A Re-Assessment of Residual Seasonality in GDP

Executive Summary

First-quarter GDP growth has been weak relative to the other three quarters in recent years (Figure 1). Even after the Bureau of Economic Analysis (BEA) published revised GDP estimates in July (the agency calculates substantial “benchmark” revisions to economic data every five years), average Q1 GDP growth remained 0.74 percentage points below the Q2 to Q3 average in the 2010- 2017 period (Figure 2). Consistent weakness in Q1 growth has led to questions about whether residual seasonality exists in seasonally-adjusted (SA) GDP data. Using GDP data published before benchmark revisions, the BEA itself has found statistical evidence to support residual seasonality.1 In this report, we conduct a re-assessment using the most recent GDP data from before and after benchmark revisions.

Our statistical analysis rejects the hypothesis of residual seasonality in revised GDP data in all of the sample periods we examine. This implies that analysts should look for other reasons to explain weak growth, and forecasters should not expect Q1 GDP growth to be weak in future years based on seasonality alone.

The Backdrop: Weak Q1 GDP Growth

In recent years, Q1 GDP growth has been consistently weak relative to growth in the year as a whole. This comes despite the fact that the BEA publishes SA GDP data. Seasonal adjustment is meant to capture the underlying trend in data by using statistical techniques to remove the influence of predictable fluctuations from calendar, holiday and weather effects.

Download the full report