floating point – Adding two numbers in half-precision IEEE754 standart

I have to add two numbers using IEEE754: -14.1875 and -7.4375.

I managed to convert them to half-precision numbers in IEEE754:

1 10010 1100011000 (-14.1875)

1 10001 1101110000 (-7.4375)

As the exponent is different, I corrected the second number so it would have the same exponent as the first number (by shifting the mantissa by 1 position to the right):

1 10010 (0,) 1110111000

Now it seems that I should somehow get this result:

1 10011 0101101000 (-21.625)

I don’t really understand how I should work with exponents — I had 10010 and 10001 and somehow should get 10011 in the result. Adding up to that, directly summing mantissas (1,1100011000 and 0,1110111000, in two’s complement gives me the result of 0,0010100000, which I do not see how correlates to 0101101000. Could anyone explain to me how do I perform the final step of calculations?