Why is to_sql() slower than iterrows() and cursor.execute()?

Question

In this code simple cursor.execute() is used which takes much less time than sqlalchemy

insert = 'INSERT INTO {} ('.format('TCHOTEL_BO') + ', '.join(df_hotel.columns) + ') VALUES '
df_hotel['uploaded']=['0']*df_hotel.shape[0]

for i,row in df_hotel.iterrows():
    values = map((lambda x: "'"+str(x)+"'"), row[:-1])
    inserting = insert +'('+ ', '.join(values) +');'
    cursor.execute(inserting )
# cursor.commit()

Time taken by this 0:00:36

In this code sqlalchemy.engine is used which is taking too much time

df_trips.to_sql('TCTRIPS_BO', con = engine, if_exists = 'replace', chunksize = 1000, index = False)

Time taken by this 0:01:21

Can anyone please tell me why sqlalchemy is taking too much time although it is more efficent ways that must be much faster that simple cursor.execute method because for every iteration it creates connection with database