Econometrics for dummies, by roberto pedace (john wiley sons, inc , hoboken

Econometrics for Dummies by Roberto Pedace
(John Wiley & Sons, Inc., New Jersey, 2013), pp. xvi + 342
Hoa Thi Minh Nguyen1
February 2015

Econometrics for Dummies aims to provide students with a ‘short and simple version
of a first-semester course in econometrics’. The book covers the basic parts of a standard
undergraduate econometrics textbook, and is a good reference text in that sense. But
more so, and following the style made popular through the Wiley “for Dummies” series,
this book gives priority to simplicity in presentation and explanation to make econometrics more manageable. This feature alone should help draw the attention of students who
are learning econometrics for the first time.
Econometrics for Dummies is organized much like a standard undergraduate econometrics textbook, with 19 chapters grouped into 7 parts. It begins with a review of
key concepts in probability and statistics, and then discusses the classical linear regression model (CLRM), its assumptions and the use of the Ordinary Least Squares (OLS)
estimator — the most popular estimation technique in econometrics. Discussion and
treatment of violations of CLRM is provided in part IV, followed by an introduction to
the Maximum Likelihood Estimator (MLE) applied to qualitative and limited dependent
variable models. A brief introduction to models used for time series, pooled cross-section
and panel data is provided in part VI before an useful conclusion that outlines good and
bad habits in doing empirical research closes the book.
1

Crawford School of Public Policy, Crawford Building (132), Lennox Crossing, Australian National
University, Canberra, ACT 2601, Australia. Email:

1


The main strength of the book is its ‘straightforward manner’ which simplifies econometrics so that it is accessible for beginners. Padace achieved his stated purpose of the
book by using simple language, a ‘busy readers–oriented’ writing style, coupled with
stand-out icons such as “Remember”, “Warning”, “Tip”, etc. to help readers manage
and skim the book easily. The guide on the use of STATA along the way is a useful bonus

to facilitate application of econometric techniques. The writing style and presentation of
this book is in stark contrast to many typical econometrics textbooks which tend to exhaust students’ desire to learn econometrics through an emphasis on more sophisticated
technique.
However, providing simple and straightfoward tips in the field like econometrics is a
real challenge. While Pedace’s effort is laudable, some substantive errors such as the ones
below are a true drawback.
1. Confusion in key concepts:
(a) Estimator versus estimate: The former is a rule for combining data to
produce a numerical value for a population parameter while the later is a
numerical value obtained by applying an estimator to a particular sample of
data. It is crucial that students can distinguish between the two concepts,
which are typically explained in a thorough manner in standard econometrics
textbooks. However, in Econometrics for Dummies, the explanation here is
simply wrong. For example, ‘When you calculate descriptive measures using
sample data, the values are called estimators (or statistics)’ (pg. 40, under
the icon “Remember”); ‘The calculus and .... result in easy-to-use formulas for
calculating the regression coefficients (estimates of the slope and intercept)’
2


(pg. 80, under the icon “Remember”).
This confusion is not simply a typo and may explain why the same notation
is used for estimators and estimates throughout in the book (for example, pg.
77, 80-81).
(b) Population regression models (PRM) versus sample regression models (SRM): Explanations on PRM and SRM are confusing. For example,
PRM is explained as: ‘... the relationship you’ve assumed in ..... may contain
errors when a specific observation is chosen at random from the population. This is known as the stochastic population regression function ...’ (pg.
65); while SRM is explained as: ‘In most applications, ... you’ll need to ...
work with sample data to estimate your PRM...’ (pg. 67); ‘When using crosssectional data, you assume that the observations represent a random draw
from your population of interest’ (pg. 69).

2. Lack of precision or potentially misleading materials:
Durbin-Watson (DW) test versus Breusch-Godfrey (BG) test: Padace
describes the key difference between the two tests such that the former is for AR(1)
and the latter is for AR(q) where q is ‘some number greater than or equal to 1’
and ‘known as Durbin’s alternative statistic’ with q=1’ (pg. 221). In fact, the
BG test is superior to DW not only in the order of autocorrelation but also for
its application in models where the lag(s) of the dependent variable is (are) used
as an independent variable(s) [Breusch(1978)]. Furthermore, the DW test requires
satisfaction of CLM assumptions and can be used with small sample data while such
satisfaction of CLM assumptions is not required for a BG test, an asymptotically
3


justified test [Wooldridge(2012)]. Given the substantial practical disadvantage of
DW test, the suggestion that DW test is ‘the most popular test for autocorrelation’
can be misleading.
The least squares principle: ‘The least squares (LS) principle states that the
SRF should be constructed (with the constant and slope values) so that the sum
of squared distance... is minimised.’ (pg. 76). This strong statement could be
misunderstood to imply that a model has to have both a constant and slope(s).
While it is rare to have a model without a constant, the LS principle, which aims
to minimise the squared residuals when fitting a model to data, is not violated in
this case.
Justification of OLS technique: The author apparently over-emphasises the
properties of OLS solutions in terms of its popularity in econometrics (pg. 76).
These properties include things such as the regression line always passes through
the sample means; the mean of the residual is zero; the residuals are uncorrelated
with the observed values of independent variable(s), etc. While those properties
are good, OLS is popular largely for meeting criteria that are considered important
for most researchers in estimator selection including unbiasedness, consistency and

efficiency when certain assumptions are satisfied, in addition to computational ease.
Failure to understand these estimator selection criteria would prevent students from
understanding why there is a need for alternative estimators such as Method of
Moments and MLE.
Estimate and interpret coefficients of logit and profit models: Coefficient
estimates from logit and probit models might have been explained with more care
4


due to non-linearity of these two models. For example, results on coefficient estimates are shown in Econometrics for the Dummies using the Stata post-estimation
command ‘mfx’. While this command is handy, it would be very useful for students
to know that it calculates the marginal effect at sample means (the default option)
and that results generated would be different if the effect is evaluated at different
values of independent variables, all due to the non-linearity of the logit and probit
models.
3. Lack of consistency in notation:
Some rules in notation set at the beginning of the book are not followed. For
example, the rule set on page 77: ‘Notice how the mathematical representation
of the SRF uses hats (ˆ) above the coefficients and error term. I use this symbol
to denote that these numbers are estimates of their true population values’ is not
followed in page 227.
Explanations on notation are also missing at times. For example, what d stands for
in the formula for Thiel’s estimator (pg. 224)?
Overall, Econometrics for the Dummies is a student-friendly reference book. Padace
has achieved his stated purpose of the book, namely to make a hard subject like Econometrics manageable for beginners. Should more attention be paid to substance and the
language used to give advice and suggestions to students? I have no doubt about this.
If done, hopefully in the next edition, I also have no doubt that this book would win the
hearts and minds of many students who are starting to learn econometrics.

5

References
Breusch(1978). Breusch, T. S., 1978. Testing for autocorrelation in dynamic linear
models. Australian Economic Papers 17 (31), 334–355.
Wooldridge(2012). Wooldridge, J., 2012. Introductory econometrics: A modern approach. Fifth Edition. Mason: South-Western Cengage Learning.

6