IDENTIFICATION OF DEMAND DRIVERS

This model exhibited extremely small error in each year, less than 0.5 margin points, with the exception of one particular year⁴. Due to the extremely good fit of the model (and strong statistical measures) we were comfortable in opining that this model fairly represented the dependence of the airline industry’s profitability on 1) the strength of the economy 2) the mood of consumers and 3) the impact of extraordinary events.

Our conclusions, based upon our findings, were:

1) It is possible to identify the demand drivers of the airline industry. It is volatile and unpredictable, but understandable. This has implications for leverage for the players in this space, and the importance of prudence. Our regression on the airline industry indicates quite clearly that the current operating performance is consistent with the Industry’s ever-present dependence on (and vulnerability to) changes in real GDP, Consumer Confidence, and extraordinary events.
2) The adverse effect, on the industry, as a whole of adverse events which create a “fear to fly’. Our regression coefficient suggested at the time that an average 2.4 point decrease in operating margin were result of wars and other Black Swan events.
3) The profitability of the U.S. airline industry will recover as GDP and Consumer Confidence recover, assuming the absence of any extraordinary shocks to the industry, ceteris paribus.
4) A separate regression analysis concluded that 80% of NWA’s profitability was driven by its industry, and 20% was unique, likely its dominance of the Minneapolis hub and its Pacific routes. We concluded that NWA enjoyed about a 20 basis point advantage in operating margins over its industry.

Footnotes

¹ For readers who are not familiar or comfortable with multi-variate regression, we offer the following analogy. Fuel efficiency of an automobile is affected by both the horsepower of the engine and the weight of the car. However, engine horsepower and car weight are related. In many cases, horsepower and car weight carry much of the same information, and fuel efficiency can largely be predicted using one or the other. A “radar” chart would reveal such a relationship, and would better predict fuel efficiency than either variable used alone. Regression is a statistical method which calculates an algebraic equation which best explains observed relationships. It is, in essence, a radar chart taken much further.

² After conducting this work (1992), we found a study by Robert Decker of Duff & Phelps, which also used regression analysis. We discussed our work and findings with him. Mr. Decker did not examine Consumer Confidence, nor did he attempt to look at the industry as a whole. He did study GDP and found that it is 97% correlated to Manufacturing and Trade Sales.

³ Businesses set travel budgets based upon last year’s profitability. This accounts for why GDP lagged a year correlated with this year’s business travel. Current consumer confidence drives holiday travel, and thus, current consumer confidence is the other demand driver.

⁴ We find this to be a very curious outcome; the year in question was the last year of the economic expansion. We posited the question, “was the consumer already pulling back on discretionary expenses in that year, in spite of all indicators?”. Robert Decker at Duff & Phelps would likely agree. His opinion was that airline profitability can sometimes predict GNP. This logic has an appeal. If consumers pull back on discretionary purchases, the economy will suffer.

⁵This case study was subsequently cited (2012) in a pioneering Airbus funded PhD thesis: “How the consumer confidence index could increase air travel demand forecast accuracy” (2012), p176.Teyssier, Narjesse https://dspace.lib.cranfield.ac.uk/bitstream/1826/7907/1/Narjess_Teyssier_Thesis_2012.pdf