Answer: Not necessarily.
The case below was studied in my M.Sc. dissertation. In the problem of a
real option F to develop a project with value V and with development cost
equal D, so that in case of option exercise the payoff is V
Consider that, in addition to the project value (V) uncertainty represented by the volatility sV, the investment cost (D) is also uncertain with volatility sD. Assume that both (V and D) follow correlated geometric Brownian motions.
What happen if we start to add cost uncertainty into this model? The answer depends of the correlation and of each individual volatility.
The chart below shows three different correlations between the two processes. The correlation is expected to be positive for the general case.
The chart above shows that for these positive correlation and for cost
volatility lower than ~9% p.a., the addition of cost volatility reduces
the total uncertainty of the process!
Only for independent (r =0) or for negative correlation, that the cost uncertainty will increase the total uncertainty (see the 3D chart below). However, in most cases the correlation is positive as in the picture above.
The 3D chart shows a whole range of correlations. The equation for the total uncertainty is also displayed in the picture.
For the equation displayed in the chart for the "total uncertainty", see for example:
This case is simple thanks to the property of homogeneity of degree one
of the differential equation, so that is possible to work with the ratio
V/D for the underlying asset (with its "total volatility") and
the ratio F/D for the option.
Homogeneity of degree one means that F(V/D, D/D) = (1/D) F(V, D), and it permits to use V/D in the maximization problem. The threshold (V/D)* is homogeneous of degree zero in V and D, see the classical paper of McDonald & Siegel (1986, p.713).