Merge only returns partial data despite matches

Question

I am attempting to merge two data frames on Asset_NM, GRADE, and Time_Period, subsets of the data frames are below.

In the complete data frames, values of Time_Period range from 0-360, in both data frames. However, when I merge the data frames, I am only returned records with values of Time_Period that are multiples of three. So for example, instead of values of Time_Period of 0,1,2,3 etc up to 360 it returns 3,6,9,12 etc up to 360.

ex.df

ENTITY_NM ALM_PORTFOLIO Asset_NM        GRADE FINANCIAL_ELEM_ID CF_DATE Time_Period
743    A        E   Corporate-Domestic     A      100      8/15/2017           0
744    A        E   Corporate-Domestic     A      430      9/15/2017           1
745    A        E   Corporate-Domestic     A      430     10/15/2017           2
746    A        E   Corporate-Domestic     A      430     11/15/2017           3
747    A        E   Corporate-Domestic     A      430     12/15/2017           4
748    A        E   Corporate-Domestic     A      430      1/15/2018           5
749    A        E   Corporate-Domestic     A      430      2/15/2018           6
750    A        E   Corporate-Domestic     A      430      3/15/2018           7
751    A        E   Corporate-Domestic     A      430      4/15/2018           8
752    A        E   Corporate-Domestic     A      430      5/15/2018           9
753    A        E   Corporate-Domestic     A      430      6/15/2018          10
754    A        E   Corporate-Domestic     A      210      7/15/2018          11


ex.df2

      Asset_NM                GRADE  Months_Out Aug 2017 Sep 2017 Oct 2017 Time_Period
723 Corporate-Domestic     A          2      1%       1%       1%           3
724 Corporate-Domestic     A          3      1%       1%       1%           4
725 Corporate-Domestic     A          4      1%       1%       1%           5
726 Corporate-Domestic     A          5      1%       1%       1%           6
727 Corporate-Domestic     A          6      1%       1%       1%           7
728 Corporate-Domestic     A          7      1%       1%       1%           8
729 Corporate-Domestic     A          8      1%       1%       1%           9
730 Corporate-Domestic     A          9      1%       1%       1%          10
731 Corporate-Domestic     A         10      1%       1%       1%          11`

exmerge <- merge(x = ex.df, y = ex.df2, by = c('Asset_NM','GRADE','Time_Period'))

             Asset_NM GRADE Time_Period ENTITY_NM ALM_PORTFOLIO FINANCIAL_ELEM_ID CASH_FLOW_DATE Months_Out Aug 2017 Sep 2017
1  Corporate-Domestic     A           3         A             E               430    11/15/2017          2       1%    1%
2  Corporate-Domestic     A           6         A             E               430     2/15/2018          5       1%    1%
3  Corporate-Domestic     A           9         A             E               430     5/15/2018          8       1%    1%
  Oct 2017
1       1%
2       1%
3       1%`

I am almost certain that the problem has to do with the Time_Period column in ex.df, which was derived to be either 0, during the present month, or 1 for each month out from the present:

`ex.df$Time_Period <- ifelse(ex.df$FINANCIAL_ELEM_ID == 100,0,(as.yearmon(ex.df$CASH_FLOW_DATE, format = "%m/%d/%Y")-FORECASTdate)*12)
`

with

FORECASTdate <- as.yearmon(Sys.Date(),format = "%m/%d/%Y")

The strange part is that when I don't derive Time_Period with the formula above, and use, say:

ex.df$Time_Period <- seq(0,11,1)

I get the expected results, so I really don't know what to make of that. Thanks in advance for your help, this has me stumped.