Form and functions

A guide to technical writing in economics

by

Emmanuel S. de Dios

2003
Quezon City

1
Table of contents
1. Introduction 1

2. Paper, font, and layout 2

3. Authorities 8

4. Quotation and punctuation 12

5. Footnotes 16

6. References 18

7. Voice and tense 23

8. Mathematical writing 27

9. Statistical tables 38

10. Diagrams and graphs 43

11 Envoi 51

12. Acknowledgements 51

13. References 52

Appendix: Frequent mistakes and disputed usage 53

2
1. Introduction
This booklet has been written primarily to help students in the last stages of writing
your final papers, theses, and dissertations in economics. But students and
professionals in other fields could find it useful as well.

Writing in economics and a number of other fields today requires a familiarity with a
various methods of exposition, including prose, mathematics, the presentation of
econometric and statistical tables, as well as diagrams and graphs. Each of these has
peculiar conventions; yet most students are hard-put to find a single source that
gives reliable advice on different situations. The result is a good deal of guesswork
and improvisation on the student’s part about proper form, leading to unevenness or
plainly bad style. This booklet seeks to fill some of that gap. The advice it gives is
necessarily wide-ranging, but this work makes no claim to exhaustiveness. The
selection of topics simply reflects the author’s judgment regarding matters that
ought to concern writers of technical papers.

The assumption throughout is that your aim as a writer is to convey your thoughts as
directly and as precisely as possible, eliminating unnecessary difficulties for readers.
Whether this aim is in fact attained will depend on many other factors, of course,
including the expository skills you have acquired thus far, the breadth of your
knowledge, and your subjective disposition. Logical consistency, lucidity, eloquence,
and wit will take many years to attain, if at all, and this manual does not pretend to
give you a magic formula for them. Nonetheless, you could certainly facilitate
comprehension of your work by avoiding egregious deviations from conventions
regarding form, language, and reasoning, and if the visual appearance of your work
helps the orderly working of your reader’s mind. Literary writing might frequently
tolerate and at times even encourage deviations from norms. The usual purpose of
technical writing, however, is not to call attention to itself but to focus the reader’s
attention on the subject being discussed, which is often difficult and complex
enough. For this reason, as a technical writer, you may wish to be aware of and
observe standard usage and formats expected in your work.

This is not a manual about printing or typography either, although it will often
emphasize the importance of visual appearance to the extent that it aids clarity and
reassures the reader. The purpose is not to instruct you in all the finer points of
selecting particular fonts, or of designing and laying out the final printed page of the
volume or journal in which your work appears. After all (think of the division of
labor) it cannot be your aim to supplant the printer or the book designer.

1
Nonetheless, scholars would be well-served if the iterated transactions between
themselves and their readers, evaluators, and editors could be minimized.
Misunderstanding is especially likely where both the reading and publication of
technical work is still in its infancy and fraught with high transactions costs for
writers of technical manuscripts. Again the contribution of this booklet will be
indirect: if writers themselves can master the most important conventions, and if
these can be set down sensibly for evaluators and publishers to examine, then the
pain and effort involved in putting out technical pieces would be greatly reduced all
around.

At the outset, however, two cases should be distinguished: the first is the final draft
of a paper, thesis, or dissertation that is to be submitted for academic evaluation.
Here the final draft the writer prepares is the final form readers will see. Hence, a
greater demand is placed on the author to make the copy accurate and reader-
friendly.

Quite different rules apply, however, to a paper being prepared for submission to
journals or to publishing houses. Such cases, it will surprise some to know, call for
less rather than more formatting on the author’s part, since the final copy will in any
case be recast by the journal or publisher using its own conventions or house-style.
In the rest of this manual, the icon
will be used to call attention to suggestions
that apply only to submissions to journals or publishing houses.

2. Paper, font, and layout

Preparing a final draft begins with choices that appear mundane and even
innocuous. But such details as the size of paper to use, the typeface, and the physical
ordering of the contents of the work bear perhaps small but significant consequences
that pervade the entire work and determine its overall or “gross” look. Paying
attention to such aspects will suggest a deliberate mind and a careful eye with a
concern for the readers’ ease that is more likely to be rewarded with their attention.

2.1 Paper size and margins. For most types of work considered here, either of two
standard sizes of bond paper may be used: (a) the international standard DIN A4,
which is paper1 measuring 210 mm. by 297 mm. (or 8.27 in. by 11.69 in.), or

1DIN stands for Deutsche Industrie Normenauschuss, the standards-setting body in Germany.

2
(b) “letter-size” paper, which measures 8½ in. by 11 in. (or 215.9 mm. by 279.4
mm.).

For papers that will not be bound, including term papers, leave uniform white space
of 32 mm. (about 1¼ in.) on either side of the main text, and 25 mm. (about 1 in.)
above and below. These margins should not be exceeded even by the text of the
footnotes. For final copies of dissertations and theses destined for binding, increase
the left margin to 40 mm. (ca. 1½ in.). Use only one side of each sheet in all cases.

2.2 Font size and type. If your word-processor will permit it, use a font that is either
11 or 12 points in size for the main or “body” text. This roughly corresponds to the
pica size in typewriters. Two broad categories of typeface may be distinguished:
those with serifs and those without (“sans-serif”). Serifs are the small bases on which
the vertical elements of “m” rest, as well as the backward-pointing slabs on the “i” or
“l”. Sans-serif typefaces (m and l) dispense with these embellishments. Times Roman
and Courier are popular examples of serif types; on the other hand, Gothic and
Arial are sans-serif types.

Opinion is almost universal that serif types are easier to read than sans-serif types2.
If a choice is at all possible, therefore, compose the main text of your work using a
serif typeface. If you wish, you may still use sans-serif types to set off headings, titles,
and captions.

Again, your word-processor permitting, use the following font sizes for the purposes
indicated:

Title on front page 20 points, boldface

Chapter heads 16 points, boldface

Chapter subheads 16 points

Normal text or body text 11-12 points

Super- and subscripts 9-10 points

Indented quotations 10 points

Footnote reference numbers 8 points
Footnote text 9-10 points.

2Tufte [1983:183]) quotes Albers as saying that “ophthalmology has disclosed that the more
the letters are differentiated from each other, the easier is the reading”, so that “the
fashionable preference for sans-serif in text shows neither historical nor practical sense.”

3
2.3 Front matter. In dissertations, theses, and final undergraduate papers, the
following introductory material should be presented in the order indicated:
(a) a title page containing the full title of the work, the author, the date of
submission, and the relevant version of the following explanatory phrase in
normal text:
A [paper/thesis/dissertation] submitted to the
University of the Philippines School of Economics
in partial fulfillment of the requirements for
[Econ 199/the degree of Master of Arts in Economics/
Master in Development Economics/Doctor of Philosophy in Economics]

(b) an abstract on a separate page;

(c) a signed waiver on a separate page that reads as follows:
Relevant portions of this work may be quoted and used for
research and other scholarly purposes, provided the proper
citation is made.

(d) a table of contents; and

(e) a list of figures and tables, if relevant.

A page of acknowledgements is optional and may be placed between the abstract and
the waiver. Some students seem to regard the acknowledgements page as an
opportunity to indulge in effusively personal and idiosyncratic messages. This urge is
understandable, especially after the long struggle to complete the dissertation
(respectively, thesis or final paper). Nonetheless, it is an urge that must be resisted.
In keeping with the rest of the work, the tone used in the acknowledgements is best
kept simple and professional, and the section itself must be quite brief. Personal
messages are better expressed in a separate note – say, hand-written on a copy of the
paper or thesis sent to the person being thanked. To paraphrase Madame de Stäel,
do not include in books matters that belong only in conversations, nor in
conversations, matters that belong only in books.

For term papers and shorter work, the elements (b), (c), (d), and (e), as well as the
acknowledgements page, may be omitted.

For journal submissions, omit (c) and (d) but include a simple title page (a) with
the author’s name, affiliation, and address; include an abstract (b) on a separate
page, and below it, on a separate line, after the phrase “Key words:” write in at most
five (5) important terms or concepts that best indicate the content of your paper.

2.4 Chapter heads, and subheads. For dissertations, theses, and final undergraduate
papers, use the following conventions:

4
 Chapter heads: I, II, II, and so on. These should be centered at the top of the
page, followed by the title of the chapter on a separate line.
Subheads: 1. , 2. , 3. , and so on. Write the subhead text flush left on the same
line.
Sub-subheads: 1.1, 1.2, 1.3., and so on. Write the text flush left on the same
line.
Shorter papers, including most journal submissions, may dispense with chapter
headings; use the convention for subheads that uses Arabic numerals instead.

2.5 Text and headings. Avoid situations where the numeral for a superior heading is
followed immediately by a numeral for a subheading (Ex. 1(a)). Text must always
introduce the next set of headings (Ex. 1(b)).
Example 1
(a) awkward
2. Price differentials
2.1 Price discrimination
A straightforward application of the standard theory of price
discrimination to the market for widgets suggests that the profit-
maximizing widget-maker with market power will charge different
prices...

(b) improved
2. Price differentials
There are two possible sources of differentials in widget prices:
price discrimination and cross-subsidization. ...

2.1 Price discrimination

A straightforward application of the standard theory of price
discrimination to the market for widgets suggests that the profit-
maximizing widget-maker with market power will charge different
prices...
2.6 Paragraph spacing. For the typical paragraph of body text, use spacing of one-
and-a-half (1½) lines. Clarity and the provision of enough space for possible
marginal comments by advisers, readers, or referees are the main purposes of
spacing. But there is no need to consume more paper than needed.

Paragraphs should be separated by two (2) lines of space. This means adding about 8
points of spacing before a new paragraph, in addition to the 1½ lines of spacing
within paragraphs. (Two-line spacing between paragraphs can be maintained even
when body text is single-spaced.)

5
2.7 No indentation. The increased line-spacing between paragraphs is sufficient to
mark off one paragraph from another, thus eliminating the reason for indenting the
first line of each paragraph. This argument applies a fortiori if you are preparing a
manuscript with single-spaced body text.

2.8 Displayed equations. Displayed equations must be indented one tab-stop

(13 mm. or about ½ in.), from the left margin. If, as is typical, the equation forms
part of a paragraph, then the space between it and any other part of the paragraph
will also be one-and-a-half lines, as will be the space between any two equations. In
Ex. 2 below, the text and the equations belong to the same paragraph, so that the
spacing is consistently one-and-a-half lines.
Example 2
The time-paths of the variables Y, K, and L can be obtained easily, since all
variables grow at the rate n in equilibrium. ... The solution, k = k0 for any
root k0 of (5), yields steady-state growth:
Y = y0L0 exp(nt) K = k0L0 exp(nt) L = L0exp(nt),
where n is the growth rate of the labour force.

Obviously when a previous paragraph ends in an indented equation, the first line of
the succeeding paragraph must be separated by two lines as in Ex. 3.
Example 3
The time-paths of the variables Y, K, and L can be easily obtained, since all
variables grow at the rate n in equilibrium. ... The solution, k = k0 for any
root k0 of (5), yields steady-state growth:
Y = y0 L0 exp(nt) K = k0 L0 exp(nt) L = L0 exp(nt).

Only equation (5) remains to be interpreted, since it is the basis of the

steady-state solution. Since y = Y/K, the steady-state growth solution exists,
if and only if the initial capital output ratio is n/s....

In Ex.3, the space between the line of equations and the preceding lines of text is
smaller than the space between it and the succeeding line of text. This is because the
equations form part of the first paragraph and not the second. In your word-
processor, this effect is achieved by first writing the equations as a separate
paragraph and setting the paragraph-spacing equal to 0 points.

2.9 Spacing after punctuation marks. Use no more than a single letter- space to
separate words from preceding or succeeding punctuation marks. This goes for the
space between sentences as well.

6
2.10 Justification. Almost all word-processing programs now contain the option of
aligning the body text flush with both margins, that is, the margins on both sides of
the page are observed exactly. This practice, called “justification” in the printer’s
trade, results in a manuscript whose appearance approximates that of the final
printed page in a book. In particular, “right- justification” refers to the exact
alignment of text with the right margin of the page, apart from the more usual left-
justification.

You should take time to consider, however, that (particularly right-) justification
carries with it some additional tasks. The effect of justification is to force the lines of
a paragraph to occupy the entire space between margins, no more, no less, a fact that
is not without its own problems.
Example 4
(a) Agreeing with von Hayek, Herbert Simon [1981:42]
considers the most striking characteristic of the
market mechanism to be “the way it reduces and
localizes informational and computational
requirements”.

(b) Agreeing with von Hayek, Herbert Simon [1981:42]

considers the most striking characteristic of the
market mechanism to be “the way it reduces and
localizes informational and computational
requirements”.

(c) Agreeing with von Hayek, Herbert Simon [1981:42]

considers the most striking characteristic of the
market mechanism to be “the way it reduces and
localizes informational and computational require-
ments”.

As may be seen from Ex. 4(a), right-justification has led to unappealing and
distracting white spaces, especially in the lowermost lines, a phenomenon Donald
Knuth calls “badness”. The example also suggests possible ways of dealing with it.
The problem clearly occurs when the chosen font is too large for the length of the
typical line, which is particularly likely to occur in column formats, or when margins
are wide. (The margins of the example have been artificially narrowed to illustrate
the problem.)

The paper size, margin-widths, spacing and fonts suggested above should minimize
badness. If it should occur nonetheless, possible solutions include (i) picking a

7
smaller font, (ii) staying away from right-justification altogether (Ex. 4(b)), or (iii)
syllabicating some long words (Ex. 4(c)). As Ex. 4(b) shows, however, even (ii) will
not always free you completely from the need for syllabication. While avoiding right-
justification removes white spaces in the middle of lines, long, non-syllabicated
words can still lead to distracting white space on the right. The best advice is
therefore to give your work a final once-over and, with the help of your word-
processor or a good dictionary, to hyphenate long words on the right in order to
minimize uneven white spaces.

2.11
Particularly for journal submissions, use neither right-justification nor

hyphenation. Indeed many journals will explicitly ask you to refrain from doing so.
The reason is that it imposes extra work for the publishers who must then undo any
formatting you have adopted in order to make it conform to the journal’s house style.

2.12 Widows and orphans. A “widow” is the last line of a paragraph that stands
alone at the top of a page. An “orphan” occurs when the beginning line of a
paragraph stands alone at the bottom of a page. Neither is pleasing to the eye, and
both should be avoided. When widows occur, try to include two or three more
preceding lines from the preceding page, or economize on some paragraph spaces on
the preceding page to push the widow up. Orphans, on the other hand, may be forced
onto the next page with the rest of its paragraph; alternatively, delete some spacing
on the current page to pull up some extra lines. Most currently available word-
processing software will allow you to set defaults to avoid widowed and orphaned
lines.

2.13 In widely-used word-processing software, it is possible to define “style-sheets”

so that the formats suggested here are automatically applied.

3. Authorities
Various conventions for citing authorities exist, but in economics and most other
social sciences, the general trend has been away from the tedious and repetitious
conventions of the past and towards a reasonable compromise. The main purpose of
citations is to assist the reader in locating and referring to earlier work, but the
system obviously breaks down if too large a burden is imposed on the writer who
must supply the information. When transactions costs are too high, supply and
demand fail to meet and no market emerges. Past conventions required writers to
master the finer points of using loc. cit., op. cit., opp. cit., idem, ibid. vide supra, and
so on, a system that was more of a punishment to both the writer and reader. The

8
following adopts a system built on conciseness and clarity regarding the most
essential information.

3.1 In citations within the main body of your work (body text), write the author’s
surname and the year of publication, the latter being separated by square brackets.
Should you wish to cite the page as well (something you should always seek to do),
place the numerals between the brackets and after year of publication, separating the
two with a colon without a space
(Ex. 5).
Example 5
Arrow [1959:145] writes that …

Some publications adopt the following styles:

Arrow [1959, p. 145] writes that …
Arrow (1959, p. 145) writes that …

Using a colon, as suggested here, is obviously more efficient, however, since it saves
on space without sacrificing clarity. The use of the square brackets also has a reason.
First, it calls sufficient attention to the citation. Second, however, you will sometimes
need to cite some author’s work within a parenthetical remark, and you will find that
the use of square brackets avoids the ambiguity otherwise caused by two sets of
round parentheses (Ex. 6):
Example 6
It has long been recognized that preferences are most generally represented
as vectors. (This was first pointed out by Chipman [1960].)

The wisdom of using square brackets becomes obvious if you try to rewrite the
preceding using round parentheses instead. In general, try to avoid placing two
identical separators adjacent to one another.

A system somewhat similar to that proposed here is found in van Leunen [1978],
who suggests using bracketed numbers in bold face type, rather than authors’ names
or publication year in referring to works. The numbers are then made to correspond
to the order in of references at the end of the work. Hence, instead of “Arrow
[1959:145] writes…”, she would recommend “Arrow [2, p. 145] writes…”. This
system, however, has the disadvantage of making the reader to flip back and forth
constantly between main text and reference list. Still, this is a system used in some
venerable publications, including the Economic Journal.

9
3.2
Some journals and publishers will explicitly require you, even upon first

submission, to adopt the house-style, which may differ from the foregoing. You
must, of course, comply.

3.3 When more than one authority is to be cited, first enclose the entire enumeration
of authorities in round parentheses; then enclose the respective years and page
numbers in square brackets. Separate the sources enumerated using semicolons.
Example 7
The usual law of supply and demand can be thought of as a minimization of
the aggregate value of excess demand (Debreu [1951]; Arrow and Debreu
[1954]).

3.4 Quotations inlaid. Use quotation marks following the suggested rules for
placement (rule 5 below) when quoting phrases or sentences inlaid in the main body
the work.
Example 8
(a) Arrow [1969] points out that “market failure is not absolute; it is better
to consider a broader category, that of transactions costs”.

(b) The most striking characteristic of the market mechanism for Simon is
“the way it reduces and localizes informational and computational
requirements” [Simon 1981:42], an opinion already voiced earlier by the
Austrian economists.

The citation reference should occur immediately after the closing quotation mark
and before any succeeding punctuation mark.

3.5 Quotation marks. Double quotation marks (“...”) are used to separate inlaid
quotations. Whenever word-processing possibilities allow, the initial quotation
marks must be distinguished from the closing quotation marks. Initial quotation
marks have the appearance of two inverted commas (“), while the closing marks look
like normal commas (”). A quotation within a quotation, on the other hand, is
separated by single quotation marks (‘...’) (Ex. 9). Not all fonts distinguish between
opening and closing quotation marks, and it is advisable to select a font that does.
Most word-processing programs, however, offer a facility for inserting these “smart
quotes” automatically.
Example 9
Georgescu-Roegen thought that as a result of the oil embargo, “it was
natural that scarcity should again become an important issue, although not
for the school that believes that ‘the market knows best’ ”.

10
Note the two sets of quotation marks at the end of Ex. 9: the single quotes refer to
the phrase “the market knows best”, while the double-quotes refer to the longer
extract, “it was natural...”. The period falls outside both closing marks, since it ends
the main sentence and refers to neither quoted fragment.

3.6 Long quotations displayed. Long quoted passages of five lines or more must be
set off as single-spaced paragraphs indented one tab-stop (130 mm.) from the left
margin. Displayed quotations require no quotation marks. Whenever the word-
processor permits it, the displayed quotation should be written using a font two
points smaller than normal text. Hence if the normal text uses an 11-point font, the
displayed quotation should be written in 9-10-pt. font. If the source is appended to
the end of the displayed extract, it must be set before the full stop (Ex. 10).
Example 10
Keynes sought to find support for his “general theory” among earlier
writers. His sympathy for the mercantilists, for example, is evident in the
following:
At a time when the authorities had no direct control over the domestic rate
of interest or the other inducements to home investments, measures to
increase the favourable balance of trade were the only direct means at
their disposal for increasing foreign investment; and, at the same time, the
effect of a favourable balance of trade on the influx of the previous metals
was their only indirect means of reducing the domestic rate of interest and
so increasing the inducement to home investment [Keynes 1935:336].
By citing the mercantilists favorably in this manner, Keynes also broke with
the prevailing free-trade orthodoxy.

Alternatively, the source may be included immediately after the author’s name is
mentioned in the main text (Ex. 11).
Example 11
In arguing that a social planner is both advocate and executor, Herbert
Simon [1981:175] likens him to an architect and, in this connection,
narrates an encounter with the Bauhaus exponent Mies van der Rohe:
I once asked Mies van der Rohe ... how he got the opportunity to build the
Tugendhat house -- a startlingly modern design at the time of its
construction. ... “Wasn’t the client shocked”, I asked, “when you put before
him your glass and metal design?” “Yes,” said Mies, viewing the tip of his
cigar reflectively, “he wasn’t very happy at first. But then we smoked some
good cigars, ... and we drank some glasses of a good Rhein wine, ... and then
he began to like it very much.”

Note from Ex. 11 that since the displayed quotation did not need to be separated with
quotation marks, quotations within it are separated by double quotation marks.

11
4. Quotation and punctuation
Most conventions related to punctuation marks are too well-known to bear
repeating. What follows reflects our judgment regarding possible problematic areas.

4.1 Punctuation in general. For most scholarly writing, periods (full stops) should
suffice to end sentences. Question marks are used for the odd rhetorical question,
while exclamation points are rarely if ever needed. Deviations from this general rule
are allowed as a matter of personal preference and style. Hence, the occasional use of
the semicolon (;), the colon (:), and the dash (–).

4.1.1 Replacing the full stop by a semicolon (Ex. 12) indicates that two otherwise
complete sentences (independent clauses) are somehow related, typically equal in
importance, or need juxtaposition.
Example 12
The high-interest rate policy sought to stave off further attacks on the peso;
it succeeded only in punishing domestic industry.

4.1.2 Replacing a full stop by a colon, on the other hand, implies a subordination of
the succeeding clause. The second clause is then regarded seen as a further
explanation or elaboration of the first (Ex. 13).
Example 13
Keynes did not reject all “classical” concepts: he continued to accept the
relevance of the demand curve for labor.

A simple test to use is the following: when two clauses are equally important, use a
semicolon; when the second clause amplifies on the first or gives specifics, use a
colon.

A colon may also be used to introduce an enumeration.

Example 14
Money has three main functions: store of value, unit of account, and
medium of exchange.

4.1.3 Elements in an enumeration should be separated by commas (Ex.15) and, when

that does not suffice, by semicolons. Semicolons may be necessary when some
elements of the enumeration themselves involve commas (Ex.15d). We are well
aware that some authorities advise dispensing with this rule, but inserting a final
comma or semicolon before “and” or an “or” in an enumeration is never wrong and is
indeed sometimes essential for clarity. To convince yourself of this, try rewriting the
Examples 15(b) to 15(d) without the final commas or semi-colons.

12
 Example 15
(a) At least three factors have been adduced to explain profit:
uncertainty, time preference, and entrepreneurial returns.

(b) Full equilibrium, therefore, rests on three factors: technology,

population and resources, and preferences.

(c) The negative relationship between aggregate demand and the price
level can be mediated through any or all of the following: the real-
wealth effect on consumption, the Keynesian interest-rate
mechanism working through investment, and the real exchange rate
or Mundell-Fleming effect.

(d) Three groups of sets are involved: (a) X, Y, and Z; (b) V ∩ W; and
(c) X/I and Y/I.

4.1.4 In any enumeration, make sure that line breaks do not separate the ordinal
number or letter (e.g., i., ii., iii., ... or a., b., c.,...) from the item it is supposed to
identify. Thus in Ex. 15d, a line break has been forced before “(c)” to keep it together
with “X/I and Y/I”.

4.2 New terms. It is sometimes necessary to introduce some unfamiliar technical

terms or expressions which have been used in other contexts or by other writers. The
first time these are mentioned, they should be set off by quotation marks. Upon their
second and subsequent mention within a shorter paper, however, the quotation
marks should be omitted (Ex. 16).
Example 16
“Moral hazard”, a term borrowed from the insurance literature, refers to the
proneness to risky behavior induced among over-insured clients. In a
standard problem of moral hazard, the principal is unable to observe the
agent’s behavior directly and the principal must instead devise an incentive
scheme to induce ...

4.3 Punctuation in relation to quotation marks. This may be one of the most
involved topics in this booklet. Two traditions in punctuation currently exist, both of
which are acceptable: the “logical” and the “conventional”. (The latter is actually
conventional only in the U.S.) Having chosen one or the other system, however, you
must adhere to it throughout. The difference has to do mainly with the position of

13
quotation marks relative to other marks of punctuation, e.g., full stops, commas,
exclamation points, and question marks.3

The logical system, which is that recommended here, inserts punctuation marks
relative to quotation marks according to sense. Hence, a period, an exclamation
point, or a question mark is placed before a closing quotation mark only if the extract
is a full sentence, but not otherwise. A quoted word, phrase, clause, or any fragment
of a sentence is no longer entitled to retain any of these concluding punctuation
marks.

A brief summary of conventions under the logical system is given below. A more
complete guide is given in Hart’s rules for compositors and editors.

4.3.1 Place periods, question marks, and exclamation points before closing quotation
marks if they apply to entire quoted sentences but not otherwise (Ex. 17).
Example 17
(a) In connection with Marshall’s treatment of increasing returns, Joan
Robinson wrote: “The more I learn about economics, the more I
admire Marshall’s intellect and the less I like his character.”

(b) Using the words of Rabbi Hillel, Arrow asks, “If I am not for myself,
then who is for me? And if I am not for others, then who am I? And
if not now, when?”

(c) Marx satirizes the backwardness of German society by enumerating

the survivals of feudalism and shouting, “But take off your shoes,
German patriot, for you are standing on hallowed ground!”

In the examples above, the full stop, the question mark, and the exclamation point
are placed within the quotation marks, since the extracts are full sentences that end
with these marks.

4.3.2 Whenever quoted sentence fragments are embedded in a main sentence, drop
any concluding punctuation marks attached to the extract. Instead, insert the

3Briefly, the conventional mode always places periods and commas within quotation marks,
whether they refer to the quoted fragment or to the entire sentence. Colons and semicolons,
however, fall outside the quotation marks. Rules are identical to those of the logical system for
dashes, question marks, and exclamation points, i.e., these marks fall within the quotes if they
refer only to the extract and outside the quotes if they refer to the entire sentence. The
differing treatment under this system of question marks and exclamation points, on the one
hand, and periods and commas, on the other, is imponderable.

14
punctuation marks applicable to the main sentence after the closing quotation marks
of the extract (Ex. 18).
Example 18
(a) The folly “where every nation wants to earn a surplus from the rest”
is what Joan Robinson called the “new mercantilism”. [full stop]

(b) Who would at that time have believed Jevons’s claim that
“economics, if it is to be a science at all, must be a mathematical
science”? [question mark]

(c) But where capacity utilization is low, high interest rates are far from
being “what the doctor ordered”! [exclamation point]

The quoted portions, “where every nation wants to earn a surplus from the rest”,
“new mercantilism”, and “what the doctor ordered”, are not full sentences. The full
stop, the question mark, and the exclamation point properly belong to the main
sentences, not the quoted matter, and therefore fall outside the quotation marks.

4.3.3 In principle the same rule applies to commas, dashes, colons, and semicolons.
You will rarely, if ever, need to include a final comma, dash, semicolon, or colon at
the end of a quoted fragment. Almost always, therefore, such punctuation will belong
to the main sentence, not to the quoted fragment, and they must fall outside the
quotation marks (Ex. 19).
Example 19
(a) By “labor-power”, therefore, Marx means the laborer’s capacity to
work, not the expenditure of work itself.

(b) In the current context, there can be only one result from “viewing
the exchange rate as anchor” — the loss of export-competitiveness.

(c) Three things are “added up”: wages, profits, and rent.

4.3.4 If the embedded quotation is a complete sentence that ends in a period,

exclamation point, or question mark (placed before the closing quotation according
to rule 4.3.1), no separate punctuation mark is needed after the closing quotation
mark.

Where quotations are broken off and interrupted by “he wrote”, “she argued”, etc., a
comma should be inserted at that point where the quotation used to contain a
punctuation mark. The comma should then be inserted before the quotation mark to
indicate that a punctuation mark was present in the original. (Note that the
conventional system is incapable of making this distinction.) Suppose the matter to
be quoted consists of the following sentence from Hicks: “Mathematics is not a

15
science. It is about concepts, and the relations between concepts, not about
phenomena.” Then either alternative in Ex. 20 is permissible.
Example 20
(a) “Mathematics is not a science,” Hicks observed. “It is about
concepts, and the relations between concepts, not about
phenomena.”

(b) “Mathematics”, Hicks observed, “is not a science. It is about

concepts, and the relations between concepts, not about
phenomena.”

In Ex. 20(a), a comma has replaced the original full stop and that this is placed
before the closing quotation mark. If, on the other hand, the quoted matter contains
no punctuation mark at the point it is broken off, the inserted comma falls outside
the quotation mark. In Ex. 20(b), since there was no original punctuation where the
extract was broken off, namely after mathematics, the first comma falls outside the
quotation mark.

5. Footnotes
In the system recommended here, bibliographic footnotes, i.e., those that merely
contain references to sources, should fall away completely, since the citations appear
in the main body of the text itself. On the other hand, explanatory footnotes, which
explain or elaborate upon some issue in the main text, would remain.

5.1 The numeral pertaining to the footnote should be in superscript and, whenever
possible, set two points smaller than body text. Following the logical system of
punctuation, if the footnote refers only to the quoted phrase, then the numeral
should be typed immediately after the closing quotation mark. Where the footnote
pertains to the entire sentence containing the quoted phrase, on the other hand, the
numeral should be placed after the full stop. In Ex. 21, the first and second footnotes
exemplify the former and latter cases, respectively.
Example 21
The folly where “every nation wants to earn a surplus from the rest”5 is what
Joan Robinson [1981] has termed the “new mercantilism”.6

5.2 Footnotes found on the same page as the annotated material are friendlier to the
reader than endnotes, since the latter would require the reader to flip back and forth.
Some book designers and publishing houses appear to be overly concerned with
“lightening” the look of a page, enough to balk at footnotes. This should be no
concern to technical writers, however.

16
One suspects that the resort to endnotes in the past was really a concession to the
humane treatment of typing secretaries, who would often struggle to adjust footnote
text to fit onto the same page as that in which the annotations occur. Secretaries and
researchers throughout the world now heave a sigh of relief with the coming of the
word-processor.

5.3 The text of a footnote should be single-spaced, using a font two points smaller
than body text (e.g., if body-text uses 11-point font, then the footnote text should use
9-point font). The first line should not be indented, but each footnote paragraph
should be separated from the next by a single space. The first footnote text on a page
should be separated from the main text by a continuous line running from the left
margin to about one-fourth of the page width.
Example 22
Annotated text:
In Marshall’s system, quantity supplied increases as long as producers find
they can dispose of their current output at a price higher than their
reservation price7, i.e., as demand-price is higher than supply price.8

Footnotes on the same page:

7That is, the minimum price at which they would be willing to dispose of the good.
8It is rarely noted that Marshall’s demand and supply curves appear to rest not on

ceteris paribus but on mutatis mutandis assumptions.

5.4 A caveat is in order regarding the length of explanatory footnotes, however.

These should never exceed three or four lines of text. If you feel it necessary to
exceed that limit, then the material is obviously too large for a footnote, and you
should consider either incorporating the explanation in the body text itself or even
resorting to an appendix.

5.5
Different rules apply when you submit manuscripts to journals and

publishing houses. Here, endnotes rather than footnotes are to be preferred, since
these allow the publisher greater flexibility to reformat the notes according to the
journal’s house-style. All endnotes should be collected in a separate section at the
end of the manuscript. Type these separately, and do not use the word-processor’s
facility for endnotes. Minimize formatting and set the text of the endnote in the same
font and paragraph style as the body text.

17
 Example 23
Annotated text:
In Marshall’s system, quantity supplied increases as long as producers find
they can dispose of their current output at a price higher than their
reservation price7, i.e., as demand price is higher than supply price.8

Endnotes on separate page (same font-size and spacing as body-text):

That is, the minimum price at which they would be willing to dispose of the
7

good.
8 It is rarely noted that Marshall’s demand and supply curves rest not on
ceteris paribus but on mutatis mutandis assumptions.

If a journal explicitly requires it, submit to its house-style in the matter of foot- or
endnotes. Guidelines can sometimes be found on the inside back-cover or the last
few pages of the journal.

6. References
Since the system suggested here cites sources in the text in abbreviated form, extra
care must be taken to ensure the accuracy of information listed under the references,
since that will be the only place the reader can find the complete information
regarding the materials used. The following system of writing references, based
loosely on the “Harvard system”, is recommended both because it is concise and easy
to remember.

6.1 General. In all cases, write the author’s (or editor’s) family name first, followed
by the personal initial(s). This is to be followed by the date of publication of the
work, enclosed in brackets exactly, the title of the work, and details of publication.
(See items 4-10 below.)
Example 24
Samuelson, P. A. [1947] Foundations of economic analysis. Cambridge,
Mass.: Harvard University Press.

If there is more than one author or editor, the names succeeding the first should be
written in the following order: personal initial(s) first, followed by the family name
(Ex. 25). All secondary authors or editors must be listed in the references. Though it
is permissible to use the phrase “et al.” or “and others” when citing the source in the
body of the text, the references cannot countenance such omissions and
abbreviations.

18
 Example 25
Laffont, J.-J. and J. Tirole [1993] A theory of incentives in procurement and
regulation. Cambridge, Mass.: MIT Press.
6.2 Spacing and indentation. Each reference item should be written single-spaced
and separated from the next entry by 1½ lines. Each entry should have the form of
hanging paragraphs, with the second and subsequent lines indented about one tab
stop from the left margin.

6.3 Second initials. As a rule, and when no ambiguity is possible, the reference may
be written using only the author’s or editor’s first personal initial, as in Ex. 24 and
25. Where some confusion may arise, such as when two authors have the same first
personal name, write in the second initial as well. In Ex. 26, it so happens that John
Maynard Keynes must be distinguished from his father, John Neville Keynes (who
was also an economist). Hence the need to include second personal name.

Where second personal initials are required, no space should separate the second
personal initial from the first. (Note: Unlike Filipino conventions, the second initial
among Anglo-Americans is usually not the maternal surname but another personal
name. (The “A” in P.A. Samuelson, for example, stands for “Anthony”.)
Example 26
Keynes, J.N. [1891] The scope and method of political economy. London:
Macmillan.
Keynes, J.M. [1935] The general theory of employment, interest, and money.
London: Macmillan.
Lewis, W.A. [1955] The theory of economic growth. London: George Allen
and Unwin Ltd.
Laffont, J.-J. and J. Tirole [1993] A theory of incentives in procurement and
regulation. Cambridge, Mass.: MIT Press.
Another situation requiring the inclusion of the second initial is where an initial
begins the author’s personal designation, e.g., W. Arthur Lewis (Ex. 26). In this case,
include the first initial as well as the second.

Some French and German personal names are combinations and therefore
hyphenated, e.g., Jean-Jacques, Jean-Paul, Hans-Jochen, Hans-Peter, etc. These
personal names may appear as initials either taking only the initial letter of the first
name, or also by writing in the initial of the second name. Hence, J.-J. Laffont.

A similar situation arises with some names derived from the Spanish, e.g., the
influence of mariolatry and the widespread of “Maria” in such combinations as Maria
Teresa, Maria Carlota, Maria Socorro, etc. The proper personal initials to use in this
case are M.T., M.C., and M.S., respectively; a simple M. would not do.

19
For a number of Asian and other authors, the second name read from left to right is
the personal name itself, while the first name is the family name or patronymic, as in
Liu Shaoqi, Mao Zedong, Deng Xiaoping, etc. There is no need in such cases for a
comma to separate the family name from the personal initial, since no inversion is
involved.
Example 27
Mao Z. [1975] Selected works. Vol. 2. Peking: Foreign Languages Press.

6.4 In general, prefixes such as “de”, “du”, “van” or “von”, such as in von Wieser, von
Neumann, von Böhm-Bawerk, etc., should be ignored. Such entries should be
alphabetized according to the main names themselves (Ex. 28).
Example 28
Neumann, J. von and O. Morgenstern [1944] The theory of games and
economic behavior. Princeton, N.J.: Princeton University Press.
In the end, however, the author’s own preference and long-standing usage must
prevail. A clue is sometimes provided by capitalization. If the prefix is capitalized,
then the name should be alphabetized using it (e.g., Descartes). Hence distinguish
between “De Guzman” or “Deguzman”, on the one hand, and “de Guzman”, on the
other. (As an example, Von Neumann’s brother appears to have preferred the
spelling “Vonneuman”, forcing a classification under the letter “V” rather than “N”.)
In most standard publications these days, all doubts may be resolved by looking at
how the author’s or editor’s family name is written in the Cataloguing-in-Publication
Data box that appears before a book’s table of contents.

6.5 Journal articles. In citing journal articles, apart from the author’s name and year
of publication, include the following minimum information: (a) the title of the
article, (b) the title of the journal, (c) the number of the volume containing the
article, and (d) the relevant page numbers (Example 29).
Example 29
Holmstrom, B. [1982] “Moral hazard in teams”, Bell Journal of Economics
13: 324-340.
The title of the article s should be written out fully, capitalizing only the initial letter
and proper nouns; the article title should then be enclosed in double quotes and
separated by a comma from the journal title. The journal title should be written in
italics, using caps and lower case. The volume number should follow immediately
after the title, and if you have NIBS, this may be written in boldface. The entry should
always end with the page numbers, preceded by a colon.

20
For a truly complete citation, however, the number of the issue within the volume
(e.g., 1 or 3) or the relevant seasonal designation (e.g., Spring, Winter, First
Semester,) should also be included. This is especially important for journals or
periodicals whose pages restart with every number (the Journal of Economic
Perspectives and the Philippine Review of Economics are some obvious examples).
The issue number should be set off from the volume number by parentheses, while a
comma should suffice as a separator for the time of year, as shown in the following
example:
Example 30
Chipman, J. and J. Moore [1973] “Aggregate demand, real national income,
and the compensation principle”, International Economic Review
14(1): 153-181.
Nelson, R. and S. Winter [2002] “Evolutionary theorizing in economics”,
Journal of Economic Perspectives 16(2), Spring: 23-46.
Page numbers in journal entries should indicate the beginning and the ending page
and should be separated from the rest of the entry by a colon. It is a current fashion
to avoid the repetition of common number-places when citing page numbers, e.g.,
1220-53, or 1220-340. While there is something to recommend this practice on the
principle of economy, it is just as likely to lead to confusion, since it unnaturally
presents a sequence from more to less digits as the eye moves from left to right. (You
can be sure this is bound to lead to future problems with machine-readers.
Remember the Y2K mischief wrought by the apparently innocuous truncation of
dates?) For the reader’s convenience and to avoid future problems, therefore, write
out the entire number, i.e., 1220-1353 rather than 1220-353.

6.6. Books. Italicize the titles of books or monographs in the references. Capitalize
only the first letter of the title. Except for proper nouns, there is no need to capitalize
“important” words (see Ex. 24, 25, and 26). Italicization is sufficient to distinguish
the title from the rest of the entry. Separate the title from the place of publication
with a full stop; separate the place of publication from the publisher by a colon. End
the entry with a full stop.

6.7 Articles in anthologies. When listing an article from an anthology or a collection

of articles (Ex. 31), cite the article’s title as you would cite one from a journal.
Separate the title from the rest of the entry with a comma, followed by the word “in”,
the editor(s) of the anthology, the title, and publisher’s information, as you would
when citing a book. Do not invert any of the editors’ family and personal names. Page
numbers are essential, and these should be the last item in the citation, separated

21
from the rest of the entry by a full stop. The same rules apply when citing
introductions or prefaces, such as the second entry in Ex. 31.
Example 31

Sonnenschein, H. [1971] “Demand functions without transitive preferences,

with applications to the theory of competitive equilibrium”, in
J. Chipman, L. Hurwicz, M. Richter, and H. Sonnenschein, eds.,
Preferences, utility, and demand. New York: Harcourt-Brace-
Jovanovich. 215-223.
Georgescu-Roegen, N. [1987] “Inroduction”, in H. Gossen. Laws of human
relations and the rules of human action derived therefrom.
Cambridge, Mass.: MIT Press. i-cxlv.
If you must cite more than one article in the same anthology, it is economical to
make a separate entry for the anthology under the editor’s name and cite only the
editor’s names in listing the article. In the example below, the anthology is
Stoneman, ed. [1995], which contains the cited articles by Metcalfe and Cohen.
Example 32
Cohen, W. [1995] “Empirical studies of innovative activity”, in P. Stoneman,
ed. 182-264.
Metcalfe, S. [1995] “The economic foundations of technology policy”, in
P. Stoneman ed. 409-512.
Stoneman, P., ed. [1995] Handbook of the economics of innovation and
technological change. Oxford: Basil Blackwell.
6.8. Multiple works and editions. When more than one work written by the same
author in the same year must be cited, differentiate these by adding a, b, ..., etc. to
the year of publication, e.g., 1959a, 1959b, and so on.

Especially when it comes to classics, new editions or translations are more likely to
be available than the originals. Yet the year of the original is often important in
conveying information about which work is involved. As an example, Gossen’s work
on marginal utility, written in 1867, was not translated into English until 1983. In
this case, write the year of the original between round parentheses immediately after
the year of the edition or translation you are using.
Example 33.
Gossen, H. [1983(1867)] Laws of human relations and the rules of human
action derived therefrom. Translated by William Smart. Cambridge,
Mass.: MIT Press.
6.9 Multiple-volume works. Some references pertain to works consisting of several
volumes in a series. It should first of all be ascertained whether a reference to the
entire series is necessary or only to a certain volumes in the series. If the entire series
is meant, then the publication dates should include all years from the appearance of
the first to the last volume, and a reference should be made to the total number of

22
volumes. Some individual volumes have their own titles, which should be mentioned
when citing each separately.
Example 34
Arrow, K. [1983-1985] Collected papers. Six volumes. Cambridge, Mass.:
Belknap Harvard Press.
Arrow, K. [1985] Collected papers. Volume 6: Production and capital.
Cambridge, Mass.: Belknap Harvard Press
Ricardo, D. [1951] Works and correspondence. Eleven volumes. P. Sraffa and
M.Dobb, eds. Cambridge: Cambridge University Press.
In the first two entries of the example, it would have been redundant to write: Arrow,
K. J. [1983-85] Collected papers of Kenneth J. Arrow, etc. The first entry is
appropriate when the entire set is used, the second when only one volume is actually
cited. The last entry illustrates the subordinate position of editors (or translators)
when compilations of the work of a single person are involved, quite differently from
the case where an anthology is involved. The editors’ names must nonetheless be
included, since other editions may exist or arise in the future.

6.10. Institutional publications. Two cases must be distinguished. The first is the
more common one where the institution issuing the work is also the publisher. The
second, which is increasingly observed, is where the issuing institution has published
the volume and let a commercial publishing house distributed it. These cases are
represented respectively by the two entries below.
Example 35
Organisation for Economic Co-operation and Development (OECD) [1992]
Technology and the economy: the key relationships. Paris.
United Nations Development Programme (UNDP) [1996] Human
development report. New York: Oxford University Press.
Note from the first entry that there is no need to repeat the name of the publisher
when it is the same as the institution issuing the work.

6.11 Foreign names, terms, and phrases and nonstandard spelling. The immediately
preceding example also points to the importance of spelling the name of institutions
using the spelling convention it prefers. The words “organisation” and “programme”
were rendered using British-English spelling, since this is how the originals appear.
In contrast, the World Trade Organization prefers the American spelling convention
and this should be respected. Short of such proper nouns, of course, you may choose
your own convention in the main body of the work, as long as you stick to it
throughout.

23
As for foreign words, it is best to consult a dictionary to ascertain which foreign
words must be italicized and which have become so assimilated into English that
italicization is no longer required. For example, “vis-à-vis” is no longer italicized, but
“tableau economique” still is. (For some other foreign phrases, refer to the Appendix
to this booklet or to a good standard dictionary.)

When a work is better known in its original foreign language, the foreign title may be
displayed as well (Ex. 36). (Initial capitalization has been adopted in the foreign title,
since this is a peculiar requirement in German. Otherwise it would have been
unnecessary. Italicization is omitted to distinguish it from the English title.)
Example 36
Marx, K. [1973(1865)] Outlines of a critique of political economy [Grundrisse
einer Kritik der politischen Ökonomie]. Translated by M. Nikolaus.
Harmondsworth, U.K.: Penguin.
6.12 Electronic sources. An area where conventions are still evolving is the citation of
electronic sources, which may come in several forms. The same general convention
should be followed as that discussed above. The author’s or organization’s name
would come first, followed by (a) the year the document was written, (b) any
institutional information regarding the site, (c) the Universal Resources Locator
(URL) address of the site (e.g., www., ftp., telnet) and (d) the specific date on which
the site was accessed. In Example 37, the Krugman article was written in 1999 (a fact
provided by the site or by the article itself), but the researcher accessed it only on 9
October 2002. Note well that the date (a) may differ from the date (d). If the
document does not bear a date, you may put “n.d.” (= no date) in lieu of the year
entry.
Example 37
Krugman, P. [1999] “Was it all in Ohlin?”,
http://web.mit.edu/krugman/www/ohlin.html Accessed 9 October
2002.
Chipman, J. [2000] “Intra-industry trade in a log-linear model”,
http://www.econ.umn.edu/~jchipman/econ840202/econ8402.htm.
Accessed 26 November 2002.
You will note one problem presented by URLs is that there is often no way to split
them, so that one will have to live with a good deal of white space towards the right
of the page. C’est la vie!

24
7. Voice and tense
Grammatical decisions regarding voice and tense of sentences can often influence
the tone of your entire work. This section lays out the possible alternatives and
suggests when each might be appropriate.

7.1 Personal or impersonal voice (active or passive). Whether to use the active or
passive voice in one's work is still a matter of personal taste. The traditional
preference for the passive voice in most technical writing was due to its placing the
author in the background, which was thought more in keeping with a “scientific”
ethos where the role of the subject or observer was unimportant. Today the active
voice has made tentative inroads into a realm formerly dominated by the passive.
This is because the active voice more closely approximates natural speech and could
actually facilitate the explanation of the method or procedure you have followed in a
proof or a piece of empirical work.
Example 38
(Passive voice)

The partial derivatives of output with respect to labor and capital are then taken.

Output is then derived partially with respect to labor and capital

(Active voice)

We then take the partial derivatives of output with respect to labor and capital.

Then one takes the partial derivatives of output with respect to labor and capital.

Any one of the above is acceptable. Again, whichever convention you choose, must
generally be adhered to throughout.

7.2 Person. Even after you have chosen to employ the active voice, you still have to
decide whether to use “I”, “we”, or “this writer/author”, or “one”. Tradition suggests
using the editorial “we” (Ex. 39b), but this convention is weakening, especially
among authors with a good deal of self-assurance. American convention in particular
is moving towards a more natural and casual style where the writer uses the first-
person singular (“I”) in referring to himself (Ex. 39c). Then the first-person plural
(“we”) is no longer used in an editorial sense, but only when there are indeed
multiple authors (Ex. 39b).
Example 39
(a) Perfect competition is assumed. (Passive voice)
(b) We assume perfect competition. (First-person plural)

25
 (c) I assume perfect competition. (First-person singular)
(d) This writer assumes perfect competition. (Third person)
A note of prudence must be sounded, however. Students submitting papers, theses,
and dissertations for review are advised for now to steer clear of the first-person
singular (“I”, “me”, “mine”) and, rather than advertising ownership, allow the
originality of the ideas to speak for themselves. The passive voice is always less likely
to raises hackles.

7.3 Tense. Either the present or the past tense may be used in parts of papers dealing
with narratives, e.g., a literature review. But once a choice is made, it should be
maintained. In the use of the present tense, it should be remembered that prior
events will take the past tense (Ex. 40(a)). On the other hand, when the principal
tense is the past, prior events take the pluperfect (Ex. 40b).
Example 40
(a) (Present + past)
Williamson [1985] fully elaborates (present) the notion of transactions costs.
Previous to this, Coase [1937] introduced (past) the concept in the context of the
theory of the firm.

(b) (Past + pluperfect)

Williamson [1985] fully elaborated (past) the notion of transactions costs. Previous
to this, Coase [1937] had introduced (pluperfect) the concept in the context of the
theory of the firm.

Obviously, however, the past-pluperfect combination is more appropriate for events

that may be regarded as historically distant:
Example 41
(a) In the Wealth of nations, Adam Smith never mentioned [past] the steam engine in
describing industrial progress, although his fellow-Scot Watt had invented
[pluperfect] the device some years earlier.

(b) Krugman [1995] refers [present] to the notion of competitiveness as “a dangerous

obsession”, but many have failed [perfect] to comprehend the theoretical basis of his
critique.

26
8. Mathematical writing
Again, clarity and concern for the reader are the purpose for insisting on some
conventions for mathematical writing.

8.1 Fonts for mathematical writing. Most word-processing programs now have
options for different styles aside from Normal Style, including Italics, Boldface, and
Symbols (NIBS, for short). If your program allows for these, then the following table
shows some conventional options.

Style Font Examples

variables and constants italics x, X, ax + b, p/q

numbers and numerals normal 2x, X1, f(0) = 0
function names italics g(x); f is increasing
mathematical operations normal 2+x
vectors lower case: bold or italics x or x
matrices UPPER CASE: bold or X′X-1 or X′X−1
italics
sets and relations UPPER CASE italics the set X; the relation R

8.2 Numbers and numerals. A literary convention that tells you to spell out counting
numbers between one and ten but to use numerals for counting numbers beyond ten.
You should remember that this rule applies only to counting numbers and should be
ignored when estimates or variable values are involved.
Example 42
There are nine papers in the literature that report estimates of 2.5 or less.

Obviously the function (x – 1)/(x – 3) is undefined when x equals 3.

In the first sentence of Example 42, “nine” is a counting number and is spelt out,
while 2.5 is an estimate. In the second sentence, “3” is not written as a word, since it
is not a counting number but a value taken by a variable.

8.3 Theorems and lemmas: If you have NIBS, use the following convention for
writing theorems (or lemmas or corollaries) and their proof. Set off the theorem in a
separate paragraph, with the theorem number or title written in boldface. Write the
theorem proper in italics. But do not italicize formulas and symbols embedded in the
text of the theorem.

27
Begin the proof as a separate paragraph, preceded by the word “Proof” written in
italics or underlined and separated by a colon. Write the body of the proof in normal
(non-italicized) font, taking care to italicize formulas and symbols, where
appropriate.
Example 43
Theorem 7. The maximum function m(x) = 1 – x, is continuous.
Proof : Take any two points x1 and x2 , with the corresponding values of the function
m(x1) and m(x2). Since m(x1) = 1 – x1 ... etc.

The general rule is to differentiate formulas from the rest of the text. Hence, if the
one is italicized, the other is not. In the statement of the theorem, m(x) = 1 – x is not
italicized, since the theorem itself is italicized. The opposite is the case for the proof.

8.4 Italicization within formulas. A particular pitfall among many writers in the age
of the word-processor is the failure to observe conventions for omitting italicization.
Remember not to italicize especially numerals, numeral subscripts, mathematical
operations, and separators, such as parentheses and brackets, as in the following
examples:
Example 44
Wrong Right

x = [x1, x2, ..., xn] x = [x1, x2, ..., xn]

f(0) > f(b) where b > 0 f(0) > f(b) where b > 0

Avoiding confusion is more important than impressing the reader; therefore

consistency is at a premium. Your reader may well understand that you do not have
the NIBS options and will adjust accordingly. But he will not understand why you
italicized the same variable in one instance but not in another. This problem is
especially likely to arise when you use “equation editors” to write your formulas and
equations, then shift back to normal text style in discussing them. Observe the
following example.
Example 45 (Bad)
The formula for the effective rate of protection ej in the case of a good using
a single input is given as follows
tj – siti
ej = 1 – s ,
i

where tj is the tariff on the good itself, ti is the tariff on the input, and si is
the share of the input in the total value of output.

28
Here, it is evident that the displayed equation was created using an equation editor,
while the rest of the text used the normal style. Note how the subscripting and the
italicization have disappeared in the paragraph discussion. When using equation
editors, be sure the fonts and sizes of the various styles correspond with those of
normal style, adjusted for mathematical convention.

8.5 Reducing tiers of symbols. Whenever possible and without sacrificing clarity, use
slashes (/) to rewrite fractions that involve several tiers. An expression such as
∂C
(p) - w
∂L
r= w
involves three tiers of symbols. This may be reduced to two tiers by writing
(∂C/∂L) - w
r= w ,

or even a single tier:

r = (∂C/∂L - w)/w .

A possible disadvantage from doing this is the loss of the information that ∂C/∂L is a
function of p. One recourse is to start out with the convention for writing the partial
derivative as CL(p), which allows you to write the whole equation on a single line, as
follows:
r = (CL(p) - w)/w .
This advice is especially warranted when you must write formulas“inlaid” in
paragraphs of text. (See 8.16 below.) Some helpful hints for reducing tiers in
formulas are contained in the table below.
Multi-tier Single-tier
x x/y
y
x x(1 – x)–2
(1 – x)2
(z – 2)2/3 (z – 2)2/3(z + 1)–1
(z + 1)
x–1 (x – 1)/(x + 1)
x+1
x–1 1–x [(x – 1)/(x + 1)] + ½(1 – x)
x+1+ 2
∂Q (∂Q/∂L) or QL
∂L
∂f ∂f f1, f2, ...
, , ...
∂x1 ∂x2

29
8.6 It is good advice [Knuth, Larrabee, and Roberts 1989] to keep subscripts and
superscripts to a minimum. To do this, it sometimes helps to hold off on definitions
until they are really needed.
Example 46
1 1 1 2 2 2 1 2 n 1 2
Let x = (x1, ..., xn) and x = (x1 , ..., xn), with x , x ∈ X⊆ Ω . Then x Rx , if
1 2 2
and only if x is preferred to x or is indifferent with respect to x .

The above example could have been simplified by writing:

n
Let x, y ∈ X ⊆ Ω . Then xRy if and only if x is preferred to y, or is
indifferent with respect to y.

8.7 Most mathematical operators should not be italicized. Among these are the
following:
derivatives Dx y
exponentials Exp(2x + 1); e2x + 1
expectations E(Y) = υ
probability Pr(Z < Z*)
limits limx →0 [(2 - x)/(2 + x)]
minimum min [u(x), u*]
maximum max [v(z), 5]
trigonometric A sin t + B cos t
functions
signum sgn x = sgn (a – by)

8.8 Proper symbols. Take the trouble to locate and use the right symbols. The
following provide some typical instances.

8.8.1 Use primes (′) and double-primes (″) in the form of f ′(x) > 0 and f ″(x) < 0,
rather than single- and double-quotation marks. These are typically available from
the “symbols” font of your word processor.

8.8.2 Avoid the form x*y, which is computerese, or the equally unsatisfactory x.y in
denoting multiplication. In most cases, xy should suffice. If you wish to place an
intervening symbol for multiplication, however, the proper one is (⋅), which is mid-
height, i.e., x⋅y and not on the writing line.

8.8.3 If you want to distinguish between the multiplication of scalars and the dot-
product of vectors, find the proper symbol that will distinguish x⋅y from xy. Better yet,
anticipate the problem by distinguishing scalars x and y from vectors x and y and

30
write xy for scalar multiplication, xy for the dot-product, and maybe XY for matrix
multiplication.

8.8.4 Do not use the letter “x” as a substitute for the multiplication sign “ד. The
latter symbol is available in the symbols-font of most word-processing programs or
(if you are really desperate) may be approximated by a sans-serif x.

8.8.5 It is useful and sometimes important to distinguish identities from other types
of equalities by the triple “≡”, especially if the sense is not sufficiently evident from
the context. Another type of equality is the definition, which is sometimes
distinguished by an equality sign with df or def written above it.
Example 47
Y≡C+I+G+X−M
df
Π = TR − TC

8.8.6 The inequality symbols ≥ and ≤ are not different from the corresponding
symbols ≥ and ≤ when scalars are being compared. There are authors, however, who
make distinctions between these when vectors are being compared. If x and y are
two vectors, then x ≥ y means that each component of x is at least as great as the
corresponding component of y.

8.9 Enumeration. Remember to repeat the same symbol to the left and to the right of
the ellipsis mark (...) in an enumeration, allowing for space on either side.
Example 48
n = 1, 2, ..., m
x1 + x2 + ⋅⋅⋅ + xn
Rk = R × R × ⋅⋅⋅ × R (k factors) or (k factors of R)
Note that unlike the ellipsis mark separating commas, the ellipsis marks separating +
and × in the above example are mid-height (⋅⋅⋅) and not at the bottom (…). The
former mark may be available in some equation editors, or may be produced as a
succession of the same dots that are used to signify multiplication. On typewriters,
the same effect may be produced by moving the carriage a half-line up and typing
three successive full stops.

8.10. Proper reference. Scrupulous mathematicians may flinch, but a convention has
established itself among economists which permits the use of the same symbol for
the dependent variable and the function that governs it. Hence, one typically sees the
following usage:

31
 Example 49
We suppose profits z to be a function of price p, that is, z = z(p).

In Ex. 49, z is both the dependent variable and the name of the profit function. In
most cases, this usage leads to no further difficulties. But not always. Suppose one
wanted to describe the behaviour of both the function and the variable, as in the
following:
Example 50
The solution to the problem, min px, s.t. u(x) ≥ u, where u is concave, yields
the expenditure function e(p, u). Writing this for various u allows us to
write spending as a function of prices and the utility level, e(p, u).

Here, there is some ambiguity over what u in e(p, u) refers to (i.e., the function itself
or a particular value of the function). In general your notation should allow
distinctions between the dependent variable, the function itself, and the value of the
function at specific values of the independent variables. More properly, therefore,
use different characters to refer to each. Sometimes a change in case is enough to get
rid of the issue:
The solution to the problem, min px, s.t. U(x) ≥ u, for some arbitrary u,
where U is concave, yields the expenditure function e(p, u), which permits
us to write spending as a function of prices p and the utility level u.

8.11 Dashes, hyphens, and minus signs. In good math books, there are four different
symbols that resemble each other: a hyphen (-), an en-dash (–), an em-dash (—),
and a minus sign (−). Hyphens connect compound words, e.g., “short-run” or “Neo-
Keynesian”. En-dashes are used to denote ranges of numbers, as in “pages 1–10”.
Em-dashes are used as punctuation — to set off phrases for emphasis, illustration, or
clarification. Minus signs are used in mathematical formulas. A conscientious
typesetter would distinguish between these four symbols. It is a somewhat
frustrating feature of most word-processing programmes, however, that they do not
produce the proper minus symbol when the minus-operation key is hit.4

8.12 Variable names. In the choice of variable names, it is best to hold off on the
Greek symbols and try to get by with the roman alphabet as much as you can.

4 In Word™ for Windows™, for example, hitting the minus (NUM –) sign produces a hyphen
“-” instead. The proper minus symbol is available only from the Symbols palette. (Hint: As a
practical matter, however, the en-dash (–) is close enough in value to be almost
indistinguishable from the minus symbol. In Word™ this may be obtained by hitting
CTRL+NUM –. The em-dash is produced by CTRL+ALT+NUM –.)

32
Remember, however, that convention has a way of allocating the following groups of
letters to certain roles:
Example 51
x, y, w, z, v generic variables
f, g, h function names as in f(x), g(x)
i, j, k, l, m, n (and r, s, t) indices as in xi or yt
a, b, c, d, p, q generic constants and coefficients
It makes for less confusion if you stick to this convention. To see this, consider the
confusion caused by the following “alternative” assignment.
Example 52 (A negative example).
Let j be a continuous function d of f given by j = d(f), where f is an x-dimensional
vector, f = [f1, f2, ..., fx] ... .

Similarly, in economics, the choice of variable names often facilitates understanding

and should be intuitive. Hence, by convention, the factors capital and labor are called
K and L, respectively, while land is sometimes called T (= terra). Goods are called X
and Y, or X1 and X2, or G1 and G2, and so on. In macroeconomics, consumption is
called C, investment is I, etc. (A classic example of flouting this advice is the well-
known 1955 article by T. M. Rybczynski, where factors of production are called X and
Y and goods are K and L. This proves that brilliance can excuse eccentricity, but
unless you can presume to be brilliant ex ante, it is better to hold off on the
eccentricity.)

Note also how macroeconomics writing tends to the use of upper-case letters, which
suggests the idea of aggregates. For this reason, avoid variable names involving
multiple letters, lest this be mistaken for some product. As an example, if you have
just defined I as investment, N as population, and V as the velocity of money, it is
inadvisable to designate inventories with INV, which could be interpreted as the
product of the three.

When writing mathematical formulas, take care to distinguish between letters and
symbols that look almost the same. Examples include the italicized Greek letters
µ (mu), ν (nu), and κ (kappa), which look like the italic letters u, v, and x, and the
Greek letter φ (phi), which looks like the symbol for the empty set ∅.

8.13 Twoness. Another way to save on notation is by using capital and lower-case
letters creatively. “Two-ness” is encountered in many economic models and
expositions. In international trade, for example, one has the 2 x 2 x 2 model of two
goods, two countries, and two factors. (A work that does this is Dixit and Norman’s

33
1980 book on trade theory.) The following shows how unnecessary complication may
be simplified:

Example 53
(a) (Tedious)
1 1
Let there be two goods produced in countries 1 and 2, and let (x1, x2) denote
2 2
the output of these two goods in country 1 and (x1 , x2) production in country
k
2, where xi denotes the production of good i in country k.

(b) (Simplified)
Let there be two goods produced in countries 1 and 2, and let (x, y) denote
the production of these two goods in country 1 and (X, Y) the production in
country 2.
This method fails, of course, once the relevant number exceeds two.

8.14 Variables in theoretical and empirical models. There is almost always a need to
distinguish the names given to variables in theoretical models and those in empirical
models, especially in estimating equations. Applied statistics and econometrics tend
to assign multiple-letter names to variables in equation estimation, so that a variable
for education is EDUC, while that for income is INCOME, and so on. You may have
several measures of income, so that you may try INCOME1, INCOME2, and so on.
In your conceptual model, however, you may simply have used e for education and y
for income in the abstract. The distinction between the variables in the theoretical
and their empirical equivalents is useful, since they remind you and your reader that
the variables as measured are not necessarily identical with their conceptual
equivalents.

The convention of using abbreviations for econometric variables (EDUC for

education, and so on) was imposed by the limitations of some econometrics
software, which imposed a maximum length on variable-names. Most types of
software these days have overcome this difficulty, and there is no compelling
technical reason for insisting on truncated names. The use of the full names of
variables is especially facilitated if estimated equations are reported in tabular form.

8.15 Equation numbers. In a final draft, number only the equations to which you
actually refer. Indent the beginning of each equation one tab-stop from the margin of
a normal paragraph; indent the second and subsequent lines one tab-stop further.
Align the equation-numbers along the right margin of the page.

34
 Example 54
The first-order conditions are
[u(2) – u(1)] π′(x) – π(x)(1 + π′(x)z)u′(2)
– (1 – π′(x))(1 + π′(x)z)u′(1) = 0 (2)
π(x)[1 – π(x)] u′(2) – [1 – π(x)] π(x)u′(1) = 0 . (3)

8.16 Formulas “displayed” and “inlaid”. When should you display a formula (an
equation or expression, numbered or not) rather than simply writing it into the text?
There may be several reasons for doing so. The first is simply length: if the formula,
when written with the text, threatens to occupy more than one line on the page, then
it is better displayed. This rule just corresponds to that which also says lengthy
quotations (more than five lines) must be displayed. A second reason is simply that
you may want to call attention to the formula, regardless of its length. A final
situation arises when formulas are to be manipulated, such as when successive steps
in a proof need to be shown.

In writing inlaid formulas, you should seek as much as possible to rewrite formulas
so that they are on a single tier. (See the suggestions given in Point 4 above.)

8.17 Three rules on sentences and symbols: (a) a sentence should never begin with
an equation number; (b) do not start a sentence with symbol; and (c) use words to
separate symbols contained in different formulas.
Example 55
Bad Good

(12) is the equilibrium condition Equation (12) is the equilibrium condition.

|I – A| is nonzero. The determinant |I – A| is nonzero.
(c + 1)/(c + r) then is the money The money multiplier is then (c + 1)/(c + r).
multiplier.
Let y = f(k), fk > 0, fkk < 0. Let y = f(k), with fk > 0 and fkk < 0.

8.18 Numbers should be kept on the same line as their units of measurement;
mathematical symbols should not be separated from the nouns that describe them.
(In WordTM, this is ensured by tying the space in between the number and the unit of
measurement or the symbol and the noun by holding down SHIFT and CTRL
simultaneously and hitting the spacebar.)
Example 56
Bad Improved

The device measured 12 The device measured 12 in.

in.

35
In an enumeration, however, numbers and mathematical symbols are preferably
kept together rather than having one of them run on to the next line with the unit of
measurement or the descriptive noun.

Bad Improved

The devices measured 12 and The devices measured

18 in. 12 and 18 in.
...where xi and xj are the ith ...where xi and xj are the
and jth factors... ith and jth factors...

When referring to the plural of a number or a variable, do not use an apostrophe to

separate the number and the “s” that follows. The different forms of the former and
the italics font of the latter are enough to distinguish them from the “s”. Hence, write
“culture of the 1990s” rather than “culture of the 1990’s”, the “xs in the equation”
rather than “x’s in the equation”. But do use an apostrophe in the context of the
possessive case, as in “the year 1990’s economic problems” or “x’s various values”,
although in both cases an alternative formulation would have been be preferred (i.e.,
respectively, “the economic problems of 1990” and “the various values of x”).

8.19 Breaking expressions and equations. If a displayed equation or expression is

overly long and there is a need to break it to observe the margins, place the break
immediately before an operation such as + or – , or an equality sign = occurs, as
shown in the following example:
Example 57
E = C(Y – T) + I(r)
+ G + X(e, y*) – M(e, y)

If, on the other hand, an inlaid expression or equation must be broken, it must be
split after the relevant sign for an operation or equality.
Example 58
If one continues to assume full employment and perfect capital
mobility, then the current account deficit is (Y0 – T(Y0) –
C(Y0)) + (T(Y0) – G0) – I(r*), or equivalently, S(Y0) – I(r*).

8.20 When displaying successive transformations of expressions on either side of an

equality, align the equality signs of successive statements:

Example 59
Y = c0 + cY + I0
Y – cY = c0 + I0

36
 (1 – c)Y = c0 + I0
c 0 + I0
Y = 1–c

In many word-processors, this effect can be produced either through a judicious

selection of tab stops, through equation editors, or through the use of (invisible)
table formats.

8.21 Separators. There is no hard-and-fast sequence in the use of separators for

terms or factors in algebraic expressions, but here is one that is easy to remember:
round parentheses, square brackets, and braces. If required, you may move on to
larger round parentheses, larger square brackets, and larger braces, and so on, i.e.

{[({ [(x +y) ]})]}

8.22 Theorems, lemmas, and corollaries all have in common that they are
statements requiring formal proof, often of a logical or a mathematical nature.
Beyond this fact, however, a certain hierarchy exists. A theorem, strictly speaking,
should involve a result that is thought to be fundamental in nature. An example in
mathematics is the mean-value theorem, or Brouwer’s fixed-point theorem. A
proposition or a claim has the same formal status as a theorem. The different name
owes more to (innate or assumed) modesty on the writer’s part than to any real
difference in substance. A further step down the modesty ladder is the claim, which
must again be treated as a theorem in all respects. Once a style is chosen, however, it
must be maintained. Fundamental results must be consistently labeled either
theorems, or propositions, or claims, but not any two in a single work. (Some writers
refer to so-called “folk theorems”, referring to statements which are generally
regarded to be true but have not been formally shown to be so.)

Lemmas (lemmata being the older plural form) are auxiliary statements (also
requiring proof) that are subsequently needed as parts of a larger proof for a
theorem. At times, however, lemmas may subsequently attain a prominence by
themselves, so that they may be cited without reference to the original proof in which
they played a part. An example of this is Sperner’s lemma in the algorithms for
determining equilibrium and Hotelling’s lemma.

A corollary is a statement (again requiring proof) that follows as a consequence of an

earlier proven theorem or lemma. Hence, lemmas generally precede theorems, while
corollaries succeed them.

37
9. Statistical tables
Tables in economics and other disciplines pose their own problems, since they
inevitably combine numbers (statistics) with letters (typically labels). The basic
difficulty with formatting tables is that good form with numbers often differs from
good form with letters. For example, labels need to be centered relative to the table’s
columns. On the other hand, numbers need to be formatted respecting numerical
place-values.

Using a popular software program, the default-option format typically produces the
following:
Example 60
Table 5. Tax and expenditure ratios
(1985, 1990, 1995)
Year Taxes Expenditure Taxes/GDP Expenditure/GDP
(in million pesos) (in million pesos) (in percent) (in percent)
1985 61253 87390 10.71 15.28
1990 151700 255755 14.08 23.74
1995 309978 384700 16.27 20.19
Source: Philippine statistical yearbook 1998. National Statistics Office.

This is hardly worthy of a final draft, however. Exactly what is wrong with such a
table? First, note the improper alignment of the numbers. This results from the
flush-left alignment carried over from the text-formatting which cannot, of course,
be allowed to remain. Centering the figures would not help either, unless all number
strings were of equal length. Opting for a flush-right alignment would solve the
problem of aligning numerical place-values. But, as Ex. 61 below shows, this is still
unsatisfactory, since the numbers continue to be misaligned relative to the column
headings.
Example 61
Table 5. Tax and expenditure ratios
(1985, 1990, 1995)

Year Taxes Expenditure Taxes/GDP Expenditure/GDP

(million pesos) (million pesos) (percent) (percent)
1985 61253 87390 10.71 15.28
1990 151700 255755 14.08 23.74
1995 309978 384700 16.27 20.19
Source: Philippine statistical yearbook 1998. National Statistics Office.

Another bad feature is that the row-widths have been determined by the length of
the labels in the first row, although the number strings are much shorter in length.
This feature is most obvious and disagreeable in the last two columns. The resulting
general perception is one of lopsidedness. Finally, the gridlines cause the eye to
wander to and fro across the table, causing one to lose sight of the numbers, in what
38
is known as the “Moiré effect”. The following two examples reformat the same
information with a number of improvements.
Example 62 (column-table)
Table 5. Tax and expenditure ratios
(1985, 1990, 1995)

Taxes Expenditures Ratio of Ratio of

Year (in million (in million taxes to expendi-
of pesos) pesos) GDP tures
( in to GDP
percent) (in
percent)
1985 61,253 87,390 10.71 15.28
1990 151,700 255,755 14.08 23.74
1995 309,978 384,700 16.27 20.19
Source: Philippine statistical yearbook 1998. National Statistics Office.

Example 63 (row-table)
Table 5. Tax and expenditure ratios
(1985, 1990, 1995)

1985 1990 1995

Taxes (millions of pesos) 61,253 151,700 309.978
Expenditures (millions of pesos) 87,390 255,755 384,700
Ratio of taxes to GDP (percent) 10.71 14.08 23.74
Ratio of expenditures to GDP (percent) 15.28 23.74 20.19
Source: Philippine statistical yearbook 1998. National Statistics Office

Examples 62 and 63 illustrate two types of table: column-tables and row-tables. The
distinction depends on whether the data-categories are organized in column- or in
row-form, requiring the reader’s eye to move vertically or horizontally, respectively.
In both examples, however, the removal of the gridlines has produced a more
subdued visual effect. The numerical entries have been positioned more or less in the
middle of each column, paying attention however to the proper alignment of decimal
places. Still, as between the two examples, the row-table is aesthetically more
satisfying. It is less top-heavy and therefore allows the reader’s eye to move in a more
natural direction, from left to right. These and a few other suggestions are
summarized below.

8.1 Shun top-heaviness. Present data in row- or column-tables depending on which

type will result in a more compact form that draws more attention to the data. In the
case of very long time series, a horizontal presentation obviously cannot be made
compactly, so that a column-format may be advisable. On the other hand, if the
category-titles are somewhat lengthy while the data series are rather short (as in Exs.
62 and 63), it makes no sense to pile up the labels in the table’s first row, since this

39
would make the table top-heavy. In the end, therefore, there is a trade-off between
the length of the data series and the length of the labels.

8.2 Category labels and dimensions. The complete names of variables or category
labels must be specified in the table. Hence, rather than “Taxes/GDP”, write “Ratio
of taxes to GDP”(e.g., Ex. 60 versus Ex. 62 or 63). The units or dimensions of the
data (percent, thousands of pesos, thousands of households, etc.) must also be
included in the relevant column or row of category labels. Alternatively, this may be
mentioned in the table title, e.g., “Table 2. Real GDP growth (1970-2000; in
percent)”.

In column-tables, center the category labels within the column. In row-tables, write
category labels flush left.

8.3 Align numerals properly. Set the alignment so that the longest string of numbers
is more or less centered in its own column. Then observe the proper alignment of
decimal places for data within the same category. Sometimes this can be done by
pulling the right margin of the column leftward then using right-justification.

Where tables involve figures of different categories and units, column entries should
not be right-aligned but rather written flush-left (Hart’s rules). Ex. 64 gives another
illustration of this rule. Note that the first row deals with absolute sums of money,
while the last two deal with percentages. Right-aligning the entries would lead to the
wrong impression that they all dealt in the same units.
Example 64
Table 4. Data for selected regions, 1998

Region 1 Region 2 Region 3 Region 4

Population (thousands) 58,000 30,000 25,000 16,000

Poverty incidence (percent) 25.0 23.0 16.0 14.0

GDP (millions of pesos) 15,000 12,000 11,000 10,000

Source: Philippine statistical yearbook 1998. National Statistics Office.

8.4 Alignment of tables. Relative to a page, tables may be constructed either flush
left or centered. Most popular word processors, however, now make it easier to
center tables5. Writing them flush left (or right) on the other hand will require you to

5This is in stark contrast to the era of the typewriter, when centering tables was virtually a
major craft skill.

40
fiddle around manually with the width of columns. Unless you are willing to go take
the extra trouble, therefore, it is expedient to adopt center-alignment.

8.5 Source of data. This should always be indicated at the bottom of the table (Ex.
63). (Another reason for adopting center-alignment of tables is that it then becomes
easier to center the information on the data source, as shown in Ex. 63.)

8.6 Annotations. It sometimes becomes necessary to annotate either the labels or the
data in a table. When only one (at most two) annotations are needed, asterisks (*)
and double asterisks (**) will suffice as annotation symbols. When more are needed,
use superscript letters (a, b, c, ...) as annotation symbols for numbers and
superscript numbers (1, 2, 3,...) as annotation symbols for labels.

8.7 Keep the use of gridlines to a minimum, since they contribute to Moiré effects.
Instead, allow for sufficient space to keep data entries and categories separate and
distinct. You may also experiment with different shades and tones (but again,
sparingly) to separate rows and columns.

8.8 Econometric results. The older tradition in reporting econometric results was to
write out the estimated equation as a displayed equation in the body text. This made
enough sense when the number of independent variables used was small and
diagnostic tests to be reported were few. The latter were usually confined to the
standard errors or t-statistics and adjusted R2 (the Durbin-Watson statistic already
being somewhat fancy).

Practice has since evolved, however, to favour the use of tables to display
econometric results. This method is especially apt when the number of regressors is
large, or when different specifications of the equation have been estimated. A table
such as that in Ex. 65 is an acceptable format and incorporates some of these
suggestions. If need be, such tables may occupy a whole page and may be reported
using a landscape (horizontal) lay-out. Some pointers follow:

(1) Display the name of the dependent variable prominently in the title of the
table.

(2) wLet the columns of the table refer to different specifications seeking to
explain the dependent variable, with the regressors being read from the left-
most column. The various statistics should also be listed in tabular form.

(3) Write out the names of the dependent and independent variables fully
(first column of Ex. 65). Do not use the abbreviations or code-names you

41
 have used in the econometrics or statistical software, e.g., YSUB1, CURDEF3,
DUM_CRIS8485, and so on. Remember this will be gibberish to your
readers, and you should not inflict it upon them.

(4) In addition to using the full names of variables, you will want to include a
separate table of variable definitions. For example, a variable called
“Inflation 1” may refer to inflation in the CPI, while “Inflation 2” may refer to
that in the GDP deflator.

(5) In statistical or econometric work, it is always prudent to include a table

of descriptive statistics for the sample, showing the means, standard errors,
minimum and maximum values of the variables used.
Example 65

Table 11.
Regression Equations
Dependent variable: Total investment as percent of GDP
(1957-1992)

Equation 1 Equation 2 Equation 3

Variable Coefficient t-statistic Coefficient t-statistic Coefficient t-statistic

Investment (lagged) 0.830203 6.1616 0.851514 9.1669 0.954097 13.0177

Cumulative current acct. -0.01886 -0.8602 -0.01714 -1.1020

deficits

Budget deficit -0.08453 -0.2490

Pre-election dummy -1.93822 -2.3258 -1.83723 -2.7461 -1.7726 -2.6027

Inflation -0.06165 -1.5278 -0.04989 -1.5179 -0.06364 -1.9513

Predicted GDP 0.686405 2.6374 0.577958 2.519 0.566809 2.4821

Terms of trade -0.01937 -1.3598 -0.02132 -2.0099

Price of investment goods -1.86047 -1.0085 -1.94722 -1.3593 -2.19253 -1.7814

Constant 11.87336 1.2387 12.72905 2.0023 8.647615 1.5296

R-squared 0.911112 0.913489 0.90146

Adjusted R-squared 0.880343 0.895678 0.88779

S.E. of regression 1.728786 1.660114 1.721809

Sum of squared residuals 77.70627 93.7033 106.7266

Log likelihood -64.9313 -76.4472 -79.18

Durbin-Watson statistic 1.855692 1.927641 1.817249

Akaike info criterion 1.324973 1.183416 1.218315

Schwartz criterion 1.764839 1.514401 1.466553

F-statistic 29.61137 51.28768 65.87077

42
10. Diagrams and graphs
Graphs and diagrams are a powerful aid to intuition and indeed were at one time
regarded as principal tools of analysis, as found, say, in the economics handed down
by Alfred Marshall .6 Notwithstanding the increasing resort to more general
algebraic and analytic methods, economics is still the social science that relies most
intensively on graphical techniques. Geometric argument, using slopes, intercepts,
and areas, is still part of the tool kit of economists.

As for statistical graphs, it will interest some to know that the time-series graph, the
bar chart, as well as the pie-graph, were all invented and first used by a Scots
political economist, William Playfair, whose Commercial and political arithmetic
[1786], appeared only ten years after the more famous work of his compatriot Adam
Smith. This bit of trivia should strengthen the confidence of economists and would-
be economists in using graphs liberally in order to help themselves and their readers
visualize information or follow an argument.

10.1 Graphs versus tables. A limited number of data points is always better
presented in the form of a table rather than a graph. This minimizes nondata ink and
in fact yields more information, since readers can read off the exact values of the data
themselves. Graphs, on the other hand, are more suited to showing a large amount of
data efficiently. For example, in order to show the “boom-bust” cycle of the
Philippine economy, a huge table showing GDP growth rates from 1947 to 2000 is
less effective than a simple time-series line graph.

10.2 Less is more. Precisely because graphs and diagrams in economics are not mere
embellishments but an aid to argument and intuition, they should contain no more
and no less than is needed to convey the points being made. This caution is especially
important today when computer software makes it increasingly easy to incorporate
various decorative elements in charts.

Ex. 66 below is an illustration of what can easily be done using computer software.
Three-dimensional effects are only a button-click away. It is far from evident,
however, that “enhanced” presentations such as these are superior to a plain old two-
dimensional graph, or even a simple table showing all the annual values.

6Joan Robinson’s Economics of imperfect competition is a classic example of geometric

argument.

43
The “enhancement” in fact makes it even more difficult to read off the values of each
observation. It is rich in what Tufte calls “nondata ink”, that is, ink used to convey no
information. If you consider the example, there are actually only twelve pieces of
data, namely, six years and six growth rates, but very little of the ink used gives
information regarding what those data points are.
Example 66 (Bad)

GDP growth

6
5
4
3
2
1
0
-1
-2
90 91 92
93 94
95 96
97 98 99 '00

Example 67 (Better)

Real GDP growth

percent 1990-2000
6
5
4
3
2
1
0
-1
-2
90 91 92 93 94 95 96 97 98 99 '00

44
 Example 68 (Even better)

Real GDP growth

(1990-2000)
percent
7
6
5
4
3
2
1
0
-1
-2
90 91 92 93 94 95 96 97 98 99 '00

Exs. 67 and 68 are improvements over 66. The ink has been considerably lessened
and a more subdued format allows the eye to focus on the data themselves. Note how
the year-labels have been pushed to the bottom of the graph rather than attached to
the abscissa. This is done so that the data points do not hit the labels. (Tufte [1983]
is partial to a format similar to that displayed in Ex. 68.) Indeed, it may be argued
that a bar graph is more appropriate than a line graph in this case, since the latter
would inaccurately suggest that data points were available for intra-year intervals,
when in fact the data are annual figures.

10.3 Accurate drawing. When drawing two- or three- dimensional diagrams,

observe the mathematical convention that assigns the abscissa (horizontal axis) to
measurements of the independent variable and the ordinate (vertical axis) to
measurements of the dependent variable. This is the rule, after all, notwithstanding
some famous exceptions. The most obvious one is the famous Marshallian demand
and supply diagram, which measures quantities supplied and demanded on the
horizontal axis, and price on the vertical axis. Yet the common functional
representations of demand and supply curves, namely, D = f (P) and S = g(P), clearly
indicate that price is the independent variable while the quantities D and S are the
dependent variables. (Why this is so is a piece of Dogmengeschichte.7)

7Why the Marshallian cross should be an exception may be a puzzle to most students. The
answer, of course, is that Marshall originally posited price as the dependent variable, being
determined by quantity demanded and quantity supplied, a formulation that is entirely
consistent with the diagram now in use. The reversal of the causation occurred when most
economists adopted L. Walras’s formulation of supply and demand, i.e., quantities being
dependent on price, without abandoning Marshall’s diagram which had by then become
famous. This is another example of path-dependence.

45
10.4 Integrating graphs and tables with text. In final drafts, tables, statistical
graphs, and diagrams should be integrated with the text. This means the relevant
table or graph should generally follow immediately after the paragraph in which it is
first mentioned. On the other hand, a table should never be truncated so that its rows
spill onto the next page. If the latter threatens to occur, simply carry over the entire
table to the next page and replace the space it would have taken up with text in order
to avoid white space between the printed text and the bottom of the page.

Whenever its size warrants, a table or figure should be printed on a separate page.
The page containing the table or figure should then follow immediately after the page
where it is first used or referred to. This rule applies even if two or more tables or
figures are mentioned in a single page of text. Hence, for instance, if the text on page
11 mentions Table 8 and Figure 2, each of which occupies a separate page, then page
12 must contain Table 8, and page 13 must contain Figure 2.

10.5
For journal submissions, tables and diagrams should not be integrated in

the main text but rather collected in the last pages of the manuscript. Each figure and
each table of significant size must be printed on a separate page. Two and at most
three small tables may be combined on one page. Items must then be clearly labelled
(Tables 1, 2, ..., and Figures 1, 2, ...) and collated in the following order: tables first,
then statistical or analytical figures. The position of the table or figure in the main
body of the text is then indicated by a caption centred on the page: (Place Figure n
about here).
Example 69
... Table 10 gives comparative data for provinces showing GDP per capita and poverty
incidence.
[Place Table 10 here.]
The weak relationship between the two suggests the importance of
distributional factors...

10.6 Ensure that the formatting used in the text is maintained in diagrams. For
example, if a variable is italicized in the text, it must also be italicized in the
diagrams.

10.7 As much as possible, diagrams and graphs must be scaled to be wider than they
are higher [Tufte 1983]. The soundness of this advice rests on the eye's being
accustomed to viewing the horizon, the ease of labeling, the emphasis on causal
influence, and centuries of experience. A height-width ratio between 1:1½ and 1:2 is
recommended, with the former being the rule. This range is an obvious

46
approximation of the “golden ratio”8, which is 1: 1.618… . In the figure below, the
shaded rectangle is in the proportion 1:2, while the unshaded rectangle has the ratio
1:1½. Ex. 67 and 68 above show time-series graphs using the suggested 1:1½ scale.

1 : 1½

1:2

If a graph inherently demands a particular scale, then obviously that scale should be
used. An example is the “Keynesian cross” in the standard macroeconomic
treatment, which relies on a 45-degree line to determine equilibrium. Another is the
Lorenz curve displaying the inequality of a distribution. Then, of course, there is
Brouwer's fixed point theorem, which states that for a continuous function f defined
over [0,1] with a closed range f(x) ∈[0,1] there exists at least one point x0 where x0 =
f(x0). These examples obviously require square graphs, so that the suggested
proportions cannot be applied.

10.8 The axes of each graph should be clearly labeled by the variable being measured
and the units of measurement. First, axis labels should read in the natural order,
from left to right and not -- as is increasingly common -- from top to bottom or
bottom to top. Second -- a somewhat more controversial suggestion by Tufte [1983]-
- the usual ticks or divisions of the value axis are dispensed with and the values of the
statistical observations themselves are displayed instead. This practice is of greater
aid to the reader, since in addition to portraying the time trend, he is able to retrieve
the actual values of each observation, obviating the need for a table in addition to the
graph. Of course, this advice will be difficult to follow in the case of large data sets.

8One way of deriving the golden ratio is first to generate the Fibonacci series, 0, 1, 1, 3, 4, 7, …,
where the next number in the series is derived as the sum of the preceding two. Skipping 0,
the golden ratio is the limit of the quotient between the larger and the smaller of two adjacent
numbers in the Fibonacci series, or the irrational number 1.61804….

47
10.9 Considering the rich structure of theoretical models in economics, it is
surprising that most statistical graphs occurring in the discipline are no more than
simple time series (using lines or bars), with the horizontal axis corresponding to
time and the vertical axis referring to some variable in question. There are various
reasons for this, but the most important is that most functional forms are
multivariate and cannot be readily observed as a two-dimensional relationship.
Whenever the two-variable functional relationship is strong enough, however, two
graphical forms can be used with striking effect. The first is a two-axis formulation
of a time series. Ex. 71 shows an approximately inverse relationship between two
variables through time.
Example 71

GDP (% ) Inflation (% )
8 20

6 18

4 16
14
2
12
0
10
85

87

89

91

93

95

97

99

'01
-2
8
-4
6
-6 GDP (Left scale)
4
-8 Inflation (Right scale)
2

-10 0

Often even better than the above, however, is the scatter-plot diagram, which shows
the relationship between two variables. Ex. 72 uses the same data as Ex. 71 but more
clearly illustrates the approximately negative relationship between output growth, on
the one hand, and inflation, on the other. This has been done by suppressing the
time dimension and using one of the variables (GDP growth) as the horizontal axis.
The time dimension may be recovered by labeling the x-y plots with the relevant time
periods.

48
 Example 72
Inflation (% )
20
18
16
14
12
10
8
6
4
2
GDP grow th (%)
0
-10 -8 -6 -4 -2 0 2 4 6 8

10.10 Graphs and tables of real (deflated) magnitudes are almost always superior to
those depicting only nominal magnitudes. For example, a graph of the change in
nominal GNP by itself is almost useless for most purposes. (One exception is the
relationship between money stock and nominal GNP, which gives the income-
velocity of money.) Values across heterogeneous entities are also more meaningful
when scaled. In evaluating saving behavior across countries, for example, it rarely
makes sense to compare the absolute level of saving in countries of different size.
More relevant is a comparison of saving rates across countries, with the size of GNP
being the scale factor. Nominal magnitudes are usually meaningful only when scaled.
For example, nominal money stock and nominal GNP on their own are not terribly
significant, but their quotient is interesting (e.g., as a measure of monetary
tightness).

10.11 Pie charts are rarely essential. Whether flat, three-dimensional or “exploded”,
pie charts are not particularly useful in technical papers. Neither are bar charts that
merely show a cross-section distribution without a time dimension.9 There is no
information contained in these charts that is not better conveyed by a table -- with
less ink to boot. Possibly the only purpose of pie-charts and cross-section bar charts
is to dramatize or give some visual “feel” to numbers to an audience that is
numerically numb or has a poorly developed sense of magnitude. Such an audience,
however, is certainly not what you should presume to confront when writing
technical papers. Save the pie charts for public presentations.

9The inferiority of cross-section bar charts was recognized even by their inventor, William
Playfair.

49
10.12 Avoid using grid lines on your graphs. If you must use them, lighten the tone to
avoid distracting Moiré effects.

10.13 Ensure the mathematical accuracy of diagrams. This is a matter of scruple. It is

too often forgotten (even by some reputable authors), that the spatial relationship
between a linear demand curve and the marginal revenue curve is not arbitrary but
dictated by mathematical relationships. The right-hand diagram in Example 73 is
obviously misdrawn, but one finds something like it in a popular textbook on
international economics. Part of this is admittedly due to the limitations of ordinary
drawing tools on word-processing program. Still, this is no excuse to display a
diagram that is plainly wrong.
Example 73

P P

/ /

MR D MR D
Q Q
0 a/2b a/b 0
If the (inverse) demand curve is P = a – bQ, total revenue is R = aQ – Q2, and the
associated marginal revenue curve is MR = a – 2bQ.Graphically, therefore, the
vertical intercept is the common point (0, a), while the horizontal intercepts are
(a/b, 0) and (a/2b, 0) for the demand curve and the marginal revenue curve,
respectively. This implies that the MR curve bisects the distance a/b and indeed any
horizontal line from the vertical axis to the demand curve. A drawing such as that on
the right above is thus patently wrong.

10.14 It is almost unavoidable to use letters in referring to specific elements such as

points, lines, or segments found in diagrams and drawings. Unlike variables and
function names, these elements should not be italicized, either in the drawings
themselves or in the text (Ex. 74). By contrast, the labels D and MR of the curves in
Ex. 73 are italicized, since these are the graphs of functions with the same names.
Example 74
The wage rate has been shown to equal f(k*) – k*f′′(k*), which in Figure 2 is
equal to AC – OC(AB/DC) =AC – OC(AB/OC), the latter following from DC
= OC. Thus in the figure the real wage is AC – AB = BC.

50
11. Envoi
This booklet has sought to provide guidance in various situations that technical
writers may confront. Some readers will be unaccustomed to the degree of attention
to detail demanded here and possibly raise questions over whether it is worthwhile
to spend this much time and effort on what are at bottom “mere” matters of form. To
borrow F. Lloyd Wright’s phrase, however, “form and function are one”, and very
often (though admittedly not always) slovenly form betrays a slovenly attitude in
matters of substance as well. At the very least, it indicates a certain disregard for the
reader which the latter may feel justified in repaying in kind.

Ultimately, one should hope that such a booklet as this should become redundant –
that is, if the guidelines it contained were to be as self-evident as, say, ending
sentences with full stops – so that it would be otiose to commit them to ink. But then
again, such an expectation is bound to be frustrated by the pressures of an
increasingly fast-paced world, inevitable human limits, and of course the emergence
of lively new usages, which continually challenge today’s conventions in order to
become tomorrow’s.❑

Acknowledgements
Sincere thanks are due, first of all, to a former colleague, Michael T. Alba of De La
Salle University, who constantly encouraged me to finish this project and who
thoroughly shares this booklet’s concern for educating students in good technical
writing and uncompromising attention to detail. A second debt is to Professor
Rolando A. Danao of the University of the Philippines School of Economics, who
combed through an earlier draft and, with his usual thoroughness, made valuable
suggestions. Remaining errors are the author’s sole responsibility.

51
References
Burchfield, R., ed. [1996] Fowler’s modern English usage. Third edition. Oxford:
Clarendon Press. (A redefined Fowler’s that is more broad-minded and
tolerant than the original; some would say excessively so.)
Fowler, H. ed. [1996(1965)] Modern English usage. Second edition. Oxford: Oxford
University Press. (A classic, whose advice is always sensible and informative
although, alas, since Fowler’s time, no longer infallible.)
Oxford University Press [1983] Hart’s rules for editors and compositors at the
University Press Oxford. Thirty-ninth edition. Oxford, New York. (A valuable
guide, although apt to cause some confusion among those reared on
American .English spelling and conventions.)
Kirszner, L. and S. Mandell [1999] The Holt handbook. Fifth edition. Forth Worth:
Harcourt Brace. (Contains good pointers on style and on doing research in
general, but uses the US “conventional style”, which differs from that
recommended here. Quite up to date, however, on conventions covering
electronic sources and documentation.)
Knuth, D., T. Larrabee, and P.M. Roberts [1989] Mathematical writing.
Mathematical Association of America. (Contains many suggestions of a high
quality and standard, compiled playfully as a set of notes in classes on
technical writing conducted by Donald Knuth, computer scientist and creator
of the mathematical writing software TEX. Language advice here leans
heavily on Van Leunen [1978].)
Strunk, W. and E.B. White [2000] Elements of style. Fourth edition. New York:
Longman. (If you have been fortunate, this will have been the style book you
were reared on in high school. It contains a few simple rules that have helped
many a perplexed student.)
Tufte, E. [1983] The visual display of quantitative information. Chesire,
Connecticut: Graphics Press. (Contains a brilliant exposition of principles to
be followed for graphic displays.)
Van Leunen, M.-C. [1978] A handbook for scholars. New York: Alfred Knopf. (Very
sensible but insists on the “conventional” approach typical of the US, as
opposed to the logical system advocated here.)

52
 APPENDIX

Frequent mistakes
and disputed usage
What follows are usages, occurring in economics writing and elsewhere, where
mistakes often occur, or where convention is less than clear-cut. The selection is a
judgement of what is typically beyond the pale of spell- and grammar-checks, and
what are pitfalls peculiar to local writing.
à la. Italicized, with a grave accent over the first “a”. EXAMPLE: Assume a
consumption function à la Keynes.
a posteriori (“from what comes after”, Latin). Two words, not italicized. (See a
priori.)
a priori (“from what comes before”, Latin). Two words, not italicized. In economics,
this adjective or adverb is typically used of statements that follow from
assumptions or definitions, hence not proceeding from empirical
observation. By contrast, a posteriori refers to statements that follow by
inference from observations.
abbreviations and acronyms. Full stops may be omitted in abbreviations
consisting entirely of capital letters (small capitals being preferred). Hence,
RP instead of R.P., and USA instead of U.S.A. Similarly, one has EU, UK, DIN,
GNP, BOP, and so on. Where words are truncated to include not only the
initials but also some succeeding letters, however, global capitalization must
be suspended and full stops inserted, such as for philosophiae in Ph.D. or
scientiae in D.Sc. (as contrasted with MA and AB), or economics in BS (Econ.),
or even the plain Dr., Ms., mm., in., m. and ft. Some quarters press for
dispensing with the full stops even in these latter cases (hence PhD, DSc, Dr,
Mr, mm, in), but this is far from being universally accepted. Acronyms are
abbreviations that can be pronounced as words, rather than the letters having
to be spelled out. Examples are ASEAN, AFTA, NAFTA, NEDA, AIDS, CALABARZON,
UNESCO, UNIDO, SLEX, and UNICEF. In some instances, one may even dispense
with the capitalization but for the first letter. Hence, Unesco, Unicef, Unido,
and at times Calabarzon, Socsargen, Neda, Slex, are acceptable. (Some
acronyms such as radar and laser have already joined the common nouns.)
This rule is not universal, however, since for example AIDS continues to be
spelled entirely in upper-case letters. The matter will at times turn on how an
entity wishes to style itself. When in doubt, remember it is never wrong to
capitalize all letters of an acronym. As a rule, fully write out the meaning of an
abbreviation that deals with the names of organizations, agencies, or
concepts when they are mentioned for the first time, enclosing either the
definition or the acronym itself in parentheses. Hence: Third-quarter gross
national product (GNP) rose 4.5 percent in the third quarter. NFIA (net factor
income from abroad) was responsible for growth during most of the year.
The abbreviation may subsequently be used alone.
accelerate/decelerate. Economics writers must frequently translate quantitative
arguments or events into plain prose, and are tempted to look for synonyms
and alternative expressions to prevent the prose explanation from becoming
monotonous. In this honest effort, however, errors may creep in, especially
when the discussion moves from levels to rates of change. To say that (the
rate of) inflation accelerated, for instance, is not the same thing as saying
that inflation rose or that the price increases accelerated. If π is the rate of
inflation, the latter statements simply mean dπ/dt > 0, such as when inflation

53
 rises from 8 percent to 10 percent per annum. On the other hand, the
inflation accelerating may be interpreted to mean d2π/dt 2 > 0, which would
happen if, discretely speaking, from t to t + 1 inflation had risen by only one
percentage point but it had risen by two percentage points from t + 1 to t + 2.
To use accelerate scrupulously to refer to dπ/dt > 0, one would have to say
that price increases accelerated. Similar caution applies when one talks
about any sort of growth rate, including the GDP growth rate.
ad hoc (“to this”, Latin). Italicize. The phrase connotes “for the purpose at hand” and
is used as an adjective. The term has gained a negative connotation in
economics that borders on the sense of being improvisational and not being
solidly grounded in theory. We resort to the ad hoc assumption of linearity to
arrive at the following form. Derivatives include ad-hoc-ery and ad-hoc-ism,
written at times without hyphens.
adapt/adopt. It is a common mistake not to distinguish sufficiently between these
two. To adopt something means to take over something wholesale; to adapt
something, on the other hand, means to change parts of it to suit a wholly or
slightly different purpose. It is almost always inadequate to simply say
something was adapted without further explanation of what other purpose it
was used for. When adapt is used, the question almost always is: adapted to
what? Such a question does not necessarily arise in the case of adopt. WRONG
or AMBIGUOUS: Marx adapted Ricardo’s theory of value. BETTER: (1) Marx
adopted large parts Ricardo’s theory of value. (2) Marx adapted Ricardo’s
labor theory of value to his critique of capitalism.
addendum (“to be added”, Latin). Has the same meaning as appendix, but the
latter is more standard. Plural: addenda. Do not italicize.
advertise. Thus spelled. NOT advertize.
advice/advise. The first word, written with “c”, is a noun, while the latter with an
“s” is a verb. Hence The economist gave the candidate some policy-advice
(noun). That is, The economist advised the candidate on policy (verb). (See
also noncount nouns.)
age. As a verb, its progressive form is ageing or aging, with a preference for the
former.
agenda. Formerly a plural form, this is now regarded as a singular noun, in the
sense of a list of things. Hence the plural agendas refers to many such lists
[Burchfield 1996]. To refer to the many elements in such a list, one must now
say items on the agenda. Note the correct preposition, however: items are on
the agenda, not in it.
-al. A trend currently gaining ground is the omission of the terminal or medial -al in
a number of modifiers. Hence, educational workers→ education workers;
agricultural systems → agriculture systems; agriculturalist → agriculturist;
graphical design → graphic design; lexicographical ordering →
lexicographic ordering. Economy largely favors this change, and it may be
adopted whenever meaning is not sacrificed. (But see also
economics/economic/economical.)
allocate. The proper idiom is allocate to or allocate towards, NOT allocate on.
EXAMPLES: In competitive equilibrium, resources are allocated to their most
efficient uses. Not enough resources were allocated to (NOT on) human
development priorities. (See also invest in)
alternate v. alternative. The word to describe things that are replacements or
reserves for something that is in customary or traditional use is alternative,

54
 not alternate. In customary use, the latter adjective means coming after the
occurrence of some other type of object, as in the case of alternate colors,
alternate layers. WRONG: alternate energy sources, alternate technologies.
RIGHT: alternative energy sources, technologies, etc. [Burchfield 1996].

analyse/analyze. For genuine etymological reasons, analyse is to be preferred,

even if analyze is established in US English.
and. In an enumeration of three or more elements, contrary advice notwithstanding,
do not dispense with the final comma or semicolon before and. It is never
wrong and avoids needless ambiguity in technical writing. EXAMPLE: Smith’s
minimalist state provided national defense, justice, peace and order, and
public works.
appendix. Plural, appendices.
apostrophes. You will rarely go wrong if you follow Strunk and White’s rule on
possessives, which is to use ‘s almost everywhere, even if the name ends in s.
Hence, Keynes’s, Walras’s, Marx’s, Hicks’s, and von Mises’s. More than this,
however, Hart’s rules suggests that ‘s be used in all one-syllable and two-
syllable words (which would include Marcos’s wealth and Congress’s
reaction). A simple apostrophe is to be used, however, where the word is
more than two syllables, ends in s, but is not stressed on the penultimate (last
but one) syllable, e.g., Lákatos’ methodology of scientific research
programmes, Heródotus’ history.
arguable/arguably v. disputable/disputably. Arguable is a positive expression,
unlike disputable which has a negative connotation. EXAMPLES: It is arguable
that the use of game theory was one of the most significant developments in
industrial organization in the last two decades. It is disputable whether the
neo-Keynesians are still true to Keynes’s original vision. In the first example,
the writer is inclined to agree with the proposition, in the second, to disagree
with it. The corresponding adverbs are arguably and disputably.
as far as… is concerned. Other variants: as far as ...goes, or insofar as ... is
concerned. Emulating a creeping tendency among speakers of US English,
some writers truncate the phrase as far as X is concerned, or in so far as X is
concerned to simply as far as X. This is at best a colloquialism and at worst a
mistake. EXAMPLE (WRONG): As far as last year’s employment growth, much
of it is really self-employment. CORRECT: As far as last year’s employment
growth is concerned (or As far as ... goes), much of it is really self-
employment. Burchfield [1996: 184] notes that the verb is needed.
Ultimately, however, these are wordy formulas that frequently can (and
should) be eliminated. ORIGINAL: There is little that has been done as far as
poverty alleviation is concerned. BETTER: Little has been done about poverty.
ORIGINAL: As far as the current account balance is concerned, a deficit was
registered. BETTER: The current account was in deficit.(See also
concerned.)
as. A number of writers seem to think this word is an exact substitute for “since” or
“because”: As the government failed to control the deficit over three
quarters, the fall in bond-prices was expected. This statement may be greatly
improved, however, by replacing as with since or because, since a causal
relationship is obviously implied. The use of as is warranted when a mere
correlation is implied: As agriculture fell victim to El Niño, industry
recovered from the previous year’s recession.

55
assumption. The idiom is under the assumption, NOT on the assumption, e.g.,
Under the assumption of perfect competition, profit maximization implies
that price equals marginal cost.
autarky versus autarchy. The former, which is the proper term in international
trade, refers to self-sufficiency; the former refers to self-rule. Neither is
necessary or sufficient for the other.
avail. There are two forms in which this word is used as a verb, one form involving a
reflexive pronoun. EXAMPLE: The poor availed themselves of subsidized
social services. Soon enough you will find that the reflexive pronoun may be
omitted altogether without any loss in meaning: The poor availed of
subsidized social services.
based. The idiom, when referring to abstract concepts, is based ON, e.g., based on
assumptions, based on experience, based on previous discussion, based on
criteria, etc., and NOT based in (or based at). The latter two are generally
used only when referring to physical location, e.g., The European
Commission is based in Brussels. The repair facilities are based at the
airport.
between versus among. It is wrong to think that between should be used only in
relationships between two, while among is to be used when there are three or
more. The Oxford English Dictionary (quoted by Burchfield [1996:106])
notes that between “is still the only word available to express the relation of a
thing to many surrounding things severally and individually, among
expressing a relation to them collectively and vaguely”. Burchfield
enumerates the OED’s division of the uses of between: “of simple position; of
intervening space; of relation to things acting conjointly or participating in
action; of separation.”. This is especially important in verbally expressing a
partition: Total time T is divided between paid work (W), unpaid work (H),
and leisure (L), i.e., T = W + H + L. By contrast, one would say that profits
were distributed among the shareholders. In graphs, one has no alternative
except to say that x lies between the two points y and z, but also that it lies
between three points y, w, and z.
between X and me/you/him/her/us/them. These are the proper usages, since
between is a preposition that takes the relevant pronoun as an object. The
frequently encountered usages between Pedro and I/ he/she/they are
colloquial but grammatically wrong. WRONG: Just between you and I, his
latest theorem sounds fishy. CORRECT: Just between you and me .... WRONG:
Differences between Kalecki and he could not be resolved. CORRECT:
Differences between Kalecki and him could not be resolved. (See also
who/whom.)
biannual versus biennial. Biannual means “twice a year”, while biennial has
always been used in the sense of “every two years”. Burchfield [1996] notes,
however, that these distinctions are less than clear-cut. Semestral can be
used rather than biannual, where applicable (note, however, that a twice-
yearly occurrence does not always mean something takes place every
semester.)
ceteris paribus (“other things being equal”, Latin; pronounced keteris pahribus).
Use italics. Increasingly, however, people get by just as well by writing the
equivalent English phrases, everything else being equal or holding
everything else constant. (See also mutatis mutandis.)
circumstances. The indisputably correct form is in the circumstances. But under
the circumstances is now also becoming accepted [Burchfield 1996].

56
compare with/compared to. The phrase compared with emphasizes the
similarity between two objects: The bursting of the real-property bubble in
the Philippines has been compared with that which occurred in Thailand.
The phrase compared to, however, calls attention to differences: Compared
to Malaysia, the Philippines was more inward-looking in the 1970s.
compatible. This is construed with the preposition with, NOT to. This time-series
program is compatible with (NOT to) a Windows™ environment. (See also
to.)
comprise versus compose. Both verbs relate smaller entities to a larger entity
containing them. Used as a transitive verb in the active voice, however,
comprise takes only the larger entity as a subject. On the other hand the
subject of compose is the smaller entities that make up the larger entity.
EXAMPLES: In a simple economy, output comprises consumption and
investment. In a simple economy, consumption and investment compose
output. In set-theoretic terms, let A and B be a partition of C, that is, A ∪ B =
C with A ∩ B =∅. One may then say that A and B compose, or make up C.
Alternatively, one may say that C comprises A and B. One often encounters
the following, however. DISPUTABLE: General government is comprised of the
central or national government and local government units. ALSO
DISPUTABLE: General government is comprised by the central or national
government and local government units. CORRECT: General government is
composed of the central or national government and local government
units. ALTERNATIVELY: General government comprises the central or
national government and local government units.
concerned. The verb itself is probably overused, (see as far as... is concerned).
Distinctions must also be made between being concerned with something,
which means being pre-occupied or having to deal with it and being
concerned over or about something, which means to be worried over it.
Economics is concerned with numbers. [Better yet: Economics concerns
numbers.] The researcher was really concerned about (or over) the numbers
when he saw them.
consider (as); consider to be. One often encounters the following usage: The
central bank considers inflation as the primary threat to sustained growth.
In almost all cases, including this one, as is unnecessary. IMPROVED: The
central bank considers inflation the primary threat to sustained growth..
This practice likely owes to the apparent similarity between consider and
regard. In the latter case, as cannot be dispensed with: The central bank
regards inflation as the primary threat to sustained growth.
consideration. A common mistake is the construction, give consideration on. The
proper construction is give consideration to, as in: In implementing
spending cuts, the government gives special consideration to essential social
services.
contrary. The phrases on the contrary or to the contrary should not be used at the
beginning of a sentence [Burchfield 1996]. One may replace these with in
contrast, or on the other hand.
contrasted with/in contrast to. Avoid the following awkward construction: The
low inflation this year may be favorably contrasted to the surge in prices last
year. IMPROVED: The low inflation this year may be favorably contrasted with
the surge in prices last year. ALTERNATIVELY: The low inflation this year is in
sharp contrast to the surge in prices last year.

57
cope with versus cope up. Sentences similar to the following are often
encountered: The poor cannot cope up with inflation. CORRECTED: The poor
cannot cope with inflation. To cope with something means to deal adequately
with it or to adapt to it. To cope up, in AmE, means to measure up or to be
good enough in the abstract, as in They speeded up the production line, but
the workers could not cope up. A careful consideration will often reveal that
the word up may be dropped in all such formulations without changing the
sense of the sentence. The odd formulation cope up with is likely the result of
confusing keep up with cope up.
corrigendum (a thing to be corrected, Latin). Plural corrigenda. No italics.
criteria. Take care to note that this is the plural form of criterion. EXAMPLE: The
criteria for a country to be called a NIC are ambiguous at best. Burchfield
[1996] notes some evidence of some use of criteria as a singular noun
(following data and agenda), but this usage is disputed.
critique. Originally a noun, this term is now legitimately used as a verb: Joan
Robinson critiqued Marshall’s attempt to squeeze an essentially dynamic
problem into the shoebox of comparative statics. But criticize often does just
as well and is less likely to raise eyebrows.
data. This is the plural of datum and therefore governs plural verbs and pronouns:
The data are not regularly available, nor are they always current. In US
English, however, the word is increasingly used as if it were a singular noun,
by the same token that news is singular.
dates. Most periods are written in forms such as the following: the 1984-85
recession; the 1991-1992 recession, etc. When referring to the behavior of a
statistic over a period, be careful to avoid the following error: The investment
rate rose from 12 to 18 percent from 1985-1995. The dash is not a substitute
for the word “to”. Instead one should have written: The investment rate rose
from 12 percent in 1985 to
18 percent in 1995, or ... in the period 1985-1995.
demand (for). Used as a verb, demand takes a direct object. Hence: The workers
demanded a wage increase (NOT demanded for a wage increase). WRONG
(Read in some establishments): Please demand for a receipt. RIGHT: Please
demand a receipt. The noun, however, requires a preposition: The workers’
demand for a wage increase was particularly ill-timed. The same rule
applies to order (for). (See also: request (for) and seek (for).)
depreciate/appreciate. One frequently encounters the formulation: The
exchange rate depreciated. This is wrong, strictly speaking, since depreciate
means for the price of a thing to be reduced. Since the exchange rate is
already a price (of one currency in terms of another), this is tantamount to
the absurd statement that the price of a price was reduced. Clearly, the only
acceptable form is: The currency (peso, baht, etc.) depreciated, which
amounts to saying The exchange rate rose (or fell), depending on how the
exchange rate is quoted. It is clearly a matter of indifference whether one
quotes the exchange rate as (a) peso-dollar rate (e.g., 40 pesos to the dollar),
or (b) a dollar-peso rate (e.g., 1/40 dollars per peso). The measure of the rate
of depreciation (or appreciation) will differ, however, depending on how the
exchange rate is quoted. If e denotes the local currency price of foreign
currency (e.g. pesos per dollar), then method (a) gives (e1 - e0)/e0, where e0
and e1 are past and present values, respectively; on the other hand,
convention (b)gives [(1/e0) - (1/e1)]/(1/e0) = (e1 - e0)/e1; the difference lies in
the denominator. Hence a depreciation of the peso from P25 to P40 would be

58
 measured as 15/25, or 60 percent, under method (a), and 15/40 = 37 percent
under method (b). (See also overvalued/undervalued exchange rate.)
different from/different than/different to. The undisputed form is different
from, as in The pattern of Taiwanese growth is different from that of Korean
growth.” “Different than” is disputed; “different to” is simply wrong. (See
below.)
differently than/from. In US writing and speech, one often encounters something
similar to the following: South Korea industrialized differently than Taiwan,
or The policy was implemented differently than in the past. This returns one
to the dispute over different from and different than (see preceding). This
problem may be avoided by saying either of the following: South Korea’s
industrialization was different from Taiwan’s. (See Burchfield [1996].)
discuss/discussion (about). This verb takes a direct object, and it is definitely
wrong to append about to it. WRONG: The open-market committee discussed
about the new T-bill issue. CORRECT: The open-market committee discussed
the new T-bill issue. It is correct to refer, however, to the open-market
committee’s discussion about the new T-bill issue. (See also request (for)
and demand (for).
distortionary. For taxes, policies, etc., use distortive instead.
divide by/divide into. In the sense x ÷ y, one may write x divided by y, or x
divided into y.
due to. In an earlier, more innocent time, grammarians such as Fowler warned
against the practice of using due to as a prepositional phrase, as in the
following example: Due to the doubling in interest rates, investment fell by
20 points. Following the rule, one would have had to rewrite the sentence as
follows: The 20-point fall in investment was due to the doubling in interest
rates. Fowler was pushing a rule under which due could be used only to relate
two nouns. Noun X was due to Noun Y. If one wanted to relate a noun to a
verb, one ought to use owing to, or because of, or since. Hence: Owing to the
doubling in interest rates, investment fell by 20 points. Burchfield [1996]
notes, however, that this will soon be another example of a “forgotten battle”.
Some may still want to play safe, however, just in case the adviser or reader is
of an older school.
e.g. (= exempli gratia, “for (the sake of) example”, Latin). Separate from the rest of
the sentence with commas, and do not italicize. A sentence may begin with
the English version, “For example”. It is inelegant, however, to begin
sentences with “E.g.”.
economics/economic/economical. An economics teacher, an economic teacher,
and an economical teacher do not refer to the same thing. The first denotes
the person’s specialization; the second his sensible attitude towards the use of
any type of resource; and the third his thrifty or parsimonious nature. Hence,
an economics teacher will not necessarily be economical, nor even economic.
The same thing goes for an objects or persons modified by these adjectives,
e.g., an economics/economic/economical strategy, suggestion, person, and so
on.
endogenous. Thus spelled; NOT endogeneous.
end-user, end-product, end-result. Remember the hyphens.
equilibrium. Plural: equilibria. IDIOM: in (or out of) competitive equilibrium, NOT
under competitive equilibrium.

59
erratum. (error, Latin) Plural: errata. No italics.
et al. (et alia, or et alii, and other persons, Latin). No italics. Treat the et as the
usual English “and”. Hence when used to refer to “the authors Herrin et al.”,
no comma should precede; but in the form “the authors Herrin, Solon, Alba,
et al.”, insert commas after each element in the enumeration.
etc. (= et cetera, “and other things”, Latin). No italics. In an enumeration where two
or more members are mentioned, it should be preceded by a comma, e.g.,
Wealth taxes, income taxes, head taxes, etc. are notoriously difficult to
collect. On no account is it correct to write X, Y, and etc., since et itself means
and.
evidence. See noncount nouns.
ex ante/ex post (“before/after the fact”, Latin). Italicize.
exogenous. Thus spelled, NOT exogeneous.
expand/expound. It is a frequent local mistake to confuse the two. While
expansive can mean verbose and long-winded, the verb expand cannot be
used to mean “to conduct a discourse”. Use expound instead, e.g., The central
banker expounded on (or upon) the merits of single currency.
fait accompli (“accomplished deed or fact”, French), Plural: faits accomplis. Italicize.
first of all. This is acceptable, but to follow it with second of all, third of all, etc. is
ridiculous. Instead, use, second, third, etc. (See below.)
first(ly), second(ly), third(ly), etc. Some authorities, including Burchfield
[1996], frown on “firstly, secondly, thirdly,…”. Indeed, closer reflection will
reveal the extra letters to be unnecessary; hence, first, second, third, ... .
floors and ceilings. These are metaphors that refer to stipulated maximal or
minimal values of certain variables, usually to be enforced as a matter of
policy. Hence, a “price ceiling” refers to the maximum allowable level of a
price, say p0. Implicitly, therefore, the target zone for a price is p ≤ p0.
Analogously, a “wage floor”, such as a minimum wage, w0, entails requiring
that w ≥ w0. (See also target.) Some verbs do not sustain these metaphors
and will jar the sensibilities, e.g., The price-floor for palay was raised. In the
New Fowler’s, Burchfield warns against combining the metaphorical use of
floors and ceilings with verbs that properly pertain to levels, e.g., extend,
lower, raise, waive.
-fold. n-fold means n times, as in a threefold blessing, a twofold reason, a manifold
answer, etc. An ambiguity arises, however, when the formula is used in a
sentence such as Average income increased threefold. The sense meant is
probably that some quantity x (in the example, income) increased to 3x over a
period of time. The phrase could also be interpreted to mean, however, that a
doubling of the quantity took place three times, i.e., x rising to 2(2(2x)) =
23x. Hence an n-fold increase could be taken to mean an increase from x to
2nx.
for. In the sense of “since”, for must be preceded by a comma: Demand for foreign
exchange was high, for a depreciation was anticipated.
given. A frequent formula in scholastic writing is the following: Given such and
such, then so and so. On the other hand, given so and so, then such and such.
This pattern is more suited to logic texts than to real prose. Used once or
twice, it is tolerable but it rapidly becomes tedious. There are many other
ways of saying the same thing

60
gross national/domestic product. Do not capitalize the initial letters except in
the abbreviations GNP and GDP.
hence. Literally meaning “from here” or “from this”; a synonym for “therefore”.
May also be used in an elliptical sentence: Hence the answer.
heteroskedastic/homoskedastic. Spelled with a “k”. The older spelling with a
“c”, i.e., heteroscedastic-homoscedastic, has all but disappeared.
homogeneous. To be distinguished from homogenous. The first is the quality of
having the same form, such as when a function is said to be homogeneous
of degree one. The second means having a common ancestry.
hyphenation. Hyphens become necessary when compound adjectives occur before
the nouns they modify. Compound adjectives are combinations of several
words functioning as a single unit to modify a noun. EXAMPLES: (a) The exam
consisted of multiple-choice questions. (b) Steady-state income growth
equals that of population. (c) A uniform five-percent tariff makes sense not
on static-efficiency but on political-economy grounds.
i.e. (= id est, “that is”, Latin). Separate from the rest of the sentence with commas.
Do not italicize. This should be distinguished from “e.g.” What follows “i.e.”
must be an exhaustive enumeration of the set, while “e.g.” requires only a
subset.
impact. This word is now increasingly used to mean “short-run effect” or
“immediate effect”, but it should be really be treated more cautiously, if not
altogether avoided. One also sees it being used in the plural, e.g., impacts of
liberalization. Even more surprising is its use as a modifier, to mean
immediate or short-run, as in impact-effects of the WTO agreement. This is
redundant, since impact itself already means a strong effect or influence.
implication. Properly construed with the prepositions for or on. WRONG: A
devaluation has serious implications to inflation. RIGHT: A devaluation is
thought to have serious implications for (or on) inflation.
in lieu of (“in place of”, French) No italics.
indefinite pronouns. A recurring question is whether indefnite pronouns like
anybody, everyone, everybody, no one, nobody, none take singular or plural
forms of verbs. The answer is that in general, they all take the singular form:
Hardly anybody computes EPRs anymore. Everybody wants a piece of the
action. None of the investors is willing to take the plunge. Some
complications are caused, however, by a desire to avoid awkwardness or
sexist labels, leading to the use of plural forms in the same sentence.
Consistency would require one to write Everyone wants to have his say on
the matter. But to avoid the charge of sexism, one might attempt the
awkward Everyone wants to have his or her say on the matter. But many
writers now say Everyone has their say on the matter. Here, wants is a
singular verb, but their is a plural pronoun, an obvious inconsistency. This is
not wrong, although one would wish some happier formulation had been
tried. (See also sexist language.) Some writers appear to pre-empt the
charge by adopting the female singular pronoun as the default form:
Everyone has her say on the matter. It is all very well calling attention to
one’s political correctness, but such cases as Everbody must be held liable for
her crimes is unflattering at least and confusing at worst. BETTER: One should
be held liable for one’s crimes. (See also one and sexist language.)
index. In the sense of a statistic, such as a price index, or a production index, the
plural is indices not indexes.

61
indifferent. There are two common meanings of indifferent: (1) to express
subjective opinion, as in Consumers are indifferent to the blandishments of
advertising, and (2) to describe absolute inferiority, as in x is an indifferent
product. Neither of these is how economists use the term. Instead they use it
to refer to a binary relation between two objects x and y, neither of which is
superior to the other. The frequent mistake is to extend the construction of x
is superior to y and x is inferior to y to this case and to say that x is
indifferent to y, probably influenced by sense (1). But this is mistaken and
misleading, since x is not capable of subjective opinion regarding y. The sense
being sought is the quality of being subjectively not different. This is best
expressed by writing that x is indifferent relative to (or with respect to, or in
relation to) y, OR, x and y are indifferent.
inflation. Refers not just to any increase in the general level of prices but to one
that is sustained. (See also accelerate/decelerate.)
interested. CORRECT: The government was interested in selling off assets to raise
revenue. DISPUTABLE: The government was interested to sell off assets to raise
revenue. WRONG: The government was interested on selling off assets to raise
revenue
invest, investment. The appropriate preposition is in NOT on. WRONG: Nations
should invest more on human capital. CORRECTED: Nations should invest
more in human capital (OR: should invest in human capital).
laissez-faire. (“let [someone] do”, French). The variant laisser faire is the infinitive,
i.e. “letting [someone] do”. In English, these terms are frequently used only
as modifiers (usually adjectives), as in laissez-faire doctrines, or laisser-faire
policies. The dash between the words then becomes indispensable. Italicize.
last/next. In the forms last (resp. next), March or last (resp. next) Tuesday, these
two words are used to refer to the most proximate past (resp. future)
occurrence of the month or day in question. A frequent mistake, however, is
to use them in referring to certain events that are essentially unique, and
which therefore by definition have no successor or predecessor. EXAMPLE:
Last December 2000 the central bank decided to intervene in the foreign
exchange market. December 2000 has, of course, already occurred and will
never come again. Hence to say last December 2000 is wrong, since it
suggests that there is a current or a future December 1999 aside from the one
just past. If this year is 2000, then one can say, Last December the central
bank...., but this obviously would not do if we were already in the year 2001.
In that case, one could write: In December 2000 the central bank... .
longitude. So spelled. NOT longtitude
majority of.... This phrase cannot stand alone but must always be preceded by the
word a or the. WRONG: Majority of the bids were below the reservation price.
CORRECT: The majority (or a majority) of the bids were below the
reservation price. Strictly speaking majority refers to a number that is more
than half of the whole. It is not a synonym for “most”, since the latter may
also refer simply to a “plurality”, which is the number of those possessing the
most frequently occurring characteristic in the population. Whether a
majority of X or the majority of X should take the singular or plural depends
on whether X itself is a singular or plural noun. Hence, a majority of
Congress (singular) was against granting special powers, but a majority of
the candidates (plural) were eminently unqualified.
market (on/in the market). For something to be on the market refers to the fact
that it is for sale, as in a glut of condos on the market. On the other hand, in

62
 the market is used of agents (people) who wish to buy or sell something:
Portfolio managers are in the market for non-bluechip stocks.
“masteral”. Many will be surprised to learn that this word is a solecism that exists
nowhere in standard English. Use master’s or MA/MS instead. Hence: She
received her master’s (or MA, or MS , NOT masteral) degree from the UP.
maximize the potential of... This is confusing at best and technically wrong at
worst. What writers typically mean is to fully use the potential of X, or to
fully exploit the potential of X. If one insists on using maximize, then one
could say maximize the use of X. (In a strict mathematical sense, maximizing
the potential of something means not using it at all.)
minuscule. So spelled. NOT: miniscule.
most. In the phrase most of X, a singular or plural must be used depending on
whether X itself is singular or plural. Hence, Most of the inflation is
(singular) cost-push in nature, while Most of the proofs were (plural) trivial.
(See also majority.)
much versus many. Use many when the entity being described is countable (i.e.,
denumerable); otherwise use much. EXAMPLES: Many of the economists
(countable) at the conference were doing applied work. Much of their work
(noncountable), however, was simply based on casual empiricism.
Obviously, much will appear in the case of noncount nouns (see below).
mutatis mutandis. (changing what needs to be changed, Latin, pronounced myutatis
myutandis). This is the less famous opposite of ceteris paribus. Hence, one
might distinguish between a ceteris paribus demand function and one that is
mutatis mutandis, depending on whether the price change is assumed to
affect wealth. Italicize.
N.B. (= nota bene, note well, Latin) Capitalize, separate with full stops; do not
italicize. Separate with commas from the rest of the sentence.
noncount nouns. Some nouns have no plural forms. The attachment of s to them
is a commonly observed mistake. The following gives examples and suggests
approximations of a plural form.
advice pieces or bits of a.
clothing pieces or articles of c.
education levels or stages of e.
equipment pieces of e.
evidence pieces or bits of e.
furniture pieces of f.
homework a lot of h.
information pieces or bits of i.
knowledge pieces or bits of k.
learning bits of l.
luggage pieces of l.
merchandise pieces of m.
news pieces or bits of n.

nontradable. No hyphens needed. Spelled thus, NOT nontradeable.

nor versus or. Confusion sometimes arises when writers juxtapose two alternatives,
both of which are negative. CORRECT: The economy does not need an interest-
rate cure, nor does it need a strong currency. ALTERNATIVELY: The economy
does not need an interest-rate cure or a strong currency. The economy needs
neither an interest-rate cure nor a strong currency.

63
numbers and numerals. As a general rule, spell out the numbers one, two, …, ten
in any part of a sentence when they are understood as counting numbers.
Counting numbers beyond ten need not be spelled out. When beginning a
sentence, however, it is obviously unhelpful to start with a numeral.
EXAMPLE: 21 out of 72 provinces earned lower revenues after devolution.
IMPROVED: Twenty-one out of 72 provinces earned lower revenues after
devolution. It must be stressed that this rule does not apply when the
numerals involved are estimates or values taken by variables in
mathematical formulas. Hence: The derivative vanishes when x equals 3. An
elasticity greater than 1 is not uncommon for this market.
numéraire. (“cash”, French). Italicize on first use, with the accent over the first e. In
the sense of Walras, it meant a unit of account.
on the one hand…on the other hand. All jokes notwithstanding, economists
still have only two hands. This formula can therefore be used only if there are
exactly two competing viewpoints, for after using up one hand and then the
other, there cannot be any more. The two parts of this formulation may be
used in the beginning or in the middle of sentences, but in either case, they
must be separated from the rest of the sentence by commas. EXAMPLE: On the
one hand, the depreciation is thought to be good for exports. It may, on the
other hand, lead to unstable inflationary expectations. The alternative
formula that omits the (i.e., on one hand... ) is no longer current. When more
than two aspects need to be considered, one must change tack and consider
writing, first, second, third, etc. (See first(ly), second(ly), third(ly), etc.)
one. Using this generic pronoun is one way to avoid having to directly address the
reader (you), use the first person (I or we), or a passive form. It is
particularly useful in describing procedures. EXAMPLES: You [One] can
perform the same exercise to test for openness. It is best if we start [one
starts] from the facts. Note, however, that once used, one should not be
mixed with the forms he, her, it, his, hers, and its in the same sentence.
WRONG: One cannot have his cake and eat it too. If one cannot stand the
heat, he should get out of the kitchen. CORRECTED: One cannot have one’s
cake and eat it too. If one cannot stand the heat, one should get out of the
kitchen. (See also indefinite pronouns.)
optimize and maximize. It bears repeating that the two are not always the same.
To optimize means to find the best use of something. But the greatest use
(maximization) is not always the best use. Optimization means either
maximizing or minimizing something. In most applications, optimization
really means maximizing something subject to constraints. Hence it is seldom
a question of pure (i.e., unconstrained) maximization, notwithstanding the
impression that bigger is always better.
overvalued/undervalued exchange rate. The exchange rate is a price, namely
the price of one currency in terms of another. It is not the price itself that is
overvalued but the object bearing the price, i.e., the object's price is too high.
Since overvalued is equivalent to overpriced, it is redundant to say that the
price (namely, the exchange rate) is overpriced. Hence, use overvalued or
expensive currency, or alternatively, say that the exchange rate is too low (or
too high, depending how it is quoted). As for how the exchange rate is
quoted, be sure that “high” or “low” correspond to what you mean. In the
Philippines, exchange rates are customarily quoted as pesos per dollar, or
pesos per yen, or pesos per DM. In this convention, therefore, a lower
exchange rate means less pesos being required to buy each unit of foreign
currency, hence a stronger currency, or an appreciation. Likewise, a higher

64
 exchange rate then means a weaker currency. Ironically if one says “the peso
plunges”, therefore, one really refers to a higher exchange rate.
pari passu (“equally and simultaneously”, Latin). Italicize.
per annum (Latin) Literally, “every year”. Do not italicize. Variants are annually,
yearly, and per year.
per capita (“for each head”, Latin). Do not italicize. Obvious variations are per
head and per person, and per man, although the last is not gender-kosher
(See sexist language.) Income per capita rose by 5 percent in 1995.
per se. (“by itself”, Latin). No italics.
Philippines. In referring to the country or political entity, use the word as you
would a singular noun. Hence: The Philippines opens from its experiment
with import substitution. In referring to the Philippines as a geographical
form consisting of many islands, use Philippine islands instead. The
Philippine islands differ considerably among themselves.
plural forms. Academic and professional writing are littered with nouns of Latin
and Greek origin, the formation of whose plurals may cause some
bewilderment. In some cases, the words have become so adapted to English
that their plurals are formed by simply adding –s or –es Using the plurals in
the original language would then appear didactic. Since there is no hard-and-
fast rule, the most that can be provided is a list.

agendum agenda (see also agenda)

alumna (female) Alumnae
alumnus (male) alumni
analysis analyses
apparatus apparatuses
axis axes
basis bases
calculus calculi
colloquium colloquia
crisis crises
criterion criteria
curriculum curricula
datum data
ephemera ephemeras
formula formulas, NOT formulae
forum forums NOT fora
genus genera
hyperbola hyperbolas, NOT hyperbolae
infimum infima
lacuna lacunas OR lacunae
lemma lemmas, OR (less frequently) lemmata
maximum maxima
memorandum memorandums, collectively, memoranda
minimum minima
nucleus nuclei
parabola parabolas, NOT parabolae
optimum optima
phenomenon phenomena
polyhedron polyhedra OR polyhedrons
radius radii

65
 simplex simplexes
spectrum spectra
stimulus stimuli
stratum strata
supremum suprema
syllabus syllabuses, NOT syllabi
syllabuses syllabuses
symposium symposia
tableau tableaux
-thesis (e.g., synthesis, -theses
hypothesis, etc.)

prevent. The accepted construction is prevent from, followed by a present

progressive from, not prevent to. WRONG: The government was prevented to
sell the asset by a court order. RIGHT: The government was prevented from
selling the asset by a court order. (See also prohibit.)
progressive, regressive. In reference to a tax system, the corresponding qualities
are progressiveness and regressiveness, not “progression”/
”regression”, or “progressivity”/”regressivity”.
prohibit. The proper preposition is from, not to, as in the following: A subsidy tied
to investment prohibits Firm A from changing its commitment. The
mistaken construction prohibited to is probably a result of a confusion
between prohibit and forbid. One is forbidden to do something, but
prohibited from doing something. (See also prevent.)
q.e.d./Q.E.D./QED (quod erat demonstrandum, “which was to be shown or
demonstrated”, Latin) Contrary to facetious graduate-student tradition, this
phrase does not mean “quite easily done”. The phrase is still used to mark the
end of mathematical or logical proofs, although it is increasingly regarded as
quaint and is rapidly losing ground. The phrase may be written in lower case,
separated by full stops, or capitalized, with or without stops. Increasingly
nowadays, however, the end of a proof is marked by simply placing the block
“
” after a space behind the last full stop of the paragraph.
reason that versus reason why. A dispute used to rage over whether “reason
why” was acceptable in the following: Slow output growth is the reason why
tax collections are low. Some would insist this should be rewritten as Slow
output growth is the reason that tax collections are low. In fact, however,
both are acceptable. What does constitute a MISTAKE is the following: The
reason why tax collections are low is because output grew slowly. In most
cases you will find you do not even need the phrase. BETTER: Tax collections
are low because output growth is slow. OR: Slow output growth has caused
low tax collections.
recessionary/recessive. In reference to a downturn in economic activity, the
appropriative adjective is recessionary. EXAMPLE: The economy displayed
recessionary tendencies in the second quarter. The term recessive, on the
other hand, occurs in biology and genetics. (But see distortionary and
routine.)
regress. The verb to regress in a statistical or econometric context is a back-
formation from the noun regression, as in regression analysis or linear
regression. In current usage, the verb takes the preposition on or against
EXAMPLE: In another experiment, we regressed earnings against (OR on)
years of schooling, age, and other demographic variables.
regressive. See progressive.

66
remains to be versus remains (as). The distinction between the two is important,
as the following shows. Some economists contend that the current account
deficit remains (as) the country's principal economic problem. But whether
the balance on current account will actually worsen remains to be seen. A
frequent error is the use of remains to be as the equivalent to remains (as) or
is. The following mistake is typical: (WRONG) Poverty remains to be one of the
country's pressing problems. What this writer fails to realize is that he has
just managed to say exactly the opposite of what he means. CORRECTED:
Poverty remains one the country's pressing problems, or Poverty continues to
be one the country's pressing problems.
request (for). One frequently sees the following mistake: (WRONG) The peasants
requested for a meeting with the President. CORRECT: The peasants
requested a meeting with the President. ALTERNATIVELY: The peasants made
a request for a meeting with the President. In short, when request is used as
a verb, it takes a direct object, and the word for is unnecessary. Used as a
noun, however, it requires the preposition. See also demand (for), seek,
search (for)/search (of).
respectively. (abbreviation, resp.) A useful device that uses something akin to the
distributive property. The return to capital (respectively, labor) is expected to
rise (respectively, fall) with an increase in the tariff on importables. The
income flows corresponding to land, labor, and capital are rent, wages, and
profits, respectively. Separate from the rest of the sentence by commas.
result in/resulting in. Notwithstanding all you see and hear, the proper phrase is
NOT result to or resulting to, but rather result in and resulting in. WRONG:
Experience shows that currency depreciation results to inflation. CORRECT:
Experience shows that currency depreciation results in inflation.
rise/fall. Forestalling boredom and avoiding repetitiveness in economics and
business writing will often test the limits of one’s knowledge of synonyms.
The following gives a few synonyms for rise and fall.
increase decrease
expand contract or recede
gain lose
go up go down
[be] on the uptrend [be] on the downtrend
flow ebb OR recede
strengthen weaken
grow positively grow negatively
improve worsen OR deteriorate
Some synonyms for showing no change are: to remain (OR be) steady,
constant, unchanged, flat, or to move sideways. (See also,
accelerate/decelerate.)
risk versus uncertainty. Arrow writes, “The meaning of uncertainty is that the
agent does not know the state of the world.” Frank Knight would have argued
that being able to assign probabilities to alternative states of the world
implied a state of knowledge akin more to risk than to uncertainty.
routine/routinary. The latter can be totally expunged. In all cases routine serves
equally well both as noun and adjective. EXAMPLE: The routine function of
teaching is the minimum expected of an instructor (adjective). The computer
refused to perform the routine(noun).
saving versus savings. The early effort to convince the profession of the need to
distinguish between these two is unfortunately faltering. Still, it is good to

67
 remember that saving is the act, a flow-concept, while savings are the result
of the act and are a stock. Hence, notwithstanding what other authors say, a
function such as S = s(Y) is a saving function and S/Y is the saving rate or
saving ratio. On the other hand, in the sense of raising the amounts saved,
one must speak of savings mobilization.
seek versus search. The verb seek takes a direct object, without any need for a
preposition. WRONG: The underemployed refers to those who are seeking for
additional work. CORRECT: The underemployed refers to those who are
seeking additional work. The confusion arises because to search for, which
requires the preposition for, is a synonym for to seek, which however requires
no preposition. Hence: The economic managers continued to seek [OR search
for, OR look for] the right mix of monetary and fiscal policies.
settle at/on/in. An economic variable or statistic settles at a certain level when it
attains a certain value after a period of some fluctuation, e.g., After a period
of instability, the exchange rate settled at 49 pesos to the dollar. To settle on
something, on the other hand, means to reach agreement on it, e.g.,
Management and the unions settled on a 230-peso wage. To settle in, finally,
refers to the physical fact of acclimatizing oneself to some environment:
Locators at the special zone have finally settled in.
sexist language. Some language reinforces unwarranted sexual stereotypes and it
is good to be conscious of this and how it might give offence. (1) A good part
of the problem has to do with occupational stereotypes, and it is a simple
matter to detect these words and replace them with gender-neutral terms.
EXAMPLE: Businessmen are worried that the impeachment will be a long
drawn out process. Instead you may use the business community, or
business people. Below are some terms likely to arise in economics and their
possible replacements:
businessman business (in general);
business community; business people
chairman chair or chairperson
craftsman artisan; crafts practitioner
fisherman fisher (sg.) or fisherfolk (pl.)
foreman supervisor
housemaid domestic help
housewife housekeeper; homemaker
landlord landowner; property owner
layman layperson; the uninitiated
man human being; humans (pl.) people
(pl.);
Mankind humankind; humanity
salesman sales representative; salesperson
seaman seafarer; mariner
tradesman trader (sg.) OR tradespeople (pl.)
workingman; workman worker

Even this is imperfect. (1) Some words with precise historical meanings, like
master (as in the master-slave relation), guildsmaster, journeyman,
patrician, or for that matter even landlord in the sense of an agrarian
landlord-tenant relation, have no gender-neutral counterparts, and it would
cause infinitely more trouble and confusion to replace them. (2) Once one is
on this gender-neutral course, problems next arise with noun-pronoun
agreement. Since the terms in the right column are gender-neutral, what
pronoun is appropriate? He/him/his? or she/her/hers? EXAMPLE: Each

68
 worker self-selects and chooses a job most appropriate to his abilities. There
are four options: his, her, his or her, and their. None of them is entirely
satisfactory: the first because it restores the gender bias; the second because
it calls attention to itself; the third because it is stilted; and the fourth
because it is uncommon in this part of the world and openly flouts the verb-
subject rule. The recommended compromise in this inherently unsatisfactory
state of affairs is to seek refuge in the plural: Workers self-select and choose
jobs most appropriate to their abilities. But then again, this is also
unsatisfactory, since it dulls the point about individual decisions being made.
(Oh, well.) One confronts a similar problem in dealing with grand statements
like Man is a social animal. Here one might perhaps say A human being is a
social animal or Humans are social animals. As another example, what were
formerly known as the rights of man are now called into human rights. (The
French, however, have predictably refused to change droits de l’homme into
droits humaines.) Again, whether or not you choose to ride on the
bandwagon, the important thing is consistency of style.
sic (“thus”, Latin). Italicize. This is used to indicate that the original spelling or usage
has been retained. But good form demands that attention should not even be
called unless some scholarly purpose is served by doing so. Let magnanimity
rule. For example, original oversights in verb-subject agreement, verb tenses,
typographical errors, and so on, are typically inessential to meaning, so that
the writer might as well correct these errors himself, rather than draw
inordinate attention to them. It is a different matter, however, when the issue
itself revolves around various interpretations caused by such original
mistakes or ambiguities.
similar(ity). The correct preposition is to not with: German exports were similar to
British exports in the late 1800s. Because of the similarity of German exports
to British exports, the Germans needed to undersell the British.
since. Used as a conjunction, this word denotes causality. Poverty incidence in the
population is higher than the proportion of poor families, since poorer
families tend to be larger. A comma is indispensable in capturing the sense of
causality and distinguishing it from since as implying mere temporal priority.
Compare the following examples: (a) Effective protection rose since input
tariffs fell. (Priority in time). (b) Effective protection rose, since input tariffs
fell. (Causality).While (a) merely states that effective protection rose after
input tariffs fell, (b) attributes this directly to the fact.
solve/solve for. One solves a problem or an equation but one solves for the
unknown, for the equilibrium value, for the time path, and so on. Hence, one
solves f(x) – c = 0 for the variable x.
species/specie. The former is the biological term and is used both for singular and
plural forms. The latter refers to metal coinage, as in payment in specie and
has no plural.
substitute for/with/by. As a verb, substitute is not transitive and has not been
regarded as such since the 17th century [Burchfield 1996]. Hence it is
INCORRECT, or at least nonstandard, to say Rice substituted maize during the
crisis. The commonly observed passive version is also WRONG: Rice was
substituted by maize during the crisis, as is the variant Gasoline was
substituted with diesel fuel. The source of all these mistakes is the wrong
impression that the verb substitute is identical in all respects with the
transitive verb replace. (Note that putting replace in lieu of substitute in all
the previous examples would make them correct, and that is probably what
should have been done.) The only indisputably correct use of the verb is the

69
 combination substitute for, hence: Rice substituted for corn during the
crisis, or People substituted rice for corn during the crisis.
suffixes following names. One often encounters such phrases as Keynesian
cross, Marshallian demand curve, Walrasian equilibrium, Marxian theory
of value and so on. These are formed by appending -ian to the names of
persons. In most names ending in consonants, the suffix -ian is simply
appended. This is also true where names end in i, hence Jacobi → Jacobian,
Hess → Hessian. Names ending in y should be treated as if they ended in i.
Special problems are posed by names ending in -a, -o, -ow, e, and eu. In the
case of -o, the ending is usually dropped. Hence, Ricardo → Ricardian rent.
Where one has -ow or -ou, the w or u is turned into a v. Hence Pigou →
Pigovian tax, and by extension (though hardly observed), Arrow → Arrovian
paradox and Debreu → Debrevian exposition. Where the name ends in a or
e, the ending may be dropped and replaced with i: Lange → Langian
socialism. These coinages are not always followed, however, and at times
suffixes are simply omitted: Lucas supply function (rather than Lucasian,
though there is a Lucasian Professor at Cambridge); Leontief (rather than
Leontiefian) production function. (See Burchfield [1996].)
supply and demand. These two terms are Siamese twins but require different
prepositions, an often forgotten fact that leads to such common mistakes as
“the demand and supply of good x”. Remember that one refers to the supply
of a good and the demand for it. Hence in a combination, one must write the
supply of and demand for a good.
target. (1) A target, as in a target exchange rate, is some value of a variable that is
desired or hoped for. The metaphor, derived from shooting or archery,
requires one to come as close as possible to the desired value of the variable,
whether the difference is positive or negative. A large excess over the target
should be regarded as inferior to a small deficiency, since it is the absolute
distance that matters. For this reason, it is not, strictly speaking, appropriate
to rejoice when actual growth has exceeded the target, since overshooting a
target is just as much a failure as falling short of it. Instead one might say
growth exceeded expecations or was better than expected. Similarly, to write
that inflation has inched up but is still below target, somehow implies that
the writer would have wished inflation to be higher, which was presumably
not what was meant. In the latter, it is perhaps better to say that inflation
was within range of the target. (See also floors and ceilings.) Be careful of
formulations that stretch the metaphor to the point of being ridiculous, such
as expanding the t., doubling the t., and so on. (In other ways this metaphor
has been “spoilt” see Burchfield [1996: 614].) (2) A target can also mean the
intended recipient (typically a beneficiary) of a particular policy, e.g., the
poor are the target of this policy. This is somewhat odd usage, since typically
the purpose of targeting something is to kill or hurt it. If this jars
sensibilities, one might try writing of focused instead of targeted programs.
tâtonnement. (“groping about”, “trial and error”, French) As used by Walras, this
word has a technical meaning: a system in which trading occurs only at prices
that clear all markets. Italicize, and use a circumflex accent.
that versus which. Ubiquitous grammar-checks in word-processing software have
mechanically taken up Fowler’s plea and now invariably make a big fuss over
using that for restrictive clauses and which for nonrestrictive clauses.
EXAMPLE 1: (Restrictive) The likeliest scenario is a large depreciation that
will push up inflation. EXAMPLE 2 (Nonrestrictive) The likeliest scenario is a
large depreciation, which will push up inflation. The difference is that

70
 Example 1 underscores the distinction between depreciations that push up
inflation and those that do not. Example 2, on the other hand, emphasizes the
occurrence of depreciation per se, whether it causes inflation or not. That
always occurs in restrictive clauses, but which may occur in both restrictive
and nonrestrictive uses. In Example 1, that may be quietly replaced with
which without doing too much violence to the sense. The most important tool
distinguishing the two is the comma. It is generally wise to follow Fowler’s
advice to use that for restrictive and which (preceded by a comma) for
nonrestrictive clauses. Burchfield [1996:724] notes some complications,
however. An important one is that using that will sometimes force you to end
your clause with a preposition. The researcher fumbled to reproduce the
estimates that his conclusions rested upon. This is certainly less elegant than
the equivalent which-construction: The researcher fumbled to reproduce the
estimates upon which his conclusions rested. Both, by the way, are restrictive
clauses. The lesson, then, is that while a “rule” may exist, one cannot be
dogmatic about the matter.
the. Indispensable in the Philippines, the United States, the Seychelles, the United
Nations, the Fed, the UK, the EU, the Bangko Sentral, the Bundesbank, etc. A
frequent mistake on the labels of products is the caption Made in Philippines.
(See also majority.)
therefore and thus. Despite an apparent similarity, there is a subtle difference in
the proper use of these two terms. Therefore means “as a result”, while thus
implies “by these means”, or “in this manner”. AWKWARD: The nominal
exchange rate rose. Thus the peso-equivalent of firms’ foreign liabilities
expanded. Here, thus should have been replaced with therefore, since the
second event was a consequence of the first. CORRECT: Higher-yielding
varieties required outlays that only richer farmers could afford. The
diffusion of new technologies thus reinforced existing inequalities. When
therefore is embedded in a long sentence, separate it from the rest of the
passage with commas.
to. Often misused owing to the inappropriate literal translation of the Tagalog sa:
interested to (interesado sa); implications to (may implikasyon sa);
resulting to (nagreresulta sa); prohibited to (pagbabawal sa). The correct
prepositions are rather: interested in, implications for; resulting in;
prohibited from, and so on.
tradable. Thus spelled; not tradeable.Hence also nontradable.
translate as/translates into. These are useful phrases to indicate the recasting of
an argument, concept, or exposition into a form that is simpler, more
transparent, or more cogent. A slight nuance may be made between
translates as and translates into. A translates as B implies that the
statements A and B are more or less equivalent. EXAMPLE: The value of
marginal product curve translates as the demand curve for labor. On the
other hand, A translates into B connotes that a greater simplification has
occurred. It is more or less the same as “reduces to”. EXAMPLE: The 23-point
increase in the price index translates into an 8 percent rate of inflation. The
widespread use of translates to (rather than as or into) is idiomatically
wrong.
verbal ballast. A strong temptation among technical writers is to load up on verbal
ballast, usually in an unconscious desire to sound impressive. As a result, a
good deal of verbiage creeps in that is really unnecessary to the argument.
Examples are the following: It is important to realize that... , This is far from
saying that... , As far as ... is concerned,.. , It can be said that... , It cannot be

71
 denied that... , and This is not to say that... . Such phrases can often be
dispensed with without a loss in meaning.
vis-à-vis (face-to-face, French). This now means “in relation to”. Italicize with a
grave accent over “a”.
viz. (videlicet, it is permitted to see, Latin; where z is the medieval Latin symbol for
the abbreviation of -cet.) This is a synonym for “namely” or “that is”. Separate
from the \sentence with commas. Use no italics.
who versus whom. This matter has unfortunately proven so confusing to such a
large number of people that the rule is now threatening to collapse. The rule
has always been the following: use whom whenever the answer to the
question is “him” or “her”, and use who whenever the question may be
answered by “he” or “she”. EXAMPLE: Who do you think is right? (Answer: He
or she is right.) Whom did Keynes regard as the main proponent of the
classical view? (Answer: Keynes regarded him [Pigou] as the main
proponent, etc.) The rule applies even in relative clauses: Marshall conceded
no debt to Jevons, while generously praising Cournot, whom he read avidly.
(Marshall read him [Cournot].) Hicks’s work preceded the work of Arrow,
with whom he later shared the Nobel Prize (Answer: Hicks shared the prize
with him [Arrow].)
with regard to... Together with other well-worn formulas such as, as far (or
insofar) as X is concerned, with respect to X, and so on, this phrase is rapidly
becoming trite. It suffers the additional misfortune of being often mistakenly
written (or spoken) in the form with regards to X. Acceptable alternatives are
possible, of course, including: as for X, regarding X, on the subject of X, and
as far as X is concerned. In most cases, however, it is both possible and
desirable to dispense altogether with these and kindred phrases. EXAMPLE:
With regard to the appropriate level of the exchange rate, economists have
differing opinions. REWRITTEN: Economists differ over the appropriate level
of the exchange rate.
END

72