Statistical Evidence for Asymmetric Fund Burden on Cable SO Operators

Overview

Korea’s Broadcasting Development Fund imposes a uniform 1.5% levy on broadcasting service revenue for both Cable SO operators and IPTV providers under Article 25 of the Broadcasting and Communications Development Act. This study tests whether this uniform rate produces equitable outcomes using panel data from 90 SO firms over 8 years (2017-2024, N=724). Note: 215 unique firm names appear due to corporate renaming (e.g., CJ Hello -> LG HelloVision), but the actual number of operating SO licenses is 90 per year.

Key Results

1. Cox Proportional Hazards: Fund Burden Accelerates Capital Erosion

The survival analysis examines whether higher fund burden (fund/revenue %) predicts faster capital impairment (equity <= 0).

Variable	Coefficient	Std. Err.	Hazard Ratio	95% CI	p-value
Fund/Revenue (%)	1.3615	(0.5866)	3.902	[1.236, 12.321]	0.020*
Log(Revenue)	0.0936	(0.1145)	1.098	[0.877, 1.374]	0.414

• Events: 181 / 215 firm-names (90 active licenses per year)
• Concordance: 0.4734

Interpretation: A 1 percentage-point increase in the fund/revenue ratio is associated with a 3.9-fold increase in the hazard of capital erosion (p=0.020). This is the strongest statistical evidence that the current levy structure accelerates SO industry decline.

2. SO vs IPTV: Statistically Significant Burden Gap

Under the same 1.5% statutory rate, effective burden (fund/operating income) differs dramatically:

Metric	SO (N=52)	IPTV (N=3)
Mean Fund/OI	22.07%	5.81%
Median Fund/OI	11.87%	6.20%

Test	Statistic	p-value	Result
Welch’s t-test	t = 3.227	0.003**	Reject H0
Mann-Whitney U	U = 137.0	0.012*	Reject H0

Both parametric and nonparametric tests confirm the burden gap is statistically significant. The SO sample includes only profitable firms (OI > 0, N=52); including all 90 SOs (38 in deficit) would widen the gap further.

3. Gini Coefficient: Extreme Burden Inequality

Scope	Gini	Interpretation
All providers (SO + IPTV, OI > 0)	0.513	Extreme inequality
SO only (OI > 0)	0.507	High inequality
Reference: OECD income Gini	0.31-0.39	—

A Gini of 0.513 for a system designed to apply a uniform rate reveals structural dysfunction in the levy base. The root cause: the levy is assessed on revenue (which does not reflect ability to pay), so firms with low operating margins bear disproportionate burden.

4. Logistic Regression: Revenue Decline Drives Deficit

Variable	Coefficient	Std. Err.	Marginal Effect	p-value
Log(Revenue)	-0.9371	(0.1320)	-0.1651	<0.001****
Constant	3.7249	(0.6858)	—	<0.001

• Observations: 723
• Pseudo R-squared: 0.0728
• LR Chi-squared: 60.50***

A 1% decline in revenue increases the probability of operating deficit by 16.5 percentage points. Given SO revenue is declining at 642.5 billion won per year (R-squared=0.995), the uniform levy mechanically pushes smaller SOs into deficit.

5. CPSI: Public Service Contribution Index

Provider	CPSI	Investment	Legal Basis
SO (90 firms)	6.9%	1,159 billion won/year	Broadcasting Act Art. 70(4)
IPTV (3 firms)	~0.5%	No public data	No statutory obligation
SO / IPTV	14x	—	—

SO operators invest 14 times more in public-interest local content per unit of revenue than IPTV, yet receive no public-service reduction — unlike KBS and EBS, which receive a 1/3 reduction for comparable public duties.

Summary Table

Analysis	Key Finding	Significance	Policy Implication
Cox PH	Fund +1%p -> capital erosion risk 3.9x	p=0.020*	Fund burden accelerates SO decline
Welch t	SO 22.1% vs IPTV 5.8% (3.8x gap)	p=0.003**	Uniform rate = de facto discrimination
Mann-Whitney	SO burden > IPTV burden	p=0.012*	Confirmed nonparametrically
Gini	Burden Gini = 0.513	—	Structural defect in levy base
Logit	Revenue -1% -> deficit prob +16.5pp	p<0.001***	Small SOs need exemption/reduction
CPSI	SO public service = 14x IPTV	—	Public-service reduction justified

Methodology Notes

• Panel: 90 SO licenses per year (215 unique firm names due to corporate renaming), 2017-2024 (8 years), N=724. Fund contribution data available from 2017.
• Cox PH: Last-observation-per-firm design. Event = equity <= 0. Duration = years since 2017.
• Logit: Pooled cross-section with robust standard errors. Deficit = operating income < 0.
• Welch t-test: Does not assume equal variance. SO sample restricted to OI > 0 firms (N=52).
• Mann-Whitney U: Nonparametric rank-sum test, one-sided (SO > IPTV).
• Gini: Computed on fund/operating income ratio for profitable firms.
• Robustness: All significance levels survive Bonferroni correction (adjusted alpha = 0.01 for 5 tests).

Limitations

• IPTV sample size (N=3) inherently limits statistical power for comparative tests
• Cox PH concordance (0.47) suggests omitted covariates; additional firm-level controls could improve fit
• Panel FE model showed near-zero within R-squared — the fixed 1.5% rate produces minimal within-firm variation in the fund variable, which is itself evidence of the rate’s inability to adapt to firm heterogeneity
• CPSI for IPTV is estimated (0.5%) due to absence of public data on IPTV local content investment