Hardship and collective violence in France, 1830 to 1960.

We challenge the standard argument whach treats collective violence as an expression of the dissatisfactions felt by populations experiencing hardship after periods of relative well-being. We propose an alternative explanation in which struggles for political power are the central features. Tfme-series analysis of year-to-year fluctuations of collective violence in France from 1830 through 1960 fail to yield significant results for a variety of models designed to represent major arguments in the recent literature stressing the effects of short-term hardship. Similar analyses representing the effects of governmental repression yield results corresponding to our .expectations. So far we have not been able to incorporate adequate measurements of the other major powerstruggle.variables into the time-series analysis. But we take the results of this preliminary investigation as a warrant to continue in that direct ion. HARDSHIP @D COLLECTIVE VIOLENCE I N F W C E , 1830 . t0 .1960 ConsSdering . .. t h e s c a t t e r e d , unsys temat ic and con t r ad ic to ry c h a r a c t e r of t h e a v a i l a b l e evidence, t h e i d e a t h a t ha rdsh ip causes c o l l e c t i v e viol ence has gained s u r p r i s i n g currency. I n r e c e n t yea r s few s c h o l a r s have propounded a simple mechanical r e l a t i o n s h i p between t h e two. Y e t a t l e a s t one v a r i a n t of t h e no t ion has a c t u a l l y gained adherents . That i s t h e exp lana t ion of c o l l e c t i v e v io l ence (and o t h e r forms of p r o t e s t o r re b e l l i o n , whether v i o l e n t o r n o t ) as a response t o a gap between expectat i o n s and achievements. That exp lana t ion can e a s i l y be made t r u e by d e f i n i t i o n -f o r example, by l e t t i n g t h e v io l ence i t s e l f s tand a s t h e fvidence of un rea l i zed expec ta t ions . It can a l s o be made i r r e f u t a b l e but t r i v i a l , simply by au tho r i z ing a n . e t e r n a 1 sea rch f o r one mqre gap t o account f o r t h e v io l ence a$ hand. There Is, however, a c r e d i b l e , weighty and sometimes t e s t a b l e form of t h e argument which reasons from short-run pardship t o p r o t e s t v i a t h e cumulation of i n d i v i d u a l d i s s a t i s ' I f a c t i o n s . We cha l lenge t h e e n t i r e l i n e of argument. Men do, indeed, o f t e n become angry when o t h e r people v i o l a t e t h e i r expec ta t ions . Under some cond i t i ons short-run hardship does, we concede, p r e c i p i t a t e r e b e l l i o n . But we do no t t h ink t h e r e is any gene ra l connect ion between c o l l e c t i v e v io l ence and hardship such t h a t an observer could p r e d i c t one from t h e o t h e r . We doubt t h a t t h e d i v e r s e e v e n t s which go by t h e names of p r o t e s t , c o l l e c t i v e behavior , r e b e l l i o n and v io l ence have anything more i n common than t h e f a c t t h a t a u t h o r i t i e s disqpprove of them. And we.suppose t h a t the principal, immediate causes of collective violence are political: collective violence results from changes.in the relations between groups of men and the major concentrations of coercive power in their environ. ments. This paper says little about the political analysis of collective violence, and-much about hardship..-.Here.we seek merely to show that plausible versions of the expectation-achievement argument fail to explain a the year-to-year fluctuation..in collective violence over an important span of one country's history, while one eminently political variable -the extent of governmental repression -does provide a partial explanation of that fluctuation. Other reports of our.work lay out the political analysis more fully, provide some-evidence of its validity, and.treat'a number of alternative argumentsnot mentioned here. We do not think 1 . for a moment that this particular-investigation disposes of all possible relationships between collective violence and hardship,. or that it comes close to establishing the priority.of politics. At .our most expansive, we claim no more than to have lodged enough doubts against the expectationachievement theories of collective violence to recommend a moratorium on their use as explanations until they have received further tests, and to have provided enough support for a .political-process theory.to justify the investment .of new efforts in.its.elaboration and verification. Lest we.be suspected of battling straw ben, let us mention a few much-cited statements which fo1low:the line of argument we.reject. James C. Davies begins by.speaking.about revolutions, but soon,extends his formulation to a wide. variety of violent events: ". . . revolution is most likely to take place.when a prolonged.period or.rising expectations and rising gratifications is.followed by a.short period of sharp reversal, during which the gap between expectations and gratifications quickly . . widens and becomes intolerable;. The frustration that develops, when it is intense and .widespread in the society, seeks outlets in violent action." (Davies 1969: 547; see also Dav.ies 1962, 1971). In,addition to revolutions in a strict sense of the term, Davies. explicitly applies the scheme to draft riots, student protests, the "Black Rebellion of the 1960s" and the Nazi seizure of power. Despite his insistence that the definitive evidence for this argument must come from observations of attitudes, he is willing to use changes in income, education, economic growth, farm productivity and civil rights as indicators-of expectations and gratifications. More important for present purposes, in.analyzing the Nazis and several other cases, he offers evidence of rapid economic decline after long expansion as substantiation of his argument. Ivo and Rosalind Feierabend (1966) offer two formulations which are germane to the relationship between hardship and collective violence. First, they argue in essence that the higher the ratio of want formation to want satisfaction, the greater a country's propensity to "instability." In one study, literacy and urbanization represented want formation, GNP, caloric intake, physicians, telephones, newspapers and radios represented want satisfaction, and thirty different domestic conflict measures for 1955 to 1961 -ten of them explicitly involving damage' to persons or objects, and a number of others iidplying it.-went into the index of instability. Second, they propose that "the faster (the slower) the rate of change in the modernization process within any given society, the higher (the lower) the level of political=instability within that society." (Feierabend and Feierabend.1966:. 263) In this case, the yearly percent rate of change from 1935 through.1962 in caloric intake, literacy, primary and postprimary education, national income, cost of living, infant mortality, urbanization and radios per thousand population served as indicators of the rapidity of modernization. This time there were two measures of instability: a) the aggregate index mentioned earlier, b) the variance of that index over single years from 1955 through 1961. Their formulation differs from Davies', but it clearly permits predictions from fluctuations in economic well-being to levels of collective violence. Ted Gurr, finally, proposes .that "...a psychological variable, relative deprivation, is the basic precondition for civil strife of any kind, and that the more widespread and intense deprivation is among members of . . . . . ... .. a population, the greater is. the..magnitude of strife in one or another form." (Gurr 1968: 1104; see also.Gurr 1969, 1970) Gurr's models and measurements are more elaborate than-those of Davies or the Feierabends. For present purposes, the essential argument is that both persisting and short-term deprivation have direct, positive effects on the magnitude of civil strife, with allowance .for the effects of legitimacy, coercive potential and soc i a l s t r uc tu r a l . f a c i1 i t a t i on . "Persisting deprivation" combines weighted measures ofeconomic discrimination, political discrimination, potential separatism, dependence on private foreign capital, religious cleavages and lack of educational opportunity. l1Short-term.deprivation" combines declines in foreign trade, inflation, declining rates of growth in GNP, qualitative-.reports of.adverse economic conditions, new restrictions on political participation and representation and new "value-depriving policies of governments". "Magnitude of civil strife" cumulates and weights information about individual conflicts, most of them involving attacks on persons or objects. (We will neglect the complicated measurements of legitimacy, coercive.potentia1 and social-structural facilitation, although they raise intriguing and serious methodological problems.) Gurr, toorreasons from short-run hardship to protest via the cumulation of individua1,dissatisfactions. Although these investigations.,are open to serious.attack on theoretical, technical and substantive grounds, we will not offer a critical assessment of them here.* Our purpose in sketching the three arguments and their implementation is to provide a rationale for our own choice of models and indexes representing the line of reasoning we wish to challenge. We,have taken one critical segment of the expectation-achievement argument, sought to represent it in terms fdthful to the usual formulation of that argument, and tried to test it -thoroughly against excellent data concerning year-to-year fluctuations in collective violence within one country over a long period of time. In theLresearch reported here, we have not represented "expectations" in any "direct or convincing way. We have, instead, inferred changing expectations from fluctuating "achievements" in a manner similar to that sometimes employed by Gurr, Davies and many other advocates of expectation-achievement explanations of collective violence. All the data are yearly aggregate measures for France during the period from 1830 through 1960. Our measure of collective violence is the estimated number of participants in disturbances.in continental France as a whole. Disturbances are continuous interactions involving at.least one group of fifty qr more persons in the course of which someone seized or damaged persons or objects over resistence. hey exclude acts.of 3 international war. The disturbances studied consist of every event,meeting our criteria detected by trained readers of two national newspapers for each day from 1830 through 1860 and 1930 through 1960, plus each day of a randomly-selected three months per year from 1861 through 1929. Once events qualified in this way, we collected information about them frpm a wide variety of sources: other newspapers, published court proceedings, .annual reviews of politics. French national and departmental archives, secondary historical works, and others. We then recorded a,: , great many characteristics of the disturbances, including estimates of the . number of participants, in machine-readable, form. In order to produCe 4 . a continuous series.over the 131 year period, we have performed two extrapolations' which tend to .reduce the variance somewhat: 1) we have ' . estimated the number of participantqin the roughly 6 percent of disturbances where we had insufficient information for a numerical estimate as the mean of all those others in .the same year that we were able to estimate numerically; 2) we'quadrupled our annual figures for the period from 1861 through 1929, in which we..had studied only a quarter of all the months. Altogether, then, we are dealing with 1,989 disturbances and an estimated 3.2 million participants. As one might expect, the number of disturbances and the number of participants vary greatly from one year to another, but vary closely together. Figure 1 represents the numbers of.disturbances and of par5 ticipants in.five-year moving averages for easy legibility. (The analysis itself, however, uses single-year data.) As the figure shows, very high levels of collective violence came around the revolutions of 1830 and 1848, at the beginning of the twentieth century and in the mid-1930s, while exceptionally low levels prevailed in the 1850s and during the two World Wars. Sometimes the transition came abruptly. In the extreme case, there were 93 disturbances and some,90,600 participants in.1851, followed by 2 disturbances and an estimated 950 participants in 1852. Without exception the large, abrupt shifts of this kind mark a major rearrangement of the national structure of political power in France. In'1851-52, the crucial events were Louis Napoleon's coup d'etat, the widespread but unsuccessful insurrection it incited, and the installation of a.police state under the man who was. to become Napoleon 111. Our indicators of hardship and well-being are all economic: 1) an index of food prices, 2) an index of prices of manufactured goods, 3) an index of industrial production. Following the usual practice in expectation-achievement -investigations, we take high levels on the first two variables and low levels .on the third as indicating hardship for the population as a whole. More precisely, we accept,short-run rises in prices and declines in industrial production as evidence of increasing hardship. he conventional argument, which we adopt for the purposes of this inquiry, is that the population compares current experience with that of the immediate past, and therefore suffers "relative deprivation" when . . the economy turns down. Davies and Gurr, among many others, use that . . reasoning explicitly; it also seems consistent with the arguments of the groups of cross-sectional studies represented here by the work of the Feierabends. We make our test of this argument in a roundabout way: not by constructing a single model and rejecting the argument if the model fails, but by.testing a set of models incorporating the,relative-deprivation argument. If none of these models.fits, we can safely reject (for our data) the usual versions of.the argument. Since our.data are measured over time, we have employed econometric time-series techniques. Each of these series -the participant in dis.turbances and the econo'mic .indicators -manifests a trend verified by the non-parametric techniques described in Malinvaud (1966: 390-92). We "detrended" the series using the method of first differences (b X=X :-X :-I), for these reasons : 1) only complicated t t and intuitively meaningless polynomial expressions could,account for the trend in these relatively long -time-series; 2) Detrending using first differences reduces the seriaL-correlation.of the residuals; and 3) most importantly, -detrending usipg first-di.fferences.rather than fitting a. polyndmial function of time is more faithful to current theories of relative deprivation.' By .including a polynomial expression for trend, we would in fact .by treating as ."deprivation years" any years (and only those years) where, for example, obser.vations on.the price index were above the predicted value.' So, in effect, our deprivation measure would depend on -the magnitude (and more importantly, on. the sign) .of the difference from the trend expression, but not,necessarily on the-difference from the preceding year (the measure which the theory implies). The method of first differences, by measuring relative deprivation as the change from one.year to the next, erases this problem. Our first model is one which specifies all of the economic.,deprivation predictors as independent variables, of.the following form: