The Score That Does Not Exist
Social credit, reputation-scored lending, and the predatory machine China shut down while the West kept it running. Every claim verified against primary sources.
서방이 두려워한 사회신용점수의 실체, 그리고 중국이 멈춘 약탈적 대출 기계에 관한 검증 보고서
The unified numerical “social credit score” that Western media attributes to every Chinese citizen does not exist. Specialist researchers at MERICS, Yale Law School’s Paul Tsai China Center, MIT Technology Review, and WIRED all reach the same conclusion: the real system is fragmented, mostly aimed at businesses, and built around court judgment-defaulter lists, not a citizen score.
The reputation-scored lending machinery that inspired the fear was pioneered in Western fintech: Zopa in London in 2005, Prosper in San Francisco in 2006. It flowed into China through platforms seeded and staffed by Goldman Sachs alumni. When that machinery turned predatory, Beijing regulated it to zero: over 6,000 peer-to-peer lenders at the 2015 peak, zero operating platforms by November 2020, and the largest IPO in history suspended to force lenders to hold real capital.
The West took the opposite path. A former US Treasury Secretary now chairs the private equity firm that owns a subprime installment lender a former employee described to the Washington Post as “a way of monetizing poor people.” This is the last quotation this paper takes from that source, and the only one it needs.
- No unified citizen score exists in China; the PBOC refused to license the private scoring pilots of Alibaba and Tencent in 2017 to 2018, and Tencent’s consumer score was pulled almost immediately after launch.
- Reputation-scored peer lending launched in the West first: Zopa (UK, March 2005), Prosper (San Francisco, 2006), LendingClub (2006 to 2007). China’s first P2P platform, PPDAI, followed in 2007.
- Goldman Sachs led Alibaba’s first institutional round in October 1999; Alibaba and Tencent are both Cayman Islands exempted companies, confirmed verbatim in their SEC filings.
- China’s crackdown was total: P2P platforms fell from over 6,000 to zero by November 2020, and Ant Group’s roughly $34 to 37 billion IPO was suspended on 3 November 2020, days after new draft rules required online lenders to fund at least 30 percent of joint loans. Ant was funding about 2 percent.
- The same predatory model still runs in the West, from Mariner Finance to buy-now-pay-later, while Goldman’s Abacus CDO settlement ($550 million, 14 July 2010) remains the reference case for what the machine does when unregulated.
Every factual claim in this paper survived a source-level verification pass against SEC filings, regulatory records, court outcomes, and first-tier journalism. Three claims commonly attached to this story failed verification and are excluded: that the 2008 crash was deliberately engineered (unsupported), that Alibaba’s president is a former Goldman Sachs CEO (false; Michael Evans was a vice chairman), and that BlackRock and Microsoft were “hired by the G7” to counter the Belt and Road (garbled; the verifiable facts are the G7’s PGII initiative of 2022 and the March 2025 BlackRock-led consortium agreement for CK Hutchison’s ports at a $22.8 billion enterprise value, a deal later stalled and overtaken by Panamanian court action). What remains is what survives hostile cross-examination.
The score that does not exist
The single nationwide citizen score is a Western invention about China, not a Chinese invention about citizens.
I spent years watching Western coverage describe a dystopia where every Chinese citizen carries a number that opens or closes their life. Then I read the people who actually study the system. Vincent Brussee at MERICS, in his February 2022 analysis, calls the citizen-score narrative “more bogeyman than reality” and documents a system that is lowly digitalized, highly fragmented, and aimed primarily at businesses. Jeremy Daum of Yale Law School’s Paul Tsai China Center, who has translated the underlying Chinese regulations for years at China Law Translate, finds that meaningful restrictions attach almost entirely to one instrument: the court judgment-defaulter list, which is a debt-enforcement tool, not a behavioral score. MIT Technology Review in 2022 and WIRED in 2019 independently reached the same conclusion.
Here is the part the coverage skips. There were real consumer scoring programs in China. They were private. In 2015 the People’s Bank of China permitted eight private firms, including Ant’s Sesame Credit and a Tencent unit, to pilot personal credit scoring. In 2017 to 2018 the PBOC reviewed the pilots, cited conflicts of interest, and refused to license any of them, creating the state-backed Baihang Credit instead. Tencent’s consumer score was pulled almost immediately after a brief launch. Sesame Credit survives as what it always functionally was: a loyalty and rewards program.
So the one entity in this story that examined a commercial citizen-scoring product and said no was the Chinese central bank. Hold that thought, because the next question is where the scoring-and-lending machinery came from in the first place.
Where the machine was actually built
Reputation-scored peer lending is a Western fintech invention that reached China two years after San Francisco.
The world’s first peer-to-peer lender was Zopa, launched in London in March 2005 by a founding team out of Egg, the internet bank Prudential plc created. Prosper launched in San Francisco in 2006, and its original model was precisely the thing Western audiences would later fear from Beijing: borrowers carried ratings and reputation data, and strangers priced their access to money against it. LendingClub followed in 2006 to 2007. China’s first P2P platform, PPDAI, appeared in 2007.
The timing matters. These platforms scaled into the 2008 crisis and its aftermath, when households on every continent needed credit and regulators were busy elsewhere. The algorithms were complex, the capital requirements were nonexistent, and the lenders had no skin in the game. The United States eventually forced its platforms under securities regulation. The model kept moving.
The men who carried it east
China’s big tech platforms were seeded, staffed, and structured by Wall Street, a fact recorded in their own SEC filings.
Henry Paulson ran Goldman Sachs as chairman and CEO from 1999 to 2006, and the Daily Telegraph reported in July 2008 that he had visited China more than 70 times in that career, building relationships with the Chinese elite. He was sworn in as US Treasury Secretary on 10 July 2006 and served to 20 January 2009, pressing China on financial opening through the Strategic Economic Dialogue he founded. His March 2008 “Blueprint for a Modernized Financial Regulatory Structure” was a domestic reorganization plan; the China-facing push ran through the Dialogue. Precision matters here, and the precise version is damning enough.
Goldman led Alibaba’s first institutional round in October 1999, roughly $5 million for a stake reported between 28 and 33 percent, with SoftBank’s separate $20 million following in early 2000. Alibaba Group Holding Limited is, per its own SEC filings, an exempted company incorporated under the laws of the Cayman Islands on 28 June 1999. Tencent Holdings is likewise a Cayman Islands exempted company. Tencent’s president, Martin Lau, came from Goldman Sachs Asia investment banking, where he was an executive director and COO of the telecom, media and technology group, and before that from McKinsey & Company. Alibaba’s president since August 2015, Michael Evans, was a Goldman Sachs vice chairman from February 2008 to December 2013 and chairman of its Asia operations.
McKinsey, for the record, paid $573 million to 47 states, DC, and five territories in February 2021 and $650 million to the US Department of Justice in December 2024 over its work turbocharging Purdue Pharma’s OxyContin sales, with a former senior partner pleading guilty to obstruction. Lau had no role in that work. But when someone tells you China’s tech giants are alien to the Western financial system, the org charts and the incorporation papers say otherwise. They are the Western financial system, offshore edition.
The companies Washington describes as instruments of the Chinese party-state are Cayman Islands corporations whose presidents came from Goldman Sachs.
Source basis: Alibaba & Tencent SEC filings · company officer biographies
What China did when the machine turned predatory
Faced with a metastasizing predatory lending sector, Beijing regulated it out of existence and stopped the largest IPO in history to do it.
By 2015, China had over 6,000 P2P lending platforms. The sector produced the Ezubao fraud: a Ponzi scheme that raised over 50 billion yuan, roughly $7.6 billion, from about 900,000 investors before collapsing. Founder Ding Ning received a life sentence on 12 September 2017. The regulatory response escalated from registration requirements to a November 2019 order converting P2P lenders into small-loan companies, and by November 2020 Chinese authorities reported the number of operating P2P platforms at zero. Not reduced. Zero.
Then came the main event. Ant Group’s dual Shanghai and Hong Kong listing was set to raise between $34.4 and $37 billion, the largest IPO ever attempted. Ant’s consumer lending arms, Huabei and Jiebei, were generating nearly 40 percent of its revenue in the first half of 2020, and Ant was funding only about 2 percent of the joint loans it originated with banks. The risk sat on the banking system; the fees sat with Ant. On 24 October 2020, Jack Ma stood up at the Bund Summit in Shanghai and attacked Chinese banking regulation and the Basel framework. On 2 November, regulators published draft rules requiring online micro-lenders to fund at least 30 percent of joint loans from their own capital. On 3 November, the Shanghai Stock Exchange suspended the IPO, days before listing. Ma then largely disappeared from public view for months.
You can read that sequence as authoritarian caprice, and much of the Western press did. Or you can read the rule: if you originate the loan, you hold real capital against it. That is not exotic. That is what bank regulation is for, applied to entities that had been enjoying bank economics without bank obligations.
What the West did instead
The same predatory model China shut down continues to operate in the West, run in part by the officials who managed the 2008 crisis.
Timothy Geithner was president and CEO of the Federal Reserve Bank of New York from 2003 to 2009 and Treasury Secretary after that. In March 2014 he became president of Warburg Pincus, the private equity firm whose fund acquired Mariner Finance, a subprime installment lender, in 2013. The Washington Post’s investigation of 1 July 2018 documented Mariner’s model of mailing unsolicited live checks to cash-strapped households and quoted a former manager trainee’s description of the business, cited in the summary above. Mariner disputes the predatory characterization. The Post’s reporting, and the business model it describes, are on the record.
The reference case for the machine’s behavior when unregulated remains Goldman’s Abacus 2007-AC1: the SEC alleged Goldman failed to disclose that the hedge fund helping select the mortgage portfolio was betting against it, and Goldman settled on 14 July 2010 for $550 million, then the largest SEC penalty against a Wall Street firm, acknowledging its marketing materials contained incomplete information. Sixteen years ago to the day, as it happens. Meanwhile buy-now-pay-later lending has spread through Western checkout pages with a fraction of the scrutiny China applied to Ant.
One more thread, worded exactly. The G7 launched the Partnership for Global Infrastructure and Investment in 2022 as its explicit answer to the Belt and Road. In March 2025 a BlackRock-led consortium agreed to acquire CK Hutchison’s Panama ports and a 43-port global portfolio at a $22.8 billion enterprise value, a transaction Washington framed as reclaiming the canal from Chinese influence; Beijing pushed back, and by early 2026 Panamanian court action had annulled the concessions. The West’s answer to Chinese infrastructure lending was to have an asset manager buy the infrastructure. The deal did not hold. The instinct is instructive.
The discipline that made the difference
China’s escape from both shock therapy and platform-lending capture came from the same habit: test small, verify, then scale.
Isabella Weber’s 2021 book How China Escaped Shock Therapy documents how close China came in the 1980s to big-bang price liberalization, and why it pulled back in favor of gradual dual-track reform. The famous phrase, crossing the river by feeling for the stones, is popularly credited to Deng Xiaoping; Weber traces its economic use to Chen Yun as early as 1950. Chen’s birdcage economy captured the same idea: the market bird flies freely, inside a frame that keeps it from flying into the window. The P2P shutdown and the Ant capital rule are the same method applied forty years later. Run the experiment. Watch what it does to people. Keep what works. Kill what preys.
Contrast the prevailing Western doctrine. Peter Thiel’s 2014 Wall Street Journal essay, adapted from Zero to One, is titled “Competition Is for Losers” and argues that monopoly is the condition of every successful business. China’s technology sector runs the other way: when DeepSeek released its R1 model on 20 January 2025, it was one of at least eight serious frontier AI labs in China, alongside Alibaba’s Qwen, ByteDance, Zhipu, Moonshot, Baidu, Tencent, and others. On 27 January 2025, Nvidia fell 17 percent in a single day, erasing roughly $589 to 593 billion in market value, the largest one-day loss for any company in history, in a selloff widely attributed to DeepSeek’s release. The country the West calls a monolith runs on brutal internal competition. The economy the West calls free is run by men who write essays against competing.
I assess that the “social credit score” narrative functions as projection. The scoring of human beings for credit access was built in London and San Francisco, financed into China by Goldman Sachs alumni through Cayman Islands vehicles, and shut down at scale by exactly one government: China’s. The West retained the machine, staffed it with its former crisis managers, and pointed at Beijing.
The recommendation is not admiration. It is precision. Any policymaker, fund executive, or journalist repeating the citizen-score claim should be asked for the primary Chinese regulatory document establishing it. Per the specialists who read those documents for a living, no such document exists. The burden of proof belongs on the claim, and the claim cannot carry it.
서방 언론에는 오래된 이야기가 하나 있습니다. 중국의 모든 국민이 하나의 점수를 부여받고, 그 점수가 기차표와 직장과 인생의 문을 열고 닫는다는 이야기입니다. 이 이야기는 수많은 기사와 방송을 통해 상식처럼 굳어졌습니다. 그러나 상식이 된 이야기일수록 근거를 물어야 합니다. 본 보고서는 그 근거를 일차 자료로 확인한 결과를 담았습니다. 미국 증권거래위원회 공시, 규제 당국의 공식 문서, 법원 판결, 그리고 해당 제도를 전문적으로 연구해 온 학자들의 분석이 그 자료입니다.
돌다리도 두들겨 보고 건너라 했습니다. 본 보고서가 두들긴 돌은 세 가지입니다. 첫째, 그 점수는 실재하는가. 둘째, 점수로 사람을 재는 대출 기계는 어디에서 만들어졌는가. 셋째, 그 기계가 사람을 해치기 시작했을 때 누가 멈추었고 누가 방치했는가.
첫째 돌부터 확인합니다. 전 국민에게 부여되는 단일한 수치 점수는 존재하지 않습니다. 독일 메르카토르 중국연구소의 2022년 2월 분석, 예일대 로스쿨 폴 차이 중국센터의 다년간 법령 번역 연구, 그리고 2019년과 2022년의 주요 기술 전문지 검증이 모두 같은 결론에 이르렀습니다. 실제 제도는 통합된 국민 점수가 아니라, 대부분 기업을 대상으로 하는 신용 기록과, 법원의 채무 불이행자 명단이라는 채권 집행 수단으로 이루어져 있습니다. 파편적이고, 전산화 수준도 낮으며, 개인의 일상을 채점하는 단일 체계와는 거리가 멉니다.
민간의 소비자 점수 실험은 실제로 있었습니다. 2015년 중국 인민은행은 알리바바 계열의 즈마신용과 텐센트 계열사를 포함한 민간 기업 여덟 곳에 개인 신용평가 시범 사업을 허용했습니다. 그러나 2017년부터 2018년에 걸쳐 인민은행은 이해상충을 이유로 단 한 곳에도 정식 인가를 내주지 않았고, 국가 주도의 바이항신용을 대신 설립했습니다. 텐센트의 소비자 점수는 출시 직후 철회되었고, 즈마신용은 사실상 마일리지 성격의 우대 제도로 남았습니다. 상업적 국민 점수를 심사하고 거부한 유일한 기관이 바로 중국의 중앙은행이었다는 사실은, 서방의 이야기와 정반대의 자리에 놓여 있습니다.
둘째 돌에서 이야기가 뒤집힙니다. 점수로 사람을 재어 돈을 빌려주는 기계는 서방 금융이 먼저 만들었습니다. 세계 최초의 개인 간 대출 회사는 2005년 3월 런던에서 출범한 조파이며, 창업진은 영국 푸르덴셜이 세운 인터넷 은행 에그 출신이었습니다. 2006년 미국 샌프란시스코의 프로스퍼는 차입자에게 등급과 평판 자료를 붙여 낯선 이들이 그 사람의 신용을 값 매기게 했습니다. 중국 최초의 개인 간 대출 회사는 그보다 늦은 2007년에야 등장했습니다.
이 기계를 동쪽으로 옮긴 손길도 기록에 남아 있습니다. 골드만삭스는 1999년 10월 알리바바의 첫 기관 투자를 주도했고, 알리바바와 텐센트는 미국 증권거래위원회 공시에 명시된 대로 모두 케이맨 제도에 설립된 회사입니다. 텐센트의 총재 류츠핑은 골드만삭스 아시아 투자은행 부문과 매킨지 출신이며, 알리바바의 총재 마이클 에번스는 골드만삭스 부회장을 지낸 인물입니다. 중국 공산당의 도구라 불리는 기업들의 회사 등기부와 임원 이력서는 서방 금융의 이름들로 채워져 있습니다.
셋째 돌은 가장 무겁습니다. 그 기계가 사람을 해치기 시작하자, 멈춘 쪽은 중국이었습니다. 2015년 절정기에 육천 곳이 넘던 개인 간 대출 회사는 단계적 규제 끝에 2020년 11월 영업 기준 영 곳이 되었습니다. 약 구십만 명에게서 오백억 위안을 끌어모은 이쭈바오 사기의 주범은 2017년 9월 무기징역을 선고받았습니다. 그리고 2020년 11월 3일, 역사상 최대 규모였던 앤트그룹의 상장이 거래 개시 며칠 전에 중단되었습니다. 마윈이 10월 24일 상하이에서 금융 규제를 공개 비판한 직후, 규제 당국이 공동 대출의 최소 삼십 퍼센트를 자기 자본으로 부담하라는 규정 초안을 내놓은 데 따른 것입니다. 당시 앤트그룹의 자기 부담은 약 이 퍼센트에 불과했습니다. 대출의 이익은 회사가 가지고 위험은 은행에 떠넘기던 구조에, 은행의 책임을 물은 것입니다.
같은 시기 서방은 반대의 길을 걸었습니다. 2008년 금융위기 당시 뉴욕 연방준비은행 총재였던 티머시 가이트너는 2014년 사모펀드 워버그 핀커스의 사장이 되었고, 그 펀드가 소유한 대부업체 매리너 파이낸스의 영업 방식은 2018년 7월 워싱턴포스트의 탐사 보도로 세상에 알려졌습니다. 형편이 어려운 가정에 수표를 우편으로 보내 빚을 지게 하는 방식이었습니다. 중국이 문을 닫은 기계가 서방에서는 위기를 관리하던 이들의 손으로 계속 돌아가고 있습니다.
세 개의 돌을 모두 두들긴 결론은 이렇습니다. 존재하지 않는 점수를 두려워하라는 이야기는, 실재하는 기계로부터 눈을 돌리게 하는 이야기였습니다. 사람을 점수로 재는 대출 기계는 서방에서 태어나 서방의 자본으로 동쪽에 이식되었고, 그것을 국가 차원에서 멈춘 정부는 지금까지 단 하나뿐입니다.
판단의 기준은 감정이 아니라 문서입니다. 전 국민 단일 점수를 규정한 중국의 일차 법령 문서를 제시할 수 있는 사람은, 그 문서를 직업으로 읽어 온 연구자들 가운데 아직 아무도 없습니다. 근거를 대야 할 쪽은 이야기를 퍼뜨린 쪽입니다. 주체강은 남의 지도가 아니라 스스로 확인한 물길을 따라 흐릅니다. 이 보고서가 그 물길의 돌 하나가 되기를 바랍니다.
Sources
Social credit system: Vincent Brussee, “China’s social credit score: untangling myth from reality,” MERICS, 11 Feb 2022 · Jeremy Daum, China Law Translate / Yale Law School Paul Tsai China Center, multi-year translations and analyses · Louise Matsakis, WIRED, 2019 · MIT Technology Review, 2022 · PBOC pilot authorization (2015), licensing refusal and Baihang Credit establishment (2017 to 2018).
P2P lending origins and China crackdown: Zopa launch, March 2005, founding team ex-Egg (Prudential) · Prosper, San Francisco, 2006 · LendingClub, 2006 to 2007 · PPDAI, 2007 · Ezubao: Xinhua via CNBC, life sentence for Ding Ning, 12 Sep 2017; c. 50 billion yuan from c. 900,000 investors · CBIRC/regulatory statements: P2P platforms at zero, Nov 2020.
Ant Group: IPO prospectus and exchange notices, Nov 2020 · Jack Ma Bund Summit speech, 24 Oct 2020 · draft online micro-lending rules (30 percent joint-loan funding floor), 2 Nov 2020 · Shanghai Stock Exchange suspension, 3 Nov 2020 · Ant H1 2020 revenue mix (credit c. 40 percent), per prospectus.
Corporate structures and officers: Alibaba Group Holding Ltd SEC filings (Cayman Islands exempted company, incorporated 28 Jun 1999; Goldman-led 1999 round; SoftBank 2000) · Tencent Holdings filings (Cayman Islands) · Martin Lau official biography (Goldman Sachs Asia; McKinsey) · Michael Evans official biography (Goldman Sachs vice chairman, Feb 2008 to Dec 2013).
US officials and firms: Henry Paulson: Goldman chairman and CEO 1999 to 2006; US Treasury 10 Jul 2006 to 20 Jan 2009; 70+ China visits per Daily Telegraph, Jul 2008; Blueprint for a Modernized Financial Regulatory Structure, 31 Mar 2008 · Timothy Geithner: NY Fed 2003 to 2009; Warburg Pincus president, Mar 2014 · Washington Post, “A way of monetizing poor people,” 1 Jul 2018 (Mariner Finance; company disputes characterization) · SEC v. Goldman Sachs (Abacus 2007-AC1), $550M settlement, 14 Jul 2010 · McKinsey opioid settlements: $573M (Feb 2021), $650M DOJ (Dec 2024).
Economic doctrine and AI: Isabella M. Weber, How China Escaped Shock Therapy, Routledge, 2021 (Chen Yun, stones phrase, 1950; birdcage economy) · Peter Thiel, “Competition Is for Losers,” WSJ, 12 Sep 2014 · DeepSeek R1 release, 20 Jan 2025; Nvidia one-day loss of c. $589 to 593 billion, 27 Jan 2025 (Reuters/LSEG; Forbes), causal attribution widely reported and disputed by at least one analyst · G7 PGII, 2022 · BlackRock-TiL consortium agreement for CK Hutchison ports, Mar 2025, $22.8B enterprise value; concessions annulled by Panamanian court action by early 2026.
주체강 · JCK-WP-2026-0714 · Unclassified · Prepared by Jesse James · Radical transparency: every entity named in this paper, including the author’s, is public.
주체의 흐름을 따라 · Navigate the Current
This report is a private strategic intelligence briefing. Every factual claim survived a source-level verification pass against SEC filings, regulatory records, court outcomes, and first-tier journalism. Claims that failed verification were excluded and are disclosed in the Method section above.