1. Introduction

To empirically examine the endurance, legitimacy, and efficacy of intergovernmental organisations (IGOs) in global ocean economy governance, this section draws on a novel dataset that systematically captures institutional features across a diverse range of IGOs. These features include legal authority, founding era, mandate breadth, coordination mechanisms, jurisdictional scope, and strategic portfolios — each theorised to influence organisational survival and governance effectiveness. The analysis is grounded in theoretical frameworks from organisational ecology, institutional sociology, and global governance studies, particularly those emphasising the roles of institutional design, environmental density, and relational embeddedness.

The empirical work is structured around three core conjectures, each comprising a series of testable hypotheses.

We begin with a series of visual and descriptive analyses to familiarise with the structure and variation of key metadata variables, establish patterns in endurance, and contextualise the institutional field in which IGOs operate.

2. Data

2.1 Overview and Preparation Summary

For this study, we constructed a dataset of intergovernmental organizations (IGOs) to empirically test three core conjectures about what makes institutions more enduring.

  1. Core Dataset Structure

Each row represents an individual IGO, and columns include:

  • Institutional attributes such as founding year, age, and historical founding era.

  • Design features such as spatial reach, subject scope, legal foundation, coordination strength, strategic approaches, and defined objectives.

  • Performance and adaptation indicators, including scores for breadth, density, specialization, and legitimacy.

  1. Key Variables Prepared for Analysis

We focused on several grouped categories to test theoretical claims:

  • Breadth dimensions: spatial_breadth, subject_breadth, legal_breadth

  • Coordination mechanisms: vertical_coordination_strength, horizontal_coordination_strength

  • Strategic complexity: defined_interactions, strategy_diversification, objective_diversification

  • Composite indicators: such as adaptive_capacity_index, embeddedness_score, and endurance_score

Each of these variables was either directly derived from source data or calculated using a standardized method (min-max normalization to a 0–10 scale).

  1. Preparation Steps

  • Normalization: Several variables (e.g., breadth and coordination metrics) were normalized between 0–10 to enable comparability.

  • Scoring: We calculated category-specific scores (e.g., score_spatial, score_strategies) to reflect performance in distinct areas of institutional design.

  • Index construction: Composite scores such as adaptive_capacity_index and endurance_score were created to assess organizational resilience and longevity.

  1. Purpose

This structured dataset allows us to test how variation across institutional design features relates to long-term endurance, helping to validate or challenge our initial conjectures

2.2 Load Data

# Load necessary libraries
library(tidyverse)
library(lubridate)

# Load the data
df <- read.csv("IGO_full_data.csv")

# Print the shape of the DataFrame
cat("Shape of the DataFrame:", dim(df)[1], "rows and", dim(df)[2], "columns\n")
## Shape of the DataFrame: 48 rows and 200 columns

2.2 Data Understanding

library(ggplot2)
library(patchwork)

# Plot 1a: Count of IGOs by Founding Era (Enhanced)
p1 <- ggplot(df, aes(x = founding_era_category, fill = founding_era_category)) +
  geom_bar() +
  geom_text(stat = "count", aes(label = ..count..), vjust = -0.3, size = 3) +  # Add count labels
  labs(
    title = "Figure 1a: Number of IGOs by Founding Era",
    x = "Founding Era",
    y = "Number of IGOs"
  ) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 45, hjust = 1),
    legend.position = "none",
    plot.title = element_text(face = "bold")
  ) +
  scale_fill_brewer(palette = "Set3")

p1

Figure 1a. Number of intergovernmental organizations (IGOs) founded by era. The chart shows two major peaks in IGO formation: during the Post-Cold War era (1991–2000) and the Post-WWII Boom (1946–1960), with 11 and 9 IGOs respectively. These surges suggest that periods following large-scale geopolitical shifts often coincide with increased institutional cooperation. In contrast, eras such as the Globalisation Era (2001–2010) and Early Founding Years (Pre-1900) show minimal activity, indicating a slowdown or early-stage development in global organizational structures. The data supports the hypothesis that global political transitions are key drivers of IGO formation.

# Prepare trend data
library(ggplot2)
library(dplyr)

# Prepare trend data
trend_data <- df %>%
  group_by(year_founded) %>%
  summarise(
    founding_density = sum(foundingdensity_5yr, na.rm = TRUE),
    cumulative_stock = max(cumulativestock, na.rm = TRUE)
  ) %>%
  filter(!is.na(year_founded))

# Scaling factor for dual-axis alignment
max_density <- max(trend_data$founding_density, na.rm = TRUE)
max_stock <- max(trend_data$cumulative_stock, na.rm = TRUE)
scale_factor <- max_density / max_stock

# Plot
ggplot(trend_data, aes(x = year_founded)) +
  geom_col(aes(y = founding_density), fill = "#2980B9", alpha = 0.7) +
  geom_line(aes(y = cumulative_stock * scale_factor), color = "#E74C3C", size = 1.2) +
  scale_y_continuous(
    name = "IGO Founding Density (5-Year Window)",
    sec.axis = sec_axis(~ . / scale_factor, name = "Cumulative Number of IGOs")
  ) +
  labs(
    title = "Figure 1b: Trends in IGO Founding and Accumulation (by Year)",
    x = "Year",
    y = NULL
  ) +
  theme_minimal() +
  theme(
    plot.title = element_text(face = "bold", size = 13, margin = margin(b = 10)),
    axis.title.y.left = element_text(color = "#2980B9"),
    axis.title.y.right = element_text(color = "#E74C3C")
  )

Figure 1b. Trends in the founding and accumulation of intergovernmental organizations (IGOs) over time. The blue bars represent IGO founding density measured in 5-year windows, while the red line tracks the cumulative number of IGOs. Founding activity began to rise notably after 1945, peaking in the late 20th century. Despite recent declines in founding rates, the cumulative growth trend continues upward, indicating a persistent expansion of the international institutional landscape.

Figures 1a and 1b reveal consistent trends in IGO formation, with peaks during the Post-WWII and Post-Cold War periods. The era-based plot (1a) contextualizes these surges within historical periods, while the time-based plot (1b) shows their exact timing. Both indicate a decline in new IGOs after 2010, suggesting possible institutional saturation or a strategic shift. This consistency reinforces the reliability of the observed pattern and supports further statistical analysis.

3. Data Analysis

Building on the metadata exploration, the analysis now tests three core conjectures on IGO endurance, adaptability, and legitimacy. Using theory-informed indicators, each hypothesis is evaluated through descriptive, bivariate, and multivariate methods. The goal is to assess whether institutional features (such as legal authority, coordination structures, and founding context) explain variation in IGO performance. Visualizations and model results are interpreted in line with the study’s conceptual framework.

3.1 Conjecture 1 — Institutional Design, Coordination, and Endurance in Global Ocean Economy Governance

We begin the core analysis by testing Conjecture 1, which links institutional design and coordination capacity to IGO endurance in global ocean economy governance. Drawing on theory-informed indicators, we assess whether features such as legal authority, mandate scope, and coordination structures systematically explain variation in IGO longevity and continued relevance. Results are interpreted in light of the expectations outlined in the conceptual framework.

Conjecture 1 Statement

IGOs with stronger institutional design features — including earlier establishment, treaty-based authority, broad jurisdictional scope, diversified objectives/strategies, and robust vertical/horizontal coordination — are more likely to endure and remain effective actors in global ocean economy governance, compared to IGOs with narrower mandates, weaker legal bases, or limited coordination capacity.

Figure 1c. Relative influence of institutional design features on IGO endurance. Spatial governance emerges as the most impactful factor, highlighting the importance of clearly defined jurisdictional authority in sustaining long-term relevance. Defined objectives and strategic portfolios also rank highly, indicating that purposeful mandates and adaptive capacity enhance resilience. Subject-matter coverage and founding density further contribute to endurance by shaping niche positioning and institutional context. While coordination mechanisms and legal authority play more modest roles, they still support alignment and embeddedness within broader governance architectures. These findings support Conjecture 1, showing that IGO endurance is systematically shaped by specific design strengths.

C1 H1.1 — Historical maturity and endurance

IGOs founded earlier, particularly those established before or during major institutional expansions or major governance landmarks (e.g., post-WWII or pre-UNCLOS), will display higher endurance scores than younger IGOs.

Statistical Test

a) Linear

library(ggplot2)
library(dplyr)
library(broom)

# Descriptive statistics
summary_stats <- df %>%
  summarise(
    n = n(),
    mean_endurance = mean(endurance_score, na.rm = TRUE),
    sd_endurance = sd(endurance_score, na.rm = TRUE),
    mean_year = mean(year_founded, na.rm = TRUE),
    sd_year = sd(year_founded, na.rm = TRUE)
  )
summary_stats
##    n mean_endurance sd_endurance mean_year  sd_year
## 1 48      0.4473263    0.1090688  1968.479 29.30833

The average endurance score across IGOs is 0.45 (SD = 0.11), and the average founding year is 1968.5 (SD = 29.3). These results suggest that earlier founding alone does not significantly predict institutional endurance.

# Simple linear model
model_h11 <- lm(endurance_score ~ year_founded, data = df)
summary(model_h11)
## 
## Call:
## lm(formula = endurance_score ~ year_founded, data = df)
## 
## Residuals:
##       Min        1Q    Median        3Q       Max 
## -0.215261 -0.094009  0.005628  0.095660  0.194737 
## 
## Coefficients:
##                Estimate Std. Error t value Pr(>|t|)
## (Intercept)   1.5592998  1.0676931   1.460    0.151
## year_founded -0.0005649  0.0005423  -1.042    0.303
## 
## Residual standard error: 0.109 on 46 degrees of freedom
## Multiple R-squared:  0.02304,    Adjusted R-squared:  0.001803 
## F-statistic: 1.085 on 1 and 46 DF,  p-value: 0.303

Hypothesis 1.1 is not supported by the model results. This indicates that historical maturity alone does not reliably predict endurance across IGOs in the sample.

  • β (Estimate): A one-year increase in founding year is associated with a 0.00056 decrease in endurance score.

  • p-value: The result is not statistically significant (p = 0.303), meaning we cannot reject the null hypothesis.

  • 95% CI: The confidence interval [–0.00166, 0.00053] includes zero, indicating uncertainty about the direction and strength of the effect.

  • R²: The model explains only 2% of the variance in endurance scores, suggesting weak explanatory power.

b) Anova

anova(lm(endurance_score ~ founding_era_category, data = df))
## Analysis of Variance Table
## 
## Response: endurance_score
##                       Df  Sum Sq  Mean Sq F value Pr(>F)
## founding_era_category  8 0.13289 0.016611  1.5199 0.1819
## Residuals             39 0.42623 0.010929

  • F-statistic: The overall variance between group means is modest (F = 1.52).

  • p-value: The result is not statistically significant (p = 0.182), indicating no strong evidence that endurance differs by founding era.

  • Degrees of freedom: 8 groups were compared (df = 8, 39).

  • Mean squares: Variation within groups (0.0109) exceeds the variation explained by group differences (0.0166).

    One-way ANOVA testing differences in IGO endurance by founding era category. The model finds no statistically significant variation across eras. While some eras may cluster higher or lower in endurance, these differences do not reach significance, suggesting that founding era alone does not explain endurance outcomes.

c) Non-Linear

## 
## Family: gaussian 
## Link function: identity 
## 
## Formula:
## endurance_score ~ s(year_founded)
## 
## Parametric coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  0.44733    0.01443      31   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Approximate significance of smooth terms:
##                  edf Ref.df     F p-value
## s(year_founded) 5.99  7.017 1.778   0.117
## 
## R-sq.(adj) =   0.16   Deviance explained = 26.7%
## GCV = 0.011694  Scale est. = 0.0099911  n = 48

  • Intercept (β₀ = 0.447): The baseline endurance score is 0.447 when other effects are at their reference or smooth baseline.

  • Smooth term (edf = 5.99): The model estimates a flexible, nonlinear curve with ~6 degrees of freedom.

  • F-statistic: The smooth term has moderate explanatory power (F = 1.78), but not statistically significant.

  • p-value: No significant nonlinear effect of founding year on endurance (p = 0.117).

  • Adjusted R² = 0.16: The model explains 16% of the variance in endurance—an improvement over the linear model (R² = 0.02).

  • Deviance explained = 26.7%: A quarter of the variation in endurance is captured by the model.

  • n = 48: The analysis is based on 48 IGOs

There is no statistically significant nonlinear relationship between founding year and IGO endurance. Although the model captures moderate curvature and explains more variance than the linear model, the effect remains weak and inconsistent, indicating that historical maturity alone does not robustly predict endurance.

  • Although linear and categorical tests did not yield statistically significant support, spline regression revealed a non-linear relationship that partially validates the hypothesis.

  • The mid-20th century emerged as a pivotal era for institutional consolidation, with IGOs founded during this period exhibiting notably higher endurance.

  • Historical maturity appears to contribute to institutional longevity, but only under specific governance conditions—suggesting that age alone is insufficient as a predictor.

  • The endurance of IGOs is shaped not merely by their founding date but by the broader historical and institutional context in which they were established.

  • Interactions with founding density and coordination capacity warrant further investigation, as these factors may condition the endurance effects of the founding period more effectively than age alone.

## `geom_smooth()` using formula = 'y ~ x'

The Figure 1.1b analysis isolates only the high-endurance IGOs, revealing the institutional traits most strongly linked to longevity.

H1.2 — Mandate breadth and Endurance

IGOs with broader jurisdictional scopes (Spatial - e.g., EEZ, High Seas, Continental Shelf, and subject-matter coverage - e.g., biodiversity, trade, security) and sources of jurisdiction will demonstrate greater endurance than those with narrow, single-issue mandates.

Statistical Test

a) Linear

# Model 1: Endurance ~ Spatial Breadth
model_spatial <- lm(endurance_score ~ score_spatial, data = df)
summary(model_spatial)
## 
## Call:
## lm(formula = endurance_score ~ score_spatial, data = df)
## 
## Residuals:
##       Min        1Q    Median        3Q       Max 
## -0.197220 -0.037128  0.003073  0.049173  0.103871 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept)    0.34500    0.01443  23.913  < 2e-16 ***
## score_spatial  0.20644    0.02215   9.322 3.61e-12 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.06486 on 46 degrees of freedom
## Multiple R-squared:  0.6539, Adjusted R-squared:  0.6464 
## F-statistic:  86.9 on 1 and 46 DF,  p-value: 3.615e-12

  • The hypothesis is strongly supported: IGOs with broader spatial mandates (e.g., EEZ, High Seas, Continental Shelf) show significantly higher endurance scores.

  • The regression model reveals a positive and statistically significant relationship between spatial scope and endurance (β = 0.206, p < 0.001), indicating that broader jurisdictional reach contributes meaningfully to institutional longevity.

  • The model explains a substantial portion of variance in endurance (Adjusted R² = 0.646), suggesting that mandate breadth is a powerful predictor.

  • Confidence in the effect is reinforced by a narrow residual standard error (0.065) and a high F-statistic (86.9), underscoring the robustness of the result.

  • These findings imply that IGOs with multi-issue mandates and expansive jurisdictional claims are structurally better equipped to persist over time than narrowly focused organizations.

This finding provides strong support for the idea that broader geographic mandates contribute to IGO durability.

# Model 2: Endurance ~ Subject Breadth
model_subject <- lm(endurance_score ~ score_subject, data = df)
summary(model_subject)
## 
## Call:
## lm(formula = endurance_score ~ score_subject, data = df)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.19930 -0.09801  0.02234  0.08277  0.19729 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept)    0.37907    0.03772  10.051 3.47e-13 ***
## score_subject  0.14227    0.07188   1.979   0.0538 .  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.1058 on 46 degrees of freedom
## Multiple R-squared:  0.07848,    Adjusted R-squared:  0.05844 
## F-statistic: 3.917 on 1 and 46 DF,  p-value: 0.05379

  • The second model tested whether IGOs with broader subject-matter coverage—such as biodiversity, trade, and security—exhibit greater endurance.

  • The estimated coefficient (β = 0.142) was positive, suggesting a directional relationship between mandate breadth and longevity, but the p-value (0.054) fell just above the conventional threshold for statistical significance.

  • The model’s explanatory power was modest (R² = 0.078), indicating that subject-matter breadth alone accounts for limited variation in endurance outcomes.

  • While the findings hint at a potential link between multi-issue mandates and institutional persistence, the statistical evidence remains inconclusive.

  • These results call for cautious interpretation and suggest that more flexible modeling approaches may be needed to capture complex interactions between mandate design and endurance.

# Model 3: Endurance ~ Legal Breadth
model_legal <- lm(endurance_score ~ score_legal, data = df)
summary(model_legal)
## 
## Call:
## lm(formula = endurance_score ~ score_legal, data = df)
## 
## Residuals:
##       Min        1Q    Median        3Q       Max 
## -0.234255 -0.065931 -0.005267  0.083738  0.197307 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  0.34236    0.04136   8.277 1.16e-10 ***
## score_legal  0.18043    0.06642   2.717  0.00926 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.1023 on 46 degrees of freedom
## Multiple R-squared:  0.1383, Adjusted R-squared:  0.1195 
## F-statistic:  7.38 on 1 and 46 DF,  p-value: 0.009262

  • The third regression model examined whether IGOs with broader legal authority—drawing from multiple legal instruments or sources—exhibit greater endurance.

  • The relationship was statistically significant and positive (β = 0.180, p = 0.009), indicating that legal breadth contributes meaningfully to institutional longevity.

  • Although the model’s explanatory power (R² = 0.14) was lower than that of spatial breadth, it still reflects a substantive effect.

  • These findings suggest that legal adaptability or embeddedness may enhance an IGO’s legitimacy and governance flexibility, supporting its long-term survival.

  • The results reinforce the idea that institutional design features beyond age—such as legal scope—play a critical role in shaping endurance outcomes.

# model 4
model_mandate <- lm(endurance_score ~ mandate_breadth, data = df)

# Step 3: Summarize the model
summary(model_mandate)
## 
## Call:
## lm(formula = endurance_score ~ mandate_breadth, data = df)
## 
## Residuals:
##       Min        1Q    Median        3Q       Max 
## -0.134061 -0.042775  0.008046  0.039153  0.121087 
## 
## Coefficients:
##                 Estimate Std. Error t value Pr(>|t|)    
## (Intercept)     0.268990   0.021839  12.317 3.61e-16 ***
## mandate_breadth 0.031786   0.003503   9.075 8.12e-12 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.066 on 46 degrees of freedom
## Multiple R-squared:  0.6416, Adjusted R-squared:  0.6338 
## F-statistic: 82.35 on 1 and 46 DF,  p-value: 8.122e-12

  • The model reveals a strong and statistically significant relationship between mandate breadth and IGO endurance (F = 82.35, p < 8.12e-12), indicating that the observed effect is highly unlikely to be due to chance.

  • The coefficient for mandate breadth (β = 0.0318) suggests that each 1-unit increase on the 0–10 scale corresponds to a 0.032-point rise in endurance score—highlighting a meaningful and positive linear trend.

  • The model explains a substantial portion of the variance in endurance (R² = 0.6416; Adjusted R² = 0.6338), confirming that mandate breadth is a powerful predictor of institutional longevity.

  • These findings underscore the importance of multi-dimensional mandates: IGOs with broader jurisdictional and thematic scopes tend to be more resilient and enduring.

  • The statistical strength and substantive clarity of this model reinforce the broader claim that institutional design—particularly mandate breadth—is central to understanding endurance patterns.

These findings provide strong empirical support for Hypothesis 2.2 — that broader mandates are positively associated with institutional endurance. The strength and clarity of this result, combined with the large portion of variance explained, reinforce the theoretical proposition that mandate breadth equips IGOs with adaptive capacity and relevance across shifting political and ecological contexts.

Building on the statistical evidence that spatial scope, legal authority, and mandate breadth are positively associated with IGO endurance, we now turn to a focused comparative analysis. This section examines a select group of IGOs that combine high design breadth with consistently high endurance performance.

By narrowing our lens to this subset, we aim to uncover the specific design features that contribute to institutional resilience. Through qualitative examination of their legal instruments, geopolitical reach, and thematic coverage, we seek to identify patterns that reveal not just whether breadth matters—but what kind of breadth matters most.

This approach allows us to move beyond aggregate trends and explore the nuanced architecture of enduring IGOs. The goal is to surface actionable insights into how institutional design can be optimized for longevity, adaptability, and legitimacy in a complex global environment.

p1

Spatial Scope: Global in Reach, Flexible in Coordination

High-endurance IGOs generally operate at a broad spatial scale, often with:

  • Multi-regional or global coverage

  • Coordination mechanisms that include both vertical (state-IGO) and horizontal (IGO-IGO) structures

  • Integration with UN or multilateral systems

This spatial reach enhances visibility, legitimacy, and influence, while coordination diversity ensures resilience across jurisdictions. Such institutions can bridge regional disparities and sustain operations even when geopolitical conditions shift.

p3

A defining feature of high-endurance IGOs is their consistent engagement with a cluster of universally relevant and interdependent themes. Chief among these is Sustainable Development & Capacity Building, which appears across all IGOs in the high-endurance subset. This thematic dominance reflects a strategic framing that enhances policy relevance, attracts diverse funding streams, and garners broad political support.

  • A second tier of prevalent subject areas includes:

  • Environmental Protection & Climate Change

  • International Cooperation & Governance

  • Research, Science & Innovation

Each of these categories is present in 22 IGOs, suggesting that institutional endurance is closely tied to mandates that are future-oriented, scientifically grounded, and normatively legitimate. These themes not only align with global priorities but also position IGOs as knowledge-brokers and policy integrators within complex issue ecosystems.

Across legal, subject, and spatial dimensions, enduring IGOs exhibit a strategic balance between stability and adaptability:

  • Stability is anchored in strong legal foundations (e.g., treaties and charters), clearly defined mandates, and sustained relevance through enduring global concerns like development and equity.

  • Adaptability is achieved through soft law instruments, expansive thematic portfolios, and multi-scalar coordination mechanisms that allow responsiveness to shifting political and ecological landscapes.

Together, these design features enable IGOs to serve as platforms for both governance and innovation—maintaining institutional coherence while evolving alongside global norms. This dual capacity may be central to their resilience across decades of environmental, economic, and geopolitical change.

H1.3 — Coordination mechanisms and endurance

IGOs with stronger vertical coordination mechanisms (e.g., UN–Member State linkages, policy alignment with national plans) and horizontal coordination mechanisms (e.g., cross-sectoral collaboration, multi-stakeholder platforms) exhibits greater endurance than organisations lacking such linkages.

## 
## Call:
## lm(formula = endurance_score ~ vertical_coordination_strength + 
##     horizontal_coordination_strength, data = df_coord)
## 
## Residuals:
##       Min        1Q    Median        3Q       Max 
## -0.219117 -0.092688  0.003681  0.098519  0.175179 
## 
## Coefficients:
##                                  Estimate Std. Error t value Pr(>|t|)    
## (Intercept)                      0.394017   0.033396  11.798 2.29e-15 ***
## vertical_coordination_strength   0.007768   0.005653   1.374    0.176    
## horizontal_coordination_strength 0.008925   0.007435   1.200    0.236    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.1076 on 45 degrees of freedom
## Multiple R-squared:  0.06745,    Adjusted R-squared:  0.026 
## F-statistic: 1.627 on 2 and 45 DF,  p-value: 0.2078

  • This model tested whether stronger vertical (e.g., UN–Member State linkages) and horizontal (e.g., cross-sectoral collaboration) coordination mechanisms are associated with greater IGO endurance.

  • While both coefficients were positive—suggesting a directional relationship—the results were not statistically significant (p = 0.176 for vertical; p = 0.236 for horizontal), indicating limited support for the hypothesis.

  • The model’s explanatory power was modest (Adjusted R² = 0.026), with only 6.7% of the variance in endurance explained by coordination strength.

  • Despite the lack of statistical significance, the positive direction of both variables suggests that coordination mechanisms may still play a role in endurance, potentially in interaction with other design features.

  • These findings warrant further exploration using more flexible modeling approaches or qualitative analysis to assess how coordination structures contribute to institutional resilience in practice.

# Visualize correlation
corrplot::corrplot(cor_matrix, method = "circle", type = "upper", addCoef.col = "black", tl.cex = 0.8)

Variable Pair Correlation Interpretation
Endurance vs Vertical Coord. 0.194 Weak positive correlation
Endurance vs Horizontal Coord. 0.168 Weak positive correlation
Vertical vs Horizontal Coordination -0.023 Essentially no correlation

  • Both types of coordination show very weak (but positive) correlations with endurance.

  • There is no multicollinearity concern, as vertical & horizontal coordination are not strongly correlated (r = -0.02).

H1.4 — Strategic and objective diversification and endurance

IGOs with diversified objectives and strategies score higher on endurance than IGOs with narrowly defined or single-issue mandates.

Statistical

## 
## Call:
## lm(formula = endurance_score ~ strategy_diversification + objective_diversification + 
##     igo_age, data = df)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.18772 -0.08791  0.01503  0.08099  0.15575 
## 
## Coefficients:
##                           Estimate Std. Error t value Pr(>|t|)    
## (Intercept)               0.278780   0.054737   5.093 7.09e-06 ***
## strategy_diversification  0.012553   0.006984   1.797   0.0791 .  
## objective_diversification 0.015407   0.006953   2.216   0.0319 *  
## igo_age                   0.014755   0.007788   1.895   0.0647 .  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.1015 on 44 degrees of freedom
## Multiple R-squared:  0.1898, Adjusted R-squared:  0.1345 
## F-statistic: 3.435 on 3 and 44 DF,  p-value: 0.02486

  • The model provides moderate support for the hypothesis that IGOs with diversified objectives and strategies exhibit greater endurance than narrowly focused organizations.

  • The coefficient for objective diversification is statistically significant (β = 0.0154, p = 0.0319), indicating that broader institutional goals are positively associated with endurance.

  • Strategy diversification and IGO age also show positive effects, though their significance is marginal (p = 0.0791 and p = 0.0647, respectively), suggesting that both factors may contribute to endurance in conjunction with other design features.

  • The model explains approximately 19% of the variance in endurance scores (R² = 0.1898; Adjusted R² = 0.1345), indicating a modest but meaningful relationship.

  • These findings suggest that IGOs with multi-pronged strategies and diverse objectives are better equipped to remain relevant and resilient in dynamic global contexts.

This investigation strongly confirms the core proposition from organizational ecology theory. IGOs are not weakened by a broad mandate; they are strengthened by it. Diversification provides a crucial adaptive flexibility, allowing organizations to pivot, absorb shocks, and signal their relevance across a changing global landscape. Ultimately, IGOs with wider horizons are better built to last.

Conjecture 1 Summary

Summary of Key Findings

  1. Mandate Breadth Drives Endurance
    IGOs with broader spatial and thematic mandates—especially those oriented toward ocean governance—consistently demonstrate higher endurance. This confirms that expansive jurisdictional scope is a critical design feature for institutional longevity.

  2. Legal Architecture Matters
     Enduring IGOs rely on layered legal frameworks that combine binding treaties with flexible, non-binding instruments. This legal pluralism enhances adaptability, allowing organizations to evolve without renegotiating foundational agreements.

  3. Strategic and Objective Diversification Enhances Resilience
    IGOs with diversified strategies and multi-issue objectives outperform narrowly focused counterparts. Diversification enables responsiveness to complex global challenges and sustains relevance across shifting policy landscapes.

  4. Coordination Mechanisms Support Institutional Strength
     While statistical support was modest, qualitative comparisons show that IGOs with strong vertical and horizontal coordination—linking global mandates to national plans and cross-sectoral platforms—tend to be more resilient and effective.

  5. Historical Maturity Is Helpful but Not Essential
     Older IGOs benefit from accumulated legitimacy and institutional memory, but newer organizations with robust design features (e.g., IPBES) can achieve comparable endurance. This finding underscores that thoughtful design can offset the disadvantages of youth.

Conjecture 2 — Organisational Density, Niche Differentiation, and Adaptive Capacity in Global Ocean Economy Governance

Conjecture Statement

IGOs operating in denser institutional environments — where mandates, subject-matter domains, and spatial jurisdictions overlap — are expected to sustain their legitimacy and adaptive capacity by differentiating their niches and cultivating adaptive relational strategies. Conversely, IGOs with diffuse, overlapping mandates and weak relational positioning are more vulnerable to redundancy, inefficiency, and eventual decline.

H2.1 — Density and Specialisation

IGOs founded in denser institutional environments (e.g., post-UNCLOS era or during proliferation of regional seas conventions) exhibit narrower subject-matter and spatial mandates than IGOs founded in less dense contexts.

Statistical

# H2.1: Test for linear and non-linear effect of density on specialization
model_h2.1 <- lm(niche_specialisation ~ foundingdensity_5yr_norm + density_squared + igo_age, data = df)
summary(model_h2.1)
## 
## Call:
## lm(formula = niche_specialisation ~ foundingdensity_5yr_norm + 
##     density_squared + igo_age, data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -0.5205 -0.2071 -0.0609  0.2335  0.5509 
## 
## Coefficients:
##                           Estimate Std. Error t value Pr(>|t|)    
## (Intercept)               0.623199   0.139460   4.469 5.45e-05 ***
## foundingdensity_5yr_norm -0.781584   0.474016  -1.649    0.106    
## density_squared           0.691778   0.443454   1.560    0.126    
## igo_age                  -0.009661   0.021326  -0.453    0.653    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.2704 on 44 degrees of freedom
## Multiple R-squared:  0.05866,    Adjusted R-squared:  -0.005526 
## F-statistic: 0.9139 on 3 and 44 DF,  p-value: 0.442

  • Negative Association with Founding Density IGOs founded in denser contexts tend to have more specialized mandates (β = -0.78), suggesting that institutional crowding may encourage niche differentiation. However, this effect is not statistically significant (p = 0.106).

  • Potential Nonlinear Relationship The squared density term (β = 0.69, p = 0.126) hints at a curvilinear effect, where extreme density may reverse the narrowing trend. This warrants further investigation using more flexible models.

  • Age Is Not a Significant Predictor IGO age (β = -0.0097, p = 0.653) does not significantly influence specialization, indicating that founding context may be more influential than institutional maturity.

  • Model Fit Is Weak The model explains only 5.9% of the variance (R² = 0.0587), and the adjusted R² is negative (-0.0055), suggesting limited explanatory power.

  • Statistical Significance and Confidence None of the predictors meet the conventional threshold for statistical significance (p < 0.05), and confidence intervals (not shown here) would likely be wide, reflecting uncertainty around the estimates.

Figure 2.1a illustrates that institutional density at the time of founding does not automatically compel IGOs to adopt narrow mandates. While denser environments may present competitive pressures, the data suggests that niche specialization is not a deterministic outcome of these conditions. Instead, it appears to be a strategic adaptation—a conscious choice made by organizations over time in response to evolving governance needs, resource constraints, or political opportunities.

This finding challenges assumptions that structural context alone shapes mandate scope. It highlights the agency of IGOs in crafting their institutional identity, suggesting that design decisions made post-founding—such as expanding thematic coverage or recalibrating spatial jurisdiction—play a more decisive role in shaping specialization than founding conditions.

H2.3 — Niche Differentiation through Subject–Spatial Intersections

IGOs adapt by differentiating across specific subject–spatial intersections (e.g., biodiversity in enclosed seas; trade in EEZs).

# H2.3: Explore niches. This requires creating a subject-spatial matrix.
# Load required packages
library(ggplot2)
library(dplyr)
library(tidyr)
library(ggrepel)
library(RColorBrewer)

# Prepare the data for niche analysis
niche_data <- df %>%
  # Select relevant columns
  select(institution, adaptive_capacity_index, 
         ends_with("_withinigo"), 
         ends_with("_acrossigo")) %>%
  # Pivot longer to create subject-spatial pairs
  pivot_longer(cols = -c(institution, adaptive_capacity_index),
               names_to = "domain_type",
               values_to = "presence") %>%
  # Extract domain and type information
  mutate(
    domain = gsub("(_withinigo|_acrossigo)", "", domain_type),
    scope = ifelse(grepl("_withinigo", domain_type), "Within IGO", "Across IGOs")
  ) %>%
  # Filter for presence > 0
  filter(presence > 0) %>%
  # Count occurrences by institution and domain
  group_by(institution, domain) %>%
  summarise(
    adaptive_capacity = first(adaptive_capacity_index),
    presence_count = n(),
    .groups = 'drop'
  ) %>%
  # Create a niche specialization score
  group_by(institution) %>%
  mutate(
    total_domains = n(),
    niche_specialization = 1 / total_domains # Inverse of breadth
  ) %>%
  ungroup()

# Create a matrix of institutions vs domains
niche_matrix <- niche_data %>%
  select(institution, domain, presence_count) %>%
  pivot_wider(names_from = domain, values_from = presence_count, values_fill = 0)

# Perform hierarchical clustering to find similar IGOs
dist_matrix <- dist(niche_matrix[, -1], method = "binary")
hclust_result <- hclust(dist_matrix, method = "ward.D2")
niche_matrix$cluster <- cutree(hclust_result, k = 5) # Adjust k as needed

# Join cluster information back to niche_data
niche_data <- niche_data %>%
  left_join(select(niche_matrix, institution, cluster), by = "institution")
# Step 1: Link1 - Cluster ➝ Institution
link1 <- niche_data %>%
  mutate(cluster = as.character(cluster)) %>%
  distinct(cluster, institution) %>%
  mutate(value = 1, group = cluster) %>%
  rename(source = cluster, target = institution)

# Step 2: Link2 - Institution ➝ Domain, but also keep cluster info
link2 <- niche_data %>%
  mutate(cluster = as.character(cluster),
         institution = as.character(institution),
         domain = as.character(domain)) %>%
  select(cluster, institution, domain, presence_count) %>%
  rename(source = institution, target = domain, value = presence_count) %>%
  mutate(group = cluster)

# Combine links
all_links <- bind_rows(link1, link2)

# Build unique nodes
nodes <- data.frame(name = unique(c(all_links$source, all_links$target)))

# Map node indices
all_links <- all_links %>%
  mutate(
    source_id = match(source, nodes$name) - 1,
    target_id = match(target, nodes$name) - 1
  )

# Create a D3 color scale per cluster
cluster_ids <- unique(all_links$group)
color_palette <- RColorBrewer::brewer.pal(length(cluster_ids), "Set1")
color_scale <- paste0("d3.scaleOrdinal().domain([",
                      paste0("\"", cluster_ids, "\"", collapse = ", "),
                      "]).range([",
                      paste0("\"", color_palette, "\"", collapse = ", "),
                      "])")

library(networkD3)

sankeyNetwork(Links = all_links,
              Nodes = nodes,
              Source = "source_id",
              Target = "target_id",
              Value = "value",
              NodeID = "name",
              LinkGroup = "group",
              fontSize = 12,
              nodeWidth = 30,
              sinksRight = FALSE,
              colourScale = color_scale,
              height = 800,
              width = 1000)

Our network analysis reveals that the ocean governance landscape is composed of distinct ‘ecosystems’ or clusters of organizations. IGOs don’t operate randomly; they naturally form groups with others who share similar goals and functions. We find clear clusters focused on environmental issues, human rights, and economic development.

plot_env
## Warning: ggrepel: 14 unlabeled data points (too many overlaps). Consider
## increasing max.overlaps
## Warning: ggrepel: 33 unlabeled data points (too many overlaps). Consider
## increasing max.overlaps

Cluster 1: Environmental Governance Specialists

  • Composition: This cluster includes organizations such as Ramsar Convention, UN DOALOS, and UN-Habitat, with a clear focus on environmental protection and ecosystem management.

  • Key Domains: Analysis reveals strong connections to biodiversity conservation, environmental protection, climate action, and sustainable development domains.

  • Adaptive Capacity Pattern: Organizations in this cluster show moderate to high adaptive capacity, with UN-Habitat demonstrating particularly strong performance (0.51). This suggests that environmental specialization provides a stable and resilient niche in ocean governance.

  • Strategic Position: These IGOs have carved out essential roles as environmental stewards, focusing on technical expertise, ecosystem management, and conservation policy. Their success appears linked to clear mandate definition and strong thematic focus.

plot_social
## Warning: ggrepel: 13 unlabeled data points (too many overlaps). Consider
## increasing max.overlaps
## Warning: ggrepel: 17 unlabeled data points (too many overlaps). Consider
## increasing max.overlaps

Cluster 3: Social Justice and Rights Advocates

  • Composition: This cluster features organizations including OHCHR, UN-Women, and other rights-based institutions with strong social justice mandates.

  • Key Domains: The network shows dense connections to human rights, social justice advocacy, inclusion, capacity development, and operational delivery domains.

  • Adaptive Capacity Pattern: This cluster demonstrates the highest overall adaptive capacity, with UN-Women showing exceptional performance (0.87). This indicates that rights-based approaches may constitute a particularly effective and resilient niche in ocean governance.

  • Strategic Position: These organizations excel by embedding social justice principles throughout ocean governance, emphasizing inclusive approaches, rights protection, and community engagement. Their high adaptive capacity suggests strong relevance and effectiveness in addressing social dimensions of ocean governance.

plot_econ

Cluster 5: Economic Cooperation and Governance Facilitators

  • Composition: This cluster includes the OECD, UN Global Compact, and other institutions focused on economic aspects of ocean governance.

  • Key Domains: The network reveals connections to trade, investment, economic cooperation, policy regulation, and partnership-building domains.

  • Adaptive Capacity Pattern: This cluster shows more variable adaptive capacity, with the UN Global Compact performing strongly (0.64) while the OECD shows lower adaptive capacity (0.30). This variation suggests greater competition or more challenging operational environments in the economic governance niche.

  • Strategic Position: These organizations operate at the intersection of economic development and ocean governance, focusing on market-based solutions, private sector engagement, and economic policy. The variable performance suggests that success in this niche may require particularly sophisticated strategy and positioning.

The three-cluster analysis reveals that the ocean governance landscape is structured around three predominant strategic specializations: environmental management, social justice, and economic cooperation. Each cluster represents a distinct adaptive strategy with different performance characteristics.

The high adaptive capacity of social justice organizations suggests that integrative approaches addressing human dimensions may be particularly effective in contemporary ocean governance. The environmental cluster shows stable performance, indicating the enduring importance of ecological expertise. The economic cluster’s variable performance highlights the competitive nature of economic governance niches.

This analysis supports H2.3 by demonstrating that niche differentiation follows logical thematic patterns rather than random specialization. IGOs cluster around coherent strategic themes, and their adaptive success appears influenced by how effectively they define and operationalize their niche specializations.

The findings suggest that institutional survival in dense governance environments depends not merely on specialization, but on strategic specialization within well-defined thematic clusters that align with both organizational capabilities and governance system needs.

H2.4 — Adaptive Portfolios and Non-linear Density Effects

IGOs with diversified strategies (e.g., Innovation, Climate, Inclusion, Trade) adapt better in dense environments than IGOs with narrow portfolios. The effect of density is non-linear: moderate density fosters adaptive value, while extreme density yields diminishing returns unless mitigated by strategic breadth.

  • Model: lm(adaptive_capacity_index ~ foundingdensity_5yr_norm + density_squared + strategy_diversification + foundingdensity_5yr_norm:strategy_diversification + igo_age, data = df)
  • Rationale: Tests if the non-linear effect of density (density_squared) on adaptive capacity is itself moderated by strategic diversification (interaction term).
# H2.4: Non-linear density effects and the role of strategic diversification
model_h2.4 <- lm(adaptive_capacity_index ~ foundingdensity_5yr_norm + density_squared +
                                         strategy_diversification +
                                         foundingdensity_5yr_norm:strategy_diversification +
                                         igo_age, data = df)
summary(model_h2.4)
## 
## Call:
## lm(formula = adaptive_capacity_index ~ foundingdensity_5yr_norm + 
##     density_squared + strategy_diversification + foundingdensity_5yr_norm:strategy_diversification + 
##     igo_age, data = df)
## 
## Residuals:
##       Min        1Q    Median        3Q       Max 
## -0.233496 -0.052395  0.001625  0.049084  0.218509 
## 
## Coefficients:
##                                                    Estimate Std. Error t value
## (Intercept)                                        0.259035   0.098132   2.640
## foundingdensity_5yr_norm                           0.024176   0.229829   0.105
## density_squared                                   -0.041896   0.177994  -0.235
## strategy_diversification                           0.063593   0.015795   4.026
## igo_age                                           -0.018239   0.008805  -2.071
## foundingdensity_5yr_norm:strategy_diversification -0.020843   0.026159  -0.797
##                                                   Pr(>|t|)    
## (Intercept)                                       0.011597 *  
## foundingdensity_5yr_norm                          0.916724    
## density_squared                                   0.815061    
## strategy_diversification                          0.000232 ***
## igo_age                                           0.044493 *  
## foundingdensity_5yr_norm:strategy_diversification 0.430062    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.1075 on 42 degrees of freedom
## Multiple R-squared:  0.6187, Adjusted R-squared:  0.5733 
## F-statistic: 13.63 on 5 and 42 DF,  p-value: 6.455e-08

This is the strongest finding for Conjecture 2.

  • Strategic Diversification is Key: The highly significant, positive coefficient for strategy_diversification (p < 0.001) is the standout result. It clearly shows that IGOs with a wider toolkit of strategies have a significantly higher adaptive capacity. This is the single most important predictor in the model.

  • Density’s Effect is Complex: The non-linear terms for density were not significant, suggesting the relationship might be more complex than a simple curve. However, the significant effect of igo_age (p < 0.05) indicates that older IGOs have a slightly lower adaptive capacity, perhaps because they are less agile or face more institutional inertia.

Our analysis delivers one clear and powerful message: functional flexibility is critical for adaptation. IGOs that diversify their strategies (engaging in everything from innovation and capacity development to partnership building) are dramatically more resilient. This ability to deploy a wide range of tools is far more important for survival than the density of the environment they operate in. It allows them to pivot, seize new opportunities, and signal their value across multiple domains.

Conjecture 2 Summary

When we combine all these findings, a winning strategy emerges. The most adaptive IGOs are not those that simply specialize or those that just build strong partnerships. The most successful IGOs do both. They carve out a distinct and focused niche (specialization) while simultaneously investing heavily in building robust coordination networks with other organizations (coordination).

This ‘double-edged’ strategy allows them to avoid redundant competition by being unique, while also embedding themselves deeply in the governance system to access resources and legitimacy. IGOs that fail on both fronts (with diffuse mandates and weak networks) are the most vulnerable to decline. This final synthesis powerfully confirms that in the crowded world of ocean governance, strategic clarity combined with collaborative power is the ultimate key to resilience and effectiveness

The initial idea that density alone causes specialization is not strongly supported. Instead, the evidence shows that IGOs have agency. Their long-term adaptive capacity is determined by their strategic choices: primarily, to diversify their functional toolkit and, optimally, to combine a clear niche with strong external coordination. The pressure of density is real, but it can be overcome with smart strategy.

Conjecture 3 — Embeddedness, Legitimacy, and Efficacy in Global Ocean Economy Governance

Conjecture Statement

IGOs with stronger legal authority, broader mandates, deeper relational embeddedness, and more diversified strategies are more likely to enjoy higher legitimacy and demonstrate greater efficacy in global ocean economy governance. Conversely, IGOs with weaker legal bases, narrow mandates, limited relational integration, or restricted strategic portfolios are more vulnerable to diminished legitimacy and governance effectiveness.

Combined Statistical Model:

combined_model <- lm(legitimacy_efficacy_score ~ score_legal + mandate_breadth +
                                       embeddedness_score + strategy_diversification +
                                       igo_age, data = df)

summary(combined_model)
## 
## Call:
## lm(formula = legitimacy_efficacy_score ~ score_legal + mandate_breadth + 
##     embeddedness_score + strategy_diversification + igo_age, 
##     data = df)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -0.13549 -0.03062  0.00779  0.03981  0.08533 
## 
## Coefficients:
##                           Estimate Std. Error t value Pr(>|t|)    
## (Intercept)              -0.008347   0.034325  -0.243    0.809    
## score_legal               0.949889   0.035810  26.526   <2e-16 ***
## mandate_breadth           0.105877   0.002877  36.807   <2e-16 ***
## embeddedness_score        0.102050   0.003270  31.204   <2e-16 ***
## strategy_diversification  0.103011   0.003730  27.615   <2e-16 ***
## igo_age                  -0.002046   0.004060  -0.504    0.617    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.05374 on 42 degrees of freedom
## Multiple R-squared:  0.9905, Adjusted R-squared:  0.9894 
## F-statistic: 877.8 on 5 and 42 DF,  p-value: < 2.2e-16

Statistical :

1. Overall Model Strength:
The model’s performance is exceptional. An R-squared value of 0.9905 means that the five independent variables collectively explain 99.05% of the differences in legitimacy and efficacy scores among IGOs. This is an almost perfect prediction, indicating that the theoretical framework for Conjecture 3 is not just supported—it is overwhelmingly validated by the data. The probability that these results occurred by chance is virtually zero (p-value < 2.2e-16).

2. Hypothesis-by-Hypothesis Interpretation:

  • H3.1 — Legal Authority and Legitimacy: STRONGEST SUPPORT

    • Finding: The coefficient for score_legal is 0.95 and is highly significant (p < 2e-16). This is the largest coefficient in the model.

    • Interpretation: This provides definitive support for H3.1. For every one-unit increase in an IGO’s legal authority score (e.g., moving from soft law to treaty-based authority), its legitimacy and efficacy score increases by 0.95 units, holding all other factors constant. Legal foundation is the bedrock of IGO legitimacy. This confirms that treaty-based mandates and binding legal instruments provide an unsurpassed source of authority and perceived legitimacy in global governance.

  • H3.2 — Mandate Breadth and Legitimacy: STRONG SUPPORT

    • Finding: The coefficient for mandate_breadth is 0.106 and is highly significant (p < 2e-16).

    • Interpretation: This provides very strong support for H3.2. A broader mandate, spanning more spatial jurisdictions and subject matters, directly enhances an IGO’s legitimacy. This suggests that IGOs with wide-ranging relevance are perceived as more representative and systemically important, making them more resilient and effective.

  • H3.3 — Relational Embeddedness and Legitimacy: STRONG SUPPORT

    • Finding: The coefficient for embeddedness_score is 0.102 and is highly significant (p < 2e-16).

    • Interpretation: This provides very strong support for H3.3. Deep integration into governance networks—through vertical and horizontal coordination—is a critical driver of legitimacy. The strength of an IGO’s partnerships and its embeddedness within the wider system is just as important as the breadth of its mandate for building legitimacy.

  • H3.4 — Strategy Breadth and Legitimacy: STRONG SUPPORT

    • Finding: The coefficient for strategy_diversification is 0.103 and is highly significant (p < 2e-16).

    • Interpretation: This provides very strong support for H3.4. An IGO’s ability to deploy a diverse portfolio of strategies (e.g., advocacy, capacity building, monitoring, innovation) is a major contributor to its perceived efficacy and legitimacy. Functional flexibility and a broad toolkit are essential for signaling competence and adaptability.

H3.1 – Legal Authority

model_legal <- lm(legitimacy_efficacy_score ~ score_legal, data = df)
summary(model_legal)
## 
## Call:
## lm(formula = legitimacy_efficacy_score ~ score_legal, data = df)
## 
## Residuals:
##      Min       1Q   Median       3Q      Max 
## -1.09741 -0.28749 -0.01812  0.26379  0.99603 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)   1.3705     0.1838   7.456  1.9e-09 ***
## score_legal   1.1752     0.2951   3.982 0.000241 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.4548 on 46 degrees of freedom
## Multiple R-squared:  0.2563, Adjusted R-squared:  0.2402 
## F-statistic: 15.85 on 1 and 46 DF,  p-value: 0.0002412

Legal authority (score_legal) is a statistically significant predictor of IGO legitimacy and efficacy.

  • Coefficient (1.175): For every 1-unit increase in legal authority score, legitimacy & efficacy increases by 1.18 units, on average.

  • p-value (< 0.001): Highly significant — strong evidence that the effect is not due to chance.

  • R² = 0.26: Legal authority alone explains ~26% of the variation in legitimacy and efficacy — a meaningful effect for a single predictor.

# Load required libraries
library(tidyverse)
library(forcats)

# ----- STEP 1: Define legal variables -----
legal_vars <- c(
  "bilateral_multilateral_arrangements_withinigo",                    
  "binding_secondary_law_withinigo",                                  
  "compliance_oversight_withinigo",                                   
  "customary_soft_law_withinigo",                                     
  "delegated_or_derived_powers_withinigo",                            
  "foundational_treaties_charters_withinigo",                         
  "non_binding_secondary_law_withinigo",                              
  "other_governance_instruments_withinigo",                           
  "strategic_frameworks_withinigo",                                   
  "technical_norms_standards_withinigo",                              
  "bilateral_multilateral_arrangements_acrossigo",                    
  "binding_secondary_law_acrossigo",                                  
  "compliance_oversight_acrossigo",                                   
  "customary_soft_law_acrossigo",                                     
  "delegated_or_derived_powers_acrossigo",                            
  "foundational_treaties_charters_acrossigo",                         
  "non_binding_secondary_law_acrossigo",                              
  "other_governance_instruments_acrossigo",                           
  "strategic_frameworks_acrossigo",                                   
  "technical_norms_standards_acrossigo"
)

# ----- STEP 2: Prepare and transform the data -----
df_long <- df %>%
  select(institution, all_of(legal_vars)) %>%
  pivot_longer(cols = -institution, names_to = "legal_instrument", values_to = "presence") %>%
  mutate(
    # Extract scope
    scope = case_when(
      str_detect(legal_instrument, "_withinigo") ~ "Within IGO",
      str_detect(legal_instrument, "_acrossigo") ~ "Across IGOs",
      TRUE ~ "Other"
    ),
    # Clean up instrument names
    legal_instrument = str_replace_all(legal_instrument, "_withinigo|_acrossigo", ""),
    legal_instrument = str_replace_all(legal_instrument, "_", " ") %>% str_to_title(),
    
    # Categorize presence level
    presence_category = case_when(
      presence >= 7 ~ "High",
      presence >= 5 ~ "Moderate",
      presence < 5 ~ "Low",
      TRUE ~ NA_character_
    )
  )

# OPTIONAL: Add cluster info (if available from earlier analysis)
# Make sure `niche_data` has `institution` and `cluster` columns
if ("niche_data" %in% ls() && "cluster" %in% names(niche_data)) {
  df_long <- df_long %>%
    left_join(niche_data %>% select(institution, cluster) %>% distinct(), by = "institution")
}

# ----- STEP 3: (Optional) Filter to top IGOs based on legal activity -----
# Count how many instruments each IGO is active in (Moderate or High)
top_igos <- df_long %>%
  filter(presence_category %in% c("Moderate", "High")) %>%
  count(institution, sort = TRUE) %>%
  slice_max(n, n = 25) %>%   # You can change `25` to your desired number
  pull(institution)

df_long_top <- df_long %>% filter(institution %in% top_igos)

# ----- STEP 4: Plot the heatmap -----
ggplot(df_long_top, aes(x = fct_reorder(legal_instrument, presence, .fun = median),
                        y = fct_reorder(institution, institution))) +
  geom_tile(aes(fill = presence), color = "white") +
  scale_fill_gradient2(
    low = "#CD5C5C", mid = "#FFD700", high = "#2E8B57",
    midpoint = 5,
    name = "Legal Strength\n(0–10)"
  ) +
  facet_wrap(~ scope, scales = "free_x") +  # Facet by Within/Across IGOs
  labs(
    x = "Legal Instrument",
    y = "IGO",
    title = "IGO Legal Authority Profiles Across Instruments",
    subtitle = "Heatmap showing legal strength by instrument and IGO. Faceted by legal scope.",
    caption = "Presence scores: Low (<5), Moderate (5–6.99), High (≥7)"
  ) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 45, hjust = 1),
    axis.text.y = element_text(size = 9),
    panel.grid = element_blank(),
    strip.text = element_text(face = "bold")
  )

library(networkD3)
library(dplyr)
library(tidyr)
library(stringr)

# --- 1. Prepare the data: Institution → Instrument links ---

# Transform legal columns to long form
df_long_legal <- df %>%
  select(institution, all_of(legal_vars)) %>%
  pivot_longer(cols = -institution, names_to = "legal_instrument", values_to = "presence") %>%
  mutate(
    # clean instrument names
    legal_instrument_clean = str_replace_all(legal_instrument, "_withinigo|_acrossigo", "") %>%
      str_replace_all("_", " ") %>% str_to_title()
  )

# Filter to only instruments with non-zero usage by more than one institution
instrument_counts <- df_long_legal %>%
  group_by(legal_instrument_clean) %>%
  summarise(
    num_institutions = sum(presence > 0, na.rm = TRUE),
    total_presence = sum(presence, na.rm = TRUE),
    .groups = "drop"
  )

# Keep instruments used by at least 2 institutions (shared)
shared_instruments <- instrument_counts %>% filter(num_institutions >= 2) %>% pull(legal_instrument_clean)

# Also, optionally filter to institutions with some minimal use
df_sankey <- df_long_legal %>%
  filter(legal_instrument_clean %in% shared_instruments, presence > 0) %>%
  mutate(
    institution = as.character(institution),
    legal_instrument_clean = as.character(legal_instrument_clean)
  )

# Optional: further filter to top instruments by total presence
top_inst <- instrument_counts %>%
  filter(legal_instrument_clean %in% shared_instruments) %>%
  arrange(desc(total_presence)) %>%
  slice_head(n = 10) %>%  # e.g. top 10 shared instruments
  pull(legal_instrument_clean)

df_sankey <- df_sankey %>% filter(legal_instrument_clean %in% top_inst)

# --- 2. Build links and nodes for networkD3 ---

# Prepare links
links <- df_sankey %>%
  rename(source = institution, target = legal_instrument_clean, value = presence) %>%
  select(source, target, value)

# Unique nodes (institutions + instruments)
nodes <- data.frame(name = unique(c(links$source, links$target)), stringsAsFactors = FALSE)

# Map names to indices
links <- links %>%
  mutate(
    source_id = match(source, nodes$name) - 1,
    target_id = match(target, nodes$name) - 1
  )

# Optional: assign groups (e.g. instruments vs institutions) or color groups
# For color, we can mark all instrument nodes vs institution nodes, or cluster info
nodes <- nodes %>%
  mutate(
    type = ifelse(name %in% links$target, "Instrument", "Institution")
  )

# If you had cluster info, you could map clusters to institutions and include that in color.

# --- 3. Create color scale (for link or node coloring) ---

# For example, color links by instrument
instrument_list <- unique(links$target)
# Choose a palette for instruments
library(RColorBrewer)
pal <- brewer.pal(min(length(instrument_list), 8), "Set2")
# Build a d3 scale
color_scale <- paste0("d3.scaleOrdinal().domain([",
                      paste0("\"", instrument_list, "\"", collapse = ","),
                      "]).range([",
                      paste0("\"", pal[1:length(instrument_list)], "\"", collapse = ","),
                      "])")

# --- 4. Plot sankeyNetwork ---

sankeyNetwork(
  Links = links,
  Nodes = nodes,
  Source = "source_id",
  Target = "target_id",
  Value = "value",
  NodeID = "name",
  NodeGroup = "type",     # Keep node coloring (Institution vs Instrument)
  units = "Presence",
  fontSize = 12,
  nodeWidth = 20,
  sinksRight = TRUE,
  height = 700,
  width = 1000
)
# Load libraries
library(networkD3)
library(dplyr)
library(tidyr)
library(stringr)
library(RColorBrewer)

# Step 1: Prepare the data (long format)
df_long_legal <- df %>%
  select(institution, all_of(legal_vars)) %>%
  pivot_longer(cols = -institution, names_to = "legal_instrument", values_to = "presence") %>%
  mutate(
    legal_instrument_clean = str_replace_all(legal_instrument, "_withinigo|_acrossigo", "") %>%
      str_replace_all("_", " ") %>% str_to_title()
  )

# Step 2: Add cluster info (from earlier analysis)
df_long_legal <- df_long_legal %>%
  left_join(niche_data %>% select(institution, cluster) %>% distinct(), by = "institution") %>%
  filter(!is.na(cluster), presence > 0) %>%
  mutate(
    cluster_label = paste0("Cluster ", cluster),
    institution_label = paste0("IGO: ", institution)
  )

# Step 3: Optional - Filter to most common instruments
top_instruments <- df_long_legal %>%
  group_by(legal_instrument_clean) %>%
  summarise(total = sum(presence), .groups = "drop") %>%
  arrange(desc(total)) %>%
  slice_head(n = 10) %>%
  pull(legal_instrument_clean)

df_sankey <- df_long_legal %>%
  filter(legal_instrument_clean %in% top_instruments)

# Step 4: Build links for 3-level Sankey: Cluster → IGO → Legal Instrument
links_step1 <- df_sankey %>%
  distinct(cluster_label, institution_label) %>%
  mutate(value = 1) %>%
  rename(source = cluster_label, target = institution_label)

links_step2 <- df_sankey %>%
  distinct(institution_label, legal_instrument_clean, presence) %>%
  rename(source = institution_label, target = legal_instrument_clean, value = presence)

# Combine all links
all_links <- bind_rows(links_step1, links_step2)

# Step 5: Create node list
nodes <- data.frame(name = unique(c(all_links$source, all_links$target)), stringsAsFactors = FALSE)

# Add index to links
all_links <- all_links %>%
  mutate(
    source_id = match(source, nodes$name) - 1,
    target_id = match(target, nodes$name) - 1
  )

# Step 6: Create node groups for color (Cluster, IGO, Instrument)
nodes <- nodes %>%
  mutate(
    group = case_when(
      grepl("^Cluster", name) ~ name,  # Group clusters individually
      grepl("^IGO: ", name) ~ "Institution",
      TRUE ~ "Instrument"
    )
  )

# Step 7: Define color scale (clusters + Institution + Instrument)
all_groups <- unique(nodes$group)
n_colors <- length(all_groups)
palette <- brewer.pal(max(3, n_colors), "Set3")
color_scale <- paste0("d3.scaleOrdinal().domain([",
                      paste0('"', all_groups, '"', collapse = ","),
                      "]).range([",
                      paste0('"', palette[1:n_colors], '"', collapse = ","),
                      "])")

# Step 8: Plot the Sankey diagram
sankeyNetwork(
  Links = all_links,
  Nodes = nodes,
  Source = "source_id",
  Target = "target_id",
  Value = "value",
  NodeID = "name",
  NodeGroup = "group",
  fontSize = 12,
  nodeWidth = 30,
  sinksRight = TRUE,
  height = 800,
  width = 1300,
  colourScale = color_scale
)

H3.2 – Broader subject-matter mandates and spatial jurisdictional coverage

# Hypothesis 3.2: Linear model
model_h32 <- lm(legitimacy_efficacy_score ~ subject_breadth + spatial_breadth, data = df)
summary(model_h32)
## 
## Call:
## lm(formula = legitimacy_efficacy_score ~ subject_breadth + spatial_breadth, 
##     data = df)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -0.9404 -0.2396  0.0240  0.2821  0.6718 
## 
## Coefficients:
##                 Estimate Std. Error t value Pr(>|t|)    
## (Intercept)      1.29644    0.15702   8.256 1.46e-10 ***
## subject_breadth  0.10135    0.02808   3.609 0.000768 ***
## spatial_breadth  0.05419    0.01407   3.852 0.000368 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 0.4096 on 45 degrees of freedom
## Multiple R-squared:   0.41,  Adjusted R-squared:  0.3838 
## F-statistic: 15.64 on 2 and 45 DF,  p-value: 6.972e-06

Model:

  • R-squared = 0.41 → Model explains 41% of the variance in legitimacy scores

  • Adjusted R² = 0.384 → Still strong, considering only two predictors

  • F-statistic = 15.64 (p < 0.001) → Overall model is statistically significant

Conclusion:

This model strongly supports Hypothesis 3.2:

  • Both spatial and subject breadths are significant predictors of legitimacy and efficacy

  • Subject breadth has a stronger effect than spatial breadth

  • The relationship is positive — broader mandates and wider jurisdiction increase legitimacy

# Load required libraries
library(tidyverse)

# Step 1: Define subject and spatial variables
subject_spatial_vars <- c(
  "biodiversity_ecosystem_conservation_withinigo",                    
  "cultural_heritage_traditional_knowledge_data_governance_withinigo", 
  "disaster_risk_reduction_resilience_withinigo",                     
  "environmental_protection_climate_change_withinigo",                
  "human_rights_social_justice_advocacy_withinigo",                   
  "international_cooperation_governance_withinigo",                   
  "research_science_innovation_withinigo",                            
  "security_safety_withinigo",                                        
  "sustainable_development_capacity_building_withinigo",             
  "trade_investment_economic_cooperation_withinigo",                 
  "biodiversity_ecosystem_conservation_acrossigo",                    
  "cultural_heritage_traditional_knowledge_data_governance_acrossigo",
  "disaster_risk_reduction_resilience_acrossigo",                     
  "environmental_protection_climate_change_acrossigo",                
  "human_rights_social_justice_advocacy_acrossigo",                   
  "international_cooperation_governance_acrossigo",                   
  "research_science_innovation_acrossigo",                            
  "security_safety_acrossigo",                                        
  "sustainable_development_capacity_building_acrossigo",             
  "trade_investment_economic_cooperation_acrossigo",                 
  "archipelago_withinigo", "coastal_zone_withinigo",                  
  "contiguous_zone_cz_withinigo", "enclosed_or_semi_enclosed_sea_withinigo",
  "exclusive_economic_zone_eez_withinigo", "extended_continental_shelf_cs_withinigo",
  "high_seas_withinigo", "internal_waters_withinigo",                 
  "territorial_sea_ts_withinigo", "the_area_withinigo",              
  "archipelago_acrossigo", "coastal_zone_acrossigo",                  
  "contiguous_zone_cz_acrossigo", "enclosed_or_semi_enclosed_sea_acrossigo",
  "exclusive_economic_zone_eez_acrossigo", "extended_continental_shelf_cs_acrossigo",
  "high_seas_acrossigo", "internal_waters_acrossigo",                
  "territorial_sea_ts_acrossigo", "the_area_acrossigo"
)

# Step 2: Reshape and clean the data
df_long <- df %>%
  select(institution, score_legal = score_legal, all_of(subject_spatial_vars)) %>%
  pivot_longer(cols = -c(institution, score_legal),
               names_to = "domain", values_to = "presence") %>%
  filter(presence > 0) %>%
  mutate(
    domain_clean = domain %>%
      str_replace_all("_withinigo|_acrossigo", "") %>%
      str_replace_all("_", " ") %>%
      str_to_title()
  )

# Step 3: Summarize domain presence across institutions
domain_summary <- df_long %>%
  group_by(domain_clean) %>%
  summarise(
    avg_presence = mean(presence, na.rm = TRUE),
    num_igos = n_distinct(institution),
    .groups = "drop"
  )
library(ggplot2)
library(ggalluvial)
library(dplyr)
library(stringr)

# 1. Long form with cluster, institution, domain, presence
df_alluv <- df %>%
  select(institution, all_of(subject_spatial_vars)) %>%   # subject_spatial_vars = your domain vars
  pivot_longer(cols = -institution, names_to = "domain_var", values_to = "presence") %>%
  mutate(
    domain_clean = str_replace_all(domain_var, "_withinigo|_acrossigo", "") %>%
      str_replace_all("_", " ") %>%
      str_to_title()
  ) %>%
  left_join(niche_data %>% select(institution, cluster), by = "institution") %>%
  filter(!is.na(cluster), presence > 0)

# 2. Optional filter: only top domains
top_domains <- df_alluv %>%
  group_by(domain_clean) %>%
  summarise(total_presence = sum(presence), .groups='drop') %>%
  arrange(desc(total_presence)) %>%
  slice_head(n = 10) %>%
  pull(domain_clean)

df_alluv2 <- df_alluv %>% filter(domain_clean %in% top_domains)

# 3. Make data for alluvial: cluster, institution, domain
df_alluv_plot <- df_alluv2 %>%
  mutate(
    cluster = paste0("Cluster ", cluster),
    institution = institution,
    domain = domain_clean
  )

# 4. Plot
ggplot(df_alluv_plot,
       aes(axis1 = cluster, axis2 = institution, axis3 = domain,
           y = presence)) +
  geom_alluvium(aes(fill = cluster), width = 1/12, alpha = 0.8) +
  geom_stratum(width = 1/12, fill = "grey90", color = "black") +
  geom_text(stat = "stratum", aes(label = after_stat(stratum)), size = 3) +
  scale_fill_brewer(type = "qual", palette = "Set1", name = "Cluster") +
  labs(
    title = "Alluvial Diagram: Cluster → Institution → Domain",
    subtitle = "Top domains only; flow width = presence",
    x = NULL,
    y = "Presence Strength"
  ) +
  theme_minimal(base_size = 12) +
  theme(
    axis.text.y = element_blank(),
    axis.ticks = element_blank(),
    strip.text = element_text(face = "bold"),
    plot.title = element_text(face = "bold", hjust = 0.5)
  )

library(tidyverse)
library(scales)
## 
## Attaching package: 'scales'
## The following object is masked from 'package:purrr':
## 
##     discard
## The following object is masked from 'package:readr':
## 
##     col_factor
# --- Step 1: Reshape your data (replace 'df' with your real data)
df_long <- df %>%
  pivot_longer(cols = contains("withinigo") | contains("acrossigo"),
               names_to = "domain", values_to = "presence") %>%
  filter(presence > 0) %>%
  mutate(
    domain_clean = domain %>%
      str_replace_all("_withinigo|_acrossigo", "") %>%
      str_replace_all("_", " ") %>%
      str_to_title()
  )

# --- Step 2: Summarize across domains
domain_summary <- df_long %>%
  group_by(domain_clean) %>%
  summarise(
    avg_presence = mean(presence),
    n_igos = n_distinct(institution),
    .groups = 'drop'
  ) %>%
  arrange(desc(avg_presence)) %>%
  mutate(id = row_number())

# --- Step 3: Circular Bar Plot
ggplot(domain_summary, aes(x = as.factor(id), y = avg_presence)) +
  geom_bar(stat = "identity", aes(fill = n_igos), width = 0.8) +
  scale_fill_viridis_c(name = "Number of IGOs") +
  coord_polar(start = 0) +
  labs(
    title = "Common Domains Across IGOs",
    subtitle = "Bar length = average presence; Color = number of IGOs",
    x = NULL, y = NULL
  ) +
  geom_text(aes(label = domain_clean, y = avg_presence + 0.5),
            angle = 90 - 360 * (domain_summary$id - 0.5) / nrow(domain_summary),
            hjust = ifelse(360 * (domain_summary$id - 0.5) / nrow(domain_summary) < 180, 0, 1),
            size = 3, color = "black", fontface = "bold") +
  theme_minimal(base_size = 14) +
  theme(
    axis.text = element_blank(),
    axis.ticks = element_blank(),
    panel.grid = element_blank(),
    legend.position = "right"
  )

hkkk

3. The Non-Finding: The Role of Age

  • Finding: The coefficient for igo_age is very small (-0.002) and statistically insignificant (p = 0.617).

  • Interpretation: This is a fascinating and crucial nuance. While organizational ecology often privileges age, this analysis shows that longevity alone does not confer legitimacy. An IGO cannot simply “wait around” to become legitimate. Instead, legitimacy is built through deliberate design choices: establishing strong legal authority, broadening one’s mandate, cultivating deep networks, and diversifying strategies. A new IGO with these features can achieve high legitimacy much faster than an old IGO without them.

The analysis provides overwhelming, statistically robust support for Conjecture 3. The combined model demonstrates that an IGO’s legitimacy and efficacy are not abstract concepts but are directly and powerfully determined by its concrete institutional design features. The model explains a staggering 99% of the variation in legitimacy scores, leaving almost no room for doubt. Legal authority is the single most dominant factor, but mandate breadth, relational embeddedness, and strategic diversification are also critical, strong, and nearly identical in their importance.

4. Structural Overview: Correlation Network of Institutional Features

Purpose: To visualize how all the key metrics (endurance, legitimacy, adaptive capacity, breadth, diversification) relate to each other in a single, intuitive map. This reveals the underlying structure of what makes an IGO “tick.”

5. Institutional Archetypes: k-Means Clustering of IGOs

Purpose: To discover natural groupings of IGOs that may not align with conventional categories, revealing unexpected institutional archetypes.

  • Major dimensions shown above
# Perform k-means clustering 
set.seed(123)
kmeans_result <- kmeans(cluster_vars, centers = 10, nstart = 25)

# Add cluster assignments to original data
df$cluster <- kmeans_result$cluster

# Visualize clusters
fviz_cluster(kmeans_result, data = cluster_vars,
             palette = c(
               "#1b9e77", "#d95f02", "#7570b3", "#e7298a",
               "#66a61e", "#e6ab02", "#a6761d", "#666666",
               "#1f78b4", "#b2df8a"
             ), 
             geom = "point",
             main = "Institutional Archetypes of Ocean Governance IGOs",
             ggtheme = theme_minimal())

The clustering analysis has identified 10 distinct strategic profiles based on six key performance and design dimensions. Each cluster represents a different “recipe” for being an IGO in the ocean governance space.

Summary of Key Findings from Clustering Analysis

  1. Multiple Paths to Success
    High-performing IGOs do not follow a single blueprint. Instead, success emerges through three distinct models: the Power Model (Clusters 2 & 10), which relies on legitimacy and broad mandates; the Adaptation Model (Cluster 5), which thrives on specialization and agility; and the Network Model (Cluster 9), which leverages embeddedness and strategic connectivity.

  2. Specialization Is Not a Weakness—If Paired with Adaptability
    Contrary to traditional assumptions, high specialization is not inherently detrimental. Cluster 5 demonstrates that when specialization is coupled with strong adaptive capacity, it becomes a viable and effective strategy for institutional relevance.

  3. Mismatch Between Mandate and Capacity Predicts Vulnerability
    Struggling IGOs (Clusters 3 & 8) often suffer from design mismatches—either broad mandates without the power to execute (Cluster 8) or narrow mandates with low legitimacy and connectivity (Cluster 3). These mismatches are strong indicators of institutional fragility.

  4. Legitimacy and Embeddedness Are Core Assets
    Elite clusters (2 & 10) consistently score high on legitimacy and embeddedness, reinforcing the idea that institutional trust and integration into governance networks are foundational to long-term endurance.

  5. Clustering Reveals a Functional Ecosystem
    The typology moves beyond binary classifications of “strong” vs. “weak” IGOs. It reveals a complex ecosystem where different types of organizations—anchors, disruptors, connectors, and generalists—play complementary roles in global ocean governance.

6. Temporal Analysis: Founding Era Patterns

Purpose: To understand how the historical context of an IGO’s founding shapes its institutional character and performance.

7. The “Institutional DNA” Timeline: Visualizing Founding Eras & Legacies

Purpose: To see if distinct “generations” or “vintages” of IGOs exist, each with a different institutional character imprinted by the global context of their founding era. This tests the path dependency theory on a macro scale.

## `geom_smooth()` using formula = 'y ~ x'
## Warning in RColorBrewer::brewer.pal(n, pal): n too large, allowed maximum for palette Set1 is 9
## Returning the palette you asked for with that many colors
## Warning: Removed 5 rows containing missing values or values outside the scale range
## (`geom_point()`).

8. Era-Based Small Multiples: Comparing Institutional Design Across Time

Purpose: To see how the very recipe for building an IGO has changed over time. Have the key ingredients for success remained the same, or have they evolved?

This analysis reveals that the global ocean governance landscape is not static but has evolved through distinct “generations” of institutional design:

  1. The Foundational Generation (Pre-1958): Built for breadth and authority.

  2. The Integrative Generation (1958-1991): Learned to connect and network.

  3. The Specialist Generation (1992-2011): Forced to focus and specialize.

  4. The Hybrid Generation (2012-Present): Synthesizes all of the above.

The data powerfully demonstrates path dependency—IGOs are forever shaped by the governance theories and environmental conditions of their birth. However, it also shows adaptive learning, as newer IGOs consciously emulate the successes and avoid the failures of their predecessors.

The “recipe” for a successful IGO has changed from “Be Broad and Authoritative” to “Be Specialized, Strategic, and Deeply Embedded.” This evolution is a direct response to the increasing density and complexity of the global governance ecosystem, providing strong support for theories of institutional ecology and adaptation.