Well run Central IT (CIT) does a more thorough job for the organization... taken as a whole, and filtered through operational (not stated) organizational priorities.
Well run Shadow IT (SIT) can get faster, more responsive results.
Either, poorly run, gets poor results.
This is similar to the tragedy of the commons, in that any one department is better served for any one immediate need by doing it themselves... usually at a poorer cost-risk-benefit to the organization as a whole, if there are overlapping requirements, regulatory or legal issues, disaster recovery issues, and so on factored in.
Well run CIT that is not extremely well budgeted cannot (and usually should not) be as responsive to any one requesting department's own SIT.
CIT needs to balance the priorities of the entire organization - individual departments do not.
CIT needs to operate more efficiently; which means if Department A wants X, perhaps Departments F and Q want X, too, CIT should gather X requirements and see if we can do X only once. SIT just "does it", quick and dirty.
CIT needs to plan for maintenance and ongoing costs (financial, manpower, compute resource contention, etc.) over the next N years. SIT rarely cares about anything but their own immediate results.
CIT needs to keep regulatory and legal requirements in the forefront. SIT often doesn't know... or doesn't care to know... or doesn't care... about those requirements, again being focused on their own internal desires.
CIT cares about disaster recovery based on how much the organization does. SIT rarely considers this.
CIT needs to make informed choices about purchasing based on the needs and operational priorities of the organization as a whole. SIT does not and can not do that.
SIT generally knows that area of the business well, and can translate very vague requirements into something close enough to what the requester expected to pass muster. CIT needs much more detailed requirements, may have internal QA processes, and hopefully cares more about true accuracy measurement.