CO, TAXES AND TRADABLE QUOTAS,
EXPERIMENTAL EVIDENCE OF BIASED DECISIONS
Carla Susana Assuad
Erling Moxnes
System Dynamics Group
University of Bergen, Norway
Fosswinckelsgt.6, 5007 Bergen, Norway
http://www. ifi.uib.no/sd/
ABSTRACT
Developed countries should find cost-effective ways to decrease Green House Gas
(GHG) emissions to comply with their Kyoto Protocol targets by year 2012. The target
can be achieved either by domestic emission reduction or by buying quotas in
international markets. Policy makers have to choose between these policy options and
decide to what extend and when to use them. In democratic countries these choices may
be constrained by limited information and misperception among voters and politicians.
Biases in decisions may occur because the choice of policy is complicated by dynamics
and uncertainty. To explore the possibility of misperceptions we perform a laboratory
experiment where subjects are asked to make these difficult policy decisions during the
12 years preceding 2012. Biases were found.
Key Words: Taxes, misperception, dynamics, climate policy, laboratory experiment
1. INTRODUCTION
Developed countries should find cost-effective ways to decrease Green House Gas
(GHG) emissions to comply with their Kyoto Protocol’ targets by year 2012. The target
can be achieved either by domestic emission reduction or by buying quotas in
international markets. Policy makers have to choose between these policy options and
decide to what extend and when to use them. In democratic countries these choices may
be constrained by limited information and misperception among voters and politicians.
Biases in decisions may occur because the choice of policy is complicated by dynamics
and uncertainty, raising the questions: will nations meet their targest and at what costs?
To explore the possibility of misperceptions we perform a laboratory experiment where
subjects are asked to make these difficult policy decisions during the 12 years preceding
2012.
ee ¥ A A _ ,
The Kyoto Protocol is a convention where 156 countries have committed to reduce green-house gas emissions of at
least 5% from 1990 levels by the year 2012. More information is available at: www.unfecc.int
The greatest challenge is to deal with the task of replacement of existing capital.
Replacements take place at long intervals of time, thus introducing delays. It will also
take time to reduce emissions through retrofits. Appropriate policies must take account
of these delays. Few of the existing papers deal with these dynamics, when discussing
policies for emission reductions’. The most relevant paper seems to be Lecocq ef al
(1998). They studied the impact over time of CO) emission abatement policies. They
use an energy model with two sectors (flexible: housing and rigid: transport). They
conclude that there is a need of early actions to reduce emissions in sectors that need
time to replace old equipment. This is a rare, however, key insight for the problem we
pose in this study.
It seems unlikely that subjects get much guidance from the ongoing debate where the
dynamics of replacement are hardly ever mentioned. From past works like Funke
(1991), Moxnes (1998, 2004) and Sterman (1998a) there is evidence of misperception
of dynamics, for instance misperceptions of delays (Brehmer, 1989, Sterman, 1998b)
and inability to respond properly to feedbacks (Sweeney and Sterman, 2000). People
misconceive and mismanage complex problems and the policies to address those
problems (Meadows, 1999). Besides, there are some experimental studies that support
the hypothesis that people do not have a complete understanding of taxes. Eriksen and
Fallan (1996) study the influence of tax knowledge on attitudes towards taxations. They
found that knowledge accounts for improved perception of fairness of the system.
Bartolome (1995) also found that people misperceive the difference between marginal
and average tax. This evidence and the lack of common information on the dynamics,
make us suspect the existence of learning problems. Hence, we hypothesize biases and
misperceptions when people address the problem of reducing GHG’s.
Using laboratory experiments we investigate the existence of biases when people have
to decide about emission policies (emission taxes and international emission trading) to
comply with an emission target. If we reveal misperceptions this should motivate a next
step to formulate policies to correct biases.
The laboratory experiment is designed as follows. Information about abatement cost is
given to the subjects in terms of a curve showing the long term costs of emission
reductions. We consider two treatments. In treatment | subjects are asked to reduce
emissions for a country by imposing a tax from 2000 until 2012 to reach a given
emission target. Trade in quotas is not allowed. Starting from the same design, we add
in treatment 2 a market for emission quota trading between countries (subjects). Each
market has five players with the same conditions (symmetric). Since the game is
symmetric, proper actions to reach the target emissions require the same tax policy as in
treatment |. In both treatments the discount rate is the same and players are punished if
+ Some studies focus on estimating curves of marginal abatement costs based on aggregate macroeconomic models or
engineering approaches (Ellerman and Decaux, 1998; Criqui ef al., 2002). Other studies analyze and compare
different climate policies using cost-effective and cost-benefit analysis (Yohe and Wallece, 1996; Nordahus, 1994;
Kolstad, 1996). This is namely research regarding emission trading for instance, laboratory experiments have been
used to study market efficiency (Bohm and Carlén, 1999), however, these experiments do not include much
dynamics. The majority of these studies are focused on the discussion of what decision-makers should do and
providing simple heuristics among for makers and politicians.
they do not comply with the target at the end of the period. They get paid according to
how well they perform.
The main question pertains to the participants’ understanding of the dynamics of the
system. Do people increase taxes early enough to reach the emission target? Do people
understand the delays in reducing emissions? Will trade of emission quotas among
countries influence how people set taxes? Interesting observations result. We find that
people tend to set too low taxes to reach the emission target. Market prices for quotas
become significantly higher than taxes, and differences in tax levels motivate quota
trade.
This paper is organized as follows: In section two we describe the method. In section
three we show the results. Discussions are presented in section 4. Conclusions are given
in section 5.
2. THE METHOD
We use a laboratory experiment which allows players to make yearly decisions, and
which simulates the consequences for emissions and for the quota market from year to
year. In line with our hypothesis about misperceptions of the delays involved, it is
particularly important that the simulator captures the dynamics of the emission
reductions. The dynamics we capture here arise from the capital stock turnover and
from retrofits. Furthermore, the model represents the market for quota trading between
countries. To simplify the experiment, the model is highly aggregated.
2.1 MODEL
In order to reach the emission targets the decision makers have two possibilities; they
can either reduce emissions in their own country (domestic reductions) or buy emission
rights in the international quota market (paying for reductions abroad). Reductions in
domestic emissions can be obtained either by replacing worn out capital equipment by
more efficient equipment or by improving the existing equipment (retrofits). The first
measure is called replacement, it is relatively cheap, its potential is limited by the yearly
discard rate, and the emission reductions do not have a lasting effect (an automobile
replaced in 2000 may have to be replaced again in 2011). Retrofit, the second measure,
is more expensive but can be executed in less time than replacement (is independent of
the discard rate) and we assume it has a lasting effect (irreversible within our time
horizon, for instance extra insulation).
2.1.1 Replacement
In general GHG’s from human activities are produced by energy consumption and by
agricultural and industrial processes*, which are influenced by the level of economic
° Carbon dioxide is released to the atmosphere when solid waste, fossil fuels (oil, natural gas, and coal), and wood
and wood products are burned, Methane is emitted during the production and transport of coal, natural gas, and oil.
Methane emissions also result from the decomposition of organic wastes in municipal solid waste landfills, and the
raising of livestock. Nitrous oxide is emitted during agricultural and industrial activities, as well as during
activity. To simplify, we set economic activity (gross domestic product, GDP) constant.
This is a conservative design since it makes it easier to reach the target and the task is
simplified, i.e. it should reduce the tendency towards misperceptions. The rate of
production in a country depends on the existing capital stock and so does the GHG
emissions. The coflow structure’ in the stock and flow’ diagram in figure | illustrates
the dynamics of replacements.
Although models of GHG emissions are often very detailed, they do not include the
dynamics described in Figure 1 (Lecocq, 1998). Lecocq’s STARTS model is an
exception. Although the Lecocq model is built for a different purpose than ours, we
choose a similar design.
Average Life Time
Capital
© Stock )
Investment Rate <_Seraping rate
Relative Emissions
from Capital
Emissions
Emission Order Rate Stk _J Scraping Emissions Rate
from Replacement fom Replacement
Emissions index for
new equipment
Domestic Climate
Policy- Tax
Figure 1 Replacement Dynamics
In figure 1 the Capital is accumulated in a stock (upper rectangular box) which increases
by investments and decreases by the scraping rate (pipe with valve). Scrapping equals
the capital divided by the average life time. When the investment rate (inflow) is equal
to the scraping rate (outflow) the stock is in equilibrium, meaning that no changes take
place in the capital stock. People are investing exactly what is discarded.
Each sector of the economy that constitutes the capital stock, like transport, industry,
commerce, housing and energy, emits CO, and other green house gases. Therefore, a
flow of the capital equipment is a flow of emission capacity as well. While capital is
going in and out of the stock, emission capacity has the same behavior as the capital.
The lower component of figure | illustrated that dynamics.
combustion of solid waste and fossil fuels. Very powerful greenhouse gases that are not naturally occurring include
hydrofluorocarbons (HFCs), perfluorocarbons (PFCs), and sulfur hexafluoride (SF6), which are generated in a variety
of industrial processes.
® Coflow structures are used to keep track of the attributes of various items as they travel through the stock and flow
structure of a system. (Sterman, 2000)
7 Diagram introduced by Forrester (1961)
The investment rate (SIR) times an emission index for new equipment (EINE) defines
the inflow to the stock of emissions (ES). Each unit, leaving the stock of capital, also
leads to a removal from the stock of emissions. The emission index for what is removed
is given by the relative emissions from capital (RER) (average emission intensity of the
existing production capital) times the scrapping of capital, which in a no growth
economy equals the investment rate (SIR). This structure is very useful because it keeps
track of the emission changes due to every new investment and every scrapping.
The differential equation to describe the stock-and-flow diagram is the following
“ES ~ (siR* EINE)-(SIR* RER) (dl)
EINE is determined by the climate policy applied, in this case the carbon tax level.
Thus, emitters are assumed to choose more efficient equipment as the tax increases. If
the tax is set equal to zero, the inflow and the outflow of the emissions stock are the
same and the stock stays constant.
A weakness of the above model is that implicit lifetimes of equipment will be widely
distributed. To get lifetimes that are more narrowly distributed we split up the stocks
into six cohorts. The average life time is set equal to 20 years.
2.1.2 Retrofit
Besides waiting for the capital to be discarded and replaced, existing equipment can be
improved by retrofits. This will be an option for equipment with long remaining
lifetimes and prohibitive costs of early retirement. Retrofits are modeled as follows. The
tax policy and the costs of retrofits define a desired level of retrofits. Since it takes time
to decide, plan and carry out retrofits a actual retrofits follow desired retrofits by a
delay. Retrofits are assumed to be irreversible within the time frame of the experiment.
Thus, if taxes are reduced, all retrofits that have already taken place, will stay in place.
2.1.3 Emission Trading Structure in the Model
The market is modeled as a symmetric network game with five players or countries.
Each player decides on a bid-offer curve each year. These curves are such that at high
hypothetical prices, subjects sell and at low hypothetical prices they buy. A computer
routine ensures that the market equilibrates each year. This routine is like giving
reservation prices to a broker who operates in a perfect market. Once the equilibrium
price is found, the quotas are assigned according to the bid-offer curves for the
individual players.
2.2 EXPERIMENT DESIGN
2.2.1 Task
The experiment starts in year 2000 with initial emissions of 4000 Mtons of CO
equivalents for each player. The goal for the participants is to reach a target of 3000*
Mton of CO» equivalents by 2012 with as low costs as possible. A punishment of 200
$/ton has to be paid if the target is not reach by year 2012. Two treatments are
considered.
Figure 2 shows the player interface for treatment 1. The upper box displays the current
year. The players enter their tax rates in the second box. After the tax rate has been
entered, the subject asks the simulator to advance on year, new decisions are made and
so forth and so on. The third box gives information about the current emissions, and
relates this to the target. The fourth box gives information regarding the total cost
achieved by the participant and the payoff. The payoff depends on the total costs, and
information in the fourth box is only shown at the end of the game (2012).
Figure 3 shows the interface for treatment 2. It is similar to the interface for treatment 1,
with the following exceptions. In the information box there is also information about the
players holding of emission quotas. If a player has bought quotas, the need for
reductions in 2012 is reduced. A negative holding of quotas implies that the need for
reductions increases. A graph is added in the upper left box. By using the mouse the
players are able to move the line and create a bid-offer curve, for instance to buy at low
prices and sell at high.
Subjects were randomly seated in cubicles. In treatment 2 they did not know who they
were competing with. They were paid privately one by one. The appendix shows the
exact instructions given to and read aloud for the participants. Note that the instructions
did not quantify or mention delays. The participants were, however, told that the
underlying computer program was highly realistic with the exceptions mentioned. They
were asked to fill in data for the current year in a handout table to keep a record of the
history, and as a backup of the data.
8 The GHG emission reduction constrains used for this study are base on the estimations of the marginal abatement
cost curve done by Ellerman ef al (1998) in the EPPA model. The region selected to be reproduced by the model is
the European Union (EC-12), must reduce emissions in 2012 to 92% of the quantity they were emitting in 1990
(reference year). According with EPPA the target for EC-12 reductions is 2773 M ton CO». In EPPA the projected
emissions for 2012 are 3901 M ton CO», therefore EC-12 should reduce 1128 M ton by year 2012. In the experiment
we used approximations to this numbers in order to simplify the calculations for the participants.
5 Tax Next Year
‘ [Ston CO2 eq]
‘ ce
S| _. saaterenissonsinaiid
i =
| sestordomesteRetutann2012
g,| cmcruewecosnmens [9]
pe Your Peyatis wx [3]
Figure 2. Treatment 1 interface
g
e |g Tax Next Year
g (8 [Siton CO2 eq ]
g 2 C44
Price [$ton C02 eq]
Quirent Year Emissions 400] Surta Price Last Year |
tacos
ee
E | ca garieacimies 3
sew Recon 012 ald C4
Efe pg] rescmaneconmmen [a] Yarra voc [a]
Figure 3. Treatment 2 interface
2.2.2 Optimal Tax and a Feasible Feedback Strategy
To identify the optimal sequence of taxes, subjects have to solve a complex dynamic
optimization problem with limited information about the model constraints. We start by
assuming full information and use Powersim Solver to find the optimal sequence of
taxes and quota price. Figure 4 shows the results. There is a small difference between
the two prices because they represent different "quality products". The emission quota
maintain its face value until 2012. The tax will lead to some “unnecessary” reductions
in that a few early replacements will have to be repeated before 2012. The optimal path
also reflects an interest rate of 4 percent p.a. on the “loans” needed to finance the
emission reductions.
150 ——“optimattax
Optimal price
90
60
$/ton CO2 equi.
2000 2002 2004 2006 2008 2010
Year
Figure 4. Optimal tax and quota price
In the experiment players do not have the time, tools and most likely not the abilities to
optimize. However, we will show that a simple feedback strategy can be used to get
very close to the optimal tax sequence and total costs. The feedback strategy, however,
requires a minimum understanding of the delays involved.
Step No. 1. Set the initial tax
We start by 40$/ton CO equivalent, which is quite a bit lower than the optimal tax, and
on the low side given the cost curve in the introduction.
Step No. 2 Three years fixed tax
The initial tax is kept for three years. The reductions for the first year are 140 Mtons,
for the second year are 89 Mtons and 62 Mtons for the third year. The effect drops as
retrofits approach their desired level, given by the initial tax rate. The reduction in the
last year denotes a maximum yearly reduction per year in the remaining 9 years if the
initial tax rate is maintained. Assuming an average reduction of 50 Mtons per year and
multiplying with 9 years, we get an expected future reduction of 450 Mtons. Together
with the reduction during the first three years, 291 Mtons, we project a total reduction of
741 Mtons in 2012. This is not enough, and the tax rate must be increased. A linear
approach suggests a tax increase of around 25 percent to reach the target. Since the cost
curve is curving upwards (convex), the tax rate should be increased more than that.
Step No. 3 Repeat
Go back and repeat step 2 with a new tax rate and get data for three new years. Repeat
until 2012 has been reached.
This quite simple procedure leads to a result close to the optimal cost level.
2.2.3 Subjects
The experiment was carried out at the University of Bergen (UiB), Norway, and
Universidad Nacional Sede Medellin, Colombia, with bachelor and master students
from the economics departments.
Thirty subjects from Norway and forty three from Colombia completed the experiment.
Treatment | was accomplished by 28 subjects and treatment 2 by 45 subjects. To avoid
learning effects, no subjects participated more than once.
2.2.4 Hypothesis
In the experiment three hypotheses were tested regarding the tax, the quota prices and
the effect of the market on the tax.
Hi. Average tax equals optimal tax
The alternative hypothesis is a downward bias due to ignorance of delays. There could
also be a downward bias if people dislike taxes or tax increases.
H2. Tax in T1 equals tax in T2
This hypothesis holds that the existence of an emission quota market will not influence
taxes. In T2 all players face the same increasing abatement costs (symmetric game).
Hence, there is no reason to trade, and the tax rates should not be influenced by the
trade option.
We do not state a clear alternative hypothesis. On the one hand, if the market produces
an unbiased quota price, this should serve to bring up a tax rate that would otherwise be
biased downwards. Subjects should react to a large difference between the two “prices”
for emission reductions. On the other hand, the existence of a quota market could be
seen as a safety valve, or as an option to avoid having to use taxes or having to deal
with an option with uncertain outcomes. Quotas have precise numerical values.
H3. Difference between quota price and tax equals difference between optimal
quota price and tax
According to economic theory there should not be different prices for one product.
Taxes and quota prices should differ somewhat according to our optimization. The null
hypothesis says that there is no difference.
The alternative hypothesis is that there is a difference.
3. RESULTS
We present pooled data for the two places where the experiments were carried out. One
outlier in T2 is removed because decisions suggest that the subject have misunderstood
the instructions. Therefore the other four subjects in the same market has to be removed
as well.
Only Tax Policy
Figure 5 shows the median tax for the subjects in Tl together with the optimal tax
(dotted line). We see that the median tax is lower than the optimal tax in all periods
except for the last three. Figure 6 shows the p-value for each year when testing if the
median is significantly different from the optimal tax (sign test). The tax in TI is
significantly different from the optimal tax from 2000 until 2008. Hence we reject H1
for the early years of T1.
160 = + = = Optimal Tax
Median Tax Tt
140
$/ton CO2 equi.
8
60
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Figure 5. Median tax T1 (no quota market)
0,7
0,6 — = «Significance level-0,05
0,5 pvalue Tt Tax <> Optimal
Tax
0,4
0,3
0,2
0,1
o4
2000 2002 2004 2006 2008 2010
Figure 6. P-value median tax T1 vs. optimal tax
Tax with trade option
Figure 7 shows the median tax for the subjects in the treatment with quota market T2
compared to the optimal tax (dotted line). The median tax is lower than the optimal tax
in all periods. Figure 8 shows the p-values. The tax in T2 is significantly different from
the optimal tax during the whole period, except for the last year. Hence, H1 is reject for
nearly the entire period when a trade option exists.
= = = = Optimal Tax
‘Median tax t2
$/ton CO2 equi.
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Figure 7. Median tax T2 ( with quota market option)
On
— - - Significance level-0,05
pvalue T2 Tax <> Optiral
Tax
ot
2000 2002 2004 2006 2008 2010
Figure 8. P-value median tax T2 vs. optimal tax
Tax with and without trade option
Figure 9 shows taxes from T1 and T2. The main difference is that with a trade option,
taxes do not increase that rapidly towards the end of the period. P-values shows that the
difference is significant after 2008. Hence we cannot reject H2 before 2008. After 2007
we reject H2, the trade option works to reduce domestic taxes.
— Median tax 2
— Median Tax Tt
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Figure 9. Median tax in T] and T2
Quota Prices
Figure 10 shows the median quota price over time (thick solid line). During all periods
the median price is higher than the optimal quota price (dotted thin line). To test H3 we
compare the difference between the median quota price and median tax to the same
difference based on optimal quota prices and taxes. The p-values in figure 11 shows that
the difference is statistically significant except in years 2008 and 2009. Hence we reject
H3 with the exception of the two years.
200
$iton CO2 equi.
— — Bemchmark Tax
—— Median tax t2
20 ---~Bemchmark Price
Price Median Pooled
2000 2002 2004 2006 2008 2010
Figure 10. Median tax T2 ( with quota market option)
0,2
— = «Significance level-0,05
P value delta price Pooled ' Delta optimal
0,1
2000 2002 2004 2006 2008 2010
Figure 11. P-value median tax T2 vs. optimal tax
Inefficiencies
Figure 12 shows individual quota holdings in year 2012 (Y axis) as a function of
domestic reductions by tax in year 2012 (X axis). The ideal situation is that all
individuals reduce domestically by 1000 and engage in no trade. All observations to the
left or to the right of this point represent inefficiencies. Given that subjects engage in
trade, the second best situation is that they compensate for inoptimal domestic
reductions by buying or selling exactly the quota needed to meet the emission reduction
target. A line is draw to illustrate the frontier where the target is reached. Those subjects
that are under the line do not reach the target and those over the line have excess quotas.
While most of the subjects come quite close to the line, there are a few exceptions.
Speculation seems to be a dominating reason for these deviations. When quota prices do
not rise above previous highs at the end, speculators end up with excess quotas.
Quota Holding by 2012
so} - Deficit
5000 .
Emissions reduced by Tax by 2012
Figure 11. Inefficiencies Aiming the target
Overall Performance
Table 3 summarizes to what extent the subjects reached the target. Average emissions
are higher than the target in both treatments (low p-values), and the average is higher
when the trade option is present.
Target Emissions | Average Emission| COW? | UPPeF |p vatue | 7% 7
9 a Emission|Emission| Emissions | Emission
2012 2012 target
s s <target |s > target
T1 3 000} 3 090} 2.982! 3315] 0,017] 79 %| 21 %|
T2 000} TT 2110} 3 723] 0,008} 48 %| 53 %|
Table 3. Subjects Performance in Reaching the Target
The average costs over treatments is summarized in Table 4. The most remarkable
finding is the much higher costs in T2 than in T1. This is not surprising since all trade
serves to reduce overall efficiency in a symmetric game. Cost also increase since the
trade option leads to lower taxes in the last years before 2012.
Optiial Cast ‘Average Total Subjects Reaching the Target
Cost Average [Lower [Upper [Average _|Lower [Upper
TH 44674] 67568 60961] _31715| 80277| _—72304| —59331| 85278
T2 TIO7q SUBST] 296 522] 112620] 406777) Bo 62e] 36442] 162871
Table 4. Total Costs over Treatments and Subjects
4. DISCUSSION
TAX
Do people increase taxes early enough to reach the emission target?
Do people understand the delays in reducing emissions?
Do people have appropriate mental models to address the problem of reducing
emissions to reach the target in time using tax as a domestic policy. Works like Lecocq
et al (1998), Schowon et al (2004), and Green Paper (2004) provide arguments for early
reductions. Emissions should be reduced by steady capital stock renewal rather than by
sudden retrofits as the 2012 deadline approaches. It is very costly to reduce emissions
rapidly because technologies and specific emissions are imbedded in costly and long
lived capital.
Having the correct mental models means that subjects are aware of the time it takes to
replace the existing high emission capital equipment by low emissions technology.
Hence, subjects should begin to reduce emissions using the tax from the very beginning
of the period. All subjects in both treatments set a positive tax the first year. However,
the median tax was significantly lower than the optimal tax, except for the last few
years.
The fact that subjects set low initial taxes, suggest a lack of correct mental models.
Such lack could arise from:
1. A dislike of taxes:
The processing of new information depends on the stock of old information, or familiar
images, Camerer (1995). If people have mental models saying that emission taxes are
not effective, there are reasons to believe that people do not like taxes. There is
common perception of taxes as having a negative effect on economic growth. Emission
taxes are criticized because they require firms not only to pay abatement costs, but also
taxes their unabated emissions (Vollebergh et al., 1997 as in IPCC, 2005). Recent
papers, however, argue that emissions taxes are more cost-effective than direct
regulation and may even lead to higher employment (Wellisch, 1995; Hoel, 1998 and
IPCC, 2005). The intuition is that with taxes emissions constitute a rent because the
firms have to pay for emitting (IPCC, 2005). Besides, some experts criticize emission
taxes because they do not guarantee a particular level of emissions. During the Kyoto
negotiations there were much discussion around taxes whether they are effective or not.
Some countries were very skeptic about the implementation of taxes.
Our subjects, however, are not likely to have very advanced ideas about the economic
effect of emission taxes beyond the first and direct effect on costs. In T1 subjects have
the tax as the only option to reach the emission target. If they dislike taxes very much
they could just play the do-nothing-strategy and pay the punishment. In T1 the subjects
did use the tax. Towards the end of the period, the average tax in the sample even
exceeds the optimal tax. Hence, at least towards the end the needs seem to dominate an
eventual dislike.
In T2 players could use the market as an alternative option. All the strategies in T2 were
mixing policies using the tax and the market. There was no player that did not use the
tax, however, there was only one player who bought almost all the reductions the first
year in the market and fixed the tax at zero for 5 years. This indicates that this person
did not realize the potential for profit of the investments (using the tax) in emission
reductions for sale. The average price in the market was all the time higher than the
average tax. That suggests that players prefer to pay more to avoid the tax, maybe
because there is uncertainty about the quantity reduced by taxes. However the fact that
subjects in T2 did not set significantly lower taxes than subjects in T1 in the first years
make us doubt the importance of their dislike of taxes.
2. Underestimation of delays
When people are not aware of delays, they typically expect to see immediate feedback
from the decisions made. Sterman (1989b) found that subjects underestimated the time
lag between placing and receiving orders in a supply line. In the first weeks subjects
failed to allow sufficient orders in the pipeline to achieve their desire inventory level.
Subjects increment orders when they realize that the current inventory is not large
enough to fulfill the demand. They create an overshoot of the desire inventory because
they ignore the supply line.
Although it is not the same problem, the replacement delay in our experiment is like the
ignored supply line in Sterman (1989b). Subjects underestimate the time needed to
reduce emissions. The low initial tax and the tax panic at the end of T1 accounts for
such underestimation. Being unaware of the replacement delays subjects may consider
the possibility of postponing the reductions to the last years and save on discounted
cost.
3. Inability to respond properly to outcome feedback over time
The emission reductions from taxes can not be determined with certainty from the
beginning of the experiment. The subjects see the marginal cost curve but they do not
have exact information about the delays or the split between replacement and retrofit.
Hence, a feedback strategy is needed.
With very low initial taxes, our results do not show clear patterns of increasing tax in
order to reach the emission targets. The panic tax in T1 in the last years may indicate
learning when the signals are strong enough, however, may also reflect simple time
pressure. In T2 there is not such a panic and the tax has a steady growth. The emissions
market seems to remove much of the time pressure, and little genuine learning seems to
take place.
2. Punishment is not a salient anchor for taxes
Interestingly punishment does not seem to be a salient anchor for taxes.
EFFECT OF MARKET OPTION ON TAX
Will trade of emissions quotas among countries influence how people set taxes?
1. Inefficiencies
When subjects have different levels of understanding they opt for different tax levels
and trade is motivated. Inefficiency arises when subjects use the trade option to reach
the target.
2. Two prices for two nearly identical “products”
Surprisingly, economic students price two quite similar “products” differently. The
question arises whether they really understand how taxes work to produce incentives for
profitable emission reductions. With a market price higher than the optimal, the optimal
tax rate is actually somewhat higher in treatment 2 than the benchmark for no trade.
This is because the players could make profits from selling quotas produced by high tax
rates. When there is no market and subjects have different cost curves, they should
have different taxes and different reductions. But, when there is a market and they have
the same cost curve the tax should be the same.
3. Market as a complementary measure
The average tax in both treatments is nearly the same; therefore the market did not have
a big influence on the tax. In the last years subjects from T1 are stressed from the
bility of not reaching the target, because they did not set high enough taxes at the
beginning, therefore they increase the tax. In T2 there is no strong increase of taxes at
the end, even if they did not set high enough taxes at the beginning. The use of both
polices show a clear strategy of mixing policies. The market was not a substitute for the
tax but a complementary measure.
4. Market Speculation
16
Smith et al (1988) found that speculation in an asset market drives the prices beyond the
optimal price, generating a bubble towards the end. The emission quota market can be
seeing as an asset market. Works by Moxnes (2003) and Anderson et al (2005) show
excessively high quota prices due to speculation, indicating that speculation could have
played some role for quota prices moving above the optimal market price. However, we
cannot know that from this experiment.
Implications of Misperception of Feedback
In the experiment subjects have instant information feedback of what is happening in
the system regarding the emission reduction achieved. However, they seem to have
problems in finding proper actions to correct the emission path. In real life the time lag
between the time measures are implemented and the information about the reductions
achieved is very long. Besides, the information is not hundred percent reliable, IPCC
methodologies for some GHG’s are still in development and the national emissions
update has a delay of around one year (or more for some countries). Hence, the policy
makers could even have more problems then subjects in the experiment to act properly
for correcting the emissions path.
Emissions
[Target Emission [Diference fray norton |TRADIN |iom 2000 to| Average annual
2012 Mton|2002 Mton|Emission |G Ic 2002 Mion | growth GDP
lotco2z fofco2 |2002-Target ISCHEME
lof coz
2000[ 2007
Austria Bi 7H 22 a, 3 1
Belgium 115] 146 ED YES 2,4 1
Denmark 5 35 10/13 EUR - 1997/YES 3,3 1
Finland 33] Ba 1/16 EUR-1999 3,6 1
France 374 407 Eg al 4 2
Germany 798 838 a Ss, 3 1
ireland 29] a5 16 3,12 a
italy 388 449 60 o| 3 2
Luxembourg 8 10 3 1,8 1
Netherlands EE] 256 58|ND-2000___|YES. 7,4 1
Norway 35] 46 11/12 EUR - 1999]YES 4,2 1
Portugal 56] 67 11 3,4 2
Spain 260 Bat a 2a 3
Sweden 37] 35 “240 EUR- 1991 [YES o|4 1
Switzerland aA a 3 2] 3 1
[United Kingdom 525) 353 28[ND-2002___|YES. 3,3 2
Table 3. Current State of CO2 Emissions European Union (source: www.unfcc.com,
www.ecia.com , www.worldbank.org, www.oecd.org)
EMISSION REDUCTIONS IN EUROPEAN UNION
- Current State
In the European Union only 5 out of 15 countries are using taxes for reducing emissions
(table3). Sweden is an interesting country because it has the highest tax rate, the
country is already below the target and has zero emissions growth rate from 2000 to
2002. Countries like Spain, Italy and Germany are far away from the target and are not
implementing taxes or trading schemes for reducing emissions. From 2000 to 2002 few
countries are achieving reductions at very low rates. Countries like Spain, Portugal,
Ireland and Austria are far from the target and are increasing their emissions instead or
reducing. The European Union Emission Trading Market which started activities in
February 2005 is reporting prices of 21 EUR /ton of CO,’. This is somewhat higher than
the prices obtained in our experiment.
- Implications
From the data showed above we have evidence that countries typically are using even
lower tax rates and higher market prices than in our experiment. In the experiment
subjects fail to understand the dynamics of emission reductions in a simplified
environment. In real life the environment is not that simple, adding complexity to the
task of reducing emissions and increment the likelihood of bias in the decision making.
9
www.nordpool.com
18
REFERENCES
Anderson, CM. and Sutinen, JG. (2005). “A Laboratory Assessment of Tradable
Fishing Allowances”. Marine Resource Economics,vol.20 issue |
Baranzini, Andrea; Goldemberg, Jose and Speck Stefan. (2000). “The future of
Carbon Taxes.” Ecological Economics. Vol. 32. p.p. 395-412
Bartolome, AM. (1995). “Which tax rate do people use: Average or Marginal”.
Journal of public economics. Vol. 56. pag 79-96
Bohm P., Charlen B. (1998): “Emisién Quota Trade Among the Few:
Laboratory Evidence of Joint Implementation Among Commited Countries” Resource
and Energy Economics. Vol. 21, p.p. 43-66
Brehmer, B. (1989):”Feedback Delays and Contorl in Complex Dynamics
Systems” in P. Millin and E Zahn (Eds.), Computer Based Maganement of Complex
Systems,Springer Verlag, Berling, 1989.
Criqui, P., S. Mima, and L. Viguir (1999): “Marginal abatement cost of CO2
emission reductions, geographical flexibility and concrete ceilings: an assessment using
the POLES model”. Energy Policy Vol. 27, p.p. 585-601
Ellerman, D. A. and A. Decaux. (1998): “Analysis of Post-Kyoto CO2 Emission
Trading Using Marginal Abatement Curves”. Report 40, Massachusetts Institute of
Technology - Joint Program on the Science and Policy of Global Change.
Eriksen, Knut and Fallan, Lars. (1996) “Tax knowledge and actitudes towards
taxation; a report on a quasi-experiment” Journal of Economic Psychology. Vol 17. pag
387-402
Forrester, J.W. (1961) Industrial Dynamics. Cambridge: MIT Press; Currently
available from Pegasus Communications: Waltham, MA.
Funke, J. (1991): “Solving Complex Problems: Exploration and Control of
Complex Systems” Complex Problem Solving: Principle and Mechanisms. Edited by
Sternberg, R and Frensch, P, NJ Lawrence Erlbaum Associates.
Hoel, M., (1998): “Emission taxes versus other environmental policies.”
Scandinavian Journal of Economics, 100(1), 79-104.
Holtsmark, B and Hagem C. (1998) “Emission Trading under Kyoto Protocol”.
Cicero Report 1998:1. Available on: www.cicero.uio.no
IPCC. (2005). Third Assestment Report. Available on: www.ipcc.ch
IETA, (2005). “Market Mechanisms” [cited 2005] Available from World Wide
Web: (http://www.ieta.org/ieta/www/pages/index.php?IdSiteTree=26
Klepper, G and Peterson, S. (2004) “Marginal Abatement Cost Curves in
General Equilibrium: The Influence of World Energy Prices” The Fondazione Eni
Enrico Mattei Note di Lavoro Series Index. Nota di lavoro 136.2004. Available on:
http://www.feem.it/Feem/Pub/Publications/WPapers/default.htm
Kolstad, C.D. (1996): “Learning and stock effects in environmental regulations:
the case of greenhouse gas emissions”. J. Environ. Econ. Manage. Vol. 31, p.p. 1-18.
Lecocq, F, Hourcade J.C, and Doung M. (1998): "Decision Making Under
Uncertainty and Inertia Constrains: Sectoral Implications of the when Flexibility."
Energy Economics. Vol. 20, p.p. 539-55.
Meadows, G and R Weisenmayer. (1999): "Identifying and addressing student's
alternative conceptions of the global warming: The need for cognitive conflict." Journal
of Science Education and Technology Vol. 8, p.p. 235-239.
Moxnes, e. (2003) individual transferable quotas versus auctioned seasonal
quotas, an experimental investigationsnf-report, 49/03, snf, bergen.
Moxnes, E. (2004). "Misperceptions of Basic Dynamics : The Case of
Renewable Resource Management." System Dynamics Review 20(2): 139-162.
Moxnes, E. (1998). "Not Only the Tragedy of the Commons: Misperceptions of
Bioeconomics" Management Science 44(9): 1234-1248.
Moxnes, Erling and Saysel, Ali. (2005) “Misperceptions of Global Climate
Change: Information Policies.” Unpublish work.
Nordhaus, W.D. (1994): “Managing the Global Commons: The Economics of
Climate Change”. The MIT Press, Cambridge.
Smith, V, Suchanek, G and William, A. (1988),“Bubbles, crashes and
endogenous expectations in experimental spot asset markets” Econometrica Vol 56 No.
5. pp. 1119-1151.
Springer, Urs and Varilek, Matthew. 2004. “Estimating the price of tradable
permits for greenhouse gas emissions in 2008-2012” Energy Policy. Vol. 32. pag 611-
621.
Sterman, J. D. (2000). Business Dynamics: Systems Thinking and Modeling for
a Complex World. Boston: Irwin/Mc Graw-Hill.
Sterman, JD. (1998a): “Misperception of Feedback in Dynamic Decision
Making” Organizational Behavior and Human Decision Process. 43(3), p.p. 301.335
Sterman, JD and Sweeney LB. (2002) “Cloudy Skies: Assessing Public
Understanding of Global Warming” System Dynamics Review. Vol. 18 issue 2. pag.
207-240
Sterman, JD and Sweeney LB. (2000) “Bathtub dynamics: initial result of a
system thinking inventory” System Dynamics Review. Vol. 16 issue 4. pag. 249-286
Sterman, JD. (1998b): ” Modelling Managerial Behavior: Misperception of
Feedback in Dynamic Decision Making Experiment” Management Science 35(3): 301-
339
Tietenberg, T., 2000: Environmental and Natural Resource Economics, 5th ed.
Addison-Wesley, Reading, MA.
Vollebergh, H.R.J., JL. de Vries and P.R. Koutstaal, 1997: Hybrid carbon
incentive mechanisms and political acceptability. Environmental and Resource
Economics 9, 43-63.
Wellisch, D., 1995: Locational choices of firms and decentralized environmental
policy with various instruments. Journal of Urban Economics 37(3), 290-310.
Yohe, G.W., Wallace, R. (1996): “Near term mitigation policy for global change
under uncertainty: minimizing the expected cost of meeting unknown concentration
thresholds”. Environ. Model. Assess. Vol. 1, p.p. 47-58.
20
Appendix
The Kyoto treaty experiment
The purpose of the Kyoto treaty is to reduce world emissions of greenhouse gases to
limit potential future climate change. According to the Kyoto treaty, the countries that
have signed the agreement must reach certain targets for their greenhouse gas emissions
by the year 2012. The targets can be reached in two ways:
1. Countries can reduce their domestic emissions, or
2. They can buy emission quotas from other countries, which in turn must reduce their
domestic emissions below their agreed targets to make up for the quotas they have sold.
In this laboratory experiment the world is split in 5 identical countries (or regions). You
will each be playing the leader in one of these countries, making all decisions for the
country by yourself. Your goal is to reach the target with the lowest possible cost.
Each year you have two decisions to make: You set a tax rate for emissions of
greenhouse gases in your own country and you make bids to buy or to sell emission
quotas in a market where all 5 countries interact. From one year to the next, the
computer calculates how much the domestic emissions have been reduced due to the tax
and the amount of quotas you have bought or sold in the market and at what price.
The experiment starts in year 2000 and the emission goal should be reached by 2012.
You cannot reach the target without incurring costs. At the end of the game you will
receive a payoff that depends on your total costs. The payoff can vary from NOK 70 for
very high total costs to NOK 150 for very low total costs.
Your total costs depend on three factors:
1. Domestic emission reductions cost money for those who have to make the reductions.
The tax income for the government is of no concern here; just assume that it is returned
to the tax payers as reductions in other taxes. It is assumed that all emission-reduction
projects that cost less than the tax rate will be implemented each year. Thus, the higher
the tax rate, the larger the costs for domestic emission reductions and of course the
larger the emission reductions.
2. If you buy emission quotas in the market you generate a cost. If you sell quotas, you
decrease your costs.
3. If you do not reach the emission target by domestic reductions or quotas in 2012 you
will be punished with an extra cost of 200 $/ton CO2 equivalent for the excessive
emissions. Note here that Greenhouse gases are measured in equivalent units of CO2
(tons of CO2 equivalents)
Think about the costs as being paid by loans for which you have to pay a 4 percent
interest per year. Thus, your total costs in 2012 will include both the direct costs and the
interests you have to pay on the loans. Hence, an early reduction in emissions will be
more costly than a later and otherwise similar reduction.
To simplify the experiment we assume that there is no economic growth. Furthermore,
all emissions reductions must take place with equipment that exists today, there is no
technological improvement over time. Your emissions in year 2000 are 4000 million
tons of COz equivalents. Your emission target for year 2012 is 3000 million tons of
COz equivalents. Thus, the needed reduction is 1000 million tons of CO2 equivalents.
21
The experiment is based on studies that have estimated the lowest possible marginal
costs of total domestic emission reductions by 2012. The minimum costs require that an
optimal sequence of taxes is used. See future values of the minimum marginal costs in
the graph below. For your information, a tax of 130 $/ton CO2 equivalent corresponds
to approximately a doubling of current energy prices.
Estimate Marginal Cost for Different Total Emission Reduction by
2012
wa
we
we
3* 00
gS a
Bs
22.
w
°
by (abv oy “(ooh me me “We “om “elo Con se
‘Total Emission Reductions in 2012
[Million tones of CO2 equivalents]
Each year the computer computes the emissions reductions that follow from chosen tax
rates. You should assume that the computer program is highly realistic except for the
simplifications already mentioned.
How to play
The PC screen is divided in three sections: decisions for the present year, information
about the last year, and total costs and payoffs in year 2012. The game progresses in the
following sequence: Look at information from last year, make decisions, press the
button “Accept Decisions”, the game progresses to the next year, you look at the new
information and so on.
DO NOT PRESS “Accept Decisions” BEFORE YOU HAVE CHECKED YOUR
DECISIONS - THERE IS NO RETURN ONCE YOU HAVE ADVANCED TO THE
NEXT YEAR.
Decisions:
You set the tax rate by entering a number in the Tax box.
You make bids for quotas by clicking on the curve and then dragging it to where you
want it. At sufficiently low quota prices (to the left in the diagram) you will probably
like to buy quotas - if so, the curve should be in the buy region (higher than zero). At
sufficiently high prices you will probably like to sell quotas and the curve should be in
the sell region (lower than zero). You have to specify the entire curve, such that the
computer program knows how much you want to buy or sell at all possible quota prices.
The curve may be flat or declining, it cannot bend upwards at any point (if you do, you
get an error message and have to change it). An upward bending curve is like saying
that you want more quotas the more expensive they are - that does not make sense.
22
When all players have entered their curves (and their taxes and have clicked on “Accept
Decisions”), the computer program finds the quota price that equilibrates the market,
that is, total sales equal total purchases.
Information last year
Current Yearly emissions
Target for emissions in 2012
Need for Domestic Reduction or Quotas
Your total Quota Holding
Need for domestic Reduction by 2012
Quota Price last Year
Global Emissions (Sum of emissions for all five players)
Quotas Bought Last Year
Quotas Sold Last year
Information in year 2012
Total Cumulative Cost
Your Payoff
Thank you and Good luck!!
23
