How does your local airport's number of passengers flying through compare to how many that would be expected? I did a regression analysis, that I can publish here online. The following data shows the expected number of passengers per year flying through an airport and compares it to the actual number. Negative numbers show that there are fewer passengers flying through the airport than predicted which means airlines should probably add service to get those numbers up. The results are from 81 Metro Areas in the United States (100 largest airports in the country). The data calculates the amount of traffic at an airport based on the following explanatory variables:

Metro Population
# of Airports in a Metro Area
Regional Per Capita Income
Whether it is a vacation destination or not
Whether it is a hub of a major US airline
Location Correction Factor


The data shows the following:

Underserved US Metro Areas:

*New York
*Philadephia
*Washington DC
*Detroit
*Phoenix
*Ontario
*Seattle
*St. Louis
*Tampa
*Pittsburgh
*Cleveland
*Cincinnati
*Sacramento
*Kansas City
*San Antonio
*Columbus
*Indianapolis
*Norfolk
*Providence
*Austin
*Milwaukee
*Jacksonville
*Mmemphis
*Louisville
*Richmond
*Hartford
*Oklahoma City
*Buffalo
*Birmingham
*Salt Lake City
*Rochester
*Tuscon
*Tulsa
*Albany
*Dayton
*Omaha
*Grand Rapis
*El Paso
*Greensboro
*Sarasota
*Little Rock
*Syracuse
*Charleston
*Colorado Springs
*Wichita
*Fr. Meyers
*Madison
*Des Moines
*Jacksonville
*Pensacola
*Spartanburg
*Maui
*Kauai

Overserved US Metro Areas

*Los Angeles
*Chicago
*Dallas
*Houston
*Miami
*Atlanta
*Boston
*San Francisco
*Minneapolis
*San Diego
*Baltimore
*Denver
*Portland
*Orlando
*San Jose
*Las Vegas
*Charlotte
*Nashville
*New Orleans
*Raleigh-Durham
*Honolulu
*Albuquerque
*Boise
*Spokane
*Reno
*Anchorage
*Savannah

I'm keeping my actual equation and model confidential. You may request it and data if you desire. Please note that this model is 85% correct, so there is error. Here is the regression predicting passenger traffic in each metro area in the United States.

Negative Numbers= Fewer passengers fly through the airport than expected

(Cllick on Link for High Resolution Version)



Feel free to post your response and critiques.

The New England Regional Airport System Plan revealed that PVD leaked about 40-45% of their catchment area to BOS, and had the highest leakage rate og all of the larger new england airports (obviously HVN and ORH had the highest rates for their cactment areas)

Hopefully this will be reduced in the 5-10 year timeframe with the eventual addition ot FL and B6, and perhapp Ryanair, plus the amtrak/MBTA station at the airport, there will always be SOME leakage of pax from the catchment area, but PVD is double each of the other major airports.

Quoting RL757PVD (Reply 1): The New England Regional Airport System Plan revealed that PVD leaked about 40-45% of their catchment area to BOS, and had the highest leakage rate og all of the larger new england airports (obviously HVN and ORH had the highest rates for their cactment areas)

According to my calculations, Providence is one of the most underserved destinations. It has 44% less traffic than would be expected. That's a big difference.

Also, I can probably calculate all the percentages if people want me to.

Quoting RoseFlyer (Reply 2): According to my calculations, Providence is one of the most underserved destinations. It has 44% less traffic than would be expected. That's a big difference.

Eventually BOS will check in full, and PVD will eventually get more options, so hopefully over time it should ballance out

What PVD should have now that they dont currently have IMHO:

RDU 3x RJ (AE)
ATL at least 50% more Delta seats, at least 2x mainline, plus 3x Airtran
DEN 1x (UA/F9/WN)
FLL 1 additional flight (total of 2)
PBI 1x
RSW 1x (seasonal 2x)
IAH/HOU 1x
DFW 1-2x
STL 2-3x RJ (AE) or 1x WN

Nothing outrageous, and actually PVD has had most of those, but AA and DL did the damage on RDU, ATL and DFW for BOS. NK left behind huge pax #'s for FLL and RSW. If that list only DEN, PBI and STL havent been done.

If/(when really) FL comes to PVD they can take care of ATL, FLL, PBI and RSW, for the most part with zero competition, since FLL is big enoguh for both them an WN, and DL wont respond much since they care more about BOS anyways.

Hopefully in the next 5 years AA will return RDU and DFW. AA did the biggest "about-face" at PVD after having 4x MD80 for years, the added a DFW, then additng 4x RDU RJ and another ORD (6x mainline 4x RJ) within 2 years they cut it down to 3x E140... WTF is that all about? Clearly not a PVD market based decision, and obviously part of a regional (BOS) plan.

Quoting RL757PVD (Reply 3): Eventually BOS will check in full, and PVD will eventually get more options, so hopefully over time it should ballance out

PVD benefits from being a much easier airport, but BOS isn't that inconvenient. It's also not going through slot control problems like LGA and JFK are, so AA is not really motivated to boost PVD traffic. People appreciate all the frequencies and choices that BOS offers. Still however, there appears to be a larger market than is served. Service should increase.

Hey RoseFlyer, interesting comparison however I have a couple of questions.

Basically trying to figure out why you say "Los Angeles" is over served?

I assume you are using MSA's for the population right? Are you looking at all airports in the MSA?
For instance the Los Angeles-Long Beach-Santa Ana MSA has at 5 commercial airports LAX, BUR, LGB and SNA and PMD which just recently received service again.

So when you say Los Angeles is over served do mean LAX, or the entire MSA?
Also how do correct for the position that LAX ends up being the onestop regional answer for many from San Diego northward towards Bakersfield? In essence it serves a much larger market then simply greater Los Angeles.

Also how do you account for the each areas propensity of travel? For instance geographic and socia-economic reasons cause certain regions to travel more then others. For instance the SF Bay Area has one of the highest propensities for travel in the country on per-capita basis.

Also how do you correct for places that have disproportionate amount of entertainment/tourism travel. Places like Orlando obviously receive many time their population in tourist visitors.

Not trying to throw rocks at you, just curious at your methodology.

[Edited 2007-11-09 11:18:28]

The metro population figure for New Orleans is way off.....it's closer to 1.4 million IIRC.....that combined with the fact that MSY airlines are increasing service puts this survey into the "interesting, but I'm not going to base any decisions on it" pile.

Tom at MSY

Quoting RoseFlyer (Thread starter): The following data shows the expected number of passengers per year flying through an airport and compares it to the actual number.

Very interesting analysis. One question I have, is where does the expected passengers number come from?

Quoting RoseFlyer (Thread starter): Please note that this model is 85% correct, so there is error.

Where does this figure come from?

Quoting RoseFlyer (Thread starter): *San Diego

Considering this airport is running at about an 80-85% load factor and you call it "over-served" you might want to re-evaluate your formula. Based on your numbers, you appear to place far too much emphasis on local traffic/population base. I’ve seen this projection method before. It is incredibly flawed and only partially works when individual mature city pairs are examined, not an entire market and then it only projects the level of originating traffic between the pairs excluding external traffic. For certain, the method only applies to short haul regional transportation, specifically public transportation such as light rail and bus not aviation. Tell Richard Carson (who it came from) I said hello, and that his congestion offset (no growth) analysis is flawed and that he's still an idiot.

[Edited 2007-11-09 11:41:46]

Quoting RoseFlyer (Thread starter): *Pensacola

This city is just one example of several on the underserved list that had more flights to more destinations and the flights were not supported. For example; CO to EWR via JAX; NW to MSP, DTW, DL to MSY/DFW..

Quoting Laxintl (Reply 5): I assume you are using MSA's for the population right? Are you looking at all airports in the MSA? For instance the Los Angeles-Long Beach-Santa Ana MSA has at 5 commercial airports LAX, BUR, LGB and SNA and PMD which just recently received service again.

In the airport traffic figure for Los Angeles, LAX, SNA, LGB and BUR were counted. PMD was not. These are the airports contained within the Los Angeles, Long Beach, Santa Ana metro area as defined by the US census. The populations are all from the 2006 US Census Bureau estimate.

Quoting Laxintl (Reply 5): Also how do you account for the each areas propensity of travel? For instance geographic and socia-economic reasons cause certain regions to travel more then others. For instance the SF Bay Area has one of the highest propensities for travel in the country on per-capita basis.

There is an income effect which is the per capita income of that area. This variable actually isn't very significant in the model, but is incorporated to show a difference between income and travel propensity for passengers in Little Rock Arkansas compared to New York.

I could only use statistics for per capita income since I could not find any other data to model propensity to travel.

Quoting Laxintl (Reply 5): Also how do you correct for places that have disproportionate amount of entertainment/tourism travel. Places like Orlando obviously receive many time their population in tourist visitors.

Yes there is a variable that corrects for tourism/vacation destinations. This is a yes or no variable. Destinations like Orlando, Miami, Las Vegas, Phoenix, etc have this extra variable included.

Quoting Bobnwa (Reply 7): Very interesting analysis. One question I have, is where does the expected passengers number come from?

That is the basis of the model. I took data from 81 metro areas where I had USDOT information on number of passengers per year. I then wrote an equation based on the six factors that I chose to be able to predict the number of passengers. This is where the expected passengers number comes from. I then compared expected passengers to actual passengers to find out if a market is over served or underserved. I will not post my equation or data on here, but will supply it on request.

Quoting Bobnwa (Reply 7): Where does this figure come from?

The 85% number comes from how accurately the expected values related to the true values of passenger counts. Using the R-squared value from the residuals, I found an accuracy of 85%. I can explain more on request.

Quoting Tom in NO (Reply 6): The metro population figure for New Orleans is way off.....it's closer to 1.4 million IIRC.....that combined with the fact that MSY airlines are increasing service puts this survey into the "interesting, but I'm not going to base any decisions on it" pile.

I was surprised by the result from New Orleans as well. The population number that I used is from the 2006 estimate from the US Census Bureau. That was the most accurate number I could find for each city. If you notice though, the difference between estimated and actual traffic for MSY is very close. MSY is about spot on.

However in this model, New Orleans was not listed as a tourist destination. If it was, then that would show that the city is very underserved. It was a judgement decision on my part to make it that way.

Quoting Boeing7E7 (Reply 8): Tell Richard Carson (who it came from) I said hello, and that his congestion offset (no growth) analysis is flawed and that he's still an idiot.

I have no idea who you are talking about. I calculated and formulated my model exclusively by myself with my own methodology.

Quoting Boeing7E7 (Reply 8): Based on your numbers, you appear to place far too much emphasis on local traffic/population base.

Well, that is how the regression worked. I chose a linear regression because it worked the best. When I used a logarithmic variable for population, the R-squared term went down, which means the error increased. So I don't think I put too much reliance on population. I used EVIEWS software to predict the terms.

Quoting Boeing7E7 (Reply 8): It is incredibly flawed and only partially works when individual mature city pairs are examined, not an entire market and then it only projects the level of originating traffic between the pairs excluding external traffic.

Well the model has an 85% accuracy for predicting traffic, which is pretty good, so I don't think that is incredibly flawed. There conclusions are definitely not certain.

Also the idea of being overserved or underserved based on expected traffic might be flawed. I thought it is a logical conclusion. Obviously you don't.

Quoting RL757PVD (Reply 3): Quoting RoseFlyer (Reply 2): According to my calculations, Providence is one of the most underserved destinations. It has 44% less traffic than would be expected. That's a big difference. Eventually BOS will check in full, and PVD will eventually get more options, so hopefully over time it should ballance out What PVD should have now that they dont currently have IMHO: RDU 3x RJ (AE) ATL at least 50% more Delta seats, at least 2x mainline, plus 3x Airtran DEN 1x (UA/F9/WN) FLL 1 additional flight (total of 2) PBI 1x RSW 1x (seasonal 2x) IAH/HOU 1x DFW 1-2x STL 2-3x RJ (AE) or 1x WN Nothing outrageous, and actually PVD has had most of those, but AA and DL did the damage on RDU, ATL and DFW for BOS. NK left behind huge pax #'s for FLL and RSW. If that list only DEN, PBI and STL havent been done. If/(when really) FL comes to PVD they can take care of ATL, FLL, PBI and RSW, for the most part with zero competition, since FLL is big enoguh for both them an WN, and DL wont respond much since they care more about BOS anyways. Hopefully in the next 5 years AA will return RDU and DFW. AA did the biggest "about-face" at PVD after having 4x MD80 for years, the added a DFW, then additng 4x RDU RJ and another ORD (6x mainline 4x RJ) within 2 years they cut it down to 3x E140... WTF is that all about? Clearly not a PVD market based decision, and obviously part of a regional (BOS) plan.

I do agree that PVD could use some more service. No doubt PVD sees leakage to BOS, especially for international travel and West Cost travel.

This may sound crazy to some, but I look at it (Rhode Island and Massachusetts) as one big market due to the clost proximaty to one another. Eventually what would be nice in my eyes is one large airport serving the entire region. This would call for a brand new airport to be built between both Providence and Boston and I believe the entire region would benefit greatly. Similar to how CVG is located in Kentucky, an large airport between Boston and Providence would help the entire region.

In any event, I find the numbers you posted a little off. In the actual field, is that real data? I assume you are talking about O&D to and from the airport and if so, the numbers for BOS as way off. Mind you people living in the New England region, (MASS, RI, NH, etc) have a higher travel rate than the national average. I would also argue that the numbers for PVD or off too.

Quoting RoseFlyer (Reply 10): There is an income effect

Not just income, but geography and access to other forms of transportation also come into play.

For instance a place like Anchorage is nearly hostage to air transport, while a city in the Northeast corridor such as PHL has access to alternate means of getting around to its top O&D markets that avoid planes.

Quoting RoseFlyer (Reply 10): Destinations like Orlando, Miami, Las Vegas, Phoenix, etc have this extra variable included.

Make sure to add Los Angeles into this. Per the Convention on Visitors Bureau, LA saw near 25million visitors of which 4.6million were foreign.


My personal take is that Los Angeles is still not "over served", for the simple fact that while you only consider the straight Los Angeles - Long Beach - Santa Ana MSA.
5 other MSA very much bleed into this including Riverside-San Bernardino-Ontario, San Diego-Carlsbad-San Marcos, Oxnard-Thousand Oaks-Ventura, Bakersfield and Santa Barbara-Santa Maria-Goleta which all look to LAX as regional transportation hub.

I think you structured your model pretty well. Nice idea RoseFlyer.

One question, what is the "location correction factor" ?

Quoting Laxintl (Reply 13): Not just income, but geography and access to other forms of transportation also come into play. For instance a place like Anchorage is nearly hostage to air transport, while a city in the Northeast corridor such as PHL has access to alternate means of getting around to its top O&D markets that avoid planes.

I understand where you are coming from. However I couldn't find any way to model differences like that. They are factored in the residual, but that's all. This is a source of error. I don't have a good model for propensity to travel. If you have any data sources that you can think of, then I'll gladly change my model.

Quoting Laxintl (Reply 13): 5 other MSA very much bleed into this including Riverside-San Bernardino-Ontario, San Diego-Carlsbad-San Marcos, Oxnard-Thousand Oaks-Ventura, Bakersfield and Santa Barbara-Santa Maria-Goleta which all look to LAX as regional transportation hub.

Three of those metro areas are independant and have their own airports in this model. I believe that the Riverside-San Bernardino-Ontario market is a problem in this model since ONT is the only airport in this region, but isn't well served. People from this metro area may travel into other metro areas to fly.

Quoting Laxintl (Reply 13): Make sure to add Los Angeles into this. Per the Convention on Visitors Bureau, LA saw near 25million visitors of which 4.6million were foreign.

Actually, I'm not putting Los Angeles as a vacation destination. This is for multiple reasons. The majority of the reason why Los Angeles is a tourist destination is because it is a big city. This is similar to New York.

The reason for it being a tourist destination is different from Las Vegas, or Orlando or Miami. LAS, MCO, FLL etc are vacation destinations that attract tourists not related to the city population and residents but because of things like beaches, amusement parks and casinos. I feel that Los Angeles' propensity for tourism is not best modeled by being a vacation destination. For the sake of the model, the error term is higher if Los Angeles is a vacation destination than if it is not.

Really interesting stuff for somebody that doesn't really believe in "underserved" or "overserved" markets (I think the ideas are an a.net invention!). I'd be curious to see how you came up with your numbers...

Quoting Flighty (Reply 14): One question, what is the "location correction factor" ?

The location correction factor is for cities on the coast or not on the coast. Cities on the coast tend to have a higher proportion of passengers flying. This is what the model shows. It isn't a very significant variable though. It is a less powerful explanatory variable.

Quoting RoseFlyer (Reply 11): I have no idea who you are talking about. I calculated and formulated my model exclusively by myself with my own methodology.

Your results are nearly identical, +/- half a percent.

Quoting AirCop (Reply 9): This city is just one example of several on the underserved list that had more flights to more destinations and the flights were not supported. For example; CO to EWR via JAX; NW to MSP, DTW, DL to MSY/DFW..

It's still undeserved... Finally got nonstop service to Chicago - ORD. Still no nonstop service to the Washington D.C Area (IAD,DCA,BWI) or the New York Area (LGA,EWR,JFK)..Both top markets/high passenger yield.

Quoting Boeing7E7 (Reply 18): Your results are nearly identical, /- half a percent.

Could you lead me to these results that you are talking about? I'd be interested to see how others have formulated similar models. I think it's an interesting topic.

Quoting RoseFlyer (Thread starter): Underserved US Metro Areas: *New York *Philadephia *Washington DC

See: Amtrak of some of that

Quoting RoseFlyer (Thread starter): Overserved US Metro Areas *Los Angeles *Chicago *Dallas *Houston *Miami *Atlanta *Boston *San Francisco *Minneapolis *San Diego *Baltimore *Denver

The hub cities will always be overserved, to support feed from smaller markets that are dependant on the hub and spoke system and cannot support point-to-point flying.


what might be the most useful peice of information is to find that % at which each market is over/underserved and rank them by %

I am glad to see that Salt Lake City is on the Underserved list! Maybe this will help the skeptics wake up and realize that DL's SLC-CDG flight can be successful!

It's an interesting analysis, though it really depends on your definition of overserved. Many of the airports on your "overserved" list are that way because they are major hubs which will by definition generate far more traffic than an O+D airport would. They appear to be overserved, but in reality many of the seats in these markets aren't being offered for the local traffic, rather they exist for the connecting traffic.

A few of your airports are also skewed because you are using too restrictive population numbers. Your using MSA population which isn't an exact proxy for air service area. For example, SAV has far more than 300K in its air service area. It's closer to 500K. BWI has more than 2.6million because you have to include some of the DC metro area which routinely uses BWI.

Of course, no model is perfect and I've never seen one model that would be universally accurate for all of the top 100 airports. There's simply too many unique variables that drive air service in each market.

You really need to put the table in a format besides JPG. It's text data.

There's an error for San Juan.

Rather than using fixed boundaries for areas, you should consider a model where the amount of service that an airport provides to an area decreases with distance, with areas overlapping to some degree. That would avoid such absurdities as including IAD, but not BWI, with DC. Combine DC & Baltimore and you get an almost perfectly served (by your definition) market.

It would be good to find some way of reducing the bias towards hubs.

Quoting RL757PVD (Reply 21): what might be the most useful peice of information is to find that % at which each market is over/underserved and rank them by %

I'll look into doing that. It's possible, but I need to do some more work.

Quoting Analog (Reply 24): You really need to put the table in a format besides JPG. It's text data.

I know that really frustrated me. A.net doesn't allow spreadsheets or tables to be posted easily. I don't know the proper coding to make a table.

Quoting Analog (Reply 24): There's an error for San Juan.

Yes there is. I don't know what happened there. It worked before.

Quoting Analog (Reply 24): Rather than using fixed boundaries for areas, you should consider a model where the amount of service that an airport provides to an area decreases with distance, with areas overlapping to some degree. That would avoid such absurdities as including IAD, but not BWI, with DC. Combine DC & Baltimore and you get an almost perfectly served (by your definition) market.

Good idea. That went beyond the scope of my initial work, but if I want to continue making a better model, then I should correct for multiple airports and markets.

Quoting Analog (Reply 24): It would be good to find some way of reducing the bias towards hubs.

Actually I already have a variable regarding a hub bias. However I think there is more work to improve it. One way might be to bias hubs for if they are a major hub or not for a specific airline. There is bias towards having the smaller hubs look like they are underserved and the larger hubs to be overserved. However I don't think there is any way to not have Atlanta come out as biased.

Quoting RoseFlyer (Reply 20): Could you lead me to these results that you are talking about? I'd be interested to see how others have formulated similar models. I think it's an interesting topic.

I have the file that was once available on line. Fair warning, this guy is seriously misguided on the subject matter of aviation.... He knows enough to sound like he knows a thing or two, but makes statements that leaves even the casual aviation observer shaking thier head. While his theories may apply to other things, they can not be extended to the entire planet as he believes is the case.

http://www.econ.ucsd.edu/~rcarson/

Some of his tasty ignorance:

The Carlsbad-Palomar airport is a logical choice. This airport already has commercial service to Los Angeles and Phoenix. By adding only 800 feet, its runway would be as long as the one at John Wayne Airport in Orange County, which handles over half as many passengers as Lindbergh.

Over a cliff????

The airport authority insisted that any airport site had to support two 12,000-foot runways with 4,300 feet of separation between them to allow a pair of fully loaded 747s to take off or land simultaneously. In reality, there is no likelihood of a 747 ever being used to fly passengers in and out of San Diego, so why require even a single 12,000-foot runway?

Runway separation has to do with IFR handling and operational redundancy, not the aircraft type. Also, a 777-200ER/LR needs 11,600-feet of takeoff distance.

Whopper statement:

While Lindbergh Field may be the busiest single runway airport in the United States, Gatwick Airport in London, like Lindbergh, has a single runway that is not long enough to launch a 747. The largest plane landing at Gatwick is the same 767 that routinely serves Lindbergh. In 2005, Gatwick handled 33 million passengers versus 17 million at Lindbergh. How? Mainly by averaging more passengers per flight.

Gatwick has about 10% of their operations in aircraft larger than a 767, 2% in 747's.

http://www.signonsandiego.com/uniontrib/20060907/news_lz1e7carson.html

Quoting Boeing7E7 (Reply 26):

Thanks, it is always interesting to see researchers in the aviation field. I do promise that his work had nothing to do with mine. However data is data, and it is definitely possible for different people to get teh same regression data if they used the same explanatory variables.

I do agree that some of those comments are outlandish.

Quoting RoseFlyer (Reply 15): If you have any data sources that you can think of, then I'll gladly change my model.

Yes there is definately more to the equation then just a few data sets such as MSA population and current airport volume.

I've worked with such material in route planning and matter of fact seen the elaborate methodology a unnamed US carrier used to identify markets for service.

There are a host of variable to consider including easier to capture material such as population demographics and earnings, regions economic activity and industry mix, physical geography constraints, access to other forms of transportation, historical travel patterns, market segment mix, infrastructure issues, plus more complex things to identify such as , market contestability, market and fare elasticity, stimulation factors, S curves etc...

So while your research makes some conclusions, I feel the results as woefully incorrect in the case of many cities such as Los Angeles whom on your shortlisted criteria ends being the number one "Over Served" market analyzed.

Quoting RoseFlyer (Thread starter): Here is the regression predicting passenger traffic in each metro area in the United States.

I don't see Fresno or Bakersfield in your analysis. Which DOT source did you use for passenger traffic?

It looks like you are using the 2006 population estimates so Fresno should be between Raleigh and New Orleans. Bakersfield should be just behind Albuquerque, just ahead of Grand Rapids.

Quoting RoseFlyer (Thread starter): *Indianapolis

Definately under served, but IND is getting more service, but still no B6

Very interesting... anyone else notice that WN is big in 7 of the top 10 over served markets in the list? (Not WN bashing and assuming the list are in order of most overserved markets)

Quoting Laxintl (Reply 28): I've worked with such material in route planning and matter of fact seen the elaborate methodology a unnamed US carrier used to identify markets for service.

You are correct that route planners use much more elaborate models. I just did a simple one with 6 statistics with widely available data. Most people however do not get to see many models for service at airports since they are proprietary. I thought people would be interested in reading this one.

Quoting HPAEAA (Reply 31): anyone else notice that WN is big in 7 of the top 10 over served markets in the list? (Not WN bashing and assuming the list are in order of most overserved markets)

I noticed that too. I'm not surrpised though. Southwest brings competition and lowers fares which would increase passenger counts.

Quoting FATFlyer (Reply 29): I don't see Fresno or Bakersfield in your analysis. Which DOT source did you use for passenger traffic?

I used data for the top 100 largest airports in the United States by passengers boarded in the last year. Fresno and Bakersfield are not included in that list.

Quoting RoseFlyer (Reply 32): I used data for the top 100 largest airports in the United States by passengers boarded in the last year. Fresno and Bakersfield are not included in that list.

Thats the source I thought you used. FAT is about the 105 or 106 largest airport. Given that Fresno is about the 55th largest MSA I'd be curious what your model spits out.

This is the BTS summary info for FAT. They round the arrival and departure numbers but its only a few hundred off. But these numbers are only domestic.
Arrival 595k
Departure 595k

And for Bakersfield
Arrival 163k
Departure 165k

Quoting Boeing7E7 (Reply 26): Gatwick Airport in London, like Lindbergh, has a single runway that is not long enough to launch a 747.

He might want to tell the airlines that run 747s in and out of there all the time about that. I'm sure they'd be interested in knowing that they've been using runways that their planes can't use.  

Quoting Boeing7E7 (Reply 26): Also, a 777-200ER/LR needs 11,600-feet of takeoff distance.

That would be at MTOW, right? I've seen 777s get off the ground for NYC-Europe flights in much less than that.

-Mir

MSY is definitely under served. It is primarily a tourist/convention city. The airlines are still adding flights since Katrina, but at an excruciating slow pace. Actually MSY was under served before Katrina in many markets. The US Census Bureau numbers are way off for 2006 and that is part of the city's struggle during recovery because FEMA and others want to US Census Bureau numbers which are estimates with very few actual on the ground surveys being conducted.

But it is very interesting to see what you have come up with, especially since oil at/near $100 Barrel and talk of the possibility of aircraft being grounded.

How is the Raleigh-Durham area considered overserved? If anything they would be served well. You have to take into account the Raleigh-Cary AND the Durham MSA.. but I don't think I would call it overserved.

First of all, there is less than 2% connecting traffic at RDU (less than 1000 pax/year) nearly everything at RDU is Origination and Destination.

Second, it is a truely business and leisure destination (last reported stats were 51%/39%).

Third, it is one of the highest transplant rates in the country...

Lastly, we still don't have our SFO, SEA, CDG, and FRA flights yet.. so until those happen.. WE ARE UNDERSERVED!

And btw.. where do you get the expected pax froms? MSA expectations for 2006? 2010? 2005? or RDU traffic expectations?

Quoting ERJ170 (Reply 36): And btw.. where do you get the expected pax froms? MSA expectations for 2006? 2010? 2005? or RDU traffic expectations?

Expected passengers is from a regression that I run. It's not the expected passengers from any source, but a calculated estimante.

Quoting Mir (Reply 34): That would be at MTOW, right? I've seen 777s get off the ground for NYC-Europe flights in much less than that.

Yes that would be at MTOW, but SAN is about as far away from the destinations that need service as they can get. NYC-Europe is like SAN-Halifax. Not that the place would be flush with service, but the demand is there for London, Paris, Frankfurt, Amsterdam,Tokyo and Manila for sure. Possibly Munich, Guam, Seoul, and Sydney/Aukland as well.

The routes are between 5,500 and 6,500 nm. A 777-200ER can do 5,500 at MTOW. SAN kicks it a 40k weight penalty to start and it's downhill from there on the rest of the routes. The -200LR fares much better where the max payload range from SAN is 6,000nm. Not a lot of orders out there though.

Hi Roseflyer,

I agree with all the airports on the list. I don't agree that the pop. of the ALB metro area is 800,000. You just included part of the metro area(Albany,Schenactady,Troy) The total metro area of the Capital District is 1,147,850.

I think that ALB should get the following flights in the coming years. Note these are JUST the terminating pax. For example ALB only has 180 terminating pax to ATL. Hundreds more connect.

DFW/IAH-about 100-150 pax, Could do well on a E145XR. yet only CO has the AC.
ALB-FLL probably on FL/WN/B6
ALB-PHX probably on US/WN
ALB-RDU probably on AE
ALB-MIA on AA
ALB-STL
ALB-BNA
ALB-PBI
ALB-DEN
Another fLIGHT TO LAS

Why hasn't B6 come to IND yet. It has been rumored many times that they were coming to IND, yet they never do. although WN is great it's fairs out of IND are often as high or higher than the non- LCC's. Another thing is that IND has limited international service-just last year we finally got a seasonal once weekly on F9 and NW to CUN and AC Jazz finally gave us a 2x daily CRJ to YYZ (upgraded from Dash 8).

I guess I'm not understanding this. How can two huge o/d locations like MCO and LAS be over served while a small and weak o/d location like MEM with a hub be under served? How do you over server origination and destination traffic? If that was the case loads would be poor. It isn't like a hub where poor o/d numbers can be masked by feed.

Interesting, TPA is listed as underserved with SRQ as well.
As Louis Miller, TPA Chief has stated SRQ catchment makes up 15-16% of the TPA enplanements.
This is a huge part of thier business.

Given the strong market overlap I would question if TPA is underserved.
What would be the impact on the TPA market if the leakage didn't exist?

Quoting Indy (Reply 41): How can two huge o/d locations like MCO and LAS be over served while a small and weak o/d location like MEM with a hub be under served?

Yes with the nature of the model, the largest airports often will be overserved. MEM counts as a hub which means its passenger count should be high, although it isn't that high. It's a source of error in the model. Again it is only 85% correct. The models that the airlines use themselves are probably 99%.

Quoting RL757PVD (Reply 21): Quoting RoseFlyer (Thread starter): Underserved US Metro Areas: *New York *Philadephia *Washington DC See: Amtrak of some of that

Actually, A LOT. Those are the top 3 Amtrak stations by annual passengers, and the next closest competitor is more than a million passengers behind Philly. In Washington's case, the 3.5 million Amtrak passengers more than make up for the underservice, and in NYC, Amtrak gets you within a million if you include both NY Penn Station and Newark Penn. Amtrak also cuts down 30% of PVD's underservice. Amtrak is even more damaging when you consider that a lot of what it siphons off is high-paying business travelers on the Acela. I'd suspect that has something to do with your underservice for PHL in particular, given the comparative proximity to downtown of PHL and 30th St. Station.

Quoting Analog (Reply 24): That would avoid such absurdities as including IAD, but not BWI, with DC. Combine DC & Baltimore and you get an almost perfectly served (by your definition) market.

For statistical sake, that's the best solution, although there are some issues with non-overlap of markets. If you're going to combine NYC into one huge market though, you've gotta do it for DC. For people not familiar with the Washington/Baltimore relationship/feud/etc, this is a big sore point for a lot around here. Remember, BWI isn't in Baltimore.

Quoting RoseFlyer (Reply 43): MEM counts as a hub which means its passenger count should be high, although it isn't that high.

Don't use a binary dummy variable then. Use a "hub factor" with a range of 0-1 and place airports along the continuum with 1 being all connecting and 0 all O&D...or reverse it if you'd like. There's also something in general going on with the hub numbers. Every single carrier's primary trans-atlantic hub is in the "under" category (IAD, EWR, DTW, JFK). This may have something to do with interrcorrelation with the location correction factor, so you might want to look at that.

Finally, instead of using per capita income alone you might want to add in the regional gross economic product divided by population, as that will give a big of a taste of the corporate travel market, which often drives service to certain areas.

Good analysis, interesting read.

Quoting RoseFlyer (Reply 43): Again it is only 85% correct. The models that the airlines use themselves are probably 99%.

Judging by the number of airlines with problems I think perhaps yours is more accurate lol. If this model is accurate then NY has some real big problems.

Thanks for a very nice analysis.

This certainly drives home the need for some of the new ULCC's that have stated up , or expanded lately.

Quoting ERJ170 (Reply 36): Lastly, we still don't have our SFO, SEA, CDG, and FRA flights yet.. so until those happen.. WE ARE UNDERSERVED!

If there was a compelling reason to offer those flights, I'm sure somebody would. I don't see how the absence of four particular destinations makes an airport underserved, especially when two of those are transatlantics that most likely couldn't survive without subsidies.

-Mir

Nice work RoseFlyer. While we can argue all day about parts of your model, it's great to put some numbers out there.

One thing I'd be very curious to see (and which might help you refine your model) is the correlation between the difference between the actual and expected levels of service in a given market and the average ticket price in that market. Obviously average ticket price is data that's not all that easy to get, but maybe an analysis of, say, 25 randomly chosen dates and destinations might be able to stand in for the real thing.

My own experience suggests that your data may not be far off with respect to the cities I often fly to. Those that are most dramatically underserved do indeed seem to have the highest ticket prices.

Quoting ERJ170 (Reply 36): Lastly, we still don't have our SFO, SEA, CDG, and FRA flights yet.. so until those happen.. WE ARE UNDERSERVED!

I wouldnt hold my breath on CDG or FRA. Im amazed LGW has hung on as long as it has. But since it is subsidized, its a good one for AA. I would like to see AA do RDU-LAX/SFO, but beyond that I dont think AA will through anything International at RDU (with the possible exception of Mexico).

As for the hub airports, they will always come up overserved. Especially Atlanta, DFW and Chicago. It seems the data is biased toward them.

Great data RoseFlyer I enjoyed reading it!