Assignment Data driven decision making BUSI 49675

Case study

Carol Katz works for the Occupational Safety and Health Administration (OSHA) within the Department of Labour. Her particular area of responsibility is air safety processes and, in particular, the integrity of safety processes at airports. The current system of assuring and monitoring safety standards involves unannounced inspections and audits of airports each month. There is a rolling system of fines applied to airports that violate standards, depending on how many times they violate. Carol wonders how well the system works. There are still plenty of violations and according to the last 18 months of records roughly 4% of inspections result in a formal violation notification. She has also noticed that there are more violations towards Christmas which is a high traffic month. However, the summer holiday months are also high traffic and do not have higher violation rates.

A naive question would be why airports violate at all, when they not only know they are being inspected but are also doing the “wrong thing” with their business. Presumably higher safety standards cost money and they are making some kind of a trade-off. She needs to write a report assessing the current system and recommending an alternative system if appropriate.

The current system involves regular inspections and fines contingent upon the outcome of these inspections. Inspection regime Under the current regulation system, OHSA send an inspector to each airport each month on a day that is not announced. The airports are assessed on their processes, not on outcomes. For instance, a “near-miss” logged by traffic control does not count as a violation. This is because some airports have much higher traffic levels and more difficult conditions than others. The focus is on the work force and managers adhering to their process. For instance, air traffic controllers must have mandated 16 hours rest periods between shifts, especially after late shifts. A second example is refuelling equipment, which must be checked and maintained every 50 uses. Checking this requires auditing of the most recent logs, comparing these with the equipment on the ground and also some interviews with technicians. There are over a dozen similar items on the safety checklist.

At the end of the inspection an airport is declared to have passed or failed. If they fail then a violation is recorded for that month. The fine imposed depends on the number of violations during a calendar year, and also on the size of the airport. Airports are divided into three classes – small, medium and large. The fine regime for medium airports, i.e., the scheduled fines (in units of $1000) depend on the number of safety violations according to the table below. These are total fines for the year. Notice that no fine is imposed for the first violation during a year. The second fine incurs a K$100 fine, the third K$250, the fourth K$500 and the fifth K$1000. Each further violation is fined a further M$1 but this number of violations is probably unlikely. Each calendar year the inspection regime begins afresh. The results of safety inspections are not available to the general public but they can be obtained by the carriers.

Airports have the power to improve their safety standards. For example, air controllers are in short supply. And this can lead to violations of the mandated rest periods. To hire a few more would cost money in overheads and entitlements and also probably require better salary and conditions to attract them. Routine maintenance is a similar story. More and better technicians would reduce the chance of a safety violation

For medium airports, OHSA have estimated that it would cost around M$2.5 per annum to reduce the rate of violations from 100% to zero. What they mean by this is that there is a certain level of staffing and tooling that would virtually guarantee a violation each and every month. Raising staff and tooling levels would virtually guarantee no violations. The monetary cost of moving from one state to the other is estimated at M$2.5. It would also obviously require a commitment from management, which is not costed.

For the sake of simplicity, Carol decides to assume that the relationship between violation rates and expenditure on staff and tooling is linear. Thinking a little more about violate rate, she decides that she will measure this by the probability p of a violation in any given month. So the relationship between safety cost and violation rate is

Safety Cost  2500  (1 – p)

The problem

Carol has to assess how well the system is working and identify any strengths or weaknesses. She knows that the system is far from perfect. Violations are still happening.

To try to assess the system she decides to put herself in the shoes of the airport. She says to herself: If I were running an airport and subject to this safety inspection regime I would worry about safety violations because they potentially cost me money, and also possibly reputation with the carriers. On the other hand, reducing the chance of these violations costs money. The chance of a fine depends on the probability of violation p. So does the safety cost. What value of p would minimise my cost?

She begins by writing down the fine table and asking:

What is the chance of an airport having different possible numbers of fines during the year? This obviously depends on p. Why do not I start off assuming that p=5%, since this is close to what I have seen recently.

She needs to fill out the probabilities for each possible number of fines in a year. Once she has done that she needs to combine the probabilities with the fines to come up with a typical or average safety violation cost from the airport’s point of view. Then she has to consider the safety which also depends on p. If she could put this all into excel, she could then experiment with different values of p. She might then be able to explain why there are 5% violations. Perhaps it is simply a rational reaction of the airport managers to the regulations.

You are asked to prepare an individual audio-narrated presentation that addresses and analyses the following issues:

Create a spreadsheet that will be helpful for Carol to use to decide on probability p of violation. Consider different probabilities of violations p and explain how you created the spreadsheet. For example, consider p=0.05, two values of p greater than 0.05 and two values of p smaller than 0.05.

What is the mean number of violations and mean fine that the airport pays for different values of p?

What is the safety cost that the airport has to pay for different values of p?

Discuss the costs that would include both safety cost and fine that will be incurred for different values of p.

Which probability of violations p seems reasonable to implement at the airport?

Create any graph that will be useful for descriptive analysis to discuss the problem.

Why is it important to rely on data in this context? What benefits may it bring?