European Integration: Trade Data and Measurement Problems
Published online by Cambridge University Press: 22 May 2009
Savage and Deutsch's relative acceptance (RA) index does not actually measure nations' preferences for transactions with one another. Using a modified measure and 1928–71 data for international merchandise trade, I conclude (1) that EEC members' preferences for one another have increased only modestly since the establishment of their Common Market; and (2) that other groups of Western industrial countries have shown a greater rise in intragroup trade preferences in the same period than the EEC. There follows a discussion on the relationship between intraregional transactions preferences and regional political integration. Two conclusions are reached: (1) high intraregional transactions saliences are necessary but not sufficient preconditions for political integration; and (2) rising intraregional transactions preferences must follow political integration. Trade data suggest that, for the EEC, intraregional saliences are probably high enough to permit integration; but slowly increasing preferences suggest that only a little integration can actually have occurred.
- Copyright © The IO Foundation 1974
1 Savage, I. Richard and Deutsch, Karl W., “A Statistical Model of the Gross Analysis of Transaction Flows,” Econometrica 28 (July 1960): 551–72.CrossRefGoogle Scholar For a briefer description of the index and a reference to a computer program by which it can be calculated, see the note by Alker, Hayward R. Jr., in Behavioral Science 7 (October 1962): 498–99.Google Scholar
2 Deutsch, Karl W., “A Comparison of French and German Elites in the European Political Environment,” in Deutsch, Karl W., Edinger, Lewis J., Macridis, Roy C., and Merritt, Richard L., France, Germany and the Western Alliance (New York: Charles Scribner's Sons, 1967), p. 223.Google Scholar The trend in Deutsch's thinking can be seen in the fact that in early 1962 he felt that “another five or ten years may show how much more ready Europe will have become to surmount what-to-day still look like formidable obstacles in her creative and lasting unification” (“Supranational Organizations in the 1960's,” Journal of Common Market Studies, 1, no. 3 , p. 215). Five years later, however, he felt that such progress had not been made and would not be for “far more than ten years.… Greater unity of action, as well as deeper emotional ties in Western Europe are growing very slowly. In terms of European unification, we may expect after 1975 a possible resumption of the advances of the 1950's, but not much earlier” (“A Comparison of French and German Elites in the European Political Environment,” pp. 300, 301).
3 Alker, Hayward Jr., and Puchala, Donald J., “Trends in Economic Partnership: The North Atlantic Area, 1928–1963,” in Quantitative International Politics, ed. J., David Singer (New York: The Free Press, 1968), p. 310.Google Scholar
4 See, for example, Inglehart, Ronald, “An End to European Integration?,” American Political Science Review 61 (March 1967): 91–105CrossRefGoogle Scholar; Inglehart, , “Public Opinion and Regional Integration,” International Organization 24 (Autumn 1970): 764–95.CrossRefGoogle ScholarPuchala, Donald, in “Patterns in West European Integration,” Journal of Common Market Studies 9 (December 1970): 117–42CrossRefGoogle Scholar, also discusses mass attitudes and offers a useful little bibliography (p. 129) of other such studies. Three of the most important recent works that both provide and analyze elite attitude data are: France, Germany and the Western Alliance; Scheinman, Lawrence and Feld, Werner, “The European Economic Community and National Civil Servants of the Member States,” International Organization 26 (Winter 1972): 121–35CrossRefGoogle Scholar; and Peterson, R. L., “Personnel Decisions and the Independence of the European Communities,” Journal of Common Market Studies 10 (December 1971): 117–37.CrossRefGoogle Scholar
5 See Lindberg, Leon and Scheingold, Stuart, Europe's Would-Be Polity (Englewood Cliffs, N.J.: Prentice-Hall, 1970Google Scholar), especially chapter 3; Nye, Joseph S., “Comparative Regional Integration: Concept and Measurement,” International Organization 22 (Autumn 1968): 855–80, especially pp. 869–70CrossRefGoogle Scholar; Fisher, William E., “An Analysis of the Deutsch Sociocausal Paradigm of Political Integration,” International Organization 23 (Spring 1969): 254–90CrossRefGoogle Scholar; and Caporaso, James, “Fisher's Test of Deutsch's Sociocausal Paradigm of Political Integration: A Research Note,” International Organization 25 (Winter 1971): 121–31.CrossRefGoogle ScholarCaporaso's, “Theory and Method in the Study of International Integration,” International Organization 25 (Spring 1971): 228–53CrossRefGoogle Scholar, also discusses policy output measures and relates them to transaction measures.
6 See, for example, Puchala, Donald, “International Transactions and Regional Integration,” International Organization 24 (Autumn 1970): 732–63CrossRefGoogle Scholar, especially pp. 755–61. Puchala's “Patterns in West European Integration,” pp. 125–29, also uses regional institutionalization data but focuses specifically on French-German relations.
7 Note that the requirement that a given country buy (provide) the same percentage of every other country's exports (imports) takes into account the requirement that no country can export to or import from itself. This condition is sometimes overlooked. For instance, Russett, Bruce M., in one of the first applications of the Savage-Deutsch index, says that “the chances of the consignments' going to country B depend only on the proportion of total world exports that go to B” (Community and Contention: Britain and America in the Twentieth Century [Cambridge, Mass.: The M.I.T. Press, 1963], p. 35).Google Scholar He should have said the proportion of world exports, excluding B's, that go to B.
8 “Transaction Data and Analysis: In Search of Concepts,” International Organization 26 (Autumn 1972): 659–80. In appendix 2 I attempt to prove this mathematically. Appendix 1 contains, for those readers who want to familiarize themselves with the Savage-Deutsch null model, an abbreviated derivation of it. Appendix 3 contains a brief comment on the possibility that the Savage-Deutsch model implicitly contains a different notion of independence than origin- and destination-independence as defined above.
9 The only case in which the Savage-Deutsch null model does happen to express origin- and destination-independence is when all the marginals are equal, as in Hughes' example for time 1.
10 Again, the only instance in which they are is, I believe, that in which all the real marginals are equal, as in Hughes's example for time 1.
11 In real trade matrices, the difference between real and hypothetical marginal values seem usually to be only a few percent. Of course, transaction totals for the whole matrix will be, for either independence assumption and for both simultaneously, equal to real world values.
12 The reader who is familiar with the Savage-Deutsch null model will note that similarity to the formula for simultaneous origin- and destination-independence. They are identical in form and differ only in the meanings of the symbols (see appendix 1).
13 I have recently discovered that this formula was mentioned in 1963 by Bruce M. Russett, who in turn attributed it to a suggestion by Thomas Synott. Russett, however, refers to it only as “a modified form of the Savage-Deutsch formula” whose main virtues are that it “simplifies computation immeasurably and gives almost identical results.” In the first place, strictly speaking there is no Savage-Deutsch “formula” but only a system of n quadratic equations in n unknowns. More to the point, the Synott formula is not merely an approximation or simplification of the Savage-Deutsch “formula”; it is fundamentally a better measure of origin- and destination-independence than the Savage-Deutsch null matrix. Furthermore, as I will try to demonstrate later, it does not in fact give results that are “almost identical” to those obtained by the Savage-Deutsch method (see Russett's Community and Contention, p. 35, footnote 14).
14 Actually a good case can be made for the following: When, as they often do, the two independence assumptions yield nearly identical results, the cell-by-cell average of the two independence matrices yields the better of the two measures of simultaneous origin-and destination-independence.
15 The United States, Canada, the United Kingdom, Ireland, France, West Germany, Italy, the Netherlands, Norway, Denmark, Iceland, Sweden, Finland, Switzerland, Austria, Japan, Australia, and New Zealand. Since such a list is always somewhat arbitrary, perhaps I should point out that Japan was considered to be Western and industrialized, Ireland was considered to be industrialized while Spain and Greece were not, and Finland was considered to be Western while Yugoslavia was not. The Union of South Africa was not included, though it may well have been. In general, this article uses all the countries used by Savage and Deutsch, plus three from the Pacific—Japan, Australia, and New Zealand—plus Austria.
16 Keep in mind that lumping Belgium and Luxembourg together excludes Belgian-Luxembourg trade for the analysis and reduces the number of founding members of the Common Market from six to five.
17 It is generally accepted that tariff changes do not fully take effect until a year or two after they occur. Thus 1960 and 1970 may be better markers for the beginning and the end of the construction of the EEC customs union than 1958 and 1968 or 1969.
18 Of course, it is possible that EEC members' preferences for one another's trade would actually have shrunk over the last decade if the EEC had not existed. Thus, the data above do not really prove the near irrelevance of the EEC for its members' trade preferences. Yet we cannot overlook the fact that intra-EEC preferences have, relatively speaking, been quite modest. This conclusion is strengthened by the fact that during the years before the construction of the Common Market, the data give much the same picture as in the years since 1960. In other words, the advent of the EEC seems to have had little effect on the standing of its members' intragroup RA scores relative to those of other groups of Western nations.
19 For those familiar with Brams's “Transaction Flows in the International System,” I should point out a problem in the use of words. What Brams called relative salience is what I am calling preference. Brams's absolute salience is close to what I am calling simply salience or, later, foreign salience. Both his absolute salience and my salience reflect countries' total international transactions as well as their preferences for one another. However, Brams measured his absolute salience by taking the difference between real dyadic transactions and the value expected on the basis of the Savage-Deutsch null model. I have chosen different measures because they are, I think, easier to grasp, easier to compute, and have been more widely used since Brams's, article first appeared in the American Political Science Review 60 (December 1966): 880–98.CrossRefGoogle Scholar
20 It is also possible to distinguish between foreign salience and general salience. In this article salience refers to foreign salience, that is, the salience of one country among all foreign countries. General salience refers to the salience of A's transactions with B among all of A's transactions, including purely domestic ones. Foreign salience could be measured as I have suggested in the text. General salience could be measured by A's trade with B as a percentage of A's (or B's) GNP. Both types of salience, however, would reflect countries' total transactions as well as their preferences for one another.
21 Community and Contention, p. 37. Note that implicit assumption that equal numbers of incoming messages will receive equal attention.
23 “Tourist Traffic and Integration Potential,” Journal of Common Market Studies 2 (March 1964): 254–55.
24 “A Comparison of French and German Elites in the European Political Environment,” pp. 215, 219, 218.
26 “Trends in Economic Partnership: The North Atlantic Area, 1928–1963,” p. 310.
27 Ibid., pp. 314–315. This conclusion is buttressed by two pages of salience data, e.g., intra-EEC trade as a percentage of EEC trade with nonmembers, 1953–63.
28 “An End to European Integration?,” p. 411. Unfortunately, much of Inglehart's (and others') criticism of Deutsch has been blunted by the false assumption that RA indices are lowered by rapidly growing GNP, national productivity, and total international transactions. GNPs and some “general rise of … productivity” simply have nothing to do with the computation of the RA. The index is based only on international, not on intranational, transactions. It is true that rapidly growing dyadic transactions will not produce a rising RA if they reflect only generally rapid growth in the worldwide transactions of the countries involved. But it is also true that the RA index will rise if growing international transactions are accompanied by increasing preferences. In short, the RA will always measure preferences, regardless of any trends in overall transaction levels. I also cannot keep from mentioning that my data do not agree with Inglehart's on one point. In 1928 American exports to Western Europe were 35 percent below the indifference expectation value, and US imports from Europe were 58 percent below. In 1957 these figures were 52 percent and 57 percent below, respectively. This indicates virtually no change in American acceptance of European goods and a fall, but not by anything like a factor of 2, in European acceptance of American goods.
29 “Comparative Regional Integration: Concept and Measurement,” p. 864.
32 Nye is one of the few who have suggested using absolute magnitudes to measure transaction flows. Even if (for trade data) inflationary effects were eliminated—and this is rarely done—absolute magnitudes do not systematically reflect preferences at all. Thus, if one doubled all dyadic trade and all national GNPs, no measure of commercial preference would change but all absolute magnitudes would of course double. Even salience measures, while they are not pure preference measures, do at least reflect preferences, along with other things (ibid., p. 864).
33 Europe: An Emergent Nation?, pp. 39–40.
34 The angle of the slice presumably refers to the degree of intraregional preference, while the size of the pie means region members' total foreign transactions.
35 “Theory and Method in the Study of International Integration,” p. 235. He also agrees with Lindberg and Scheingold that the exclusivist or preferential view of integration is as inapplicable to the nation as it is to the region. “[I]t is no longer fruitful to view community and national systems as antagonistic or in terms of one replacing the other. The involvement and co-operation of national elites, the utilization of national channels, and the preservation of national political structures, far from posing a threat to integration at the supranational [international?] level, may actually reinforce it” (ibid., p. 237).
36 “Transaction Data and Analysis: In Search of Concepts,” International Organization 26 (Autumn 1972): 680.
37 Ibid., p. 666. Of course, strictly speaking it is impossible for a measurement to impose “constraints” on a concept. A measure either does or does not measure (correlate highly, but not necessarily linearly, with) a concept. Perhaps RA indices do not measure social integration but they cannot constrain it.
38 Ibid., p. 676. See also Hughes's earlier “Transactional Analysis: The Impact of Operationalization,” especially pp. 134–135.
39 See Chadwick, , “A Brief Critique of ‘Transaction Data and Analysis: In Search of Concepts’ by Barry Hughes,” International Organization 26 (Autumn 1972): 683.CrossRefGoogle Scholar Chadwick may have added that RAs can rise, not only because A comes to buy more of B's goods without buying more of C's and D's, but also because A continues to buy the same amount of stuff from B while buying less from C and D. Since Hughes and Chadwick also discussed the constant sum feature of the RA index, it may be appropriate to point out that, mathematically, the RA may be thought of as summing to zero for any matrix of countries if each dyadic RA score is weighed by the expected values.
40 By exclusive I mean that while other factors may help to produce transactions increases, such increases cannot occur unless also preceded by increased political integration.
41 Without reviewing the academic pedigree of the phrase international political integration, I will simply define it as the sustained ability of a group of nations to cooperate in important (or in increasingly important) ways without war. Whether this cooperation is achieved rationally (by a distribution of net benefits that is on the whole fair) on emotionally (by the creation of loyalties to a new central government or a sense of common identity among the peoples of the new larger unit) is not part of this definition. Nor is the issue of whether political integration is primarily a mass or an elite process or the question of whether the new unit will be intergovernmental or supranational in its formal structure part of this definition. Nor, finally, does this definition address itself to such matters as the role of core areas, one or two leadership countries, force or other kinds of threats and negative sanctions in the integration process.
Of course, no one has really proposed that integration should really be defined entirely in terms of transactions. Even Deutsch never really went beyond the position that transactions were simply the best indicator of integration. But it is just a short step from saying “best indicator” to saying “only indicator,” and an only indicator is really just an operational definition. Nye has suggested the use of transactions as the best indicator of integration, but for social, not political, integration (“Comparative Regional Integration: Concept and Measurement,” pp. 863–64).
42 For a definition of general and foreign saliences, see footnote 20. One may also distinguish between foreign and general preferences. In foreign preferences, a country's domestic economy is excluded by the assumption that no country can trade with itself, i.e., the cells in the principal diagonal of the real and null transactions matrices are assigned zero values, as in the Savage-Deutsch model. In general preferences, this assumption is dropped. Throughout this article I, as did Savage and Deutsch, have used only foreign preferences.
43 This point was made by Inglehart in “An End to European Integration?,” p. 413.
44 “Tourist Traffic and Integration Potential,” pp. 251–52.
45 “Comparative Regional Integration: Concept and Measurement,” pp. 862–63.
46 The argument can be carried a step further. There may be a transactions salience threshold beyond which, we can say, the region must be integrated, for otherwise the regional polity simply could not cope with the tensions, issues, and messages that must accompany increased transactions. This amounts to arguing that failure to integrate is (or causes) a series of barriers to further transactions when saliences rise above a certain level. For economic matters, this reasoning makes some sense. Since national governments are ultimately responsible for the performance of the economy, increased regional interdependence (as measured by salience) may at some point become unacceptable unless it is accompanied by greater regional cooperation, whether supranational or intergovernmental.
47 An underlying assumption here is that international transactions will tend to take the path of least resistance, other things being equal. Thus, intraregional integration, by removing barriers to transactions within the region, inevitably leads to increases in such transactions. Incidentally, regional political integration in the area of economic policy may not only facilitate transactions by removing official barriers to them but more directly as well, as for example when governments agree to make more of their official purchases from other region members.
48 “International Transactions and Regional Integration,” p. 762.
50 On page 763, Puchala suggests that intense transaction flows may be one of the “preconditions during regional integration.” Exactly how transactions can be a precondition for but not a cause of integration I am not sure. I can only assume that by “precondition” he means necessary but not sufficient precondition while by “cause” he means sufficient but not always necessary cause.
51 It will be interesting to see what effect, if any, recent exchange-rate changes and the accompanying financial-economic turmoil have had on our measures of intraregional salience and preference.
52 By saying that Atlantic or Western integration is weak, I by no means mean to imply that recent Western attempts to improve the international economic decision-making process have failed. The swap network, the creation of Special Drawing Rights in the IMF, and the rapidly expanding network of forums and procedures for consultation and cooperation (monthly central bankers' meetings in Basle, the GATT, IMF and OECD, and even UNCTAD, and their many subgroups) are, taken together, an attempt at intergovernmental political integration which, in my judgment, rivals the EEC in long-term importance. Yet the decisions of this new international economic polity, while they may be dominated by Western officials, are effective worldwide. GATT, IMF, and UNCTAD are, after all, global, not Western, institutions. One could not, therefore, expect their successes to be reflected in intraregional trade data, even for a large region like the industrial West.
53 An Econometric Study of International Trade Flows (Amsterdam: NorthHolland Publishing Company, 1966).
54 Gallup polls in the mid-1960s indicated favorable attitudes in Norway toward Norwegian membership in the Common Market, but these were significantly changed (at least enough to tip the scales in the other direction) during last fall's referendum campaign on a specific (and not unfavorable) set of entry conditions. To take another example, de Gaulle's popularity in the polls rose during his boycott of EEC institutions, though French attitudes at this time favored supranational integration—the very thing against which de Gaulle was fighting.
56 To take just one example, the failure of various EEC decisional outputs on economic and monetary union to be headed—Commission proposals ignored, Council decisions violated—has actually increased the value of any policy output index, for an effective proposal or decision need never be repeated, while an ineffective one may have to be done over and over again.