Revenue Passenger Miles (RPM)

1
Revenue Passenger Miles (RPM)

• Brandon Briggs, Theodore Ehlert, Mats Olson,
  David Sheehan, Alan Weinberg

2
What are RPM?

• The leading indicator of the health of the
  airline industry
• Nearly universal application
• Measures passenger traffic
• Number of seats sold multiplied by distance
  traveled
• Distances are fixed
• Expands only by airline capacity
• Accurately reflects changes in demand
• Does not rely upon sales figures
• Insulated from inflationary concerns

3
Characteristics of Airline Industry

• August is the peak month
• RPM always decline in September
• Highly cyclical

4
Effect of 9/11 on RPM

• Significant drop in September RPM
• Air travel was shut down for several days
• RPM bottomed out for several months post-9/11
• Long-term impact on RPM
• RPM depressed below pre-9/11 levels for 3 years
• What would the graph of RPM look like if 9/11
  hadnt occurred?

5
Histogram of RPM

• Not significantly different than normal
• Multi-peaked

6
Correlogram of RPM

• Seasonal trend in PACF
• Possible cyclical trend ACF

7
Unit Root Test of RPM

• No unit root
• Does not approximate white noise
• Affected by large drop in 200109
• Add intervention variable (STEP)

8
Box-Jenkins Model I

• Step function
• Parse data for pre-9/11 and post-9/11 trends to
  account for precipitous drop in RPM
• First difference
• Seasonal difference
• Drop first difference
• Negative coefficient on autoregressive term
• Over-differenced

9
Box-Jenkins Model II

• SRPM C SSTEP AR(1)

10
Box-Jenkins Model II

• Residuals still not orthogonal (Q-Stats)
• Add
• MA(12)
• MA(15)
• AR(2)

11
Box-Jenkins Model III
12
Box-Jenkins Model III

• Orthogonal, normal, slightly kurtotic
• Fitted values match actual, even 200109
• No autocorrelation (Breusch-Godfrey)
• ARCH/GARCH not needed

13
Forecast (200703 200802)

• Peaks are trending upward
• The forecast seems to fit well

14
Forecast (200703 200802)
Shown with 95 Confidence Interval
15
Long-term effects on RPM

• Added a linear trend from data 199601 200108
• Linear trend represents mean value for RPM if
  9/11 did not occur
• RPM is trending at a lower mean post-9/11
• Post-9/11 trend has greater acceleration than
  pre-9/11, suggesting RPM is catching up

16
Conclusion

• RPM drops 29.8B from 200108 200109
• Difficult to measure short-term impact of 9/11 on
  demand as measured by RPM due to complete
  shut-down of airports
• May follow our study with daily RPM analysis
• RPM drops 184.2B from 200109 200208
• Confidence Interval /-13.6B
• 25 decline
• Definite long-term impact of 9/11 on RPM
• Does not accommodate impact of post-9/11
  recession
• Multiply SSTEP by month and sum over twelve
  months