The pattern of summit performance identified above presents a three stage progression, in which high levels of effectiveness in the first period (1975-1980) are followed by a decline to low levels in the second period (1981-85), and a return toward moderate levels in the third period (1986-1988). This particular uneven pattern, with its three different levels and two shifts in direction, severely compromises the explanatory value of any theory or factor requiring a unilinear progression, or at least one without reversals, throughout the summit's fourteen year history. Such theories include simple models - such as memory loss, hegemonic decline, or international institutionalization from both subjectivist and objectivist schools. Their lack of fit inspires a search for more complex explanatory models, including ones that might combine objectivist and subjectivist variables, but that still respect the limited testing possible with a three case dependent variable, with only rather rough values attached to each case.

The most straightforward offering of the subjectivist tradition suggests that concerts decay, in a continuous process, as memories of the shock that led to their formation fade. Such a conception is contradicted by the revival of summit performance in the late 1980's. It also contradicted by the very rapid drop from high levels at the opening of the decade, for there is little reason to assume that a five year period is the theoretically appropriate time after which memory loss will become pronounced.

An adaptation of this "memory loss" model stresses the replacement of the founding generation by those with no first-hand experience of the seminal shock that gave rise to the concert. This "founders departure" model seems at first glance to have some merit. The steep drop in performance at the turn of the decade was indeed accompanied by a major leadership transition, as Thatcher replaced Callaghan, Mitterrand replaced Schmidt, and Reagan replaced Carter. But even here Schmidt, the founder (and co- author with Giscard), continued through 1982, and Trudeau, who had been present since 1976, returned from a one-year absence in 1980. More importantly, a leadership transition in the largest powers, the United States and Japan, earlier in the 1970's caused no comparable degradation. Indeed, as early as 1978 a majority of the summiteers at the table were non-founders and at the 1979 summit a full three-quarters were. Yet these two summits are generally regarded as the high point of summitry, along with Rambouillet. And the leadership in the United States, Japan, the United Kingdom, France and Germany remained constant through the 1985 transition from low to medium performance. At the ministerial level one might note the earlier return of founder George Shultz to summit-relevant office as US Secretary of State. But even here it was his colleague, Treasury Secretary James Baker, who was not present at the creation, that was in the vanguard of the latter US change in policy and accompanying revival of summit performance.

A variant of these cognitively-based models stresses the short-term learning that leaders experience from one summit to the next. The simplest version, that of cybernetic learning, suggests that leaders will try to keep summit performance within acceptable limits, and hence the summit as an institution alive. They will do so by reacting to failed summits with a determination to do better next time. Conversely, they will relax after strikingly successful summits, which suggest they have no further lessons to learn. There is some historical evidence to suggest that this process in fact takes place in the summit sequence. The Trilateralists were inspired to outperform the failure of Puerto Rico (and in the case of the US prove their Republican successors inadequate) in London next year. And the failure of Versailles provoked a similar determination to do better next time. But this hypothesis is empirically contradicted, even in the Putnam and Bayne data, by the trilogy of back-to-back successful summits from 1977 to 1979, the unsuccessful trilogy from 1980-82 and the failed pair of 1984-5.

If these cognitively - based models do not generate an explanation of the observed trend, or any particular pattern of their own, perhaps they can be combined with objective factors to improve the result. In particular, it can be argued that the memory loss after the seminal shock model is critically affected by the presence or absence of a successor physical shock that is as severe as, and substantively similar to the original shock. Especially if this successor shock is highly similar, and recalls a proximate rather than remote seminal event, it should remind leaders of the distant disaster and inspire them to concerted action of the high standards of their vibrant, youthful years.

To test this "objective reminder" model, Table E lists the shocks to which the seven power summit responded after its formation, and pairs them with the seminal or formative shocks which most closely resemble each. The data suggest there is some correlation between the similarity of seminal and successor shock on the one hand, and the success of the summit's response to the successor shock on the other. The memory of the facts (and failure) of the 1973 oil shock appear to have inspired the summiteers to generate a successful response to the second oil shock (of a similar sort, from the same source) six years later. This interpretation is sustained by the heavy attention given energy issues at the Bonn summit of 1978. Moreover, the accompanying hostage and Afghanistan shocks, which had no similar seminal shock predecessor, met with a failed summit response. Three years later, the dollar shock, with no similar predecessor, generated a failed response, while the 1982 recession, with medium levels of similarly, produced a fair response.

However the final three shocks - third world debt 1982, US deficits 1983 and the stock market crash of 1987 - all appear to contradict this pattern. They show an inverse relationship between similarity and successor success. Yet these anomalies in the third stage can be attributed to two factors: their greater distance in time from the most similar seminal shock (leaving a less vivid memory to be reminded of); and (in the case of the earlier 1982 debt shock), the success of the summit's response to the seminal cognate, the NIEO (on the assumption that failure is remembered and success is forgotten). Although there are too few cases to confidently assess an explanation which combines three factors (seminal- successor shock similarity, distance in time, and success of seminal shock response), this analysis suggests that this interaction of subjective and objective factors may have some explanatory value. It also underscores the creativity of the summit system m responding successfully to the October 1987 stock market crash, with no seminal shocks (and only the distant memory of a similar crash over half a century ago) as a guide.

The introduction of objective factors, and the anomaly of 1987, directs attention to those core factors of interest and capability that lie at the heart of structural realism, and of objectivist conceptions of international concert. In the most austere version of structural realism, the interests of the participating units are assumed to be constant (e.g. survival), and to derive directly from configurations of relative capability (Waltz 1979). In contrast, in Putnam and Bayne's analysis, interests are portrayed as autonomous creations of national - level processes, but are judged to be incapable of independent identification for the purposes of analysis. The problems of independent specification of interests are indeed formidable. However overcoming them is a necessary part of the analysis of concerts in particular, and of co-operation under anarchy in general. For as Oye reminds us, the most simple explanation of successful international policy co-ordination is harmony of interests among the participants, while the most simple explanation of deadlock is differences of interest rather than a failure of coordination (Oye 1986). Unless it can be demonstrated first that interests differed, but that co-ordination and co-operation resulted, there is nothing left for summits, concerts, and the mechanisms of co-operation under anarchy to explain.

In order to specify interests independently, objectively, and as national level variables, this analysis examines the political party affiliation of the participating heads, on a simple conservative - non conservative party dimension. Such a variable is not to be confused with the subjective "moral and ideological consensus" that is said to sustain concerts (Morganthau 1967:446). For these factors are cognitive elements that reside in the minds of the individual leaders at the table and their publics back home. In contrast, it is the party affiliation rather than the leaders themselves that are important here. Thus summit countries may rapidly rotate individuals through the head position, or even send a mere minister to cover the slot (as the Japanese have done) without affecting the operation of this variable. And while there is considerable slippage between a party's place on the political spectrum and the resulting interests it identifies and articulates at the summit table, there is enough continuity to identify gross patterns on a cross-national basis over 14 years.

Table F displays the pattern. It indicates, in the first instance, that there has never been party unity at the summit. Thus the less interesting situation of co-operation-under-harmony has never arisen. It shows, further, that there is nothing inherent in the content of either conservatism or nonconservatism as ideologies that breeds summit success or failure. This contradicts suggestions that conservatives dislike government intervention in markets at the international as well as national level, and thus avoid even trying to co-ordinate policy at their summits. The data indicate, moreover, that there is little relationship between summit success and the dominance of the summit by leaders from either conservative or non-conservative parties. The dominance of a summit by leaders of a single party does not in fact make it easier to arrive at consensus on a co-ordinated approach, regardless of its particular content. Here the successful summits of 1978 and 1979 had as much party dominance (by non-conservatives), as the failed ones of 1981 and 1982. Moreover the highest degree of party compatibility at a summit has come in 1986-1988 (from conservatives). But there are few to argue these have been the most successful years of summitry on record. Indeed, if the uniqueness of the Toronto summit of 1988 lies in its status as the only virtually party-unanimous summit on record (Francois Mitterrand being the only spoiler), there are grounds to believe that harmony does indeed make the agony of policy co-ordination, and the time and trouble of summits, unnecessary.

Two additional variables, however, may interact with party dominance to produce summit success. The first, an international institutional variable, is a doubling of the weight of the party affiliation of the leader who is hosting the summit, in recognition of the disproportionate prerogatives and power that hosting provides. The second, a relative capability variable, is a doubling of the weight of the party affiliation of the President of the United States, in recognition of the fact that the United States throughout summit history has deployed at least twice as much national capability as the next most powerful summit member, Japan.

Table F indicates that the international organizational factor of summit hosting makes no discernible difference. However the relative capability adjustment of doubling the United States' weight does produce one predicted result. It shows the three summits of 1980, 1981 and 1982 were unique in being "deadlock" summits, with an equal weighting of conservative party representatives and non-conservative party representatives at the summit table. With less party dominance than in the previous period, this data is consistent with the decline in summit performance from the first to the second period. However it still fails to explain why the summits from 1983 onward, and especially those of 1986-8, with their strong common party dominance, did not generate summit successes that surpassed those of the first five years.

This conclusion directs attention to variables that deal with relative capability alone. The most general, and potentially most powerful, of these models is that dealing with overall relative capability among the powers of the system. Figure 1 displays the results of such an analysis of the summit seven, using Charles Doran's index of overall national capability (Doran and Parsons 1980). These data show that in all three time periods the seven-power summit has been characterized by large inequalities in capability, even as it has had a variable record of success. Moreover, and also in contrast to the predictions of the "concert - equality" model (in which effectively equal capabilities among the members breed concert success), these data show that growing inequality of capabilities among the summit seven characterized the 1976-80 period of high summit success. Moreover the equalization of capabilities (driven by a declining US and rising Japan) marked the 1980-85 period of poor performance. Indeed, the data suggests support for the hegemonic stability model, in which summit success depends on a strengthening (as well as strong) United States, and the leadership which such strength and self confidence enables and inspires the US to provide.

This emphasis on an increase in US capabilities relative to its summit partners as a recipe for summit success is confirmed by the data in Table G. It reports the ranking of the summit seven countries in their GDP/GNP growth performance over the summit's first twelve years. These data, in which the rise or fall in GNP captures the dimension of each country as egoists, and the ranking captures their status as competitors, show the US ranking relatively highly from 1975 to 1978, relatively poorly from 1979 to 1982, and relatively highly from 1983 to 1986. Such a pattern is consistent with the observed pattern of summit performance if one allows for a one to two year lag in growth ranking translating into summit performance. This seems reasonable, as a relatively poor growth performance one year will harm one's reputation at next year's summit, and even then can be explained away as a temporary one-year aberration until it becomes a trend the following year.

Taken together, these data suggest it was the weakening in US overall capabilities, and its low ranking among its summit colleagues in GNP growth, that caused the failed summits of the early 1980's. Figure 2, which shows US growth (domestic demand) relative to its OECD partners during the 1980's, confirms this view. It shows that the dramatic and sustained increase in relative US growth (and thus the self confidence that soaring domestic demand bred) after 1983 coincided, after a one or two year time lag, with the revival in summit performance in the late 1980's.

The presence of these time lags however, raises once again the subjective question of how changes in real capabilities enter into the perceptions and predispositions of the leaders that must chose to consult, co-operate and co-ordinate at the summit table. Given the above findings about the absence of potent long memories (without objective reminders) on the part of the sumiteers, these leaders can be assumed to have short term memories, with the annual summits lengthening them by only one year. Thus dramatic changes in performance indicators that take place within a single year, that are highly visible, and that can be reasonably thought by the leaders to have an important effect on overall relative national capabilities, can be assumed to be the dominant factors that leaders focus on, and act on the basis of, when they go into their summit each year.

Figures 3, 4, and 5 report the data on three such short-term, perceptually-potent performance indicators: the US real exchange rate, the US external deficit, and the US ratio of debt to GNP. All three have been the focus of sustained, vigorous debate at the summit table and throughout the western world during the 1980's. Moreover all three are so intensely focused upon and argued about precisely because they are thought to, and do, have an important effect on real capabilities. Exchange rates have a dramatic short-term effect on relative GNP by multiplying or dividing the real capability a country has relative to its colleagues outside. Similarly, external deficits drain or augment countries' foreign exchange reserves (or exchange rate), while the debt ratio determines how much of the shortfall can ultimately be financed at home, rather than abroad.

Each of these three measures shows a high degree of correspondence with the observed three - stage progression of high-low-medium summit performance. And each does so, as hypothesized, with few serious time lags or leads. On the most potent measure - US exchange rates - the five years following Rambouillet were characterized by a falling US dollar (and thus falling relative GNP in current dollar denominated terms). The five years following 1980 were marked by the soaring US superdollar, which gave the US an almost 50% boost in relative current dollar-denominated GNP within a five year period. This gave the US the immense self-confidence that came from knowing that all the world wanted and trusted its currency once again, and the propensities toward unilateralism which rising relative capabilities breed. Most recently, the period since 1985 has been marked by an equally dramatic and sustained plunge in the US dollar. The stability of the slopes may be partially seen as an effect, rather than a cause of summit success (in this case the summit system's successful exchange rate management). But in both 1981 and 1985 the trend changed before the summit altered its policy direction or approach to intervention and exchange rate management.

The level (as well as the direction) of the US dollar also matches the varying high-low-medium-level of summit performance over these three periods. Thus, it was only with the Versailles summit of 1982 that the US dollar rose above the highest and generally stable level of the 1975-80 period. Here unprecedentedly high US dollar values produced unprecedentedly bad summits. Although the trend changed in 1985, the Toronto summit of 1988 was the first one of the late 1980's in which the dollar fell beneath the highest levels of 1975-80, and in which the incentive to generate summit successes of the high quality of Rambouillet, Bonn I and Tokyo I thus arose. If one accepts the classical view that the seven power summit is above all about money, that it was created in the wake of 1971 essentially to rescue a too-weak US dollar, and that its great early successes came in US dollar stabilizing package deals, then this pattern makes eminent sense. Moreover, especially if one awards high marks to the 1988 Toronto summit, one can look forward to a new generation of weak-dollar driven and thus strikingly successful summits as the 1990's dawn.

Figure 4, which reports the data on the second short term capability indicator - the US external deficit - supports this analysis. It shows that the successful summits of the 1975 to 1978 period were immediately preceded by a sharply worsening (and largely oil import-driven) trade deficit. The generally less successful summits of 1979 to 1982 were preceded by a generally improving US external balance. And the summits from 1983 onward were preceded by one steadily deteriorating to unprecedented lows. The anomaly of 1979 can perhaps be accounted for by the uniquely successful response of the Tokyo summit to the second oil shock, which was in turn due to the unique presence of a proximate first shock of very similar characteristics. The lethargy of the 1983 to 1985 summits, in the wake of a steadily deteriorating US external deficit positions, remains unexplained.

Finally, as Figure 5 shows, US debt as a percentage of its GNP generally rose from 1974 to 1977, fell from 1978 to 1981, and rose from 1982 onward. Such a pattern is again consistent with the predicted pattern, again with the anomalies of Tokyo I and the summits of 1983-6.

---

Source: Prepared for the annual meeting of the International Studies Association and the British International Studies Association, London, March 29-April 1, 1989. Unpublished in print. Reproduced by permission of Professor John J. Kirton.