# DECS-S-0034 — SIM Case Study: Global Drivers of Mobile Phone Subscriptions

**Author:** Group 3B Analysis — DECS‑S‑0034 Statistics for Business Decision Making  
**Date:** July 8, 2026  
**Course:** Sasin School of Management, DECS‑S‑0034  
**Data:** World Bank API (201 countries, 2024–2025)  

---

## 1. Introduction

### 1.1 The Business Problem

Mobile telecommunications is one of the world's most important industries, yet the drivers of adoption remain poorly understood. Why do some countries — like the United Arab Emirates (203 mobile subscriptions per 100 people) — have twice the penetration of others? And what role do GDP, internet access, fixed-line infrastructure, tourism, and demographics play?

This case study analyses 201 countries using World Bank data to determine which factors significantly predict mobile subscription density. The findings have direct implications for telecom operators deciding where to invest, governments crafting digital policy, and manufacturers forecasting device demand.

### 1.2 Thesis Statement

**Internet penetration is the only statistically significant predictor of mobile subscription density across 201 countries; GDP per capita, landline infrastructure, tourism volume, and young dependency ratio have no significant explanatory power after controlling for internet access. The model explains 18.3% of variance (R² = 0.1831, Adjusted R² = 0.1622), suggesting that most of the variation in mobile adoption is driven by unobserved factors such as regulatory quality, market competition, and cultural preferences.**

### 1.3 Assumptions

1. **Complete case analysis:** All 201 countries have complete data for all six variables (verified from World Bank API).
2. **Linearity:** The relationship between predictors and mobile subscriptions is approximately linear.
3. **Independence:** Observations (countries) are independent of each other.
4. **Homoscedasticity:** Residual variance is constant across fitted values (verified via residual plots).
5. **Normality of residuals:** Residuals are approximately normally distributed (n = 201 satisfies CLT).
6. **No endogeneity:** Predictors are exogenous — internet penetration is not simultaneously determined by mobile penetration in a way that biases the OLS estimator.
7. **Excel Data Analysis ToolPak:** All regression and test outputs are formatted to match Excel's Analysis ToolPak output style.

---

## 2. Background

### 2.1 The ASEAN Mobile Market

Southeast Asia is one of the world's fastest-growing mobile markets. The ASEAN region has:

- **11 countries** ranging from ultra-developed Singapore (94% internet penetration) to emerging Laos (64%)
- **Diverse income levels:** from Singapore's US$88,000 GDP/capita to Myanmar's ~US$1,200
- **Young populations:** average young dependency ratio of 32.1 vs global average of 40.3
- **High mobile penetration:** ASEAN mean of 124.2 subscriptions/100 people, above the global mean of 117.0

The "more SIM cards than friends" phenomenon — where individuals own multiple SIMs for price arbitrage, data plans, and network coverage — is common across ASEAN. Thailand leads the region with 187.8 subscriptions per 100 people.

### 2.2 Theoretical Framework

Economic theory suggests several possible drivers of mobile adoption:

| Theory | Implication |
|---|---|
| **Income effect** | Richer countries should have more mobile subscriptions |
| **Network effect** | Higher internet penetration creates demand for mobile data |
| **Substitution effect** | Mobile may replace fixed-line phones (negative correlation) |
| **Tourism effect** | Tourists buy local SIMs, boosting subscription counts |
| **Demographic effect** | Younger populations adopt new technology faster |

We test all five simultaneously using multiple regression.

---

## 3. Evaluation of the Case

### 3.1 Descriptive Statistics (Excel ToolPak Style)

The dataset contains 201 countries with complete observations across all six variables.

| Statistic | Mobile/100 | GDP/capita | Internet% | Landlines/100 | Tourism | YoungDep |
|-----------|-----------|------------|-----------|---------------|---------|----------|
| Mean | 116.97 | 23,667.39 | 70.41 | 14.98 | 4,955,028 | 40.33 |
| Standard Error | 3.66 | 2,550.74 | 1.91 | 1.16 | 918,754 | 1.39 |
| Median | 116.38 | 9,474.02 | 81.47 | 10.79 | 903,000 | 33.32 |
| Std Deviation | 51.91 | 36,163.02 | 27.01 | 16.39 | 13,025,589 | 19.68 |
| Range | 570.19 | 287,767.75 | 100.00 | 95.48 | 117,108,100 | 84.42 |
| Minimum | 3.96 | 233.82 | 0.00 | 0.00 | 900 | 14.69 |
| Maximum | 574.15 | 288,001.57 | 100.00 | 95.48 | 117,109,000 | 99.11 |
| Skewness | 3.88 | 3.60 | -0.97 | 1.69 | 5.36 | 0.86 |
| Kurtosis | 30.66 | 18.23 | -0.27 | 3.87 | 35.52 | -0.31 |
| Count | 201 | 201 | 201 | 201 | 201 | 201 |

**Key Observations:**

1. **Mobile subscriptions** are highly right-skewed (skewness = 3.88) with extreme outliers like UAE (203) and Macao (574). The median (116.38) is close to the mean (116.97), but the distribution has heavy tails (kurtosis = 30.66).

2. **GDP per capita** shows extreme inequality — the maximum (Monaco, $288,002) is over 1,000 times the minimum (Burundi, $234). Variance is 1.3 billion USD².

3. **Internet penetration** is left-skewed — most countries cluster above 80%, with a long tail of low-penetration countries pulling the mean down to 70.4%.

4. **Tourism** is extremely right-skewed (5.36) — major destinations like France and Spain receive orders of magnitude more visitors than small nations.

### 3.2 Correlation Matrix (Excel ToolPak Style)

| | Mobile | GDP | Internet | Landlines | Tourism | YoungDep |
|---|---|---|---|---|---|---|
| **Mobile** | 1.0000 | 0.1646 | 0.4120 | 0.2239 | 0.0658 | -0.3794 |
| **GDP** | 0.1646 | 1.0000 | 0.4531 | 0.5951 | 0.1332 | -0.4691 |
| **Internet** | 0.4120 | 0.4531 | 1.0000 | 0.4627 | 0.2435 | **-0.7785** |
| **Landlines** | 0.2239 | 0.5951 | 0.4627 | 1.0000 | 0.2180 | -0.5959 |
| **Tourism** | 0.0658 | 0.1332 | 0.2435 | 0.2180 | 1.0000 | -0.2495 |
| **YoungDep** | -0.3794 | -0.4691 | **-0.7785** | -0.5959 | -0.2495 | 1.0000 |

**Interpretation:**

1. **Internet_pct has the strongest correlation with Mobile (r = 0.4120).** This is the most promising predictor in the regression.

2. **Young dependency ratio is strongly negatively correlated with Internet (r = -0.7785).** Countries with more young people have significantly lower internet penetration — a development gap, not a behavioral effect.

3. **GDP and Internet are moderately correlated (r = 0.4531)** — potential multicollinearity, but VIF analysis (not shown) confirms it remains below the threshold of 5.

4. **Tourism has virtually zero correlation with Mobile (r = 0.0658)** — tourists buying local SIMs does not materially affect national subscription statistics.

### 3.3 Hypothesis Tests

#### F-Test: Two-Sample for Variances — ASEAN vs Non-ASEAN Mobile Subscriptions

| | ASEAN | Non-ASEAN |
|---|---|---|
| Mean | 124.20 | 116.55 |
| Variance | 762.50 | 2,808.22 |
| Observations | 11 | 190 |
| df | 10 | 189 |
| **F** | **3.68** | |
| F Critical (0.05) | ~1.90 | |

**Conclusion:** F = 3.68 > F_crit = 1.90. **Reject H₀: variances are unequal.** Non-ASEAN countries have significantly more dispersion in mobile subscriptions. This is expected — non-ASEAN includes the full range from South Sudan to Macao.

#### T-Test: Two-Sample Assuming Unequal Variances — ASEAN vs Non-ASEAN

| | ASEAN | Non-ASEAN |
|---|---|---|
| Mean | 124.20 | 116.55 |
| Variance | 762.50 | 2,808.22 |
| Observations | 11 | 190 |
| Pooled Variance | 2,705.42 | |
| Welch df | 14.7 | |
| **t Stat** | **0.83** | |
| P(T≤t) two-tail | 0.4171 | |
| t Critical two-tail | 2.1448 | |

**Conclusion:** |t| = 0.83 < t_crit = 2.14. **Fail to reject H₀.** There is no statistically significant difference in mean mobile subscription density between ASEAN and non-ASEAN countries. The 7.65-point difference is attributable to sampling variation, not a structural regional effect.

#### Hypothesis Test: μ_ASEAN > 100 (One-Tailed)

| | Value |
|---|---|
| H₀: μ_ASEAN ≤ 100 | — |
| H₁: μ_ASEAN > 100 (one-tailed) | — |
| Sample Mean | 124.20 |
| Standard Error | 8.33 |
| **t Stat** | **2.91** |
| t Critical (α=0.05, df=10) | 2.228 |

**Conclusion:** t = 2.91 > t_crit = 2.228. **Reject H₀.** ASEAN countries have mobile subscription density significantly above 100 per 100 people. This provides statistical evidence for the "more SIMs than people" phenomenon in Southeast Asia.

---

## 4. Proposed Solution / Changes

### 4.1 Simple Linear Regression (Excel ToolPak Style)

**Model:** Mobile_subs_100 = β₀ + β₁·Internet_pct

| REGRESSION STATISTICS | |
|---|---|
| Multiple R | 0.4120 |
| R Square | 0.1697 |
| Adjusted R Square | 0.1655 |
| Standard Error | 47.4215 |
| Observations | 201 |

| ANOVA | df | SS | MS | F | Sig F |
|---|---|---|---|---|---|
| Regression | 1 | 91,474.64 | 91,474.64 | 40.6771 | <.0001 |
| Residual | 199 | 447,511.43 | 2,248.80 | | |
| Total | 200 | 538,986.07 | | | |

| Coefficients | Coef | Std Err | t Stat | P-value | Lower 95% | Upper 95% |
|---|---|---|---|---|---|---|
| Intercept | 61.2216 | 9.3594 | 6.5412 | <.001 | 42.7649 | 79.6783 |
| Internet_pct | 0.7918 | 0.1242 | 6.3779 | <.001 | 0.5470 | 1.0367 |

**Equation:** Mobile = 61.22 + 0.7918 × Internet_pct

**Interpretation:** For every 1% increase in internet penetration, mobile subscriptions increase by 0.79 per 100 people. A country with 100% internet penetration would be predicted to have 140.4 mobile subscriptions per 100 people. Internet alone explains 17.0% of the variance in mobile adoption.

### 4.2 Multiple Linear Regression (Excel ToolPak Style)

**Model:** Mobile = β₀ + β₁·GDP + β₂·Internet + β₃·Landlines + β₄·Tourism + β₅·YoungDep

| REGRESSION STATISTICS | |
|---|---|
| Multiple R | 0.4279 |
| R Square | 0.1831 |
| Adjusted R Square | 0.1622 |
| Standard Error | 47.5168 |
| Observations | 201 |

| ANOVA | df | SS | MS | F | Sig F |
|---|---|---|---|---|---|
| Regression | 5 | 98,706.55 | 19,741.31 | 8.7434 | <.0001 |
| Residual | 195 | 440,279.52 | 2,257.84 | | |
| Total | 200 | 538,986.07 | | | |

| Coefficients | Coef | Std Err | t Stat | P-value | Lower 95% | Upper 95% |
|---|---|---|---|---|---|---|
| Intercept | 92.1651 | 25.4623 | 3.6197 | <.001 | 41.9535 | 142.3766 |
| GDP_per_capita | -0.000095 | 0.000119 | -0.7930 | 0.4288 | -0.0003 | 0.0001 |
| **Internet_pct** | **0.6090** | **0.2022** | **3.0121** | **<.003** | **0.2103** | **1.0078** |
| Landlines_100 | 0.1100 | 0.2874 | 0.3828 | 0.7023 | -0.4568 | 0.6769 |
| Tourism_arrivals | -0.0000003 | 0.0000004 | -0.7256 | 0.4689 | -0.0000 | 0.0000 |
| Young_dep_ratio | -0.4095 | 0.3007 | -1.3617 | 0.1748 | -1.0025 | 0.1835 |

**Equation:** Mobile = 92.17 - 0.000095×GDP + 0.6090×Internet + 0.1100×Landlines - 0.0000003×Tourism - 0.4095×YoungDep

### 4.3 Interpretation of Findings

| Predictor | β Coefficient | t-value | Significant? | Business Implication |
|---|---|---|---|---|
| **Internet_pct** | **+0.6090** | **3.01** | ✅ YES (p<0.003) | For every +1% internet penetration → +0.61 mobile subs/100. **This is the only meaningful driver.** |
| GDP_per_capita | -0.0001 | -0.79 | ❌ No (p=0.43) | Once you control for internet access, income doesn't matter. A poor country with good internet will have high mobile adoption. |
| Landlines_100 | +0.1100 | +0.38 | ❌ No (p=0.70) | Fixed-line infrastructure is irrelevant to mobile adoption. No evidence of complementarity or substitution. |
| Tourism_arrivals | -0.0000003 | -0.73 | ❌ No (p=0.47) | Tourists buying local SIMs has no detectable impact on national statistics. |
| Young_dep_ratio | -0.4095 | -1.36 | ❌ No (p=0.17) | Younger populations do not significantly drive higher mobile penetration after controlling for internet. |

**F-test:** F(5,195) = 8.7434, p < 0.0001. The model as a whole is significant — but only one predictor individually achieves statistical significance.

**R² = 0.1831:** The model explains 18.3% of the variance in mobile subscriptions. This is a weak model by predictive standards — **82% of the variation is unexplained by these five economic variables.**

### 4.4 Why GDP Is NOT Significant

The most surprising finding is that GDP per capita has no significant relationship with mobile adoption. This contradicts economic intuition (richer countries should buy more phones). Three explanations:

1. **Mobile phones are an essential good, not a luxury.** Even the poorest countries (Burundi, GDP = $234) have 56 mobile subscriptions per 100 people. Mobile penetration reaches saturation before income effects become visible.

2. **Internet access mediates the GDP effect.** GDP and Internet are correlated (r = 0.45). When both are in the model, Internet captures the shared variance, leaving GDP with no unique explanatory power.

3. **Pre-paid SIM cards make access nearly free.** In most developing countries, a SIM card costs pennies. The marginal cost of an additional subscription approaches zero.

### 4.5 95% Confidence Intervals

| Group | n | Mean | 95% CI | Interpretation |
|---|---|---|---|---|
| ASEAN | 11 | 124.20 | [105.65, 142.75] | ASEAN mobile penetration is significantly above 105 but could be as low as 106 |
| Global | 201 | 116.97 | [109.80, 124.15] | Global mean is tightly estimated (±7 points) |
| ASEAN > 100 | 11 | 124.20 | One-tailed t=2.91 > 2.228 | Confirmed: "more SIMs than people" is true ❌ |

---

## 5. Recommendations

### 5.1 For Telecom Operators

1. **Investment priority: Internet infrastructure > GDP-rich markets.** A developing country with expanding internet access (e.g., Cambodia at 69% and growing) offers better ROI than a stagnant high-GDP market with saturated internet.

2. **Ignore tourism for demand forecasting.** The near-zero coefficient (β = -0.0000003, p = 0.47) means tourist SIM sales are noise, not signal. Base capacity planning on resident population and internet growth rates.

3. **Don't chase the "youth premium."** Young populations do not drive statistically higher mobile adoption after controlling for internet. Marketing budgets targeting youth may have lower marginal returns than assumed.

4. **The ASEAN premium is real but not universal.** ASEAN mean is 124.2, above global mean 117.0, but the difference is not statistically significant (p = 0.42). Thailand (187.8), Malaysia (155.8), and Singapore (155.7) drive the regional average up; Myanmar (63.4) and Laos (64.4) bring it down.

### 5.2 For Policymakers

1. **Internet access policy IS mobile policy.** The single most effective lever for increasing mobile penetration is expanding internet access. Every 1% increase in internet penetration predicts 0.6 more mobile subscriptions per 100 people.

2. **GDP growth alone won't close the digital divide.** The non-significant GDP coefficient means that economic growth without internet infrastructure investment will not meaningfully increase mobile adoption.

3. **Address the 82% unexplained variance.** The model's low R² (0.18) strongly suggests that factors beyond economics drive mobile adoption. These likely include:
   - **Regulatory quality** (spectrum allocation, licensing)
   - **Market competition** (number of operators, MVNO policy)
   - **Cultural factors** (multiple-SIM behavior, smartphone culture)
   - **Geographic factors** (island nations, population density)

### 5.3 Limitations and Future Research

| Limitation | Impact | Suggested Improvement |
|---|---|---|
| Cross-sectional (single time point) | Cannot detect trends or causality | Use panel data (2010–2025) with fixed effects |
| Only 5 predictors | R² = 0.18, 82% unexplained | Add regulatory, competitive, and cultural variables |
| OLS assumes linearity | May miss threshold effects | Test quadratic terms for GDP and Internet |
| ASEAN subsample n = 11 | Wide confidence intervals (±19 points) | Supplement with national telecom regulator data |
| Tourism measured as raw arrivals | Doesn't capture SIM purchase behavior | Use tourist expenditure or SIM registration data |

---

## Appendix A: Excel-Ready Analysis Output

All statistical output above is formatted to match Excel's Data Analysis ToolPak. The regression tables, ANOVA, F-tests, and T-tests can be directly pasted into an Excel workbook.

---

## Appendix B: ASEAN Country Profiles

| Country | Mobile/100 | GDP/capita | Internet% | Landlines | Tourism(M) | YoungDep |
|---|---|---|---|---|---|---|
| Brunei | 135.2 | 34,960 | 98.1 | 28.3 | 0.3 | 30.2 |
| Cambodia | 119.5 | 2,730 | 68.9 | 0.3 | 1.3 | 52.7 |
| Indonesia | 130.5 | 5,190 | 69.2 | 4.2 | 4.0 | 43.1 |
| Laos | 64.4 | 2,330 | 64.4 | 17.0 | 0.9 | 53.4 |
| Malaysia | 155.8 | 13,060 | 97.7 | 24.2 | 4.3 | 32.5 |
| Myanmar | 63.4 | 1,240 | 44.0 | 1.0 | 0.4 | 38.9 |
| Philippines | 120.0 | 4,130 | 77.9 | 4.8 | 1.5 | 47.2 |
| Singapore | 155.7 | 88,240 | 94.3 | 33.0 | 2.7 | 15.8 |
| Thailand | 187.8 | 7,770 | 89.5 | 4.3 | 6.7 | 20.9 |
| Timor-Leste | 106.9 | 1,420 | 44.0 | 0.2 | 0.1 | 48.1 |
| Vietnam | 137.7 | 4,650 | 78.1 | 2.4 | 3.8 | 31.3 |

---

*Report prepared using World Bank API data (live retrieval, July 8, 2026) and Python statistical computation. All regression coefficients verified against Excel Data Analysis ToolPak conventions.*