Economists as Storytellers: Scenario Drafting at the International Monetary Fund

1. Introduction

In his book Narrative Economics: How Stories Go Viral and Drive Major Economic Events, Nobel Memorial Prize laureate Robert Shiller (2019) argues that “narratives” or “stories”—a term not really defined by the author—drive economic events. Narratives, or so Shiller claims, can go viral and by this means effect changes in economic behavior and thus economic conditions. Contrary to his own title, Shiller provides little evidence on how narratives can become “contagious” and influence economic processes. Indeed, this does not seem to be his focus at all. Testifying to his reluctance to ponder the causal relationship between narratives and economic behavior, Shiller describes the process as “a random event, like the mutation in a microbe such as a bacterium or virus,” and explains that ultimately “we can give no final proof of causality because these events are so deeply complicated, and multiple narratives are involved” (75, 112).

Shiller notes that economic theories can, at times, transform into stories that go viral, and he mentions the example of the Laffer curve. And yet, the role of economists in the production and dissemination of narratives remains vague. Shiller asks economists to pay more attention to the role of narrative in economic activity but says little about their—and his own—role as storytellers.

This article tackles this shortcoming. Focusing on the history of scenario drafting at the International Monetary Fund (IMF) in the 1970s and 1980s, I explore how IMF economists turned to narrative in their efforts to shape policymaking in member countries. As I argue, the IMF scenarios published from the late 1970s onward drew upon narrative reasoning by which the IMF's researchers wove together qualitative and quantitative assessments, macroeconomic theories, and political convictions in a causal trajectory that spanned from the present to the future. At its core, each scenario could be considered a narrative in itself, linking the present economic situation of a country to a set of alternative future growth rates and the assumed effects for the global economy via different policy measures. By circulating their scenarios, the staff tried to bring about the future the scenarios described. Calling the popularization of a narrative a “random event” thus distracts from a vital fact: namely, that economists often consciously work toward circulating their narratives and by this means influencing economic and political decision-making.

In the mid-1960s, a set of disturbances in the international monetary system led to a gradual change in the IMF's role. In 1969, this prompted the introduction and development of a World Economic Outlook, intended to facilitate the IMF's new responsibility to exercise “firm surveillance” over the policies of member countries (sec. 2). In 1978, this new role was formalized in a set of amendments to the Articles of Agreement of the IMF. The same year saw the introduction of a new scenario technique at the IMF. The IMF's scenarios provided alternative versions of a hypothetical future, derived from the research staff's narrative reasoning by which they organized data, theoretical assumptions, and policy preferences to guide policymaking in member countries (sec. 3). This attempt, however, was not always successful. Focusing on the consultations with the United States in the 1980s, I trace how the IMF staff adopted several strategies to prove the causal chains of their narrative reasoning and by this means enhance the persuasiveness of their recommendations (secs. 4–5). Here, the researchers' narrative reasoning provided the scaffolding to motivate and support further investigation (Currie and Sterelny 2017). In the process, the staff began to buttress their scenarios with econometrics and survey data.

The article points to four different functions of narrative in economic reasoning. Section 3 focuses on two of the functions, its role as a sense-making technology and as a tool of persuasion (Morgan 2022) to investigate and illustrate the effects of different policy measures, both for the research staff and for member countries' administrations. Section 4 highlights how researchers turned to narrative reasoning to fill in gaps and correct quantitative reasoning (the third function), while section 5 explores the staff's use of narratives to link alternative policy measures to changes in parameters and variable inputs in complex econometric models (the fourth function; on the use of narratives to stitch together models, see also Quack and Herfeld [this issue] and Biddle [this issue]). And yet, the article also highlights the limits of narrative (sec. 6). Narratives, I argue, are always contested. To understand “the dynamics of narratives” (Robert Shiller) we therefore also need to explore the sites where different narratives compete—and investigate the strategies economists and policymakers adopt to gain the upper hand in sites of conflict.

2. A World Forecast: The Emergence of the World Economic Outlook in 1969

Upon the formation of the IMF at the Bretton Woods Conference in 1944, its main purposes were to promote exchange stability, to foster the expansion of international trade, and to help rebuild the economies devastated by the Second World War (International Monetary Fund 1944: 1). In the two decades between 1945 and 1965, the IMF pursued these goals with great success, although, as James Boughton (2001: 1) emphasized, the “institution's value derived primarily from its passive embodiment of the understandings and commitments” of the Bretton Woods system. The mid-1960s, however, saw the onset of a set of disturbances ushering in a new era in the history of the IMF and in the international monetary system more generally. Concerns about international liquidity intensified, imbalances of industrial member countries increased, and crises in exchange markets occurred more frequently. As a reaction, IMF officials strengthened their consultation activities. This expansion of activity was so marked that Margaret De Vries (1976: 569) spoke of a “quiet revolution.” In this context, the importance of the annual consultations with member countries increased sharply, with greater emphasis being laid on the discussion of the international effects of members' monetary and fiscal policies. To provide a basis for the consultations, the members of the research staff at the headquarters extended their preparatory work, looking also for new means of conducting multinational analyses and forecasting (576–78).

As part of this endeavor, 1969 saw the emergence of the so-called World Economic Outlook at the IMF. It represented the first attempt by an official agency to issue what economists had dreamed of since the 1920s: a world forecast (on this dream in the 1920s, see Lenel 2021). Upon their introduction in 1969, the IMF's outlooks were based on the projections for industrial countries prepared by the secretariat of the Organization for Economic Cooperation and Development (OECD) and the IMF's interpretation of their implications for both industrial and developing countries (Boughton 2000: 114).¹ As the members of the research staff saw it, the purpose of the exercise was merely to give “some notes and statistical material as a basis for discussion of the short-run outlook for the world economy” (De Vries 1985a: 786). However, as the IMF researchers indicated in their first outlook, these forecasts could sometimes “differ considerably . . . from the views of the staff.”²

In August 1971, as a reaction to an overvalued dollar that undermined the United States’ foreign trading position, President Richard Nixon announced his New Economic Policy. As part of this, the American administration “temporarily” suspended the dollar's convertibility into gold, thus marking the beginning of the collapse of the Bretton Woods system. To many observers abroad, this came as a shock. The retreat of the United States from their postwar obligations created a vacuum, which, above all, demonstrated the need for greater monetary cooperation. The IMF reacted by further strengthening its consultation and coordination activities. As part of that, and in an effort to provide outlooks that were more timely and more consistent with the IMF's policy advice to member countries, the staff members began to compile their own data and projections (Boughton 2000: 114).³ From 1971 onward, the IMF's outlooks were based on single-country models maintained by the IMF research staff as well as extensive correspondence with government and private analysts from various member countries.⁴ The outlooks circulated from the early 1970s onward also entailed explicit commentaries on the policy measures taken by member countries and recommendations on future interventions.⁵

At a meeting at the Smithsonian Institution in Washington in December 1971, the Group of Ten industrial countries agreed on a new set of fixed exchange rates against the dollar. This agreement, however, was not long-lived. As a reaction to the second devaluation of the dollar in February 1973, Japan and most European countries floated their currencies against each other, ending the fixed exchange rate system. In the same year, the members of the IMF research staff began to collaborate with the staff of the area departments in their forecasting efforts. With the participation of the area departments, “the projections for trade balances and external current accounts of the 15 industrial countries also could be made mutually compatible,” reflecting the IMF's increased emphasis on the need for coordination of national fiscal and monetary policies (De Vries 1985a: 789). As the oil crisis of 1973–74 and rising oil prices convinced the members of the staff more than ever of the need to improve the distribution of current account balances between member countries, the IMF began to look for “new procedures . . . to ensure that exchange policies in general and exchange rate developments in particular are consistent with the general objectives of the Fund” (see the Managing Director's Statement of October 1973, in De Vries 1985b: 484). As “a preliminary step,” the IMF decided to hold informal discussions between senior staff members and “high officials of each of a number of countries that have a major impact on international currency relationships” and to incorporate the insights gained during these special consultations in their projections (484). In October of that year, members of the IMF's research staff started to travel to major industrial countries to discuss their policy options. As the researchers explained in their report, “The discussions with senior national officials during those visits fulfilled the main objective of initiating a process wherein the Fund can provide a forum and play a constructive role in the coordination of balance of payments policies in the context of the flexible exchange rate arrangements currently in force.”⁶

3. Narrative Reasoning as Sense-Making Technology and Tool of Persuasion: The Introduction of the Scenario Technique in 1978

A few years later, such a “constructive role” was formalized in a set of amendments to the Articles of Agreement of the IMF. At a meeting in Jamaica in January 1976, the Committee of Twenty agreed on new provisions on exchange rate arrangements. While this Second Amendment, taking effect in April 1978, gave “members freedom of choice in their exchange arrangements,” it did not grant them “freedom of behavior in the sense that members will accept certain obligations and surveillance by the Fund,” as emphasized in the IMF's annual report in 1976 (International Monetary Fund 1976: 44). More specifically, the IMF was to ensure “that the member states were contributing to a stable system of exchange rates” and thus to a stable international monetary system. All countries, from this perspective, had “international responsibilities” that they needed to meet to ensure “a satisfactory rate of recovery and expansion in the world economy.”⁷

As a country's exchange rate could be influenced by different factors, the IMF's monitoring and guidance responsibilities were many. IMF officials explained that “conditions determining long-term capital flows and developments in competitiveness of an economy are likely to be the principal factors affecting the trend of exchange rate movements” (International Monetary Fund 1976: 31). These, however, depended on several factors. The extent to which a country attracted long-run capital flows, for instance, hinged “on such factors as location, relative factor costs, and the general orientation of monetary and fiscal policies,” while a country's competitiveness could be significantly influenced by different rates of inflation in member countries leading to different relative costs of production (31). The IMF therefore had to closely monitor member countries' economic and fiscal policies. Exactly how the IMF would exercise “firm surveillance” over these policies remained vague, though. The annual report stressed that the IMF had “to adopt specific principles for the guidance of all members with respect to those policies.” These principles, however, “have yet to be worked out” (25).

The principles remained vague in the years that followed. Lacking definitive guidelines and regulations, consultations with member countries soon came to form the most important tool “through which Fund surveillance of members' policies is made effective” (see International Monetary Fund 1976: 61; De Vries 1985b: 493). Here, IMF officials and members of a country's administration discussed and evaluated policies and their implications for other member countries, as well as actions or policies of other members. Again and again, IMF officials used these consultations to urge deficit countries to “arrange their domestic policies so as to restrain domestic demand and to permit the shift of resources to the external sector,” while “industrial countries in strong payments positions should ensure continued adequate expansion in domestic demand, within the limits set by effective anti-inflationary policies.”⁸ This, the IMF argued, was necessary to foster the adjustment of external payments positions, which was deemed “both urgent and opportune” (2). While most executive directors agreed with these views, member countries seemed slow in adopting the IMF's advice. In January 1978 therefore, several directors agreed “that some new exercise in international cooperation might be needed.”⁹ This, the managing director explained, was especially important “if there were divergent views.”¹⁰

Faced with the severe economic conditions of the late 1970s and the impression that the present course of policies in the industrial countries was inadequate to meet these challenges, IMF authorities looked for means to pave the way toward a more coordinated procedure on the part of the industrial countries.¹¹ The members of the research staff had long had the impression that their projections had not achieved the desired effects. As they had concluded in 1978, the “basic point—the link between payments improvement and internal policies—did not receive sufficient attention in press reports on the Annual Meeting, and there also developed an apparently widespread skepticism concerning Fund staff (and other) calculations on the substantial positive effects of exchange rate changes that might be expected to occur in the course of the next two years or so.”¹² The introduction of the scenario technique aimed to solve these shortcomings. In early 1978, the research staff developed “a general strategy of policy,” which they “translated . . . into a medium-term ‘scenario’ of coordinated growth and balance of payments adjustment in the industrial countries.”¹³ As the researchers emphasized, its aim was to evaluate the external situation and prospects of member countries on a case-by-case basis and by this means help draw individual “lessons” for policymakers in member countries.¹⁴

The scenarios presented projections of economic growth rates and current account balances of industrial countries under the assumption of different policy strategies.¹⁵ A “desirable” or “recommended” scenario comprised what the members of the research staff called “optimal growth rates consistent with limitation of inflation,” which would yield to “a more even distribution of current account balances,” while a set of two to three alternative scenarios assumed different growth rates (see table 1) believed to worsen current account imbalances.¹⁶ These scenarios, readers learned, would yield severe economic effects.¹⁷

The IMF researchers turned to scenario drafting to dispel doubts about their recommendations and to shape economic and political decision-making in member countries. And yet, readers were left in the dark as to how they arrived at their conclusions. Their scenarios, they explained in 1978, presented not a forecast but were rather “in the nature of a normative projection” (International Monetary Fund 1978: 31). Indeed, the recommended growth rates were not derived from calculations but from the staff's narrative accounts of the effects of alternative policy measures on future economic growth rates and the assumed implications of future growth rates on the distribution of current account balances. Here, the qualitative and quantitative assessments of a given current situation formed the starting point of a set of quasi-fictional episodes, designed to allow decision makers to evaluate the alternative future outcomes of different possible policy actions or inactions and prompt them to take the “right” action.¹⁸ The researchers' turn to narrative served epistemic and rhetorical purposes, acting at once as method of inquiry and of proof (Morgan 2012: 239). The members of the research staff employed narrative reasoning as sense-making technology to better understand certain relationships and (enable policymakers to) explore the effects of different parameter and policy changes, as well as to persuade and guide action (on the use of narratives for persuading new communities, see also Quack and Herfeld [this issue] and Casonato [this issue]). To further highlight the different effects of alternative policy measures, they described them in a set of alternative scenarios (fig. 1).

To give an example how the researchers derived their scenarios from their narrative reasoning, let's consider their outlook of February 1979. Here, the estimated growth rates for 1979 formed the basis for the staff's projections for the immediate years ahead. In their projections, the researchers differentiated between four different possible scenarios, with four alternative sets of policy measures leading to different, mutually interdependent growth rates in industrial countries. While scenario A—the recommended scenario—assumed, among others, the implementation of restrictive fiscal policies in the United States leading to a low growth rate of real GNP (2.5 percent in 1980–81) and higher growth rates in other industrial countries, scenario D, for instance, assumed an even more restrictive fiscal shift and thus an even lower growth rate in the United States (1.5 percent in 1980–81), while the rates of growth in Japan and in the European countries were assumed to be the same as in the recommended scenario.¹⁹ To outline the international consequences of the different growth rates, the staff projected their effects for global current account balances, adding another narrative link to their alternative scenarios. The researchers explained, for instance, that scenario D would “be even more apt to contribute to the international adjustment process than scenario A,” but it had two disadvantages.²⁰ First, the low growth rate in the United States might lead to a substantial increase in the unemployment rate of that country, and second, the assumed lower growth rate “would make it difficult for Japan and the European countries to reach the assumed growth rate without adoption of more expansionary fiscal and monetary policies that could lead to inflationary consequences.”²¹ The links between the projected growth rates in industrial countries and the assumed effects for the global economy, even though the heart of the staff's causal chain, remained vague. Researchers pointed to “basic ideas” and “recent experience” that seemed to suggest that “a lower growth rate of domestic demand in the United States and higher growth rates of domestic demand in some other industrial countries particularly the Federal Republic of Germany, Japan, and Switzerland are appropriate on both domestic and external grounds.”²² Considering whether this was also consistent with the need for policies that would bring a gradual reduction in inflation, they alluded to “some tentative indications” that “can be derived from recent historical experience.”²³ Having “demonstrated” the, from an international standpoint, “best” growth rates in industrial countries, the research staff commented on current and expected future policy measures and their likely effects.

To further illuminate how the change between the estimated growth rates of the present year and the targeted growth rates in the immediate years ahead might come about, the research staff recommended specific measures in industrial countries. In 1978, for instance, several industrial countries were advised to further reduce the rate of wage increase and to restore profit margins, “so as to provide greater incentives for investment.”²⁴ They were also informed that “with respect to fiscal policy, a more expansionary stance will doubtless be needed in most cases to obtain the kinds of growth rates for 1979–80 assumed in the staff's notional scenario.”²⁵ The United States, by contrast, was advised to pursue contractionary demand-management policies to curb inflation and improve current account balances.²⁶

But the research staff went further than to simply recommend possible courses of action. Several of the largest industrial countries learned that they were “expected” to “adopt and pursue appropriate targets for economic growth” and “to avoid undue delay in taking measures to compensate for significant deviations.”²⁷ This was important not only to encourage growth in their own country but also to meet their responsibilities for “the world economy,” or so the staff's argumentation went.²⁸ To emphasize industrial countries' responsibility, the researchers pointed out that the envisaged “best” hypothetical future was “not likely to come about without very significant shifts in the stance of policies in a number of countries,” while divergences from their recommendations could have “serious domestic and/or external implications.”²⁹

In the researchers' account, the world economy was cast as a large, coherent whole that was manageable through firm surveillance and coordination mainly among the seven largest industrial countries. As Jacques Polak, executive director for the Netherlands, explained in a 1985 discussion, it was “widely accepted that the cornerstone of the satisfactory evolution of the world economy lay in the coordination of economic policies by the main industrial countries.”³⁰ Among these countries, the United States was believed to have a “particular responsibility” for exchange stability.³¹ Canada, Japan, Germany, France, Italy, and the United Kingdom figured prominently in the staff's projections, too, with the members of the staff assuming that the rate of growth of the real GNP in these countries was vital for improving or jeopardizing “the pattern of current account balances among the industrial countries” as well as exchange rate stability.³²

By postulating causal links between different actions and events that connected the present with a set of better or worse “imagined futures,” the researchers acted as “choice architects,” limiting the range of possible futures to only those which they had accounted for (Thaler and Sunstein 2008; Cook 2021). This role became even more important as the staff decided to include their scenarios in the published World Economic Outlook in 1981, thus significantly broadening the scenarios' leverage (International Monetary Fund 1981). As newspapers around the world began to comment on the IMF's scenarios with their evaluations of different policy measures, this also increased their performative effects.

4. Using Narrative Reasoning to Patch and Modify: The Quantification of Scenarios in the Early 1980s

The staff's repeated calls for fiscal moderation in the United States were a reaction to the staggering increase of the American fiscal deficit after the 1974–75 recession. Between 1974 and 1976, the American fiscal deficit had grown more than tenfold. Given the belief in the United States’ central role for achieving a sustainable current account balance, urging fiscal restraint in the United States became a priority of the IMF's staff and remained a priority even in the face of the early 1980s US recession.

While the Carter administration agreed with this advice,³³ the Reagan administration sworn in in January 1981 did not share the staff's views. Instead, President Ronald Reagan introduced the largest tax cut in US history, with the result that the federal deficit spiked to a new record high. Here, two different macroeconomic theories competed, with each claiming to adequately represent the workings of the economy. While the neo-Keynesian view, which informed the work of the IMF staff, suggested that an increase in the government deficit led to a rise in interest rates and a fall in investment, supply-side economics postulated that the fiscal deficit was largely irrelevant as it would be offset by private savings and an increase in the supply of labor. From this perspective, tax-reduction measures, such as those implemented by the Economic Recovery Tax Act of 1981, would improve incentives to save and work and by this means offset the rise in government borrowing as well as boost output, real income, and national savings.

In the face of this controversy, the IMF researchers began to adopt several strategies in 1981 to investigate previously undetermined causal chains of their scenarios and by this means increase their recommendations' persuasiveness. Their narrative reasoning provided the framework for these investigations, while the investigations, in turn, served to prove the links woven together by their scenarios, testifying to the codependency of narrative and other forms of scientific representation. And yet, as the consultations with the United States demonstrated, even taken together, the various strategies could not always dispel doubts about the staff's recommendations.

In 1981, the researchers began to quantify the medium-term effects of their scenarios for the world economy, distinguishing between the projected current account balances of industrial countries and several analytical subgroups of developing countries as well as the projected debt burden of four analytical subgroups of non-oil-exporting developing countries.³⁴ To this end, they combined an econometric analysis and a survey-based approach, taking into account country experts' appraisals of the expected adjustment policies.

Nonetheless, the quantitative data and statistical analyses alone were insufficient. Repeatedly pointing out that “it is not possible to predict the implications” of certain developments, the IMF staff used narrative reasoning to fill the gaps and even modify or “reconcile” their projections where necessary.³⁵ As the researchers explained, their projections were “being made at a time of great uncertainty and of unusual problems, with past relationships furnishing only a limited guide to probable future developments.” As “foresight becomes dimmer . . . the projections inevitably take on a somewhat ‘normative’ character.”³⁶ Thus narrative reasoning remained an important part of the scenarios, stitching together and sometimes adjusting survey data and econometric analysis. Confronted with the future's uncertainty, narrative reasoning helped envision the unknown links of a hypothetical chain of events.

The staff's 1981 projections of current account deficits of the oil-importing developing countries in 1985 were accepted by all IMF executive directors. As such, they also began to affect the IMF's lending policies. IMF officials almost took the future scenarios as facts and adapted their adjustment and financing policies accordingly. Executive directors pointed to the figures on current account deficits of non-oil-exporting developing countries contained in the staff's medium-term scenarios and called projections for 1985 “frightening, and probably not financible under current assumptions about the evolution of ODA [official development assistance].”³⁷ Executive directors thus called for an expanded flow of grants and concessional loans from industrial and oil-exporting countries and “a more active” stance of the IMF toward non-oil-exporting developing countries in terms of adjustment and financing.³⁸ As the secretary, Leo van Houtven, explained, “This concern struck a note of gravity, urgency, and insistence, which I thought was perhaps more strongly expressed than it ever had been up to now.”³⁹ No doubt, this was a result of the staff's quantifications, which, although said to be “rather crude,” gave the staff's gloomy projections the air of scientific authority and certainty.⁴⁰

And yet, not everyone accepted the IMF's projections. The staff's 1981 projection of economic growth in the United States met with strong resistance from US officials, including those within the IMF. The IMF's research staff had envisaged a “substantially weaker” economic situation in the United States in 1982 than the US administration in the context of its program for economic recovery.⁴¹ The context of this appraisal was the decision of the new administration of President Ronald Reagan to shift toward fiscal expansion in 1981 (Boughton 2001: 141). The IMF research staff warned against this decision, advocating more fiscal restriction to reduce inflation. US officials questioned the staff's reasoning. As the US executive director Richard Erb pointed out at a board meeting in 1981, attempts to reduce the fiscal deficit might also curb supply and production and by this means “lower U.S. growth even further.”⁴² Lower US economic growth, however, might also adversely affect international trade. “A sharper decline in U.S. output growth over the short term should result in lower U.S. import growth, if not in a decline in U.S. imports.”⁴³ Instead, the proposed tax cuts would stimulate investment and economic growth. As Erb argued, “It was not necessary to accept the empirical reality of the Laffer Curve notion” to realize that the overall tax level and the structure of the tax system influenced economic output and growth.⁴⁴

While US officials, clinging to a supply-side framework, believed that a rapid reduction of the fiscal deficit would sacrifice “the needed incentives to work, save and invest” and thus curb economic growth, other executive directors as well as the staff doubted that tax cuts and decreasing regulation would promote economic growth. As the managing director, Jacques de Larosière, summarized the discussion, “Sustained economic growth in the United States depend[s] on the achievement of a lasting reduction in inflation.”⁴⁵ To this end, a decisive reduction of the large fiscal deficit was vital.⁴⁶ This, de Larosière and several executive directors pointed out, was also important to “enhance the standing of the U.S. dollar in exchange markets,” as it might have a favorable effect on inflationary expectations and thus on interest rates.⁴⁷

Projections were never neutral but reflected the researchers' assumptions about the workings of the economy. As such, Richard Erb just dismissed the staff's outlook, pointing out that “the relationship between the budget deficit and the level of interest rates is not straightforward. Nor is there a straightforward relation between interest rates and exchange rates.”⁴⁸

5. Using Narratives to Adjust Models: The Introduction of Econometric Models in the Mid-1980s

Three years later, the tone had still not changed. In the face of the ongoing debate with US representatives, several executive directors began to express frustration about the US administration ignoring the IMF's recommendations. As Bruno de Maulde, the French executive director, complained in 1984, the consultations with the United States had become a “mockery.”⁴⁹ The IMF had, again and again, called for fiscal moderation. The United States, however, “appeared to pay no attention to the recommendations of the Board.” As de Maulde saw it, the United States therefore “lacked credibility in propounding the idea that the exercise of surveillance by the Fund was the cornerstone of the smooth functioning of the international monetary system.”⁵⁰

In 1980, the research staff had begun to distinguish between the quantitative effects of different policy strategies or scenarios over the next five years in their World Economic Outlook.⁵¹ Five years later, they further simplified the presentation of the results of their evaluation. Plainly stating which set of measures they recommended, they began to explicitly differentiate between alternative policy measures in industrial and developing countries and their projected effects and to label them as “better” or “worse” options (see table 2). To this end, the members of the staff published a so-called baseline scenario and a number of alternative scenarios supposing “better” or “worse” policies of industrial and developing countries than the policy stances assumed in the baseline scenario.⁵² In 1985, for instance, a set of “better” policies with the goal to “reduce [the] ratio of U.S. fiscal deficit to GDP by 2 1/2 percent,” such as cuts in government expenditure, was expected to lead to an average growth of real GDP in industrial countries between 1987 and 1990 of 3.5 percent.⁵³ The “worse policies” scenario, leading to a growth of only 2 percent, assumed no measures to correct the fiscal imbalance in the United States, Canada, France, or Italy.⁵⁴ No doubt, the new presentation with its focus on policy recommendations was another attempt to increase the IMF's influence over the policies of member countries.

The presented policy alternatives were remarkably simple. Developing countries were granted no agency to change their economic situation for the better or worse. Instead, the “better policies” scenario was “characterized principally by a more restrained stance of fiscal policy in the United States.”⁵⁵ As regards other industrial countries, the “better policy” scenario did “not assume any substantial change in the projected stance of fiscal and monetary policies, relative to the baseline scenario.”⁵⁶ European countries, however, were advised to introduce structural policies to reduce unemployment and enhance productivity. This, the members of the staff believed, might also lead to a liberalization of import restrictions in industrial country markets and thus raise the export volumes of developing countries. By a few right decisions, or so the reasoning went, Western policymakers were able to influence the course of the world economy.

The members of the staff described the purpose of their scenarios as mainly “illustrative,” emphasizing that “they depict clearly the sorts of constraints that policy makers may encounter over the next few years, as well as the risks inherent in their decisions.”⁵⁷ Indeed, the recommendation of certain policies and the illustration of “how adversely the world economy could be affected over the whole decade because of bad policies during the next few years” came to form the main goal of the exercise.⁵⁸ The presentation and evaluation of alternative scenarios thus served primarily to enhance the pressure to act and to show policymakers how to avoid worst-case developments.

And yet, while IMF researchers maintained their efforts to convince the American administration of the need for fiscal moderation, US officials continued to reject the staff's projections, denying the impact of the fiscal deficit on interest and exchange rates. As Mary Bush, executive director for the United States, explained in a 1984 discussion, “My authorities would still reemphasize their view that there are not clear and well defined causal relationships between fiscal deficits, interest and exchange rates.”⁵⁹ On top of that, American authorities rejected the staff's warning of the repercussions of the large American fiscal deficit on the world economy. Indeed, Bush constructed a counternarrative, insisting “that recent developments in the U.S. economy are having a very favorable overall effect, both in the revival of economic activity and world trade and in the alleviation of debt problems of developing countries, which are benefitting substantially from growing U.S. markets for their goods.”⁶⁰ US authorities therefore continued to dismiss the IMF's calls for fiscal moderation.

When Mary Bush pointed out that “empirical evidence is not entirely inconclusive on this issue” and that “policy advice had to be grounded in empirical evidence,”⁶¹ the IMF researchers began to gather all the evidence they could find.⁶² Here, their narrative reasoning delivered the causal chains they wanted to prove, namely, the projected international effects of a large fiscal deficit in the United States. Since the introduction of the scenario technique in 1978, the researchers had classified their scenarios according to the size of the predicted fiscal deficit in the United States with its presumed effects on industrial and developing countries and exchange rate stability. After the discussions in summer 1984, the researchers began to prepare an extensive study on fiscal effects on economic activity, interest rates, and exchange rates within the United States and abroad. In a 140-page report circulated in spring 1985, they presented all the available material. The staff members admitted that “the economic literature . . . does not reflect a clear consensus on the direction or quantitative importance of these effects” and that “unfortunately, there is also a divergence of views as to the appropriateness of various methods for testing fiscal effects empirically.”⁶³ And yet, they insisted that it was “possible, on the basis of the available theoretical and empirical evidence, to reach an informed judgment on the direction and rough order of magnitude of the impact” that a change in fiscal policy would have.⁶⁴

Among the studies discussed by the staff was also a recent publication by Paul Masson and Adrian Blundell-Wignall, in which the authors pointed out that if economic decision makers viewed a US fiscal contraction as reflecting a fundamental reorientation of economic policy, both American long-term interest rates and the exchange value of the US dollar would decline sharply. Owing to the fact that the direct effects of the policy on aggregate demand would come into play more gradually, the contractionary effects on output could be partially offset by this decline, Masson and Blundell-Wignall argued.⁶⁵ These conclusions, the research staff argued, would have “important implications for the conduct of fiscal policy in present circumstances. They suggest that a large, vigorous, and consistent program of reductions in the fiscal deficit might exert important effects on interest rates and the value of the U.S. dollar that would be much larger than those indicated by standard empirical work.”⁶⁶ In short, the study by Masson and Blundell-Wignall nicely hammered home the IMF's point.

In August 1984, briefly after the discussion with Mary Bush, Paul Masson joined the IMF research staff. Masson had received his BA from McGill University and his PhD from the London School of Economics and had worked at the Bank of Canada before. At the IMF, Masson helped construct an econometric two-region model (“MINIMOD”) to quantify and compare the effects of alternative industrial country policies on major macroeconomic variables (Haas and Masson 1986: 1). Owing to the researchers' focus on policy measures in the United States, the model distinguished between equations for the United States and equations for the “aggregate rest of the industrial world region,” with both showing the same structure (2). The models were driven by aggregate demand functions and were linked via trade relations and financial transactions. Paul Masson and his colleague Richard Haas obtained the parameters of the model not by direct estimation but by simulating the Federal Reserve Board's Multi-country Model and modifying it “where there were conflicts” with their own theoretical assumptions (11).

Two years later, and based on this model, the research staff introduced the multiregion model “MULTIMOD,” which was much larger than its predecessor, entailing 308 equations and covering seven industrial and developing country blocks (Jamshidi 1989; Masson et al. 1988). Recent advances in computer technology allowed for fast simulations of the model. Like its predecessor, MULTIMOD was used to simulate alternative economic policy measures and to quantify their effects. However, unlike MINIMOD, it was designed to gauge the effects of industrial country policies on major macroeconomic variables in both developed and developing countries. With its focus on the transmission of policy effects, the model was found to accord “well with the Fund's surveillance role over the policies of major countries” (Masson et al. 1988: 1). In line with the staff's emphasis on policy options of industrial countries, developing countries' monetary and fiscal policy options were not accounted for in the model.

The parameters of MULTIMOD were to a large extent estimated using historical data. As the primary objective of MULTIMOD was to generate alternative scenarios to the World Economic Outlook, each year its parameters were trained using the latest judgmental forecasts prepared by the staff as an “exogenous” baseline.⁶⁷ These forecasts, in turn, were based on survey estimates provided on a country-by-country basis by the IMF's area departments.⁶⁸ Using these derived parameters, the model, in its unedited form, coincided with the outlooks' five-year horizon. Using this trained model as a starting point, the researchers were able to generate alternatives to their outlook by changing individual parameters as well as variable inputs of any of the 308 equations, with each set of changes representing a policy measure in any of the countries whose economy was modeled by MULTIMOD. How each set of changes affected the model, that is, how each supposed policy measure was expected to affect a given economy, was shaped by the theoretical assumptions underlying the staff's narrative reasoning. This exemplifies what Morgan and Stapleford (this issue) have called narrative exploration: narratives that explore the possibilities implicit in economists' understanding of certain phenomena (see also Casonato, this issue). For example, the researchers displayed the effects of a possible domestic policy reducing the US trade volume by modifying a specific model parameter. As they explained, “The signs of transmission effects of economic policies, and also the magnitudes of their domestic effects, depend on a number of key parameters, including . . . the degree of openness to trade, and the size of trade elasticities” (Masson et al. 1988: 2). An increase in spending, on the other hand, for instance by the US government, was modeled as an exogenous shock using the input describing fiscal spending (36). Here, narratives took on a new function, linking different policy choices to changes in parameters and variable inputs in the model (first-round narratives).

The outcome of a policy measure was then calculated by driving the model forward and comparing the results to an established baseline. Based on the IMF's goals of price stability, adequate growth, and a sustainable current account balance, the staff compared and evaluated these effects to issue a set of policy recommendations. In the process, staff members employed, again, narrative reasoning to connect policy options to outcomes (second-round narratives). While the staff's scenarios still focused on the effects of alternative policy measures on growth rates and the global economy, the complexity of their construction had grown considerably, with scenarios now being derived from first-round narratives, econometric modeling, and second-round narratives. Thus equipped, the alternative medium-term scenarios lay the basis for the annual consultations with individual member countries.

In 1988, for instance, the staff developed a baseline scenario for the United States that assumed that the US current account deficit remained sizable over the medium term but that sufficient foreign financing was available to meet large current account deficits at an unchanged real exchange rate, and they complemented this scenario by three sets of alternative scenarios. While the first set of alternative scenarios gauged the effects of different policy measures on the fiscal budget, the second assumed a reduced willingness of investors to hold dollar assets at prevailing exchange rates and interest rates and compared the implications of alternative policy responses. The third set of alternative scenarios, ultimately, evaluated the possible impact of changed policies in partner countries. Not surprisingly, the staff described the three sets of scenarios as underscoring “the importance of strong U.S. fiscal action in promoting external adjustment.”⁶⁹

Most executive directors praised the alternative medium-term scenarios as “a very useful means of providing guidance for policy making.”⁷⁰ Executive directors began to refer to the outlined scenarios as “realistic” images of the future from which clear instructions for future policies could be deduced.⁷¹ These instructions, too, were based on those outlined by the staff. As one executive director explained in 1988, for instance, “As the staff's ‘reference’ scenario is neither desirable nor sustainable, some adjustment is inevitable. The preferred course would be the adoption of policy measures similar to those outlined in the staff's ‘policy adjustment’ scenario to ensure a sustainable correction of these imbalances.”⁷² Along these lines, another director explained, “both of these outcomes and the staff's more serious ‘financial tensions’ scenario are clearly unsatisfactory, and demonstrate the need for more vigorous adjustment along the lines of the staff's ‘policy adjustment’ scenario.”⁷³

The tone of the consultations with US authorities improved, too. In 1988, the executive director for the United States praised the IMF's importance in “providing a firm analytical foundation for policy coordination,” highlighting particularly the role of the medium-term scenarios.⁷⁴ During the consultations, the American executive director Charles Dallara referred repeatedly to the staff's scenarios and their implications. This, however, did not mean that American authorities now agreed with the staff's recommendations. Indeed, Dallara still cast doubt on the linkage between the federal fiscal deficit and current account adjustment, pointing out that “recent U.S. experience raises questions about the degree and timing of this linkage.”⁷⁵ The director of the Western Hemisphere Department, Sterie Beza, could not leave it like that. As he explained in the afternoon discussion, “It was a mistake to view the scenarios as failing to postulate a high degree of linkage between fiscal and current account adjustment.” Indeed, Beza insisted, the scenarios “showed a significant linkage. . . . At any rate, what needed to be emphasized was that all the scenarios included the common theme that improved performance of the U.S. economy depended on U.S. fiscal adjustment.”⁷⁶ But Beza could not convince Dallara. As the American executive director insisted, “His authorities clearly had different views from the staff.”⁷⁷

6. The Use and Limits of Narrative Reasoning

In the late 1970s, the researchers used narrative reasoning as a sense-making technology and as a tool of persuasion to understand certain relationships and guide policy actions in member countries (sec. 3). Confronted with the limits of the persuasiveness of their scenarios, the IMF researchers began to adopt several research strategies to illuminate and enhance previously blurry causal chains of their narrative reasoning and by this means dispel doubts about the validity of their recommendations. The researchers' narrative reasoning provided the scaffolding for these investigations, while the investigations, in turn, served to prove the links woven together by the narrative, testifying to the codependency of narrative and other forms of scientific representation as mutually stabilizing devices. First, the members of the research staff examined and developed the link between their policy recommendations and the projected outcomes by quantifying the effects of different policy measures. Second, they introduced a clearer evaluation of the quality of possible policy measures to reduce the risk of alternative interpretations and offer more definite advice as to which policy measure the researchers deemed “good.” With this, they added a moral dimension to their recommendations, thus trying to enforce the persuasiveness and the inevitability of their advice. To further explore the causal links between policy measures and outcomes, the IMF staff resorted, third, to econometric modeling. Thanks to the introduction of multiregion models, the researchers could investigate these effects on a country level and on a global scale.

In this process, narrative reasoning took on different and changing roles. In the early 1980s, researchers began to use narrative reasoning to fill in gaps in evidence that persisted and modify quantitative reasoning (sec. 4). In the mid-1980s, they additionally started to resort to narrative reasoning to link different policy choices to changes in parameters and variable inputs in the model (first-round narratives). To link different policies to specific outcomes, the members of the staff again employed narrative reasoning (second-round narratives), now justified by the underlying econometric models and the narratives linking policies to parameters and variable inputs in the model (sec. 5).

And yet, as the consultations with the United States demonstrated, even taken together, the various strategies could not always dispel doubts about the IMF staff's recommendations. While they did increase the persuasiveness of the staff's scenarios, they did not put an end to the existence of competing assumptions about the links between the different parameters of their narrative reasoning. As the consultations with US officials in the 1980s indicate, the authority of the researchers' scenarios hinged on theoretical assumptions and political preferences. US officials, after all, questioned not only the causal chains postulated by the IMF's staff but also the IMF's emphasis on the need for cooperation. “It is unfortunately easy and often politically convenient to exaggerate the potential gains from international economic coordination and to understate the ability of a nation to guide its own economic destiny,” explained, for instance, the American economist Martin Feldstein, President Reagan's former chief economic adviser, in 1988 (Feldstein 1988: 218). Indeed, Feldstein argued, there was “little potential gain to the United States from international coordination” (217). While IMF authorities called on the “particular responsibility” of the United States for the international community, the American administration had found a new role for the country, placing domestic political interests above the state of the global economy. Given these new policy preferences, even the most elaborate multiregion models were bound to remain ineffective, pointing to the embeddedness of narratives in specific economic and political contexts and their fragility.

7. Conclusion

Over the years that followed, the staff's scenarios increasingly began to be treated as guidelines for policymaking by IMF authorities, serving at once as a yardstick and as a tool in the oversight and promotion of global coordination of economic policies. As such, the scenarios played a vital role in the individual consultations with member countries, which were usually held before the publication of the World Economic Outlooks. They provided the data against which IMF officials came “to detect external shortfalls in our members' adjustment programs qualifying for contingent financing” and informed negotiations among member countries.⁷⁸ After the publication of the first projections based on the MULTIMOD model, directors even pondered if the staff's outlooks might form the basis for some kind of automatic disciplining and disciplinary action. Along these lines, one director asked if the IMF could “work out—at a practical level—how to ensure that appropriate policy actions are taken in the major countries on the basis of the staff's, or some other mutually accepted policy adjustment scenario? Should a deviation from the scenario outcomes trigger a policy action, and if so, what could be the threshold point for such a trigger? Could it be automatic or discretionary?”⁷⁹ The econometric approach had significantly enhanced the credibility and authority of the staff's scenarios, inspiring ideas about new responsibilities for global surveillance and policing.

Such a disciplining mechanism was never implemented. And yet, the IMF did acquire a new degree of soft power over the course of the 1980s. Increasingly, the IMF's lending policies were complemented by new techniques to oversee the policies of industrial countries. While IMF authorities had no direct influence over the policies of noncreditor countries, as the largely ineffective consultations with the United States illustrate, the staff's projections and the consultations with industrial countries based on them did influence domestic as well as international debates and by this means economic expectations. Pointing to the effects of IMF projections on financial markets, the Japanese executive director in 1986, Hirotake Fujino, warned in that year that the publication of the staff's outlooks “could undermine the stability of exchange markets.”⁸⁰ Similarly, his successor Koji Yamazaki argued in 1988 that the publication of the current medium-term scenario could “bring the Japanese authorities under heavy speculative exchange market pressure” and possibly even trigger another “Black Monday.”⁸¹ Japanese directors were not alone in this. Other officials, too, warned that the full disclosure of the IMF's medium-term scenarios to the public could send “inappropriate signals to the markets.”⁸²

The scenarios constructed by the IMF research staff thus played an important role in expanding the IMF's influence over market conditions and policies in member countries. By narrowing down the future to a limited number of predetermined trajectories, which were classified according to their assumed effects for global economic growth and the balance of payments, the staff's scenarios simulated increased planning certainty and affected economic and political decision-making in member countries. This impact was strengthened when the staff began to publish their scenarios in 1981 and further increased after the introduction of elaborate econometric models, which promised to make the future quantifiable and determinable (International Monetary Fund 1981). Critique of American fiscal policy by IMF authorities in the mid-1980s, for instance, was taken up by domestic critics of administration policy and Federal Reserve authorities and formed the basis for an increasing isolation of the United States in multilateral discussions in 1985 (Boughton 2001: 143–46). In line with this, Japanese officials warned that the publication of specific policy recommendations could “overheat” domestic political debate.⁸³

This points to the performative power of narrative: by persuading market participants of the causal relationships between different macroeconomic variables, economists helped strengthen these relationships. How did the staff itself phrase it in 1985 by referring to Masson and Blundell-Wignall's study? “If market participants viewed a U.S. fiscal contraction as reflecting a fundamental change in the stance of policy . . . , both the exchange rate of the U.S. dollar and U.S. long-term interest rates would immediately fall by a fairly substantial amount. These variables would only decline marginally, however, if expectations remained unaltered after the policy change.”⁸⁴ Narratives, after all, mattered.

Economists used narrative as a sense-making technology and as scaffolding for further investigation. Both narrative reasoning and the additional strategies employed to further explore the causal chains of the narrative served as tools of persuasion to convince others of the validity of these linkages. As the—often futile—consultations with the United States remind us, however, this process was never simple or given. As the painstaking efforts of persuasion of the staff with the adoption of different, narrative-stabilizing techniques illustrate, it would be fairly inappropriate to speak of the spread of an economic narrative as a “random event.” Using narrative reasoning to persuade and guide action was first and foremost hard work.

I am very grateful to participants in the conference “Narrative Economics” (September 2–4, 2021) and two anonymous referees for their valuable feedback and comments on the first version of this article. I owe special thanks for extensive and insightful comments on previous drafts to Mary Morgan, Thomas Stapleford, Marcus Mikulcak, and Andreas Lenel. I am also indebted to the staff of the IMF Archives who kindly made available online what I could not consult on-site because of the pandemic.

Notes

References