The Moorish Wanderer

Quantitative Tales from Moroccan Politics


suite du post précédent, comparaison des caractéristiques des groupes parlementaires entre 2007 et 2011 en utilisant l’analyse de composantes principales. Les résultats démontrent une cohérence entre les hypothèses évoquées plus tôt sur les déterminants du populisme, et permettent aussi une détermination des ‘types’ de groupes parlementaires.

The opposition-turned government PJD displays a significant degree of heterogeneity, just as USFP, but oddly enough, not Istiqlal (PI)

The graph speaks a thousands words: it scatters the selected intake of both 2007 and 2011 parliaments and assigns scores to these (the methodology can be found here, with the Stata functions described here as well)

The score plot displayed above (whose results are listed below this post) shows interesting results as to how our parliamentary caucuses between 2007 and 2011 are listed. These results challenge in part the common wisdom about Moroccan politics; If anything, I would gladly discuss the graph with the PSU/AGD leadership because as far as the pre-2011 elections go, there was potential for greatness. Most of its caucus is close to the bigwigs in parliament (USFP, PAM and the moderate PJD elements) and if it was not for its unrepresented leadership in parliament: Mohamed Sassi should have put a lot more fight in it when he stood for Rabat parliament in 2007, 3000 votes was certainly not enough, and yet splintered votes could have been gathered up:

RABAT MOUHET (212,644 voters)
Party                            |Votes      %  Seats
Constitutional Union             |   3,250  06.4    -
Independence Party               |   2,715  05.4    -
Party of Justice and Development |  14,267  28.2    2
Popular Movement                 |   5,571  11.0    1
Socialist Union of Popular Forces|   5,367  10.6    1
Others (less than 6%)            |  19,384  38.3    -
Total                               50,554          4

Well, that was about the democratic/radical left. The strange thing however (though it is no surprise, given the intellectual leadership controlling PAM caucus and party structures) is how fractured the ‘left-wing caucus’ -if there ever was: PPS caucus, the second-largest sub-caucus after USFP has been so diluted in its membership -as far as the selected parameters are concerned- it is the farthest on the map; ideology, populism as well as parliamentary leadership (neither Ismail Alaoui nor Nabil Benabdellah have succeeded in their respective bids for a seat in 2011 and 2007) so I should perhaps stop considering PPS to be anything near a nature fit in the grand ‘left-wing coalition’ (I was warned to that, but hey, I am a faithful follower of Saint Thomas) so we are left with PT (Benâtik) and PGVM (Fares-Zaidi) in parliament.

RNI-UC does not seem to be such a great fit after all; I had to make do with the available information (the parliament website displayed only caucuses on the database for the 2007 parliament) but then again, the score plot does not seem to splint them apart: most of these are grouped into two sub-groups, the closer one to PAM and others being made up mainly of RNI members, and perhaps those sympathetic to a strong alliance with the said party. (RCU stands for Rassemblement Constitutionnel Unifié, RNI and UC caucus together, though I cannot say if UC used to back up the government when their RNI bretheren were part of it) On the other hand, there is also the effect of RNI going over to the opposition after 2011; this in fact is the main determinant with the RCU cloud is split.

MP stands at odds with the idea of ‘large party’. While its caucus remained constant between 2007 and 2011, and switched sides quite often over the same period of times, it stood far away from the other large parties (those with representatives elected on the national ballot) which makes it similar to PPS, with murky ideology and no purpose as to its existence (its versatile nature in August 2009 and before the 2011 elections) makes it an establishment party, with some degree of internal cohesion, yet with no particular ideology (they should perhaps work on that Amazigh regionalism a bit more)

At least 5 groups can encompass 2007-2011 caucuses.

USFP and PJD’s caucuses ‘splintering’ is partly due to the same effect observed for RCU (their respective government/opposition swap between 2007 and 2011) and partly due to the more heterogeneous nature of their respective caucuses. From what I have heard (of reliable sources) party discipline is very stern in parliamentary proceedings – i.e. members are expected to vote the way their leadership wants, and pressure is eventually exerted when needed, especially for majority members. Yet USFP and PJD members (after they got into office) regularly challenge their leadership; if additional data about the voting record of each member were made available, the same analysis can be conducted to produce more finessed results.

In the finally analysis, it is possible to group caucuses between 2007 and 2011 into 5 large super-groups, the distances between each points providing a measure of distance according to the selected parameters.
This shows for instance a PAM-USFP alliance is not that stupid nor treacherous, and even a second PAM takeover on PSU is plausible enough.

Factor analysis/correlation                        Number of obs    =      599
    Method: principal-component factors            Retained factors =        4
    Rotation: (unrotated)                          Number of params =       26
         Factor  |   Eigenvalue   Difference        Proportion   Cumulative
        Factor1  |      2.19880      0.66075            0.2749       0.2749
        Factor2  |      1.53806      0.42188            0.1923       0.4671
        Factor3  |      1.11617      0.09060            0.1395       0.6066
        Factor4  |      1.02558      0.06937            0.1282       0.7348
        Factor5  |      0.95621      0.28168            0.1195       0.8544
        Factor6  |      0.67453      0.30509            0.0843       0.9387
        Factor7  |      0.36943      0.24820            0.0462       0.9848
        Factor8  |      0.12123            .            0.0152       1.0000
    LR test: independent vs. saturated:  chi2(28) = 1304.63 Prob>chi2 = 0.0000

Factor loadings (pattern matrix) and unique variances
        Variable |  Factor1   Factor2   Factor3   Factor4 |   Uniqueness 
    populist_l~d |   0.8871   -0.2900    0.0011    0.0209 |      0.1285  
        district |  -0.0184    0.1063   -0.3451    0.5929 |      0.5178  
        ideology |   0.5594   -0.6318   -0.0388   -0.0258 |      0.2858  
           union |   0.7074   -0.0433    0.5114    0.2527 |      0.1724  
    leader_par~t |   0.5344    0.4837   -0.4356   -0.1917 |      0.2539  
       e_machine |   0.5529    0.6013   -0.2825   -0.1882 |      0.2175  
          gender |  -0.0193    0.0504    0.3258   -0.6823 |      0.4254  
         gov_opp |   0.0800    0.6661    0.5986    0.2673 |      0.1202  
Scoring coefficients (method = regression)
        Variable |  Factor1   Factor2   Factor3   Factor4 
    populist_l~d |  0.40347  -0.18852   0.00100   0.02043 
        district | -0.00837   0.06909  -0.30922   0.57808 
        ideology |  0.25441  -0.41076  -0.03474  -0.02513 
           union |  0.32171  -0.02815   0.45818   0.24637 
    leader_par~t |  0.24306   0.31447  -0.39026  -0.18694 
       e_machine |  0.25147   0.39092  -0.25306  -0.18353 
          gender | -0.00879   0.03278   0.29189  -0.66528 
         gov_opp |  0.03636   0.43307   0.53631   0.26060