These kind of problems are analyzed by economics of organization but to point toward a specific theory you need some specific source of the problem/reason for why the long run costs can’t be calculated.
Note it’s not necessary about ease of calculation, even if the productivity calculation would require some fancy mathematical model the firm could just hire some economist to do it. So there must be some reason why it can’t be done properly (uncertainty, information asymmetry etc). The only exemption would be where savings from hiring the economist are too low to do the calculation in the first place.
For example, if you assume principal is risk averse and simply the cost of printer is certain but cost of the forgone productivity is uncertain then you will get different way of analyzing the problem and use different theories/models than when for example the long run costs simply can’t be observed by principal due to some information asymmetry. For example, If a an agent request printer, the principal, who does not have full information on the task, must also take into the account that maybe agent requests the printer not only for productivity purposes but also for some personal benefit - many people use printers for personal tasks as well, and thus the full cost of not buying extra printer might not be possible to calculate.
Or there might be simply bad incentives throughout the organization where actually the person responsible for managing organization is the agent and principals are shareholders and then again due to information asymmetry you could have problem of focusing too much on only observable performance.
There could be some behavioral reasons for this as well like hyperbolic discounting.
So depending on the situation you would use asymmetric information theory, game theory, behavioral theory etc. If you are looking for solution to specific problem I would recommend putting economics of organizations + name of the problem into google scholar to get results relevant to specific problem).