Origins of Terms in International Economics

Here I record what I have been able to learn about the origins of some of the terms we use in international economics, both who introduced their meanings and who first gave them their names, if those are not the same people.

If I attribute a concept or the term for it to a particular author, that means I have personally checked the source and seen it used there in the way that I describe. However, if I say or imply that this was the first use of a concept or term, I obviously cannot always know that for certain. For some of these, I have searched in Google Scholar to find the first use of the term identified there. But I don't know how complete that tool is (it seems remarkably good), and in any case I can't find who may have used a term orally with their colleagues or students without publishing it earlier. If you know of prior uses that should be mentioned, please let me know, preferably by e-mail to

Atlas method
The Atlas method gets its name from its use in the World Bank Atlas, more recent editions of which are called the Atlas of Global Development.

The current method seems to have been used first in the 1985 edition of the Atlas, where it says "...the procedures for estimating gnp in U.S. dollars differ from those used in previous years." Previous editions had used average prices and exchange rates from a three-year base period, whereas the new procedure was to use "the simple average of the exchange rates for the current year and for the two preceding years; the latter two exchange rates are adjusted for differences between domestic and U.S. inflation." The most important difference between the new and old methods was that the old one used a difference conversion for a given year in successive editions of the Atlas, making comparisons difficult. I am unable, from reading details of the prior procedure, to tell whether the new one also causes other differences. In any case, it seems that the subsequent use by others of the "Atlas method" has been to use the method introduced in World Bank (1985).

In an effort to determine whether the Atlas method had a prior history, I have used Google Scholar to search for "Atlas method" over various years. The term itself was used with other meanings prior to its use by the World Bank, as well as subsequently: a method of measuring the maturity of skeletons; a method in meteorology for detecting weather events; a tool for measuring the "HLB value of a reagent," whatever that is, by Atlas Chemical Industries; and an exercise program by body builder Charles Atlas.

The earliest mentions of the World Bank's Atlas method were in 1980, where I found two. One of these (by M.Y. Smith, "Romania: forecasting and development," in Futures) mentions "following the World Bank Atlas method of adjusting official Romanian national accounts data." The second (by Davies, Grawe, and Kavalsky, "Poverty and the Development of Human Resources: Regional Perspectives," World Bank) has a table of GNP per capita data labeled "IBRD Atlas method, US dollars."

From these it appears that the term "Atlas method" predated the version of World Bank (1985) and referred to the method used in prior editions, which seems to have been slightly different.

Balance of trade
Price (1905) examines the origins of this concept, the exact wording of which appeared in 1615 and the concept of which, without the wording, can be found as early as 1381 in England, when writers were concerned that by importing a greater value than it was exporting, England was losing money -- i.e., gold and silver. Somewhat before the term "balance of trade" appeared, similar concerns were said in 1601 to be due to "overbalancing of foreign commodities." From the discussion by Price, it appears that "balance of trade" in this early use referred to a situation in which values of exports and imports were equal, rather than today's use measuring the extent to which they are unequal.

Fetter (1935) dates the term to 1623, apparently disagreeing with Price that its use in 1615 was comparable. His main concerns are with the common attribution that a positive balance of trade is "favorable" and with whether the term includes only trade in goods or instead extends beyond that to include other payments such as we today would include in the balance on current account or even balance of payments. It appears that early writings used the term variously in each of these senses.

[I was alerted to the articles by Price and Fetter by Obstfeld (2012).]

Canonical model of currency crises
Krugman (1997) gives credit for this model, which he seems in this source to be the first to call the "canonical model," to apparently unpublished "work done in the mid-1970s by Stephen Salant, at that time at the Federal Reserve's International Finance Section," focusing on schemes to stabilize commodity prices, and described briefly in Salant and Henderson (1978) where it is applied to the market for gold. Krugman then drew on that work for his model of currency crises in Krugman (1979b), and it was refined by Flood and Garber (1984).

CES function
Arrow et al. (1961, pp. 225-226) described their empirical motivation to "derive a mathematical function having the properties of (i) homogeneity, (ii) constant elasticity between capital and labor, and (iii) the possibility of different elasticities for different industries." They named it the CES function and estimated it across industries and countries.

One of the co-authors, Minhas1962, tried to rename it the "homohypallagic" function, deriving from Greek homo=same and hypallage=substitution. He credited the idea for this name to Emmanuel G. Mesthene of the Rand Corporation. The name did not catch on.

Mukerji (1963) also tried to rename it the "SMAC function," using the initials of the four authors of Arrow et al. (1961) -- Solow, Minhas, Arrow, and Chenery -- but that too failed to catch on.

The CES function did not play a major role in international trade theory during the first decade or two after its introduction, perhaps because trade theorists had a proud tradition of deriving results without specifying functional forms. It came into its own, however, in the Dixit-Stiglitz function used by Krugman (1980) as central to incorporating monopolistic competition into the New Trade Theory. The innovation here was to make the number of products (varieties) variable.

Comparative advantage
Ruffin (2002) credits the concept of comparative advantage and the law of comparative advantage to Ricardo (1951-1973), in a discovery that Ruffin dates to early October 1816. The law was developed in Ricardo's celebrated chapter on foreign trade, while the term "comparative advantage" seems to have first appeared in a later chapter (Ricardo (1951-1973), Vol I, p. 263). In crediting Ricardo, Ruffin disagrees with Chipman (1965) who credits Torrens (1815). From what I see in this debate, Torrens deserves credit for first stating the possibility that a country will import a good in which it has an absolute advantage, even though he seems not to have recognized its importance, and he certainly did not work out the full conditions needed for this to happen, as Ricardo did.

Continuum of goods
The first to model trade with a continuum of goods were Dornbusch, Fischer, and Samuelson (1977), who also use that term in their title. They cite an unpublished paper by Charles Wilson, also dated 1977, that further explores their model, but in the published version of that paper, Wilson (1980) credits them with having suggested this modification of traditional trade theory. This approach was admired but not used extensively by others until Eaton and Kortum (2002) replaced deterministic continuous functions for productivity with stochastic distributions in what is now the widely used EK Model.

Currency area
Mundell (1961, p. 657) spoke of "...defining a currency area as a domain within which exchange rates are fixed...". Perhaps because the exchange rates among separate national currencies are seldom if ever truly fixed, the term has come to mean a group of countries that share a common currency. Mundell also coined the term "optimum currency area" which is now more commonly expressed as optimal currency area.

Depression, The Great Depression
Mendel (2009, an intern at the History News Network, reports that the term "depression" was being used for an economic downturn as early as US President James Monroe in 1819, referring to what has been called the Panic of 1819 and its bank failures and currency depreciation. Monroe also used the phrase great depression in his 1820 Fourth Annual Message. Other presidents after Monroe -- including Grant, Hayes, and Coolidge -- used the terms as well, prior to Herbert Hoover, who is often credited with introducing the term in preference to the more common "panic."

Although Hoover referred to the world has experiencing "a great depression," he did not give it the name "The Great Depression." That seems to have been coined by Robbins (1934).

Diversification cone
Dixit and Norman (1980, p. 52) attribute this to Lerner (1952) and McKenzie (1955). I see nothing in Lerner to justify this. McKenzie, however, makes considerable use of the concept in the form of a set of factor endowments within which factor price equalization occurs, though he does not give it a name. Since he projects factor requirements and factor endowments onto a simplex, his set appears as a triangle, though a cone is implicit. I do not yet know who may have preceded Dixit and Norman in using this term.

Dixit-Stiglitz utility
This refers unambiguously to the model of monopolistic competition introduced by Dixit and Stiglitz (1977) and to the similar model of Spence (1976) that leads this sometimes to be called Spence-Dixit-Stiglitz utility. What is less clear is exactly what is meant by either term. The term "Dixit-Stiglitz utility" did not appear in publication until 1987, when it then appeared more than once but with slightly varying meanings. The varying meanings have continued since then, but the commonality among them is
    U={Σinxiρ}1/ρ       0<ρ<1
with the interpretation that n, the number of products or varieties, is variable and that σ=1/(1-ρ)>1 is the elasticity of substitution among them. It is this interpretation of variable n that allows this function to display very simply a preference for variety. If x=xi, i=1,...,n, then U reduces to
so that an equal increase in n and fall in x (keeping Σx constant) increases U to the extent that ρ<1.

Dixit and Stiglitz (p. 298) start with a somewhat general form for utility, u=U(x0,V(x1,x2,x3,...)), where the range of products is left unspecified but greater than the number of products actually consumed, n. They then consider several special cases, the one usually adopted by others being a CES form for the function V: u=U(x0,{Σinxiρ}1/ρ). What is crucial and distinctive about either form is that the number of products (often called varieties) consumed, n, is variable.

Neary (2004) notes that later users of Dixit-Stiglitz utility combined three assumptions that Dixit and Stiglitz themselves mentioned but never used in combination: symmetry of V in xi, CES form for V, and Cobb-Douglas form for U. He therefore suggests that the following should be called "Dixit-Stiglitz lite":

    u = x0(1-μ)Vμ,       V={Σinxiρ}1/ρ

In fact later users have often omitted the numeraire good, x0, or replaced it with other goods. And the symmetric CES function with variable n has, by itself, come to be what is most commonly regarded as the Dixit-Stiglitz utility function, or sometimes the Dixit-Stiglitz subutility function. In that case the Dixit-Stiglitz utility function appears identical to the CES function, the only difference being the interpretation of n as variable.

The function has also been used frequently for production, with the xi as intermediate inputs, following Ethier (1982). The number of varieties is variable and contributes to the value of the function, which in this case is output. However, Ethier made the role of n in productivity explicit with a second parameter, and his production function was:

    M = nαin(xiρ/n)}1/ρ,       0<ρ<1, α>1
If α were equal to one, this function would not increase with a rise in n and an equal fall in a common x=xi, so α>1 is what generates Ethier's international returns to scale when the xi are traded. Ethier's function reduces to the Dixit-Stiglitz form when α=1/ρ>1, whose positive benefit from variety (greater n) is then directly related to the elasticity of substitution, σ=1/(1-ρ).

DUP activity
Bhagwati (1982) introduced this acronym for directly unproductive profit-seeking activity. After listing a variety of activities that fit this description, including rent seeking, revenue seeking, and others, he said (p. 990), "Thus, these are aptly christened DUP activities."

Dutch disease
Term was coined by The Economist in an article "The Dutch Disease" in the issue of November 26, 1977, pp. 82-83, which included the passage "... in the words of Lord Kahn [1905-1989], 'when the flow of North Sea oil and gas begins to diminish, about the turn of the [21st] century, our island will become desolate.' Any disease which threatens that kind of apocalypse deserves close attention." The article attributes the problems of the Dutch economy (an external appearance of strength but internally high unemployment and a declining manufacturing sector) to "three causes, only one of them external." These are (1) a strong currency; (2) high industrial costs; and (3) use of government gas revenues to increase spending rather than investment. As used since, the term has been focused primarily on the real exchange rate. The term was used by Corden and Neary (1982), whose reference to it as "... sometimes referred to as the 'Dutch Disease'" suggested that it had passed into common usage.

Edgeworth-Bowley box
The origins of this were examined by Tarascio (1972). The diagram was first drawn by Pareto (1906), based originally, though only very partially, on a diagram of Edgeworth (1881). Edgeworth's diagram was not a box at all, and was drawn on axes more approptiate to an offer curve than to exchange of fixed quantities of goods or factors. Edgeworth's purpose was to define and depict the contract curve, which today we almost always draw within the box diagram.

Bowley's name was added to the name of the diagram as a result of Bowley (1924), who drew indifference curves for two individuals, one rotated clockwise 90 degrees and the other counterclockwise, thus forming the outline of a box, within which he showed Edgeworth's contract curve. That is probably why his name came to be associated with Edgeworth's. However, Bowley did not claim originality, and while he cited Edgeworth for the contract curve, he neither named nor attributed the box diagram. Indeed, in his own diagram showing exchange, his focus was on the internal portion and he did not extend the axes of his two indifference maps far enough to touch or cross, and therefore did not actually produce a box. And had he done so, his box would have been a mirror image of the one we normally draw today.

It was Pareto (1906), writing in Italian that was soon translated into French, who had actually been the first to draw and use the box diagram. It is unclear whether his contribution was known to Bowley and to others writing in English until later. His diagram, displaying indifference curves for two consumers, one drawn conventionally and one rotated 180 degrees, formed the box very much as we know it today, for exchange between consumers. With each consumer endowed with only one of the two goods, he showed a trade equilibrium as a common tangent to two indifference curves that were also tangent to a price line from the consumers' endowment point.

I have searched in Google Scholar for "Edgeworth box," "Edgeworth-Bowley box," "box diagram," and the joint appearance of "Edgeworth" and "box." The last of these gets many hits, of course, but none of them are about the Edgeworth box, until Stolper and Samuelson (1941). Their Figure 2, p. 67, had labor and capital on the axes and isoquants for two industries inside. Of this they said:

    This is done in Fig. 2 which consists of a modified box diagram long utilised by Edgeworth and Bowley in the study of consumers' behaviour.
Since this classic paper did not find its way into publication immediately (see Deardorff and Stern (1994)), it seems very likely that Saumuelson introduced his version and name of the box to his colleagues and students in the years before this. Whether he himself originated it or picked it up from others as an oral tradition, I do not know.

Based on all of this, it appears that the box applied to consumption, as well as the Edgeworth production box, have both often been called just the Edgeworth Box, even though Edgeworth never drew either. Calling it the Edgeworth-Bowley box is only slightly less erroneous, since Bowley's version of the box was incomplete and perhaps accidental.

Pareto was more deserving of having his name on the consumption version of the box diagram than either Edgeworth or Bowley. Stolper and Samuelson, if they needed further recognition, should share credit for the application to production that has played such a large role in international trade theory. And it seems likely that they, too, were the ones who led us to call it the Edgeworth or Edgeworth-Bowley box ever since.

Flying Geese
The name Flying Geese Model or Paradigm derives from a graph of Akamatsu (1961), (but 1937 in Japanese) that resembles a formation of flying geese. The graph shows paths over time of a developing country's imports, production, and exports of a product, similar to the product cycle.

As used to mean a splitting up of production processes, the term fragmentation was first introduced by Jones and Kierzkowski (1990), who start their analysis by noting (p. 31) that increasing returns and specialization encourage a growing firm to "switch to a production process with fragmented production blocks connected by service links.... Such fragmentation spills over to international markets." (Italics in original.) Many other terms have been used with the same, or related, meanings, as listed here, but "fragmentation" seems to have caught on most widely.

Gravity model
The gravity model of bilateral international trade flows first appeared independently in Tinbergen (1962) and Pöyhönen (1963), but neither used the word "gravity." Tinbergen's formulation was very similar to the formulation of the basic model used today:
    Eij = α0Yiα1Yjα2Dijα3
where Eij is exports from country i to country j, Yi,Yj are their national incomes, Dij is the distance between them, and α0...3 are constants. Pöyhönen's formulation differed from this slightly, including country-specific multipliers cicj and replacing Dij with (1+γDij).

Pöyhönen was member of the same Finnish research team as Pulliainen (1963) whose formulation was more like Tinbergen's. While he also does not call this a gravity model, he does mention the parallel: "The results of our empirical study show that the structure of international trade is capable of description in terms of gravitational theory." (p. 88).

This analogy to gravity was actually resisted by another early user of the model, Linnemann (1966). He followed the above authors in setting out an equation much like the above, with Y's replaced by "potential supply" and "potential demand" which he went on to explain interms of both GDP and population. But in his footnote 43 (pp. 34-35) he remarked, "Some authors emphasize the analogy with the gravitation law in physics, and try to establish that [α3=−2]. We fail to see any justification for this." And in the rest of his book the only mention of "gravity" regards the "centres of gravity" of the countries considered, used for defining distance between them.

The first to call this a gravity model seems to have been Waelbroeck (1965), which includes (p. 499) "Hypothesis 2: The gravity model: distance, export push, and import pull" and has the equations from the other three authors. Waelbroeck notes that "There is, as has been pointed out, an odd similarity between formulae (6) and (7) and the law of gravity, with Yi and Yj playing the role of masses, and this justifies christening the model as the gravity, or G, model." He does not say where it was "pointed out," but it seems likely that he was referring to Pulliainen (1963).

Predating all of this explicit application to bilateral trade between countries, however, the term "gravity model" was used in other social science contexts, and models of this form were used, under other names, in other applications. Bramhall and Isard (1960), in a chapter of a volume on regional science, discuss "gravity, potential, and spacial interaction models -- which for short we shall term gravity models." Similar to other earlier applications that do not seem to have used that name, they formulate the number of trips between areas with different populations using populations instead of GDPs.

Another later source, Glejser and Dramais (1969), cite gravity models as having been used for a long time in literatures on migration, tourism, and telephone calls as well as trade. In a series of papers starting with Zipf (1946), Zipf applied what he called "The P1P2/D Hypothesis" to inter-city movements of freight, persons, information, and perhaps more. Stewart (1947) included in his "Empirical Mathematical Rules Concerning the Distribution and Equilibrium of Population" a formula similar to gravity, but called it "potential." He also cited a much earlier author, Reilly (1929), who provided a "law of retail gravitation," but that was for explaining the market regions covered by cities of different sizes, not the transactions between them.

Great Moderation
This term, as applied to the moderating of economic fluctuations from the 1980s to 2007, seems first to have been used by Stock and Watson (2003). They used the word "moderation," not capitalized and without the adjective "great," throughout the paper, but the title of their section 3, p. 170, was "Dating the Great Moderation."

The term was picked up, and probably made much more visible, by Ben Bernanke in his Remarks at the meetings of the Eastern Economic Association in Washington, DC, February 20, 2004, while he was a member (but not yet Chair) of the Board of Governors of the Fed. See Bernanke (2004). He cites several authors as having documented the decline in volatility, the first being Kim and Nelson (1999) who cite McConnell and Perez-Quiros (2000), in a 1999 Fed working paper, as having documented the decline in a linear formulation rather than a structural shift. McConnell and Perez-Quiros, in turn, start their paper with "The business press is currently sprinkled with references to the 'death' or 'taming' of the business cycle in the United States." Neither of these papers use the term "Great Moderation," or even the word "moderation."

So it appears that the phenomenon represented by the Great Moderation was noted gradually over time and then documented by a number of scholars. The name for it as well as one of the more rigorous documentations of it were by Stock and Watson (2003).

This term first appeared in print in Buiter and Rahbari (2012). DeTraci Regula, in an undated posting on, suggests that the term was coined by the second author, Citigroup's Ebrahim Rahbari. She also points out the prior existence of, an e-mail storage and organizing tool. I'm told by someone who worked in the EU prior to 2012 that the term was in use there as early as 2010.

Hirschman index
This index of trade concentration first appeared in Hirschman (1945). Michaely (1958) misunderstood it as being identical to the Gini coefficient and called it that in his application to exports, while acknowledging that Hirschman had also used it for that purpose. In fact Hirschman's formula, H=sqrt[Σ(xi/x)2], is not the same as the Gini coefficient. As Hirschman (1964) pointed out, his formula reflects not just unequal distribution but also fewness, its value rising the smaller is the number of goods in the summation. I have calculated both measures in a spreadsheet and can confirm that they do indeed yield different values and that the Hirschman index does indeed fall as the number of goods rises, while the Gini coefficient does not. In spite of this, several others followed Michaely in calling it the Gini coefficient or Gini index, until Hirschman published his correction in 1964.

Hirschman (1964) also pointed out that his index differed from what had come to be called the Herfindahl index of industrial concentration only by Hirschman's inclusion of the square root. Herfindahl had introduced his index in Herfindahl (1950). In spite of the fact that Herfindahl himself acknowledged Hirschman's prior contribution, his index was named the Herfindahl Index by Rosenbluth (1955) and the name stuck, even though Rosenbluth later tried to correct the error. Today (February 2016), the Herfindahl index is said by Wikipedia to be "also known as Herfindahl-Hirschman index." A Google search for "Herfindahl-Hirschman index" get "about 96,700 results" while a search for "Herfindahl index" gets "about 144,000".

As for the Hirschman index itself and its use for quantifying concentration of trade, it is difficult to search for it without the Herfindahl, but I was able to find a few sources that use it. For example, Ng (2002), p. 587 says "The related measure used by UNCTAD is the concentration index, or Hirschman index (H)" and then provides the formula above.

Immiserizing growth
The term "immiserizing growth" was used by Bhagwati (1958) and it seems unlikely that anyone used it before him, since he seems to have coined the word "immiserizing."

As for the concept, Bhagwati credits Johnson (1953, 1955) with identifying a form of immiserizing growth and also with working out the conditions for Bhagwati's form of it in an unpublished note. Long before both of them, Edgeworth (1894, p. 39-40) had shown, though only by example, that increased production of exports could so reduce their relative price that the country would lose or, as Edgeworth put it, be "damnified by the improvement." Perhaps he should have called it "damnifying growth."

Edgeworth in turn credits Mill (1821) with noting the possible worsening of the terms of trade, though Mill apparently incorrectly equated this worsening with a necessary decline in welfare. (I have not read Mill and am taking Edgeworth's word for this.)

Magee (1973) seems to have been the first to do careful theoretical analysis of this phenomenon of the trade balance first worsening before it improves after a devaluation. But he was certainly not the first to use the term, as he cited a passage from the 1972 Wall Street Journal describing "what economists call a J-curve" in reference especially to the aftermath of the devaluation of the British pound in 1967.

A search through Google Scholar finds the term "J-curve" used frequently in other contexts, but the first use of the term applied to effects of a currency devaluation seems to have been at a 1971 conference (the proceedings of which I have not yet seen) and by Posner (1972). Both of these use the term as though it is already familiar, so I suspect that it had entered common use before this in the economic press.

Kaldor-Hicks criterion
Kaldor (1939) was the first to state this criterion, but he was followed in the next issue of the same journal by Hicks (1939) who built on Kaldor and developed the idea more fully. One could easily, based on reading Kaldor and the fact that Hicks did not claim to have had the idea himself, conclude that this should be called simply the "Kaldor criterion." Several authors in the next few years did attribute it solely to Kaldor. Most notable was Scitovszky (1941), who pointed out that the "principle enunciated in Mr. Kaldor's first-quoted article" (p. 77, footnote 1) could in certain cases justify both a policy change and its reversal.

It was Little (1949a, 1949b) who first called it the Kaldor-Hicks criterion. In 1949a, he credited Hicks with going beyond Kaldor by explicitly using it as a criterion for an increase in welfare, and then, throughout the last half of the paper, called it the Kaldor-Hicks criterion. Later the same year Little (1949b) addressed Scitovszky's criticism, and used that terminology throughout.

Lerner diagram
The Lerner Diagram was first drawn by Lerner in an unpublished seminar paper in 1933. He used unit-value isoquants together with unit isocost lines to show the relationship between goods prices and factor prices in the H-O model. That paper was reproduced, "as it was originally written" according to the journal editor, as Lerner (1952). It appears that Findlay and Grubert (1959) were the first to make extensive use of the diagram, attributing it as "a diagram introduced by Mr. A. P. Lerner in his brilliant paper on factor price equalization in international trade." They did not happen to christen it the "Lerner diagram," however, and the first use of this (based on a Google Scholar search) was by Findlay (1971).

Some (including myself, until I learned better) have called it the Lerner-Pearce diagram, giving credit also to Pearce (1952). Bierwag (1964), in footnote 2, p. 57, says "This diagram is often called the Lerner-Pearce diagram..." citing Lerner (1952) and the Pearce's comment in the same issue of the the journal, and says "it has been widely used in international trade theory." That may be, but he cites only Findlay and Grubert (1959) plus the preface to the Japanese edition of Harry Johnson's International Trade and Economic Growth, which I have not attempted to find.

In fact, although Pearce (1952) was debating Lerner regarding the likelihood of factor price equalization, he used unit isoquants, not unit-value isoquants, for the purpose. Since these do not align in equilibrium with a single unit isocost line, they cannot be used in the same way, and they do not achieve the essential simplicity of Lerner's construction. Pearce did use the diagram with unit-value isoquants in his comment on Lerner (1952), but there he wass clearly following Lerner.

Marshall-Lerner condition
The condition was first stated in words by Marshall (1923) as characteristic of two offer curves that intersect in an unstable equilibrium, which he showed in his Fig. 20, p. 353 (appearing here as point E in Figs in the case Both Very Inelastic):
    For they assume the total elasticity of demand of each country to be less than unity, and on average to be less than one half, throughout a large part of its schedule. Nothing approaching this has ever occurred in the real world: it is not inconceivable, but it is absolutely impossible. [p. 354, bold italics added]
Although first published in 1923, the opening footnote to Appendix J in which it appears explains that Marshall had done much of the work between 1869 and 1873, with "somewhat later dates" for "attempts to assign definite measures" and was privately printed and circulated in 1879. So this first statement of the M-L condition dates back at least that far, perhaps half a century before its formal publication.

The next appearance of the condition was twenty years later still, in Lerner (1944). The context was very different from that of Marshall, as Lerner was not looking at the market for international exchange. Rather, his purpose was to determine whether a mechanism for maintaining full employment through the gold standard would be stable. The mechanism would start with a fall in the price level due to the outflow of gold associated with a negative trade balance. To be successful, that fall in price level would need to reduce the trade deficit, thus increasing aggregate demand. But then, "There are other circumstances that render the automatic maintenance of full employment still more precarious." His reason was that the direct effect of a fall in prices is to decrease the value of a given quantity of exports, not increase it, and this then needs to be offset by sufficient increase in export quantity and/or decrease in import quantity in order to cause net exports to rise. For this to be true, he says,

    The critical point is where the sum of the elasticity of demand for imports plus the elasticity of demand for exports is equal to unity.
This then, in words, was precisely the statement that has come down to us as the Marshall-Lerner condition. He also points out that a fall in the value of the currency would serve the purpose no better, as it would be subject to exactly the same result for the same reason. This latter interpretation, separated from the concern with employment, captures what is probably the most familiar statement of what the condition is about: the condition for a devaluation to improve the balance of trade.

A number of authors have chosen to name the condition after Robinson as well as Marshall and Lerner, presumably on the grounds that Robinson (1937) preceded Lerner in examining the same question of a devaluation and the balance of trade. Indeed, she did it more formally in a mathematical footnote that derived the change in the trade balance as

    k{Eq[εf(1h)/(εf+ηh)] − Ip[ηf(1−εh)/(ηf+εh)]}
where k is the percentage devaluation, Eq and Ip are the values of exports and imports, εf and ηh are the elasticities of demand and supply respectively for exports, and εh and ηf the elasticities for imports. Setting this to be greater than zero is not, of course, what we know as the Marshall-Lerner condition. It is not hard to get that condition from it, however: simply set the trade balance to zero (Eq=Ip) and take the limit as both supply elasticities (ηh and ηf) go to infinity. The above result becomes kEq(εf+εh−1), which will be positive if
    εf+εh > 1
Perhaps Robinson knew this and didn't think it worth mentioning, since the two conditions needed for its validity are quite restrictive, especially the assumption of zero trade balance which removes any need to reduce it by devaluation. In any case, in her second edition of the same work, Robinson (1947), she cited Lerner (1944) and stated the familiar result.

There is one other source that might have been expected to reach the result before Lerner (1944): Machlup (1939-40), who dealt in great detail with "The Theory of Foreign Exchanges." He applied supply-and-demand analysis to the exchange of currencies, and he noted that the supply curve for foreign exchange could be backward bending, changing some of market's comparative static implications. However, he did not mention that if both demands were sufficiently inelastic, then the equilibrium would be unstable. Therefore he did not find his way to the third familiar implication of the Marshall-Lerner condition: the stability of the market for foreign exchange.

From all of this, the name "Marshall-Lerner condition" seems appropriate, since both of those authors stated the condition, independently and in different contexts, and no other author seems to have done so.

Who gave it the name? Searching the literature around this time, I find Polak (1947) citing the condition, but only as "the well-known formula." Likewise Haberler (1949) presents the condition and includes the following:

    This condition is usually expressed in terms of elasticities of demand for imports and of demand for exports. It is now often referred to as the "Lerner condition", although it has been mentioned by Marshall and formulated with even greater precision later by Mrs. Robinson.
However, the first I have found to call it the Marshall-Lerner condition was Hirschman (1949). His purpose was to show that the condition is actually incorrect for determining the effect of a devaluation on a non-zero trade balance (as was implicit in Robinson (1937)), and he concludes:
    Our results permit the following conclusions:
      (a) The "Marshall-Lerner" condition for devaluation to have a favorable effect on the trade balance (sum of the two elasticities larger than unity) holds only when imports are equal to exports.
His use of quotation marks around the term suggests he was introducing it. And indeed a search in Google Scholar for "Marshall-Lerner" in the period 1900-1950 finds Hirschman's as the only valid entry. I conclude that it was Hirschman who gave it the name that stuck.

Meade Index
Meade (1955a) did not put his calculation into the form of an index, except in an appendix, but rather suggested adding up the increases in trade and the decreases in trade separately, each weighted by tariffs, and concluding that there had been a gain from trade (in his context of formation of a customs union) if the former were larger than the latter.

In his example of the duty on Dutch and Belgian beer, Meade said (p. 66): "What we need to do, therefore, is to take all the changes in international trade which are due directly or indirectly to the reduction in the Dutch duty on Belgian beer; value each change at its supply price in the exporting country and weight it by the ad valorem rate of duty in the importing country; add up the resulting items for all increases of trade and do the same for all decreases of trade; if the resulting sum for the increases of trade is greater than that for the decreases of trade, than [sic] there is an increase of welfare; and vice versa."

Meade's main point was that one should not look only at the product on which the tariff is being reduced, but rather at all changes in trade that will be caused, both positive and negative, by that change. Of course if tariffs on all other products were universally zero, then the contributions to his calculation for them would also be zero. Hence, this is a simple way of taking account of the second-best nature of a tariff reduction when tariffs on other products are not zero.

In his Appendix II (pp. 120-121), Meade formalized his calculation and called it "an index of the change in world welfare," which he derived as

      dU = uΣi{dxi(pici)}

where U is world utility, u is the common marginal utility, dxi is the change in trade of good i, pi is price to consumers, and ci is price (i.e., cost) to producers.

It seems to have been Vanek (1965, p. 15) who first called this the Meade Index.

Offer curve
The offer curve, without the name, was first developed by Alfred Marshall. It was first published in an Appendix attached to Book III, Chapter VIII, of Marshall (1923), but as he explained in an opening footnote,
    Much of it had been designed to form part of an Appendix to a volume on International Trade, on which a good deal of work was done, chiefly between 1869 and 1873. ....the main body of the present Appendix is reproduced with but little change in substance from that part of the mss. which was privately printed and circulated among economists at home and abroad in 1879.
Marshall himself said he should share credit with others, but he names only Auspitz and Lieben (1879), "in which use is made of diagrams similar to mine, which they had constructed independently." I have viewed that work, which is in German, and did indeed see many diagrams similar to Marshall's. But based on Marshall's above statement, he was the first.

He did not however name it. He referred to his curves throughout only as OE and OG, for England and Germany respectively. Only once did he even use the world "offer" in connection with these curves, saying "E will be prepared to offer only OM′′ of her bales in return for P′′M′′ bales from G."

The best candidate I find for having named the offer curve is Edgeworth (1894). This article was published in three parts, in the second of which, Edgeworth (1894) Part II he showed the curves in various configurations and used the verb "offer" to explain what they represented. Then in the third part, Edgeworth (1894) Part III, discussing a similar diagram of Auspitz and Lieben (1989), he says "Accordingly their supply- or offer- curve is never inelastic in our sense of the term...." That, to me, qualifies Edgeworth as having introducing the term.

Several subsequent authors also used the term, perhaps following Edgeworth or perhaps coming to it on their own. Some used it not to represent international trade, but rather for the supplies and demands of individual consumers. Bowley (1924) however was quite explicit, both in using the diagram which he said had been used "by many writers in the fundamental treatment of foreign trade" (p. 5), and in calling it the offer curve (p. 7). He cited Edgeworth's Mathematical Psychics for the concept and mathematics of the equilibrium, but apparently not for the offer curve.

The only other author who deserves mention here, however, is Lerner (1936). In this classic article on the symmetry between import and export taxes, his argument was built entirely on offer curves. In his opening sentence he stated as one of his purposes that he "demonstrates the applicability of Marshall's 'offer curve' apparatus to the elucidation of this problem." (p. 306) His use seemed to assume that readers were already familiar with both the diagram and the name for it, although his use of the quotation marks around it suggests that he didn't see the name as already universally adopted. But the use of the term in this still widely cited paper has surely secured its place in the lexicon of economics.

Peso problem
The term is often attributed to Milton Friedman, who apparently commented on the market for the Mexican peso in the early 1970s and explained Mexico's high (relative to the US) interest rate by the concern that the peso would be devalued, which it later was. It is not clear that Friedman actually used the term "peso problem," however. Paul Krugman, in his blog on July 15, 2008, says that the term was coined in the "MIT grad student lunchroom," perhaps by him or perhaps by Bill Krasker, who he says "published the first paper using the term" in Krasker (1980).

Policy space
Although the term or a variant began to be used in UNCTAD discussions and documents in the early 2000s, it was defined explicitly in the São Paulo Consensus of UNCTAD (2004, p. 2): "...the space for national economic policy, i.e., the scope for domestic policies, especially in the areas of trade, investment and industrial development, is now often framed by international disciplines, commitments and global market considerations."

Purchasing power parity
It was Cassel (1918, p. 413) who introduced the term in the context of discussing how exchange rates should be reset after World War I and the large differences in inflation that occurred in different countries, especially Sweden and England. "...the rates of exchanges should accordingly be expected to deviate from their old parity in proportion to the inflation of each country. .... I propose to call this parity 'the purchasing power parity.'" The idea, though not the name for it, is much older than that, said to date back at least to the 16th century.

Rent seeking
Rent seeking was introduced to the trade literature by Krueger (1974), who defined it generally but applied it to quantitative restrictions on trade. She noted (p. 291) that government restrictions on economic activity "give rise to rents ..., and people often compete for the rents." She called this competition rent seeking, a term that she apparently coined and that has caught on hugely.

Second-best argument for protection
The introduction of the term "second best" in the context of protection was by Meade (1955b), who included four chapters on "The Second-Best Argument for Trade Control" with subheadings "The raising of revenue," "The Partial Freeing of Trade," "Domestic Divergences," and "Dumping as a complex case."

The classic paper establishing that trade policy is only second best in general for dealing with domestic distortions is by Bhagwati and Ramaswami (1963), who do not cite Meade. However, they also do not use the term "second best," and the point of their contribution is to argue for policies superior to trade policies. The term "second best" was adopted by Lipsey and Lancaster (1956), attributing it to Meade, but their application to tariffs is concerned only with the optimal level of one tariff when another is non-zero. By 1969, in Bhagwati et al. (1969), Bhagwati too was using the "second-best" terminology, but again his and his co-authors' main point was that tariffs are not even second best but only at most third best when other policies can be adjusted. Thus trade economists have generally used the second-best argument against protection rather than for protection.

If policies superior to tariffs are not available, however, the argument may become one in favor of protection. Thus in its simplest form, a government that is unable to levy any other kind of tax but requires revenue in order to function will use tariffs no matter how far down the list from first-best tariffs may lie. This has presumably been understood since long before the distorting effects of tariffs were examined by economists. And even here, the argument is subject to the caveat that the benefits from the government activity must outweigh the welfare losses due to the tariff.

This is equally true in more complex cases. The infant industry argument depends on distortions that prevent infant industries from reaping the full benefits of their production. The first-best policy is to correct or offset that distortion, perhaps by a production subsidy. But if production subsidies are unavailable (not just rejected politically, since that should in principle apply even more to a tariff, if it were fully understood), then a second-best tariff will be beneficial if not too large.

Technology gap model
Those who write about technology gap models routinely cite Posner (1961) as the first of several papers with this idea. However, Posner's paper includes neither the word "technology" nor the word "gap." It was Hufbauer (1966) who elaborated Posner's idea and spoke of a "technological gap account" of trade. Krugman (1986) may have been the first to formalize the model to modern standards, and he certainly used the words. One source cites the same Krugman (1986) paper but listed as "Conference of the International Economic Association, Sweden, 1982," suggesting that the switch from "technological" to "technology" may have originated then with Krugman. I have found no earlier use of the term "technology gap."

Terms of trade
The phrase "terms of trade" was first used with more or less its modern meaning by Marshall (1923), p. 161. In an example involving countries E and G, he spoke of "the amounts to which E and G would be severally willing to trade at various 'terms of trade'; or, to use a phrase which is more appropriate in some connections, at various 'rates of exchange.'" He then explained his preference for the new term on the grounds that "rates of exchange" could be understood to connote monetary exchange rates, while he meant the rate at which goods are traded for other goods.

Having introduced the expression in the book, Marshall then used it in subsequent discussions, but he did not use it exclusively. He seemed to alternate between "terms of trade" and "rate of interchange," two expressions that seemed to be synonyms as he used them.

There is slight uncertainty as to whether this was Marshall's first use of the expression. This is because it also appears in Appendix J of the same book, which a footnote explains was largely written much earlier, between 1869 and 1873, and which was "privately printed and circulated among economists at home and abroad in 1879." (p. 330). However, Appendix J with only very few exceptions does not use "terms of trade," but rather alternates between "rate of interchange" and "exchange index." It seems likely that the few (I only found two) occurrences of "terms of trade" in that appendix were added when it was presumably revised for its 1923 publication. This is supported by the fact that "terms of trade" does not appear at all in the 1920 8th edition of Marshall's (1890) Principles.

Was Marshall the first to use the term? Taussig (1927) said so, citing Marshall (1923). And I have confirmed that Mill (1848) did not use the term. That of course leaves open a great many others who might have. But from the way Marshall introduced the term, it appears that he at least thought it was new.

Variations on the Terms of Trade

Taussig (1927), after explaining Marshall's preference for "terms of trade" over "rate of exchange," went on "to reduce still further the possibilities of misunderstanding" by refining the expression as barter terms of trade, emphasizing that it referred to the rate at which goods are exchanged for other goods. Taussig also distinguished net and gross barter terms of trade, the latter allowing for total amounts paid even when they differ from prices due to trade imbalances that might arise from, say, reparation payments.

Viner(1937) argued that the classical economists were concerned not just with the rates at which goods exchanged for one another, but also with the rates at which factors exchange, through their production of goods and trade. He therefore introduced the factoral terms of trade, both single and double.

Finally, Dorrance (1948) suggested income terms of trade as an alternative to all the others, which he argued gave a misleading indication of the extent to which a country was gaining from trade when markets were in disequilibrium, as had become more common in the mid-20th century. A rise in a country's barter terms of trade, due to a rise in its prices relative to the price of imports, could be harmful if it mainly caused a fall in the quantity it was able to sell. The income terms of trade, because it relates the value of exports -- price times quantity -- to the price of imports, will correctly record a decline if the price increase is more than offset by a quantity decrease.

To summarize, let pX, X, AX be the price, quantity, and productivity of factors producing exports respectively and pM, M, AM be the same for imports. We then have:

Commodity terms of trade
    = Net barter terms of trade:
    NBTT = pX/pM
Gross barter terms of trade: GBTT = M/X
Single factoral terms of trade: SFTT = (Px/Pm)×Ax
Double factoral terms of trade: DFTT = (Px/Pm)×(Ax/Am)
Income terms of trade: ITT = PXX/PM

Up or Down

As defined above, a rise in the terms of trade is an "improvement," in the sense that the country is getting more in return for what it exports. That has been the convention followed by most authors, who have defined it as pX/pM, but not all. Some have defined it as pM/pX, in which case a decline in the terms of trade is an improvement.

Marshall himself treated the terms of trade as a property of the transaction, not of either country, and thus his terms of trade was just the relative price of two goods, the identities of which were arbitrary. Also, a great many of those who have used the concept of the terms of trade have not needed to define it quantitatively, since they could speak simply of the terms of trade improving or deteriorating. But others have needed to incorporate the terms of trade into an economic model or report it as empirical data, and for either purpose it was necessary to choose either pX/pM or pM/pX.

The first to do this was Taussig (1927), who spoke of and reported data for the terms of trade of Great Britain, Canada, and the United States. For each he chose to define it as pM/pX, with the result that when his curves declined, that was beneficial for the country. Viner (1937) made the opposite choice, remarking in a footnote (p. 558) that

Viner's choice has been followed by economists studying both international trade and international development ever since.

However, the opposite choice has often been made by economists studying international monetary, macroeconomic, and financial issues. This seems to have started with Obstfeld (1980, p. 463) who had "... the terms of trade, defined as the price of foreign consumption goods in terms of home goods." This was not his choice when writing with Rogoff in the definitive textbook of the field, Obstfeld and Rogoff (1996), who on p. 25 said "In general a country's terms of trade are defined as the price of its exports in terms of its imports." But the same two authors, writing later, reverted to pM/pX, as have many (but not all) authors in that subfield writing since.

Thank-you note
As a policy to respond to a foreign subsidy. I attribute this to Paul Krugman fairly early in his career. I have, however, been unable to track down where he actually said it. I once asked him directly, but he couldn't recall.

Third World
"But the term third world did not originally refer to geopolitics. The first to use it in its modern sense was Alfred Sauvy, a French demographer who drew a parallel with the 'third estate' (the people) during the French revolution. In 1952 Sauvy wrote that 'this ignored, exploited, scorned Third World, like the Third Estate, wants to become something, too.' He was paraphrasing a remark by Emmanuel-Joseph Sieyès, a delegate to the Estates-General of 1789, who said the third estate is everything, has nothing but wants to be something. The salient feature of the third world was that it wanted economic and political clout." From "Seeing the World Differently," The Economist, June 10, 2010.

Trade deflection
Shibata (1967, p. 151) defines this as a "redirection of imports from third countries through the partner country with the lowest tariff, with the sole aim of realizing tax advantage by exploiting the rate differentials between the member countries within an economic union." He notes that trade deflection had previously been defined in the Stockholm Convention, which established the European Free Trade Association, but somewhat more narrowly and differently as arising from difference in tariffs on raw materials or intermediate inputs that allows a final good to be exported from one member to another of an FTA.