Market efficiency & arbitrage opportunities in the FTSE-100 option market : an application on the Put-Call Parity with high frequency data
This thesis examines Put Call Parity (PCP) deviations in the LIFFE FTSE-100 Options quoting system and tests the two competing hypotheses put forward in the literature. Our dataset covers the period of July 1994 to March 1997 and contains 357,985 and 431,145 observations (for the European and the American types) resulting in 40,124 and 57,382 PCP deviations respectively. We calculate PCP misspricings using the model proposed in Kamara and Miller (1995). The model used here accommodates market imperfections but does not include taxes. The model also allows for the immediacy risk and the early exercise risk associated with evidence of Put-Call Parity deviations documented in the literature. We find evidence of significant deviations, net of costs, throughout the period. We test misspricings in both American and European contracts for the same period and equal contract parameters and find evidence supporting both hypotheses where appropriate. The level of deviations found suggest that other factors could attribute to their identification, we propose liquidity-related factors such as inventory constraints. We assume that persistent deviations from the PCP, which are not supported by the option pricing theory are indications of market inefficiency. In well functioning markets we expect that larger PCP deviations will be removed from the system first. We fit a Cox Proportional Hazard model and test the significance of the level of deviations as a covariate. We find the degree of deviations to be a significant factor in the duration of the misspricings for the majority of the observations. We conclude that under these evidence the market is not characterised as inefficient. The last part of this thesis models the PCP deviation series as a sequential stochastic process. We fit around the process the Autoregressive Conditional Duration Model, as proposed by Engle and Russell (1995) and modified by Bauwens and Giot (1997). We conclude that the model offers an adequate representation for this high frequency, irregularly spaced series. Keywords: Put-Call Parity, High Frequency, Duration, Cox (AND) Proportional (AND) Hazai-ds, Irregularly Spaced Observations, Autoregressive (AND) Conditional (AND) Duration (AND) Model.