Homework 3 Solutions

Ηλεκτρονική - Συσκευές

10 Οκτ 2013 (πριν από 4 χρόνια και 9 μήνες)

96 εμφανίσεις

UNC
-
Wilmington

ECN 321

Department of Economics and Finance

Dr. Chris Dumas

1

Homework 3

Solutions

1)
Economics is the study of the rational allocation of resources under constraints to meet objectives.

2
)
The word "data" is plural. Data are
measured,
recorded

observations
.

3)
A theory is a pattern believed to be observed in d
ata. A model is a formal, precise statement of a
theory. The three elements of a model are: variables, parameters and operators.

4)
Inductive Reasoning is

the process of finding a pattern/theory/model from

data. Deductive reasoning
is the process of

finding the implications/forecasts/predictions/theorems that come from a model

5)
The
ceteris paribus

assumption is the modeling assumption that everything in the world outside the
model is assumed to remain constant.
Each thing left outside the model
is assumed to remain constant
where it was when the data were collected for the model.

6)
Ockham's Razor is the

model
building principle
that
states that
unimportant details should be "cut
away," or discarded, from a model. In practice, applying Ockham'
s Razor means that a model
-
builder
should initially attempt to create a simple model that
capture
s

only
the key
variables in

the situation
under study.

7)
Statistical Regression Analysis is typically used to determine the best values for the parameters in

a
model.

8)
In mathematical models, operators are the "plus signs, minus signs, multiplication symbols," etc. in the
equations of the model.

The operators in a model are determined by using Nested Modeling.

9)
L
inear

relationships are important becau
se they are
common
; however,

it is
almost always
more
important to correctly identify and analyze
nonlinear
(rather than linear) relationships in a model,
because nonlinear relationships are most often the
source

of
surprising

model results, such as
"explo
ding" relationships among variables.

10
)
Theorems are results
/forecasts/predictions

that are derived from theories
/models
; in other words,
theorems are the implications of theories
/models
.
Theorems are used (1) as forecasting/prediction tools,
and (2) t
o test models/theories by comparing the theorem predictions against new data.

11)
A Hypothesis (plural Hypotheses) is a pattern in the data that
is suspected to

indi
cate a true
relationship among

variables.
A Spurious Hypothesis is a hypothesis based on
a pattern in the data that is

discovered to have occurred by chance and that does not reflect a true pattern among variables.

12)
Statistical hypothesis testing

is used in two ways: (1) to test the reliability of the parameter estimates
in the model, an
d (2) to determine whether "real world" observations are so far from the theorems
(predictions) derived from a model that the model should be revised. [In (2),
Statistical hypothesis
testing

methods are used to calculate Confidence Interval numbers for t
heorems/model predictions. If
real
-
world observations are inside the Confidence Interval numbers, then we continue to have confidence
in our model. If real
-
world observations are outside the Confidence Interval numbers, then we lose
confidence in our mo
del, and it's
"
Back to the drawing board!" to revise the model.]

UNC
-
Wilmington

ECN 321

Department of Economics and Finance

Dr. Chris Dumas

2

13)
Economists use "Natural Experiments."

Such experiments are called "natural" because they occur
by chance; they are not "set up" by scientists
.
As an

example
, consider a situation in w
hich all factors
that affect the purchasing behavior of two groups of consumers are the same, except for one factor, say,
price.

This situation is a natural experiment.

Differences in behavior between the two groups
(differences in quantities consumed, f
or example) may be explained by differences in the one factor that
differs between the groups (in this example, price), because all other factors are the same for the two
groups.

14
)
Sensitivity analysis determines the sensitivity of model results (the d
ependent variable) to changes in
the values of model parameters.