By Dripto Mukhopadhyay
In research fraternity, forecasting is always known as “Thankless Job”. The reason being it is one of the most difficult exercise since forecasts depends on large number of assumption about future over and above the assumptions involved in the econometric modelling itself. Being fortunate enough to work on forecasting relating to various sectors ranging from petroleum demand to luxury car to carbon emission, I know the amount of effort and skill goes behind any forecasting exercise, if it is a serious business. Even after that, many of the times the researcher find their forecasts off the target extensively mostly because of externalities. At times I feel that except a “Fortune Teller”, no scientific researcher ever can guarantee about the forecasts. The forecasts can change drastically because of small amount of change in any of the multiple assumptions goes into forecast because of macro-scenario in a dynamic world.
However, a researcher always wants to understand how his forecasts are matched with actual scenario after a few years of the forecasts were made. I did a forecasting for foreign tourists arrivals to India in the year 2008-09 for Indian Institute of Tourism and Travel Management (IITTM) as a consultant. The paper was published later in the “Indian Tourism Statistics”, the only government publication on tourism statistics of India. The forecasts were made from 2010 to 2014. Since recently the latest Tourism Statistics published for the year 2015 contains data for 2014, I felt like matching the accuracy of the forecasts I made in 2008.
The research paper covered 6 countries and all the regions of the world. The data used was various macro economic parameters, household disposable income and certain dummy variables relating to policy and other localised incidences like terrorism etc. The comparison between forecasts made in the year 2008 and the actual foreign tourists arrival to India. The details of the accuracy level of the forecasts is given in Table 1. Country-wise details and region-wise details are given in Table 2.
For any secondary data collected in a large scale and at a macro level, it is always considered that results are extremely accurate if lies within plus/minus 10% deviation level. An accuracy level till 85% (where the deviation is plus/minus 15%) is considered as acceptable for any valid decision making purpose. The numbers presented in Table 1 provides the details of forecast numbers from 2010 to 2014 for all countries and regions covered under the study. It suggests that 62% forecast numbers in the study is extremely accurate when compared with the actual FTA (Foreign Tourist Arrival). If we consider the acceptable limit with 85% accuracy, it goes up to 77% of the forecasted data points. Overall, this results suggest that FTA forecasts made in 2008 was fit to the expectations out of any forecasting exercise. Keeping in mind the global economic recession during end of 2008 and the continuing volatility of the global economy, these results suggests that decision making and policy making can depend on forecasts to a large extent if the methodology used is robust.
Table 1: Details of Accuracy Level of Forecasts
(Forecasts made in Year 2008 for the years 2010 to 2014)
Forecast VS Actuals
% Forecast points
|On target||100% correct||
|Highly accurate||More than 90% accuracy||
|Acceptable||Accuracy level 85% to 90%||
|Low on accuracy||Accuracy level less than 85%||
Table 2: Regions and Country of Details of Forecasts and Deviation of Actual Foreign Tourists Arrivals to India