We use administrative micro-level data on the full population of all 2,537,713 divorce cases in Germany over the 2006 to 2020 period to empirically show that marriages with weddings at year end last considerably shorter than does the average (median) marriage. Specifically the average (median) difference (conditional on divorce) between marriages with December weddings and marriages with weddings in all remaining months is 466.1 (534) days with overall length of marriage being 5275.7 (4542) days.
We attribute this empirical observation to the German marriage tax benefit which is granted for the entire calendar year if the couple is married for at least one day therein. In response to this incentive, couples use forward shifting of the ’legal’ wedding date as a tax planning strategy to collect the marriage tax benefit for the year(s) prior to their ’counterfactual’ (i.e. in the absence of any tax benefit) date of the wedding. We theoretically model this decision and argue that, in essence, couples forego sufficient courtship time to collect the marriage tax benefit, and hence, mismatching in long-term relationships results.