The U.S. Census Bureau recently announced a revision to its monthly construction spending survey results that affected more than 10 years of data. Although the revisions were minor for many periods, some major errors appeared as recently as the third quarter of 2014. For instance, the originally released Private Residential Construction data that quarter showed a 3.5% decline in spending year-over-year. However, the revised data now indicates that spending grew 10.9%. There is a big difference between -3.5% (recession) and a healthy 10.9% improvement in spending. Since many businesses rely on this Census data as a proxy for the health of the construction industry and as a leading indicator for the economy as a whole, projections based solely on these numbers would have been completely off base.
Data revisions are part of life when looking at and using economic data.
As organizations receive more and better survey information, they revise initial findings to reflect their new understanding of the industry. Given the nature and inherent challenges to relying on survey data, it is important to look at multiple data sources and variables to get the complete picture. Yet despite this propensity toward revisions, it is surprising that many businesses often settle for one or two data sets as a measure of overall demand, leaving decision makers with a high and unnecessary risk of making projections based on inaccurate data.
There is no shortage of data and sources when it comes to measuring the economy.
Here at Prevedere, we collect approximately 2 million data sets from over 200 sources for our clients. For those affected by residential construction, for example, we not only factor in the U.S. Census Bureau’s survey referenced above, but also separate surveys of U.S. Building Permits, Housing Starts, Completions, Remodeling and Architectural Billings, to name a few. A quick look at the same third quarter 2014 period that the Census originally reported at -3.5% shows that Housing Starts were up 12.8% year-over-year and Architectural Billings (which precedes construction) accelerated throughout the first half of the year. Not to mention, signals from Disposable Personal Income, Consumer Sentiment and Retail Spending, among others, also painted a much more positive picture of the U.S. economy than the error-filled Construction data report. A business that looked at all of these factors would have taken the negativity in the Census survey with a grain of salt.
Looking at only one data series presents an unnecessary and substantial risk of making poor strategic decisions.
No one wants to be the business leader that was planning 2015 off of the falsely negative U.S. Census Bureau numbers. Improved forecast accuracy requires multiple data sources, making predictive models more likely to capture the various and always-changing economic headwinds and reducing exposure to data revisions in a single indicator.
While traditionally such models and improved accuracy required the cumbersome task of sifting through millions of data sets in order to achieve a diverse and substantial understanding of the economic conditions, these models are now obsolete. Prevedere was built so businesses can efficiently mine millions of economic data points to identify those that accurately predict future demand for their products and services. Contact us for a demo today to see how we’ve helped companies like Hershey, RaceTrac and Wendy’s leverage big data to improve performance.