sra

ner

ihe

e n v i r o nm en t a l s c i e n c e & p o l i c y 1 2 ( 2 0 0 9 ) 2 5 7 2 6 9

s a

available at www.sciencedirect.com

.el1. Introduction

Global climate change issues are high on the agenda for both

the scientific community and policy makers. According to the

fourth assessment report of the Intergovernmental Panel on

Climate Change (IPCC), significant reduction effort in green-

house gas (GHG) emissions is needed in order to limit long

term global temperature increase. Although most of this

temperature rise would take place in the second half of the

century, actions should be taken in the near future in order to

adopt the necessary steps for the massive infrastructure

change required, and to give the right signals to the sectors

responsible for most of the anthropogenic GHG emissions.

The pulp and paper sector faces a threefold challenge from

the climate change perspective. On the one hand, it is a very

energy-intensive sector. Producing 1 tonne of paper requires

517 GJ of process heat, depending on the paper type and on

the technology applied. Therefore the energy content of the

different paper grades is comparable to that of other energy

intensive products, such as cement or steel. On the other

hand, the most important natural resource for paper-making

is biomass, mainly wood and other fibre resources, the use of

which is by internationally accepted definitions assumed to be

CO2 neutral. An additional factor whichmakes themodelling

of the sector even more complex is that self-generated

electricity and heat play an important role in the energy

balance of the sector. These three aspects make the sector

unique from an energy modelling and climate change

perspective, and put it in the focus of attention of the climate

research.

According to its relative importance, numerous attempts

have been made in the literature to model the pulp and paper

market at the global scale and with an outlook to its long term

energy consumption and GHG emissions. These global forest

sector models are the Global Forest Products Model (see

Tomberline et al., 1998) of FAO, the Global Forest Sector Model

Published on line 6 March 2009

Keywords:

Pulp and paper sector

Climate change

Bottom-up modelling

JEL classification:

L73

Q54

with a focus on energy consumption and carbon emissions. It is an annual recursive

simulation behavioural model with a 2030 time horizon incorporating several technological

details of the industry for 47world regions. The long time horizon and themodular structure

allow the model users to assess the effects of different environmental, energy and climate

policies in a scenario comparison setup. In addition to the business as usual developments

of the sector, a climate commitment scenario has been analysed, in which the impacts of

changing forest management practices are also included. The climate scenario results

reveal that there is a significant carbon reduction potential in the pulp and paper making,

showing a number of specific features: the central role of the fibrous resource inputs and the

potential impact of increased waste wood and black liquor based heat generation.

# 2009 Elsevier Ltd. All rights reserved.

The ideas expressed in this paper are those of the authors and do not necessarily represent the views of the European Commission.* Corresponding author. Tel.: +34 954 488236; fax: +34 954 488279.E-mail address: laszlo.szabo@ec.europa.eu (L. Szabo).

1462-9011/$ see front matter # 2009 Elsevier Ltd. All rights reserved.doi:10.1016/j.envsci.2009.01.011A world model of the pulp andenergy consumption and emis

L. Szabo a,*, A. Soria a, J. Forsstrom b, J.T. Kea Institute for Prospective Technological Studies (IPTS), Directorate Ge

c. Inca Garcilaso s/n, Expo Building, E-41092 Seville, SpainbTechnical Research Centre of Finland (VTT), P.O. Box 1000, Vuorim

a r t i c l e i n f o a b s t r a c t

This article introduce

journal homepage: wwwpaper industry: Demand,ion scenarios to 2030

nen b, E. Hytonen b

al Joint Research Centre, European Commission,

ntie 5, Espoo, Finland

bottom-up global model of the pulp and paper sector (PULPSIM)

sevier.com/locate/envsci

of the European Forest Institute (EFI-GTM) (Kallio et al., 2004),

and theWorld Forest Products Model (WFPM) of JIRACS (2003).

important market interactions, both on the demand and the

supply sides, including the trade issues. Demand for the final

e n v i r onm en t a l s c i e n c e & p o l i c y 1 2 ( 2 0 0 9 ) 2 5 7 2 6 9258While the first two models are based on mathematical

programming, the last uses a simulation approach. The EFI-

GTMmodelmainly focuses on the forest industry (Kallio et al.,

1987), and is based on themodel developed in the IIASA during

the late 1980s.

Numerous references in the literature model certain

segments of the markets, focusing either on an individual

country or on certain characteristics of the sector. Thus

national or regional models exist for various countries. The

most recent one has been developed for the USA (Ruth et al.,

2000; Davidsdottir and Ruth, 2004) and is based on an

econometric analysis. It specifies a dynamic model for the

paper sector, analysing the impacts of different climate

policies (carbon taxation and investment-led policies).

Another regional model is NAPAP partial equilibrium eco-

nomic model of the North American region developed by the

USDA Forest Product Laboratory (see e.g. Ince, 1998), that gives

long term outlook (till 2050) for the sector based on a price-

endogenous linear programming system. Mollersten and

Westermark (2003) investigate the CO2 reduction potential

for the Swedish paper industry, based on different carbon

emission reducing measures. Farahani et al. (2004) present an

article dealing with the techno-economic potential of a new

technology the black liquor gasification-combined cycle

(BLG/BLGCC) on the pulp and paper sector in the US and

Sweden. According to their findings, the excess electricity that

this new technology can produce may lead to a significant

reduction of CO2 emissions in the sector. Technologically

detailed analysis on black liquor gasification can be found in

an article by Eriksson and Harvey (2004), in which the

performance of the BLG technology is compared in different

mill powerhouse configurations.

Another interesting series of publications are activity-

based analyses. Farla et al. (1997) make a cross-country, cross-

time comparison of energy efficiency developments in the

pulp and paper sector in eight OECD countries. They find that

energy efficiency improvements played a key role in limiting

energy consumption during the 19731991 period. The articles

of Brannlund et al. (1998) and Bruvoll et al. (2003) implement a

nonparametric frontier method (DEAdata envelopment

analysis) that allows to measure the effects of environmental

regulations on the performance of the individual firms in two

Scandinavian countries.

Themain purpose of this article is to present a global paper

and pulp model (PULPSIM) that attempts to capture both the

technological aspects and the market developments of the

sector.1 The overall objective of constructing this model is to

synchronize the technological details with an appropriate

economic framework on a global level. In particular, in order to

analyse specific policies, e.g. in the context of climate change

policies,more attention is paid to the detailedmodelling of the

energy consumption and the GHG emission of the sector.

Therefore in order to overcome some of these boundaries,

the specification of the PULPSIMmodel incorporates the most

1 The PULPSIM model has been written in VENSIM 5.4 software.

It operates on a year to year recursive simulation basis, running upto the year 2030.paper grades are derived through the intensity of use

hypothesis (see e.g. Van Vuuren et al., 1999), according to

which the per capita income determines the commodity

intensity of the different regions. This dynamic evolution of

commodity intensity is essential in long term models, as the

relationship between income and consumption is not deter-

mined by a single elasticity, but it is redefined over the full

time horizon. This provideswith amore sensible picture of the

commodity demand and, moreover, it allows reflecting the

substitution away from the commodity, as the use of

alternative materials could take a higher share over a long

time scale.

The next element incorporated in the PULSPSIM model is

international trade, based on product separation within paper

grades. The imported and domestically produced products are

differentiated and their demand shares reflect these price

differentials. Not only the final products and the forest

resources are traded in the model, but many of the

intermediate goods, such as sawmill products and chemical

pulps as well. Importing and exporting regions are also

distinguished according to the past developments of the

markets.2 This approachhas its own limitations,mainly that if

a region is classified as an importer, it will stay in this group in

the whole period. Another method widely used in empirical

trade modelling is the product differentiation by origin

approach with a constant elasticity of substitution (CES)

function, however it still does not solve the small share

problem: those countries, which have small shares of imports

will continue to have small shares in the future, as it is difficult

to drive them out from the CES function corner solution. In

order to model the major world trade flows, four regional

markets (Africa and Middle East, America, Asia, Europe) were

set up. World market prices include distance dependent

transport costs, as this cost element could play an important

role in some product and market segments. The global trade

market is cleared through the allocation of import demand to

exporters (see Section 3 for details). These allocation proce-

dures are constructed to check that capacities and the needed

resources are available.

The mentioned market modules introduce the economic

rationale in the model in a long time scale, with an attempt to

provide with a reasonable ground for assessing different

policy options or scenarioswith long term impacts, such as the

carbon constrained future scenarios analysed in this article. It

has to be noted however that our approach is still not

accounting for the paper-containing products export and

import, so it ismost probably underestimating the importance

of the trade effects in the continuously growing international

trade market of the paper products.

This article is structured in five sections. The next section

discusses themain features of the pulp and paper sector at the

global scale. Section 3 describes the PULPSIMmodel. In Section

4 the most important features of the reference scenario are

introduced concerning the future development of the pulp and

2 Individual countries, however, could be exporter in one pro-

duct and importer in another grade according to their pastrecords.

e n v i r onm en t a l s c i e n c e & p o l i c y 1 2 ( 2 0 0 9 ) 2 5 7 2 6 9 259paper market. This section includes the analysis of a climate

commitment scenario, and also summarises the results of the

sensitivity analysis of the model, while Section 5 concludes.

Appendix A in Supplementary Data gives more details on the

material and energy flows of the sector, Appendix B in

Supplementary Data provides with a technical model descrip-

tion, while Appendix C in Supplementary Data presents the

full sensitivity analysis.

2. Pulp and paper industry overview

Section 2.1 gives a focused description of the technologies/

processes used in the sector, with an insight to the materials

used, as they are the determinant factors explaining the

differences in energy consumption. Section 2.2 presents the

trends in the global paper supply together with the patterns in

the international trade in various paper products and

resources.

2.1. Paper-making processes

The forest industry comprises fibre-based chemical industry

producing pulp, paper and fibre based board and sawmills

industry producing sawn wood, plywood and wood based

panels. The industry is a capital intensive one, and the

production technology is based on well known principles and

readily available technology. Thus, the future of the global

structure of the forest industry is expected to be based on the

availability of suitable raw material and the cost patterns in

different areas.

Papermaking consists ofmany and complex processes, but

the twomost important steps are pulping and paper finishing,

and the technology choice depends on the final use (or grade)

of the paper. The twomainmethods of pulping aremechanical

and chemical processing. These pulps are not substitutes but

complements in papermaking. Their qualities differ and they

are used in different grades of papers. Pulping methods differ

from each other also in energy and wood use. Chemical

pulping uses twice as much wood per tonne compared to that

of mechanical pulping. Modern chemical pulp mills are self-

sufficient in their energy consumption as half of the wood is

dissolved and used as fuel in the chemical recovery phase.

Mechanical pulping uses wood very efficiently but at the

expense of much higher electricity consumption. Depending

on the local situation the global competitiveness can be

improved by choosing the suitable product and processing

option.

The mechanical and chemical wood processing industries

are interlinked through the common resource base and

because the by-products of the mechanical processes can be

used as raw material in chemical processing. In addition,

wood can be useddirectly as a fuel. All these competinguses of

wood have their effects on the pulp and paper sector when

simulations extend to the future.

In addition towood, recycledpaper formsan important raw

material input in the paper-making process. For example

Europe has a leading position in reuse, asmore than 50% of thepaper is recycled. Although recycling is both economically and

ecologically sound, recovered paper cannot be efficiently usedin all paper grades, nor can it be used indefinitely. In spite of

this, paper recovery is expected to grow fast in the future in

order to keep up with the demand increase. This will most

probably lead to a continuous price increase of recovered

paper. Limitations for fibre recovery are set by fibre length,

quality and usability. Fibre shortens up every time it is used

and at some point, usually after 46 cycles, it is too short to be

used in papermaking. Therefore a certain amount of virgin

pulp input will always be needed to meet the quality demand

set for the products. In the mix of resources for paper making,

recycles fibre typically replaces for mechanical pulp in

newsprint and in some board grades. Because of its economic

and environmental advantages, more and more recycled fibre

is expected to be used as raw material in the paper making.

2.2. Present trends in paper consumption and trade

North America, Europe and Asia account today for more than

90% of total paper and paperboard consumption (360 million

tonnes in 2004), with almost equal shares amongst them.

Oceania, Africa and Latin America together account for less

than 8%. World paper and paperboard demand is expected to

grow about 2.1%/yr until year 2020 and the growth will be

fastest in Eastern Europe, Asia (except Japan) and Latin

America (Jaakko Poyry Ltd., 2006; Diesen, 1998). This growth

rate covers high variation amongst the countries: in the

developing regions it is expected to exceed 4%, while in the

mature markets (North America, EU, Japan) this rate is

expected to be around 0.51%. This means that developing

Asia (mainly China) will play a crucial role in the future paper

market.

The pulp and paper market is a heterogeneous and

extensively interlinked market. It means that international

trade not only takes place in all the different product

categories, but also in semi-products and raw materials as

well. In addition to the fibre resource, the chemical pulp and

the recycled materials are also traded internationally. Pulp

and paper producers specialise: some of them are only

engaged in pulp production, while others manage the full

production cycle from fibre resources to final paper grades. As

a result, one can observe a situation where many countries

specialise in certain final products, and still rely on imports in

other paper categories. This means that the paper and pulp

market is highly interconnected through international trade,

which increases the complexity of themodelling work. Supply

is dominated by ten countries, giving around 60% market

share in the sawn-wood, pulp and paper production. Six of

them cover more than half of the exports in these products,

which show the high concentration level of resource supply

and production.

Fig. 1 illustrates relative positions of the ten most

important countries within the market. The USA and Canada

are present in allmarketswith significant shares,which is also

true for Sweden and Finland but with considerably less

proportions. Apart from them, Germany is a major European

player represented on the export markets of the final paper

products. While Russia is a important actor in the resource

markets, its weight in pulp and paper making is lower. Chinafaces similar challenges in paper making as in many other

energy intensive products, such as iron and steel and cement:

the model is merged with the POLES model (see Criqui, 1996

for an introduction to POLES).

3. Model overview

e n v i r onm en t a l s c i e n c e & p o l i c y 1 2 ( 2 0 0 9 ) 2 5 7 2 6 9260while its production capacity is significant, it still relies to a

great extent on imports to cover its rapidly growing demand.

There are two issues in this market picture with direct

implications for the pulp and papermodel. First, themodelling

approach should incorporate multiple markets for all the

production chain, from thefibrous resource, through the semi-

products to the final product markets. Following this route

allows the modeller to capture a realistic picture of the whole

papermarket, including the energy consumption of the sector,

which is distributed in all of these production steps. This calls

for a modularly structured model, which is built up in the

PULPSIM global pulp and paper model.

The second issue highlighted in this market overview is

the resource and capacity availability. As the main wood

resource owners (Russia, North America and Latin America)

and the emerging markets in paper consumptions are

distinct from the traditional producing centres, the risk of

emergence of supply bottlenecks with sudden price peaks in

certain segments of the markets (e.g. in recycled pulp,

fibrous resources etc.) is high. Additionally the wood and

other fibrous biomass have limited resource potential if they

are produced in a sustainable manner. According to the

results of the Global Fibre Resource Model of FAO (1998)

Russia and North America are the main wood based fibre

resource owners together with Latin America. However, in

this latter one the available potential will be more restricted,

Fig. 1 Pulp and paper market shares in 2002 (based on

data from the Finnish Forest Research Institute).as according to this study the ratio of natural forest not

available for harvesting is the highest, ranging from 60 to

90% in the countries belonging to the region compared to the

50% world average. This availability is restricted by geo-

graphical inaccessibility as well as by the assumed sustain-

able forest management practices. This resource faces a

competing use (biomass to electricity, biomass to liquid

fuels) and it has to face markets with enormous growing

potential. This demand growth comes not only from the

developing regions, but also from developed countries which

are expected to increase their demand for biomass. This is

the case, for instance, in Europe which pursues environ-

mental policies aiming at multiplying biomass use in the

long term, which in turn puts pressure on the resource

resulting in an increasing price trend. The setup of PULPSIM

model reflects these potential bottlenecks within the paper

sector, however does not account for the other competing

uses of biomass yet. This connection will be provided whenThis section describes the functioning of the model in a non-

technical way.3 In the model the world is divided into 47

countries/zones. Some of them correspond to an individual

country; some others are aggregation of several neighbour-

ing countries sharing similar socio-economic characteris-

tics. The 47 zones are grouped together, for reporting

purposes, into 5 regional markets, corresponding to Europe,

Asia, North America, South America, and the Rest of the

World (ROW), including Africa, Middle East, Oceania, Japan,

Russian Federation and Rest of the new Independent States

(RIS).

3.1. Material and technology characterisation

Paper product classifications are usually made according to

the intended use of the product, e.g. printing, tissue, case,

writing papers. However from the energy consumption point

of view, the pulping method used is the key determining

factor. As most of the statistical information is mainly

collected by use categories, even the product categories (paper

grades) should be selected carefully for the model in order to

capture the variation in energy consumption and tomatch the

required categories the statistics not necessary provide.

Considering all these factors, modelling three paper grades

newsprint, fine and board have been selected in PULPSIM as a

compromise. Newsprint means both newspaper4 and all types

ofmechanical papers (i.e. papers containingmechanical pulp),

fine means coated and uncoated wood-free papers including

tissue papers, and board embraces all types of wrapping,

packaging and board papers.

More categories wouldmeanmore precision in the product

part of themodel, but it would have been increasing themodel

complexity and the difficulties in verification, while most of

the differences in energy consumption are still captured by the

chosen categories. With these three categories the main

differences in energy consumption are captured (see Table 1),

while the generally used paper categories can be classified into

these grades in a straightforward manner, according to the

pulp type used for the paper grade.

Differences in energy use in paper making are mainly due

to the variation in pulp shares used in the distinct paper-

making processes. Pulps differ substantially in resource use.

Mechanical pulp uses a lot of electricity but only about half of

the wood compared to that of chemical pulp. Mechanical

pulping produces heat as a by-product and it is used as drying

steam in paper processing.

3 Description of the main model equations can be found inAppendix A.2 in Supplementary Data.4 Newsprint paper is nowadays being made almost 100% out of

recycled paper. But as in the model several paper grades aremerged (e.g. magazine papers) under the name newsprint, a

11% share of chemical pulp use is assumed in this grade as well(see Supplementary Data, Fig. 6).

different regions. Using the intensity of use hypothesis for

paperdemandmeans thatdemand isnot onlydrivenbyasingle

e n v i r onm en t a l s c i e n c e & p o l i c y 1 2 ( 2 0 0 9 ) 2 5 7 2 6 9 261In chemical pulping, half of the wood material remains

dissolved and used as fuel. Recycled paper pulp uses only a

fraction of energy compared to the other pulps. All the wood

residues of the processing chain can be used as fuel. In Table 1,

the average pulp and energy consumptions are shown for the

different grades.

Raw wood use, energy use and by-product fuels produced

show high differences amongst the different paper products in

Table 1. Board productionwith a large recycledfibre share leads

toverysmall rawwoodandenergyuse.Newsproductionstands

out with its electricity consumption and fine paper production

has the largest wood and steam consumption. On the other

hand, its fuel production reflects the amount of wood used.

The autonomous energy efficiency improvements (AEEI-s)

found in the literature report values in the range of 0.30.5% for

the sector, while the total energy efficiency improvements

reported for the 1980s and 1990s were in the range of 11.6%

(Farla et al., 1997; Martin et al., 2000).

The table gives a clear indication that the chosen level of

disaggregation is necessary in order to capture energy

consumption in the sector with the appropriate accuracy. It

also suggests that the three pulp categories (mechanical,

chemical, recycled) must be included in the model in order to

address different impacts on the different paper grades with

energy taxation and carbon constraints. Raw wood use differs

by a factor of almost four times between the different paper

grades, and energy use differs by a factor of two. Eliminating

this distinction using average numbers on energy consump-

tion would undermine the ability of the model to account for

the different impacts on the different products in the future.

As the detailedmaterial flows and processes are core assets of

the model, a more detailed description is given in Appendix A

in Supplementary Data.

Table 1 Material and energy use in paper making.

Pulp shares News Fine Board

Mechanical pulp 0.60 0.00 0.13

Chemical pulp 0.11 0.74 0.14

Recycled paper pulp 0.23 0.14 0.62

Raw wood use, m3 2.3 3.9 1.1

Energy, MWh

Electricity 2.2 1.4 1.4

Steam 1.6 4.6 2.4

Fuels produced 1.0 3.5 0.73.2. Paper demand and global trade

Commodity intensity is frequentlydescribedasafunctionof the

national per capita income, as their consumption patterns

follow a Kuznets-like inverse U-shaped curves. These types of

functions have been considered for various materials for

different regions, many of them being energy intensive

products. Studies related to this field are numerous, see e.g.

Van Vuuren et al. (1999) for steel, Mannaerts (2000) for paper,

aluminium,steelandvariouschemicalproductsandSzaboetal.

(2006) for cement. This approach was also followed in deriving

the paper consumption in the PULPSIM model. The GDP per

capita, the lagged value of the paper consumption per unit of

GDPandtheproductpricedeterminepaperconsumptionfor theincome and price elasticity. Using two behavioural parameters

on the per capita income (see parameter alpha and beta in

Appendix B.2.1 in Supplementary Data for the formulation)

allows for capturing the saturation level at different income

levels. Thismeans that after a certain level incomegrowthdoes

notentailmorematerial consumptionperunitofGDP.Reaching

this point income growth is decoupled from more intensive

material consumption. Obviously, this approach has its limita-

tions as well. As the estimation or the main parameters are

based on the historic data, sudden shifts (e.g. newly emerging

technologies) could also change consumption patterns radi-

cally. In the case of the pulp and paper sector the wide-spread

diffusion of electronic offices and practices the printing paper

use was expected to shrink, however there are only few

countries that experienced this reduction in levels (e.g. Sweden,

Japan) in the last few years. Additionally printing paper

represents only a smaller share in total paper consumption.

Total demand for each product consists of domestic and

import demands. Both are forecasted according to global

consumption trends and domestic capacity available. The

globalmarket is cleared throughallocating the import demand

onto exporters. The allocation is carried out by applying a

simple computing algorithm that replicates with satisfactory

results supply allocation under scarcity conditions. If the

exporters cannot meet the import demand then the global

total import demand is rationed and the scarcity is distributed

for all, according to a tatonnement process.

The international trade module has a two-stage hierarchical

system. It is hierarchical, as first the import quantity is

determined for each 47 model regions, as a share of the total

domestic demand. It is a dynamically updated share, where

the driver is the distinct price development of the domestic

and international markets. Second, the export is shared out

among the exporting regions of the given product, which

depends on the shares of exports of the earlier period, and on

their cost-competitiveness. Additionally the import algorithm

is two-staged, as import from closer markets and from the

global market is split. Since the transportation costs play an

important role in the total import costs, regional supply gets

first priority, and the remaining non-satiated demand is met

by import from the global market.

As not only the final goods are traded internationally the

model captures trade in the other goods as well, including the

rawmaterials (rawwood, recycled paper and sawn-wood) and

in an intermediate product (chemical pulp). For each traded

product, the import and export regions are defined.5 In the

model, a country can only be either an importer or an exporter

of a traded commodity. But a country can be an importer in

one market and an exporter in another.

3.3. Production module and capacity planning

The process module is the core of the PULPSIM model,

determining the production routes of the different pulp and

5 Four trading zones were set up: Africa and Middle East, Asia,

America, Europe. The grouping of the individual countries to thezones is shown in Appendix C in Supplementary Data.

paper-making technologies. It uses an iterative optimisation

routine to fulfil its threefold task. First, it ensures the

technological material balance, where fixed coefficient mate-

rial use is assumed. Second, it takes care of balancing

The price of a resource differs from the production costs if

the demand of the resource approaches to the supply limit, in

e n v i r onm en t a l s c i e n c e & p o l i c y 1 2 ( 2 0 0 9 ) 2 5 7 2 6 9262production in such a way that domestic production (if exists)

covers domestic demand for domestically produced paper

grades, intermediate demand of raw materials and inter-

mediate goods and export demand for the various paper

goods. Third, the routine searches for bottlenecks on the

production chain, which would limit the production possibi-

lity frontiers. In the short term these bottlenecks can result in

shortages in some segments of the production chain.

As data on the time structure of capacities is generally

missing for many of the modelled regions a vintage approach

of capacity accounting and planning was unfeasible to

introduce in themodel. Instead, the average economic lifetime

of the equipments (ranging from 14 to 20 years) was used to

account for the average lifetime of themachinery pools, which

were contracted every year by the average retirement share. A

backward looking expectation approach6 was used to deter-

mine the expected demand in the t + 10th years, and the

comparison of the expected demand and the expected

capacity availability will determine the need for new production

capacity. In themodel it is assumed that capacity ismaintained

during its lifetime, but old capacities are upgraded (retrofitted)

if its cost-effective compared to building new capacities. Both

the investment and variable costs are considered at any point

in time to determine the economic rationale of both options

(that are not mutually excluding), and depending on the type

of equipment available for potential retrofitting towards an

upgraded technology, the associated retrofitting costs and the

investment costs required for new installation. Investment

and retrofitting costs are annualised through the economic life

of the equipment.

The requirement for new capacity is based on the expected

production, installed capacity and retired capacity. The

installed and retired capacity is a straightforward aggregation

over time, while for estimating the expected production, a

trend of the last years production is used.

The specific investment costs of the retrofitting are

assumed to be a share of the overnight investment costs in

new capacity, this share depending on the specific technology

considered and the economic lifetime of the replaced and

replacing equipment.

The total production cost of a given process is the sum of

costs of the resources and the fixed and variable costs of

production. Fixed costs comprise annualized investment costs

and fixed operation and maintenance costs, which are

assumed to be a given share of the investment costs. The

exogenously prescribed variable costs include variable opera-

tion and maintenance costs, whereas resource costs include

fibrous raw material, i.e. wood, recycled paper, market pulp,

and other production inputs, such as chemicals, filling

materials etc.

6 Based on the average of the last ten years growth rates of thedemand, and assuming that this rate is maintained in the future.This myopic approach reflects the uncertainty the investors face

concerning their future expectations. Obviously pursuing a per-fect foresight approach is unrealistic here.which case its price increases. This price rise affects the costs

of the processes that use it. In this way the higher price goes

through the production chain and the end product price rises,

affecting its demand in the next time step.

3.4. Energy consumption and carbon emissions

Every process is (dynamically) characterised by a specific fuel,

steam and electricity consumption. The values of these

coefficients are based on the techno-economic characterisa-

tion of the different production routes. Typically, specific

energy demand decreases when new technological solutions

are taken in use.

Energy demand is divided into three categories: electricity,

heat (steam), fuel. These demands form three different

balances. The model development faced more limitations

when constructing the energy consumption module. The

International Energy Agency Energy Balances (2006) give

detailed information on the fossil fuel, combustible renewable

and waste energy and electricity used in the pulp and paper

sectors for the OECD countries. It also provides data for the

non-OECD countries, however it seems to be less reliable, as

many of the renewable and waste energy use is missing. An

additional constraint was, that data needed on the technology

level, where only rough estimations were available for the

developing regions. To overcome this problem, in the missing

regions the average energy use values of Table 1 were used

with a weighting scheme, where the weights are given by the

shares of the different technologies in place to closer match

the reported total energy use. This undoubtedly gives bias to

our energy consumption estimations and it only highlights the

pressing needs for more reliable data in the field. The third

statistical limitation was on the shares of on-site and grid

based electricity consumption, as IEA does not split the

electricityuse in thesector. In themodel average efficiencies for

the recovery boilers are assumed, which improve in time

according to the average industrial boiler efficiency improve-

ments of POLES, and only the remaining electricity need is

bought from the grid. This question becomes more important,

when accounting for the total carbon emission for the sector. In

themodel a full accounting approach is used, so in case of grid

electricity, emissions are also calculated, based on the POLES

projectionsonthe futurefossil fuelmixofelectricitygeneration,

and these emissions are attributed to the sector. Naturally, this

calculation is valid only for the stand-alone version of the

model, when coupled with POLES in the future, this split has to

bemade. Anadditional bias in themodel is that considering the

present functioning of the European Emission Trading Scheme

(ETS), the carbon taxes on electricity is paid by the power sector

and not the by sector demanding the electricity.7

Heat balance forms the core of the energy production

model. Electricity generation by CHP, if applied is a by-product

of steam production. In pulpmills themain energy generating

unit is a soda recovery boiler. It uses spent cooking liquor, or

7 The extent to which power producers could pass the price

increase of the carbon taxation on the consumer is out of thescope of this study.

black liquor, as a fuel. This boiler is a central part of the

chemical recovery device. In addition to the black liquor, wood

Future paper demand is calibrated on the historic values

following a procedure, where the logic of the intensity of use

Fig. 2 Regional paper demand and supply.

e n v i r onm en t a l s c i e n c e & p o l i c y 1 2 ( 2 0 0 9 ) 2 5 7 2 6 9 263wastes and bark frommechanical processing are used as fuels

in separate boilers to produce steam. Black liquor and wood

wastes are used first. After these, fossil fuels come in last to

meet the rest of the demand.All the fuels that are used for heat

production can also be used for electricity generation.

The emissions considered here are only those due to the use

of fossil fuels. In the version of the model described in this

paper, only CO2 emissions have been accounted for. If the

emissions are taxed, the price of each fossil fuel is modified

according to their carbon content. The new relative prices flow

through the production system as higher energy cost, raising

product prices accordingly.

3.5. Model data and calibration

Data of the model come from various information sources

listed in the references of this article, with the FAOSTAT

database being the most relevant to our work. Historical

values on pulp and paper production, trade are available there

for all paper grades and sub-product categories. This database

could also serve as the basis for the calculation of the apparent

consumption.Table 2 Socio-economic variables and the energy/carbon int

1980 1990 2

Population (million person)

OECD 1110 1178 12

Developing 3309 4060 48

GDP/capita (Ks)OECD 14.0 17.4

Developing 2.0 2.4

Paper Cons/GDP/capita (kg/Ks)OECD 8.6

Developing 5.1

Heat demand/paper prod (toe/t)

OECD 0.3

Developing 0.7

CO2 emissions/paper prod. (tCO2/t)

OECD 0.7

Developing 0.7hypothesis (introduced in Section 3.2) was reversed, and the

parameters alpha and beta (in Appendix B.2.1 in Supplemen-

tary Data) were calculated to minimise the deviation from the

historic consumption trend by using the GDP (in ppp terms)

and population values. Trade shares and production shares

are also calibrated to the historic values, however the long

term driving parameters (trade elasticity, price elasticities) are

exogenous. This called for a sensitivity analysis of these

parameters, where the main findings are presented in Section

4.3 and the detailed tables are placed in Appendix C in

Supplementary Data.

The historic trends in these main driving variables, and

their future projections for the Business as Usual scenario are

summarised in Table 2.

4. Model simulation results

4.1. Business as usual scenario

The Business As Usual (BaU) scenario shows an uninterrupted

growing trend for theworld paper demand. The average yearlyensities in the reference.

000 2010 2020 2030

30 1261 1281 1288

07 5530 6216 6794

20.4 25.7 31.6 37.8

3.4 4.9 6.6 8.4

8.6 8.6 7.5 6.6

5.1 5.0 4.9 4.7

0.3 0.3 0.2 0.2

0.8 0.8 0.8 0.7

0.7 0.7 0.7 0.7

0.8 0.8 0.8 0.7

growth rate is projected to be 2.1%, with the highest average

growth rate in Asia (4.1%) and the lowest in Europe (1%). In

World energy, technology and climate policy outlook (WETO,

European Commission, 2003), representing a delayed action

500 ppm concentration scenario. The delayed action means,

that in the early stages up till 20202025, depending on their

future development developing world countries face rather

soft commitments (or targets), and participate in the

reduction scheme through e.g. the clean development

mechanisms.

The scenario set-up can be summarised as follows. SFM

impacts are modelled through two impacts, first by introdu-

cing exogenously supply constraints of fibrous resources for a

9 The pulp and paper sector is already included in the ETS in thefirst trading period of 20052007, with paper and pulp capacitieshigher than 20 tonnes per day (EC Directive 2003/87).10 It is planned to connect the model to the POLES world energysimulation model in the near future, however for the policy exer-cise presented in this paper, the connection is limited to certain

e n v i r onm en t a l s c i e n c e & p o l i c y 1 2 ( 2 0 0 9 ) 2 5 7 2 6 9264spite of their high growth rates, South America and the Rest of

the World regions will still account for a small proportion of

the world total paper consumption in 2030. Undoubtedly Asia

will be the key actor in driving the future paper demand. China

and India will be accounted for most of this growth, where

China alone will represent more than 50% of the total Asian

demand in 2030. Fig. 2 illustrates these trends in paper

demand and supply for the aggregated regions of the model

and for the main paper grades. Amongst the grades board

papers show the highest increase (2.8%/year), while the

growth rates in the fine and news paper category are more

moderate.

In paper production the trends are similar to those of the

demand side, where the international trade to some extent

modifies the picture. Global trade also shows a strongly

increasing trend between 1.5 and 4% annual growth rates in

the different paper categories, and the trade patterns also

change significantly. The resource owners, mainly Latin

America and Russia steeply increase their shares in the raw

material part of international trade, they account for 75% of

raw wood export by 2030. Asia also becomes more active on

the end products part, increasing its share 30% on certain end

product paper in the export market. Due to its very intensive

demand growth, Asia also covers a significant share of its

consumption by import, so Asia is projected to be an

important actor on both side of international trade.

The pattern of energy use also varies significantly across

regions.8While Europe is only slowly increases its total energy

use in the sector, which is in line with its production trends,

the Asian and American regions show dynamically growing

energy demand in the Business as Usual scenario. The Asian

region has a massive contribution to the world energy

demand. There are high differences in the distribution of

the fuel used. Europe and South America already use close to

50% of fibrous resources (waste wood and black liquor) in

covering their heat demand, while Asia is lagging behind in

waste wood use. See Fig. 3 for more details. The two leading

countries in waste wood utilisation are Finland and Sweden

with close to two third shares in energy use, while this

potential is limited in other countries like Germany and Italy,

as their production is more focused on the end-products with

less access to waste wood resources.

Carbon emissions from the sector also rise with a yearly

average rate of 2%, where the most significant increase comes

from Asia, increasing its present 25% proportion close to 40%

by 2030.

4.2. A climate commitment scenario

In addition to the BaU run, a policy scenario was also run with

the model. This is a transition scenario representing a carbon

reduction committed future, in which the effects of the

changing forestry management practices are also included. In

this setup the sector is expected to contribute to the carbon

8 In the sector heat accounts for more than 90% of the energy

used, so referring to energy use in the paper sector mainly meansheat consumption.emission reductions required to achieve a GHG stabilisation

scenariomainly in two channels. Firstly, as the sector is the big

consumer of forestry products the expected changing

practices in forest management that are foreseen to take

place globally it has to cope with some disturbances on its

main raw material supply chain. The impacts of introducing

sustainable forest management (SFM) practices globally could

arrive to the sector through two main channels. First, supply

could be reduced in a transitional period, and additionally

there could be certain price increases due to the new forest

management practices. Due to the lack of quality data and the

long time scale involved it is difficult project these changes,

however the global forestry model of FAO (Global Fibre Supply

Model 1998) gives indication of the extent of these impacts.

Secondly, there is a direct impact on the sector, through its

high fuel use, brought by carbon constraint policies including

trading schemes (such as the European Trading Scheme) or

through carbon taxation.9 As the model is a stand-alone

version sector model,10 the carbon reduction target cannot be

determined within the model. To overcome this difficulty, the

carbon constraint commitment is represented by an exogen-

ous carbon tax, derived from the latest European Commission

Fig. 3 Heat use in the paper sector (for years 2000, 2015,

2030).exogenous variables: GDP, population projections and energyprices of POLES.

cen

)

w m

n a

nt

ons

lat

81.

00.

00.

05.

00.

00.

54.

e n v i r onm en t a l s c i e n c e & p o l i c y 1 2 ( 2 0 0 9 ) 2 5 7 2 6 9 265Table 3 Scenario assumptions.

Constraints Reference s

Carbon valueEurope 030 s/tC (20062030Carbon valuerest of the World None

Wood and other

fibrous resources restriction

Reference level of ra

availability

Wood and other fibrous

material price

According to BAU, o

1% growth in consta

a Year of introduction in brackets.

Table 4 The demand side effects of the BaU and carbon c

Population (Million) GDP/popu

EU

2030 Ref./2000 Ref. 103.5% 1

2030 only CV/2030 Ref. 100.0% 1

2030 Carb. C/2030 Ref. 100.0% 1

Asia

2030 Ref./2000 Ref. 132.0% 3

2030 only CV/2030 Ref. 100.0% 1

2030 Carb. C/2030 Ref. 100.0% 1

N. America

2030 Ref./2000 Ref. 129.0% 1certain transitional period, and additionally set the minimum

price increase to a range indicated by the FAO (1998) study.

Setting these limits exogenously are necessary in ourmodel, as

thefibre resourcemarket isonlymodelled inourmodel through

exogenously given available potentials, so not in the details as

GFSM does. These potentials are harmonised to the ones of

GFSM, while the constraints themselves are put in the logging

capacities (see Table 3 for values), that are already organic part

of our PULPSIM model. Introducing the GFSM estimates in the

scenario makes the model behaviour more realistic on the

resource market as well. The direct financial impacts of the

carbonconstraintcommitment is representedbyacarbonvalue

(CV) introduced in two stages. Europe faces the CV already by

2006, while the other regions start to face it only after the first

Kyoto period. It is linearly increasing froms0 tos140 per tonneof Carbon by 2030, a value which represents a median value of

thedifferent estimates for the500 ppmconcentrationpathway,

according to the delayed actions set-up.

The following table summarises the assumptions made to

arrive at the carbon constraint scenario.

In order to analyse separately the effect of the carbon value,

an analytical scenario was also run, which only differs from

the reference run by the CV. In the graphs and in the

subsequent text this scenario is named as the Only CV

scenario, whereas the mixed instruments policy case is called

Carbon Constraint scenario (abbreviated to Carbon const.).

The modelled carbon policy case triggers important effects

in the sector. Total paper production and heat consumption

runs parallel in the BaU and the Carbon Constraint scenario in

2030 only CV/2030 Ref. 100.0% 100.

2030 Carb. C/2030 Ref. 100.0% 100.ario Climate Commitment Scenario

0140 s/tC (20062030)0140 s/tC (20132030)

aterial Transitional resource supply

constraints introduceda

Asia: 2040% (2020)

Europe: 6% (2012)

North America: 12% (2012)

South America: 2040% (2020)

verage yearly

term

Minimum price increase introduced to

the transitional period

traint scenarios.

ion (kE/cap) Paper demand/GDP (kg/KE) Demand (Mt)

3% 75.3% 141.2%

0% 98.6% 98.6%

0% 98.1% 98.1%

7% 84.5% 341.1%

0% 96.6% 96.6%

0% 95.7% 95.7%

4% 70.9% 141.3%all of the five regions. In general, the carbon value alone

reduces paper production only to a minor extent in the range

of 23%. If it is applied together with the SFM practices, CO2emissions further reduce (as is shown by the following

Tables 4 and 5), but paper demand is reduced by more than

6% on world level, of which around half could be attributed to

the supply-side effects, and the generated demand responses.

Furthermore, the different paper grades seem to be reduced

proportionally.

More remarkable are the changes amongst the energy

fuels. Fossil-based energy consumption is reduced, while the

black liquor andwastewoodbased energies gain higher shares

in both the pure carbon taxation and mixed policy cases. This

is more apparent in the case of America and Asia, but holds

true for all the regions. South America and the ROW regions

(including Russia) even overtake Europe in the waste wood

and black liquor use, but it is not a surprise: they have

abundant source of the necessary raw material.

As the study focuses on the future energy consumption of

the sector, and incorporates measures both from the demand

and supply side of the sector, the following tables gives insight

to the most important effects, that take place in the model in

the three most important regions (Europe, Asia and North

America). These regions are not only themost important ones,

with more than 80% of the total world production, but also

represent regions with different characteristics for both their

economic development and their paper sector structure.

The analytical overview presented in the next part shows

the core mechanism of the model, based on an identity

0% 97.5% 97.5%

0% 95.7% 95.7%

Table 5 The supply side effects of the BaU and carbon constraint scenarios.

Paperproduction (Mt)

Energy use*/paper(toe/t)

Fossil fueluse/energy (toe/toe)

Emisson/fossilfuel (t/toe)

CO2emissions (Mt)

88.0% 103.4% 125.8%

e n v i r onm en t a l s c i e n c e & p o l i c y 1 2 ( 2 0 0 9 ) 2 5 7 2 6 9266decomposition approach. The carbon emission of a region is

decomposed in the following way:

CO2 Emissionsr Paper productionr Energy userPaper productionr Fossil fuel user

Energy user CO2 emissionsrFossil fuel user

This identityallowsforcapturingthedrivingforcesthatdrive

the future CO2 emissions in the sector. To capture the effects of

the demand side measures a further identity is introduced:

Paper demandr Populationr GDPrPopulationr Paper demandr

GDPr

International trade gives the connection points between

the two equations. Following this route, all the demand and

supply side effects could be captured in the future develop-

ment of the sector. Tables 4 and 5 shows the result of the

decomposition, where the first five columns deal with the

demand side changes and the last five with the supply side

effects taking place between 2000 and 2030. In the table the

percentage change values are presented, which compare the

changes in the reference scenario in time (between 2000 and

2030) and between the scenarios for 2030.

The first two columns summarise the exogenous driving

EU

2030 Ref./2000 Ref. 146.7% 94.1%

2030 only CV/2030 Ref. 98.1% 100.8%

2030 Carb. C/2030 Ref. 99.7% 99.6%

Asia

2030 Ref./2000 Ref. 337.3% 96.7%

2030 only CV/2030 Ref. 97.4% 100.9%

2030 Carb. C/2030 Ref. 95.8% 101.3%

N. America

2030 Ref./2000 Ref. 180.2% 94.1%

2030 only CV/2030 Ref. 98.6% 100.8%

2030 Carb. C/2030 Ref. 87.3% 101.2%

* Includes heat and electricity consumption.forces for the model, the population and GDP per capita

projections, and as the table shows these are extremely

influential driving forces.11 GDP per capita triples in Asia, but

also in Europe and North America more than 50% growth is

projected over the period of 20002030. This is coupled with a

30% population growth in Asia and North America (while

stagnating in Europe), and puts high pressure on the paper

demand on the global level. This pressure is only partly

relieved by dematerialisation represented in the third

column but these values are in the range of 2030% reduction

only in the BaU. Therefore the overall effect is a rapidly

increasing paper demand for Asia, whichmore than triples its

demand for paper, while 3040% increase also observable for

the developed world.

11 Projections of GDP and populations are POLES data, where GDPis based on CEPII forecasts, population data is based on the UNWorld Population Prospects medium variant (see European Com-mission, 2003 for more detailed description).In the policy case the impacts are minor on the demand

side if we consider that the values in the table represent the

effects for the CV over 15 years (starting in 2013 in the post

Kyoto period for regions outside Europe), and the introduc-

tion of the SFM practices covering 1520 years of a transition

period. In the carbon constraint policy case, a minor 35%

reduction in demand is envisaged, so the assumed policy has

minor effects on the demand side.

Table 5 is related to the product supply and to the energy

consumptionemissionpart of themodel, the core of analysis

of this study. Asia strongly accelerates capacity expansion, so

its production keeps pace with the steeply increasing demand

as the first column shows in the table.12 The developed regions

also significantly increase their production till 2030, but with a

much lower pace. As North America has the capacity and the

proximity of fibrous resources it follows a more dynamic

increase in production relative to Europe.

The energy consumption per unit of paper produced does

not change dramatically in the modelled period, the assumed

climate commitment policy has a rather minor impact. More

importantly, the effects of carbon taxation are also diluted, as

significant part of the energy consumption is based on heat

derived from waste wood and black liquor (3050%), assumed

to be carbon neutral. However the directions of the changes

97.0% 95.1% 91.1%

99.4% 95.0% 93.8%

96.4% 109.4% 343.8%

86.7% 91.0% 77.5%

86.1% 91.0% 76.0%

94.1% 109.5% 174.8%

68.0% 90.7% 61.3%

71.2% 90.7% 57.1%are remarkable. In the reference case all the regions reduce the

specific energy consumption of its paper production, but only

to a minor extent (36%). As the third column shows, all

regions follow a carbonisation trend in the reference case,

meaning increasing fossil fuel (mainly coal and oil) use. This is

a presently observed trendmainly in Chinaan action aiming

to reduce its rapidly growing energy dependency on oil. Given

the cost advantages of coal, the domestic availability and the

lack of a carbon constraint in the BaU scenario, it is not a

striking result for this region. This trend is true for Europe and

North America as well, as the numbers indicate, however to a

12 It is uncertain whether Chinawishes to pursue the same policyfor paper as for cement and steel and to cover its whole demand bydomestic production. No special treatment is provided in themodel to reflect this policy, so the results presented are basedon the model rationality. However if China goes along on thispath, it would significantly change the trade and most probablythe production patterns in the sector.

lesser extend for Europe (3.4%).What ismore surprising is that

the pure effect of the CV case is a further increase the specific

energy consumption. The changes in fossil fuel shares explain

e n v i r onm en t a l s c i e n c e & p o l i c y 1 2 ( 2 0 0 9 ) 2 5 7 2 6 9 267this rather unexpected effect. As the fossil fuel use is reduced

by the application of the carbon value, the sector uses more

waste wood fuels and black liquor to produce heat and

electricity, where the efficiencies are lower than for the fossil

fuel-based heat generation in the sector.13

Application of the full policy set-up assumed in the sector,

however, wouldmore significantly reduce the fossil fuel share

(up to 28% in North America), than the overall energy use. It

would also add to the effect of the de-carbonisation triggered

by the CV. The table shows the potential of the assumed policy

to generate fuel changes in the sector, which is a mixed effect

of changing the distribution amongst the fossil fuels (mainly

from coal to natural gas), and from fossils to wood and black

liquor-based heat and electricity generation. The effect is 10

11% reduction in emission per unit of fossil fuel used for Asia

and North America, while around 5% in Europe indicating that

in the baseline this region already has the cleanest energy

mix from the CO2 reduction potential point of view. The use of

recycling paper is also the highest here,meaning less room for

further manoeuvres.

The last column shows the resulting CO2 emissions (the

other GHG emissions are not yet modelled by PULPSIM). The

picture is alerting: CO2 emissions from the sector triple in the

forthcoming 30 years in Asia, if the Business as Usual trends

continue, North America emits more by 75%, but even the

European paper sector would produce 25% more CO2 gases in

2030 as now. However the model shows a high potential to

save in emissions in the modelled policy scenario. The carbon

taxation alone could reduce the emissions in the regions by 9

38%, and the full policy setup arrives to a slightly higher

reduction potential in some regions. The numbers in the last

column show that the measures other than CV could play a

role in achieving more stringent GHG reduction targets within

the sector. For example in Europe where the CV has less

significant impacts due to the already cleaner energy mix

further reduction could only be attained if other, mainly

technology based measures are also introduced.

4.3. Model sensitivity and verification analysis

As many of the driving parameters in the PULPSIM model are

exogenously set, based by literature review and expert

judgements, a detailed sensitivity analysis should reveal the

uncertainty range around its output. Additionally, as the

model has a long term outlook to the sector and as it is a new

application, it calls for an extensive check on these para-

meters. This sensitivity analysis also fulfils other purposes.

First, it gives an uncertainty range about the model results,

providing for relative ranges for the rather precise values

reported in the earlier chapters. Second, it helps to identify

those features of the model where further elaboration would

result in the highest gain in precision.

13 More widespread use of the black liquor gasification technol-

ogy would reverse this effect, as the study of Farahani et al. (2004)suggests.Carrying out a reasonable sensitivity analysis was a

challenging task, as it must accommodate with the recursive

simulation approach followed in the PULPSIMmodel. It means

inpractice, thatweperturb the system in twodifferentways: on

static, initial-value parameters as well as on dynamic assump-

tions (see Appendix C in Supplementary Data for details). The

analysis followed a two step approach: first individual para-

meters are shocked, than a multivariate analysis was carried

out to check the composite effects, if all the chosen parameters

are allowed to change simultaneously.

According to the sensitivity results the price elasticity, the

consumption level parameters and the life time of the

equipments are the most influential parameters, showing

the expected signs (directions) of changes. Concerning the

multivariate analysis, the model shows high variance for the

extreme values, but the standard deviation is within a 15%for the analysed variables. This is a reassuring range, if we

consider the long time scale involved in the model runs and

the ranges in which the parameters allowed to change.

A model verification experiment was also carried out in

order to verify the model functioning, when the last eight

years data covering the period of 19982005 were removed

from the input data file of the model. In general the model

arrived to a difference of 5-10% during the eight year periodof 19982005 for paper production, demand and export, which

equals to a yearly average 11.5%difference in terms of growth

rates. The difference pattern seems random, which means

that there is no systematic under- or overestimation by the

model, and the variations are not amplified radically in the

later periods. Only the international trade values show higher

fluctuations, which is a common feature of most economic

models, also reflecting the higher past fluctuations in the trade

data. The deviation is still under 20%, which means only

slightly above 0.5% a year (see Appendix C in Supplementary

Data for more details).

5. Conclusions

The article introduces a global bottom-up model of the pulp

and paper sector incorporating fundamental economic dri-

vers, in particular, product demand and international trade.

The PULPSIM model scrutinises the sector at the technology

level going down to the various process modules and, at the

same time, also describes the whole paper market with its

segmented and highly interlinked sub-markets. This techno-

logically rich representation of the sector can provide for

consistent long-term future narratives or storylines, bene-

fiting not only from the technological part of the sector, but

also from the considered market mechanisms.

The development of the sector is analysed under three

scenarios. The business as usual (BaU) scenario is comple-

mented with a carbon constraint one, where the sector faces

not only an increasing carbon tax in time, but reduced

resource availability caused by the introduced sustainable

forest management practices. These measures represent a

climate committed future for the sector compared to the

described reference path. In order to isolate the effect of thecarbon taxation, a third scenario considers only the increased

carbon value.

e n v i r onm en t a l s c i e n c e & p o l i c y 1 2 ( 2 0 0 9 ) 2 5 7 2 6 9268The simulation results show that the sector has a

distinctive responsiveness to the climate change policy in

comparison to other energy intensive sectors. In particular, as

wood and other fibrous resources serve as energy resources,

the sector could use this option to reduce its exposure to

carbon-constrained future developments. Yet implementing

this strategy involves risk as well: the resource must be

attained from sustainable forestry, which could be endan-

gered by the dynamic growth of the paper demand. Addition-

ally, biomass use faces increasing competition from other

sectors, such as power generation or transport.

The second insight from the article comes from the

analysis of the future trends in the CO2 emissions of the sector,

and the measures aiming at limiting its growth. In the BaU

scenario the sector is projected to increase its emission from

220 Mt of CO2 to over 400 Mt at world level between 2000 and

2030. The Carbon Constrained Future scenario shows a high

potential in carbon savings in the sector, up to 3040%

compared to the BaU.

The results also show that carbon taxation (or an emission

permit system) is a sharply targeted instrument reducing

carbon emissions while other aspects of the market function-

ing are less affected (e.g. minor reduction in demand for final

products). Other measures, such as efficiencymeasures or the

promotion of Black Liquor Combined Cycle technology, have

significant impact on the carbon performance of the sector.

The process of model development also revealed some

vulnerable points in the approach used: e.g. the pressing need

for more reliable data on certain fields (data on technology

levels and on the forest resource markets) and the need to

incorporate in the interaction amongst the forest resource

using sectors (e.g. power generation, pulp and paper,

construction). To by-pass this later difficulty exogenous

information sources are used in the model. The sensitivity

analysis carried out on these parameters of the model shows

that these technologically richmodels have their utility in long

term modelling, however as many of the driving parameters

possess high uncertainty range, the model should be

constrained to comparative scenario analysis, and the results

should not be treated for predictive purposes.

Future developments of the PULPSIMmodel could be along

the following directions. First of all, themodel is planned to be

linked to the POLES energymodel in order to take into account

the market interactions with the global energy markets.

Additionally, the improvement of the specification of the

international trade of forest and paper products would be

further explored.

The contribution of this paper to the existing model and

literature is twofold. First, it seeks to arrive to a modelling

scheme,where theglobal paper industry couldbecharacterised

in sufficient details for alternative scenario analysis. The

modelling approach applied in this model (e.g. technology

characterisation, setting up the pulp and paper categories) is

bounded fromtwosides: dataavailability and thepressingneed

formore technological details. Second, the proposedmodelling

solution seeks to answer questions concerning the sectors

contribution to the climate change issues. As the problem can

only be analysedwith long termmodels, theymust be ready inthe forthcoming years to sensibly contribute to the climate

change debate. Long term modelling requires, that the higheruncertainty involved in this modelling approach be treated

explicitly, and it is attempted to be achieved in our paper by a

detailed sensitivity and verification analysis.

Appendix A. Supplementary data

Supplementary data associated with this article can be

found, in the online version, at doi:10.1016/j.envsci.2009.01.011.

r e f e r e n c e s

Brannlund, R., Chung, Y., Fare, R., Grosskopf, S., 1998. Emissionstrading and profitability. Environmental and ResourceEconomics 12, 345356.

Bruvoll, A., Torstein, B., Larsson, J., Telle, K., 2003. Technologicalchanges in the pulp and paper industry and the role ofuniform versus selective environmental policy. DiscussionPapers No. 357. Statistics Norway; Research Department. Seealso: http://www.ssb.no/publikasjoner/DP/pdf/dp357.pdf.

Criqui, P., 1996. International markets and energy prices: thePOLES model. In: Lesourd, J.B., Percebois, J., Valette, P. (Eds.),Models for Energy Policy. Routledge, London, pp. 1429.

Davidsdottir, B., Ruth, M., 2004. Capital Vintage and climatechange policies: the case of US pulp and paper.Environmental Science & Policy 7, 221233.

Diesen, M. (Ed.), 1998. Economics of the Pulp and Paper Industry.TAPPI and the Finish Paper Engineers Association. ISBN952-5216-012.

Eriksson, H., Harvey, S., 2004. Black liquor gasificationconsequences for both industry and society. Energy 29,581612.

European Commission, 2003. World energy, technology andclimate policy outlook 2030WETO. Office for OfficialPublication of the European Communities, Luxembourg.ISBN 92 894 4186 0.

European Commission, EC 2003/87 directive establishing ascheme for greenhouse gas emission allowance tradingwithin the Community and amending Council Directive 96/61/EC. Official Journal of the European Union 2003.

Food and Agricultural Organisation of the United Nations (FAO),1998. Global Fibre Supply Model. See also: http://www.fao.org//DOCREP/006/X0105E/X0105E00.HTM.

Farahani, S., Worrell, E., Bryntse, G., 2004. CO2 free paper?Resources Conversation and Recycling 42, 317336.

Farla, J., Blok, K., Scipper, L., 1997. Energy efficiencydevelopments in the pulp and paper industry: a cross-country comparison using physical production data. EnergyPolicy 25, 745758.

Ince, P.J., 1998. Long-range outlook for U.S. paper andpaperboard demand, technology, and fiber supplydemandequilibria. In: USDA Forest Service, Forest ProductsLaboratory Proceedings.

Jaakko Poyry Ltd., 2006. World paper markets up to 2020. Know-how Wire Magazine, January 2006. Jaakko Poyry Ltd. Seealso: http://forestindustry.poyry.com/linked/en/news/KnowHowWire_January2006_internet.pdf.

JIRACS, 2003. Development of World Forest Products Model(WFPM). JIRACS Newsletter No. 35.

Kallio, M., Dykstra, D., Binkley, C. (Eds.), 1987. The Global ForestSector: An Analytical Perspective. John Wiley & Sons, NewYork.

Kallio, M., Moiseyev, I., Solberg, B., 2004. The Global Forest

Sector Model EFI-GTM. The Model Structure. EuropeanForest Institute, Internal Report No. 15.

Mannaerts, H., 2000. STREAM: substance throughput related toeconomic activity model: a partial equilibrium model formaterial flows in the economy. Netherlands Bureau forEconomic Policy Analysis, CPB Research Memorandum, No.165. The Hague. ISBN 90 5833 042 7.

Martin, N., Anglani, N., Einstein, D., Khurshch, M., Worrell, E.,Price, L.K., 2000. Opportunities to improve energy efficiencyand reduce greenhouse gas emissions in the U.S. Pulp andPaper Industry. Ernest Orlando Lawrence Berkeley NationalLaboratory. LBNL-46141. See also: http://www.energystar.gov/ia/business/industry/LBNL-46141.pdf.

Mollersten, K., Westermark, M., 2003. Potential and cost-effectiveness of CO2 reductions through energy measures inSwedish pulp and paper mills. Energy 28, 691710.

Ruth, M., Davidsdottir, B., Laitner, S., 2000. Impacts of market-based climate change policies on the US pulp and paperindustry. Energy Policy 28, 259270.

Szabo, L., Hidalgo, I., Ciscar, J.C., Soria, A., 2006. CO2 emissiontrading within the European Union and Annex B countries:the cement industry case. Energy Policy 34, 7287.

Tomberline, D., Zhu, S., Buongiorno, J., 1998. The Global ForestProducts Model (GFPM): Users Manual and Guide toInstallation. FAO Working Paper No. GFPOS/WP/02. ForestryPolicy and Planning Division, Rome.

Van Vuuren, D.P., Strengers, B.J., De Vries, H.J.M., 1999. Longterm perspectives on world metal usea system dynamicsmodel. Resources Policy 25, 239255.

International Energy Agency (IEA) (2006) Energy Balances.

Statistics Finland, http://www.stat.fi/.OECD, http://www.sourceoecd.org/.Eurostat, http://www.europa.eu.int/comm/eurostat/.Finnish Forest Research Institute, http://www.metla.fi/

julkaisut/index.htm.TAPPI, http://www.tappi.org.

Laszlo Szabo is a scientist at the Institute for Prospective Tech-nological Studies, DG JRC, where he deals with energy and climaterelated research issues. He is an economist, holding a PhD inenvironmental management. His work focuses on the modellingof the energy intensive sectors and assessing their long termclimate change impacts.

Antonio Soria is a scientist at the Institute for Prospective Tech-nological Studies, DG JRC, where he coordinates the work of theenergy and climate research group. He holds a PhD in industrialengineering, with main interest in long term modelling of energyand climate change issues and techno-economic characterisationof emerging technologies.

Juha Forsstrom holds a masters degree in engineering from Hel-sinki University of Technology. He has been working at VTT since1984. His main research interests are in energy system modellingand energy economics.

Janne T. Keranen is a physicist from Jyvaskyla University andholds an MSc in applied physics in papermaking technologies. Hehas worked from 1999 at VTT as research scientist focusing indrying and quality of paper.

e n v i r onm en t a l s c i e n c e & p o l i c y 1 2 ( 2 0 0 9 ) 2 5 7 2 6 9 269d a t a b a s e s a n d o t h e r d a t a s o u r c e s

FAO, http://www.fao.org/.World Resources Institute, http://www.wri.org/ and http://

earthtrends.wri.org/searchabledb/index.cfm?theme=9.Eemeli Hytonen is a research scientist at Technical ResearchCentre of Finland, VTT. He holds a MSc in applied physics. Hisresearch interests are in modelling based process design andintegration.

A world model of the pulp and paper industry: Demand, energy consumption and emission scenarios to 2030IntroductionPulp and paper industry overviewPaper-making processesPresent trends in paper consumption and trade

Model overviewMaterial and technology characterisationPaper demand and global tradeProduction module and capacity planningEnergy consumption and carbon emissionsModel data and calibration

Model simulation resultsBusiness as usual scenarioA climate commitment scenarioModel sensitivity and verification analysis

ConclusionsSupplementary data

ReferencesDatabases and other data sources