df1                                       df2
     day     number   id   recordDay         day     number   id    recordDay
2017-03-21     17      1   1990-01-01     2016-03-21    6      2   1991-02-01
2017-03-22     19      1   1990-01-01     2016-03-22    8      2   1991-02-01
2017-03-23     21      1   1990-01-01     2016-03-23   10      2   1991-02-01

pd.concat(df_dict).to_csv('data.csv', index = False, data_format = '%Y-%m-%d)