On the Reliability of International Forest Sector Statistics: Problems and Needs for Improvements

Article On the Reliability of International Forest Sector Statistics: Problems and Needs for Improvements A. Maarit I. Kallio 1,*ID and Birger Solberg 2 1Bioeconomy and Environment, Natural Resources Institute Finland, Latokartanonkaari 9, 00790 Helsinki, Finland 2Faculty of Environmental Sciences and Natural Resource Management, Norwegian University of Life Sciences, Universitetstunet 3, 1433 Ås, Norway; birger[email protected] *Correspondence: [email protected]; Tel.: +358-295-325-434 Received: 6 June 2018; Accepted: 3 July 2018; Published: 5 July 2018   Abstract: Statistics on production and trade of forest products form a basis for the analyses of the economic and environmental performance of the forest sector and for projecting future developments in the sector and related markets. The forest product statistics FAOSTAT by Food and Agriculture Organization of the United Nations (FAO) are perhaps the most widely used data source for such analyses. This study aimed at systematically scrutinizing this important dataset and identifying regions and product categories where improved data on wood and forest industry production and trade would be needed in order to set the modeling and analyses employing these data on a more solid basis. The consistency of the data on production, imports, and export volumes of wood was examined with some simple tests and with the aid of linear programming. The results show that the data related to the supply of wood chips and particles are unreliable in many countries. Improving these data would be important in order to assess the cascading wood input–output flows, to evaluate resource potentials, and to specify more reliable coefficients for wood use in forest sector models. Moreover, the data show remarkable inconsistencies, even of the magnitude of millions of cubic meters, between the apparent supply of wood (harvests + net imports) and forest industry production in many regions. Errors and uncertainties of such magnitude have important consequences on the results of any analysis using the data and call for special attention by the data users. Keywords: data; errors; forest products; forestry; statistics; FAOSTAT; modeling 1. Introduction The forest product statistics FAOSTAT of Food and Agriculture Organization of the United Nations (FAO) on the production and trade of wood and wood products [ 1 ] provide an important and valuable basis for the economic and environmental analyses and projections of the developments in the forest sector. Consequently, these statistics are employed for various sectoral analyses using partial economic equilibrium models, such as the global forest product model (GFPM [ 2 ]), the global forest sector model EFI-GTM (e.g., [ 3 ]), the global biosphere management model (GLOBIOM [ 4 ]), and many others. Typically, the FAOSTAT data are used for defining wood harvest volumes and apparent consumption levels for the forest industry products for the first period modeled (base year) in such models. Depending on the model, the FAOSTAT data may also be employed, for instance, to define inertia constraints for the trade in forest products, to define the base year production capacities for the forest industry, and to calibrate the wood-input coefficients and production costs for the forest industry. Furthermore, many other prognoses and analyses ([ 5 , 6 ]) concerning consumption and the markets for forest products are based on these data. Among the various other research purposes, the data have been used to support studies on carbon dynamics in wood products [ 7 ], the water footprint of forest Forests 2018,9, 407; doi:10.3390/f9070407 www.mdpi.com/journal/forests Forests 2018,9, 407 2 of 17 products [ 8 ], the forest footprint of the European Union [ 9 ], and the role of trade in deforestation [ 10 ], to name but a few examples of the recent studies. The FAOSTAT data could also be useful for the Intergovernmental Panel on Climate Change (IPCC) for tracking the production, trade, and apparent consumption of harvested wood products and roundwood [ 11 ] and for the development of climate change mitigation policies (e.g., [11,12]). The reliability of these data is thus of crucial importance. This study aims at systematically scrutinizing them and identifying the most important product groups and geographical areas where improvements would be required to set the modeling, analyses, and forecasting of the forest sector on more solid ground. The data on production, imports, and export quantities are explored first with some simple tests and then with the aid of linear programming (LP). In the latter case, it is tested if it is possible to find reasonable wood input coefficients for various industrial products so that the volume of wood biomass available according to the statistics has been adequate to supply the raw material needs for the reported production quantities in the forest industries. The tests identify considerable inconsistencies in the data for several countries. 2. Material and Methods 2.1. Data 2.1.1. Data on Production and Trade The main data source and also the subject for our analysis was the FAOSTAT forestry database. In addition, the United Nations Economic Commission for Europe (UNECE)-FAO production statistics [ 13 ] were used to gain information on the division of plywood production to softwood and hardwood grades in the countries within the domain of those statistics. The data in FAO pulp and paper production capacity surveys from the year 2000 onwards [ 14 ] were used as an aid to divide sulfate pulp production to softwood and hardwood grades, though these data do not seem to cover all countries. For pulp production, data from RISI [ 15 ] were also available, but because this dataset is not available for free, its use was limited here to inform if a country only produced hardwood or softwood sulfate pulp. Table 1presents the products considered and the sources for the data on production and trade that were already available or estimated within this study using linear programming. In the FAOSTAT data, the category “wood chips and particles” includes wood reduced to small pieces, which issuitable for pulping, particle board or fiberboard production, and wood for use as a fuel or other purposes. It excludes wood chips made directly in the forest from roundwood. The category “wood residues” includes wood processing co-products other than those belonging to wood chips and particles, such as sawmill rejects, slabs, edgings and trimmings, veneer log cores, veneer rejects, sawdust, residues from carpentry and joinery production, and wood residues that will be used for production of pellets and other agglomerated products. Table 1. Annual production and trade data used. D = Data given in FAO forest product statistics (FAOSTAT [ 1 ]), U = data given in UNECE/FAO [ 13 ] for Europe and North America, e = data estimated based on the other data using linear programming. C = coniferous, NC = non-coniferous, Sa = sulfate, Si = sulfite, Bl = bleached, Unbl = unbleached. UNECE/FAO, the United Nations Economic Commission for Europe-The Food and Agriculture Organization of the United Nations. Product Product Production Imports Exports Industrial roundwood C D D D Sawlogs and veneer logs C D e e Pulpwood C D e e Other industrial roundwood C D e e Industrial roundwood NC D D D Forests 2018,9, 407 3 of 17 Table 1. Cont. Product Product Production Imports Exports Industrial roundwood NC, non-tropical D D Industrial roundwood NC, tropical D D Sawlogs and veneer logs NC D e e Pulpwood NC D e e Other industrial roundwood NC D e e Wood residues D D D Wood residues C e Wood residues NC e Chips and particles D D D Chips and particles C e e e Chips and particles NC e e e Pellets D (since 2012) Chemical pulp D Dissolving pulp D Chemical wood pulp Sa Unbl D Chemical wood pulp Sa Bl D -Bleached hardwood kraft e -Bleached softwood kraft e Chemical wood pulp Si Unbl D Chemical wood pulp Si Bl D Semichemical pulp D Mechanical pulp D Sawnwood C D Sawnwood NC D Plywood and veneer D Plywood and veneer C e, U Plywood and veneer NC e, U Particleboard D OSB (Oriented strand board) D (since 1995) MDF (Medium-density fibreboard)/ HDF (High-density fibreboard) D Hardboard D Fiberboard D 2.1.2. Data on Conversion Factors Used in LP Test Here, the term conversion factor refers to an input quantity of roundwood, wood residues, or chips and particles in solid cubic meters (m 3 ) under bark needed to produce one cubic meter of mechanical forest industry product or 1 tonne of pulp. The assumptions made regarding the possible uses of different wood categories within the forest industry are given in Table 2. The main references for the conversion factors were the UNECE and the European Forest Sector Outlook Study 2005 [ 16 – 18 ]. Table 2summarizes the conversion factors, while any available country specific conversion factors used are provided in the Supplementary Materials (Supplementary File S1: Data for reference input coefficients (excel-file)). Forests 2018,9, 407 4 of 17 Table 2. Wood categories suitable as inputs in forest industry products and the assumed typical (Ref), low (Ref low), and high (Ref high) wood input quantities in the linear programming (LP) test when country-specific reference quantities were not available. Min and Max in the parentheses are additional broader minimum and maximum input quantities applied for all regions, respectively. Slogs = sawlogs and veneer logs, Pwd = pulpwood, Res = wood residues, Chips = chips and particles, C = coniferous, and NC = non-coniferous. Sources: a = The smallest value given in UNECE [ 18 ] (after Spain and Portugal with 1.2 m 3 /m 3 for veneer sheets removed from the sample); b = Average of all values in [18]; and c = the highest value in [18]. If data were not available from [18] for (a), (b), or (c), data from [16] were used. Forest Industry Product Ref Low (Min), m3of Wood per Unit of Output Ref Typical, m3of Wood per Unit of Output Ref High (Max), m3of Wood per Unit of Output Wood Categories Suitable as Inputs Chemical wood pulp Si Un bl. 4.6 (4.0) 4.5 6.0 (6.0) Pwd (C, NC), Chips (C, NC) Chemical wood pulp Sa Un bl. 3.5 (3.2) 4.3 5.0 (5.5) Pwd (C, NC), Chips (C, NC) Chemical wood pulp Si Bl. 4.6 (4.0) 5.0 6.5 (6.5) Pwd (C, NC), Chips (C, NC) Chemical wood pulp Sa Bl. 3.5 (2.98) 4.5 5.5 (6.5) Pwd (C, NC), Chips (C, NC) Bleached hardwood kraft 3.5 (2.98) 3.8 4.3 (5) Pwd (NC), Chips (NC) Bleached softwood kraft 4.5 (4.79) 4.9 5.3 (6.5) Pwd (C), Chips (C) Dissolving pulp 5.1 (5.0) 5.65 b 6.5 (6.8) Pwd (C, NC), Chips (C, NC) Semichemical pulp 2.5 (2.3 a) 2.83 b 3 (3.2 c) Pwd (C, NC), Chips (C, NC) Mechanical pulp 2.4 (2.3 a) 2.5 b 2.9 (2.9 c) Pwd (C), Chips (C) Sawnwood C 1.6 (1.42 a) 1.77 b 2.0 (2.22 c) Slogs (C) Sawnwood NC 1.5 (1.4 a) 1.76 b 2.1 (2.5 c) Slogs (NC) Plywood and veneer sheets 1.7 (1.67 a) 2.19 b 2.5 (3.1 c) Slogs (C), Slogs (NC) Plywood and veneer C 1.9 (1.67 a) 2.0 b 2.31 (2.9 c) Slogs (C) Plywood and veneer NC 2.0 (1.67 a) 2.29 b 2.68 (3.1 c) Slogs (NC) Particleboard 1.4 (1.2 a) 1.49 b 1.7 (1.93 c) Pwd (C, NC), Chips (C, NC), Res (C, NC) OSB 1.4 (1.30 a) 1.64 b 2.0 (2.0 c) Pwd (C, NC), Chips (C, NC), Res (C, NC) MDF/HDF 1.5 (1.45 a) 1.68 b 1.8 (1.93 c) Pwd (C, NC), Chips (C, NC), Res (C, NC) Hardboard 1.8 (1.75 a) 1.97 b 2.3 (2.4 c) Pwd (C, NC), Chips (C, NC), Res (C, NC) Fiberboard 1.7 (1.5 a) 1.98 b 2.8 (3.0 c) Pwd (C, NC), Chips (C, NC), Res (C, NC) Pellets 2.3 (2.2 a) 2.54 b 2.9 (3.0) Res (C, NC) Forests 2018,9, 407 5 of 17 2.2. Methods Used to Analyze the Data 2.2.1. Simple Tests with the Raw Data First, the regional data on the production and trade of industrial roundwood, wood residues, chips, and particles were examined with three relatively simple tests i–iii, described below under the questions (i)–(iii), respectively. (i) Is the reported wood production at least as high as the reported net export (exports −imports)? If not, the region’s apparent wood consumption is negative, which signals errors in the underlying data. The data allowed this first infeasibility check to be made for four wood categories: industrial coniferous roundwood, industrial non-coniferous roundwood, chips and particles, and wood residues. (ii) Is the reported production of chips, particles, and wood residues (too) high compared to the production of solid wood products? This test identifies countries where the produced number of woody by-products is considerably higher than the produced amount of the solid wood products. The test was carried out for the category of “wood chips and particles” alone and then for the total amount of by-products labeled “wood chips and particles” and “wood residues”. If the number of by-products is much higher than the production of the principal products, there is probably something wrong in the data, despite the fact that part of the residues comes from further processing of sawnwood, veneer, and plywood to finished products, carpentry, and so on. Typically, chips and particles are obtained as a side product of sawnwood and plywood production, and they form only part of the material obtained as by-products. Also, other types of residues, for instance sawdust, are coming out. Moreover, part of the log input in sawnwood and plywood production vanishes due to shrinking. The log input coefficient under bark in the production of solid wood products is seldom above 2 (Table 2), in particular, when it comes to sawnwood that is more commonly produced than plywood and veneer. For these reasons, it is reasonable to expect that production of chips and particles should not exceed that of solid wood products. If the output of “chips and particles” and solid wood products were equal, it would indicate that little other material would come out from the process. (iii) Are the apparent levels of wood use high enough for the reported production of forest industry products? This test was done for all sawlogs and veneer logs (coniferous + non-coniferous), for pulpwood and pulpwood substitutes (coniferous pulpwood + non-coniferous pulpwood + chips and particles), and for all wood. If the assumed required minimum consumption of wood inputs exceeds the amount of their maximum availability under the rather loose requirements discussed below, it is likely that there is something wrong with the data. •Case of sawlogs and veneer logs used for sawnwood, plywood, and veneer In this test, the assumed minimum quantity of logs needed in sawnwood, veneer, and plywood production in a country was compared to the maximum potential volume of sawlogs and veneer logs available based on the statistics on wood harvest and trade. The test setting was cautious for the following reasons. Firstly, it was assumed that 1.5 m 3 of logs under bark are sufficient to produce one unit of sawnwood, veneer, or plywood. This coefficient can be considered somewhat low (see Table 2 ). Secondly, the maximum available amount of sawlogs and veneer logs in a region was defined to include sawlog and veneer logs harvests and all industrial roundwood imports to the country, despite the fact that part of the imports can be pulpwood or roundwood for other industrial uses. Thirdly, all the exports of industrial roundwood were assumed to consist of the wood grades other than sawlogs and veneer logs. Forests 2018,9, 407 6 of 17 •Case of wood material used for pulp This test asked if there may have been enough pulpwood, chips, and particles to sustain the reported wood pulp production in a country. Again, the test was rather robust in the sense that the requirements for the data passing the test were set low. It was assumed that all pulpwood, chips, and particles potentially available in a region may have been used for pulp production, although these raw materials are also used for particle board, Medium-density fibreboard (MDF), and other panel products. Furthermore, all industrial roundwood imports were now assumed to be pulpwood, while all roundwood exports were assumed to be other wood grades. It was modestly assumed that chemical pulp requires only 3 m 3 , dissolving pulp 4 m 3 and mechanical and semi-chemical pulp 2 m 3 of wood fiber per tonne of pulp output. •Case of all wood materials used for all products This test added panels (particle board, Oriented strand board (OSB), MDF, etc.) to the previous set of products and asked whether the production of this new set of products could have been produced by the wood material reported to be available. a very modest (see Table 2) wood use coefficient was applied to these panels: 1 m 3 /m 3 . For the other products, the minimum input coefficients described above were used. “Wood residues” were included to the available wood together with sawlogs, pulpwood, chips, and particles. The amount of wood available per country was defined to include net imports (imports − exports) and own harvests or production of these wood grades. The test compared only the total wood material reported to be available in a region to the conservatively defined aggregated demand for wood materials in the reported forest industry productions. It imposed no allocation rules based on the suitability of alternative wood grades (Table 2) to different products. 2.2.2. Test Using Linear Programming An LP formulation was used to improve the consistency and precision of test iii for all wood grades and products and also to identify the cases where the regional wood supply seems too high compared to the respective forest industry production. Because the exact mathematical formulation of the LP problem is space demanding, only its main principles are described here, while a complete description, including all the constraints, is provided in the Supplementary File S2: Description of the LP model (word.doc). The main principle of the LP program is that it aims to find input coefficients for the industrial products in Table 1that are as close as possible to the reference values given in Table 2(or the more country-specific figures) and which, when applied to the reported forest industry production data, match the consumption of various wood grades to their apparent availability (production + imports − exports) as well as possible. The objective function minimizes the sum of the various penalty variables that allow the data to be inconsistent with the rules defined. Three levels of penalty variables for the wood input coefficients not complying with their reference values were defined: (1) a somewhat low penalty of deviation from the point estimate reference value; (2) a higher penalty for input coefficients not being within an assumed range around the reference value; and (3) the highest penalty for not even fitting in between the absolute minimum and maximum values defined for the coefficients. The default input coefficient values are given in Table 2but were refined as shown in the Supplementary Materials when there was additional country-specific information available. When looking for the input coefficients, constraints (subject to penalty variables if violated) that aim to minimize year-to-year fluctuations of the input coefficients were also defined so that the technologies were assumed to be material saving. a regional input coefficient for a product in a given year that was higher than that in the previous year was penalized if there was production in both years. Roundwood harvests were constrained to equal the harvests reported in the statistics subject to penalty variables. The calculated value for the penalty variable in a region indicated the suggested deviation from roundwood harvests from the statistics. Forests 2018,9, 407 7 of 17 The program also divided the traded quantities of roundwood given in the statistics between pulpwood, sawlogs, and other industrial wood. It also divided the production of sulfate pulp, plywood and veneer, and chips and particles into coniferous and non-coniferous grades. These divisions were done so that the data become as consistent as possible with the wood use in the forest industry. Before doing the LP test, the data showing negative apparent consumption figures for wood in test i described in Section 2.2.1 were modified as follows: •The production volumes of wood residues and chips were increased to match with the exports. • The production volumes of other coniferous or other non-coniferous industrial roundwood were increased if the apparent consumption of coniferous or non-coniferous roundwood was negative, respectively. Such typically small adjustments were rarely needed. Similarly, noting the inconsistencies raised by test ii, the following modifications for the data on by-product outputs were made before the LP test in order to improve their consistency with the production of the solid wood products (sawnwood, plywood, and veneer). • The production volumes for chips and particles were decreased to be at most 80% of the production of solid wood products. • The production of wood residues was reduced, if necessary, so that the resulting logs input calculated as “(sawnwood production + plywood and veneer production + chips and particles production + wood residues production)/(sawnwood production + plywood and veneer production)” is at most 2.8 m3/m3. If a country then violated test i, the exports were reduced. Test i in Section 2.2.1 mostly identified cases where the production of forest industry residues was too low compared to the exports, while test ii found cases where it was too high, generally. Consequently, further corrections made based on test ii seldom affected the same countries. 3. Results 3.1. Test i: Is the Apparent Wood Consumption Non-Negative? Table 3shows the data for countries that had an apparent consumption figure of more than 5000 m 3 negative in 2015 for the four wood categories tested. The dataset for the cases for such negative apparent consumption figures during a longer period 2000–2016 is given in the Supplementary File S3: Additional results from the tests (excel-file). The problem of data indicating negative consumption levels is mostly related to the category of “chips and particles”. a negative apparent consumption of industrial roundwood occurs only with a few countries during 2000–2016. During the years of that period, only two to four countries out of the 244 tested had an apparent consumption figure of less than − 5000 m 3 for coniferous industrial roundwood and at most two countries for non-coniferous roundwood. The figures of Costa Rica and Myanmar for coniferous roundwood were negative for most of this period. These countries report exports of coniferous wood but no production or imports. In 2015–2016, New Zealand reported negative apparent consumption for non-coniferous roundwood. For “wood residues”, the number of respective annual observations with negative apparent consumption figures varied from one to eight during 2000–2016, while for “chips and particles” sometimes up to 18 countries reported data that indicate negative apparent consumption. The problems of negative consumption figures for wood residues, chips, and particles are typically related to the issue that exports are reported, but data on production or imports are lacking. Some problems seem to relate to the lack of updates in the dataset. For instance, the production data for chips and particles for Indonesia and Fiji have obviously not been updated since 2011 (data status “FAO estimate”), while the respective data on exports have been revised annually (data status “FAO official”). The highest negative apparent consumption figure in 2015 for “chips and particles” is found for Vietnam ( − 10 Mm 3 ). The production of chips and particles in Vietnam has in the recent years reported (data status “FAO estimate”) to be constant at 3.3 Mm 3 , while the export volumes of over 12 Mm 3 have Forests 2018,9, 407 8 of 17 been reported (under data status “FAO official” or “FAO unofficial”). At the same time, the sawnwood and plywood production needed to give chips and particles as by-product has not been updated, and it has been reported to be below 8 Mm 3 (under data status “FAO official” or “FAO unofficial”). Considerable gaps in the net exports and production figures of chips and particles in 2015 are also given ( e.g., for Gambia , Indonesia, Mozambique, Singapore, South Africa, and Thailand) as can be seen in Table 3. Table 3. Annual productions, exports, imports, and apparent consumption of industrial roundwood and wood residues in 2015 and in the period of 2011–2015 on average (1000 m 3 ). Only countries where apparent consumption in 2015 was <−5000 m3are shown. C = coniferous, NC = non-coniferous. Country Production Imports Exports Consumption Production Imports Exports Consumption 2015 2015 2015 2015 2011–2015 2011–2015 2011–2015 2011–2015 Industrial roundwood C Costa Rica 1 15 −14 1 28 −27 Myanmar 0 11 −11 0 33 −33 Industrial roundwood NC New Zealand 23 3 38 −11 20 4 16 8 Chips and particles Albania 2 37 −35 2 0 32 −30 Bangladesh 0 20 −20 0 20 −20 Congo 0 36 −36 0 145 −144 Fiji 210 0 244 −34 210 0 230 −20 Gambia 0 59 −59 0 25 −25 Indonesia 1788 0 2491 −703 1788 5 2675 −882 Liberia 0 33 −33 0 101 −101 Libya 0 22 −22 0 22 −22 Montenegro 0 16 −16 0 18 −18 Mozambique 0 49 −49 0 10 −10 Papua New Guinea 0 6 −6 0 7 −7 Singapore 3 160 −157 3 70 −67 South Africa 1926 7 2319 −385 2125 4 2247 −119 Thailand 4239 52 4398 −107 2512 154 4712 −2047 Vietnam 3312 3 13,347 −10,032 3290 106 12,002 −8607 Wood residues India 2 12 −10 5 3 2 Laos 1 8 −8 0 7 −7 Mexico 3 21 −18 2 19 −17 Montenegro 0 21 −21 0 10 −9 Singapore 6 29 −23 5 12 −7 3.2. Test ii: Is the Production of Chips, Particles, and Wood Residues Consistent with the Production of Solidwood Products? Table 4shows the countries where the supply of forest industry by-products (chips, particles, and wood residues) is reported to exceed the production of solid wood products in 2015. It also gives the respective production figures and resulting average theoretical wood input coefficient per unit of solid wood products produced. For some countries, particularly for Australia, the test for “chips and particles” production, which requires that “production of chips and particles < production of plywood + sawnwood” (column C compared with column D in Table 4) suggests that this data might be wrong. While often the entries related to suspicious data on chips and particles production is rated ‘unofficial’, that is not the case with Australia. Also, for example, Italy and the Netherlands report very high outputs of chips and particles compared to the production of solid wood products. To justify the figures, the activities in carpentry and other further processing of wood that would provide chips and particles must be of considerable magnitude. The total output of chips, particles, and wood residues is very high compared to the reported production of solid wood products in France and Thailand. Notably, Thailand seems to have conflicting problems with the data on chips and particles production. Test ii suggests that the production is too high compared to the production of solid wood products. Yet, test i proposed that the production Forests 2018,9, 407 9 of 17 is too low to cover the reported exports. For France to reach the almost 18 Mm 3 difference between the production of final output and by-products (column F in Table 4), its volume of activities related to further processing of wood to furniture and other products must be considerable in order for the figures to be credible. For some countries where the resulting figure for “log use coefficient” is closer to 2, such as in Brazil or Finland, the data may be realistic, because these countries also produce plywood, which can require higher wood input than sawnwood. For instance, in Finland, the log input in sawnwood production is closer to 2, and the use of sawlogs in plywood production is about 2.5 m 3 /m 3 , which means that somewhat high amounts of residues can be obtained as side-product. Also, in Sweden, the calculated log use of 2.1 m 3 per one cubic meter of solid wood products could be realistic considering that the use of wood input in sawnwood production is rather high in Sweden and that by-products are additionally obtained from further processing. Within a wider period of 2011–2016, considerable data problems can be detected (e.g., for Canada, which is not among the countries shown in Table 4focusing on the year 2015). Table 4. Production of solid wood products (column C), chips and particles (D), and wood residues (E), their difference (F), and the resulting coefficient of wood use per solid wood product output (G, sum of all outputs divided by the solid wood output) in year 2015. The countries are ranked according to the value in column (F). Country A B C = a + B D E F = C −D−E G = (C + D + E)/C Sawn-Wood Veneer and Plywood Solid Wood Products Chips and Particles Wood Residues Difference Logs Use Coefficient 1000 m31000 m31000 m31000 m31000 m31000 m3m3/m3 France 7514 367 7881 5740 19,920 −17,779 4.26 Brazil 14,797 3829 18,626 11,788 19,140 −12,302 2.66 Australia 5085 296 5381 13,962 2577 −11,158 4.07 Thailand 2850 305 3155 4239 8200 −9284 4.94 Poland 4835 445 5280 3178 6500 −4398 2.83 Italy 1470 446 1916 4800 −2884 3.51 Portugal 1134 80 1214 1275 2331 −2392 3.97 Chile 8372 1494 9866 10,208 1916 −2257 2.23 Sweden 18,174 95 18,269 9965 10,175 −1871 2.10 Finland 10,640 1207 11,847 8341 5314 −1807 2.15 Netherlands 185 185 899 808 −1522 10.22 Estonia 1770 164 1935 1870 1155 −1090 2.56 Spain 2453 489 2942 1856 1846 −760 2.26 Mali 130 27 157 480 400 −723 6.61 Lithuania 1248 111 1359 1162 774 −577 2.42 Sudan 11 11 383 106 −478 45.45 Luxembourg 78 78 422 99 −443 7.65 El Salvador 16 16 457 −441 29.04 South Africa 1966 97 2063 1926 450 −314 2.15 Latvia 3479 280 3759 2513 1549 −303 2.08 Slovenia 725 98 823 300 800 −277 2.34 Fiji 130 20 150 210 −60 2.40 Colombia 450 55 505 194 354 −43 2.08 Georgia 76 5 81 121 −40 2.49 Nicaragua 62 62 80 −18 2.29 Uruguay 487 218 705 716 4 −15 2.02 3.3. Test iii: Can the Apparent Wood Consumption Levels Sustain the Reported Production of Forest Industry Products 3.3.1. Case of Sawlogs and Veneer Logs Use to Production of Sawnwood, Plywood, and Veneer Table 5displays the deficit of sawlogs and veneer logs reported in the data for the period of 2011–2015 for countries with such deficits being larger than 50,000 m 3 in 2015. The table also gives the theoretically maximal apparent logs input (m 3 /m 3 ) for the reported solid wood production. It can be regarded to be maximal, because when calculating the deficit in the volume of logs used, it was generously assumed that all industrial roundwood imports but none of the exports are sawlogs and Forests 2018,9, 407 16 of 17 in several countries. a lack of reliable official data from some regions is obvious. As these freely available statistics provide valuable information for a wide range of businesses, media, policy making, and scientific analyses, it would be of utmost importance to further improve the data, despite the many challenges faced. Meanwhile, when using the current FAOSTAT forestry data, one should, like with any data, be cautious. For instance, if wood input coefficients and demand functions for forest industry products employed in a forest sector model are calibrated to match these data while ignoring the inconsistencies in the data, the model results are doomed to carry on these errors in projections of future supply, demand, and prices of forest products. Likewise, the currently available data do not form a reliable basis for calculating the development of the stock of harvested wood products needed in carbon stock accounting for all countries. Supplementary Materials: The following are available online at http://www.mdpi.com/1999-4907/9/7/407/s1, Supplementary File S1: Data for reference input coefficients (excel-file), Supplementary File S2: Description of the LP model (word.doc), Supplementary File S3: Additional results from the tests (excel-file). Author Contributions: A.M.I.K. was responsible for the problem formulation, data preparation, software used in the analysis, and the first analysis and draft preparation for the manuscript. B.S. participated in the analysis of the results and the co-writing of the manuscript and was also mainly responsible for funding acquisition. Funding: The research leading to these results was funded by the forest sector modeling part of the research projects for FORMIT (“Forest Management for the Mitigation potential of European forests”—financed by the EU’s 7th Research Program as project no. 311970) and POLYFORES (“Decision-making Support for Forest Ecosystem Services in Europe—Value Assessment, Synergy Effects and Trade-offs”, financed by the EU’s program FP7 ERA_NET Sumforest). Conflicts of Interest: The authors declare no conflict of interest. References 1. The FAO’s Forest Product Statistics (FAOSTAT). Forestry Production and Trade. Available online: http://www.fao.org/faostat/en/#data/FO (accessed on 5 April 2018). 2. Buongiorno, J.; Shushuai, Z.; Zhang, D.; Turner, J.A.; Tomberlin, D. The Global Forest Products Model (GFPM): Structure, Estimation, and Applications; Academic Press: San Diego, CA, USA, 2003; p. 301. 3. Kallio, A.M.I.; Moiseyev, A.; Solberg, B. The Global Forest Sector Model EFI-GTM—The Model Structure. Available online: http://www2.efi.int/portal/virtual_library/publications/technical_reports/ 15/ (accessed on 5 April 2018). 4. Havlík, P.; Schneider, U.A.; Schmid, E.; Bottcher, H.; Fritz, S.; Skalsky, R.; Aoki, K.; De Cara, S.; Kindermann, G.; Kraxner, F.; et al. Global land-use implications of first and second generation biofuels targets. Energy Policy 2011,39, 5690–5702. [CrossRef] 5. Jonsson, R. Econometric Modelling and Projections of Wood Products Demand, Supply and Trade in Europe. Geneva Timber and Forest Discussion Paper 59. 2012, p. 1. Available online: https://www.slu. se/globalassets/ew/org/centrb/f-for/pdf/econometricmodellingandprojectionsof-woodproducts.pdf (accessed on 7 April 2018). 6. Hurmekoski, E.; Hetemäki, L.; Linden, M. Factors affecting sawnwood consumption in Europe. For. Policy Econ. 2015,50, 236–248. [CrossRef] 7. Brunet-Navarro, P.; Jochheim, H.; Kroiher, F.; Muys, B. Effect of cascade use on the carbon balance of the German and European wood sectors. J. Clean. Prod. 2018,170, 137–146. [CrossRef] 8. Schyns, J.F.; Booij, M.J.; Hoekstra, A.Y. The water footprint of wood for lumber, pulp, paper, fuel and firewood. Adv. Water Resour. 2017,107, 490–501. [CrossRef] 9. O’Brien, M.; Bringezu, S. European Timber Consumption: Developing a Method to Account for Timber Flows and the EU’s Global Forest Footprint. Ecol. Econ. 2018,147, 322–332. [CrossRef] 10. Mills Busa, J.H. Deforestation beyond borders: Addressing the disparity between production and consumption of global resources. Conserv. Lett. 2013,6, 192–199. [CrossRef] Forests 2018,9, 407 17 of 17 11. The Intergovernmental Panel on Climate Change (IPCC). Datasets for use in the IPCC Guidelines FAO Data and How It Can Be Used in the IPCC Agriculture and Land Use Guidelines. Available online: https://www.ipcc-nggip.iges.or.jp/public/mtdocs/pdfiles/0910_FAO-IFAD-IPCC-Meetingreport.pdf (accessed on 5 April 2018). 12. Kallio, A.M.I.; Solberg, B.; Käär, L.; Päivinen, R. Economic impacts of setting reference levels for the forest carbon sinks in the EU on the European forest sector. For. Policy Econ. 2018,92, 193–201. [CrossRef] 13. The United Nations Economic Commission for Europe; The Food and Agriculture Organization of the United Nations (UNECE-FAO). Complete Database, all Years, all Products. Data for 1964–2016. Available online: https://www.unece.org/forests/fpm/onlinedata.html (accessed on 5 April 2018). 14. FAO. Pulp and Paper Production Capacities. Available online: http://www.fao.org/forestry/statistics/ 81757/en/ (accessed on 5 April 2018). 15. RISI. Available online: https://www.risiinfo.com/ (accessed on 3 October 2013). 16. UNECE. Forest Product Conversion Factors: Project Overview and Status. 2009. Available online: https://www.unece.org/fileadmin/DAM/timber/meetings/forest-products-conversion-factors.pdf (accessed on 5 April 2018). 17. UNECE and FAO. National wood resource balances. In Proceedings of the Geneva Timber and Forest Discussion Papers 53, Geneva, Switzerland, 31 March–1 April 2008. 18. UNECE. Forest Product Conversion Factors for the UNECE Region. Geneva Timber and Forest Discussion Papers 49. Available online: https://www.unece.org/fileadmin/DAM/timber/publications/DP-49.pdf (accessed on 5 April 2018). 19. EU FLEGT Facility. Available online: http://www.euflegt.efi.int/flegt-action-plan (accessed on 5 April 2018). © 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).