Applying Intervention Analysis to Financial Performance Data: The Case of US Airlines and September 11th

Article excerpt

Abstract

Using intervention analysis I assess the effect of the September 11 terrorist attack on the performance of the US airline industry. The estimated initial effect supports the US federal government decision to provide a $5 billion cash compensation to the airlines. However, the long-run effect is found to be much smaller than the losses reported by the industry in 2001 and 2002. Also, the analysis suggests that not all of the airlines were equally affected by the terrorist act and that investors were fairly rational pricing major airline stocks, but were less accurate with the stocks of smaller regional carriers.

Keywords US airlines * Intervention analysis * Terrorist attack * Impact effect

JEL Classification G14 * L93 * L98

1 Introduction

The tragic events of September 11, 2001 provide researchers with the natural experiment settings to study the effect of an unanticipated large-scale catastrophic event on the US economy and society. US airlines have been placed in the most vulnerable position by this devastating terrorist act. Immediately after the attack the Federal Aviation Administration shut down the National Airspace System and all air traffic in the US remained grounded for several days. US airlines have experienced substantial losses after the attack as a result of flight restrictions, enhanced security procedures and the abrupt decline in passenger traffic. Even though $5 billion in cash compensation and up to an additional $10 billion in loan guarantees were provided to the airlines by the federal government, US airlines posted total losses of $17.7 billion in 2001 and 2002.1

I assess the impact of the September 11 terrorist attack on the performance of US airlines using intervention analysis first suggested by Box and Tiao (1965, 1975) and further developed by Larcker et al. (1980), Enders et al. (1992) and others. Intervention analysis has advantages over the standard event study method first introduced by Ball and Brown (1968) and Fama et al. (1969). It allows the observed autocorrelation in the model residuals to be removed thus providing improved estimates for reliable statistical testing, and requires considering each time series on a case-by-case basis instead of forcing one model on all series when examining events that simultaneously affect several firms (Larcker et al. 1980). Also, intervention analysis provides an impulse response function to study the transitional effects of an event.

Guzhva and Pagiavlas (2004) utilize vector autoregression (VAR) model to separate the effect of the attack on US airline performance from the downturn in general economic conditions. They confirm the US government assessment that US airlines lost about $5 billion due to the attack in September-December 2001. While VAR is very useful in decomposing the effect of the attack from macroeconomic influences, it does not provide the long-run effect, which can be estimated through intervention analysis as the change in the long-run mean of a series of financial data.

I examine monthly revenue passenger miles (RPM) series to objectively assess the effect of the terrorist attack on the financial performance of US airlines. RPM are considered to be the industry standard for the airlines' financial performance evaluation and have an advantage over typical accounting and financial statements as they provide unbiased performance information and are not influenced by creative accounting techniques. I perform intervention analysis with the aggregate industry data and with individual monthly RPM series for US major2 and regional airlines to detect if the terrorist act resulted in different effects for individual air carriers. As expected, the intervention analysis of the industry data reveals a statistically and economically significant effect of the attack on the performance of US airlines. The magnitude of the effect, however, is much smaller than losses reported by the US airline industry. …