New (Blue Sky) Frontiers in Risk Management (and Markets.)
It would seem obvious to anyone who has ever boarded an airplane that weather is a primary factor in determining whether or not any given flight will take-off and/or land on schedule. The impact of adverse weather conditions is even more acute for commercial flights using increasingly congested major airport hubs, with the complexity of managing thousands of co-dependent paths within the network. This weather-driven uncertainty has a real and measurable financial impact on carriers and air travelers including (but not limited to): extra fuel, maintenance and staff costs, customer compensation (or at the very least poor customer satisfaction); and for their customers lost time and productivity (not just on delayed flights as customers are forced to ‘build-in’ the uncertainty implicit in the flight schedules to their planning.)
In the past, there was no practical way to price and manage this risk; the transaction costs and computational intensity of trading and managing such a granular and complex set of weather risks would have certainly outweighed the potential benefits. Today however thanks to accelerating technological advances in computing and communications, and taking inspiration from the creative application of technology to pricing, managing and distributing risk derivatives pioneered by innovative young companies like Weatherbill, this is clearly no longer the case. I would go so far as to posit that any CEO and/or Board of a commercial air transport business is at risk of breaching their fiduciary duty if they are not seriously contemplating how they can manage and mitigate weather risk in their operations. At the very least, it should be quantified and reported - much in the way fuel-price risk now is - and any hedging (or deliberate decision not to hedge) strategy articulated and explained to shareholders and any relevant regulatory bodies.
To help get this Boardroom debate started, Weatherbill has helpfully just published a white paper framing the problem and at a high level quantifying the risk of weather-induced delays for US commercial airlines (from the report summary:)
Between June 2003 and April 2007, over 25% of all flights in the United States were disrupted
(either cancelled or delayed). More than 55% of those disruptions (almost three million) were
due to weather- the leading cause of flight disruptions in the U.S. In an effort to educate airlines,
airports, and consumers about their weather risk, WeatherBill has identified the most sensitive
airports and airlines to adverse weather to facilitate reliable estimates of future flight disruptions.
Fifty-four major airports and sixteen larger airlines were studied. There are three main results:1. WeatherBill can statistically quantify the relationship between weather delays and
observed temperature and precipitation at major U.S. airports and airlines
2. The study shows that disruptions are more common with precipitation than temperature
3. Temperature-linked delays are seasonalWe have included a list of the top five airlines and airports with the highest & lowest
percentages of weather disruptions at the end of this summary. Those lists are already widely
available. What follows immediately are lists of airlines, airports, and their delay sensitivity to
precipitation and temperature, in minutes. WeatherBill hopes this new data will help the flight
industry and travelers better understand their weather delay risk.
I would suspect that most (all?) of the airlines would have there own detailed data on this, so it is unlikely (I would hope!) that anything in this report will come as a big surprise but I would be curious to know how (if?) they apply this data in managing their business (setting schedules, pricing, etc.) Having quantified the risk, you might think the next logical step would be to initiate a risk management program to monitor and potentially manage (through trading granular weather derivative contracts) this risk dynamically. While I think this is a sine qua non for anyone managing an airline, I believe there is a much more exciting opportunity (than simply mitigating downside risk) that arises from the ability to measure and manage economic sensitivity to weather: the opportunity to build such an advanced knowledge of operational risks into the customer proposition. So what exactly am I talking about? Well, at the simplest level to illustrate, an airline could sell ‘weather-protected’ tickets: tickets offering for example a full or partial (or variable) refund for weather related delays or cancellations. Not only would this improve the customer experience, but would be very helpful in offering a differentiated price based on time-sensitivity of the traveler: a business person going on a day-trip for a critical client meeting vs a university student traveling home for the vacation probably have a different sensitivity to the ‘cost’ of a delay or cancellation. Or maybe not. The point is, let the customer pay accordingly. It wouldn’t be perfect - there is basis risk involved - and I suspect that from a marketing/adoption point of view there would be an optimal level of complexity (accuracy) vs simplicity in structuring these sorts of deals; and obviously it wouldn’t necessarily ‘change’ the outcome (weather is weather) but those customers who most valued their time would be compensated accordingly.
This dynamic (weather-) risk adjusted approach to pricing and management would also be relevant to airports and private charter or air taxi operators. Airports - especially those run ‘for profit’ - could build weather sensitivity into their landing and operational fees. For private aviation, while generally less subject to the negative weather-related knock-on effects endemic to the large commercial ‘hub’ airports, the relative importance of weather on flight disruptions is almost certainly higher than for commercial airlines (being less exposed to the other primary causes of flight disruptions (as defined by the US DoT): carrier delays, security delays, and late arrival of an earlier leg.) Also given the use of smaller aircraft, the sensitivity to adverse weather may also be greater in some cases (than for large commercial aircraft.) This approach would also be relevant for air cargo operators and given their concentrated hub operations and expertise in highly sophisticated logistical optimization, one could argue that implementation would be easier for companies like FedEx and UPS. (One problem however might be finding enough risk capacity on the ‘other side’ for Memphis…)
The Weatherbill study points out that (due to a lack of data and complexity beyond the scope of their paper) they did not consider ‘wind’ risk in their study, although they suggest that this is likely to some extent to be embedded in the temperature risk in some regimes (ie disruptions correlated with high summer temperatures may well reflect higher convective winds and/or thunderstorms related to these high temperatures.) However, there is no (technical) reason why over time sufficient (and sufficiently granular) relevant data on windspeed couldn’t be captured and used in pricing models up to and including in real-time. This would require data-capture not just at ground stations (airports) but in the air (aircraft) but the technology exists and the cost of transmission, storage and computation have (or will soon) declined sufficiently to make using what would be an incredibly vast and dynamic data set tractable (in a way that it would not have been even 5 years ago.)
You may recall I touched on this subject (dynamic outcome-driven air transport pricing) previously; and while weather risk management would be a good start in terms of bringing airlines into the 21st century, in my opinion a much deeper and more fundamental reappraisal of their business model is called for. Fundamentally, airline seats (or on any carrier for that matter) are substantially fungible - ie a ticket gives the right for one person to travel from point A to point B. Ok, Ok … before you get the pitchforks out - yes there is a difference in value between a seat on a Gulfstream V and the middle seat in row 34 on DiscountAirways… but the differences are relatively easy to understand and so I believe would be (mostly) efficiently priced in an open market. But ok, for the sake of simplicity, let’s set aside private or charter operations for the moment, and concentrate on scheduled commercial airlines. What you have today is a fragmented and reasonably opaque primary market* and no real efficient secondary market. (*Although the advent of the internet generally, and a number of innovative start-ups specifically has vastly improved the situation from that of a decade ago…) Why is this? Practically speaking it is because the airlines don’t want to allow transferability - they are unable or unwilling to embrace the fact that what they are selling is a commodity. That they are selling the transport of packets on a network. It’s an ego thing. They think they would lose out. Putting aside the fact that most of these companies have lost billions of dollars over decades (so I’m not sure what they think they have to lose), I am convinced that by encouraging a robust and complete secondary market in airline seats, not only would consumers win (through more transparent and rational (supply/demand) driven pricing) but the airlines - at least the well-managed ones - would be huge winners. First they would be able to save money by eliminating (or redeploying more productively) the boffins they currently pay to build ridiculously complex and customer un-friendly pricing schemes in the vain hope of optimising a priori load factors and revenues, and instead be able to focus on managing their assets, optimizing their routes and schedules and making their customers happy (insofar as they could see a return on this investment from a structurally higher secondary price for their seats.)
Stop for a moment and think how fundamentally this would change the paradigm of running an airline - load factor would disappear as a relevant metric because by definition, every seat on every flight would be theoretically ’sold’ - ie would have a market price (which in some cases may admittedly be zero…) - it would make explicit the fact that an airline is actually long a portfolio of options and could - using the feedback loop implicit in a robust secondary market - seek to manage this portfolio in such a way as to maximize the premium income. Part of this strategy would involve deciding when and how many seats to sell in the ‘primary’ market, and may in some cases involve also buying - yes buying - back seats in the secondary market as demand dynamics change. I sense that many of you are still uncomfortable with the heterogeneity of the market - ie the lack of fungibility - and the impact that would have on the liquidity of a secondary market. Perhaps this analogy will help: think of the (corporate) bond market - by definition it is much more complex and heterogeneous than the ‘equivalent’ market in common equity. Not only are there differences in the credit quality between companies (think different service levels, seat coverings, entertainment systems, airport lounges…) but even for the same company their are securities with different characteristics (maturities, seniority, coupons, etc.) (think flight times, class of service, changeability, etc.) - none of which inhibits the market from operating. Furthermore, the price signals this market sends with respect to these variables are important inputs for optimizing the management and balance sheet structure of these same companies. By allowing the CFO to see the relative cost and cost volatility of having a BBB vs a AA financial structure, she is much better able to make a decision as to which is best for her shareholders. In the same way, an airline executive would be able to better understand if his investments in customer service or in-flight entertainment provided positive returns to the shareholder (based on an average per-seat premium reflected in the market.)
I imagine that a further argument against such a market would be ’security’ (ie ‘identity’): the airlines (and various government agencies) need to know who is travelling in any given seat. Well again - thanks to technology - this is a red herring argument. There is no reason to believe that in a world of ubiquitous mobile phones, electronic payments and fulfillment that this ‘problem’ is not entirely tractable. Indeed, to take the securities markets as an example - due to various ‘know-your-customer’ and anti-money laundering statues - the days of anonymously trading bearer certificates are long gone; and yet the number of participants and transactions in financial markets has never been higher. So yes, any secondary market would need to robustly and accurately identify the ultimate ticket holder but this would not be a problem. (Additionally, in the first stage I would imagine it would probably make sense to start with a market that ended T-1 - ie not allow trading in the last 24hrs before a flight - which would significantly remove or mitigate a number of potential operational risk factors arising from such a market. Once these risks were better understood and engineered around, one could imagine eventually allowing trading up to the moment the flight closes for boarding.)
You may recall that in my earlier post, I mentioned Farecast as one of a variety of companies innovating intelligently in this space. This is a company that was on my “IRWIWHHTOTI” (I-really-wish-I-would-have-had-the-opportunity-to-invest) list (for reasons I hope would be clear to my regular readers…) Unfortunately, they have just sold themselves to Microsoft. Why do I say unfortunately? (1) There is now no chance to invest in or buy the company. (2) Microsoft has a long and not-so-illustrious reputation for buying really interesting and innovative companies (good) and then having their own big-corporate antibodies attack and often kill said innovation and energy (not so good.) (See here for more thoughts on Microsoft.) I hope this doesn’t happen to Farecast (in the same way I hope CBS won’t kill last.fm) but let just say I’m cautiously pessimistic. I am entirely sympathetic to the founders - liquidity is important (you can’t pay mortgages with ‘potential’ upside) and understand how the structural constraints of the mainstream VC business model drives the logic of this kind of exit. But the combination of these factors leads from time to time to what I would consider excellent opportunities to deploy smart, unconstrained capital. Since this is something I personally have limited amounts of (alas) I will be working to convince others of the merits of this view, with the goal of being able to act on a small number such opportunities when they arise in future…
Many of the issues that would be faced in creating such a market are very similar to those faced by secondary markets in live event tickets. Whilst, I wouldn’t want to distract the founders and executives of the companies pioneering in this space (against much hysterical and illogical reactionary resistance from some of the incumbent market participants,) I wonder if they might be available in a year or two’s time to sit down with me and my partners and construct a plan to turn this vision into a reality. Or I wouldn’t be surprised if someone were already working on it. If so I’d love to know more.
