First, are we to assume that "X(s)" and "Y(s)" are the Laplace transforms of x(t) and y(t)? I don't see any reason why you just cannot calculate directly that the Laplace transform of y(t) is [itex]e^{-s}X(s)[/itex] or get it from a table of transforms.