Market outlook: The science and art of chemical forecasting

28 June 2013 15:15  [Source: ICB]

Reliable price forecasting takes into account a multitude of market drivers that are carefully weighed and balanced to create an effective, tailored model

Volatile markets have financially stressed many businesses during the past few years. Companies have been burned by sudden price fluctuations resulting in much more change than budgeted.

This causes an unexpected impact on cost, cash flow, working capital and profitability as well as credibility. In today's market, taking the past year's price average and adjusting for inflation or deflation no longer works.

 A good forecast tool tracks much more than just costs and margins

Copyright: alamy

This has led to the increased need for reliable forecasting, especially where feedstocks and/or raw materials represent a significant portion of your cost of goods sold.

As a veteran forecaster of prices, supply, demand and related economic factors, I often hear the question: "What is the most important thing for forecasting?" But there is no one-size-fits-all approach. Each product, market, region or economy is potentially unique. This being said, a good forecast starts with understanding exactly what you are forecasting. This involves questions such as:

1 What is the magnitude (dollar-wise) of the decisions being made from this forecast? This should be reflected in the amount of time and money spent on it.

2 What does the trade flow and value stream look like?

3 What fundamental drivers are behind it?

4 What are the leading indicators in front of it? Repercussions from missing a forecast can be mitigated if you see it coming soon enough.

5 What are the parallel events or alternate value streams that can influence it?

6 Which of these factors typically exert the most influence?

7 Which factors can come into play as markets change, and what are their triggers?

8 What are the time lags between their cause and effect relationship on what we are forecasting?

9 Is there seasonality or repeating patterns? Usually the answer is yes.

10 What is the degree of predictability, randomness or noise?

11 Do the market characteristics or personalities change over time?

12 Does the forecast model need to be self-correcting and adapt to market changes?

13 What are the required confidence levels or allowable margin of error? Every forecast should have statistically determined confidence intervals or high and low limits that help understand and manage the risk to the forecast.

14 What are the risks to the forecast - the price of oil, foreign exchange rates or changes in consumer spending?

An example of a leading indicator might be the stock-to-use ratio of cotton in China, which is a key indicator of global cotton demand next year. If that ratio goes up, the price of cotton traditionally has a very high probability of going down next year. When the price of cotton exceeds a certain delta over polyester - a trigger level - the price of polyester begins to see abnormal increases.

The reason is because producers that make 80/20 blend clothing (80% cotton, 20% polyester) begin to change their blend.

Over the short term, it changes a little as they tweak blends to stay within tolerances (80/20 +/-3%) but if the price delta holds for long, the apparel industry, which uses the majority of cotton and polyester, shifts from 80/20 blends to lower-cost 60/40 blends. Then polyester prices are subject to a substantial increase as demand soars. This is an example of a leading indicator, a trigger, and a parallel market or alternate value stream - all of which may be considered in a good forecast model.

Once the market is understood, knowledge of statistics is necessary. Neither a PhD in statistics nor years of experience alone will help you arrive at the best model. It takes a background in statistics, market experience and some common sense.

A forecaster must realise that costs are driven by upstream raw materials and indirectly by oil in most cases. Cost is a floor below which producers will not sell at for long. Margins are driven by supply and demand.

Often, these two factors must be forecast separately. Margins are not only key to accurate forecasting but also valuable information when negotiating prices. Imagine playing a game like poker or blackjack where you know what cards are in the dealer's hand. You may not win every time, but you are not going home broke either. Negotiations can be aided by a good forecast.

Once the appropriate statistical work is completed and the forecast is in place, it is critical to track accuracy, identify where/why misses occur and to fine-tune the model. A good forecast tool is a living model that is maintained and improves over time. More advanced models can track their own accuracy and make self-corrections by assigning more weight to factors that evolve to influence the market more. Models can easily be designed to recalibrate their forecast each period after the actual values are entered to track accuracy. Many of these topics are covered in the ICIS Petchem II training class.

In the end, forecasting is a combination of being an economist, a statistician, a market expert and last but not least, a chef who can blend all the ingredients together to come out with something worthwhile.

ICIS Forecast Reports
Combining experience in price forecasting and ICIS market data and intelligence has resulted in two new ICIS Forecast Reports for:

1) Europe - Polypropylene and 2) US - Polypropylene.

Information from such forecasts has the potential in certain cases to save $200,000 per million lb purchased, if annual purchases to stock are made at seasonal low points versus high points. Supply and demand figures for key commodity chemicals are available on the ICIS ATEC Database.

For more information on ICIS Forecast Reports, visit

About the author
James Ray is a senior consultant for ICIS, based in Houston, Texas, US. He has worked for private industry (primarily capital management), with a focus on reducing raw material costs using market intelligence and establishing more favourable contracts throughout the supply chain. Sectors include manufacturing, converting and recycling polymers for blow moulding, extrusion and injection moulding, as well as textiles. Contact him at

Author: James Ray

AddThis Social Bookmark Button

For the latest chemical news, data and analysis that directly impacts your business sign up for a free trial to ICIS news - the breaking online news service for the global chemical industry.

Get the facts and analysis behind the headlines from our market leading weekly magazine: sign up to a free trial to ICIS Chemical Business.

Printer Friendly