TUTKIMUSRAPORTTEJA - REPORTS

34

Structural Change in US Newsprint Demand: GDP and Price Elasticities. Jari Kuuluvainen*

HELSINKI 2004

TUTKIMUSRAPORTTEJA - REPORTS 34

Structural Change in US Newsprint Demand: GDP and Price Elasticities. Jari Kuuluvainen*

Helsingin yliopisto, metsäekonomian laitos University of Helsinki, Department of Forest Economics PL 27 FIN-00014 HELSINGIN YLIOPISTO, FINLAND Telephone: (+358) 09 1911 Telefax: (+358) 09 191 57984 ISSN 1236-6218 ISBN 952-10-2217-5 Helsinki University Press Helsinki August 2003

ii

iii

Jari Kuuluvainen. 2004. Structural Change in US Newsprint Demand: GDP and Price Elasticities. Abstract: A simple classical newspaper demand function for the United States for the period 1961-2002 is estimated. A structural break in the mid-1980s as pointed out by Hetemäki and Oberstainer (2002), is confirmed. Allowing for the structural break by including a time trend for technical change in consumption, a positive elasticity of demand with respect to real national product and a negative elasticity of demand with respect to price predicted by the theory are estimated. Therefore, the classical demand function can be used in short-term forecasting, but more information is needed to discover the causes of the structural break and its effects on newsprint demand.

Authors’ address: Department of Forest Economics, P.O. Box 27 00014 University of Helsinki email: [email protected] Acknowledgements: I thank Lauri Hetemäki, Anne Toppinen, Mikko Tervo and Jussi Uusivuori for useful discussions. Errors are my own responsibility.

iv

1

Introduction

Printed media is experiencing increasing competition from digital information technology. Newsprint production and trade especially may be affected. This phenomenon is new and there is still little scientific evidence on this topic, although a large body of literature on empirical models of demand exists for different forest products. The classical demand functions for newsprint assume that newsprint demand is positively correlated with income and negatively correlated with the price of newsprint. A problem with these classical demand formulations is that information technology may cause a structural change in the demand for newsprint and other printed media (Hetemäki and Obersteiner, 2001). An interesting ‘leading indicator’ for future development may turn out to be the newspaper market in the United States. The United States has traditionally been the world’s largest newsprint market where per capita newspaper consumption has been high for decades. Only recently, the signs of possible stagnation in the demand for news print have started to become visible. Even in the mid-1990s Zhang and Buongiorno (1997) concluded that electronic media had not markedly affected printed media during the period from 1961 to 1991. However, among other factors, the use of the internet exploded during the 1990s. In the year 1984 there were about 1000 internet users in the world, in 1996 the number had increased to 12 million, and is estimated to have been 171 million in 2003. More recently, Hetemäki and Obersteiner (2001) showed that the US paper demand appears to have gone through a structural change in the middle of the 1980s. They claim that a classical forest product demand model that uses GNP and price as explanatory variables cannot explain or forecast this structural change in newsprint

2

consumption. Furthermore, both price and income become insignificant determinants of the newsprint demand in 1987-2000 in the ‘classical’ demand equation. Hetemäki and Obersteiner (2001) conclude that GNP elasticity may even be negative. To improve the forecasting power of the specification, Hetemäki and Oberstainer (2001) estimate a Bayesian model and have found that this approach is better suited for forecasting than the classical model in the US case. Bolkesjø et al. (2003) used a Bayesian approach to study newsprint demand in Japan and Europe with panel data. They did not find evidence for a similar structural change to the one which seems to be taking place in the US. However, they do recommend a Bayesian approach to forecasting. In this paper, we take a different point of view. The structural change pointed out by Hetemäki and Obersteiner (2001) can be detected in the time series data but, in spite of the structural break, a classical model may still be useful at least in short-term forecasting. However, more information than just newsprint prices and income (GDP) is needed for valid long-term forecasts. As pointed out by Simangunsong and Buongiorno, (2001) and Buongiorno et al., (2001), a negative income elasticity is inconsistent with the economic theory, when inferior goods are disregarded, and may reflect a problem in the model, the statistical inference or the data. Therefore, the classical model used to forecast newsprint consumption, after the structural break, may suffer from omitted variable bias which could take into account, e.g., the possible substitution between newsprint and information technology and perhaps some other structural changes in the markets. The problem is that it is very difficult to measure relevant prices of the information technology to be included into the model (Zhang and Buogiorno 1997). To avoid omitted variable bias, Haynes (2001) included in the model, in addition to newsprint price and GDP, a print media price index, capital price, prices of TVs and

3

radios, computer prices and a demand calibration dummy. The last mentioned variable was used to improve the forecasting performance of the model. Unfortunately, detailed estimation results are not reported in Haynes (2001). This paper reconsiders the classical forest products demand model and claims that both the income and price elasticities have theoretically correct signs even after the structural break in the mid 1980s. This indicates that the connection between prices, income and consumption may not have changed, but the substitution from electronic media and other possible factors should be allowed, as was done in Zhang and Buongiorno (1997) and Haynes (2001). In this paper we allow for the structural break by using the dummy variable and a trend and by estimating the model for the whole period and the two periods separately. The models estimated before and after the structural break are not able to explain the structural change nor are they suited for longterm forecasting, but serve to demonstrate the fact that the income and price elasticities measured may be fairly persistent even during structural changes in the market, as commonly assumed in economics.

2. Derivation of the empirical model

Previous econometric studies on paper demand often assume a Cobb-Douglas production function for paper-using industry. The cost minimisation problem of the paper user can be written as (Bolkesjö et al. 2003). Min c(pn,po,y) = pnx+poz,

s.t y=axbzc,

where c(.) is production costs, pn and po are paper price and price of other inputs respectively, x and z are paper consumption and other input consumption in production, y is the production of final output, and a, b and c are the parameters of the production

4

function. Defining β1= 1(b+c) and β2= -c/(b+c), the conditional demand function for newspaper can be written as

p x(px,pz,y) = β 0 y β1 ( x ) β 2 pz Denoting the relative price of inputs by p=px/py and including the possible dynamics to the model (to be determined by the empirical data), the short-term demand function for news paper can be written as: β

β

β

x(px,pz,y)= β 0 y β1 p β 2 p −13 y −14 x −15 , from which the long run elasticities can be solved as functions of the short-run elasticities. It is convenient to estimate this relationship in logarithmic form and to use GDP as the proxy for y (cf. Simansungsong and Buogiorno 2001). Note that we have not included prices that describe, for example, IT technology and electronic media. This technical change will be taken into account using a linear trend and a dummy variable.

3. Data

The data that we use is from publicly available statistics. Newsprint quantities consumed, import and export volume and the value of newsprint are from the FAO database (FAO 2003). Domestic prices for newspaper consumption in the United States are quantity weighted averages of the export and import unit values of the year in question (Chas-Amal and Buongiorno, 2000). The GDP and producer price index for the United States were obtained from the data base of the Economic Research Institute

5

of Finland (ETLA 2003). Prices and the GDP used in the analysis are deflated using the producer price index. The observations are from the years 1960-2002.

4. Results

Model without structural break In this section we first present the results of time series models estimated using a data set on US paper consumption in 1961-2002 without allowing for the structural change in the mid 1980s in order to demonstrate the possible omitted variable bias in the price and GDP elasticities. Table 1 presents the results of a dynamic model with newsprint price and real GDP as explanatory variables of paper demand, without accounting for the structural break. According to the diagnostic tests the short-term model behaves reasonably well, apart from possible heteroscedasticity. However, the estimated static long-run model gives a positive price elasticity and a negative income elasticity and is not statistically significant. Furthermore, the Wald test accepts a common factor. This is already indicated by the coefficients of the present and the lagged values of prices and income that are close to each other in absolute terms, with opposite signs. Thus, there does not seems to be an equilibrium relationship between the consumption of newsprint and prices and income, which, however, has been the basis of many models forecasting newspaper demand since Buongiorno (1978, see Simangunsong and Buongiorno 2001 for a review). The Chowbreak point test for the periods of 1961-1986 and 1987-2002 gives test statistics of 12.73 (6,29), which is well over the critical value of 3.50.

6

Hetemäki and Obersteiner (2001) were the first to point out the fact that the equilibrium relationship between GDP and paper consumption may have broken down in US paper consumption in the late 1980s. This result is also confirmed by the present sample, if separate models are estimated using the above specification for the period of 1961-1986 and 1987-2002. The solved static long-run results of the model estimated for these two

Table 1. Estimated results for newspaper demand in the US in 1961-2002 without taking into account the structural break in the data. Variable Constant lncons −1 lnrp lnrp−1 lnrgdp lnrgdp−1

Coefficient Std.Error 0.40743 0.59241 0.97504 0.11320 -0.1317 0.08658 0.13449 0.08499 0.71568 0.25074 -0.74004 0.25851 -2.863

t-value 0.688 8.613 -1.521 1.582 2.854

Part R2 0.0137 0.6857 0.0637 0.0686 0.1933

t-prob 0.4963 0.0000 0.1375 0.1229 0.0073 0.0071

0.1942

R^2 = 0.95202 F(5,34) = 134.93 [0.0000] \sigma = 0.0449643 DW = 2.23 RSS = 0.06874092105 for 6 variables and 40 observations AR 1- 2 F( 2, 32) = ARCH 1 F( 1, 32) = Normality Chi^2(2)= Xi^2 F(10, 23) = Xi*Xj F(20, 13) = RESET F( 1, 33) =

1.2879 0.2920 3.2218 0.9403 3.0128 1.0384

[0.2898] [0.5926] [0.1997] [0.5165] [0.0229] * [0.3156]

Solved static long-run equation lncons = +16.32 + 0.1116 lnrp − 0.9756 lnrgdp (SE) (56.33) (1.856) (6.788) 2 WALD test χ (2) = 0.0237 [0.9882] COMFAC WALD Order df Chi^2 Value p-value 1 2 2.2635 [0.3225] Solved static lonr-run equation (1961-1986) lncons = 4.80964 - 0.303927*lnrp + 0.761531*lnrgdp; (SE) (0.2094) (0.0374) (0.02519) WALD test: Chi^2(2) = 1050.31 [0.0000] **

Solved Static equation (1987-2002) lncons = 17.0124 - 0.317771*lnrp - 0.638006*lnrgdp; (SE) (10.45) (0.5798) (0.7832) WALD test: Chi^2(2) = 1.5441 [0.4621]

7

periods are reported in table 1 (see appendix for dynamic models). The dynamic specifications are given in the appendix. The model for the period 1961-1986 behaves as expected from the economic theory, and the long-run relationship is statistically significant. However, for 1987-2002 the solved static model is not statistically significant and GDP has an unexpected negative sign. We note in passing that the adjustment parameter is negative but close to zero for 1962-1986, while positive (0.53) for 1987- 2002. This result is similar to Hetemäki and Oberstainer (2001), who estimated a much smaller adjustment parameter (0.16) indicating faster adjustment for 1971-1987 than that (0.66) estimated for 1987-2000. The next step is to take into account the structural break in the estimation.

Model with structural break Economists generally tend to share the view that price and income elasticities are fairly stable. Thus, even if there is a structural break in the market, the good does not necessarily turn into an inferior good that would be indicated by a negative income elasticity. Therefore, the other explanation for the negative income elasticity estimated above might be omitted variable bias. To give some insight into this question in the simplest possible method, the model in table 2 accounts for the structural break. A time trend for 1987 – 2002 and a dummy variable, which is one for 1961-1986 and zero thereafter, are included. The first lag of real GDP is omitted as its coefficient was close to zero in absolute terms with a large standard error. The dynamic specification is statistically valid, passing all the diagnostic tests, and all coefficients seem to be statistically significant, disregarding the lagged endogenous variable. The lagged endogenous variable has a coefficient, that is close to

8

zero and negative. This would mean (if significant) that newspaper demand actually adjusts to the equilibrium level within a year. Consumption and real GDP are nonstationary according to Dickey-Fuller tests, real price may be trend stationary, and the tvalues may not have their normal interpretations. More interestingly, however, the solved static model is significant with a negative price elasticity (-0.36) and a positive elasticity with respect to a real GDP of 0.76. Thus, the model estimated for the whole

Table 2. Estimation results for newspaper demand in the United States 1961-2002. Variable

Coefficient

Std.Error

t-value

t-prob

Part R2

Constant lncons −1 lnrp lnrp −1 lnrgdp trend2 D1

6.2199 -0.074308 -0.12127 -0.19770 0.81949 -0.039344 -1.1170

0.68242 0.12227 0.05132 0.05878 0.09205 0.00427 0.12580

9.114 -0.608 -2.363 -3.363 8.902 -9.205 -8.879

0.0000 0.5475 0.0242 0.0020 0.0000 0.0000 0.0000

0.7157 0.0111 0.1447 0.2553 0.7060 0.7197 0.7049

R^2 = 0.983829 F(6,33) = 334.62 [0.0000] \sigma = 0.0264964 DW = 1.87 RSS = 0.02316794676 for 7 variables and 40 observations, AR 1- 2 F( 2, 31) = ARCH 1 F( 1, 31) = Normality Chi^2(2)= Xi^2 F(11, 21) = Xi*Xj F(25, 7) = RESET F( 1, 32) =

0.0558 0.2950 1.5383 1.6128 0.5936 0.0124

[0.9458] [0.5909] [0.4634] [0.1667] [0.8414] [0.9118]

Solved static long-run equation (1961-2002) lncons = 5.79 − 0.2969 lnrp + 0.7628 lnrgdp −0.03662 trend2 − 1.04 D1 (SE) (0.2303) (0.0363) ( 0.02509) ( 0.002219) ( 0.065) 2 WALD test χ (4) = 2213.9 [0.0000] ** Solved static long-run equation (1987-2002) lncons = 1.43085 - 0.470696*lnrp + 1.47971*lnrgdp - 0.0644319*trend2; SE (3.487) (0.2662) (0.5978) (0.0251) WALD test: Chi^2(3) = 12.6104 [0.0056] **

9

period allowing for the structural break actually indicates that the equilibrium income elasticity may in fact be positive even after the structural break and that the price elasticity is negative as predicted by the theory. The absolute values of the elasticities are in line with those estimated earlier. The number of observations is too small for definite conclusions. However, it is still of interest to estimate the models separately for the two periods. The dynamicspecification for 1961-1986 is similar to that presented in table 1 and is not reported here. For the period 1987-2002 only the time trend allowing for the technical change in consumption is added. Diagnostic tests do not indicate any problems with the specification for 1961-1986, nor for the period 1987-2002, although for this period the number of observations is too small to test for heteroscedasticity. We only report the solved static models here (dynamic model with trend for 1987-2002) is reported in the appedix). Both models are statistically significant according to the Wald statistics. The elasticities estimated for the period 1961-1986 are close to those estimated for the whole period. More interestingly, the elasticities for price and GDP are greater in absolute terms for the period 1987-2002 and are of the sign expected for them, negative for price and positive for the income. The income elasticity is actually almost the same as reported by Bolkesjø et al. (2003), which was estimated for Europe for 1971-1999 using two different panel data sets. However, it must be noted that there is a strong negative correlation between the estimated trend and the real gross national product. Therefore, in order to make more definite conclusions on what the GDP elasticity is, presently prices or other variables accounting for the technical change in consumption should be identified.

10

Conclusions We estimated a simple “classical” model for US newspaper demand for the period 1961 – 2002, in order to study the effects of possible structural change in US newsprint demand on price and income elasticities. Our results confirm the structural break pointed out by Hetemäki and Oberstainer (2001). However, an inclusion of a simple linear time trend for 1987-2001 to allow for the technical change in consumption, makes the elasticity of newspaper demand with respect to real GDP positive and larger in absolute terms than that estimated for the period 1961-1986. Of course, this model is not able to detect the causes for the decreasing trend in demand for newsprint since the late 1980s. It may not be presently possible to measure the relevant relative prices or other factors that should be included into the demand specifications for paper products to avoid the omitted variable bias. Therefore, e.g., the Bayesian methods suggested by Hetemäki and Oberstainer (2001) and Bolkesjø et al. (2003) may be a solution to practical long-term forecasting. However, more work to detect the present determinants of newsprint demand and also possibly the demand for other forest industry products as well is required. Meanwhile, at least in the short term, the “classical” demand equation including the time trend for technical development may still be valid in short-term forecasting.

References: Bolkesjø, T. F., Oberstainer, M. and Solberg, B. 2003. Information technology and the newsprint demand in Western Europe: a Bayesian Approach. Can. J. For. Res., 33: 1644-1652. Buongiorno, J. 1978. Income and price elasticities in the world demand for paper and paperboard. For. Sci. 33: 185-197. Buongiorno, J. , Liu, C.-S. and Turner, J. 2001. Estimating International Wood and Fiber Utilization Accounts in Presence of Measurement Errors. Journal of Forest Economics, 7: 101-128.

11

Chas-Amil, M. and Buongiorno, J. 2000. The demand for paper and paperboard: Econometric models for the European Union. Appl. Econ. 32: 987-974. Haynes, R. 2001. An Analysis of the Timber Situation in the United States: 1997-2050. A General Technical Report PNW-GTR-560. USDA Forest Service, Pasific-North-West Research Station, Portland, Oregon, USA. Hetemäki L. and Oberstainer, M 2001. US newsprint demand forecasts to 2020. International Institute for Applied Systems Analysis, Laxenburg, Austria. Interim Report. IR-01-070. Simangunsong, B.C.H. and Buongiorno, J. 2001. International demand Equations for Forest Products: A Comparision of Methods. Scandinavian Journal of Forest Research, 16: 155-172. Zhang, Y. and Buogiorno, J. 1997. Communication Media and Demand for Printing and Publishing Papers in the United States. Forest Science, 43: 362-377.

12

Appedix Table A1. US newsprint demand for 1961-1986, dynamic model

lncons_1 Constant lnrp lnrp_1 lnrgdp lnrgdp_1

Coefficient -0.130573 5.43765 -0.126854 -0.216758 0.709269 0.151698

sigma 0.0275722 R^2 0.978055 log-likelihood 57.7307 no. of observations 25 mean(lncons) 9.14445

Std.Error 0.2007 0.9677 0.06761 0.09466 0.1601 0.2115

t-value -0.650 5.62 -1.88 -2.29 4.43 0.717

t-prob 0.523 0.000 0.076 0.034 0.000 0.482

Part.R^2 0.0218 0.6243 0.1563 0.2163 0.5081 0.0264

RSS 0.0144442935 F(5,19) = 169.4 [0.000]** DW 1.83 no. of parameters 6 var(lncons) 0.0263287

Table A2. US newsprint demand for 1987-2002, dynamic model

lncons_1 Constant lnrp lnrp_1 lnrgdp lnrgdp_1

Coefficient 0.534361 7.92164 -0.107134 -0.0408323 1.72024 -2.01732

sigma 0.0337418 R^2 0.579278 log-likelihood 33.3824 no. of observations 15 mean(lncons) 9.42731

Std.Error 0.4518 3.621 0.1616 0.1133 0.8389 0.8934

t-value 1.18 2.19 -0.663 -0.360 2.05 -2.26

t-prob 0.267 0.056 0.524 0.727 0.071 0.050

Part.R^2 0.1345 0.3472 0.0466 0.0142 0.3184 0.3616

RSS 0.010246568 F(5,9) = 2.478 [0.112] DW 1.75 no. of parameters 6 var(lncons) 0.00162365

Table A3. US newsprint demand for 1987-2002, dynamic model with trend

lncons_1 Constant lnrp lnrp_1 lnrgdp lnrgdp_1 trend2

Coefficient Std.Error 0.105008 0.3437 1.28060 3.273 -0.214312 0.1181 -0.206958 0.09476 2.34380 0.6192 -1.01947 0.6989 -0.0576660 0.01795

sigma 0.0236481 R^2 0.816304 log-likelihood 39.5976 no. of observations 15 mean(lncons) 9.42731

t-value 0.306 0.391 -1.82 -2.18 3.79 -1.46 -3.21

RSS 0.00447386532 F(6,8) = 5.925 [0.012]* DW 1.86 no. of parameters 7 var(lncons) 0.00162365

t-prob Part.R^2 0.768 0.0115 0.706 0.0188 0.107 0.2917 0.060 0.3735 0.005 0.6417 0.183 0.2101 0.012 0.5634

13

HELSINGIN YLIOPISTO - UNIVERSITY OF HELSINKI METSÄEKONOMIAN LAITOS - DEPARTMENT OF FOREST ECONOMICS TUTKIMUSRAPORTTEJA - REPORTS No: 1.

Miikka Pesonen. 1993. Japanese Market for Scandinavian Wood Products. Helsinki. 135 p. ISBN 951-45-6471-5

No: 2.

Heikki Juslin and Miikka Pesonen. 1994. Forest Products Marketing, The Current State of Research and Education. Helsinki. 65 p. ISBN 951-45-6662-9

No: 3.

Pasi Martikainen. 1994. Srategic Marketing Planning in the Finnish Forest Industries A Theoretical Marketing Planning Model and Its Empirical Testing. Helsinki. 176 p. ISBN 951-45-6661-0

No: 4.

Eija Pouta and Mika Rekola. 1994. Valuing Environmental Effects of Forest Regeneration Cuttings: A Theoretical Framework. Helsinki. 19 p. ISBN 951-45-6899-0

No: 5.

Jari Kuuluvainen and Mika Rekola. 1995. Eino Saari Commemoration seminar the History and Development of Economic Research in Multiple Use Forestry Helsinki. 32 p. ISBN 951-45-6979-2

No: 6.

Timo Hartikainen ja Kaisa Pirkola. 1996. Katsaus aihioiden käyttöön Tanskassa, Saksassa ja Alankomaissa. Helsinki. 63 s. ISBN 951-45-7350-1

No: 7.

Ritva Toivonen. 1996. Raakapuumarkkinoiden ja puun hintaraportoinnin vertailu. Comparative study about roundwood markets.and price reporting in selected countries. Helsinki. 45 p. ISBN 951-45-7477-X

No: 8.

Ritva Toivonen. 1996. Raakapuumarkkinat ja markkinainformaatio eri maissa. Helsinki. 62 s. ISBN 951-457478-8

No: 9.

Zoltan A. Kosy. 1996. Higher Education in Forestry and Forest Economics in Baltic and Nordic States. Helsinki. 42 p. ISBN 95145-7527-X

No: 10.

Mika Aho .1996. Hankintatyön verovapauden lisääminen metsäpolitiikan keinona. Helsinki. 64 s. ISBN 951-45-7545-8

No: 11.

Anne Toppinen. 1997. Incorporating Cointegration Relations in a Short-Run Model of the Finnish Sawlog Market. Helsinki. 26 p. ISBN 951-45-7918-6

No: 12.

Eija Pouta, Mika Rekola, Jari Kuuluvainen Chuan-Zhong Li and Olli Tahvonen. 1998. Referendum Model of Contingent Valuation and the Finnish Natura 2000 Nature Conservation Program: Preliminary analysis. Helsinki. 29 p. ISBN 951-45-8040-0

14

No: 13.

Kari Hyytiäinen, Olli Tahvonen. 2000. Legal Limits and Recommendations in Timber Production: On the Political Economy of the Forest Rotation Period. Helsinki. 35 p. ISBN 951-45-92610

No: 14.

Jussi Uusivuori, Jari Kuuluvainen . 2000. Substitution in Global Wood Imports in the 1990s. Helsinki. 26 p. ISBN 951-45-9438-X

No: 15.

Harri Häkkinen, Mikko Tervo. 2000. Raakapuun ostotoiminta Suomessa. Helsinki. 61 s. ISBN 951-45-9716-8

No: 16.

Jussi Uusivuori, Jari Kuuluvainen and Peter Blandon. 2001. Substitution in Japanese Wood Imports. Helsinki. 25 p. ISBN 952-10-0075-9

No: 17.

Jaana Rekikoski, Jari Kuuluvainen and Anne Toppinen. 2001. Stumpage and delivery trade in the Finnish pulpwood market. Helsinki. 32 p. ISBN 952-10-0076-7

No: 18.

Chuang-Zhong Li, Jari Kuuluvainen, Eija Pouta, Mika Rekola, Olli Tahvonen. 2001. Using choice experiments to value Natura 2000 nature conservation program in Finland. Helsinki. 18p. ISBN 95210-0192-5

No: 19.

Mika Rekola, Eija Pouta. 2001. Risk perceptions and contingent valuations – conceptual issues and a measurement experiment. Helsinki. 28 p. ISBN 952-10-0274-3

No: 20.

Mika Rekola. 2001. Lexicographic Preferences in Contingent Valuation: A Theoretical Framework with Illustrations. Helsinki. 26 p. ISBN 952-10-0273-5

No:21.

Pouta, E. 2002. Sensitivity to scope in contingent valuation of forest cutting practices. p. 28. ISBN 952-10-0523-8

No:22.

Jari Kuuluvainen, Emmi Lehtonen, Eija Pouta, Mika Rekola and Chuang-Zhong Li. 2002. Etelä-Suomen ja Pohjanmaan metsien suojelun hyötyjen taloudellinen arvottaminen. Alustavia tuloksia. 43p. ISBN 952-10-0588-8

No:23.

Mika Rekola. 2003. Lexicographic Preferences Challenge Contingent Valuation Method: What are They and How to Measure Them? In Print.

No:24.

Tapio Rantala and Eeva Primmer. 2002. Value Positions Based on Forest Policy Stakeholders’ Rhetoric in Finland. Helsinki. 23 p. ISBN 952-10-0873-3

No:25.

Jussi Uusivuori and Jari Kuuluvainen. 2003. The Harvesting Decisions when a Standing Forest with Multiple Age-Classes has Value. Helsinki. 38 p. ISBN 952-10-1246-3

15

No:26

Jari Kuuluvainen, Ibrahim Kaikai and Jussi Uusivuori. 2003. Empirical Behaviour Models on Timber Supply. Helsinki. 31 p. ISBN 952-10-1563-2

No:27

Tapio Rantala. 2003. Conceptions of Democracy of Key Informal Interest Groups in Finnish Forest Policy. 21 p. ISBN 952-101717-1.

No:28

Tapio Rantala. 2003. Metsäpolitiikan aatteelliset muutokset 1990luvulla - vanhaa, uutta ja lainattua. 32 s. ISBN 952-10-1718-X

No:29

Sami Berghäll. 2004. Relationships in the Finnish Forest Products Industry - Structure of the Perceptions of the Finnish Nonindustrial Private Forest Owner regarding Industrial Buyers of Roundwood. Helsinki. 18p. ISBN 952-10-1713-9

No:30

Sami Berghäll. 2004. Commitment - Towards a Theory of Marketing Exchanges. 25 p. ISBN 952-10-1712-0

No:31

Sami Berghäll and Martti Makkonen. 2004. A Comparison of the Perceptions of dyadic business relationships in forest and telecommunications industries. 11 p. ISBN 952-10-1715-5

No:32

Sami Berghäll. 2004. A comparison between relationship marketing and social psychological concepts of commitment. 21 p. ISBN 952-10-171-7

No:33

Ibrahim M. Kaikai, Jari Kuuluvainen and Jussi Uusivuori. 2004. Optimal Timber Stock in Finnish Nonindustrial Private Forests. 30 p. ISBN 952-10-1948-4.

No:34

Jari Kuuluvainen. 2004. Structural Change in US Newsprint Demand: GDP and Price Elasticities. 19 p. ISBN 952-10-2217-5