Another way to think about these different market behaviors is to think of them as risk that behaves in different ways. For example, market risks may be of different duration, of faster or slower onset, of greater or lesser variance (kurtosis) or greater or lesser persistence. In other words, we may be exploring the different qualities of risk rather than its quantity. If these differences are significant, they may provide considerations other than aggregate risk and return that allow corporations and investors to prefer investments in smaller markets to those in larger markets. If this is true, the different behaviors would not be transient risk factors subject to being arbitraged away in global markets characterized by APM trading, but rather the risk factors in segmented global capital markets. In such a case these different qualities of risk may provide the basis for permanent market segmentation.
If global capital markets remain segmented on the basis of these risk qualities, investors would want to diversify their securities portfolios based on their individual preferences of risk and return, searching for the optimal market portfolio composed of securities from different capital markets. Thus the risk characteristics of smaller markets might provide diversification possibilities not available in larger markets and these possibilities might be useful enough in reducing portfolio variance that they could provide a basis for the continued competitiveness of these smaller markets.
This research uses general autoregressive conditional heteroskedasticity (GARCH) and threshold ARCH (TARCH) models of the Jamaican, Taiwanese and U.S., stock markets to look for significantly different market characteristics that might provide tempting options that would motivate investors to choose one market over another on a basis other than CAPM or APT efficiency. GARCH and TARCH models ( Gourieroux) are well adapted to this task, because they allow an investor to forecast changes in risk, rather than in price as is ordinarily done. The Black-Scholes models ( Fischer & Scholes) and other options models show that the key variable which determines options value is the variability of the underlying stock's price rather than the market price of the stock. So investors have used GARCH ( Baillie & Bollerslev; Bollerslev 1986; Bollarslev, Chou & Kroner; Bollarslev & Wooldridge; Engle 1993b; and Nelson) and TARCH models to predict changing patterns of securities risk ( Baillie & Bollerslev; Bollerslev & Wooldridge; Chou, Engle & Kane; Engle 1982; and Engle 1993a), hunting for market inefficiencies ( Chan, Karoly & Stultz; Chou & Engle; and Glosten, Jagnathan & Runkle) that might be turned into large capital gains and abnormally high rates of return ( Day & Lewis and Engle & Chowdhury).