This thesis studies the impact that a change in seasonal adjustment method from X12-ARIMA run in SAS to X13-ARIMA-SEATS run in R would have on the quality of the French quarterly accounts.

The theory behing these methods is compared before analyzing how each one affects the adjustments' quality, how they react to revisions in the data and to new data points, and how they impact the difference between the evolutions of the accounts and of their indicators.

Strenghts and weaknesses are identified in both methods, leading to a compromise of keeping X12-ARIMA while incorporating new diagnoses inspired by JDemetra+.