17th Market Economy Colloquium: Do Circular Economy Policies Reduce Waste Generation?

17th Market Economy Colloquium

Date and Time: 11:00 a.m., September 18, 2026

Venue: Pureun Hall

Topic: Do Circular Economy Policies Reduce Waste Generation?

Presenter: Giman Kim, Secretary General of the Citizens Forum for Good Regulation

Discussants: Jaewook Ahn, Professor Emeritus at Kyung Hee University; Sung-no Choi, President of the Center for Free Enterprise (CFE); Yoonseok Jeong, Professor at Myongji College; and four others

Do Circular Economy Policies Reduce Waste Generation?

Giman Kim, Secretary General of the Citizens Forum for Good Regulation

1. Research Background and Problem Awareness

■ The “Recycling Paradox” and the Problem Raised

A phenomenon has been observed in which the absolute amount of waste generated is actually increasing despite advances in resource circulation and circular economy policies. Existing policies have treated “raising the recycling rate” as a core performance indicator, but there has been a lack of rigorous empirical verification as to whether a higher recycling rate actually leads directly to a fundamental reduction in waste generation. Some argue that recycling gives consumers and firms a kind of environmental moral licensing, thereby justifying original consumption and waste discharge.

■ Limitations of Existing Research

① Linear, average-centered analysis

By relying only on aggregate national or regional data (total averages), prior studies failed to properly reflect potential confounding factors such as income, consumption structure, and industrial structure, as well as composition effects.

② Distortion in the recycling-rate indicator

The metric is often measured based on the volume brought into recycling companies, creating a gap from the actual volume of recycled raw materials produced and used.

③ Elimination of the multidimensional data structure

Waste data have a multilayered structure in which two independent dimensions—“region (si, gun, gu)” and “item category (paper, plastic, food waste, glass, metal)”—intersect. Yet this structure has often been simplified into a single dimension, causing aggregation bias.

■ Research Objective

This study aims to test for the existence of Simpson’s Paradox and to identify the actual structural and micro-level causal relationship between recycling rates and waste generation.

2. Data and Research Methodology

■ Data for Analysis

① International panel

* 50 OECD countries × 9 years (2015–2023)

* Total of 408 observations

② Domestic si-gun-gu panel

* Based on raw data from the Ministry of Environment: 229 si, gun, and gu × 10 years (2015–2024)

* A balanced panel with a total of 2,290 observations

③ Domestic item-specific panel

* Five major categories of household waste: paper, plastic, food waste, glass, and metal

* si-gun-gu × year × item

* Total of 11,435 observations

■ Main Research Methodologies

① Fixed Effects Model

This identifies within-unit variation over time by controlling for unit-specific time-invariant characteristics.

② Mundlak Within/Between decomposition

This explicitly separates and estimates the temporal change effect within the same unit (Within Effect) and the structural level difference effect across units (Between Effect).

③ Cross-Classified Multilevel Model (CCMM)

By modeling the complex data structure in which regions (spatial hierarchy) and items (material hierarchy) intersect, this overcomes the problem of underestimated standard errors and improves the efficiency and statistical power of parameter estimation.

3. Main Empirical Findings

(1) Macro and aggregate-level analysis: reproduction of the “recycling paradox” at the aggregate level

■ International panel analysis (50 OECD countries)

At the aggregate level, a significant positive (+) relationship was observed between the recycling rate and per capita waste generation (Mundlak Within β = +0.0050, p < 0.05). This reflects the fact that both variables rose together due to common economic factors such as income growth and expanded consumption.

■ Domestic aggregate analysis by si-gun-gu

The fixed-effects regression analysis of the si-gun-gu panel likewise showed a positive (+) relationship in which an increase in the recycling rate within the same si-gun-gu was accompanied by an increase in waste generation (β = +0.005, p < 0.01). In the quadrant analysis of internal change within si-gun-gu as well, the “paradox zone” (upper right), where both recycling and waste increased simultaneously, accounted for the largest share at 45%.

(2) Cross-Classified Multilevel Model (CCMM) analysis: verification of “Simpson’s Paradox” and identification of the actual effect

■ Variance decomposition (ICC)

A total of 71.1% of the variation in waste generation was determined by the item dimension, while the explanatory power of the si-gun-gu (regional) dimension was only 5.9%. This shows the need for analysis that reflects structural heterogeneity across item categories.

■ Verification of Simpson’s Paradox

In the integrated CCMM model (M1), which reflects the multilayered si-gun-gu × item structure, the Within Effect of the recycling rate turned negative (-), completely reversing the sign (β = -0.0023, p < 0.001).

In other words, the positive (+) relationship observed at the aggregate level was empirically shown to be not a failure of recycling policy but a product of aggregation bias, in which level differences across regions and infrastructure (Between Effect) obscured the actual waste-reducing effect within item categories (Within Effect).

(3) Heterogeneity by item category and decomposition of mechanisms

■ Plastic: the strongest reduction effect

The Within coefficient was -0.017 (p < 0.01), showing the strongest suppressing effect on waste generation among the five categories. This indicates that establishing a plastic recycling system contributes substantially not only to post-use processing but also to reducing waste generation at the source.

■ Glass and paper

Glass showed a meaningful waste-reduction effect with a Within coefficient of -0.004 (p < 0.01), while paper showed -0.003 (p < 0.01).

■ Food waste: the main driver of the paradox

The Within Effect was only modest at -0.003 (p < 0.1), whereas the Between Effect was very high at +0.016 (p < 0.01). The structural characteristic that regions with higher income and better infrastructure also generate more food waste was the key factor driving the paradox in aggregate-level analysis.

■ Metal: reflecting market characteristics

Both the Within (+0.005) and Between (+0.005) effects were positive (+). Metal cans and similar items have a market structure in which recovery is economically attractive, so increased consumption leads to simultaneous increases in both waste generation and recycling volume.

4. Conclusion and Policy Implications

① A fundamental shift in the performance evaluation system

It is necessary to move away from evaluating performance with a single indicator centered on the aggregate recycling rate and instead adopt a multidimensional performance evaluation system that reflects heterogeneity by item category and Within performance—that is, the waste-generation suppression rate for each item category.

② Differentiated strategies tailored to item characteristics

▲ Plastic

Since it has the greatest potential for suppressing waste generation, policies should combine the creation of demand for recycled raw materials, expansion of processing infrastructure, and preemptive waste-reduction measures.

▲ Food waste

Because the recycling rate has already reached saturation (about 89–93%), policy should place greater emphasis on source reduction rather than simple expansion of recycling.

③ Improvement of substantive indicators and a shift toward waste suppression

The OECD standard for Actual Recycling should be adopted to ensure measurement accuracy, and the ultimate goal of circular economy policy should be redefined not as simply increasing recycling, but as achieving an “absolute reduction in waste generation.”

---

Original title: 제17회: 순환경제 정책은 폐기물 발생을 감소시키는가?

Author: Market Economy Colloquium

Date: 2026-09-18

Source: https://www.cfe.org/bbs/bbsDetail.php?cid=collo&pn=1&idx=29550