Submitted by: Submitted by ttavtra
Views: 377
Words: 3075
Pages: 13
Category: Business and Industry
Date Submitted: 02/07/2011 01:35 AM
ID-nummer: 0836342
0837063
TERM PAPER
at Handelshøyskolen BI
Multivariate Statistics
Eksamenskode og navn:
GRA 60361 – Multivariate Statistics
Innleveringsdato:
26.11.2010
Studiested:
BI OSLO
Table of contents
Problem 1: Arm Race Theory....................................................p. 1
Problem 2: Singular Value Decomposition..............................p. 4
Problem 3: Exploratory and Comfirmatory Factor Analysis....p. 9
Problem 4: Drinking and Driving..............................................p. 13
Problem 5: Equivalent Models...................................................p. 17
Attachment.................................................................................p. 20
∞
Problem 1: Arms Race Theory
A) Based on these assumptions formulate a model for the armament expenditures of the two nations. Draw a path diagram of this model and write the corresponding equations in LISREL notations.
US-Expenditure
E(US) = α + β1USSRt-1 + β2USt-1 + μ
USSR-Expenditure
E(USSR) = α + β1USt-1 + β2USSRt-1 + μ
B) Use the number of nuclear explosions shown in Table 1 as a proxy for armament expenditures; estimate model A using LISREL. Estimate the model separately for the periods 1950-1958 and 1962-1973 and compare the two models. To what extent are the models the same or different for these two periods?
We use a simple regression model in LISREL to estimate two separate equations for US and USSR. There are no latent variables in this system, only observable. This is a time series model. Although it is possible to estimate a consistent model using OLS, there might be problematic issues regarding forecast using OLS. OLS is more suited to estimate models using cross sectional data. We believe GLS would be a better method of estimating this model.
Estimation for period: 1950-1958:
US Expenditures
E(US) = 9.509 + 5.274USSRt-1 - 0.647USt-1 + μ
USSR Expenditures...