Thursday, January 18, 2007
What Will 2007 Hold for Printing Shipments?
The recent bullish rise in industry shipments has changed the outlook in our different forecasting models for 2007 and 2008. Forecasting is sometimes viewed as guesswork or a black art, but there are numerous tools that forecasters use, and we use as many as we can find. Through the years, we have found that the use of statistical models helps to focus the discussion and immersion in anecdotal accounts are used to enrich prognostications.
Every day, we review forecasts, and sometimes we joke there are often more forecasts than there are real data. There is one aspect of forecasts that continually befuddles us: the “consensus forecast.” This means that someone has taken all of the forecasts that were available and averaged them together. If there were true consensus, the forecasts would not have to be gathered in the first place. And if forecasting was something important, why would one want to average a bunch of bad forecasts together hoping that something good would come of it. One can't make a good stew with tainted meat and rotting vegetables.
Forecasts that are statistical in nature use historical data to forecast the future. The data are just numbers that resulted from business activities at a particular point in time. The conditions of that time are probably different than what will be in the future. While all trends have historical roots, it is much too simplistic to just extend past trends into the future. There's something comforting about “predicting the past.” It ignores that marketplaces, in the long term, are dynamic. There is one benefit, however.
When in high positions, executives can find it difficult to distance themselves from the daily heat of the corporate battle. Some managers fall into a habit of assuming the last thing that they heard was correct and was the most important thing they will hear affecting their business.
For that reason, having a statistical forecast of trends that might be embedded in one's business or market ensures a sense of detachment required for good planning.
PrintForecast.com runs three different models on the industry shipments data series we maintain. One is a straight-line model. Another detects seasonal patterns. Yet another fits an exponential curve equation to a data series. They are nothing more than arithmetic exercises, no matter how sophisticated they seem.
By constantly reviewing the literature of the markets that feed communications demand, such as what is happening in direct marketing techniques, publishing, advertising, graphic design, photography, telecommunications, broadcasting, and other fields, one arrives at a general understanding of what is happening in our industry today, and what forces will impact it in the future. It's easy to let the hyperbole of technology and its marketing affect forecast decisions. Again, including a basic statistical forecast in the process helps one to better understand the issues.
Good forecasts require a combination of approaches, but someone, sometime, has to put a stake in the ground and proclaim what the forecast is.
Our forecast models for 2007 show three scenarios. For comparison, this past year's inflation-adjusted shipments will be about $91.4 billion. Model 1 estimates that shipments in 2007 will be $92.9B. Model 2 says $92.7B. Model 3 says $82.3B. We can say with great confidence that all of the models will be wrong. But that's the first step.
Models 1 and 2 have formulas that more heavily weigh the recent past than distant history, just like that executive who values the last thing that was heard in a casual hallway conversation. Model 3 has a longer term view, and in light of the past five years trend, and not the last five months, forecasts more pessimistically—about -11% less than the other two forecasts. The consensus of the models would be $88.6B, if one takes comfort in that. It almost sounds reasonable.
So what forecast should one use? $91.4B? $92.7B? $82.3B? $88.6B? All of them. A company needs to prepare itself for whatever the marketplace can hurl its way. Knowing that Model 3's forecast of an -11% decline in shipments is part of the mix should create a process of asking and answering hard questions.
We should also remember that not a single statistical forecast produced in 2000 would have projected a scenario of industry shipments decline as it actually occurred. Any planning exercise should include the possibility of a totally unlikely industry scenario. Sometimes that's called disaster planning. No, it's part of any thorough planning process, not just disasters.
What's the official PrintForecast.com forecast? $89 billion. Any resemblance to the consensus forecast is purely coincidental. It's actually where our dart landed.
Every day, we review forecasts, and sometimes we joke there are often more forecasts than there are real data. There is one aspect of forecasts that continually befuddles us: the “consensus forecast.” This means that someone has taken all of the forecasts that were available and averaged them together. If there were true consensus, the forecasts would not have to be gathered in the first place. And if forecasting was something important, why would one want to average a bunch of bad forecasts together hoping that something good would come of it. One can't make a good stew with tainted meat and rotting vegetables.
Forecasts that are statistical in nature use historical data to forecast the future. The data are just numbers that resulted from business activities at a particular point in time. The conditions of that time are probably different than what will be in the future. While all trends have historical roots, it is much too simplistic to just extend past trends into the future. There's something comforting about “predicting the past.” It ignores that marketplaces, in the long term, are dynamic. There is one benefit, however.
When in high positions, executives can find it difficult to distance themselves from the daily heat of the corporate battle. Some managers fall into a habit of assuming the last thing that they heard was correct and was the most important thing they will hear affecting their business.
For that reason, having a statistical forecast of trends that might be embedded in one's business or market ensures a sense of detachment required for good planning.
PrintForecast.com runs three different models on the industry shipments data series we maintain. One is a straight-line model. Another detects seasonal patterns. Yet another fits an exponential curve equation to a data series. They are nothing more than arithmetic exercises, no matter how sophisticated they seem.
By constantly reviewing the literature of the markets that feed communications demand, such as what is happening in direct marketing techniques, publishing, advertising, graphic design, photography, telecommunications, broadcasting, and other fields, one arrives at a general understanding of what is happening in our industry today, and what forces will impact it in the future. It's easy to let the hyperbole of technology and its marketing affect forecast decisions. Again, including a basic statistical forecast in the process helps one to better understand the issues.
Good forecasts require a combination of approaches, but someone, sometime, has to put a stake in the ground and proclaim what the forecast is.
Our forecast models for 2007 show three scenarios. For comparison, this past year's inflation-adjusted shipments will be about $91.4 billion. Model 1 estimates that shipments in 2007 will be $92.9B. Model 2 says $92.7B. Model 3 says $82.3B. We can say with great confidence that all of the models will be wrong. But that's the first step.
Models 1 and 2 have formulas that more heavily weigh the recent past than distant history, just like that executive who values the last thing that was heard in a casual hallway conversation. Model 3 has a longer term view, and in light of the past five years trend, and not the last five months, forecasts more pessimistically—about -11% less than the other two forecasts. The consensus of the models would be $88.6B, if one takes comfort in that. It almost sounds reasonable.
So what forecast should one use? $91.4B? $92.7B? $82.3B? $88.6B? All of them. A company needs to prepare itself for whatever the marketplace can hurl its way. Knowing that Model 3's forecast of an -11% decline in shipments is part of the mix should create a process of asking and answering hard questions.
We should also remember that not a single statistical forecast produced in 2000 would have projected a scenario of industry shipments decline as it actually occurred. Any planning exercise should include the possibility of a totally unlikely industry scenario. Sometimes that's called disaster planning. No, it's part of any thorough planning process, not just disasters.
What's the official PrintForecast.com forecast? $89 billion. Any resemblance to the consensus forecast is purely coincidental. It's actually where our dart landed.
# posted by DrJoeWebb @ 5:00 AM 
