X Close

Forecasting Methodology

In order to see videos, you need javascript and flash active in your browser.

Video Demonstration

  • Forecasting Methodology Summary
  • Economic Drivers

Forecasting Methodology Summary

For nearly 30 years, RISI economists have used econometrics - the practice of applying mathematics and statistical methods to economic data - in developing a series of integrated models of the global forest products industry.

The models are tools that assist our regionally-based, grade-specific economists in interpreting data to produce their short-, medium- and long-term forecasts quarterly.

RISI currently maintains fifteen distinct models, containing over 5,000 equations. Each model is fully integrated across product areas and geographic regions to simulate the structure of the global forest products industry and capture its many complex interactions.

The accuracy of the models is tested over time by 'forecasting history' to see how the models would have predicted the past. The models are continually refined and adjusted as the structures of the various sectors evolve.

The end result is the most comprehensive and accurate forecast of demand and supply for the pulp and paper market.

Top-down, Bottom-up Approach

Each quarter the economists work through the model series for each grade that is forecast, for each quarter of the forecasted time-frame.

The results are checked against the assumptions the economists had going into the process. If the numbers are off, corrections are made in the models and the process is run again. The data will be run through the series as many times as necessary until the forecast is as accurate as it can be.

The final results tabulated in the first cycle are fed back into the beginning of the model series for the next quarter of the forecasted time-frame. (See the diagram outline for a visual representation of this process). Each subsequent model series feeds into the following quarter's model series until the end of the forecast's time-frame.

RISI's forecasting process uses a top-down approach to forecast demand (consumption) and a bottom-up approach to forecast supply (production). The end result is a forecasted price per tonne of paper by grade and world region.

The entire forecast can be represented in the following equation where each variable represents one of RISI's models:

End-use Indicator x End-use Factor = Consumption + Inventory Change = Apparent Consumption - Imports + Exports = Shipments/Capacity = Operating Rate

A final calculation is performed on Operating Rate, Production Cost and Mill Inventories to get to Price.

The below outline is a description of the process that takes RISI economists from raw market data to a forecasted price.

Economic Drivers

The forecasting process begins each quarter when RISI's 17, grade-focused economists, from across the world, join a macroeconomic meeting with RISI's lead macro-economist. Using historical data from published sources as his basis, the macro-economist presents a five-year forecast of macro-economic drivers that affect the global pulp and paper industry including: GDP, exchange rates, interest rates, chemical prices, oil prices, industrial production, unemployment rates, and others.

This meeting provides opportunity for each regional economist to learn about macro-drivers impacting his or her region, and to supply feedback on those drivers as well. This regional feedback helps the macro-economist to finalize the forecasted macro-economic variables.

After the macro-economy meeting, the regional economists gather again to discuss the paper and board forecast data. Fiber economists also attend to compare their expectations for pulp and recovered paper availability against those of the graphic and packaging paper economists. When the economists reach consensus and the graphic and packaging forecast data is finalized, the fiber economists can begin to produce their forecast.

End-Use Indicator

An important early phase of the forecast process is the collection of end-use indicator data . The end-use indicators are the drivers of demand for each major grade.

For example, the end-use indicator for folding boxboard/cartonboard, is the Industrial Production Index. This is a monthly index that measures manufacturing output for industrial goods. For use in the RISI forecast, the index is weighted to exclude heavy durable goods. This leaves non-durable goods that are packaged in folding boxboard/cartonboard - such as food and over-the-counter pharmaceutical drugs.

Other end-use data is collected from government sources, but is limited based on the type of information collected by the originating country or region. In Europe, published data is only available on non-consumer goods and is not broken down further into food and pharmaceutical end use.

Where end-use data is not available for a country - as in Central, South and Eastern European countries - GDP (as the only published historical data available) is used as the end-use indicator.

Examples of end-use indicators for each major grade:

  • Pulp - board and paper production
  • Graphic Paper - newspaper, magazine and catalog circulation, newspaper & magazine advertising sales, newspaper insert advertising, production index for periodicals, book and other printing, production index for commercial printing
  • Packaging - industrial production of food and non-durable goods, consumer spending, milk production, and others

After end-use indicator data is collected, other dimensions are then included to generate the end-use factor (amount consumed per unit of end-use indicator), to calculate true consumption.

End-Use Factor

The end-use factor is a ratio of consumption for a grade to the end-use indicator for that grade. Various drivers, such as the time trend (which captures shifts in technology over time) and the price for a grade relative to substitutes, are used to forecast the end-use factor.

Ultimately, the end-use factor indicates the amount consumed, per unit of end-use indicator. It is used in the end-use model to calculate consumption by end use for each grade throughout the forecast period.

For example, the end-use factor for folding cartonboard relates the price of folding cartonboard for the previous four quarters (adjusted for inflation) to the price index for consumer non-durables (the specific end-use market), relative to plastic film and sheet prices (the substitutes).

The end-use factor uses the same general equation for each grade forecasted. However, the specific end-use and market data used in the equation are unique to each grade, sub-grade and region and thus cannot be directly compared across grades or even countries.

The end-use factor is manually adjusted, as necessary, to account for technological or other factors the economists believe will affect the market over time.

Consumption of Paper by End Use

Consumption = End-use Indicator x End-use Factor

Consumption by end use refers to the amount of each grade and product forecasted is ‘truly' used. This may vary from the amount that is purchased, as the purchaser (end user) may buy more than what they will immediately use; holding stock in reserve.

End-user inventory

In some markets, data is available on end-user inventory. For example, printers and publishers in the graphic paper markets often make these data available publicly.

However, where the data are not available, RISI estimates end-user inventories, because of the importance they play in forecasting near-term demand. For example, large swings that can sometimes take place with panic buying will impact near-term demand.

End-user inventory is estimated historically as the difference between consumption and apparent consumption and is set at the same time as consumption.

Change in inventory model

Inventory is forecast using the change in inventory model, which uilds out a normalized inventory line across the forecast period. The economists balance recent consumption with recent end-user inventory to achieve a steady historically-based inventory to consumption ratio.

For example, if historically, inventories have been 45% of consumption, the model aims to forecast inventory change to reach this steady state.

RISI maintains a historical ratio of inventory to consumption back to 1972 for North American models and back to 1992 for European.

The economists also adjust inventory change figures, where appropriate, based on their experience analyzing the market.

Inventory change is primarily used to forecast for one or two years and is not as meaningful if it is forecasted beyond this period.

Apparent consumption

Apparent consumption = consumption + inventory change

Apparent consumption is what ‘appears' to be consumed in a country. For example what amount of folding cartonboard is purchased, rather than truly consumed. This is what RISI refers to as demand and it differs from true consumption, since people buy in bulk and create inventories.

Apparent consumption fluctuates on a quarter-by-quarter basis because of the building and use of inventories.

Grade model

Once apparent consumption for the overall grade is determined, the sub grades are calculated as shares of this total.

Published historical data are used for the base year and to forecast the demand share by grade. The relative price and availability (capacity) of different grades is factored in over the forecast period.

For example, 2010 North American apparent consumption of folding cartonboard is broken down into recycled and multi-ply cartonboard (40%); bleached cartonboard;(26%) and unbleached folding for carton uses (34%).

There is an equation for both bleached and unbleached cartonboard that consider capacity shares and relative price as the biggest drivers to calculate their share of the market. To obtain the forecast recycled share, the apparent consumption of bleached and unbleached cartonboard are subtracted from the total.

Each of the equations can be adjusted manually to account for market changes, such as shifts in policy mandates; consumer preferences for higher or lower quality grades; and technological changes in types of papers.

For example, in Europe there has been a shift from coated-recycled board to folding boxboard. Despite the higher cost and non-recyclability of the grade, safety issues linked with ink migration from coated-recycled board onto foodstuffs has changed market demand. While this is factored into the current model, if, over the forecast period, technological developments in coated-recycled board create higher demand, the Economists can adjust the equation to reflect the change.

International Demand/Supply Model

Historical, regional, import and export data are collected by each economist from customs bureaus in their region. These are fed into the international demand and supply model, which normalizes numbers globally.

The goal of the model is to obtain a net trade of 0, so that the exports to a country match up with the imports from the countries it exports to.

Trade model

This model takes the current import and export data generated from the international demand/supply model and applies forecast exchange rates, tariffs, and relevant cost and capacity expansion figures. The end result is forecast import and export data by country and by grade.

For example, in the folding boxboard/cartonboard model the inputs are: industrial production overseas, the import share of total consumption and international prices.

There is two-way feedback between this trade model and the above International Demand/Supply regional model so that any discrepancies can be worked through.

Shipments

Shipments = apparent consumption - imports + exports

The culmination of the forecast model at this point is the calculation of shipment data. The import and export data (by country and grade) from the trade model is used to calculate shipments (or domestic production).

It is here, where the top down approach using macro-economic and end use variables to forecast demand and consumption, meets the bottom up approach that forecasts supply.

Capacity model

It is here, where the top down approach using macro-economic and end use variables to forecast demand and consumption, meets the bottom up approach that forecasts supply.

The process up until this point has been focused on demand. This is where the supply forecast begins.

Current and announced mill capacity by grade from RISI's Mill Intelligence team is used as a basis. The forecast capacity model is applied to these figures and predicts assumed expansions and closures using a combination of factors:

  • Forecast prices and profitability
  • Historical capacity creep estimates
  • Changes in production to identify capacity swings

For example, it is possible that steep increases in recovered paper prices will generate rising costs for recycled mills, an expense that they are not able to pass on in the form of price hikes when markets are weak. If this should occur, the deterioration in industry profitability would likely lead to more capacity closures.

To cross-check the figures, the Economists ensure production or shipments figures are not appreciably higher than capacity figures.

Operating rate

Operating rate = shipments/capacity

The operating rate is a measure of market tightness - it is a ratio of shipments (demand) to capacity (supply). A market is considered to be tighter, the higher the operating rate for producers or the closer they are to maximum capacity being used.

A tighter market also means higher prices since there is little spare capacity or supply available for buyers to go elsewhere.

Production Costs Model

Historical and current production costs from RISI's Mill Intelligence team are used in the production costs model to forecast total costs.

The Mill Intelligence team calculates total cost as follows:

Total cost = cash cost + capital cost + SG&A cost

Cash cost = variable manufacturing costs + distribution costs

Capital cost = capital expenditure history +annual depreciation

SG&A cost = indirect mill expenses


Price Model

Price = Operating rate, production cost, mill inventories

Price is the last element to be forecast since it is a product of supply and demand.

To forecast price, the production cost of the previous two quarters is combined with the operating rate of the previous 4 quarters in the price model to estimate a forecast single price point.

The operating rate identifies how high producers can raise prices from the variable cost floor to generate a profit margin - the maximum price.

The mill cash production cost or average ‘variable cost' weighted for all sources of supply provides the cost floor or minimum price producers can charge.

As a second measure of market tightness after the operating rate, mill inventories may also be used to forecast pulp and packaging prices.

To double-check their work, economists use their knowledge of the markets, track price indexes and follow industry consolidation to feed into the price forecasts.

RISI's forecasting process as represented in the model series calculates each figure on a quarterly basis. It ends with price which is then fed back into the model series for the next quarter. Prices must feed back into the next quarter's model series because they affect profitability and thus can determine capacity closures or expansions. They also impact relative pricing in the U.S. vs. offshore markets and can be a factor in determining trade. Relative prices between grades determine some of the grade allocation in the grade model. Ultimately price is a variable that impacts future demand.