I have to assume you mean material?
Short answer no reason at all but it would be like eating candy and not brushing your teeth afterwards. After a while your going to run into a bad case of decay
In a past topic I suggested that there is a difference between inherited performance loss as compared to procedural optimization of combining mesh with material assets. When it comes to geometry UE4 is very efficient at rendering HIGH detail geometry with little impact to FPS and rendering performance and even if as a single object with a 1 meg poly count (2 meg tris) UE4 with a decent video card will have no problems of pushing polygons. You can if you wish cut the model up into modular parts and export as separate objects as in place relative to 0 0 0 origin. This way you can drop the parts package into the scene and whatever is not visible will be culled and not rendered so the total poly count is “only” equal to whats in the view port. (excluding memory footprint of course)
Material though is where your going to run into inherited performance loss as the user can easily create noticeable FPS drop because there is no real way to control the optimization as a procedural process.
It’s actually very easy math using a single texture map at 4k resolution = 4096X4096 would be equal to the same area of resolution at 4 X 2048X2048 but in the process would inherit 4 times the performance loss as to draw calls.
So as a single map object by design broken down into separate object UE4 will cull what it does not see and for all thous objects using a texture atlas would only require a single draw call and not an additional call for each material added.
For that matter if you are doing mobile or VR using a texture atlas is almost a must.