Answer - Interest impact need to be seen with reference to two countries and the category of the countries can be developed or developing or underdeveloped. In case of developed countries, if the interest rate is already at a lower level then one may ask why the euro or yen is depreciating. The answer is that in case of developed countries the weight-age of the interest rate differential is low while other factors have higher weight-age on exchange rate. The model which has been worked out here is confined to India and the US exchange behavior and analyzed with respect to past data, and results of both countries which are tested with substantial success. We may notice that even in the case of the US and in India, the weight-age worked out and is shown in a pie diagram for long term. Another diagram for short term shows different result. It clearly shows that the impact of the variable is dynamic and not sacrosanct. This model cannot be universally applicable to other two sets of countries.
Answer - Government confidence, policy, and announcements do have an impact on the short or long term. However, this cannot be quantified. But in the long run, these policy impacts definitely reflect in the number of variables and are ultimately captured in the model. With regards to the confidence it gives with the perception, we can assign a scale 1 to 10 and allocate say 5% for this factor and make an adjustment to the final result. As all other non-quantifiable factors will have, say 15%, weight then this adjustment can be made to our result that we get from the calculations.
Answer - We have incorporated this factor while formulating mathematical and other models. Prior to 1991, our economy was closed economy and FDI and capital inflows were not much due to political and regulatory constraints. As we are aware that after opening up in 1991 (Liberalization, Privatization and Globalization), the real FDI flow started from 1995–6 and in our model these differentials are considered. This has improved the dependency and correlation further. The other factors such as foreign borrowing and repayments have not been considered as we see our foreign exchange reserves for past 6–7 years have been maintained at decent levels and the differential impact on (YoY) may not be significant.
Answer - The performance of other economics does impact the exchange rate of the country. However, the measure of impact differs upon the economic, trade, political relationship, and dependency on the other economics.
Answer - The impact of RBI intervention in the market needs to be considered. If the currency shows sudden appreciation or depreciation, then it is not good for the economy. The central bank closely monitors the import, export, balance of payment position, major outflow or inflow expected in near future, economic scenario, etc., and regulates the exchange rate from time-to-time either by selling or buying the currency from the market in the same manner as it manages the liquidity by manipulating CRR, repo etc. in case of rupee.
Answer - Lending interest rate considered in the model consists of base rate plus credit spread. Base rate calculation mainly consists of risk free rate, variable overheads and margin for bank. Thus calculation of lending rate consist of
a. Risk free rate (which only covers time value of money)
b. Credit spread covering credit risk
c. Banks operational margin if any.
On the other hand inflation covers change in the purchasing power of the population.
Interest rate majorly influences production pattern on the other hand inflation influences the consumption pattern.
Although interest rate is a tool for dealing inflation but the calculation of interest rate does not include the inflation as if we deduct the inflation figure from the interest rate the remaining is not equal to risk free rate plus credit spread.
Answer - Capital inflows and outflows also impacts exchange rates. However during the period 1980 to 1991, India was closed economy and capital inflows and outflows were very limited compared to post LPG phase. Post 1995 capital inflows have increased which we have captured in the form of FDI. Surplus in capital account is defined as money is flowing into the country from the foreign countries which indirectly implies FDI and FPI.
Apart from FDI, NRI bank deposits and other sources along with FDI should also be considered which is positive for India. However this data is not readily available. However, on the other hand, with same logic, profit repatriation of MNCs to foreign countries should also be considered which will negate the benefit of foreign currency reserves as they will buy dollar and then repatriate the same to the country of origin.
We can consider all the above data, however to get the exact details of all the data from 1980 to 2016 seems difficult.
Answer - In exponential smoothing we give the impact of seasonality. Lag variables can be used in order to smoothen the short term volatility in the data so that to make the data more immune to short term hick ups. We have given weights of 70% to the value derived by respective models of current year & 30% to the value of exponential smoothing of previous year.
The same is mainly due to the frequency of release of Economic Data. Most of the economic data releases Quarter on Quarter basis after some days of the end of the particular quarter. Thus first 30% of the period of current year's exchange rate is mainly affected by the Economic data as on previous year. Therefore we have given 70% weight to current year and 30% weight to previous year values.
Answer - In our model we have applied Moving Average of 4 years. (We have assumed 4 years economic cycle for the analysis of change of exchange rate.) With an assumption of weight age of 55% to current year’s data, 30% to previous year’s data, 10% to two years prior data and 5% to three years prior data.
The assumption is made mainly because of the fact that current year's data affects the exchange rate more than the previous year's data. As the time goes by, most of the impact of older data is already priced in and adjusted in the exchange rate. Therefore, we have assumed lesser weights for the past years' data compared to recent year's data.
Answer - The change in Exchange Rate is regressed with multiple Variables.
In this regression all the 7 factors which affect exchange rates i.e. Interest, Inflation, Growth, CAD as a % of GDP, Crude oil, FDI as a % of GDP and Per Capita Income are independent variables (X1, X2,.....,X7) and dependent variable is exchange rate i.e. Y.
Therefore the regression formula becomes
Y = a + b1X1 + b2X2 +.....+ b7X7
Thus all the 7 independent factors jointly and severally affect exchange rate. If you notice that after exponential smoothing and moving average along with normal mathematical model, we have worked out the regression in each case and lastly made a comparative table which gives fair accuracy with respect to R, R square, standard error etc.
Answer - The p-value for each term tests the null hypothesis that the individual coefficient is equal to zero (no effect). A low p-value (< 0.05) indicates that you can reject the null hypothesis. In this case 0.05 is the significance level therefore the confidence level is (1-0.05) i.e. 95%.
If we formed hypothesis test then -
Null hypothesis will be - The model doesn't properly forecast the exchange rate.
Alternative hypothesis will be - The model properly forecast the exchange rate.
A low p-value (< 0.05) i.e. high confidence level indicates that you can reject the null hypothesis i.e. you can reject the statement that the model doesn't properly forecast exchange rate. P value less than 0.05 implies that there is less than 5% chance that null hypothesis will prevail.
In the regression performed by us for past data as well as on the future forecast data, p-value of some coefficients is significantly greater than 0.05 which implies that the co-efficient individually doesn't properly indicates the changes in exchange rate.
However, in case of multiple regressions there are interrelations and interdependence between the coefficients. Thus the independent coefficients affects the dependent variables not only on a standalone basis but also as on a combination basis (may be jointly or may be in permutation and combinations or may be bunch of 2-3 variables simultaneously affecting the dependent variable).
Although p-value indicates statistical significance of individual co-efficient but it doesn't indicate the statistical significance of the model as a whole. This is the major limitation of p value in case of multiple regressions.
In order to overcome this limitation of interrelations and interdependence, we can consider significance F value which indicates that the model as a whole has statistically significant predictive capability or not. If significance F value (< 0.05) indicates that you can reject the null hypothesis.
A low f-value (< 0.05) i.e. high confidence level indicates that you can reject the null hypothesis i.e. you can reject the statement that the model doesn't properly forecast exchange rate. F value less than 0.05 implies that there is less than 5% chance that null hypothesis will prevail.
Although p values for individual co-efficient doesn't show the effectiveness, significance F value of regression is well below 0.05 in each of the regression that implies model as a whole is working properly and efficiently predicting exchange rate.
Thus some analysts also recommend ignoring p values for the individual regression coefficients if the overall significant F is within the limits.