Perhaps if copying a proven design, but even then the economies of scale are against you. Now if there is no value assigned to labor, and you already have the tools, the gap is closer, but it's the "proven design" part of the equation that makes a difference IME.
There's also the cost of developing the design and building prototypes that's often (conveniently ignored), but by ignoring even the minimal cost of design, labor, quality hardware, testing and wasted materials, then yes it can appear to be less expensive.
There are also the DYI designs that, due to inexperience or lack of understanding, become "amp killers" due to not understanding reactive load limits in amps when it comes to crossover design, that bring hidden costs to the table. Yes, I run into this often enough, especially from the OCD crowd that thinks that every crossover needs multiple Zobel networks without any consideration for what ends up being lumped resonant tank circuits.
This is just my experience, based on what I have seen over the (many) years.