Effect of changes in age structure of the population on government tax revenues and predicting its changes: An approach of Mixed Frequency Data Sampling (MIDAS)

The aim of this thesis is to investigate the effect of changes in population age structure on government tax revenues and forecast its evolution using MIDAS method and time series data during the years 1367 until 1393.Changes in population age structure caused by the sharp rise in fertility in early 1360, has brought many consequences and questions. One of the questions is how changes in population age structure will affect the government's tax revenues. This paper tries to answer this question. For this purpose, by the theoretical foundations of the economy, we will specify a function for government tax revenues, where changes in population age structure is one of the explanatory variables. In this paper, by using of the method described by Ghysels, Santa-Clara and Valkanov in 2004, we estimated this function and anticipated government tax revenues.This study is done by using MIDAS method in order to estimate the specified for government tax revenues by the aid of R software.  MIDAS method allows variables with different frequencis, i.e., seasonal, monthly or weekly, put together this in one equation and it is possible to revise the forecasted value for the dependent low frequency variable as soon as new high frequency data are released. Hence the publiction of seasonal data for the variables considered, sach as government total revenue and gdp at the beginning of the year, will make it possible to forecast the government tax revenues. This forecast will help the policy makers to see is the budget will face some unbalances, relevant policy action be token from just the beginning of the year.
The statistical data used in this study are time series, seasonal, which is used to collect them from the database of time series of the central bank, economic indicators and the statistical center. Variables used: Government tax revenues in the form an annual, gross domestic product and total imports in the form season, age structure of the population in the form an annual.
Before estimating the coefficients of the model, the reliability of the variables has been investigated .The results show that in the equation specified, the effect of seasonal GDP and total imports, annual age structure (the ratio of population aged 35 to 64 to the total population) on government tax revenues are statistically meaningful. Given the positive impact of the age structure of the population aged 35-64 to the total population, it can be said that, according to Ando Modigliani's theory, since this age group has higher income, they pay more taxes and therefore have a positive impact on government tax revenues.
 To estimate this function, we used the relevant data in the period the first quarter of 1367 to the fourth quarter of 1392. Next, we forecast government tax revenues for 1393. To assess the predictive power of the model in outside of the estimation range, we did not used the data of 1393 in initial expressed relations estimation. The government's tax predicted revenues forecasted by the model is 709,365.7 b.Rials and compered to its real data which is 709,651.9 b.Rials, indicate that the model forecast is satisfactory. As can be seen, entering the data the fourth chapter of the seasonal variables used in the relationship, the prediction value is very close to the real value. Also, the coefficient of determination of the pattern is estimated at 0/9954, which indicates the high explanatory power of the model. The quantity of the test statistic is 0.77, which indicates that adverbs applied are statistically significant and sufficiently adequate. Regarding the quantity of the camera-Watson test statistic and Shapiro-Wilk's normal test, the disturbance Sentences of the pattern, are not correlated and have normal distribution.