Monetary Dynamics and the Mundell-Fleming Priority Question: Evidence from the Adaptations in

First Draft Please Do Not Quote

Monetary Dynamics and the Mundell-Fleming Priority Question:

Evidence from the Adaptations in

International Economics

Russell S. Boyer

University of Western Ontario
Warren Young

Bar-Ilan University

September 26, 2005

Paper to be presented at

Rutgers University

Department of Economics

Money, History and Finance Workshop

October 3, 2005

Copyright, Russell S. Boyer and Warren Young

Chapter from:

Boyer and Young, Open Economy Macromodels: origins, development and debates Routledge, London and NY, 2006

The authors thank Jim Boughton, Virginia Carper, Bob Flood, Dale Henderson, Peter Kenen, David Laidler, Donald Moggridge, Robert Mundell, Jacques Polak, Grant Reuber, Bob Solomon, and Bob Solow for conversations about the topic of this paper. A number of pertinent documents were available only from the International Monetary Fund’s Archives. We thank the staff there for their efforts in obtaining them for us. Jane McAndrews provided invaluable service in securing for us obscure Mundell documents from various locations in Canada. None of these individuals is responsible for the conclusions at which we arrive in this paper.

Abstract: All the chapters in Mundell’s International Economics [1968a] were altered in the process of editing their original forms. Some of the changes are extensive, and in at least one case the final version bears little resemblance to the original article. This paper investigates the nature and effects of these adaptations.

Our view is that Mundell’s Capital Mobility Paper [1963e] had attracted sufficient attention to justify a unified presentation of the research that culminated in that article. Furthermore, because of its easy accessibility, the forms of these papers in International Economics were likely to become the definitive versions in the eyes of the profession. It appears to us that editing of the papers in 1967 was undertaken with this possibility in mind, as well as with knowledge of the competition from the paper by Fleming [1962].
The adaptations of the Capital Mobility Paper done for International Economics are in addition to the changes in 1963 to make the Departmental Memorandum version of Mundell’s famous paper suitable for publication in a professional journal. These early changes included the insertion of material which omitted the names of Hume, Polak and Fleming. It therefore appears to claim propriety of results which they had previously reported.
Five years later, the contemporaneous appearance of this paper in the Caves and Johnson [1968] book of readings constrained the extent to which further alterations could be made to it. Consequently, the editing undertaken in 1967 for International Economics left the Capital Mobility Paper essentially in its originally published form (although with an added appendix). Instead other papers, especially those that appeared before 1962, were changed so as to correct defects and to provide an integration of the research agenda which led up to the famous paper. In this way the adaptations for International Economics have the effect of seeming to assert that Mundell had priority over Fleming in deriving the celebrated model.
This possible assertion seems to us to be hard to sustain. Instead our conclusion is congruent with the conventional one. Namely, that Mundell’s Quarterly Journal [1960] paper, with its exclusive interest in dynamics, should not be considered to be part of the Mundell-Fleming canon. Rather that paper should be seen as an unusual model which became the basis of the Principle of Effective Market Classification, which today has much less currency than does the famous model.

1. 
   Introduction
   

In awarding the Bank of Sweden Prize in Economic Sciences in 1999, the Nobel Committee pointed to Mundell’s series of papers during the years 1960-64 as constituting his most important contributions to international macroeconomics. It further stated that these papers are reprinted in his volume International Economics, published in 1968 (Persson, 2000, xii). This statement is problematic because the chapters in that book in all cases differ due to editing from the original publications, some to the extent that they are barely recognizable.

The Nobel committee did not come up with this statement out of thin air, but instead was probably paraphrasing what it had read in vetting reports, which most likely were written by Dornbusch [2000] and Rose [2000]. In turn, these citations were reiterating what had appeared earlier from the pens of Frenkel and Razin [1987], and of Flanders [1989, 329],¹ among others. The form by Frenkel and Razin is the most specific: “The foundations of the Mundell-Fleming model of international macroeconomics were laid a quarter century ago in the classic writings of Robert A. Mundell (1960, 1961a, 1961b, 1963, 1964; collected in 1968)…” (Frenkel and Razin, [1987, 567]). The uniformity of these citations by Dornbusch, Flanders, Frenkel and Razin, Persson, and Rose is striking, as though they arose from an orchestrated campaign.
This paper analyzes the consequences which these adaptations have The hypothesis in this paper is that the editorial decisions about the presentation of the material in International Economics were taken in order to maximize its attractiveness when viewed as a single-volume collection of works. Since such a publication would be so much more accessible, it was likely that close inspection of these articles, especially by individuals or institutions wishing to assess the entire body of work, would be via the book rather than by the original papers. This would be even more likely to occur if Mundell’s “marketing department” (Dornbusch [2000, 201]) repeated a number of times the elaborate phrasing of the citation which is given above. The first part of the citation emphasizes how early the original contributions had appeared, and the second part of the citation points the way to an easily accessible source of the apparently faithful reproductions of the works.
The view that this body of work deserved close attention was founded on the extraordinary interest that the profession had shown in the Capital Mobility Paper (CMP) (Mundell [1963e]). The decisions concerning editing therefore reduced to whether the uniqueness of that paper should be emphasized, with previous work viewed as building towards it, or whether earlier work should be highlighted more, in order to present a coherent and substantial multifaceted body of work focused on a particular topic: open economy macromodelling. These goals were not necessarily inconsistent as the papers could be reworked in order to demonstrate the continuity of the research agenda, an aspect that was lacking in the original publications.
As we show in this paper, in International Economics Mundell did alter all the earlier publications with the result that these changes made the body of work seem more consistent. The chief object of the adaptations was the earliest paper which he had published in the area of international macroeconomics titled “The Monetary Dynamics…” (MD) (Mundell, [1960]), and which, according to later publications, contained defects in its original form. In addition, adaptations to footnotes in subsequent works were made in response to these changes Finally, CMP itself was adapted in the form of the addition of a substantial amount of new material during its transformation from a working paper into its published form in 1963. Changes resulting from its later adaptation (in 1967) are comparatively quite minor.
The result of these adaptations is that the body of work in International Economicwas more cohesive and consistent than a collection of the contributions in their originally published forms would have been. That is, the papers were now integrated, as was Mundell’s stated intention (Mundell [1968a, v]). In addition, the chapters in the book were now ostensibly free of defects. The adaptations therefore constituted a valid body of work, which extended back to materials published prior to Mundell’s tenure at the IMF. The result was to lend credence to the claim that Mundell had developed the famous model before Fleming [1962] did. But such a claim must first make the argument, as Mundell does in his recollections [1999], [2001], and [2002], that the MD article is appropriately seen as part of the Mundell-Fleming canon. We feel that this is a difficult argument to sustain.
II. The Origins of CMP: the March 1963 DM and the June 1963 Adaptation
Less than two months after Fleming published his landmark work (Fleming [1962]), Mundell began writing a manuscript that would evolve into his celebrated Capital Mobility Paper, whose best known version is Mundell [1968h]. The first written evidence of his renewed activity in this line of research, which had lain dormant since 1961 (Mundell [1961a], [1961c]), comes in the form of a two-page attachment to a notice of seminar (Mundell [1963a]). This attachment lays out quite an ambitious agenda, even promising analytical comments on econometric work undertaken to test open economy macromodels. The exchange rate regime question is not explicitly considered, but there is a focus on “…the implications of capital mobility for stabilization policy…” (Mundell [1963a, 2])
Apparently these ambitions were scaled back substantially, and the first version of CMP, and in many ways the definitive one, is the Departmental Memorandum (DM) put out on March 20, 1963, approved by Fleming (Mundell [1963b]). This is the most reliable version in that it includes all the substantive results that are repeated in later versions, but here they appear in a shorter form. The familiar results of the famous checkerboard square² are fully reported in this nine-page version of the paper (although the table itself does not appear until the June 6 incarnation (Mundell [1963c, 18])).
In this March 20 DM one finds discussion of the consequences under flexible exchange rates of: monetary policy, fiscal policy, and foreign exchange policy (which, as in later versions of the paper, is found to be virtually the same as monetary policy). The results reported are repetitious of work Fleming had published four months earlier. Under fixed rates Mundell considers: monetary policy (for which he finds that the Hume-Polak ineffectiveness result holds), fiscal policy, and deficit finance. After this central section of the paper, there is discussion of “Fixed Exchange Rates and Sterilization Operations.” This material, too, appears in the later versions of this paper such as Mundell [1963c], [1963e] (CJEPS63) and [1968h] where it is included under the rubric “Other Policy Combinations.” There are no citations in this DM.
If this DM is accepted as the definitive version, then all subsequent forms of this paper are, in essence, adaptations. They are in a format more suitable for publication in a professional journal in that they contain tables, references, footnotes, and figures. But most of these additions are merely for show and spin; they do not add substantively to the forthright statement one finds in the original, March 20 version. Furthermore the added sections of the paper, which more than double its length, have the feel of being appended to the original paper, in that there is little integration of this new material with what is in the earlier version. All the additions are peripheral both in the sense that they can be ignored with no loss of substance, and also in the sense that they appear entirely at the beginning and at end of the paper. The heart of the paper is essentially unaltered, and the new material bookends what we have in the original version.
When one views the Departmental Memorandum draft of the CMP as the definitive statement of the argument, then the subsequent versions must be seen in a different light. One of the key amendments in all adapted versions of the paper is the addition of a Long Footnote, which is synedochal of not only all the changes which were made to the original paper, but also of the relationship between every paper published during the period 1961-1962 and earlier work. The dominant theme of this footnote is that the results reported in it are solely the outcome of Mundell’s extending previous work on his own model, with the latest work additionally repairing shortcomings in his earlier publications.
In all versions of CMP in which the Long Footnote appears, it is placed in the paper at the point where the newly-introduced diagrams also make their appearance: namely, at the beginning of the newly-appended material which runs to the end of the paper. (For example, in Mundell [1963e] they both appear starting at page 481.) It is appropriate that these two portions of the amendment are adjacent because the diagrams work with the footnote in order to make the argument that the results reported in CMP are completely consistent with the analysis one finds in previous Mundell papers, dating back at least until 1961.³ Furthermore the implicit claim is that the reason these results had not been reported earlier is that the author had not until that time thought about the ultimate destination of the world financial system, when it attained perfect capital mobility.⁴ But once he appreciated what the world would look like at that point, a change of definition of monetary policy was necessary. With the change of definition, the checkerboard square results suddenly appeared. Since this is such an important method of policy classification, Mundell argues that CMP provides an essential guide to policy-making in a globalized world.
The arguments in the Long Footnote are of sufficient importance that we devote the next two sections of this paper to an analysis of it.

3. 
   An Analysis of the Long Footnote in the Adaptations of CMP
   

The Long Footnote in the first adaptation of CMP, the June 1963 manuscript (Mundell [1963c, 14a-b]), covering well over one page of single-spaced typescript, had two contradictory purposes, which created a delectable tension in its construction.⁵ In the same way, virtually all the papers which eventually were included in International Economics walk a fine line between emphasizing continuity with earlier work (at the risk of unnecessary repetition) and emphasizing the novelty of the present contribution (with risk of rupturing the sense of continuity with preceding work). As we note below there is a great deal of repetition in Mundell’s research output. Since the stakes were so high concerning the possible impact of CMP on Mundell’s professional standing, this tension becomes palpable in this crucial footnote.

On the one hand, the author wished to promote a sense of continuity between his earlier work and the results in CMP. For this reason Mundell points to the similarity of the diagrammatic tools presented there to those that he had used in the past. This connection is clearest in the version of the Long Footnote that appears in the first adaptation, because it alludes to only two previous papers (Kyklos (Mundell [1961a]) and CJEPS61 (Mundell [1961c]).⁶ The diagrams from those papers do indeed appear as the upper and lower panels, respectively, of each of the two figures that are presented in CMP. ⁷ The two alterations required in order to make them identical are very minor. First, the FF locus, which had appeared in Kyklos with a positive slope (and in one case it was vertical), now is portrayed as being completely horizontal. This is a straightforward change because CMP is dealing exclusively with the perfect capital mobility case.⁸ In addition, the CJEPS61 diagram now appears mirror-reversed for convenience of exposition.
On the other hand, the author wished to emphasize the novelty of the results which CMP presented. Ideally an earlier result could be presented to serve as a foil for the more recent conclusion. This one does not find in the paper, perhaps because to have done so would have been to emphasize the narrowness and repetitiveness of his research effort. Instead one finds that some inconsistencies are acknowledged but others are not. But even where the acknowledgement is made, it appears to have been done reluctantly and without full reconciliation.
An example of a contradiction to which attention is called is the fiscal policy result for flexible exchange rates. CMP finds that fiscal policy is completely ineffective in influencing output and employment under flexible exchange rates. The Long Footnote points out that in CJEPS61 “…it was…argued that…fiscal …[policy is]…more effective under flexible exchange rates than under fixed exchange rates.” (Mundell [1963e, 481]) This inconsistency is said to be due to differences in definitions and in assumptions.
An inconsistency which goes unacknowledged is that concerning monetary policy. In both Kyklos and CMP, the conclusion is that under fixed exchange rates monetary policy is ineffective when one considers only equilibrium situations. In contrast, CJEPS61 had found that expansionary monetary policy had a positive impact on employment at least in the short-run equilibrium. (Mundell [1961c, 513])
Clearly part of the reason why we have such different conclusions is that alternative definitions are being used in the two cases. CJEPS61 defines an expansionary monetary policy as being a reduction in domestic rates of interest, whereas Kyklos and CMP consider such a policy to be an increase in the quantity of domestic credit. But there is a further difference at work here revolving around whether we are dealing with a short- or a long- run equilibrium. Kyklos seems to be reconcilable with CJEPS61 if the timeframe is short enough so as to permit a consideration of points which in Kyklos, in its original and adapted forms, are variously described as “partial equilibrium,” [1961a, 163] “’equilibrium’” [1961a, 164] “quasi-equilibrium” [1968a, 226]⁹ or “external disequilibrium” [1961a, 165].¹⁰
Despite acknowledgement of the fiscal policy inconsistency, one gets the feeling from the Long Footnote that the author does not relish making such comparisons. His favored interpretation is that for the most part any apparent inconsistency is only illusory. Here is an example of such a claim: “The apparent conflict with the present analysis lies in the different definition of monetary and fiscal policy and in the extreme assumption in the present paper of perfect capital mobility…In both cases the underlying model is (in essence) the same and would yield the same results if the same assumptions were made about capital mobility and the same definitions were used.” (Mundell [1963e, 481-482])
The explanation for this reluctance to make comparisons to earlier work is stated at the very beginning of the Long Footnote. The reason that Mundell does not wish to make an extended comparison with the analysis in CJEPS61 is that that article had as its “…main purpose…to show that commercial policy—import restriction or export promotion—was ineffective under flexible exchange rates…” (Mundell [1963e, 481]). What this discussion fails to point out is that at least two-thirds of CJEPS61 deals with monetary and fiscal policy and therefore a comparison of its results with those of CMP was quite germane.
This reluctance is expressed even more strongly in Mundell’s recollections when commenting on Fleming’s citation of his work (Mundell [2001, 223]). He says of Fleming’s choosing CJEPS61 as the one paper in Mundell’s bibliography which is used as a foil for his own work “Curiously, he chose the least relevant article to his or my topic…What must have been going through his mind to single out that paper (which showed that commercial policy was ineffective or counterproductive under flexible exchange rates but no capital mobility) as the most relevant of my papers on monetary and fiscal policy?”
Our view is that Fleming had chosen precisely the right citation of Mundell’s work, and indeed did not need to allude to any other paper in order to make the comparison. The reason is that CJEPS61 is the first of the checkerboard square articles, in that it deals with both monetary and fiscal policy, and both fixed and flexible exchange rates. It therefore provided a direct comparison with what Fleming was doing, in a way that no other macro paper by Mundell extant at the time did.¹¹
The reference to “no capital mobility” in the quotation above is a mystery to us, but one which recurs in Mundell’s recollections.¹² CJEPS61 clearly specifies that capital movements are assumed to depend on interest rates: “I assume that capital imports and exports depend on the interest rate at home and abroad, but that the latter are parameters determined by monetary policy.” (Mundell [1961c, 510]) Curiously this sentence is modified in the adaptations in International Economics: “We assume that while capital imports and exports may depend on interest rates at home and abroad, the latter are...” (Mundell [1968g, 241])
We suspect that Mundell may be playing a word game here, of the sort that we will encounter again below. While it is true that the results in CJEPS61 do hold if there is “no capital mobility,” they hold as well if capital mobility does exist (although perhaps problems arise in the polar case of perfect capital mobility). That is, zero capital mobility is one, but only one, case in which the results hold, since they hold quite generally. Or to state the point again, the degree of capital mobility is essentially irrelevant to this result.
There seems to be more going on here than just reluctance to make comparisons. Mundell probably wished to hide these particular earlier results on both monetary and fiscal policy because they contrast so strongly with what he concludes in CMP, and because they had been published only sixteen months prior to it. As we have noted the one example that is mentioned in the Long Footnote (but not in the body of the CMP) is the fiscal policy result for flexible exchange rates. As we pointed out as well, the analysis of the monetary policy ineffectiveness result under fixed exchange rates could have been added to the list of inconsistencies. That conclusion, while validated by some interpretations of the work in Kyklos, is at variance with the result reported in a paper published in the interim, namely CJEPS61. Furthermore, this ineffectiveness result for monetary policy in earlier papers (Kyklos and Barter Theory (Mundell [1962b])¹³) is quite rightly attributed to Hume. In CMP Hume’s name is not mentioned, and, although there is a passing reference to Ricardo, the conventional interpretation is that Mundell is claiming this result to be a fresh discovery which he has made.¹⁴ Given Polak’s [1957, 10] contribution in this area, it is appropriate to call it a Hume-Polak result, an expression which Mundell himself has used (Mundell [1991, 481]).¹⁵
These failures to make comparisons to earlier work are quite vexing. While the fiscal policy inconsistency is mentioned in the Long Footnote in all versions of CMP which contain it, there is only a limited degree of explanation for the differences from earlier Mundell papers and even less of an attempt at reconciliation.
As noted above, there is merely the assertion that the present model is “in essence” the same as earlier ones, and therefore these results can be derived from them. These claims of CMP cause the reader to sense that the earlier results are not relevant to the modern world of high capital mobility.¹⁶

The major claim of the Long Footnote is that Mundell had had an epiphany that high capital mobility was the direction in which the world economy was heading, and therefore an analysis of its ultimate destination was appropriate. In this world it was necessary to change the definition of monetary policy. And in turn this change generated profound insights, making clear that the checkerboard square was the preferable way of assessing the policy mix question.