Appendix for Agency proliferation and the globalization of the regulatory state: Introducing a data set on the institutional features of regulatory agencies Jacint Jordana, Xavier Fernández-i-Marín and Andrea C. Bianculli December 2018 Technical documentation of the article Agency proliferation and the globalization of the regulatory state: Introducing a data set on the institutional features of regulatory agencies published at Regulation & Governance. Description of the dataset Number of agencies by country Figure 1 shows the number of actual institutions (not number of sectors covered) in every country. There are a total of 17 sectors. Sector coverage by country Figure 2 displays the number of sectors covered by agencies in every sector (not the number of institutions). There are a total of 115 countries. Cluster analysis Number of clusters Figure 3 shows the model comparison of several different cluster analysis models based on EM for a parameterized Gaussian mixture models. The model comparison can be performed using the Bayesian Information Criterion (BIC). Specifying 6 clusters for the dataset is aligned with a number beyond which the gains in the Bayesian Information Criterion are almost negligible and the complexity of describing new clusters increases. Distribution of countries by cluster Figure 4 presents the proportion of agencies that, within each country, are classified in each of the clusters. Distribution of sector by cluster Figure 5 shows the proportion of agencies that, within each sectors, are classified in each of the clusters. Classification of countries Figure 6 presents the dendrogram with the classificacion of countries based on the means of the institutions within country to calculate the country position in each of the dimensions. The dendrogram is based on hierarchical cluster analysis with the Ward agglomerative method and with the distance matrix calculated using Euclidean distances.
Classification of sectors Figure 7 shows the dendrogram with the classificacion of sectors based on using means of the institutions within sector to calculate the sector position in each of the dimensions. The dendrogram is based on hierarchical cluster analysis with the Ward agglomerative method and with the distance matrix calculated using euclidean distances.
Mixed Factor analysis The measurement model is based on mixed factor analysis. This sections presents the JAGS code for the dimension with more variables (namely, Political independence). It is equal for all dimensions, except that for other dimensions not all parts continuous, binary and ordinal are necessary. 1 model { 2 # Measurement part 3 4 for (o in 1:O) { # Observations 5 # -- Binary variables 6 for (ib in 1:I.binary) { 7 Y.binary[o,ib] ~ dbern(pi[o,ib]) 8 logit(pi[o,ib]) <- mu[o,ib] 9 mu[o,ib] <- delta[ib,1] * (xi[o] - delta[ib,2]) 10 } 11 # -- Continuous variables 12 for (ic in 1:I.continuous) { 13 Y.continuous[o,ic] ~ dnorm(mu.continuous[o,ic], tau.continuous[ic]) 14 mu.continuous[o,ic] <- gamma[ic,1] + (xi[o] * gamma[ic,2]) 15 } 16 17 for (io in 1:I.ordinal) { 18 nu[o,io] <- lambda[io,1] + (xi[o] * lambda[io,2]) 19 } 20 21 # ------------ 22 # -- Ordinal variables 23 # -- A loop is not possible, as the number of categories is not always the same 24 25 Y.ordinal[o,1] ~ dcat(p1[o,]) # first categorical variable 26 logit(q1[o,1]) <- alpha1[1] - nu[o,1] 27 p1[o,1] <- Q1[o,1] 28 for (j in 2:3) { 29 logit(q1[o,j]) <- alpha1[j] - nu[o,1] 30 p1[o,j] <- Q1[o,j] - Q1[o,(j-1)] 31 } 32 p1[o,4] <- 1 - Q1[o,3] 33 34 # ------------... until all categorical variables are expressed 35 36 } 37 alpha1 <- sort(alpha1.0) 38 # Priors over thresholds for ordinal variables 39 alpha1.0 [1] ~ dnorm(-3, 4) 40 alpha1.0 [2] ~ dnorm(-1, 4) 41 alpha1.0 [3] ~ dnorm(1, 4) 42 # -----... untill all categories are expressed 43 44 45 # Priors for measurement part - binary 46 for (ib in 1:I.binary) { 47 delta[ib,1] ~ dnorm(0, 1)T(0,) 48 delta[ib,2] ~ dnorm(0, 1) 49 } 50 51 # Priors for measurement part - continuous 52 for (ic in 1:I.continuous) { 53 gamma[ic,1] ~ dnorm(0, 1) 54 gamma[ic,2] ~ dnorm(0, 1)T(0,) 55 } 56 57 for (ic in 1:I.continuous) { 58 tau.continuous[ic] ~ dt(0, pow(0.1, -2), 1)T(0,) 59 sigma.continuous[ic] <- 1/sqrt(tau.continuous[ic]) 60 } 61 62 # Priors for measurement part - ordinal 63 for (io in 1:I.ordinal) { 64 lambda[io,1] ~ dnorm(0, 1) 65 lambda[io,2] ~ dnorm(0, 1)T(0,) 66 } 67 68 # Priors for scores of observations 69 for (o in 1:O) { 70 xi[o] ~ dnorm(0, 1) 71 } 72 }
The factor loadings and discrimination parameters are shown in Figure 8
Country India United Kingdom Chile Brazil Austria Slovakia Romania Portugal Norway Australia Venezuela, Bolivarian Republic of Sweden Slovenia Mexico Italy Ireland Ghana France Finland China Argentina United States Switzerland Peru Nigeria Malta Lithuania Dominican Republic Denmark Colombia Bolivia, Plurinational State of Belgium Zambia Spain New Zealand Netherlands Iceland Hungary Greece Germany Estonia Bulgaria Zimbabwe United Arab Emirates Russian Federation Philippines Panama Pakistan Luxembourg Latvia Kenya Japan Egypt Czech Republic Costa Rica Canada Uganda Turkey Tanzania, United Republic of Sri Lanka South Africa Singapore Poland Hong Kong Ecuador Bangladesh Uzbekistan Ukraine Thailand Senegal Paraguay Nicaragua Malaysia Korea, Republic of Indonesia Ethiopia El Salvador Cyprus Tunisia Taiwan, Province of China Saudi Arabia Mozambique Morocco Mali Israel Iraq Iran, Islamic Republic of Algeria Viet Nam Malawi Kazakhstan Honduras Guatemala Cuba Cameroon Angola Afghanistan Yemen Uruguay Sudan Niger Nepal Congo, the Democratic Republic of the Chad Burkina Faso Syrian Arab Republic Kuwait Haiti Côte d'ivoire Brunei Darussalam Myanmar Madagascar Lao People's Democratic Republic Korea, Democratic People's Republic of Cambodia 0 5 10 15 Number of agencies Figure 1: Number of regulatory agencies (institutions) in each country.
Sector Central Bank Financial Services Telecommunications Securities and Exchange Electricity Insurance Nuclear Safety Competition Gas Pharmaceuticals Pensions Food Safety Postal Services Water Environment Health Services Work Safety 0 30 60 90 120 Number of agencies Figure 2: Coverage of regulatory agencies in each sector. -6200-6400 Figure 3: Bayesian Information Criterion (BIC) for several hierarchical cluster analysis models based on EM for a parameterized Gaussian mixture model. -6600 BIC -6800-7000 -7200 EII VII EEI EVI EEE EVE VEE VVE EEV VEV EVV VVV 1 2 3 4 5 6 7 8 9 Number of components
Country Afghanistan Algeria Angola Argentina Australia Austria Bangladesh Belgium Bolivia, Plurinational State of Brazil Brunei Darussalam Bulgaria Burkina Faso Cambodia Cameroon Canada Chad Chile China Colombia Congo, the Democratic Republic of the Costa Rica Cuba Cyprus Czech Republic Côte d'ivoire Denmark Dominican Republic Ecuador Egypt El Salvador Estonia Ethiopia Finland France Germany Ghana Greece Guatemala Haiti Honduras Hong Kong Hungary Iceland India Indonesia Iran, Islamic Republic of Iraq Ireland Israel Italy Japan Kazakhstan Kenya Korea, Democratic People's Republic of Korea, Republic of Kuwait Lao People's Democratic Republic Latvia Lithuania Luxembourg Madagascar Malawi Malaysia Mali Malta Mexico Morocco Mozambique Myanmar Nepal Netherlands New Zealand Nicaragua Niger Nigeria Norway Pakistan Panama Paraguay Peru Philippines Poland Portugal Romania Russian Federation Saudi Arabia Senegal Singapore Slovakia Slovenia South Africa Spain Sri Lanka Sudan Sweden Switzerland Syrian Arab Republic Taiwan, Province of China Tanzania, United Republic of Thailand Tunisia Turkey Uganda Ukraine United Arab Emirates United Kingdom United States Uruguay Uzbekistan Venezuela, Bolivarian Republic of Viet Nam Yemen Zambia Zimbabwe 0.25 0.50 0.75 1.00 Proportion Cluster 1 2 3 4 5 6 Figure 4: Proportion of agencies that, within each country, are classified in each of the clusters.
Central Bank Telecommunications Postal Services Figure 5: Proportion of agencies that, within each sectors, are classified in each of the clusters. Pensions Insurance Sector Gas Electricity Work Safety Water Securities and Exchange Pharmaceuticals Nuclear Safety and Radiological Protection Cluster 1 2 3 4 5 6 Health Services Food Safety Financial Services Environment Competition 0.0 0.2 0.4 0.6 Proportion Italy Uruguay Cyprus Madagascar Nicaragua Malaysia Panama El Salvador Guatemala Honduras Malawi Argentina Bulgaria Mexico United Kingdom Canada United States Mozambique Uganda Thailand Haiti Tanzania, United Republic of Lao People's Democratic Republic Slovenia South Africa Dominican Republic Venezuela, Bolivarian Republic of Kenya Malta Indonesia Philippines Turkey Ghana Paraguay Nigeria Singapore Norway Colombia Peru Iceland Czech Republic Germany Portugal Romania Greece Finland Latvia Ecuador Poland Bolivia, Plurinational State of Hungary Chile Spain France Sweden Cambodia India Switzerland Austria Luxembourg Ireland Denmark Costa Rica Netherlands Zimbabwe Belgium Israel Lithuania Brazil Australia Japan New Zealand Korea, Democratic People's Republic of Cuba Uzbekistan Saudi Arabia Korea, Republic of Côte d'ivoire Cameroon Mali Niger Ethiopia Afghanistan Yemen Iran, Islamic Republic of Syrian Arab Republic Kazakhstan Hong Kong Russian Federation Burkina Faso Bangladesh Sudan Ukraine Tunisia Estonia Morocco Slovakia Viet Nam Chad United Arab Emirates Taiwan, Province of China Congo, the Democratic Republic of the Kuwait Myanmar Brunei Darussalam Angola China Zambia Iraq Egypt Senegal Sri Lanka Pakistan Nepal Algeria 0 10 20 30 Figure 6: Dendrogram for country means.
Water Competition Insurance Pensions Electricity Telecommunications Gas Postal Services Central Bank Financial Services Securities and Exchange Nuclear Safety and Radiological Protection Environment Work Safety Food Safety Health Services Pharmaceuticals 0 1 2 3 4 5 Figure 7: Dendrogram for sector means.
Variable Personnel policy Personal status Organizational structure Budget income Budget control Budget approval Holding of ces in government Current legal instrument Agency head term of of ce Agency head renewal Agency head professional requirement Agency head dismissal Agency head appointment Agency board members term of of ce Agency board members renewal Agency board members professional requirement Agency board members dismissal Agency board members appointment Resolutions are online Minutes are online Civil society accountability - Public hearings Civil society accountability - Other Civil society accountability - Open consultations Civil society accountability - Consumers' of ce Civil society accountability - Advisory council Annual reports are online Accountable to the parliament Accountable to the ministry Accountable to the executive Revision decisions - President - Prime Minister Revision decisions - Parliament Revision decisions - Other regulatory agency Revision decisions - None Revision decisions - Minister Revision decisions - Judiciary Regulatory competencies in the sector Capacity to supervise Capacity to implement sanctions Capacity to establish prices Capacity to establish entries and exits of the market Capacity to elaborate norms Capacity to do research Capacity to do con ict resolution Factor loadings (γ, λ) and discriminations (δ) -2 0 2 4 Posterior median, 90% and 95% Highest Posterior Density Managerial autonomy Political Autonomy Public accountability Regulatory capabilities Figure 8: Factor loadings (continuous and ordinal variables) and discriminations (binary variables) for every variable in the measurement model.