To put it very simply, paper is produced from pulp. Pulp, in turn, is produced from wood, and to a lesser extent straw, bamboo, cotton and other natural sources of fibre. At each step in the mass-production, variation and diversification are deliberately created, such that the relatively undifferentiated raw-material input – wood – turns into a product highly specialized for a use-case, be that something to write on, to transport shoes in, or to wipe spilled dairy with.
The progression of chemical and material transformations can be understood as a value chain. At each step, value is created. The paper, or the intermediate pulp, can be traded and interact with other value chains, like food or consumer electronics.
Because paper is so ubiquitous, yet complicated to make and based natural raw material that is not evenly distributed over our planet, the value chain and trade is necessarily global. The industry in different countries are bound to adopt different relations to this global exchange paper and pulp, or more generally the value these products embody.
Because paper is a major industry, and because it relies on important natural resources, the industry is subject to a great deal of study. In this article, I deploy the powerful toolbox of modern Data Visualization and analysis to a data set of global production and trade of paper and pulp between 1998 and 2018. Relations and trends in the value chain of paper that are not simply a function of a handful of numbers can be discovered with that toolbox.
In this article, I will illustrate what this visualization toolbox can do, and in the process make the paper value chain, paper trade, and paper production better understood. Details on data and methods are put in the last section, along with some code snippets.
Paper Production Differentiated
To start with, I take a global view.
In the stacked area chart below I visualize the aggregate quantity of global production of paper between 1998 and 2018, as well as the paper types the aggregate is comprised of.

Over this 20-year period the aggregate production is increasing. The financial crash of 2008 and 2009 is visible in the chart as a slight dip in the growth curve.
The categories of paper types are quite broadly defined. Not all writing paper or packaging paper are identical after all. Still the groupings capture some distinct utilities, and it is at this level of detail I will perform the data analysis.
From a first inspection of the relative sizes of the areas in the above chart, the extent of what paper is being produced is evident. Paper to pack things in is clearly a major use-case.
It is informative to disaggregate global aggregates into national aggregates, literally speaking to add another dimension to the analysis. I select a specific year, and for each nation (or at least the major producers) illustrate the quantities of their paper production. The stacked horizontal bars below is for 2018.

The left-hand side diagram illustrates absolute amounts. Populous major manufacturing nations appear at the top, as with so many other manufactured outputs.
The right-hand side diagram illustrates relative amounts of the different categories of paper within any given nation. Differences in proportions of the domestic production is clearer in this view of the data. Among other things, Finland stands out for its high proportion of printing and writing paper and low proportion of case materials, and Canada stands out for its relatively high proportion of newsprint as compared to other nations.
In order to further drill down in the data, a specific type of paper can be selected, and the trends over time for several nations put in a data visualization that enable comparisons over time and space. That has been done for newsprint paper in the multiple line diagram below.

The coloured lines relate to the left-hand axis and visualize the trend for a handful of top-producing nations; the grey line relates to the right-hand axis and visualizes the trend for the global aggregate of newsprint production.
The two most significant producers in the late 1990s were Canada and the USA. However, their output has been declining over the twenty year period to become less than half of what it was. Initially, this took place without a decline in the global aggregate. The reduced outputs were substituted by increasing Chinese production. However, around the mid-2000, the trend across all nations is clearly downwards. The obvious cause is digital newsmedia, which has reduced this particular value of paper.
Global Trade of Produced Paper
Once produced in a certain nation, paper be consumed domestically or enter a different value chain of that nation, like packaging for other manufactured goods. Paper as such can also be traded across borders, another common feature of specialized value chains.
The exports and imports for 2018, understood as flows out of and into a nation, is visualized below. (Code snippet available at end of article.)

On the left-hand side, the negative flows, the dotted stacked bars illustrate the types of papers the nations export. On the right-hand side, the positive flows, the criss-crossed bars illustrate the types of papers the nations import.
Inspection of the visualization reveals that different nations exist in different relations to the global market of paper.
- Finland, Sweden, Canada, Indonesia, for example, are considerably skewed towards negative flows, or in other words, they are net-exporters.
- China, USA, France and in particular Germany, trade a great deal with paper, as both importers and exporters. That it makes sense to trade in that manner is in part a reflection of that not all paper is identical, and that, say a German use-case is better served by paper from, say a Finish mill rather than a German mill. Also, it shows mercantilism is not what it used to be.
- United Kingdom (and the many nations omitted from the diagram), import the bulk of the paper they use.
In order to visualize trends in trade flows, with the national categories and paper type categories still fully resolved, the same type of image as above is animated. Each frame in the image visualizes the trade flows for a year in the range 2000 to 2018.

Similar trends as before are evident in this data visualization. The reduced footprint by Canada, and the increased one by China on Global Trade stands out, which from a previous visualization is understood to be related to the decreased utility of newsprint. Constant features are evident too, like Sweden and Finland being major net-exporters in all years.
Another informative contrast is that between national trade flows relative aggregate production. The scatter-plot below visualizes the relevant quantities for 2018.

The national stock of paper is defined as the quantity of all paper either produced or imported in that year by a particular country. The horizontal axis denotes what percentage of the national stock that is exported, the vertical axis denotes what percentage of the national stock that is obtained through imports. Finally, the area of a disk denotes the total quantity of paper production for the corresponding nation.
Location and size of the disks reveal a number of different relations to the paper value chain and its global trade:
- Sweden and Finland, lower right, mostly produce paper to export.
- China and Japan, lower left, are major producers, but mostly produce to meet their domestic needs for paper.
- United Kingdom (and the many omitted countries that produce very little paper), upper left, meet most of their domestic needs for paper through imports.
- Germany, France, Italy and Canada, on the diagonal, meet their domestic needs with both production and imports, and they export a sizeable proportion of their production to the global markets.
These relations are mostly stable over the years under study. An alternative means to illustrate trends, other than animation, is to visualize the trace of the disks as an overlay. That has been done for an ever smaller subset of countries in the diagram below between 2000 and 2018. (Code snippet available at end of article.)

Global Trade in Pulp
So far, paper has been considered. But paper production requires pulp as input. Pulp too is an item that can be produced, exported and imported. That part of the paper value chain is characterized in the following data visualizations.
However, a limitation of the data set must be noted first.
There are mills that produce both pulp and paper. One part of the pulp produced in such mills can be dried and sold to other paper producers within or beyond the borders of the country. Another part of that pulp can be transported within the mill to an adjacent paper machine, and there transformed into paper. The data set only accounts for the former part, and neglects to account for the latter part. An understandable omission, given that the latter part is transacted entirely within a mill.
As far as the current analysis is concerned this omission means aggregate values for how much pulp a given country produces cannot be obtained, since a potentially large fraction is unaccounted for. However, global trade with pulp is contained in the data and can therefore be analyzed along the same lines as was done above for paper.
The same type of stacked bar chart for the export and import flows for pulp is shown below for 2018.

The first thing to note is that there are fewer categories of pulp in the data, which reflects in part that at this stage of the value chain, there is less variation in the product. However, as noted for the paper types above, these categories are not internally uniform (a missing distinction is pulp made from coniferous or non-coniferous wood, the former being more common in Northern nations).
Second, there are fewer countries that export any significant amount of pulp: Brazil, Canada, USA, Indonesia, Finland, Sweden and Russia. It is hardly a surprise that these are all countries with a great deal of forest within their borders. Trees are heavy and not easy to transport. A common observation in the supply chain for pulp production is that the wood transformed in the mills is predominately collected from a relatively small geographical area around the mill.
A very sizeable amount of pulp imports to China is evident. As concluded above, China is a major producer of paper mostly for domestic use. However, an appreciable amount of that paper is made with imported pulp.
Another contrast with the paper data analysis is Brazil. In the earlier scatter-plot, the disk corresponding to Brazil fell in the category of countries that produces paper mostly for its domestic use, and less for global trade. However, with respect to pulp a great amount is exported. Clearly Brazil, unlike Sweden and Finland, has elected to engage in global trade at a different stage of the value chain: monetize pulp rather than paper.
The animated version of the diagram is shown below.

It should be noted that pulp for paper production can be made in part of recovered or recycled paper. Although the fibre must originate from pulp made from wood (or other natural source) at some stage in the past, the immediate source can be other paper. Data not shown here suggests that a large portion of Chinese paper production has as its immediate source recovered paper. Therefore, the differential between amount of imported pulp and amount of produced paper must not necessarily be domestically produced pulp from a natural source.
Trade data can be further divided into the bilateral transactions, the network of exporter and importer countries that is. Quantities associated to a pair of categories is quite well visualized with a heatmap. The data below are for the wood pulp global trade in 2017.

A subset of countries that import wood pulp are shown as rows, and the subset of countries that export the greatest quantities of wood pulp are shown as columns. The saturation of the color in any given cell of the heatmap visualizes what percentage of total imports for the corresponding importer country that originates from the corresponding exporter country.
In other words, looking at a row of colours, the nations that supply pulp to the nation corresponding to said row, become quite clear. Comparing rows to one another, differences and similarities in how pulp is acquired are discerned.
Some observations:
- Countries on the European continent have similar relative distributions of their pulp imports, where they import an appreciable amount from Brazil, Sweden and Finland, and some from Canada and USA, but little to nothing from Russia and Indonesia.
- The East Asian importers acquire their wood pulp more evenly distributed from the exporting countries, Sweden and Finland excluded, which supply modest relative amounts.
- Canada and USA trade wood pulp with each other, as well as imports from Brazil.
- Brazil is a global exporter of wood pulp.
In order to drill down and view trends, one approach is to select a particular country and display time along the horizontal dimension. That has been done below for the imports of Italy.

General trends of pulp global trade already discussed are evident, among them the increasing volumes from Brazil and the decreasing ones from Canada. More subtly, the imports from Indonesia, though always fairly modest, have decreased. However, from previous diagrams, total exports from Indonesia are known to have increased. The growing Chinese demand is likely where the Indonesian pulp has been redirected.
What About the Trees?
As noted above, most pulp is produced through logging in forests near the mill in question. Global trade with wood is possible. However, the different possible uses of wood, such as furniture, construction, energy, would have to be factored into the analysis. That is beyond what I will do here.
Furthermore, logging and alterations to forests have numerous causes. As discussed elsewhere, global forest area is decreasing. On the other hand, the decrease is not uniform, in fact, Europe has over the same time period as the one in this article become covered with more forest.
Suffice to say, the paper value chain and the process of making paper and pulp are both advanced, instructive of emergent features of a global economy, and a rich data footprint is becoming available. The additional connections to the web of ecological interactions create even more space to explore.
Notes on Data and Methods
- The data is retrieved from the Food and Agriculture Organization (FAO) of the United Nations. The particular data is based mostly on official data from the different nations. FAO has invested in an easily accessible open data portal.
- All visualizations are done using the Python library Bokeh. This library offers a great deal of flexibility and control of the visualizations and their design, especially when the data processing is combined with the Python library Pandas. To guide the coding efforts with Bokeh, quick data overviews are done with Microsoft Power BI.
- Multi-dimensional data is inherently hard to investigate, and human visual perception puts limits on what a visualization can capture. Concepts from databases, like the OLAP cube, and the canonical operations of slicing, dicing and drill-down, help structure the data exploration. Data visualization principles are also helpful.
- A text written on Medium cannot utilize interaction with the diagrams. The animated diagrams are a related variant of how a continuous or ordinal variable can be disaggregated without requiring a spatial axis in the visualization.
- The GIF animations are made with the Python library imageio.
- Paper and pulp Manufacturing is a complex engineering science with numerous chemical and mechanical steps. There are introductory YouTube videos to the process. Also, as an industry, paper making is old. At least one historical argument has put paper in the top-50 of the most important innovations – the argument being that the revolutionary potential of the printing press would never have been realized if the written word kept on being put on animal skin rather than the cheaper and more flexible paper.
Code snippet: inner loop for the trace of disks diagram
Code snippet: main part of creating the export – import flow diagram