bod7 production function of the finnish pulp and paper industry

B O D 7 Production Function of the Finnish Pulp and Paper Industry

MINNA-MAARIH. KARVONEN*

Traditional neoclassical production theory analyzes the relationship in a production process between inputs and outputs which have a positive market value for the producer. The externalities of production, which have nonpositive market values, are discarded or included as the cost in a cost function. This paper studies the relationship between biological oxygen demand (BOD) emissions, an output of nonpositive value, and traditional factors of production, that is, investments, labor, output, and raw materials. An emissions production function is theoretically presented and empirically estimated with data from the Finnish pulp and paper industry. The approach is based on the observation that it is the minimization of effluents rather than, or together with, the maximization of yields, that increasingly defines the technological frontier of production processes. The empirical function estimation demonstrates the validity of the proposed novel modeling approach. (JEL 040, Q20)

Introduction

In many industries and in many countries, technological development has accelerated in the past decades. In some countries, as in the case of Finland, the growth of industrial output and technological development as its by-product has been especially pronounced in the postwar years. At the same time, increasing environmentalism has urged industrial designers and planners of new technology to incorporate aspects of environmental management and protection into the new technologies that are introduced to the market.

Emissions are an unwanted by-product of the production process of goods. They can be eliminated either by improving the process to produce less waste and emissions or by installing equipment that handles emissions in some sensible way. It is rather obvious that, for example, the amount of labor does not have a bearing on emissions reduction. The choice of raw materials and energy as well as the choice and age of technologies are decisive. This analysis focuses on the pulp and paper industry in Finland. Due to the abundance of virgin wood fiber as raw material, nonwood fibers are not used in pulping or are used only to a negligible extent. The analysis here focuses on the effects of investments on both emissions formation in the industry as well as on the use of wood fibers and the need for labor. It is assumed, a priori, that technological development is the most decisive factor in shaping the emissions profile of the industry.

The age of a mill or of its production equipment is a direct reflection of its level of technology. The latest technology is also the most advanced. Figure 1 shows the cumulative

* University of Jyv/iskyl~i--Finland.

310

KARVONEN: B O D 7 PRODUCTION FUNCTION 311

pulp production in Finland in thousand metric tons and the biological oxygen demand (BOD) emissions per metric ton.

FIGURE 1 Cumulative Pulp Production and BOD7 Emissions: 1994

18.00 16.00

o 14.00 12.00 10.00

o 8.00 6.00 4.00 © 2.00 0.00 I I I I I I

0 1,000 2,000 3,000 4,000 5,000 6,000

Pulp Production per Thousand Tons

~0

o

¢3

o Z

FIGURE 2 Normalized Water Emissions Coefficients: 1994

2.00

1.00

0.00 Mill 1 Mill 2 Mill 3 Mill 4 Mill 5 Mill 6 Mill 7

5.00

4.00

3.00

2.00

1.00

0.00

• Bop • Phospherous • Nitrogen • Absorbable Organic Halogen

Mill 8 Mill 9 Mill 10 Mill 11 Mill 12 Mill 13 Mill 14

312 IAER: AUGUST 2001, VOL. 7, NO. 3

The data reflect 14 individual mills. Figure 2 plots the normalized values per unit output of four water emissions for the same 14 pulp mills in the sample by increasing mill age, that is, mill 1 is the newest and mill 14 the oldest. In essence, Figure 2 is a plot of the age of the mill against relative emissions. The age is an estimate combining data on actual mill age with major expansion, retrofitting, and environmental investments. Clearly, the oldest mills are the most polluting.

Figures 1 and 2 are evidence that technological development, reflected here in the age of the mill, is the main determinant of the level of pollution. This paper begins with the premise that technological development, or mill age, is the determining variable with respect to the emissions profile of a modern pulp mill. Mill age is not a particularly straightforward variable to measure or use. As a result, the analysis here will focus on cumulative investments and their effects on the improvement of environmental performance.

Cumulative capital investments can be seen as a proxy variable of technical development. Indeed, the age of the production equipment is strongly associated with the amount of effluents that are emitted per each unit of output. Capital investments rather than equipment age is selected as a variable because of the ambiguities in defining the actual age of a digester or batch cooker, vis-g~-vis continuous improvement investments. There have been attempts to define the so-called technical age of a production facility. However, those types of calculations abound in ambiguities and are not employed here. The main aim here is to build a quantitative model that describes the emissions production in the industry. Since the age of the mill is difficult to define universally, it cannot be used in building the model, therefore, capital investments are substituted for age.

A production function is a technological statement or a model describing, through simplification, real life phenomena. Models are an essential tool for comprehension of complex systems and for choice making under alternative uncertain actions [Costanza and Ruth, 1998]. There is relative abundance with respect to economic environmental production (EEP) theory that ranges from the analysis of externalities [Pigou, 1932] to the inclusion of the cost of emissions as a restrictive factor [Baumol and Oates 1988; Pearce and Turner, 1990]. Natural production functions have been introduced in conjunction with studies on optimal resource harvesting, mainly of fisheries (for example, see Tahvonen [ 1989 ]). The se functions tend to describe, as does a traditional production function, the production of desired output, or the growth of the population, under polluted conditions.

Several different alternatives for presenting environmental variables, mainly emissions, in the economic theory of production have been suggested. They are helpful in demonstrating the logical sequence leading to the approach taken in this paper. Production functions are models, and based on these prior models, a new and hopefully improved model is developed here. This model presents a new approach to the problem of including environmental variables in production theory and includes emissions as a restrictive factor. The production of pollution is examined in the context of normal system operations. Next, this paper explains the data as well as the variables used in the study and their composition, demonstrates how the new model is built, and shows the empirical estimation of the model's parameters.


The Data and the Variables

This study includes Finnish integrated and nonintegrated pulp and paper mills for which time series data from 1972 to 1998 are used. The data on which the study is based has been collected by the Finnish Ministry of the Environment and the Finnish Environment Institute during permit compliance checks as well as the Finnish Forest Research Institute.l According to Finnish legislation, each mill receives a permit with plant specific standards for major airborne and waterborne emissions and effluents to be updated every three years. Compliance with permit levels is controlled by regular measurements. The measurement data over one year is first cleaned of the effects of other products, and then averaged for the mill over the year.

During the last few decades, the Finnish pulp and paper industry has been dominated by a few large companies. Today, Stora-Enso and UPM-Kymmene are the main players, which in 1998, constitute almost 80 percent of Finland's paper manufacturing capacity and almost an equal share of the pulp capacity [Finnish Forest Industries Federation, 1998]. The pulp and paper industry presents a large portion of the Finnish industrial structure and is the main environmental investor in the country.

The analysis here focuses on one effluent, BOD7, which indicates the amount of oxygen necessary for bacteria to consume the organic material released into water. Discharged organic materials cause rapid increases in the growth of microorganisms that, in turn, deplete the oxygen resources available for aquatic life [Biermann, 1993]. The chemical pulp and paper industry still accounts for roughly half of the BOD load of Finnish water bodies [Finnish Forest Industries Federation, 1998]. Publicly available data on two water emissions, BOD and solids, date back to 1950. Here, the data has been obtained directly from the Finnish Forest Industries Federation. Water emissions, especially BOD, mainly result from the pulping stage and not paper making, which is more able to utilize closed water cycles.

The pulp and paper industry is capital intensive. The natural investment cycles are relatively long, ranging on average between six and 15 years in Finland. The effects of any new investments penetrate a number of years and not merely the year in which the investment was made. In fact, the real effect can often only be felt ex post facto, after the investment has been running for a certain period of time. This is especially true in the case of most investments with an environmental impact. Spreading the effect over a number of years is partly a result of learning, that is, it takes time before the best results from a new investment can be reaped. So a new investment can increase efficiency for a few years before the effect levels off.

In this analysis, cumulative investments are chosen as one of the independent variables. It is assumed that cumulative investments reflect the development of the industry better than yearly investments, even if the latter were to be treated as lagged variables. Investment figures are obtained from the Finnish Forest Research Institute. The investment figures include traditional, both replacement and expansion investments, and environmental investments, both in end-of-pipe and process-integrated technologies. In short, all capital expenditures in the industry are included. This analysis assumed that each capital stock is used on average for 25 years, after which its value has depreciated substantially and the equipment is replaced. Thus, the cumulative investment is calculated as the sum of the investments of all past years, with an annual discount factor of 15 percent applied to all. This


depreciation rate is rather high in international comparison. Investment figures are deflated with the base year of 1998. The deflator is adapted from the official general money value or price index tables of Statistics Finland [1998]. It is not necessary to specifically use the deflator for investment goods.

Production figures are from the Finnish Forest Industries Federation [ 1998]. Most of the pulp produced in Finland is domestically made into paper and paperboard. Naturally, some double counting is introduced since pulp is used as the base raw material for paper making. However, since the two processes are separate, double counting does not affect the results to any significant extent and it appears that using total production or pulp production will only yield highly similar results.

Labor figures are obtained from Statistics Finland [1998] and include the number of persons employed, full-time per annum, in the pulp and paper industry in Finland. It does not include persons employed in supporting industries such as forestry or transportation of the goods. A renewal to meet the International Labor Organization and European Union definitions was made in the labor force survey during 1997 and 1998. Data used here has been updated since 1989 to meet these definitions. However, the effect of the change in the accounting procedure is negligible and is only mentioned.

The wood raw material figures include industrial thinnings and seed tree, shelterwood, and clear fellings. The main species are pine, spruce, and birch. Wood raw material figures are reported in cubic meters (m3). Data has been obtained from the Finnish Forest Research Institute. Table 1 shows the descriptive statistics, the means, and the variances of the variables.

TABLE 1 Descriptive Statistics of Employed Variables

Variables Description Mean Standard Deviation

E B O D 7 emissions in thousands of metric tons 197.89 143.82

K Indexed (1998 = 100) as cumulative capital 27,590.89 10,738.18 investments, depreciated at 15 percent annually, per million Finnish markka

V Virgin raw material, rn 3 3,045.85 1,136.09

L Labor, number of people 29,488.41 4,405.18

Q Combined production of pulp and paper, 15,592.85 4,131.87 in thousands of metric tons

The Theory of Production

The traditional economic production function discards environmental interventions and shows the maximal amounts of valuable outputs, which the producer can make of a given set of costly inputs [Cohen and Cyert, 1965; Mansfield, 1991; Chung, 1994]:


g ( 0 1 , . . . , O ) = f ( I 1 , . . . , I ) (1)

The boundaries of this function are defendable on the grounds that the producer makes economic decisions and that free goods do not affect the outcome of the decision. However, this function is not applicable to most industries today, where the cost of abatement of environmental interventions has become very high. For example, the rule of thumb is that the cost of installing air and water cleaning equipment adds an extra 15 percent to the investment costs of a modem pulp mill [Metsfiteollisuus, 1995], and running the equipment is a sizable and increasing cost to the mill. In response, the economic costs of the emission outputs have been included in (1) by considering them as a cost together with inputs. Such formulations are common, for example, in studies whose objective is to define an optimal tax rate for polluters [Barbera and McConnell, 1990; F/ire et al., 1993; Hetem~iki, 1996]. Work along this line has modified (1) by including unwanted outputs which have negative economic value to the firm and by grouping them with inputs [Pittman, 1983; Gollop and Roberts, 1983; Jorgensen and Wilcoxen, 1990; Barde, 1995].

The other objection to using (1) is theoretical. More than 25 years ago, Shephard [ 1974, p. 205] saw the need to include both economic variables and emissions in production functions: "since the production function is a technological statement, all outputs, whether economic goods are wanted or not, should be spanned by the output vectory." Environmental variables are rarely included in economic models in practice [Forssell and Polenske, 1998; Pesonen, 1998]. It is also argued that including an effluent in the production function, when treated as an input and priced at its tax, adds nothing essentially new to traditional theory [Uimonen, 1992].

A general EEP function of a production process [Pento, 1998a] incorporates emissions into their physical units without assigning arbitrary subjective notions such as money values. This can be expressed as the relationship between outputs with either positive or nonpositive market value and inputs that are either natural or man-made:

g ( 0 1 , . . . , O ; 0 + 1 . . . . , O ) = f ( I i , . . . ' I ; I ~ 1 ' "'" ' I k ) (2)

where outputs 01 , . . . , O are products with nonnegative economic value and Op + 1 ' "'" ' 0 • . P

are emissions and waste, which may or may not have a negative value for tile producer. Inputs I 1 . . . . , I k are man-made inputs, such as capital or labor, and inputs I k+ a ' "'" ' / are from the natural environment. The outputs of nonpositive value are not retained in the process and become environmental interventions in the form of emissions, waste, noise, and the like.

Optimization of the function in (2) requires maximizing the positive outputs (O 1 . . . . , O ) while minimizing the environmental interventions (O +. , . . . On) as well as those inputs

• . / ~ l '

(man-made and natural) that have a posmve market value (I1, ... , / , / + 1, "'", I , ) . Using (2) as a technological statement in place of (1) can be well justified by observing

how the technological frontier of a production process, or a product, is no more defined by


economic and technical variables, but is effectively constrained by the environmental demands. The economic performance of engines or power plants would be much higher were it not for the catalytic converters, scrubbers, and the like. Also, the yields of pulp mills would be higher if active chlorine could be still be used in bleaching pulp [Gullichsen, 1995]. Increasingly, the technological frontier is defined by environmental rather than economic considerations.

Traditionally, the objective has been to maximize the function output of the desired output product (or O., . . . , O ) while respecting the limits set by economic demands. Literature

1 p . . . .

concentrates on estimating (1), but the focus of this paper is on eshmatlng (2). Here, the aim is to minimize the function outputs or emissions. Within a given production infrastructure, emissions and output quantities move in the same direction: when output increases, so do emission levels, and optimal solutions in both terms are not feasible.

T h e M o d e l

Distance functions have been used in the derivation of shadow prices for the wrong created in the production process [F~ire et al., 1993; Chung et al., 1997; Hetem~iki, 1996]. The main advantage of a distance function is that it allows for modeling joint production of multiple outputs [Uimonen, 1992]. It also makes it possible to model the "wrongs" as not freely disposable and to derive shadow prices for the undesired product [F~ire et al., 1993]. Distance functions have the main drawbacks of being tedious to execute and their relatively poor empirical applicability when compared with traditional cost and production functions [Uimonen, 1992; Hetem~iki, 1996]. Distance functions rely on monetary figures, which can introduce significant bias in studies, during analysis, interpretation, and application and, thus, further reduce the applicability of results [B6ning, 1994; Vatn and Bromley, 1994; Pento, 1998b]. The results of such an analysis are difficult to interpret and have questionable, practical relevance.

Some literature moves away from monetary estimation and begins with the premise that discharged emissions (E) reflect the efficiency of a production process and defined emissions as some function of output or E = f (Q). This approach, suggested by Clift et al. [1999], defines independent variables as the functional outputs (Xl , . . . ,x i ) and the dependent variables as the environmental interventions (B 1 , . . . , B) . Inputs into the process are not included. The model then includes J in such equations:

B, = qo(X1, . . . ,Xa) . . . . ( j = 1 , , J ) (3 )

However, in this formulation, the reduction of environmental wrongs is only possible through the simultaneous reduction in the desired outputs, since no other independent variables are included [Chung et al., 1997]. This makes the approach unnecessarily restrictive.

In the early days of production theory development, natural capital variables did not restrict production, whereas capital and labor posed limitations [Boulding, 1966; Daly and Cobb, 1990; Daly, 1996]. With current legislation and constantly increasing legislative and consumer pressure, environmental restrictions severely limit the firm's production choices.


Moreover, most industries have become sufficiently automated to remove the availability of labor as a production constraint. This is not to say that labor is redundant, but that labor resources no longer constrain industrial activity. It is argued that natural resources are rapidly declining [Meadows et al., 1972; World Commission on Environment and Development, 1987; Tietenberg, 1992]. However, the observed fluctuation, both up and down in most commodity prices, cannot be attributed to decreasing natural capital stock.

Emissions are regarded here as a restrictive factor of the production process along with investment capital. In this approach, emissions are valued in their physical units, that is, not abstracting them into monetary terms. The familiar production function framework is used and the goal is to look at the implications that emissions have on marginal factor productivity. The restrictive nature of emissions would justify their inclusion as a factor input rather than an undesirable factor output. Thus, the general form of the studied function could be rewritten as:

Q = f ( K , E) (4)

However, emissions are not and should not be treated as an independent variable since that would not reflect their true nature, that is, that emissions are a function of the original production process, very much the same way as the desired output. Thus, if Q = f (K, L) and E = f ( Q ) , then it must hold that:

E = f ( K , L ) (5)

The optimization of this function involves minimizing E, emissions, which is the opposite of the traditional production function which maximizes output. This philosophical statement opens two alternative avenues: simplification or extension of the model. The extreme, simplification, is justified by economic analysis. However, to test the validity of the simplified model, it is necessary to also build and test models with multiple regressors and compare which of the models achieves greater accuracy in predicting real values. Starting with the simplest model and moving toward larger models makes it easier to see the real value of adding new regressors to the model.

Equation (5) serves as the starting point. As stated earlier, labor no longer restricts production, thus, it can be excluded from the function, reducing the equation to:

E . = f ( K ) , (6)

where Eit is emission, i, at time, t, and K, represents deflated cumulative investments at time, t. It is assumed, apriori, that a firm's investments in new types of production technology will decrease emissions per unit of output. This will lower the actual emissions and make it possible for the manufacturer to increase his production without exceeding the limits set for him by the authorities, thus effectively decreasing the restrictive impact of pollution and


increasing marginal returns of other factors of production. As a result, in this framework, any partial derivative of (6) must be negative, c3fr < O.

This first simplified model, (6), as well as all the subsequent, elaborated models are estimated assuming linearity in the variables and linearity in the parameters, but nonlinearity in the regressors. Thus, estimating (6) is done first in the linear format:

E . : 5+ x ,

and according to the Cobb-Douglas production function, approximation:

representing a first-order

Ei t = ~Z;=I Kt[I (8)

An empirical estimation of(8) requires taking logarithms, transforming the EEP function to:

/q

log E = log c~ + [3. log ~ t = l K" t (9)

To enable making predictions concerning the potential future level of emissions, the above Cobb-Douglas function with only cumulative investments may be too restrictive. As a result, another function is also estimated which includes the combined production quantities of pulp and paper (Q) alongside investments:

Eit = f (K, Q) (10)

To gain better insight into the process, (10) also serves as the basis for an elaborated, more holistic model through the inclusion of additional variables. This solution includes as many variables as can be reliably measured, for which data is available and that are variables in the production process. In this case, the emissions are modeled as a dependent variable of production, investments, labor (L), and wood raw material (V):

E, = f ( K , Q, L, V) (11)

All models are empirically estimated in multiple linear regression with the ordinary least squares (OLS) methodology. Table 2 shows the OLS estimation results, the goodness of fit and R 2 values, and results of the main statistical validation tests.

KARVONEN: B O D 7 PRODUCTION FUNCTION

TABLE 2 OLS Estimation Results for Production Functions

319

E = a +[3K E = a * K ~ E = a +~K+yQ Variables (7) (8) (10)

Constant 562.550 18.870 600.290 (12.8; 43.98) (2.42; 7.90) (16.40; 36.60)

K -0.0132 -1.380 -0.011 (.00043; -30.514) (0.24; -5.78) (0.001 ;- 15.87)

Q n/a n/a -0.006 (0.002; -3.11)

L n/a n/a n/a

V n/a n/a n/a

R 2 0.973 0.572 0.981

Adjusted R 2 0.973 0.554 0.980

DW 0.964 0.113 0.990

F-statistic* 931.125 33.353 631.680 (0.000) (0.000) (0.000)

BG LM X 2. 10.360 2t.180 10.690 (0.006) (0.005) (0.005)

White Z 2. 4.670 8.700 4.440 (0.097) (0.010) (0.488)

Variables E =o~ ,K~ , Q v

(lO) E=a+~K+yQ+)~L+coV E=a,Kf~,QV,Lx,V ~

(11) (11)

Constant 36.84 543.52 -25.54 (3.04; 12.13) (116.78; 4.65) (18.45; -1.38)

K -0.39 -0.011 -0.47 (0.206; 1.884) (0.0008; -14.489) (0.18; -2.48)

Q -2.91 -0.0045 -0.679 (0.432; -6.738) (0.0031; -1.451) (0.75; -0.91)

L n/a 0.0014 4.05 (0.0026; 0.555) (1. t9; 3.41)

V n/a -0.00462 -0.01 (0.004;- 1.1187) (0.10; -0.15)


TABLE 2 (CONT.)

Variables E :o~ ,K~ ,Q ~

(10) E:cz+~K+yQ+)~L+coV E:c~,K~,QV,Lz,V~'

(11) (11)

R z 0.852 0.982 0.903

Adjusted R 2 0.840 0.979 0.886

DW 0.453 1.0199 0.444

F-statistic* 69.000 308.740 51.442 (0.000) (0.000) (0.000)

BG LM X 2. 15.880 10.480 18.290 (0.000) (0.005) (0.000)

White )~2 * 5.510 12.690 14.940 (0.357) (0.550) (0.380)

Notes: * denotes p-values are in parentheses. DW denotes Durbin-Watson test, BG LM denotes Breusch-Godfrey Lagrange multiplier for serial correlation, White denotes the White test for heteroskedasticity, and n/a denotes not applicable. Standard errors and t-values are in parentheses.

Statistical Validity

The more holistic model, (11), with its high explanatory R 2 value but relatively few significant t-ratios suggest multicollinearity between the variables, that is, that a linear relationship exists between one or more of the regressors. Multicollinearity is a characteristic of the data and results in large variances and covariances for the OLS estimators. As a result, confidence intervals are wider and t-values are lower. However, multicollinearity does not violate any of the assumptions of regression and only results in fewer coefficients with small standard errors [Achen, 1982]. This same problem can be equally caused, for example, by a small sample size. In this case, the correlation coefficients between the independent variables are between 0.3 and 0.8 in absolute values. For all practical purposes, multicollinearity is not a problem but rather serves to validate the main argument.

Both the DW test for first-order serial correlation and the BG LM test for general, high- order serial correlation are used. The BG multiplier is assumed to follow a ;(2 distribution under the null hypothesis of no serial correlation. The BG test overcomes some of the constraining assumptions of the DW. In all except one of the explored models, there is presence of autocorrelation. This is only logical since the models employ a cumulative additive variable, investments. Since a cumulative variable is by definition nondecreasing, it is only assumed that the residuals will be time correlated.

Heteroskedasticity is tested using the White model, which calculates a regression of the residuals. The multiplier, reported in Table 2, is assumed to asymptotically follow a )~2


distribution. Results show that the Ho of no heteroskedasticity can be accepted at 95 percent confidence levels. This is expected because the data is of a dynamic time series rather than a cross-sectional sample.

Discussion

The estimation results reported in Table 2 confirm that the models have strong explanatory power, that is, they can give reasonably good estimates of the aggregate BOD 7 production function in the industry. Indeed, investments are the most decisive variable in affecting emissions formation, almost to the degree that they solely determine the emissions profile. This is especially true when investments are taken as a proxy for the age and capital structure of the equipment. This result is obvious from the strong multicollinearity of some of the additional regressors and the failure of the more holistic models to provide more accurate explanations than the simplest model.

Marginal returns have fluctuated over time. During the early years of the time period, marginal returns defined as additional reductions in emissions were relatively much higher than those of the latter years of the study. The late 1970s and early 1980s were still time for picking the "low-hanging fruit" in terms of environmental protection. As the industry progressed, it became increasingly difficult to further reduce emissions since they are already at a low level. The marginal returns of invested capital, or a partial derivative of the function (OBOD7/OK) , between 1972 and 1979 were multiples of those in latter years. Between 1972 and 1979, each markka reduced total B O D 7 emissions on average 7.0 grams, between 1980 and 1990, only 3.2 grams, and almost none, only 0.6 grams, between 1995 and 1998. This is shown in Figure 3 along with marginal returns from the viewpoint of emissions per output unit, that is, the reduction in grams per metric ton.

Figure 3 points out times when large-scale investments were made in process technologies that changed the production structure toward paper grades with higher value added, from the production of newsprint to the production of coated fine and magazine papers. 2 The result of those investments is a positive marginal return. However, for the majority of the time period, marginal returns have been negative, meaning that investments have resulted in decreasing emissions. The figure also shows decreasing marginal returns on average. Observations from the latter part of the time series are more cluttered around the zero line than those at the beginning of the time period.

Investments affect both economic and environmental aspects of production. Investments as a rule increase capacity and thus result in increased production given that capacity utilization rates stay constant, increase, or decrease only slightly. 3 If investments do not increase the total amount of production, they may result in production shifting to technically more advanced products with higher value added. Both result in an economically beneficial situation for the firm. In an environmental sense, advanced technology is more efficient and produces less effluents and waste per product unit than older, more obsolete technologies, that is, emissions per metric ton decrease. This decrease is sometimes, as in the case studied here, large enough to offset the increase in total output brought about by technology. The marginal returns of investments on pulp production capacity are shown in Figure 4. Clearly, the marginal return in a productive sense has not been dramatic. This is partly due to the fact that the investment figures reported include investments in the paper industry as well. From

322 IAER: A U G U S T 2001, VOL. 7, NO. 3

Figures 3 and 4, the posit ive environmental marginal return has decreased on average during

the studied t ime period, whereas the posit ive production capaci ty has increased and marginal

return o f investments has been relatively stable.

F I G U R E 3

M a r g i n a l R e t u r n s D e f i n e d as c3BODT/OK a n d O(BODTIt)/OK: 1 9 7 2 - 9 8

aBOD7/OK 2

"~ 0 -2 -4 -6 -8

-lO -12 -14 -16 -18

m

1972 74 76 78 80 I I I i I I I I I I I I I I I I I I I I I I I I I I I

82 84 86 88 90 92 94 96 98

1 . 5 - 1 . 0 -

o 0 . 5 - 0 . 0 -

-0.5 - -1.0 - -1.5

"-" 0 . 8 - 0 . 6 - 0 . 4 - 0 . 2 - 0 . 0 -

o -0.2 -

-0.4 - -0.6 -

~ -0.8 - -1.0

O(BOD7/ t ) / OK

I I I I I I I I I I I I I I I I I I I I I I I I I I I 1972 74 76 78 80 82 84 86 88 90 92 94 96 98

F I G U R E 4

M a r g i n a l R e t u r n s D e f i n e d as OQ/OK: 1 9 7 2 - 9 8

aQ/OK

I i I I I I I I I i I I i I i I I I I i I I I I I i I 1972 74 76 78 80 82 84 86 88 90 92 94 96 98


The environmental performance and ecoefficiency of the Finnish pulp and paper industry has increased significantly between 1972 and 1998. Emissions of BOD 7 have decreased in total amounts by 92 percent and the same trend can be seen for other waterborne effluents. At the same time, production of chemical pulp has increased by 80 percent and production of paper and paperboard, more than twofold. This results in a negative relationship between output and emissions. Technological innovation has decreased the emission-production ratio substantially both within processes and outside of them. It is precisely the increasing effectiveness of technological improvements that permit growth with minimal increases in pollution (see Carraro [ 1998] and Barbier [1999]). Technological improvements affect the ecoefficiency of an industry, as illustrated in this case by allowing growth while decreasing, not merely maintaining constant, its environmental burden. Growth has not harmed the environment. In fact, industry can now afford high levels of investments in environmental protection technologies. Productivity gain can be defined as increased output per unit of emissions [Repetto and Rothman, 1997]. Productivity has increased consistently during the last 20 years with technical advances. A similar result was found by Chung et al. [ 1997] for the Swedish pulp and paper industry between 1986 and 1990.

The decreases in B O D 7 have been achieved by increased utilization of internally recycling process water and by installing cleaning plants within the plant for water purification before it is released into communal sewage systems or lakes and rivers. Process modifications were also made by improving the bleaching processes in the oxygen and ozone stages. Increased internal recycling of important raw materials, such as certain chemicals, is also strongly encouraged.

Investment ratios of the industry remain constant at about 10 percent (of revenues), except during the recession in the early 1990s. At the beginning of the studied time period, environmental investments were mainly in the "end-of-pipe" technologies, often consisting of cleanup technologies. Later, the trend had shifted more toward process-integrated environmental investments that improve the process, causing less pollution and reducing the need for cleanup technologies. However, between 1992 and 1996, 53 percent [Statistics Finland, 1998] of all environmental investments of the forest industry were still in end-of- pipe technologies. This domination can partly be explained by the nature of the industry itself. Processes cannot be varied to a great extent. Possibilities simply do not exist for such radical changes, such as in the energy sector where different ways of producing energy (for example, low nitrogen oxide burners' are more abundant.

Conclusion

This paper presented the EEP function of BOD7 emissions for the Finnish pulp and paper industry between 1972 and 1998. The theoretical basis of such a function was found and developed with traditional economic tools. The theoretical development was empirically tested with an application in the pulp and paper industry. It was demonstrated that estimating an EEP function for BOD7 as a dependent variable of cumulative investments was both meaningful and provided easily interpretable results. Clearly, such a function exists and provides a powerful tool for analyzing the dynamics and interplay in the development of an industry. The application of EEP provided here, that is, one that relies on physical system


international measurement units instead of monetary values, successfully widens the perspective provided by conventional models and introduces a new basis for analysis.

Investments and technological change are the main drivers for most types of change in the industry. Most of the observed development in the different variables making up the profile of the industry can be attributed to technical change such as concerning emissions. This point is justified by the model parameter estimation which indicates strong linear relationships among all variables.

The EEP model provides a starting point and basis for different types of analysis. In this paper, it was only used to give an overview of the environmental effects caused by the industry during its growth and development. Applications of the model need to be further explored. The EEP may be very useful in policy setting, in general analysis of industry, and in evaluating the potential of technological innovation to enhance the performance and environmental impacts of entire industries. Another further area of research is broadening the objective of the function to simultaneously encompass economic, environmental, and social performance.

Footnotes

.

2.

.

Data are from the archives and permit decisions by the authorities for each mill. Collection and measurement are performed by the authorities. Fine and magazine papers require the use of more chemicals and fillers. They also use more pulp (up to 50 percent) than does newsprint (with an average pulp content of 10 percent). Precisely, pulp manufacturing is the main cause of BOD. The average capacity utilization rate in Finnish mills between 1980 and 1997 was 91 percent [Finnish Forest Industries Federation, 1998; Statistics Finland, 1998].

References

Achen, C. H. Interpreting and Using Regression, Thousand Oaks, CA: Sage Publications, 1982. Barbera, A. J.; McConnell, V. D. "The Impact of Environmental Regulation on Industry Productivity:

Direct and Indirect Effects," Journal of Environmental Economics and Management, 18, 1990, pp. 50-65.

Barbier, E. B. "Endogenous Growth and Natural Resource Scarcity," Environmental and Resource Economics, 14, 1, 1999, pp. 51-74.

Barde, J. P. "Environmental Policy and Policy Instruments," in H. Folmer; H. L. Gabel; H. Opschoor, eds., Principles of Environmental and Resource Economics, Cheltenham, United Kingdom: Edward Elgar, 1995, pp. 201-27.

Baumol, W. J.; Oates, W. E. The Theory ofEnvironmentalPoIicy, 2nd ed., New York, NY: Cambridge University Press, 1988.

Biermann, C. J. The Essential of Pulping andPapermaking, San Diego, CA: Academic Press, 1993. B6ning, J. A. Methoden Betrieblieher Okobilanzierung, Marburg, Germany: Metropolis Verlag, 1994. Boulding, K. E. "The Economics of the Coming Spaceship Earth," in H. Jarrett, ed., Environmental

Quality in a Growing Economy, Baltimore, MD: Johns Hopkins University Press, 1966, pp. 3-14. Carraro, C. "New Economic Theories: Impact on Environmental Economics," Environmental and

Resource Economics, 11, 3-4, April-June 1998, pp. 365-81.

KARVONEN: BOD 7 PRODUCTION FUNCTION 325

Chung, J. W. Utility and Production Functions, Oxford, United Kingdom: Blackwell, 1994. Chung, Y. H.; F~ire, R.; Grosskopf, S. "Productivity and Undesirable Outputs: A Directional Distance

Function Approach," Journal of Environmental Management, 51, 3, 1997, pp. 229-40. Cliff, R.; Frischknecht, R.; Huppes, G.; Tillman, A. M.; Weidema, B. "SETAC Working Groups: LCA

Section," SETAC Europe Newsletter, 10, 3, May 1999. Cohen, K.; Cyert, R. Theory of the Firm: Resource Allocation in a Market Economy, Englewood

Cliffs, NJ: Prentice Hall, 1965. Costanza, R.; Ruth, M. "Using Dynamic Modeling to Scope Environmental Problems and Build

Consensus," Environmental Management, 22, 2, 1998, pp. 183-95. Daly, H. E. Beyond Growth, Boston, MA: Beacon Press, 1996. Daly, H. E.; Cobb, J. B. For the Common Good." Redirecting the Economy Toward Community, the

Environment, and a Sustainable Future, London, United Kingdom: Merlin Press, 1990. F~ire, R.; Grosskopf, S.; Lovell, C. A. K.; Yaisawarng, S. "Derivation of Virtual Prices for Undesirable

Outputs: A Distance Function Approach," Review of Economics and Statistics, May 1993, pp. 374- 80.

Finnish Forest Industries Federation. Key to the Finnish Forest Industry, Helsinki, Finland: FFIF, 1998.

Forssell, O.; Polenske, K. "Introduction: Input-Output and the Environment," Economic Systems Research, 10, 2, 1998, pp. 91-8.

Gollop, F. M.; Roberts, M. I. "Environmental Regulations and Productivity Growth: The Case of Fossil-Fueled Electric Generation," Journal of Political Economy, 91, August 1983, pp. 654-74.

Gullichsen, J. "The Changing Kraft Pulping Process," proceedings from the International Chemical Recovery Conference, April 1995.

Hetem~ki, L. "Essays on the Impact of Pollution Control on a Firm: A Distance Function Approach," research paper, 609, Finnish Forest Research Institute, 1996.

Jorgensen, D.; Wilcoxen, P. J. "Environmental Regulation and U.S. Economic Growth," RAND Journal of Economics, 21, 2, 1990, pp. 314-40.

Mansfield, E. Microeconomics: Theory and Applications, 7th ed., New York, NY: W. W. Norton, 1991.

Meadows, D. H.; Meadows, D. L.; Randers, J.; Behrens, W. W. TheLimits to Growth, New York, NY: Universe Books, 1972, p. 205.

Mets~iteollisuus Ry. Ympi~ristOnsuojelun Vuosikirja, (Yearbook of Environmental Protection), Espoo, Finland: Metsgteollisuus Ry (Finnish Forestry Association), 1995.

Pearce, D. W.; Turner, R. K. Economics of Natural Resources and the Environment, Baltimore, MD: Johns Hopkins University Press, 1990.

Pento, T. "The Economic Theory of Environmental Life Cycle Inventory Models," in J. M. Baldasano; H. Power, eds., Environmental Engineering and Management, Southampton, United Kingdom: WIT Press, 1998a, pp. 103-12.

Pento, T. "Integrating Environmental Management Instruments: Material Flow Analysis," in B. Nath; L. Hens; P. Compton; D. Devuyst, eds., EnvironmentalManagement in Practice, 1, New York, NY: Routledge, 1998b, pp. 241-54.

Pesonen, H. L. "From Material Flows to Cash Flows: An Extension to Traditional Material Flow Modeling," dissertation, 24, University of Jyv~iskyl/i, 1998.

Pigou, A. C. The Economics of Welfare, 4th ed., London, United Kingdom: Macmillan, 1932. Pittman, R. W. "Multilateral Productivity Comparisons with Undesirable Outputs," Economic Journal,

1983, pp. 883-91. Repetto, R.; Rothman, D. "Has Environmental Protection Really Reduced Productivity Growth?,"

Challenge, 40, 1, 1997, pp. 46-58.


Shephard, R. "Comment on E. Diewert's 'Applications of Duality Theory'," in M. D. Intrilligator; D. A. Kendrick, eds., Frontiers of Quantitative Economics, 2, Amsterdam, Netherlands: North- Holland, 1974.

Statistics Finland. EnvironmentaI Protection Expenditure by Finnish Industry: 1996, Helsinki, Finland: Statistics Finland, 1998.

Tahvonen, O. "On the Dynamics of Renewable Resource Harvesting and Optimal Pollution Control," working paper, A67, Helsinki School of Economics, 1989.

Tietenberg, T. Environmental and Natural Resource Economics, 3rd ed., New York, NY: Harper Collins, 1992.

Uimonen, S. "Pollution Control in the Firm and the Industry: A Theoretical Study," dissertation, 62, Annales Academie Scientarium Fennicae, 1992.

Vatn, A.; Bromley, D. "Choices Without Prices, Without Apologies," Journal of Environmental Economics and Management, 26, 2, 1994, pp. 126-48.

World Commission on Environment and Development. Our Common Future, Oxford, United Kingdom: Oxford University Press, 1987.

bod7 production function of the finnish pulp and paper industry

Documents