What version are you using? It works just fine on my 11.3 installation:
In[2]:= Expectation[1-Exp[-\[Rho] ((K (1+s*K^(\[Beta]-1)*\[Epsilon])-c*s*K))],{c\[Distributed]NormalDistribution[Subscript[\[Mu],c],Subscript[\[Sigma],c]],\[Epsilon]\[Distributed]NormalDistribution[Subscript[\[Mu],r],Subscript[\[Sigma],r]]}]
Out[2]= 1-E^(1/2 [Rho] (-2 K+2 K s Subscript[[Mu], c]-2 K^[Beta] s Subscript[[Mu], r]+K^2 s^2 [Rho] Subsuperscript[[Sigma], c, 2]+K^(2 [Beta]) s^2 [Rho] Subsuperscript[[Sigma], r, 2]))

Some general pointers though:
- Specify
Assumptions if symbolic computations are failing. Very often Mathematica needs to know that certain variables are real or positive rather than complex-valued
- Do not use
Subscript for variables/parameters in symbolic computations. Sometimes it works (like it did just now), but often enough it won't and it's just not worth the bother. It's better to stick with normal symbols most of the time. See also: Can we use letter with a subscript as a variable in Mathematica?
Subscript, it works fine on 10.4. – JimB Apr 03 '19 at 14:25Subscript! – user63801 Apr 04 '19 at 06:45