Scribd
Upload
3 views

Karan AFA Assignment Presentation

The presentation discusses forecasting financial variables using ARIMA and VAR models, focusing on exports, imports, and forex reserves from 1960 to 2012. It covers the necessary conditions for both models, including stationarity and stability diagnostics, and presents forecasting outputs alongside actual results. The conclusion indicates that ARIMA performed better in 2013 and 2014, while VAR was more accurate in 2015.

Uploaded by

karan.shah2909
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
3 views

Karan AFA Assignment Presentation

The presentation discusses forecasting financial variables using ARIMA and VAR models, focusing on exports, imports, and forex reserves from 1960 to 2012. It covers the necessary conditions for both models, including stationarity and stability diagnostics, and presents forecasting outputs alongside actual results. The conclusion indicates that ARIMA performed better in 2013 and 2014, while VAR was more accurate in 2015.

Uploaded by

karan.shah2909
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 27

PRESENTATION ON

“FORECASTING THROUGH
ARIMA AND V.A.R”

BY:- KARAN SHAH


Roll No. 1011517039
OBJECTIVES

 To pick any 3 financial variables (at least 25


obs.)
 Run ARIMA & VAR on them
 Forecast using ARIMA and VAR
 Compare & Analyze results and provide
conclusion
VARIABLES CHOSEN

 Exports as a % of GDP
 Imports as a % of GDP
 Forex Reserve in $ Mn.

 Time Span:
Annual data from yr. 1960 – 2012
V.A.R (Vector Auto Regression)

 Vector Auto Regression is used when it is difficult to


identify, which among multiple variables, is a dependent
variable.
 For ex. Consumption, Expenditure and G.D.P of a country
over the time. Following are the conditions required to
run V.A.R and the corresponding results of our data:
◦ Time Series Data
◦ Stationarity
A) STATIONARITY
 The Stationarity is the property of a time
series data in which statistical properties
like mean, variance, auto correlation, etc.
are all constant over the time.
 For our data, this is achieved at first level
differencing, which can be seen from figures
on the next three slides:-
1) For Exports as a % of G.D.P
2) For Imports as a % of GDP
data
3) For Forex Reserve in $ Mn.
B) Stability Diagnostics and Lag Length Structure

 We do stability diagnosis to check if the


data is stable or not, which means whether
the data is affected by, different sample
(from same population), different sample
size or different time span or a longer time
span, or not. The lag length structure helps
us to find if the said data is stable or not.
At First lag
Lag Length Criterion
 The lag length criterion helps us to choose the
best lag for our model. The best lag means the
one that has minimum errors as compared to
other lags.
 The AIC and BIC (SC) values in the resultant
output gives us a better idea regarding the
errors of our model at different lags. The lesser
these value the better it is.
Lag Length Criterion
Output
At Second Lag:
I) Stability
II) Granger Causality/
Block Exogeneity Wald Test
 We run this test in order to know whether two or
more variables collectively, significantly affect
the remaining variable or not.

 The variables might not individually affect the


remaining variable, but when taken collectively
i.e. together they might significantly affect the
remaining variable.
Granger Causality/
Block Exogeneity Wald Test Output
G.C/B.E.W Test Contd…
III) Auto(Serial) Correlation

 The autocorrelation function is the correlation of


the residuals (as a time series) with its own lags.
 Autocorrelation is the degree of similarity between
a given time series and a lagged version of itself
over successive time intervals.
 Here, we test the degree of similarity in residual
series and the lagged version of residual series,
over successive time intervals
Auto(Serial) Correlation Output
IV) Variance Decomposition

 The variance decomposition indicates the


amount of information each variable
contributes to the other variables in the auto
regression.
 It determines how much of the forecast error
variance of each of the variables can be
explained by exogenous shocks to the other
variables.
Variance Decomp. Output
V) Forecasting Output
ARIMA (Auto Regressive Integrated
Moving Average) Model

 Conditions for applying ARIMA:-


 Data should be Time Series Data

 Stationarity must be achieved in the data (There

should be no auto correlation or partial auto

correlation between observations)


p, d, q table
ARIMA Forecasting Output
Comparison with Actual
Results
VAR
Export (% of GDP) Import (% of GDP) Forex Res. (U.S Mn.$)
YRS. ACTUAL EST. ACTUAL EST. ACTUAL EST.
2013 25.32% 25.43% 28.30% 32.02% $ 292,046.00 $328,963.90
2014 22.91% 25.86% 25.89% 32.55% $ 304,223.00 $335,833.40
2015 19.90% 26.29% 22.50% 33.08% $ 341,638.00 $342,702.90

ARIMA
Export (% of GDP) Import (% of GDP) Forex Res. (U.S Mn.$)
YRS. ACTUAL EST. ACTUAL EST. ACTUAL EST.
2013 25.32% 24.90% 28.30% 31.71% $ 292,046.00 $297,650.00
2014 22.91% 25.29% 25.89% 32.18% $ 304,223.00 $303,254.00
2015 19.90% 25.68% 22.50% 32.65% $ 341,638.00 $308,858.10
CONCLUSION
 From the above comparison, we can say
that for the yrs. 2013 and 2014 ARIMA
predicts better results and for the yr. 2015
V.A.R predicts better results

You might also like