Forecast Accuracy and Unbiasedness of Output and Current Account Growth Rates: A Panel Study of Forty Three Developed and Developing Economies

This research proposes to test the accuracy and unbiasedness of output and current account growth using consensus survey data published by fx4casts.com. Consensus survey data are used extensively by practitioners and policy makers to understand the influence of expectations on economic behaviour, and they are found to be more accurate than the individual forecasts. Consensus forecasts are of particular interest to researchers because these are the opinions of the experts. Many studies investigated the accuracy of consensus survey data on important macroeconomic variables, like, interest rate, exchange rate, GDP growth rates, inflation rates, unemployment rate, etc. There are a number of studies that use survey data on output growth to understand whether experts expectations are unbiased and whether they use all available information efficiently in forming their expectations. The general conclusion of these studies show that survey data are a biased predictor of actual change, and that they do not incorporate all available information. The question needs further investigation as suggested in the literatures because of a number of important developments at the policy level (for example increased monetary policy transparency across the world), and forecasting environment (for example, more access to information and development of sophisticated applications). Current proposed study will be particularly interesting to researchers as it covers a time period (2002 2014) when monetary authorities in many countries have become more open, and significant events, like Sept 11 attack on World Trade Centre, financial crisis of 2007-2008, and massive monetary and fiscal interventions have taken place in many countries in the world. Besides it will be a more comprehensive study (Forty three developed and developing countries) than any other earlier study using a longer time period and panel based regression methods. To test the accuracy and unbiasedness of expectations, we will conduct an extensive descriptive statistics of forecast errors and a formal unbiasedness test and also perform panel cointegration analysis. Data on the two variables will be collected from www.Fx4casts.com for this study. There are no studies using this source of data. The journal provides updated monthly consensus forecasts of these variables, where forecasts starts two years early and forecasts are revised every month. Data on actual rates are also provided in the journal. The nature of the data compilation is itself an makes the study very interesting. Other relevant data will be collected from data stream, and Bloomberg if necessary. The results of this study will enhance our understanding of forecast accuracy of output and current account growth, which will be useful to both practitioners and policy makers. The study will particularly enhance our understanding of forecast efficiency in the developing countries as there is no comprehensive study on the issue for these countries.