In 2012, the monthly Colorado Department of Labor and Employment (CDLE ) employment press releases told a story about the economy that did not agree with what happened on the streets. The initial seasonally adjusted employment data depicted huge swings in employment, ranging from an unbelievable gain of 19,500 jobs in January to an equally absurd loss of 6,900 jobs in June. This is a range of 26,500 jobs.
The initial data showed losses in two months and no growth in a third. The initial data indicated that job gains only occurred in nine months.
The benchmarked revision, released in March 2013, told a much different story. There were consistent job gains in all 12 months, rather than the erratic job growth portrayed by the initial data. That range of job growth was 7,300 jobs, from a low of 1,700 jobs added in May to a peak of 9,000 jobs added in October.
The correlation coefficient between the initial data and the March benchmark data is .56. The coefficient of determination is .31. In other words, the relationship between the two sets of data is weak. It is difficult to understand why the initial data set does such a poor job projecting employment growth.
It is important for public and private leaders to have “accurate” data available to make critical business decisions relating to their industry. In this case, it was difficult for consumers to have confidence in the business climate when the story being told by state officials did not reflect what was actually happening on the street. CDLE must revisit its priorities. Publishing credible data is much more important than conducting a media blitz for the sake of gaining exposure for the agency.
©Copyright 2011 by CBER.