risk-assessment-frameworks

# Risk Assessment Frameworks Skill ## Purpose This skill provides comprehensive risk assessment methodologies for evaluating political, institutional, and democratic risks within the Swedish political system. It integrates international frameworks (V-Dem, Transparency International, Freedom House) with CIA platform's proprietary 50+ Drools risk rules to create systematic early warning capabilities for democratic backsliding, corruption, institutional erosion, political violence, and coalition instability. ## When to Use This Skill Apply this skill when: - ✅ Conducting democratic health assessments of Swedish institutions - ✅ Identifying early warning signs of institutional erosion - ✅ Assessing corruption risk at politician or party level - ✅ Evaluating coalition stability and government sustainability - ✅ Detecting democratic backsliding indicators - ✅ Measuring institutional accountability effectiveness - ✅ Analyzing political violence risk factors - ✅ Creating risk-based intelligence priorities - ✅ Benchmarking Sweden against international democracy standards - ✅ Generating risk reports for stakeholders and media Do NOT use for: - ❌ Political persecution or targeting of legitimate opposition - ❌ Fabricating risks to manipulate public opinion - ❌ Undermining democratic institutions through false alarms - ❌ Violating privacy or conducting surveillance without legal basis ## Risk Assessment Framework Architecture ### Integrated Risk Intelligence System The CIA platform integrates four layers of risk intelligence to create comprehensive political risk profiles: ```mermaid graph TB subgraph "Layer 1: Data Collection" A1["🗳️ Behavioral Data<br/>3.5M+ votes, attendance<br/>Productivity metrics"] A2["💰 Financial Data<br/>World Bank, ESV<br/>Economic indicators"] A3["📊 Democracy Indices<br/>V-Dem, Freedom House<br/>International benchmarks"] A4["📰 Media Coverage<br/>Sentiment analysis<br/>Scandal tracking"] end subgraph "Layer 2: Risk Rules Engine (Drools)" A1 --> B1[Behavioral Risk Rules<br/>24 politician rules<br/>12 party rules] A2 --> B2[Financial Risk Rules<br/>8 corruption indicators] A3 --> B3[Democratic Health Rules<br/>6 institutional rules] A4 --> B4[Reputational Risk Rules<br/>4 scandal detection rules] end subgraph "Layer 3: Risk Aggregation" B1 --> C1[Individual Risk Profiles] B2 --> C2[Institutional Risk Profiles] B3 --> C3[Systemic Risk Profiles] B4 --> C4[Reputational Risk Profiles] end subgraph "Layer 4: Risk Intelligence" C1 & C2 & C3 & C4 --> D["🎯 Composite Risk Score"] D --> E[Early Warning Alerts] D --> F[Risk Mitigation Strategies] D --> G[Intelligence Priorities] end style A1 fill:#e1f5ff style A2 fill:#e1f5ff style A3 fill:#e1f5ff style A4 fill:#e1f5ff style B1 fill:#fff9cc style B2 fill:#fff9cc style B3 fill:#fff9cc style B4 fill:#fff9cc style D fill:#ffe6cc style E fill:#ffcccc style F fill:#ccffcc style G fill:#e6ccff ``` ## 1. Democratic Backsliding Detection ### V-Dem Integration Framework The Varieties of Democracy (V-Dem) project provides the world's most comprehensive democracy measurement. The CIA platform integrates V-Dem indicators with behavioral data. **V-Dem Core Indicators Tracked:** - **Liberal Democracy Index** - Rule of law, checks on government - **Electoral Democracy Index** - Free and fair elections - **Participatory Democracy Index** - Citizen participation - **Deliberative Democracy Index** - Quality of public discourse - **Egalitarian Democracy Index** - Equal access to power ```python from typing import Dict, List, Tuple import pandas as pd import numpy as np from datetime import datetime, timedelta class DemocraticBackslidingDetector: """ Detects democratic backsliding through trend analysis and threshold monitoring. Based on V-Dem Early Warning of Democratic Decline (Edda) methodology and combines international indices with CIA platform behavioral data. """ # V-Dem backsliding thresholds (0-1 scale) CRITICAL_THRESHOLDS = { 'liberal_democracy_index': 0.50, # Below = autocratization 'electoral_democracy_index': 0.60, # Below = electoral manipulation 'participatory_democracy_index': 0.45, # Below = citizen disengagement 'deliberative_democracy_index': 0.50, # Below = discourse degradation 'egalitarian_democracy_index': 0.55 # Below = inequality deepening } def assess_democratic_health(self, country_code: str = 'SWE') -> Dict: """ Comprehensive democratic health assessment for Sweden. Combines: 1. V-Dem historical trends (5-year analysis) 2. CIA behavioral indicators (parliamentary effectiveness) 3. International comparison (Nordic benchmarking) 4. Early warning signals (acceleration detection) """ # Fetch V-Dem data vdem_query = """ SELECT year, v2x_libdem as liberal_democracy_index, v2x_polyarchy as electoral_democracy_index, v2x_partipdem as participatory_democracy_index, v2x_delibdem as deliberative_democracy_index, v2x_egaldem as egalitarian_democracy_index, -- Component indicators v2x_judicind as judicial_independence, v2x_frassoc_thick as freedom_association, v2x_freexp_altinf as freedom_expression, v2x_elecoff as elected_officials_index, v2xlg_legcon as legislative_constraints, v2x_corr as political_corruption_index, -- Backsliding indicators v2x_regime as regime_type FROM vdem_data WHERE country_code = %s AND year >= EXTRACT(YEAR FROM NOW()) - 10 ORDER BY year DESC """ vdem_df = pd.read_sql(vdem_query, self.connection, params=[country_code]) # Calculate trends (5-year linear regression slopes) trends = {} for column in vdem_df.columns: if column not in ['year', 'country_code', 'regime_type']: X = vdem_df['year'].values.reshape(-1, 1) y = vdem_df[column].values # Simple linear regression slope = np.polyfit(X.flatten(), y, 1)[0] trends[column] = round(slope, 4) # Fetch CIA behavioral indicators behavioral_query = """ SELECT -- Parliamentary effectiveness AVG(ce.overall_effectiveness_score) as avg_committee_effectiveness, -- Party discipline (inverse of deviation) AVG(100 - pd.avg_deviation_rate) as avg_party_discipline, -- Oversight activity COUNT(DISTINCT oa.document_id) as oversight_action_count, AVG(oa.oversight_effectiveness_score) as avg_oversight_effectiveness, -- Cross-party collaboration AVG(cpc.collaboration_intensity) as avg_cross_party_collaboration, -- Voting participation AVG(100 - vbs.avg_absent_percentage) as avg_participation_rate FROM committee_effectiveness ce, party_deviation pd, oversight_activity oa, cross_party_collaboration cpc, vote_ballot_summary vbs WHERE pd.analysis_date >= NOW() - INTERVAL '2 years' AND oa.created_date >= NOW() - INTERVAL '2 years' """ behavioral_data = pd.read_sql(behavioral_query, self.connection).iloc[0] # Current V-Dem scores current_vdem = vdem_df.iloc[0] # Identify risks risks = self._identify_risks(current_vdem, trends, behavioral_data) # Calculate composite democratic health score (0-100) health_score = self._calculate_health_score(current_vdem, behavioral_data) # Early warning assessment early_warnings = self._detect_early_warnings(trends, current_vdem) return { 'country': country_code, 'assessment_date': datetime.now().isoformat(), 'current_scores': { 'liberal_democracy': round(current_vdem['liberal_democracy_index'], 3), 'electoral_democracy': round(current_vdem['electoral_democracy_index'], 3), 'participatory_democracy': round(current_vdem['participatory_democracy_index'], 3), 'deliberative_democracy': round(current_vdem['deliberative_democracy_index'], 3), 'egalitarian_democracy': round(current_vdem['egalitarian_democracy_index'], 3) }, '5_year_trends': trends, 'behavioral_indicators': { 'committee_effectiveness': round(behavioral_data['avg_committee_effectiveness'], 2), 'party_discipline': round(behavioral_data['avg_party_discipline'], 2), 'oversight_effectiveness': round(behavioral_data['avg_oversight_effectiveness'], 2), 'cross_party_collaboration': round(behavioral_data['avg_cross_party_collaboration'], 3), 'participation_rate': round(behavioral_data['avg_participation_rate'], 2) }, 'composite_health_score': round(health_score, 2), 'health_classification': self._classify_health(health_score), 'identified_risks': risks, 'early_warnings': early_warnings, 'international_ranking': self._get_nordic_comparison(current_vdem) } def _identify_risks( self, current: pd.Series, trends: Dict, behavioral: pd.Series ) -> List[str]: """Identify specific democratic risks.""" risks = [] # Check V-Dem thresholds for indicator, threshold in self.CRITICAL_THRESHOLDS.items(): if current.get(indicator, 1.0) < threshold: risks.append( f"CRITICAL: {indicator} below threshold " f"({current[indicator]:.3f} < {threshold})" ) # Check negative trends for indicator, slope in trends.items(): if slope < -0.01: # Declining more than 0.01/year risks.append( f"WARNING: Declining {indicator} (trend: {slope:.4f}/year)" ) # Check behavioral indicators if behavioral['avg_committee_effectiveness'] < 50: risks.append("Institutional dysfunction: Low committee effectiveness") if behavioral['avg_oversight_effectiveness'] < 60: risks.append("Accountability deficit: Weak oversight mechanisms") if behavioral['avg_participation_rate'] < 85: risks.append("Disengagement: Low parliamentary participation") return risks if risks else ["No critical risks detected"] def _calculate_health_score( self, vdem: pd.Series, behavioral: pd.Series ) -> float: """Calculate composite democratic health score (0-100).""" # V-Dem component (70% weight) vdem_score = ( vdem['liberal_democracy_index'] * 20 + vdem['electoral_democracy_index'] * 20 + vdem['participatory_democracy_index'] * 10 + vdem['deliberative_democracy_index'] * 10 + vdem['egalitarian_democracy_index'] * 10 ) # Behavioral component (30% weight) behavioral_score = ( (behavioral['avg_committee_effectiveness'] / 100) * 10 + (behavioral['avg_oversight_effectiveness'] / 100) * 10 + (behavioral['avg_participation_rate'] / 100) * 10 ) return vdem_score * 100 + behavioral_score def _classify_health(self, score: float) -> str: """Classify democratic health.""" if score >= 85: return "ROBUST_DEMOCRACY" elif score >= 70: return "HEALTHY_DEMOCRACY" elif score >= 55: return "FLAWED_DEMOCRACY" elif score >= 40: return "HYBRID_REGIME" else: return "AUTOCRATIC_REGIME" def _detect_early_warnings( self, trends: Dict, current: pd.Series ) -> List[str]: """Detect early warning signals of democratic decline.""" warnings = [] # Accelerating decline (second derivative) declining_indicators = [k for k, v in trends.items() if v < -0.005] if len(declining_indicators) >= 3: warnings.append( "EARLY WARNING: Multiple indicators declining simultaneously" ) # Judicial independence warning if (current.get('judicial_independence', 1.0) < 0.70 or trends.get('judicial_independence', 0) < -0.01): warnings.append( "CRITICAL: Judicial independence erosion detected" ) # Freedom of expression warning if (current.get('freedom_expression', 1.0) < 0.75 or trends.get('freedom_expression', 0) < -0.01): warnings.append( "WARNING: Press freedom and expression declining" ) # Legislative constraints weakening if (current.get('legislative_constraints', 1.0) < 0.70 or trends.get('legislative_constraints', 0) < -0.01): warnings.append( "WARNING: Legislative oversight weakening" ) # Corruption increasing if trends.get('political_corruption_index', 0) > 0.01: warnings.append( "WARNING: Political corruption index increasing" ) return warnings if warnings else ["No early warnings detected"] def _get_nordic_comparison(self, current: pd.Series) -> Dict: """Compare Sweden to other Nordic countries.""" query = """ SELECT country_name, v2x_libdem as liberal_democracy_index FROM vdem_data WHERE country_code IN ('SWE', 'NOR', 'DNK', 'FIN', 'ISL') AND year = (SELECT MAX(year) FROM vdem_data) ORDER BY v2x_libdem DESC """ nordic_df = pd.read_sql(query, self.connection) sweden_rank = nordic_df[ nordic_df['country_name'] == 'Sweden' ].index[0] + 1 if 'Sweden' in nordic_df['country_name'].values else None return { 'nordic_ranking': f"{sweden_rank}/5" if sweden_rank else "N/A", 'regional_comparison': nordic_df.to_dict('records') } ``` ## 2. Corruption Risk Assessment ### Transparency International Integration The CIA platform integrates Transparency International's Corruption Perceptions Index (CPI) methodology with behavioral indicators to assess corruption risk. ```java @Service public class CorruptionRiskAnalyzer { /** * Multi-dimensional corruption risk assessment. * * Risk dimensions: * 1. Financial irregularities (unexplained wealth, conflict of interest) * 2. Behavioral anomalies (voting patterns inconsistent with stated positions) * 3. Network corruption (connections to sanctioned entities) * 4. Transparency violations (disclosure failures, opacity) * 5. Accountability evasion (oversight avoidance, question dodging) */ public CorruptionRiskProfile assessCorruptionRisk(String politicianId) { String sql = """ WITH financial_risk AS ( SELECT p.person_id, -- Financial disclosure completeness fd.disclosure_completeness_score, fd.wealth_change_unexplained_ratio, fd.conflict_of_interest_declarations, -- Red flags CASE WHEN fd.wealth_change_unexplained_ratio > 0.30 THEN 1 ELSE 0 END as wealth_anomaly_flag, CASE WHEN fd.disclosure_completeness_score < 0.70 THEN 1 ELSE 0 END as disclosure_failure_flag, CASE WHEN fd.conflict_of_interest_declarations = 0 AND fd.business_holdings > 0