Wednesday, March 4, 2020

Managing the Network During A Crisis

In the past, most of my posts have been data-heavy. However, today's will be a break from those posts. If you follow me on Twitter (@FlyDataGuy), you likely saw a long-winded tweet providing guidance for airport professionals. I thought it would be appropriate to follow that tweetstorm with my thoughts regarding the coronavirus and how airlines are likely planning their networks with any demand changes.

You all are likely well aware of airlines liberalizing their change and cancel fees while reducing capacity to Asia and Italy as the coronavirus continues to spread. Clearly, airlines are feeling a squeeze in demand as the public reacts to the virus's spread.

Today's post will focus on what I believe is going on behind the scenes and what is weighing on professionals within the commercial groups (network planning, revenue management, marketing, and finance) as well as some considerations these groups have for operational groups. I will reference some of the internal data they may be reviewing, however, since it is... um, internal, I do not have access to it.

I do want to be entirely clear here. While I continue to have connections with professionals across a variety of airlines, you will not find any inside information here. In the unlikely event someone would actually discuss their decisions making process with me (they can't re: NDAs), I feel these types of conversations are privileged and should not be discussed openly. They would, however, help impacted professionals understand the operating conditions and constraints in which the commercial groups would be making decisions. Let's also not forget, the virus is still located in isolated regions throughout the US (we believe). As more cases are suspected and confirmed, it is likely the impact on the aviation industry will continue to grow.

So, where are we at right now? To date, outside targeted Asian and Italy flight reductions, March through August flight reductions have largely focused on domestic travel due to MAX aircraft availability. In total, domestic trips are down 1,776 trips vs 140 trips when comparing schedules published in January compared to current published schedules. MAX reductions to date have removed nearly 10x as many flights as any virus related reduction.



However, it is clear the aviation industry is under intense distress. United seems to be the most vocal regarding the impact. To date, United has reduced capacity to Asia, suspended 2020 earning guidance, canceled investor day, delayed future pilot classes, asked widebody pilots to take a reduced pay vacation, and warned future capacity reductions may be necessary.

How significant are these close in capacity reductions? Huge. Why? Costs. By the time airlines' operate their flights, a vast majority of their costs are largely locked in. Outside, of fuel, incremental maintenance, and some airport costs, cutting flights a couple of days out does not save that much money.


This is why we are not seeing widespread domestic cancellations in March. While I am sure demand is off for the month, cutting these close in flights would make a bad situation worse. As long as flights are covering their fuel and other truly incremental expenses, it is best for the airline to operate the flight rather than canceling it. Costs do not take into consideration the potential nightmare it would be to redevelop flight crew pairings and aircraft routings with near term capacity cuts.

Airline costs are often not well understood. This is often true by commercial professionals as well as the general outside world. When it comes to costs, airlines are capital intensive organizations with largely few variable costs. On the day of departure, nearly roughly 70% of an airline's costs are already locked in. Take United's 2019 cost structure, only ~30% of United's costs could be shed if they were to cancel flights across the network. Even this 30% is misleading. At most major airports, airlines are on the hook for airport expenses depending on the airline/airport operating agreement.


This 30% estimate could also range wildly at the route level. Different fleets and mission profiles may have different cost structures. The point, however, is for airlines to cancel flights due to lack of demand, demand has to really dry up. This appears what happened to United in Asia. According to the Wall Street Journal, US-based ticket sales to Asia are off a little over 50% year-over-year around the middle of February.

With softening demand, airlines have hinted there could be capacity reductions on the horizon. The commercial teams are likely actively working on these a variety of capacity reduction solutions if demand continues to soften. We are likely in the heart of spring and summer booking season. If the softness continues, airlines can start reducing flight schedules in a much more ordered manner which would allow some deferring or eliminating of costs.

When working within a dynamic environment, who do I believe is at the wheel driving decisions? I suspect most airlines have an active war room with commercial executives, revenue management, finance, and network planning. Revenue management is likely informing executives of the current booking trends and projected health.

Revenue management will be analyzing negative booking impact based on expectations of their booking curves. Revenue management would be looking at both near term and further out booking curves to understand the impact the virus is having on their operation.



With this information, executives are likely determining a few courses of action. They are likely guiding revenue management (manages ticket prices) inventory and pricing strategies. Executives from there are likely using the provided information to guide network planning on a general theme of what to cut if they believe cuts are necessary. Network planning will be taking this direction to start reducing capacity in impacted markets.

If you are an airport, your traditional network planning capacity planners are not likely the individuals actually making the decisions on which flights to cut. They might be informed and be able to weigh in on some options, but they are not the decision-maker here. Each airline has a team within network planning which manages the fleeting of schedules, as well as near term schedules decisions. These are the unsung heroes of network planning. They make the schedule operational. They also are at the front line when stuff absolutely hits the fan as well.

These teams are likely working hand-in-hand with finance to determine estimated variable costs by flight as well as revenue management of projected final booked revenue. The network planning team would then start weeding through the schedule to see which of these projected cash negative flights could be removed while keeping the schedule operations.

So here's the million-dollar question, where do I expect flights to be reduced? As always, it depends. If we continue businesses implementing travel bans, I would expect carriers to begin reducing their business networks. This would include large Northeast cities (New York/Washington), Chicago, Bay Area, Dallas, Houston, etc.  But some of these reductions would be dependent upon suspension of slot usage requirements (which IATA is requesting). Airlines are willing to operate empty flights in the short term rather than lose their slots to constrained airports.

Next, if the virus continues to cause long term demand concerns, airlines could take two approaches. First, expect to see long haul flights that overfly hubs to be targeted for reductions. Second, watch for the number of banks within hubs to be reduced. Reducing the number of hub banks would reduce the number of itineraries within a carrier's own network which competes with itself. Finally, as some outlets have started to report, entire fleets that are approaching retirement may see their retirement dates accelerated. This would help reduce some costs associated with unproductive aircraft.

As airlines continue to respond to the virus situation, it is incredibly important for those within the industry, especially airports, to understand how airlines are being impacted so you can plan ahead or you can address concerns as they arise. If you have access to GDS/MIDT data (I do not), you are at least four to seven months ahead of others in understanding the financial impact.

If you are new to analyzing airline revenue or you report entirely on load factors and passengers, please have an air service consultant help you. Trust and respect are the foundation to air service development. Actually understanding what airlines are looking at can help you build relationships with the decision-makers and lead to better results. Solely reporting on load factors and top O&Ds to airlines will not help, especially in dynamic times.

If you need help, please reach out to me. I am happy to get you in contact with some of my preferred air service development groups that can help you understand your airport's current performance and analyze the traffic results when the newest traffic data is released. I do not receive a commission, nor work for any consulting group. I'm offering so that others do not make the same mistakes I made in the past.  



Wednesday, February 26, 2020

Decoding Breeze's 121 Application

This post will focus on my interpretation of Breeze Aviation Group's Part 121 application currently under the Department of Transportation review. It is essential to understand that, while my beliefs are research-based, there is enough gray area in Breeze's application that my analysis should not be taken as absolute fact. Further, even if my interpretation is 100% correct (it won't be), by the time the airline actually flies within the next year or so, the dynamic nature of the industry could materially change Breeze's tactical market decisions.

On February 10th, Breeze Aviation Group's Part 121 (air carrier) application was posted on regulations.gov. For those interested in reading route cases, applications, and generally dull documents, this is the site for you. However, while much of the Breeze's application was (extremely) dry, there were more than a few nuggets, which, for us aviation dorks, were quite interesting. Today's post will analyze those nuggets and provide my best guess at what Breeze's network may look like within the first few months of launch.

Let's first talk about Breeze's timeline. In the filing, Breeze makes it incredibly clear they are requesting an expedited application process.
"Breeze requests that this application be processed by the use of expedited non-hearing procedures. The use of expedited procedures will serve the public interest by facilitating the introduction of Breeze's innovative services as soon as possible" (pg 6)
The expedited request should not come as a surprise. I assume that any application would request something similar, but Breeze has assembled a highly experienced group of professionals (we will touch on this later). An expedited application request should not come as a surprise. In April, Breeze will start taking delivery of their sub-leased E195s (~2 per month) from Azul and by April 2021, their C Series aircraft will also begin to be delivered (1 per month). (page 122).  Once the aircraft start to be delivered, costs can escalate quickly.  As soon as aircraft expenses start being accrued, Breeze will see a 10% increase in total operating expenses (this excludes capital expenses associated with aircraft delivery and inductions).


So, if Breeze hopes to get off the ground quickly, when are they expecting to launch the airline? Based on their provided financial data (pages 131 & 143), I believe they are planning to get Part 121 approval around August 2020. From there, Breeze is expecting to use its aircraft to operate charter missions for 1-3 months before the scheduled service would begin. This could mean schedule service would start around November 2020.

While the timeline may seem aggressive, when JetBlue filed their formal application on April 30, 1999, it took just over four months for their application to be approved. The application is only part of the process. If you read Blue Streak, you are well aware there was a significant amount of background working ongoing before the application filing. Among many items, it appears JetBlue (New Air Corp) filed for JFK slots before submitting their Part 121 certificate.  


Turning towards the growth front, it appears Breeze's capacity deployment might be more modest than JetBlue's original launch. In trying to compare apples to apples, I took a look at JetBlue's system block hours from their initial start and compared it to Breeze's scheduled service launch. With this, we can see Breeze's growth is at least planned to be much more modest than JetBlue.


This might be a fallacy to compare the system block hours between the two, however. Breeze's application makes it clear they are looking to operate with low utilization values. In all the schedules provided by Breeze, the scheduled service utilization barely cracks seven hours. JetBlue, on the other hand, likely had a much higher utilization with their fleet. (Note: I do not have OTP data back that far to calculate utilization for JetBlue). 

But where exactly is Breeze likely to operate? There are a lot of hints in the application. However, it is important to remember, in the deregulated airline industry, once a carrier receives their certificate, there are not many limitations preventing the airline from picking up their operation and moving to the other side of the country. I say this as the next section is entirely my game theory of where I would fit the pieces of Breeze's application to fit their discussed strategies, maintenance, and schedules. Exactly how Breeze constructs their network may (will likely) be materially different. 

Here are the guardrails for Breeze's network:
  1. Mid-sized markets (pg 2)
  2. Underserved markets without nonstop service (pg 5)
  3. Line maintenance in ISP and/or other locations as the carrier grows (pg 6)
  4. The first four cities and three markets will be leisure north-south markets (pg 6)
  5. Cities will be attractive for secondary leisure markets or second homeowners (pg 6)
  6. The first aircraft will be E190s
There are two areas in the application we can focus our analysis on guessing Breeze's phase one service offering. First, Breeze offered their schedule with the cities covered. Each route has a provided block time associated with it (pg 134).


Next, we can turn toward their projected system-level metrics (pg 149). These projected statistics begin as soon as the carrier starts its charter services. So we have to skip to month four when we expect Breeze's scheduled service to begin. Based on this information, we would expect the average stage length to be 1,039 miles. 


For all of my analyses, I anchor ISP as at least one point on their network. Why? Well, here's what Breeze states in their application regarding maintenance: 
For FAA certification, Breeze will conduct line maintenance at its facility in Islip,  New York (“ISP”) and has contracted with Embraer in Nashville, Tennessee for heavy maintenance. As the route system grows, Breeze will use a mix of contract and in-house maintenance providers. At all times maintenance will be conducted in accordance with Breeze’s FAA-approved maintenance program.

Line maintenance is a defined term by the FAA. From one of the FAA's web-based training modules:  
Includes routine and non-routine maintenance, bench checks, calibration, and repairs accomplished in support of day-to-day aircraft operations.

In theory, a line maintenance station would likely be on the carrier's network to route aircraft through for routine maintenance. This, however, could be the fallacy in my analysis. Breeze does state they could use a mix on other providers, which could be in any of their cities. While unlikely, Breeze could plan to use their charter division to route aircraft to/from ISP for maintenance. 


Next, I focused first on block times. Block generally points to the route distance. Using the DOT on-time performance data, we can see how the E190s are currently being operated today. Typically, we see block times include 15 minutes for the aircraft to taxi out and 5 minutes to taxi in as part of their block times. 


Once we removed the assumed taxi out/in times, we can then plot the flight times vs other observed E190s flights. Note, these are directionally average, so east/west flying shows as their average. Using this data, we can estimate the route routes would be similar in stage length ranging from around 970 miles to 1,090 miles. 


There are then three different options you could see Breeze deploy in phase one. This is where the real game theory starts. As a planner, it is crucial to understand the company's strategy to make tactical decisions. Is the company looking to concentrate at one leisure destination, one origin market, or two markets in the north and two in the south? 

My initial impression on the application was ISP would be a focus city, but this was quickly proved wrong. A very reasonable service pattern can be developed matching the block times and the 1,039 mileage statistic. However, service from ISP to MCO, TPA, and MIA area all competitive markets out of ISP, which Breeze stated they would avoid. Further, none of Florida cities would likely be considered "secondary leisure destinations". 


Next, I examined the potential of Breeze turning all flights from a single Florida city to ISP and two other Northeast cities. In total, there were roughly 490,000 route pairings for three markets to/from any Florida airport with service to any Northeast market with service. We further isolated these pairing where ISP had to be a turn and the mileage had to equal 1,039 average stage length. There were still 55 possible route pairing. 

However, when I isolated to secondary, destination airports only, one interesting solution popped up; the potential of turning on Sarasota. In Breeze's application, they state they expect to continue to grow their service in non-competitive routes. Until recently, SRQ has not seen the growth other secondary Florida markets have. That is until Allegiant started operations in 2019. And it is important to remember, many of the senior leaders at Breeze are former Allegiant executives. 


The problem I have with this solution is State College (SCE) is not a medium-sized city that Breeze describes as their target. 

So I took a look at the phase two expansion schedule which, to me hinted at a two north/two south split. Why? In the narrative, Breeze states:
In early 2021, Breeze plans to introduce more service east of the Mississippi river with flights from existing destinations to points in the Southeast and mid-Atlantic region (pg 131).


During this time, Breeze's statistics pages show only two new destinations are planned to be added. Since the mid-Atlantic is not in the central timezone, it is safe to assume all these time zone crossings would tie to the new southeast city. 

If we assume the aircraft 1 & 2 are the original service, we can expect the new southeast city would connect to at least two, maybe three Florida markets since the application continues to reference secondary leisure markets. While I am not going to speculate on the southeast city, phase two caused me to recalculate my view on phase one. Why? Whatever the new southeast city is, it appears that they should expect roughly two flights a day and it is not likely a two hour block time could easily reach the northeast from a southeast city. 

So where should we go from here for additional guidance? For those paying attention this week, the Department of Transportation finally released the Small Community Air Service grants. While I often believe many of these applications are pipe dreams, one unawarded proposal caught my eye. New Haven, Connecticut submitted a proposal that specifically called out Breeze Aviation in their application which was sent back in July.




If we believe New Haven's application to be as serious as HVN attempts to demonstrate, phase one could be a little more clear, especially if we believe in a two-city north and two-city south set up. In their application it is clear to me they have had serious conversations with the carrier. 

An HVN-SRQ, ISP-SFB, and ISP-SRQ would meet all of the objectives set out in Breeze's application. The only thing that might not wholly reconcilable is HVN's assumption of A220 (C-Series) service versus Breeze's initial E190 fleet. 


Regardless of precisely what Breeze ends up flying, it is always exciting to see additional nonstop route offerings to cities that have been reduced to just hub service over the last couple of decades. 

Wednesday, February 19, 2020

No 2/18 Post

For those who showed up today expecting a new post, thanks, but there's no post today. I took an R&R week to take care of some family business and flip through Breeze's application. There are a lot of interesting tidbits which we will discuss next week! 


Wednesday, February 12, 2020

Oklahoma City: No Longer a Fly Over City

For those that know me professionally, I lived in Oklahoma City for nearly three years. In the first year and a half, I was commuting mostly daily between Dallas and Oklahoma City by air or by car. In April 2018, Southwest reduced DAL-OKC from 4x to 3x which, for a commuter, made daily commuting difficult. For the next six months, I drove weekly to Dallas in a crashpad. I would finally throw in the towel in September 2018. And yes, if you were wondering, it was my team that decided to reduce DAL-OKC (which I entirely supported).

This post, however, is not about me. Instead, I want everyone to understand why I am digging much deeper into Oklahoma City. One of my biggest pet peeves as a planner and a general aviation dork was airlines and planners focusing on what is currently sexy. Most of my followers may consider Oklahoma a flyover state. Cities such as OKC often do not get the headlines on national releases nor the attention they deserve, but it appears at least one airline is seeing growth opportunities in Oklahoma City.

For fourteen years (2004-2017), OKC seat capacity was mostly flat outside high blimps in 2008 and 2012. However, in 2018 and beyond, the city has returned to a positive seat growth rate, which we will discuss shortly. During the same time, however, scheduled commercial flights into and out of the airport were down 17.5% (-1.4% annual growth rate). Industry gauge growth trends largely offset the decline in flights out of the airport.



While the decline in capacity could be seen as alarming, OKC's flight decline can be explained by Delta's network realignments following their DFW and post-Northwest merger hub closures. In total, Delta reduced its footprint in the city from 24 daily departures in 2004 to 10 daily departures in 2017.


While seats and originating passenger growth largely remained stagnant, the Oklahoma City economy did not. Strength in the local Oklahoma City economy helped drive airline revenue originating from Oklahoma City on relatively flat capacity growth. This revenue growth came almost entirely from higher fares and excludes baggage fees or other ancillary product fees.



A city with a strong economy and growing revenue on flat passenger growth, to me, is indicative of an airport needing new service or fare stimulation. This may not always be the case at every airport; however, here in Oklahoma City, it appears to be at least one of the indirect drivers to recent capacity moves.

After years of stagnant seat growth at OKC, American began to increase their OKC footprint. American's average daily seats between 2017 and currently scheduled 2020 seats are forecasted to be up 41%.  During the same time, American started or has announced nonstop service to all of their previously unserved hub cities. For those counting, yes, I include JFK and LGA as part of a single metro even though the airports serve very different missions on the AA network.



American's impressive growth within the city started in early 2018. Phoenix and Philadelphia were added to the city within two months of each other, followed by new DCA six months later. Now in 2020, American has rounded out its hub service with new service to Miami and New York. It should not be lost that even with American's growth, total seats by all other carriers have primarily remained flat. This could means other carriers are not seeing negative pressures from the additions to warrant capacity reductions. 



With service to all nine of American's hub cities, Oklahoma City is now one of only nine American Airlines non-hub cities (4% of all cities) that has nonstop service to all hub cities. OKC is also batting far above its weight to get here. Of the eight other cities with full hub city coverage, the average American Airlines daily departure count is over twice as large as Oklahoma City. There are 30 other American cities with more daily departures than OKC who do not have full hub coverage.



While Oklahoma City should celebrate this accomplishment, I do not believe all is well on the home front. When I worked in the industry, I repeatedly preached for communities to use their service or lose it. To me, it feels like the Oklahoma City community better start using the Philadelphia service or they are going to lose it.

What do I mean? Well, let's first take a look at all of American's flown performance for the last year. Unit revenue and load factors on the PHL flight have been soft since its launch.



Soft load factors dragged on the OKC-PHL's revenue performance during the 2018-2019 winter quarters with load factors dropping as low as 50% during the first quarter of 2019. American has quickly reduced the route throughout 2019 and early 2020 to just a single daily frequency during the off-peak months. During the second and third quarter, the performance of the route, aided in part by capacity reduction, performed near system-level RASM target for the route. 

While published 2020 schedules show second and third quarter capacity returning to twice-daily service, given the historical revenue performance, I wonder if this will hold. After I developed the graph below, American reduced their May OKC-PHL route back to a once-daily flight pattern. 




To complicate matters, American's new LaGuardia service will put further pressure on Philadelphia’s performance. Philadelphia’s top flow markets include Boston and LaGuardia, and the new LGA service duplicates itinerary options.    

While most of Oklahoma City's love has come from American, there appears to be some experimentation by other carriers to non-traditional cities. After Southwest departed the DAL-OKC market in November 2019, they backfilled the reduction with increasing Houston, Phoenix, and new daily service to Nashville. (Note: These capacity discussions and decisions were had after I departed Southwest).

Frontier has also experimented within Oklahoma City. During late 2017 and 2018, the carrier launched Orlando, San Diego, and San Antonio. Today, only Frontier's Denver and Orlando service remain, however, the experiment did point to potentially untapped demand in the San Diego market. When Frontier entered the OKC-SAN market with 3-4 times weekly service, they were able to stimulate the market to nearly 160 PDEWs. Interestingly, OKC-SAN was also part of the failed ExpressJet experiment. ExpressJet operated 1-2x daily service during almost the entirety of their experiment.


Other carriers such as Alaska and United also have increased their investment in the city. At the end of last year, Alaska up-gauged its E175 service to A320 service (doubling seats in the market). United is also investing additional capacity in the market with flight increases to Dulles and Denver. This is on top of United's Chicago product enhancements with the CRJ-550 over the CRJ/ERJ flights.

It is great to see Oklahoma City getting the attention it deserves. We are clearly batting above our size with American's hub service. But it is clear. Some risks have to be addressed if we hope to keep the impressive product offering in place. We need to support all of our carriers, whenever possible, to make sure the service that we have at our airport stays and grows with our city.

-------------------------------------

Thank you all for swinging by the blog. There will not be a post on 2/19 (unless an airline does something interesting). The following week, I am planning a deep dive into Breeze's DOT application. 

Wednesday, February 5, 2020

Feeling United's Paine: United Couldn't Compete with Alaska's Service

The tale of commercial air service in Paine Field is about as old as time. Since the late-2000s, various airlines have expressed interest in operating commercial service from the airfield. However, air service at the airport was always going to be a challenge. The airport lacked an existing public terminal and the local community was combative to the idea of commercial air service at Paine.

It wasn't until Propeller Airports built a new terminal and a lawsuit or three resolved before airlines could announce their service. Once everything was in place, Alaska quickly announced service in May 2017, followed by United over the summer, and Southwest early winter 2018. Southwest would later bow out of the running. (Really boring, but informative read regarding the history)


With so many airlines trying to access to Paine, demand for the two gate facility quickly exceeded the planned capacity of 16 daily flights. By April 1, 2020, the terminal saw 24 daily departures on just two gates. Neither of the carriers really adjusted their schedule down on off-peak days. Offhand, I can't think of another airport in the US that is turning 12 flights per gate every day. 


With 12 flights per gate per day, the operation was tightly scheduled. This could lead one to believe that the airport would be operationally challenging. This really does not appear to be the case. Since the services started, airlines operated at the airport with roughly 88% on-time rate. But, just because the airport operated on time, does not necessarily mean the airlines could schedule to fit their commercial network, which we will get into here shortly. 



Alaska's commercial strategies at PAE seem to be pretty clear: operate in top Seattle originating markets within range of their respective regional fleets. Alaska with their 18 flights blanketed seven out of the ten originating markets. Most of these markets fit nicely into Alaska's growing California network. 


United's commercial strategy seemed to follow Alaska, just via their DEN and SFO hubs. With most of Seattle's originating traveling to California, Vegas, and Phoenix, you can see why United placed most of their capacity into SFO. In theory, if the flights were well-timed and connected to the SFO bank structure, we should expect to see robust connections to all of these top markets. But we didn't.


Each of the PAE-SFO markets only connected to 4-5 of the top Seattle originating markets. The PAE flights often arrived 30 minutes to an hour after their ideal connecting bank into SFO. At a normal airport, this is often easy to fix, just adjust the departure time. However, with PAE's 12 turns per gate, adjusting departure times was likely out of the question unless Alaska wanted to play nice. But with no upside for Alaska, why would they play nice? With no ability to adjust their schedule, United's SFO schedule stayed nearly identical since they launched service in late March 2019. 


The lack of quality itineraries over SFO can be seen in the passenger data. PAE saw sizable passenger demand in its top markets. These markets largely mirrored the Seattle originating passenger demand as well as where the airlines allocated capacity. However, since United did not have quality itineraries via SFO, especially to Southern California, United only saw single-digit PDEWs to these top destinations.


Let's not forget, United was also going head-to-head in San Francisco versus Alaska. Since United operated four daily flights vs Alaska's two daily flights it would be logical to assume United would carry twice as many passengers, however, this is not the case.

When we look at the SEA-SFO market, the market is evenly split between originating in SFO vs SEA. However, in the PAE-SFO market, the market skews towards PAE 60-40. This should not be unexpected since PAE is a new, mostly originating airport. This was ultimately a huge disadvantage for United. While the two carriers equally split the SFO originating market, Alaska nearly doubled the passenger demand on the PAE side of the market. 


With the local PAE-SFO market favoring Alaska and United's inability to build connections to top PAE markets, United's revenue performance on the route suffered significantly. My models show the route was roughly 60% below what United would expect on a similar staged market.


Now, I caution anyone from using the graph above in absolute terms. I debated for quite some time about even including the chart given how negative PAE-SFO was. It is important to remember this information comes from modeled data. Airlines do not publish their actual RASM production figures. However, we can see the direction and magnitude of United's performance in the market. It was negative. Really negative. This likely explains why United killed the route even before its first birthday. 

Astute readers will likely point out that DEN is also performing below system production. While this is true, the data available to us only includes the first six months of service. It is common for new routes to underproduce revenue during their first couple of years as a market matures. So while the market might appear to be a red flag, I say it is a wait and see. Why?

DEN does not have the disadvantages that we saw with SFO. Currently, Alaska is not flying to nor over DEN. Even if we see Alaska attempt to operate to or beyond DEN, their ability to put substantial pressure on United would be much more muted than SFO. This would not preclude Alaska from operating DEN as it is still a large originating market for the Seattle area.

So what will happen with the available gate time saved from United reduction? It is hard to say. The facility was built originally for 16 flights. I have to wonder if Propeller would like to dial capacity back for a bit, but they are an investment group and investment groups typically aim to maximize returns.

If Alaska were to pick up the gate times, it would be reasonable to see them in DEN as discussed above. It would be equally possible to see them add depth to existing markets or new service to a secondary Basin market.

I would doubt that Delta would be interested in operating three flights into the airport, especially if they are locked into United's old times. As these old times likely would not fit operationally or commercially into LAX or SLC. In the unlikely event Delta was to jump in, I would not be surprised to see them operate a non-hub such as Vegas really to just be a thorn in Alaska's side. 

There's also Allegiant if the costs are low enough. They had expressed interest in the past.

There's plenty of gamesmanship that could be played here. But ultimately, we will all just keep an eye on the schedules and see who, if anyone, takes the gate times in PAE.  

Wednesday, January 29, 2020

How much has Frontier's network changed?

This week, I decided to take a step back from hardcore revenue analytics like we have seen in the past few weeks. While revenue is one of the most important metrics in determining the health of a market, I thought I’d switch to an operational and schedule analysis this week. So this week we’ll take a look at the Department of Transportations on-time performance report. I’d argue this dataset provides some of the most insightful views to the operation of an airline, but is wildly underutilized.

Why? If you are not one of those data nerds that often pull raw datasets yourself, you likely assume that the OTP information basically tells you if a flight arrives on-time or not and that is about it. But in today's post, we will use OTP information to see aircraft routing, aircraft metrics, fleet used for service, etc. There is plenty of other metrics that could also be estimated from the dataset including operational spares and turn times.

While these might not sound as sexy metrics, consider this. If you can examine an airlines' aircraft utilization to a few hundredths of a point and have a good future fleet source, then you can have confidence when estimating how much capacity a target airline may have left by fleet to deploy at any given point of the schedule.

Today, we will combine both the Frontier schedule and on-time performance information to see exactly what has changed within the Frontier network over time and what has stayed the same.

First, let's take a look at aircraft utilization. As you might expect, ULCCs push their assets longer to spread fixed costs across more ASMs. However, if you expected Frontier to be firewalling their fleet compared to previous years, you are going to be surprised.



When we look at the Frontier operated schedule from the DOT's OTP report, we can see Frontier's April schedule has been pretty consistent in their fleet deployment since 2007. There is a noticeable decline in Frontier's utilization in 2011, however, it should be noted during 2011, Republic was handling a significant share of Frontier's flying. Republic was removed from today’s data.



But what about other ULCC carriers? How does Frontier stack vs Spirit or Allegiant if they have not increased their utilization? Frontier and Spirit are largely identical in their deployments. Allegiant, interestingly, has a hard cutoff around 13 hours which you do not see at Spirit or Frontier.



If you are an inflight crew, you likely know this is only a piece of an aircraft's daily story. While I do not currently have turn time (time on the gate). Turn time and total aircraft operating can be decoded from the OTP data, however, I have not built the model for it at this time. In time. I promise. Maybe.

In addition to core aircraft utilization metrics, we can also track tail number specific movements within airline networks. Here's Grizz (N227FR) on 4/19 - 4/22.


By the way, if you are an old hardcore Frontier fan, you will be disappointed to know Flip and Larry have been retired. They were A319s living in an A321 world.

But beyond tracking our favorite fictitious Frontier tail, we can glean useful information from tracking tail numbers. Based on Frontier’s OTP data, starting in 2014, it appears we can see two distinct base schedules per year, at least in terms of aircraft routings. Really, these appear to be highs and lows for aircraft routing thru Denver. There appears to be a distinct Denver high (April thru November) and a Denver low (December thru March).

While it might seem like a general, "oh that's pretty cool", there are real-world implications here. Aircraft have to be maintained and crews need to be based. I’d be interested in learning if crews are moved to non-DEN bases during this time or if non-DEN crews are pushed harder during higher non-DEN routing periods. The chart also shows that Frontier is no longer directly dependent upon Denver to route their aircraft.

Additionally with this data, we can begin to see how many aircraft frames Frontier is deploying within their operation over any set period of time. This is different from frames within the fleet since a percentage of frames will be down for scheduled maintenance among other events.


Finally, while we often see on the forums jokes regarding Frontier's "dartboard" approach to capacity planning, what we find is what appears to be a 40%, 20%, 20% strategy. 40% mature routes, 20% maturing routes, and 20% yearling routes.

When we look on an absolute basis, most planners will see Frontier's declining absolute maturity within their routes. In other words, Frontier has cut deeply into their maturing frequencies that have existed for 3+ years for yearling frequencies (< 1-year-old) or maturing frequencies (2-3-year-old routes).


But the chart is somewhat misleading. The drop in mature frequencies is more a function of Frontier's shift towards more breadth and lower frequency within its existing network.


Absolute route counts have skyrocketed since 2016 as Frontier has diversified its network. During the peak travel demand period, Frontier operated nearly 350 routes, albeit at very low frequency. 


In relative terms, Frontier has invested 40% of its routes towards mature capacity (> 3 years old), 20% towards maturing routes (2-3 years old), and 20% towards yearling routes. This appears to flex some, but it appears to be a good rule of thumb. 


Finally, as we all have come to know Frontier over the last few years the airline has become known as the king of churn. We have all seen the press releases where Frontier has started one billion new markets across a million new cities. They are often announcing so many new cities and routes, they have to deploy their general council to announce route on their behalf. (Rant: If you have to use your general council for route announcements, you are either too thinly staffed or announcing far too many routes).

What we often do not hear is Frontier's "seasonal" service that never seems to return


Frontier will be interesting to watch over the next few years. With the amount of new frames coming from Airbus, I do wonder if the churn rate is sustainable. Finding new homes for airframes gets increasingly more difficult when the amount of frames continue to increase

In the near term, it sure feels like we are days or weeks away from another massive expansion from being announced, just in time for summer 2020 travel. 


Wednesday, January 22, 2020

Behind the Cuts: Examining Why JetBlue is Cutting Long Beach

In the early winter of 2017, a small group of us Southwest Airlines network planners were gathered in a conference room. We were given our marching orders. The 737 classic fleet was headed into retirement on September 29th, moved the 30th, and the MAX would join the fleet on October 1st. With the reduction in airframes, we were going to have to reduce frequencies and eliminate routes to keep the network operational with balancing commercial constraints. After months of debating, it was time to decide.

Cutting sucks. There’s a good chance you had to eat crow after defending the poor performance as you believed the route just needed to mature. But it was finally time. Due to aircraft constraints, pressure to get network performance up, or this route’s performance just dragged for too long without meaningful long term strategic value, today is execution day for this route. Regardless of the reason, someone is going to be pissed at your decision.

In my time as a planner, I heard it all:
  • This is the most short-sighted decision ever made 
  • I am going to have (name the ULCC) enter the city and they will take all your customers
  • You do not understand the city/market. 
  • You are lying
  • Well, I'll be calling (name the Leader) and they will be reversing this decision
Once, I heard three of these in just one 20 minute meeting.

I bring this up not for sympathy, but for understanding. I was fortunate enough to have and still maintain great industry relationships. I bring this up as JetBlue's network planners are likely experiencing this exact same reaction both internally and externally with their decision to cut Long Beach. JetBlue employees are likely being displaced or their hours reduced and front line employees may be feeling left in the dark. However, JetBlue has been struggling in Long Beach since 2008 and are executing on their fourth attempt at a city recovery.

The reality is there was little JetBlue could do with Long Beach that would make commercial and financial sense. Its commercial teams were left with no great options. It all likelihood, network planners could have been handicapped by constraints to delay what many may believe as an inevitable outcome.

With that digression, let's dig deeper into JetBlue's history and ultimate decision to reduce its operation to just 15 daily flights.

In August 2001, JetBlue entered Long Beach with twice daily JFK service, which was quickly accelerated to three times daily. After JetBlue's initial launch, B6 rapidly built up its position at Long Beach in response to their initial success in the market. This continued until 2008 when they topped out on slots and performance.


Depth to Breadth 

While the initial service into Long Beach appears to have been pretty successful for JetBlue, the performance love affair was shortlived. The Great Recession took a healthy bite into JetBlue’s revenue and in 2008 Long Beach's unit revenue production turned negative. Of the 14 routes, 12 were performing below the system RASM curve.



To counter the declining performance, JetBlue had to change something with the Long Beach service. Over a few years, JetBlue slowly began to transform Long Beach's network. The carrier believed their path to system returns lay with more markets at a lower frequency. The planners began to reduce higher frequency markets like IAD and JFK for lower frequency markets like AUS, PDX, SEA, and SFO.


Seasonalization 

Even with the breadth approach, the network performance in Long Beach still dragged on the network. In 2011, Long Beach's unit revenue performance really took a turn for the worst. At this point, most routes were double-digit negative compared to JetBlue's RASM curve. The only routes that appear to be working were back to the East Coast. All other Long Beach markets really pulled on the JetBlue network.


Without immediate action, this level of revenue performance would become quickly unsustainable. In 2012, JetBlue began aggressively cutting off-peak flying. Typically, JetBlue would reduce its seasonal schedule by 5-10% off the peak. However, in 2012 thru 2014 the carrier reduced its off-peak schedule by 20% of peak schedules. This approach trimmed JetBlue's yearly departures by as much as 10%.




Save the Slots 

In 2014, the Basin began to heat up as carriers started to fight for limited airport space. All secondary airports in the Basin, excluding ONT, are constrained by some type of noise controls. Orange County and Long Beach have slots, while Burbank has gate caps. As Orange County and Burbank started to fill, carriers began to shift their attention to Long Beach's limited slots.



Appetite for Long Beach began in earnest in late 2015 when the airport announced it would lift the noise cap to allow nine more flights into the airport. Of the nine new slots, JetBlue received three, Southwest four, and Delta two. Later, Southwest would use JetBlue’s underutilized slots to fuel additional growth in the city.

As is often the case with JetBlue’s focus city, these competitive incursions initiated a response from JetBlue. In August 2016, JetBlue announced its plans to fully utilize all of its slots. This shot JetBlue's departure count from ~23 daily flights in 2016 to 35 flights in 2017.


Plowing this much capacity into subpar routes is one heck of a way to start a turbine with cash. During this period, most routes moved further down the revenue production ladder. Long Beach now saw a sea of red as most routes were producing 20-40% below the system RASM curve.



The opportunity cost associated with operating these routes was astronomical. If these flights were deployed to routes that produced system average returns, JetBlue would have seen roughly $80M in additional ticket revenue. This value excludes all ancillary fees.



Now, I want to be extremely clear. The graph above does not state JetBlue lost $80M in Long Beach during their slot utilization ramp-up. Rather, the graph is an opportunity cost for flying underperforming routes. No one can say exactly how much money JetBlue made or lost in Long Beach except JetBlue. Please see my cost tangent if you receive airline "profitability reports". But the graph above can be used to understand the direction and magnitude of JetBlue's performance.

Slot Saving is Expensive 

After what appears to be a massive drop in Long Beach performance, JetBlue tossed in the towel. Twice. Kinda. In September 2018, JetBlue reduced flights by 30% but would not return the slots back to the airport until months later. Interestingly, the carrier decided to only exit FLL and removed frequencies across a variety of underperforming markets rather than more significant market exits.

This is the one time I’ll fault JetBlue’s logic in this entire process. Long Beach was bleeding cash and the first round of reductions really did not feel in line with the city’s performance. The first round of cuts announced by JetBlue were focused on stopping the hemorrhaging, but not on truly getting the city healthy.


The reductions in capacity greatly increased the performance of the city in terms of year-over-year unit revenue production. However, year-over-year RASM growth is largely meaningless when JetBlue's 2018 route performance is factored in.

This is where we found JetBlue until last week. Most routes continued to underperform their system RASM curve. Intra-California routes to SMF and SJC were particularly terrible. This is in line with JetBlue’s market exits.


Will Cutting to Profitability Work? 

After all the aggressive protectionism then retreating, do I think JetBlue may finally be stable within Long Beach? I give it a big maybe.

When we examine JetBlue as a whole, rather than just focusing on Long Beach, after the latest rounds of cuts, while still underperforming the system, Long Beach could be in much better financial position than it has been in quite some time. The routes that remain are generally within the tolerances of acceptable unit revenue performance. Don't get me wrong, the remaining routes are not stellar, but the entire system cannot be above system averages. Averages don't work that way.

Another thing to keep in mind, following this round of cuts, Long Beach is unlikely to have the JetBlue’s worst-performing routes attached to it. If JetBlue is looking to harvest and redistribute aircraft within their network, it appears they would have a healthy selection of other poor performers to sift through first.


Finally, there has to be consideration given to the logistics of routing crews out of the Long Beach crew base. Further cuts would make it nearly impossible for JetBlue to efficiently move crews. The costs associated with dislocating LGB based crews are might be prohibitive given the production of the remaining routes compared to extremes such as shutting down a flight crew station.


It is still very possible that we may see a reallocation of flights between the remaining markets, but I honestly do not expect any more earthshattering moves at least in the near term.

The only wildcard for Long Beach is competition, but I would not expect immediate meaningful competitive growth. Kate Kuykendall, Long Beach's public affairs officer, stated the airport has not received an official notice from JetBlue regarding their intention to relinquish slots at the airport. The airport is well aware of the announced reductions, however, until JetBlue returns slots, the airport cannot start the process of reallocating them to other carriers.

According to the Long Beach airport, JetBlue has 24 permanent flight slots. These slots have minimum usage requirements which JetBlue will dip below during the 3Q2020. This could mean, without JetBlue voluntarily surrendering their slots, it may be the 4Q2020 or 1Q2021 before slots are reallocated from this flight reduction. If the 14.8 average daily flights are the baseline for JetBlue's new slot allocation, we should expect to see JetBlue be allocated 17-18 slots, which free 6-7 slots for competitors.



JetBlue's potential delay in relinquishing slots should not come as a surprise. When JetBlue reduced capacity in September 2018, the carrier did not return its unused slots until the spring of 2019. Further, in New York, I suspect there are a lot of hard feelings following increases in curfew fines, increases in slot utilization requirements, and the city walking away from the international terminal. I do not expect JetBlue to play along nicely with its slots.

By the time the JetBlue surrenders the slots and the airport executes on the allocation process, we could be looking at early 2021 before Long Beach sees carriers backfill in the airport. In the more near term, the airport is already in the process to allocate three new noise supplemental slots. This process should be completed in February.


According to CrankyFlier, three carriers remain interested in additional Long Beach slots: Delta; Hawaiian; and Southwest. Based on JetBlue's remaining network, I believe there is limited exposure that you can reasonably see these three carriers overlapping with JetBlue.

If Delta were wanting to increase pressure in SLC, they already have the opportunity with their last slot allocation which they sent to Vegas. On the Southwest network, a reasonable person could argue for additional Las Vegas flights or maybe a long shot at Austin service. Otherwise, I’d expect the new capacity to be allocated away from JetBlue’s Long Beach network by these carriers.

Barring Alaska jumping into the mix, which is possible, but not likely, SEA, PDX, and SFO are unlikely to see additional competitive pressures. With Alaska exiting Long Beach in 2015, I do not see a large possibility of them wanting to relaunch the city.

This should for the time being bring stability to JetBlue in Long Beach.


Delta's Push to Drop Small Cities