Direct vs. Indirect Seasonal Adjustment for CPS National Labor Force Series

Thomas D. Evans


An aggregate series can be seasonally adjusted either directly or indirectly by adjusting its components and adding the results, or dividing in the case of ratios. The results will rarely be identical. To preserve additivity, it is often preferred to indirectly adjust many series. However, in some instances this may lead to an adjustment with residual seasonality, large revisions, or less stable results than direct seasonal adjustment. To test these possibilities, I directly adjust the 392 national Current Population Survey (CPS) series (that are currently indirectly adjusted) with X-13A-S and summarize the results. Key diagnostics for comparison include the spectrum and revisions. Further discussion deals with cases where a direct adjustment might be more appropriate than indirect adjustment.