Example 2
Historical trends in global forest area (chapter 10)

Back to front page                                                                                                                  To list of six examples

   Summary. In order to deny that global forest area is steadily shrinking, Lomborg uses a dataset for the period 1950 to 1994 which is incomplete and therefore gives total figures that are too low, especially for the years up to 1965. He does this by taking totals from a set of FAO yearbooks, knowing these figures to be obsolete and ignoring corrections made to them in the intervening years by the FAO itself. In the later - English - edition of the book he studiously ignores new, robust FAO data that unequivocally indicate a steady decline in global forest area.


     Ever since Lomborg´s appearance on the public scene in Denmark in January 1998, one of his main attention-grabbers has been the provocative statement that the global forest area is not declining, as people believe, but that it is constant or slightly increasing.
     The basis for this statement is the data from the FAO production yearbook. This yearbook, which has been edited once a year from 1948 onwards, contains a table describing how the total land area in the world is partitioned between various land uses, such as arable land and permanent pastures. One of the categories is called "Forests and woodlands". Lomborg has taken the figure for "Forests and woodlands" from every volume of the yearbook and used these global totals to produce the graph shown in Figure 60 (page 111). This graph indicates that the global forest area has fluctuated up and down, but with the final values slightly higher than the initial values (The time series ends in 1994).
     Is it true that the forest area has increased slightly ? And if not, what is the trick ?
     Let us repeat what Lomborg has done, and take the figure for "Forests and woodland" from every volume of the FAO yearbook. For the sake of simplicity, let us discard the first aberrant years, and stop the series after 15 years. We then have the following list of figures:



year
Area of  forests and woodlands (bn ha´s)
Rise
1950
4.007

1951
3.869

1952
3.843

1953
3.947
^
1954
3.949

1955
3.864

1956
3.839

1957
3.985
^
1958
3.987

1959
4.069
^
1960
4.046

1961
4.099
^
1962
4.065

1963
4.070

1964
4.105
^
1965
4.015


To the right of the list, I have marked those years where the forest area shows a substantial increase. Let us look closely at these increases. How have they come about ?
     There is an increase in 1957. This is due exclusively to an increase in the forest area of the Soviet Union. This does not of course mean that the forest area here has suddenly grown by 0.14 bn hectares from one year to the next but that the previous data from USSR were of poor quality, and that better data were obtained that year.
    There is an increase in 1959. Most of this stems from a sudden increase in the forest area of Canada. Of course, this does not imply a sudden forest growth from one year to the next, but a revision of the previous figures for Canada. In addition, however, a large increase is due to the inclusion of Mongolia, which has some mountain forests. Up to then, there were no data for Mongolia, so this increase is due to the inclusion of an extra country.
     The next increase was in 1961. This is due to the inclusion of Papua New Guinea, which is almost completely covered by rainforest. This was also the first year that data for this country were included, so once again, the apparent increase was due to the inclusion of an extra country.
      The increase in 1964 is largely due to an amended figure for Chad. Up to then, the forest area in Chad was given as 3m hectares, but from 1964 this was changed to 16,54m hectares. The forest area in Chad had not, of course, suddenly increased by a factor of 5; and indeed, there is little forest in Chad. What happened was that the definition of "woodland", which includes  savanna and bush steppe, was more strictly applied than before. So this "forest increase" is not an increase in forest, but in scrub, and is moreoever an increase on paper, not in reality. This, of course, means that all previous years have a misleading figure for Chad, and in principle all previously given figures should be adjusted upwards accordingly.
     As the number of such cases increased year by year, the FAO staff realised that the older figures were too flawed to be of any use. From 1971 onwards the FAO therefore not only gave recent figures, but also revised the old ones. For example, in 1972 the figure for Angola in 1953 was given as 43.2m hectares, but in 1973 the figure for Angola in 1953 was revised to 72.66m hectares. Clearly, such delayed revisions make the older figures untrustworthy.
     For the period from 1961 onwards, revised figures are presented in later reports. If we use these revised figures, then we find that the global area of forests and woodland typically declines by 0.2 % per year - throughout the whole time span studied.
     In conclusion, the data series from the FAO production Yearbook confirms that the global wooded area declined steadily by about 0.2 % per year, which is approximately the same rate as during the period 1980 - 2000 where data of much higher quality are available.
     From this it seems that Lomborg has made a mistake. He has not been sufficiently thorough and studied the data in detail.
     But no - the point is that Lomborg knows all this perfectly well. He has actually studied the data in detail. That is what he tells us in his very long note 770, in the right hand column of p. 375. He notes that global figures for the same year may vary by up to 2 %, and he notes that the countries included are not the same in all cases. So he knows that the figures have been revised in later editions, and that one should therefore only compare figures that have been subject to the same revisions - to avoid, for example, a widely different contribution from Angola in two consecutive years. He knows that. But he obviously hopes that we do not - and that we will not draw the relevant conclusions from his own Note 770. As long as that is the case, he can postulate that the data demonstrate a rising trend, and (as he constantly admonishes us) you have to respect what the data tell.
     He also neglects what is written in the introduction to the table in every volume of the FAO production Yearbook:
"It should be borne in mind that definitons used by reporting countries vary considerably and items classified under the same category often relate to greatly differing kinds of land. . . Thus the area specified is not intended to refer to or delineate "forest coverage." So, he knows that the unrevised data are misleading, and that the data are not intended to delineate forest coverage anyhow. But what does he write ?
     He writes: "Figure 60 contains the best information on the global forest area." After that, he writes negatively about the quality of the data of all the other time series, which without exception show a steadily declining trend. But the data series which has the poorest quality - and which is not intended for the use that Lomborg makes of it - that data series is not criticized in the main text. The criticism of this data series occurs only in the notes. And even here, after presenting this criticism, he immediately continues: "Using short time-span series actually risks losing the general tendency in noise created by the individual adjustments. It has therefore been important to employ the longest time-span series available, and FAO´s long series from 1950 is the only one available. Unfortunately, the FAO database only provides access to figures from 1961 onward.  . . ". So, the reason for Lomborg to prefer the data from the FAO production Yearbook is that they go right back to 1950, although he knows that FAO´s data for the preriod 1950-1961 are not reliable. It is of course not warranted to use the longest time series if you know that its data are systematically too low (for example because Papua New Guinea is only included in figures after 1961).
     All the text referred to up to now is nearly identical to the text in the Danish edition of the book from 1998. Between 1998 and 2001, however, FAO published - for the first time - a comprehensive survey of the total forest area of the world, using identical definitions of "forest" for all nations. In addition, this new report was based on much more reliable data than the production yearbook - such as satellite images. This new, authoritative report demonstrated unequivocally that the global forest area is steadily declining.
     Lomborg read this new report, which clearly disproved the text he had previously written. So what did he do ? Did he revise his text ? No, he kept his text unchanged, evidently because if he had revised it, he would have missed one of his main ponts. Thus any lingering doubt as to whether the misleading text in the Danish version was intentional was dispelled: in the English version it was clearly deliberate. In short, the section on forest areas is so grossly manipulated that it may properly be described as fraud.