopting where the green agent technology only faces
uncertainty about the ongoing innovation's private benefits,
x*
ß -1- = -tL-(r - a)k(X) (Dosi and Moretto, 1997).
For instance, by postponing abandonment of the polluting technology the
agent hopes that his rival will adopt the green technology first, thus gaining the
network benefit k(l)-k(2). (2)
As far as excess inertia is concerned, from proposition 1 we can conclude
that :
Corollary 1 While strategic behavior does not reduce the time the player
must wait before he considers adopting the green technology (strategy b), it
decelerates initiation of the bandwagon (strategy c), increasing the period of
inertia.
As long as the trigger value x*¡ is an increasing function of the private valua
tion parameter 6i we can also prove the following corollary :
Corollary 2 The higher 6 the later the agent will adopt the green technolo
gy (monotonicity property of the trigger value) :
, and x*(6>)e[jt'(0),r(0)], for i = 1,2.
Proof: See Dosi and Moretto (1996b, appendix C).
(2) Lambrecht and Perraudin (1994) find a similar result for a cost-preemption game. Moretto
(1996) extends the above results considering an option game with both war of attrition and
preemption.
1 04 REVUE D'ÉCONOMIE INDUSTRIELLE — n° 83, 1er trimestre 1998 — PUBLIC INCENTIVES TO ACCELERATE III.
ENVIRONMENTAL INNOVATION
1. The regulatory problem
For the sake of simplicity, let us assume that, under the current technology,
each firm produces an identical and constant amount of emissions per unit of
time, affecting two independent environmental media (e.g. two water bodies
located in different hydrological regions). Emissions, which cannot be avoided
unless the green technology under consideration is adopted, accumulate over
time following a deterministic process. Finally, let us assume that public
authorities wish to keep the pollutant stock below a given («critical») thre
shold (F), identical for both environmental media. Without technological chang
e, P will be reached at time T
Under our assumptions, public intervention is required if the date at which
firms would find it appropriate to voluntarily abandon the polluting technolo
gy goes beyond T. However, while under complete information it is possible to
compare the private (7) and the policy-maker's desired (ultimate) switching
time (7), the situation is different when the private innovation's benefits are
uncertain. In fact, jc's volatility implies that the firm's switching time becomes
a stochastic variable. Consequently, the regulator has to identify a policy-rule
referring to F s probability distribution.
To keep the analysis as simple as possible, we assume the regulator aims at
ensuring that, at least in terms of expected value, both firms will abandon the
current technology before the date at which T is expected to be reached.
That is :
E(T) = f-m ^
where m > 0 is a predefined constant which takes on the sense of a «safety
interval». By (1) and the definition of private switching time T, the above poli
cy-objective may be reformulated in terms of jc's trigger value, i.e. the per unit
of time operating benefit at which the technological change will take place.
Denoting with i the «social» trigger value, this is given by - _ , ~u x " x {m }
(Cox and Miller, 1965, pag. 221).
Among the various policy instruments which could be adopted to induce
firms to revise their investment plans, we will focus on the provision of invest
ment grants. For instance, subsidization of environmental innovations is quite
common, because of its popularity among industrialists as well as politicians.
As long as abandonment of a polluting technology is slowed down by invest
ment «reversibility and uncertainty about ongoing private benefits, subsidizat
ion, to be effective, has to be designed so as to adequately compensate firms
for giving up their option to wait for new information to arrive. The compen-
REVUE D'ÉCONOMIE INDUSTRIELLE — n° 83, 1er trimestre 1998 required increases even more in the presence of spillover effects on sation
switching costs, because firms' expectations of gaining the network benefit
makes the option to wait even more valuable.
However, as long as subsidies are financed though distortionary taxation, a
trade-off may emerge between the social benefits deriving from speeding up
the environmental innovation, and the social costs of regulation.
More formally, let us express the regulation's ex-post net value as follows :
B-(l + X)s(x) + R(x) (7)
where B is the social benefit brought about by accelerating abandonment of the
polluting technology; s(x) is the investment grant; X > 0 is the shadow cost of
public funds; R(x) denotes the firm's (subsidized) rental price.
It is quite obvious that maximization of (7) requires targeting the subsidy to
the agent with the lower private valuation parameter 6, rather than subsidizing
the entire industry. In fact, by accelerating initiation of the «bandwagon», the
regulator can induce the whole industry to spontaneously abandon the pollu
ting technology before reaching the pollution threshold P. However, to fully
exploit the potential regulatory benefits resulting from (network) externalities,
knowledge of the individual parameters 6 is required. If, as we assume here,
these parameters are private information, appropriate incentive mechanisms
are required, to minimize agents' information costs and, consequently, the
regulation's social costs.
2. Auctioning green investment grants
Since the regulator does not know the private valuation parameters 9h he/she
is unable to identify the true optimal trigger values x*¿(0¿). To find the best fea
sible incentive mechanism inducing both agents to abandon the polluting tech
nology the first time xt, randomly fluctuating, hits Jc, we consider a Bayesian
auction where each firm announces simultaneously its trigger level (x*7, x*2)
and the subsidy is granted to the firm which announces the lowest one.
The subsidy under consideration is formed by the sum of a fixed payment
function (individual rational transfer) - defined according to the difference
between the announcement x*( andi- plus a linear sharing of overruns which
depends on the announced trigger value. If this subsidy is incentive compatible
it will be sufficient to induce the firm announcing the lower rental price to
adopt the green technology at x , and then, by the network benefit, the other
firm will also abandon the polluting technology at jc (the case of sequential
investment will be considered in section 3.3).
Without loss of generality, we may assume that the regulator knows the
agents' conjectural distribution, so that, conditional on the information avai
lable at the time when the subsidy is offered, i.e. xt = x , for the regulator the
REVUE D'ÉCONOMIE INDUSTRIELLE — n° 83, 1er trimestre 1998 agents' optimal trigger levels are drawn independently from the continuous
distribution F(x\ , x).
• Assumption 4. f ( -* . ^
is nondecreasing in x¿ (monotone hazard-rate).
Defining the subsidy to induce adoption of the new technology at x, s^x^Xj)
as a function of the announced trigger levels xt, agent i's expected (subsidized)
rental price can be expressed as :
(x;^J)-yl(rrXJ)(x;-x)}, for ¿ = 1,2 (8)
where y^x^Xj) is the probability that agent / is selected to receive the subsidy,
with 2
for any x¡ and x7. A mechanism that induces a truth-telling Bayesian Nash equi
librium is an incentive compatible mechanism where the subsidy designed for
agent / with trigger level x, is the one he prefers in the menu [s¿(.), y¿(.)] offe
red by the regulator.
Going back the regulatory problem, while under complete information the
planner maximizes (7) by selecting the firm with the lowest trigger value x\
(and subsidizes only that agent), under incomplete information tl>e ex-ante
maximand becomes :
2 2 (9)
J=l J i=\ i = l
or :
i J i=i ¡=i
As in the above objective function Rt is considered the state variable, maxi
mization with respect to y¿ and R¡ yields :
Proposition 2 In the case of two agents whose private trigger values are
drawn independently from the same continuous distribution with monotone
hazard-rate, an optimal Bayesian auction will give the subsidy to the agent
with the lowest trigger level. The optimal choice will be :
yi(xi;xj) = l if x¡ <Xj
yi(x*;x*) = 0 if x*>x]
and the expected transfer to the agent is :
REVUE D'ÉCONOMIE INDUSTRIELLE — n° 83, 1er trimestre 1998 1 07