import pandas as pd
import numpy as np
ycap = [2015, 2016, 2017]

df = pd.DataFrame({'a': np.repeat(ycap, 5),
                    'b': np.random.randn(15)})

       a         b
0   2015  0.436967
1   2015 -0.539453
2   2015 -0.450282
3   2015  0.907723
4   2015 -2.279188
5   2016  1.468736
6   2016 -0.169522
7   2016  0.003501
8   2016  0.182321
9   2016  0.647310
10  2017  0.679443
11  2017 -0.154405
12  2017 -0.197271
13  2017 -0.153552
14  2017  0.518803

a         b     c
0   2015 -0.826946  2014
1   2015  0.275072  2013
2   2015  0.735353  2012
3   2015  1.391345  2011
4   2015  0.389524  2010
5   2016 -0.944750  2015
6   2016 -1.192546  2014
7   2016 -0.247521  2013
8   2016  0.521094  2012
9   2016  0.273950  2011
10  2017 -1.199278  2016
11  2017  0.839705  2015
12  2017  0.075951  2014
13  2017  0.663696  2013
14  2017  0.398995  2012

gp = df.groupby('a')
df['c'] = gp['a'].apply(lambda x: x-1)