Abstract

 

            Recent changes in wine bottle closures have put pressure on the cork industry, and the regions where cork is produced. Using time-series analysis, this paper examines the possible economic impact a collapse of the cork industry could have on one of the major cork producing nations, Portugal.

 

Introduction

 

            According to the Oxford English Dictionary, the use of the word ‘cork’  to refer to a piece of cork “cut into a cylindrical or tapering form, used as a stopper for a bottle, cask, etc” dates back to the 15th century.[1] The noun and the verb both convey, in modern usage, the assumption that the material used is ‘cork’ – the bark of the cork oak (quercus suber). However, changes in the wine industry over the past decade or so[2] have begun to negate this assumption.

            Nearly as old as the noun ‘cork’ is the adjective ‘corked,’ which refers to wine which was stopped with a bad cork.[3] This ‘cork taint’ gives wine “a moldy, musty, and/or earthy aroma that can mask the natural wine aroma and lessen its quality.”[4] Although the matter of quantity is very much in dispute (estimates vary between less than 1% and 8% of all wine is ‘corked’),[5] there is no doubt money is lost when wine goes off. Some would put the value at over $10 billion per year.[6]

            This loss has recently prompted the wine industry to revisit the question of what should be used to close their bottles. A number of producers have determined that cork is no longer the best option, and have converted to artificial stoppers. According to Food Engineering magazine, “about 8 percent of wines worldwide now use synthetic stoppers and screw caps.”[7]

            The conversion of the wine industry to non-cork stoppers could have a dramatic effect on the economies of cork-producing nations. This is due primarily to the age of the industry, and its integration with local economies. For example, in the regions where cork oaks are grown, the management of the forests has led to “large single-species stands”[8] of trees. Currently cork is produced in an environmentally friendly process using renewable resources.[9] It is generally accepted that if the cork production system were to fail, the stands of cork oak would be replaced not by a natural ecosystem but by another monoculture plantations system.[10] Many sources referred to eucalyptus, [11] which is a cash crop used for paper, and is one of the few crops besides cork oaks able to live in the region. Unfortunately, unlike the cork oak eucalyptus acts more like a weed than a friendly member of the ecosystem. Eucalyptus “acts like a sponge” and takes all the water out of the ground, “leaving the ground between the trees barren.” This is in contrast to the cork oak, which has allowed farmers to raise animals which graze on the undergrowth.[12] This destruction of habitat would also likely prove fatal to some of the 100 species of mammal and several hundred species of bird which inhabit this region. The most well known of these species is the Iberian Lynx, whose total population is estimated at under 500 individuals.[13] According to the World Wildlife Federation, the lynx is the “rarest cat in the world.”[14]

 

Data Availability

 

            Although the author is aware of the environmental issues associated with the cork industry in Portugal, unfortunately the body of literature and research needed to do a more robust economic evaluation of the cork industry, taking into account environmental valuations, is simply not available. In point of fact, data, in general, is not available on the cork oak. For example, the cork oak has been said to occupy 2 million hectares of land (in a citation from 1950 and 1985), but according to a 1999 paper, “real area is not known, since up-to-date inventories are not available, particularly with regard to age distribution and the density of stands.”[15] One of the few people doing work in this area, José Guilherme Borges, noted “forest statistics are frequently not available in a readily form or consistent. Data is dispersed among several institutions and often it is not either published or digitized.”[16] In addition, he noted “there is no experience in developing econometrics based forest sector models.”

 

Methodology

 

Due to the noted data issues, this paper will stick to a more traditional approach (with some creative touches to fill in gaps), using linear time series analysis to investigate the potential losses to GDP were the cork industry in Portugal to go into decline.

            One primary source of data was the International Monetary Fund, which maintains a database of most macroeconomic indicators, including GDP, interest rates, population, and money supply.[17] Information on cork came from the Department of Forest Resources at the Technical University of Lisbon.[18] Cork data was only available in yearly series, however, due to cork’s close association with exports I chose to construct quarterly data using total exports to estimate quarterly changes in cork production. Clearly this is not optimal, however, given the dearth of data sources it will need to serve.

            Basic macroeconomic theory tells us GDP can be modeled using government spending, consumer spending, investments, and exports. Slightly more advanced theory says past GDP is a factor. In addition, it was my opinion that the exchange rate might have an influence on Portugal’s economy. As such, I used an exchange rate index variable in addition to the government discount rate, government consumption spending, total exports, and cork exports. Also included were CPI and money supply (M2).

            A quick look at descriptive statistics told me these variables as they were could not be used to model as they were, due to high correlations (Appendix 1). In addition, of course, the log-log form is generally considered a more appropriate form for this sort of modeling. Finally, stationarity is required for time-series data, and as such we will take the first difference of our data. Thus our final model will be

 

Ln(YT)=ln(GT-1)+ ln(GDP T-1)+ ln(CPI T-1)+ ln(M2 T-1)+DisRate T-1

+ ln(Exports T-1)+ ln(e T-1)+ln(corkval T-1)

 

In addition, however, the number of lags needs to be determined. Due to the constraints of the number of observations, four or more lags was not possible. Based on the AIC and Schwartz criteria for determining lag length I therefore decided to go with three lags. On review of the regression, unfortunately, I determined the data was not precise enough for the simulation of quarterly data. As such I returned to yearly data for estimates.

 

Findings

 

Conclusions


APPENDIX 1 – Tables

 

 

Correlation Coefficients

5% critical value (two-tailed) = 0.3120 for n = 40

1) g

2) GDP

3) CPI

4) M2

5) DisRate

6) Exports

7) e

8) corkVal

 

1

0.9977

0.9821

0.9952

-0.7077

0.9412

-0.5731

0.8835

1) g

 

1

0.9739

0.9949

-0.7254

0.9558

-0.5696

0.9025

2) GDP

 

 

1

0.9785

-0.6598

0.8872

-0.5099

0.8285

3) CPI

 

 

 

1

-0.7227

0.9405

-0.5851

0.8921

4) M2

 

 

 

 

1

-0.7915

0.7227

-0.7433

5) DisRate

 

 

 

 

 

1

-0.5994

0.9822

6) Exports

 

 

 

 

 

 

1

-0.554

7) e

 

 

 

 

 

 

 

1

8) corkVal

 

 

AIC and Schwartz for:

Lag 3

AIC       0.000385183     SCHWARZ    0.00119684

Lag 2:

AIC       0.000351757     SCHWARZ   0.000754549

Lag 1:

AIC       0.000300022     SCHWARZ   0.000447556

 

Lag 3 model with constructed cork data:

 

Model 4: OLS estimates using the 33 observations 1990:1-1998:1

Dependent variable: d_l_GDP

 

      VARIABLE      COEFFICIENT        STDERROR       T STAT   2Prob(t > |T|)

 

   0)    const         0.0139926          0.0161906     0.864    0.412617

  24) d_DisR_1         0.000112438        0.00244001    0.046    0.964375

  25) d_DisR_2        -0.00108561         0.00240827   -0.451    0.664113

  26) d_DisR_3         0.000961631        0.00210937    0.456    0.660593

  28)  d_l_g_1         0.121319           0.750485      0.162    0.875587

  29)  d_l_g_2        -0.0498996          0.998449     -0.050    0.961366

  30)  d_l_g_3         0.470542           0.963061      0.489    0.638243

  32) d_l_GD_1        -0.0413357          0.382972     -0.108    0.916706

  33) d_l_GD_2        -0.389064           0.434225     -0.896    0.396415

  34) d_l_GD_3        -0.101830           0.410879     -0.248    0.810505

  36) d_l_CP_1         0.697740           1.36173       0.512    0.622219

  37) d_l_CP_2        -0.166324           0.809619     -0.205    0.842363

  38) d_l_CP_3        -0.0733959          1.10575      -0.066    0.948707

  40) d_l_M2_1        -0.0725857          0.308994     -0.235    0.820180

  41) d_l_M2_2         0.204298           0.309666      0.660    0.527956

  42) d_l_M2_3        -0.286772           0.272130     -1.054    0.322758

  44) d_l_Ex_1         0.0313895          0.229210      0.137    0.894457

  45) d_l_Ex_2         0.121347           0.184221      0.659    0.528584

  46) d_l_Ex_3         0.220445           0.333371      0.661    0.527026

  48)  d_l_e_1         0.00564726         0.223632      0.025    0.980472

  49)  d_l_e_2         0.00354599         0.163303      0.022    0.983208

  50)  d_l_e_3        -0.0658750          0.238390     -0.276    0.789297

  52) d_l_co_1        -0.0433887          0.207379     -0.209    0.839503

  53) d_l_co_2        -0.0562080          0.223582     -0.251    0.807844

  54) d_l_co_3        -0.196025           0.327665     -0.598    0.566223

 

  Mean of dependent variable = 0.0238988

  Standard deviation of dep. var. = 0.0162288

  Sum of squared residuals = 0.00279357

  Standard error of residuals = 0.0186868

  Unadjusted R-squared = 0.668535

  Adjusted R-squared = -0.325858

  F-statistic (24, 8) = 0.672304 (p-value = 0.786)

  Durbin-Watson statistic = 2.32533

  First-order autocorrelation coeff. = -0.173583

 

  MODEL SELECTION STATISTICS

 

  SGMASQ    0.000349196     AIC       0.000385183     FPE       0.000613738

  HQ        0.000564071     SCHWARZ    0.00119684     SHIBATA   0.000212917

  GCV        0.00144043     RICE        undefined

 

Excluding the constant, p-value was highest for variable 49 (d_l_e_2)

 

 


Bibliography

 

Álvarez-Rodríguez et al. “Cork Taint of Wines: Role of the Filamentous Fungi Isolated from Cork in the Formation of 2,4,6-Trichloroanisole by O Methylation of 2,4,6-Trichlorophenol.” Applied and Environmental Microbiology, December 2002. Vol. 68, No. 12, p. 5860-5869.

 

Anonymous, “Moreover: A Corking Row” The Economist; London; Jun 5, 1999. Vol. 351, No. 8122. P. 84

 

Borges, José Guilherme. “Analysis of the Markets for Roundwood and Forest Industry Products in Portugal.” Department of Forest Resources. Technical University of Lisbon

 

Bruce-Gardyne, Tom. “Stop your wine-ing” The Ecologist; Nov 2000. Vol. 30, No. 8, P. 57

 

Gonçalves, Eduardo. “The Algarve Tiger.” The Ecologist, Feb. 2002. Vol. 32 No. 1 p. 52-5

 

Higgins, Kevin T. “Can quality control save cork closures?” Food Engineering, Nov. 2002. Vol. 74, No. 11, P. 16.

 

International Financial Statistics - http://ifs.apdi.net/imf/

 

Oxford English Dictionary, 2d edition. Oxford University Press, 2003. OED Online

http://www.oed.com

 

Grupo de Economia e Gestão em Recursos Naturais. “Portuguese Forestry. Some Statistics.”

http://floresta.isa.utl.pt/gegref/english/statistics.htm

 

United Nations Joint Staff Pension Fund, “Pension Info – July 2001”

http://www.unjspf.org/pdf/pensioninfo-jul-eng.pdf

 

World Wildlife Federation. “On the ground in Doñana in southern Spain”

http://www.panda.org/about_wwf/where_we_work/europe/where/spain/cota_donana/iberian_lynx.cfm

 

 


”Plants vary with respect to their moisture requirements.

Xerophytes of which all drought-resistent plants and m particularly, the cork oak are typical. Highly resistant to drought conditions.[32]

 

It is a remarkable fact that 84.5% of the world’s cork is growing in exactly the kind of soil that some writers have recently advised the potential planter to avoid…[34]

 

Some Geographic and Economic Aspects of the Corkl Oak

 

Victor A. Ryan

 

Baltimore MD 1948.Crown Cork and Seal Company.



[1] Oxford English Dictionary

[2] According to The Ecologist, the first reliable plastic cork was invented in 1992. Metal closures have been around for some time.

[3] Oxford English Dictionary

[4] Álvarez-Rodríguez et al. Introduction.

[5] Anonymous, “Moreover: A Corking Row.”

[6] Álvarez-Rodríguez et al. Introduction.

[7] Higgins, Kevin T. “Can quality control save cork closures?”

[8] M.C. Varela. “Cork and the cork oak system

[9] Ibid. It is worth noting, however, that some documents referred to post-production treatments which were not as environmentally friendly.

[10] Ibid.

[11] Grupo de Economia e Gestão em Recursos Naturais. “Portuguese Forestry. Some Statistics.” – many sources made reference to eucalyptus, this group actually did some estimation of the values involved.

[12] Higgins, Kevin T. “Can quality control save cork closures?”

[13] Gonçalves, Eduardo. “The Algarve Tiger”

[14] World Wildlife Federation

[15] M.C. Varela. “Cork and the cork oak system” – interestingly, this data is not backed up by Borges, who cites 1994 data from Pereira and Martins stating total cork oak coverage is 700,000 ha.

[16] Borges, José Guilherme. “Analysis of the Markets for Roundwood and Forest Industry Products in Portugal.”

[17] International Financial Statistics - http://ifs.apdi.net/imf/

[18] Borges, José Guilherme. “Analysis of the Markets for Roundwood and Forest Industry Products in Portugal.”