Thursday, March 02, 2006
Do The Math Part 2: Every 0.1% Increase in E-Commerce = -$1.8B in Commercial Printing Sales
My post of Friday 2/24 explained how the statement “every one percentage-point reduction in GDP growth cuts printing industry revenues by more than $1 billion.” It just doesn't work out.
But since they revised GDP for the fourth quarter to +1.6%, the actual number in decline of GDP from average to the decline in print for that time period is $3.9 billion per point or 4x as bad as the rule of thumb would indicate.
But that's not statistically valid anyway. What is valid, with an r-squared of 90% (which is stunning from a statistical forecasting perspective), is the relationship of the percentage of retail sales that are conducted by e-commerce and inflation adjusted printing shipments. That is now a mere 2.2% ($20.8B for Q3-05) of retail sales, but the equation generated by using that in a linear regression model is rather incredible, showing a -$1.84 billion drop in commercial printing shipments for every 0.1% change in the share of e-commerce as a part of total retail. The absolute value of the errors using it as a forecast model is only 2%.
PrintForecast Contrarian View subscribers will get the details and what it means in this Friday's podcast.
But since they revised GDP for the fourth quarter to +1.6%, the actual number in decline of GDP from average to the decline in print for that time period is $3.9 billion per point or 4x as bad as the rule of thumb would indicate.
But that's not statistically valid anyway. What is valid, with an r-squared of 90% (which is stunning from a statistical forecasting perspective), is the relationship of the percentage of retail sales that are conducted by e-commerce and inflation adjusted printing shipments. That is now a mere 2.2% ($20.8B for Q3-05) of retail sales, but the equation generated by using that in a linear regression model is rather incredible, showing a -$1.84 billion drop in commercial printing shipments for every 0.1% change in the share of e-commerce as a part of total retail. The absolute value of the errors using it as a forecast model is only 2%.
PrintForecast Contrarian View subscribers will get the details and what it means in this Friday's podcast.
