Forecasting stock market prices in a thin security market

Stock market efficiency is a crucial concept when forecasting of future stock price behaviour is discussed. In the literature, a distinction is made between three potential levels of efficiency. Under a weak form of efficiency, information on historical price movements is of no value for predicting the future price development. Similarly, a semi-strong form of efficiency holds that no publicly available information can be successfully used in the prediction of prices. And finally, a strong form of efficiency means that the share prices fully reflect all relevant information including data not yet publicly available. Stock market efficiency has been extensively studied in different countries. On a thin security market, like in the Helsinki Stock Exchange, many anomalies and deviations from market efficiency have been obtained. This paper aims to contribute to that discussion. It is shown in the paper that both the monthly and quarterly stock market prices (the general stock market index) can be adequately forecasted using either univariate time-series analysis or multivariate econometric modelling. The univariate ARIMA-models seem to be slightly outperformed by the econometric models. It is further shown that the forecasting accuracy of the models can be improved when time-series and econometric forecasts are combined into a composite forecast. The empirical results obtained indicate an absence of efficiency on the Finnish security market.