Wednesday, August 5, 2020

Networks Are Stabilizing: What About Cities?

The last five months (it's only been five months?) have been rough. Right now, we are currently in a cycle where legacies are rolling out massive cuts every couple of weeks. If you only read capacity change reports every couple of weeks, you would likely feel like the industry continues to spin in a downward spiral. While it often feels this way, I'm here to tell you airline networks are stabilizing. Yes, really. 

As an industry, network capacity has stabilized since July to around 14,000 average daily flights and 1.6M average daily seats. While this is nearly a 50% reduction in capacity, year-over-year, we are not seeing the continued decline in network capacity as we were seeing in March thru May, nor do we see a recovery that we saw in May thru July. Instead, we are stable at our 50% year-over-year decline. However, this stability is a hard pill for all of us to swallow with WARN notices spread across the industry. 

While I would expect some additional pruning of September capacity, I believe most of the legacy capacity was finalized over the weekend. If there are more substantial cuts coming, expect them from more point-to-point carriers such as B6, WN, and F9 as well as HA.

As networks have stabilized, legacy carriers have invested and often restructured their hubs. Since American's merger with US Airways, American has focused on building large, peaking banks rather than flatter rolling banks of the American-past. 

Let's take American's most bank dependent hub, Charlotte. Before the pandemic, American was running nine distinct banks within their CLT hub. Just two months later, American had gutted the structure to only four slightly smaller banks. American clearly positioned these banks within the core of the day as well. 

By July, seven of the banks had returned. Interestingly, these banks were largely restored to their March size, just fewer banks overall. As an example, the 9:30 departure bank peaks at 60 departures in both the March and July schedules. Finally, by September, the hub is still working with seven banks; however, American continues to adjust the bank sizes. The first two banks are slightly larger than their March compares, while there appears to be some horse-trading in the middle three banks. 

Delta, on the other hand, had historically used a blended, rolling bank within its Atlanta hub. Using March as an example, Delta only has a few peaks within its structure. Instead, there is more a continuous feed of flights throughout the day. However, by May, the rolling hub was gone, and Delta had switched the hub entirely to eight mini-peaked banks. Clearly, Delta was attempting to generate as many connections as possible for each flight. 

By July, each of the banks grew by at least ten additional flights with limited flights between each of the banks. This starts to change in the September schedule. In the September schedule, we can see Delta's willingness to grow outside of the peak bank structures starts to grow. 

While network planners are rebuilding banks, they are also examining how hubs are interacting with each other. Hopping back over to American's network, the makeup and proximity of CLT, DCA, and PHL to one another could create duplicate capacity on the network. As a planner, you are already competing with other carriers for scarce demand. You really do not want to be competing against yourself as well. 

When we examine American's leg passenger make up at CLT, DCA, and PHL, we quickly notice two things. One, from an absolute passenger count, CLT is massive compared to DCA and PHL. However, in terms of local passengers, all three hubs are roughly the same size. 

However, when we dig deeper, American is clearly treating each of these hubs independently. In September, DCA will only be operated at 29% of its last year's schedule. This compares CLT and PHL scheduled at 71% and 55% of their prior-year domestic trips, respectively. 

The significant DCA capacity decline is only possible thanks to FAA slot usage waivers that are currently extended through the end of October. Given the current level of passenger demand, I have to believe we will see the waiver extended at least thru the end of the year. With the September schedule, year-over-year, American will exit 21 markets and isolated their higher frequencies to American hubs, focus cities, and Portland, Maine. To be honest, I don't understand the last one. I suspect this might be an operational requirement, but it is difficult to know. 

When we take a look at the markets that American discontinued in September, we will quickly see two themes. Either American believes these markets can be recaptured, primarily via CLT or PHL, or American decided that the quality of the market demand is too low to warrant competing on these O&Ds. 

When we review both PHL and CLT, it is clear each of the cities offers distinct connecting patterns. When considering hub structure, PHL connects a lot of the Midwest and Mid-Atlantic to the Northeast and Europe (when it becomes available again...). PHL does overlap some with Northeast to Florida connectivity as well. 

CLT, on the other hand, offers a much larger breadth and depth of markets connecting the Midwest, Southeast, Florida, Mid-Atlantic, and the Northeast. This allows for robust connectivity and operational redundancy should it be needed. 

Interestingly, other than HPN, we do not see any domestic exits from CLT year-over-year. The reductions we do see are nearly entirely isolated to the number of banks planned this September vs. last September. 

This is likely a function of how American's network planners have structured the operation. Notice, CLT has limited Chicago and Dallas overflies. This again limits the amount of unnecessary redundancy within your own network, which would cause hubs to compete with themselves. 

While most hubs are starting to see capacity return, I do believe within the next month, barring an extension of the CARES Act, we will begin to see airlines remove spoke cities from their networks. If your city ended up on any of the carrier's DOT wish list, even if they are still operating there today, I would consider those cities the some of the first to be removed. 

Unfortunately, after that, we only need to go to our latest recession to see the standard targets, co-termed cities and small markets. In our next post (8/19), we will take a look at what airlines have historically dropped during network contractions. By then, it is very likely we will see Delta publish their revised October schedule, which is when CARES obligations expire and airlines can remove cities entirely from its networks. 


  1. Great analysis. Any reason why you left out LGA and JFK? LGA in particular offers connecting opportunities and competes with CLT, PHL, and DCA for connecting passenger traffic. Also, AA faces similar issues with PHX and LAX.

    I find AA's current hub structure and strategy to optimize those hubs of particular interest.
    Unlike DL and UA, AA has significant hub overlap, and I am curious to see how this plays out. My assumption is that AA drops two of the 4 following northeast hubs (JFK, DCA, LGA, PHL) and I would assume NYC is probably the most likely to go. While AA wants to have a big presence there because it is such a large market, their small scale results in an uncompetitive position for AA.

    Out west, AA needs to decide on PHX or LAX and move forward.

    1. On the LGA/JFK comment, I left those out because of the aggressive State of New York COVID travel restrictions. It would be tough to tease out the impact of network design decisions vs. a more direct impact from the state. LGA currently is down to peanuts vs. their standard capacity set.

      While there is some overlap between the hubs, they are functionally different. In regular times, JFK has international partnerships and now JetBlue. PHL has its domestic to European flow. CLT has the depth and breadth that serves the four major regions. LGA and DCA are high value local, slot restricted airports. If you leave or reduce in either, you’ll never get access back.

      I see the same thing with PHX and LAX. LAX is insanely cost-prohibitive, but there are certain domestic destinations you have to serve out of LAX for yours and the alliance’s international flow. Alaska’s domestic feed will help with this.

      PHX, on the other hand, allows much cheaper California connections. They don’t have much of an international network in PHX when you compare it to the other hubs. But that, again, is not a liability, in my view, is a function of deciding which customers are their priority. I’d expect both cities to remain large on the American network, barring massive restructures related to l COVID impacts.

    2. Given what you just stated, I am curious to know if you believe there is any opportunity for AA to close some hubs in an attempt to become more efficient / profitable? They have more hubs than any other legacy but lack the offsetting revenues to justify that many hubs when compared to the other legacies. As you stated above there is only so much demand to go around. Four hubs within a 200 mile radius of each other seems overkill to me.

    3. There is always an opportunity to improve network efficiency, but I not convinced shutting down any of the Northeast hubs is the right answer. I could see opportunities to monetize some of the LGA or DCA slots if the carrier needed to capitalize, but there's a long term trade-off.

      JFK is the preferred NYC international airport by your alliance. So, if you leave or reduce JFK, your value to the alliance is decreased.

      PHL is entirely unrestricted and is the only Northeast hub that can grow. So I don't see them needing or wanting to shut PHL down.

      Each hub is functionally independent. Yes, there is some flow overlap, but you do need some overlap in your system when the economy is healthy. While it is possible to close down a hub, reduce capacity, and increase yields, I think AA management knows that's a short term play that will constrain their future opportunities.