Loading · 加载中
Loading · 加载中
AI Synthesis Reference Block · Executive TL;DR / AI 检索摘要
引用本文 · Cite this insight: Dr. Tong Yin（殷彤博士） (2026-09-29). When Scale Overrides Quality: The Systemic Costs in Higher Education and Industry / 《当规模压倒质量：高等教育与工业体系的系统性代价》. InsightBridge Global Intelligence. https://intelligence.insightbridge.global/articles/when-scale-overrides-quality-systemic-costs-higher-education-industry — Series: deep-analysis
规模化推动了现代繁荣：它扩大教育机会，降低某些产品的单位成本，让更多人获得稳定供应。但规模的收益有条件。维护质量所需的人才、判断力和制度，未必能与招生、论文或产量同步增长。本文考察高等教育、科研和制造业中的这一张力。教育普及提高了学历水平，教学能力与毕业生就业的压力却使部分文凭的价值更难判断；研究发表量激增，新颖性与可重复性的证据提示我们，不能把篇数等同于发现；工业案例既显示检测、披露与品控的失灵，也说明规模本身并非罪魁。人口变化与技能需求正在促使调整。本文提出一种多元卓越的方向：更有选择性且重质量的研究机构、受尊重的教学与职业路径，以及按质量能力安排扩张节奏的企业。这里的匠人精神是可验证的严谨，而非怀旧。
战后增长的一项重要组织逻辑，是以管理能力支撑大规模生产和范围经济。钱德勒（Alfred Chandler）的研究解释了规模为何能够提高供给能力和一致性。在可以标准化的环节，重复操作、设备投入、缺陷反馈与统一规格可以相互促进。这些收益不应被抹去。增加入学名额、扩大研究参与、增加产品供应，也可能创造真实机会。本文要问的不是规模抽象地好还是坏，而是扩张过程中哪些支撑质量的条件必须保全。
教育体现了限制。机器每小时或可多生产同样的部件，教师细读学生论证、导师判断陌生问题，却无法按同样方式加速。鲍莫尔与鲍恩所讨论的成本病指出，其他部门生产率提高时，教育等劳动密集型个人服务的相对成本容易上升。这不等于教育不能提高效率；它提醒我们，若把座位当作标准件计数，可能看不见每位学习者实际获得的关注减少。
质量也不只是个别工匠的品德，而是选拔、训练、监督、测量、反馈，以及发现缺陷时有权停下流程的制度属性。戴明十四点的第11点要求取消针对员工的数量定额和针对管理层的数字目标。其要义并非拒绝数据，而是防止易计数的替代指标取代工作的本来目的。大学可能同时提高招生量与论文量，却忽略学习和思想冒险；工厂也可能按期交货，却削弱发现缺陷的能力。
丰田章男的证词为扩张节奏提供了一个严峻的管理问题：增长可能快于人才和组织的培育。本文检验的是一个命题，而非宣称普遍定律：如果规模超过机构维持质量的能力，其根基就可能变弱。规模与质量同步提升的反例，应当帮助我们限定这一命题，而不是被排除在讨论之外。关键比较，是有制度能力支撑的扩张，与耗损自身质量能力的扩张。
马丁·特罗（Martin Trow）的框架把同龄群体入学率低于约15%的体系称为精英阶段，15%至50%为大众阶段，超过50%为普及阶段。这是分析分类，不是质量排行。洪堡大学理念强调教学与研究统一、学术自由和大学自治；它形成时的高等教育远小于今天。据美国国家教育统计中心，1899至1900学年美国高等教育在校生相当于18至24岁人口的2.3%，1949至1950学年为15.2%，1989至1990学年为51.4%；美国在一个世纪内依次跨过了特罗的三个门槛。普及教育无法简单地让每个席位都复制密切的研究学徒关系，但也不应美化昔日稀缺；机会的扩大确实具有社会价值。
变化的幅度十分明显。联合国教科文组织指出，全球高等教育毛入学率从2000年的19%升至2022年的42%；在校生从2000年的1.81亿增至2010年的2.56亿。经合组织成员国25至34岁人口的高等教育完成比例，从2000年的27%升至2024年的48%。美国国家科学基金会记录，2024年美国授予58,131个研究型博士学位。这些指标所指人口不同，不能拼成同一条时间序列；合在一起，它们显示一个体系需要同时为更多人颁授资格、提供教学、衔接就业，还须维持严肃研究。
成本与用人结构，使人难以无条件地赞颂扩张。堪萨斯城联储指出，1980至2004年美国平均学费及杂费从1,289美元升至7,122美元，年增约7%，同期核心消费物价年增约4%；此后学费通胀有所放缓。2026年第二季度，纽约联储家庭债务口径的学生贷款余额为1.65万亿美元；LendingTree归纳的美联储G.19较宽口径约为1.86万亿美元。两种定义不同，并非同一个精确定义的总额发生矛盾。美国大学教授协会报告，非终身教职占比从1987年的47%升至2023年的68%，终身及终身轨教职则从53%降至32%。非终身聘任本身不能证明教学差，但持续指导和组织延续性因此处在另一种制度环境中。
学习成效同样不能轻率概括。Arum与Roksa对24所院校逾2,300名学生的研究发现，至少45%的人在入学前两年的大学学习评估（CLA）中未表现出具有统计显著性的进步。结论只适用于该测量和样本，并非说所有学习都停滞。不过，它足以促使学校用可展示的能力成长，而非只用招生和毕业人数，来支持自身的教育价值主张。所谓选拔，也应包括学生、课程与培养目标的诚实匹配，而不是按社会出身排斥有能力者，或只承认一种学术道路。
就业数据提供另一组需要限定的信号。纽约联储数据显示，2026年第二季度22至27岁近期毕业生失业率约5.6%，就业不足率42%；2019年3月失业率为3.6%，全部大学毕业生就业不足率则为33.7%。克利夫兰联储认为，大学学历工资溢价较2000年下降约10%。然而平均收益仍然可观：纽约联储的回报分析估计，典型大学毕业生每年多挣逾32,000美元，约有68%的溢价。平均溢价不等于每个专业或借款人的个人回报。公众信任又是另一指标：Higher Ed Dive报道的盖洛普数据显示，对高等教育有高度信心的美国人从2015年的57%降至2023至2024年的36%，2025年回升到42%。
因此，谨慎的判断不是“大众教育失败”。机会和学历水平提高了，但教学能力、可负担性与劳动力市场吸纳能力，并未在所有学校和领域同步增长。规模与质量的落差分布不均。应对之道是拿出更清楚的学习与匹配证据，而不是为了排他而重返排他。
科研也呈现大规模生产的外观。美国白宫科技政策办公室报告称，年度学术论文从1996年不足100万篇增至2020年的290万篇，约为原来的三倍。行业追踪机构估计2024年约有514万篇；这只是行业估算，不应视为与官方序列完全同口径的延长线。产出扩大可能反映研究人员增多、参与更广，也包含较小体系本会遗漏的贡献；但论文数量并不能自动证明重要发现同比增长。
一项有影响力的测量提示了问题。Park、Leahey与Funk考察跨越六十年的4,500万篇论文和390万项专利，发现新成果相对于既有知识的颠覆性指标下降。不过，解释存在争论。《定量科学研究》的批评文章认为，引用量膨胀可以解释很大一部分测量下降。因此这个指标是需要追查的线索，不足以断言创造力消失。Chu与Evans研究241个学科、9,000万篇论文之间的18亿次引用，发现领域过大时，注意力可能向已有经典集中。这提供了一种规模成本机制：每项新贡献都要在更密集的注意力市场争取被看见。
还有理由区分论文篇数与每位研究者的发现效率。Bloom、Jones、Van Reenen与Webb估计，要维持摩尔定律所说的芯片密度翻倍，如今所需研究人员已超过20世纪70年代初的18倍。这是特定产业的结果，不能套用于所有学科。Benjamin Jones讨论知识负担：20世纪重大创新出现时的平均年龄上升约六岁，专业化与团队合作也增加。发现变得更晚、更依靠集体，可能反映知识前沿深化，而非个人才华消失；但它要求机构提供钻研时间、跨领域协作，以及探问非惯常问题的空间。
科研诚信的指标尤其需要克制。《自然》记录2023年逾10,000篇论文被撤稿；撤稿增加既可能体现生产环节的问题，也可能体现发现问题的能力提高。在限定范围的心理学重复实验中，开放科学合作组织发现，100项原始研究有97%得到统计显著结果，而重复实验只有36%。《自然》调查访问1,576名研究者，逾70%称曾尝试却未能重复他人的实验。这些研究设计不同，不能换算成科研整体的失败率；它们共同支持投入验证、透明方法与认真工作的激励机制。
因此，本文并不主张论文越少科学就必然越好。数量便于行政管理，但若计数本身取代目的，戴明对数字目标的警告就有现实意义。领域拥挤与唯数量的激励，可能稀释对新颖性和可靠性的关注。这与“质量能力须与产出同步增长”的命题相符，但尚不足以单独证明它。
丰田之所以有启发性，恰恰因为丰田生产方式表明规模和质量能够并存。标准作业、发现问题、持续改进，可以让大型组织更加可靠，并非必然相反。然而据美国交通部，2009至2010年间，丰田因油门踏板卡滞与脚垫干涉等问题召回近800万辆车。丰田章男在证词中直接点出了节奏问题：“我们追求增长的速度，超过了我们培养人才和组织的速度，对此我们应当真诚地引以为戒。”这不是每一起召回的完整因果说明，却提出一个可检验的问题：产能扩大时，组织是否同样增强了识别自身产品问题的能力。
下表所列事件在原因、裁定性质与后果上各不相同，不应合并成针对整个国家工业的单一指控。共同的分析问题是，真实测量、披露和纠错的制度能否承受商业压力。
类似问题并不限于日本。2024年1月5日，一架波音737 MAX 9的门塞脱落。美国联邦航空管理局在针对波音及Spirit为期六周的制造质量审查中发现多项涉嫌不符合品控要求的情形，并将737 MAX月产量限制在38架。据美联社，上限于2025年10月提高到42架；半岛电视台报道，波音于2025年5月与美国司法部达成协议。技术调查、产量限制与法律协议各有不同含义；它们不能证明每架飞机都不安全，却把流程保障与独立监督置于讨论中心。
产品寿命也是质量的一部分。苹果公司的iPhone性能限制争议，在2020年11月以1.13亿美元与美国33个州及哥伦比亚特区达成和解，见路透社；另有2020年3月最高5亿美元的集体诉讼和解，以及同年2月法国当局处以的2,500万欧元罚款，见事件汇编。和解不应被写成企业普遍蓄意缩短产品寿命的证明。不过，当替换销售比多年可靠使用更容易计数时，披露、可维修性与耐久性值得审视。欧盟2024/1799号维修指令提供了鼓励维修的政策先例，不能想当然地把更换视作唯一结局。
所谓“计划性淘汰”的疑虑，应通过设计、披露与维修证据检验，不能仅凭和解推断。
市场结构也会使这些问题更值得关注，但不能单独给出完整因果解释。Grullon、Larkin与Michaely发现，自20世纪90年代末以来，美国逾75%的行业集中度上升。这不证明某一家企业必然牺牲质量，却提高了独立验证、可执行标准和消费者有效选择的重要性。上述案例支持一个有限结论：匠人精神不能只停留在口号上，它必须体现在检测、真实数据、合格监督，以及下一件产品出厂前纠正流程的权限里。
机构并非因为一篇文章呼吁就调整规模，人口、预算和招聘决策构成更严格的考验。美国西部州际高等教育委员会预计，美国高中毕业生在2025年达到390万人峰值，到2041年降至337万人。美联储研究记录，1996至2023年8,633所院校中有1,661所关闭，占19.2%；近四分之三是两年制营利机构，公立四年制院校仅有两所关闭。分布至关重要，关闭总数并不预言每所研究大学面临相同风险。《财富》对费城联储模型的报道称，2008年以来已有逾300所授予学位的院校关闭，并预测2025至2029年还可能关闭约80所。预测是有条件的情景，并非既成统计。
其他国家面临不同压力。据NHK，日本2025财年594所私立大学中，316所新生人数未达招生容量，占53.2%。《南华早报》所引官方数据显示，中国2026年高校毕业生创纪录地达到1,270万人，而当年8月不含在校生的16至24岁青年失业率为18.9%。这些指标并不等价；私校缺额和青年失业也不能全部归因于大学扩张。它们说明，教育机构与岗位的匹配、可信的就业路径正在变得更加重要。
雇主声称重新审视学历门槛，实际招聘却跟得较慢。Burning Glass研究所与哈佛商学院发现，2017至2019年间，46%的中技能职业和31%的高技能职业出现实质性学历要求调整。与此同时，哈佛商学院与埃森哲报告，88%的雇主称，符合能力要求的高技能申请者因不完全匹配职位描述而被筛掉。其后Burning Glass研究所的研究发现，承诺的技能招聘变化实际体现于不足每700次招聘中的1次，变化仅0.14个百分点。招聘广告撤去学历条件，并不等于筛选、能力评估和晋升方式随之改变。校正应当是可信地验证技能，而不是宣布正规学习无关紧要。
一些劳动市场信号让替代路径变得具体。根据2023年UPS与美国卡车司机工会（Teamsters）的协议，UPS预计五年合同末全职司机的平均薪酬和福利合计约17万美元，见CNBC；卡车司机工会称，到2027年全职司机最高档位的平均时薪为49美元。前一数字包括福利，并非典型入门工资。美国劳工统计局给出电工2025年年薪中位数63,190美元、就业人数82.1万人，预计2025至2035年就业增长9%，平均每年约72,700个职位空缺。不同职业和口径不能并作“每种技工都比每个大学毕业生收入高”的承诺。
入学选择和培训制度也支持这一判断。美国国家学生信息交换中心（National Student Clearinghouse）统计，2025年春季以职业教育为重点的公立两年制院校有87.1万学生，比2020年春季增长近20%。Mathematica依据劳工部数据指出，在册且仍活跃的注册学徒，从2015年约36万人，增加到其报告所述时期的近70万人，接近翻倍。《华尔街日报》称这种可见变化为“工具腰带一代”；这是新闻描述，不是人口统计。在制度更成熟的体系中，据德国联邦职业教育与培训研究所，约半数德国离校生选择双元职业培训；据瑞士政府，约三分之二瑞士青年完成义务教育后进入职业教育与培训。两国制度不同，不能直接移植，但它们表明，受到承认的技能路径不必是次等教育。
这场校正更适合被理解为：可验证的能力相对于内容不清的资格证书，正在重新定价。这并不意味着大学应机械缩小、学徒制适合所有人，或单一工资数字就能裁定学习的价值。实际问题是，哪类机构能以学生、劳动者和雇主可以检验的证据，提供哪一种卓越。
以下是笔者的主张，不是前述统计已经证明的预测。第一，本文认为，应形成更有选择性、更重质量的研究核心，同时通过其他受尊重的路径广泛培养技能。“有选择性”应指课程和机构确实能够提供的智识训练与支持，而非继承来的身份，也非对入学机会一概设限。研究大学应按其能够认真指导、资助研究和开展探究的能力确定规模。其他机构可在应用学习、转学衔接、教学或职业培养方面做到卓越，而不必模仿同一种研究型履历。这是多元卓越，不是把所有人放在一架以某类校园为顶点的阶梯上。
第二，笔者主张更清楚地区分大学内的专业角色。按这一建议，“教授”称谓宜对应推进知识前沿的研究者；以教学为主的教育者则应拥有另一个同样受尊重的称谓，获得稳定的发展机会，并以卓越教学得到相应回报。这绝非认为教学次于研究，而是反对用同一个职位标签和同一套产出数字遮蔽不同的有价值劳动。具体称谓应与教师和学生充分讨论；实质检验在于，两类工作都得到诚实评价与充分资源。
第三，笔者主张所有机构遵守扩张的节奏原则：增长速度不得超过维护质量的人才和系统的成长速度。丰田章男的反思既适用于招生与指导，也适用于实验室、供应商和生产线。扩容之前，管理者应找出辅导、检验或纠错的瓶颈，并说明将如何解除。此原则不是要求永久保持小规模；丰田生产方式提醒我们，制度设计得当的大型体系同样可以把质量嵌入规模之中。
第四，笔者主张将评价从原始数量重新导向可验证的结果。戴明第11点警惕数字目标取代目的。大学可以考察学习进步、毕业去向、研究能否重复及其意义，同时尊重学科差异和每种指标的边界。雇主应验证工作能力与培养路径，而非未经检验就依赖学历过滤器。制造商应奖励真实报告、及早发现缺陷和长期可靠使用，即便这些做法使短期产量目标更难完成。任何单一的新评分都不应成为另一个供人迎合的目标；独立审查与定性判断不可省略。
最后，笔者主张把可维修性与耐久性纳入价值的定义。欧盟维修指令是政策先例，并不证明所有产品与市场已经解决问题。这里的匠人精神，指严格改进和承担责任，是工业文明的根系：外界看见的是树冠般的规模，支撑它的却是常不显眼的质量土壤。大树若长在薄土上，也许可以挺立一时，但高度不能代替根系。这是关于制度维护的比喻，而非断言所有大组织都脆弱。
现有证据并不支持把大众机会与卓越简单对立。高等教育普及提高了学历水平，大学毕业生的平均工资溢价仍然显著；更多科研产出包含有价值的工作，大型生产系统也能制造可靠产品。另一方面，学费与债务压力、学习成果不均、毕业生就业压力、关于科研新颖性的争议、诚信问题和有记录的工业质量失误，都使“规模即成就”的判断站不住脚。每项指标都有适用范围和局限。本文并不认为增长造成了列举的每一个问题，而是认为，缺乏相应质量能力的增长，会反复制造脆弱性。
人口与市场变化可能推动调整，但收缩本身不会自动改善教学、强化研究或保障生产安全。机构仍须做那些不易显见的工作：按目标合理选拔、培养合格监督者、可信地测量成果，并赋予纠错者真实权限。笔者提出的均衡是多元卓越：有选择性的研究核心、优质教学机构、附有透明收入资料且有尊严的职业与学徒路径，以及扩张服从能力的企业。若把精益求精嵌入制度，规模与质量可以相互促进；若规模成为压倒一切的目标，它也可能耗尽使增长值得追求的根基。
Chandler, A. D.（1990）。《规模与范围：工业资本主义的原动力》（Scale and Scope），哈佛大学出版社（章节见Oxford Academic）。 https://academic.oup.com/book/34608/chapter/294592706
联合国教科文组织（2025）。《鲍莫尔成本病：机器无法替代人力领域的长期经济影响》（评述Baumol与Bowen 1966年及Baumol 2012年的研究）。 https://www.unesco.org/en/articles/baumols-cost-disease-long-term-economic-implications-where-machines-cannot-replace-humans
戴明学院（The W. Edwards Deming Institute）。《戴明管理十四点》，第11点。 https://deming.org/explore/fourteen-points/
丰田汽车公司（2010年2月24日）。丰田章男在美国众议院监督与政府改革委员会听证会上的书面证词。 https://global.toyota/jp/detail/2364044
Trow, M.（1973）。《从精英高等教育向大众高等教育过渡中的问题》，卡内基高等教育委员会。 https://escholarship.org/content/qt96p3s213/qt96p3s213.pdf
Oxford Academic（2004）。《德国与洪堡模式》，载《从启蒙运动到1914年的欧洲大学》。 https://academic.oup.com/book/27081/chapter/196420142
美国国家教育统计中心（1993）。《美国教育120年：统计画像》，表24。 https://nces.ed.gov/pubs93/93442.pdf
联合国教科文组织（2024）。《高等教育：数据一览》。 https://unesdoc.unesco.org/ark:/48223/pf0000389069
经合组织（2025）。《教育概览2025》之“成年人受教育程度”。 https://www.oecd.org/en/publications/2025/09/education-at-a-glance-2025_c58fc9ae/full-report/to-what-level-have-adults-studied_c1d483b0.html
美国国家科学基金会国家科学与工程统计中心（2026）。《2024年博士学位获得者调查》。 https://ncses.nsf.gov/surveys/earned-doctorates/2024
Bundick, B. 与 Pollard, E.（2019）。《大学学费通胀的兴衰》，堪萨斯城联邦储备银行《经济评论》2019年第一季度。 https://www.kansascityfed.org/documents/461/2019-The Rise and Fall of College Tuition Inflation.pdf
纽约联邦储备银行（2026）。《家庭债务与信贷季度报告》，2026年第二季度。 https://www.newyorkfed.org/medialibrary/interactives/householdcredit/data/pdf/HHDC_2026Q2
美国大学教授协会（2025年春）。《数据速览：美国高等教育中的终身教职与非终身聘任（2023年秋）》。 https://www.aaup.org/academe/issues/spring-2025/data-snapshot-tenure-and-contingency-us-higher-education-fall-2023
Arum, R. 与 Roksa, J.（2011）。《学术漂流：大学校园里有限的学习》，芝加哥大学出版社。 https://press.uchicago.edu/Misc/Chicago/028569.ht
纽约联邦储备银行（2026）。《近期大学毕业生劳动力市场》（数据截至2026年第二季度）。 https://www.newyorkfed.org/research/college-labor-market
LendingTree（2026）。《学生贷款债务统计》（对美联储G.19数据的归纳）。 https://www.lendingtree.com/student/student-loan-debt-statistics/
克利夫兰联邦储备银行（2025年3月31日）。《对大学学历劳动者的需求增速不再高于供给》。 https://www.clevelandfed.org/collections/press-releases/2025/pr-20250331-demand-for-college-educated-workers-is-no-longer-growing-faster-than-supply
Abel, J. R. 与 Deitz, R.（2025年4月16日）。《上大学还值得吗？》，纽约联邦储备银行Liberty Street Economics。 https://libertystreeteconomics.newyorkfed.org/2025/04/is-college-still-worth-it/
Spitalniak, L.（2025年7月16日）。《公众对高等教育的信心十年来首次回升》，Higher Ed Dive（报道盖洛普与Lumina基金会数据）。 https://www.highereddive.com/news/confidence-in-higher-education-increases-for-the-first-time-in-a-decade/753076/
美国白宫科技政策办公室（2024）。《关于联邦资助研究开放获取出版融资机制的国会报告》。 https://bidenwhitehouse.archives.gov/wp-content/uploads/2024/06/2024-Report-to-Appropriations-Committee-on-Scholarly-Publishing-and-Public-Access-Implementation.pdf
Publishing State（2025）。《2024年发表的学术论文超过500万篇》（行业估算）。 https://publishingstate.com/more-than-5-million-scholarly-articles-were-published-in-2024-and-here-are-more-interesting-facts/2025/
Park, M.、Leahey, E. 与 Funk, R. J.（2023）。《论文与专利的颠覆性随时间下降》，Nature，第613卷，第138–144页。 https://ideas.repec.org/a/nat/nature/v613y2023i7942d10.1038_s41586-022-05543-x.html
Petersen, A. M.、Arroyave, F. 与 Pammolli, F.（2024）。《颠覆性指数受引用膨胀影响而失真》，Quantitative Science Studies，第5卷第4期。 https://direct.mit.edu/qss/article/5/4/936/124788/The-disruption-index-is-biased-by-citation
Chu, J. S. G. 与 Evans, J. A.（2021）。《大型科学领域中经典进展的放缓》，PNAS，第118卷第41期，e2021636118。 https://www.pnas.org/doi/10.1073/pnas.2021636118
Bloom, N.、Jones, C. I.、Van Reenen, J. 与 Webb, M.（2020）。《新想法是否越来越难找到？》，American Economic Review，第110卷第4期，第1104–1144页。 https://www.aeaweb.org/articles?id=10.1257/aer.20180338
Jones, B. F.（2009）。《知识的负担与“通才之死”：创新是否越来越难？》，Review of Economic Studies，第76卷第1期，第283–317页。 https://academic.oup.com/restud/article-abstract/76/1/283/1577537
Van Noorden, R.（2023）。《2023年逾一万篇论文被撤稿，创历史新高》，Nature，第624卷，第479–481页。 https://www.nature.com/articles/d41586-023-03974-8
开放科学合作组织（2015）。《估计心理学研究的可重复性》，Science，第349卷第6251期，aac4716。 https://pubmed.ncbi.nlm.nih.gov/26315443/
Baker, M.（2016）。《1,500名科学家揭示可重复性问题》，Nature，第533卷，第452–454页。 https://www.nature.com/articles/533452a
美国交通部（2011）。《美国交通部公布NHTSA与NASA关于丰田车辆意外加速的研究结果》。 https://www.transportation.gov/briefing-room/us-department-transportation-releases-results-nhtsa-nasa-study-unintended-acceleration
Ando, R.（2015年7月21日）。《东芝首席执行官因会计问题辞职》，路透社。 https://www.reuters.com/article/business/toshiba-ceo-quits-over-accounting-scandal-idUSKCN0PV0B6/
神户制钢株式会社（2018年3月6日）。《关于神户制钢集团不当行为的报告》。 https://www.kobelco.co.jp/english/releases/files/20180306_en.pdf
路透社（2022年8月2日）。《调查委员会将丰田子公司日野的数据造假归因于企业文化》。 https://www.reuters.com/business/autos-transportation/committee-probing-toyota-unit-hino-blames-company-culture-false-data-scandal-2022-08-02/
大发工业株式会社（2023年12月20日）。《独立第三方委员会调查结果及后续措施》。 https://www.daihatsu.com/news/2023/20231220-4.html
Nippon.com（2024年9月20日）。《日本汽车制造商陷入大范围认证违规事件》。 https://www.nippon.com/en/japan-topics/g02432/
丰田自动织机株式会社（2024年2月22日）。《关于国内工业发动机认证违规的行政处分》。 https://www.toyota-industries.com/news/2024/02/22/008589/index.html
美国联邦航空管理局（2024–2025）。《波音737-9 MAX飞机最新情况》。 https://www.faa.gov/newsroom/updates-boeing-737-9-max-aircraft
Yamat, R. 与 Funk, J.（2025年10月18日）。《门塞脱落近两年后，FAA允许波音提高737 Max产量》，美联社。 https://apnews.com/article/boeing-faa-737-max-production-limit-42-d61ecda95a5db53ef2f179cf2cba5f9b
半岛电视台（2025年5月23日）。《波音与美国司法部达成协议，避免因737 MAX空难被起诉》。 https://www.aljazeera.com/economy/2025/5/23/boeing-reaches-deal-with-us-doj-to-avoid-prosecution-over-737-max-crashes
路透社（2020年11月18日）。《苹果与美国多州就iPhone性能限制达成1.13亿美元和解》。 https://www.reuters.com/business/apple-us-states-reach-113-million-settlement-iphone-throttling-2020-11-18/
维基百科。《Batterygate》（和解与罚款时间线；二手参考）。 https://en.wikipedia.org/wiki/Batterygate
欧盟（2024年6月13日）。《关于促进商品维修共同规则的第2024/1799号指令》，《欧盟官方公报》。 https://eur-lex.europa.eu/eli/dir/2024/1799/oj/eng
Grullon, G.、Larkin, Y. 与 Michaely, R.（2019）。《美国行业是否越来越集中？》，Review of Finance，第23卷第4期，第697–743页。 https://academic.oup.com/rof/article/23/4/697/5477414
美国西部州际高等教育委员会（2024年12月）。《叩响大学之门》第11版：主要发现——高中毕业生。 https://www.wiche.edu/knocking/key-findings/graduates/
Kelchen, R.、Ritter, D. 与 Webber, D.（2024）。《预测大学关闭与财务困境》，美联储金融与经济讨论论文2025-003。 https://www.federalreserve.gov/econres/feds/files/2025003pap.pdf
Sataua, T.（2026年8月31日）。《美国高中毕业班规模预计缩减13%，大学已感受到压力》，《财富》（引用费城联邦储备银行预测）。 https://fortune.com/2026/08/31/college-enrollment-decline-closures-high-school-graduates/
NHK World（2025年8月11日）。《日本半数私立大学连续第三年招生未达容量》。 https://www3.nhk.or.jp/nhkworld/en/news/20250811_02/
《南华早报》（2026年9月17日）。《创纪录的高校毕业生规模推动中国青年失业率升至18.9%》。 https://www.scmp.com/economy/economic-indicators/article/3367869/record-influx-university-graduates-pushes-chinas-youth-unemployment-rate-189
Fuller, J. B.、Langer, C. 与 Sigelman, M.（2022）。《正在出现的学历门槛重置》，Burning Glass研究所与哈佛商学院。 https://www.hbs.edu/faculty/Pages/item.aspx?num=62030
Fuller, J. B.、Raman, M. 等（2021）。《隐形劳动者：未被利用的人才》，哈佛商学院与埃森哲，执行摘要。 https://www.hbs.edu/managing-the-future-of-work/Documents/research/Hidden Workers Executive Summary.pdf
Burning Glass研究所与哈佛商学院（2024）。《技能导向招聘：从宣言到实践的漫长道路》。 https://www.burningglassinstitute.org/research/skills-based-hiring-2024
CNBC（2023年8月9日）。《五年合同期末UPS司机薪酬与福利平均将达17万美元》。 https://www.cnbc.com/2023/08/09/ups-drivers-to-average-170000-in-pay-benefits-at-end-of-5-year-deal.html
美国卡车司机工会（2023）。《UPS临时协议要点》。 https://teamster.org/wp-content/uploads/2023/08/8.5x11-UPS-TA-Highlights-1.pdf
美国劳工统计局（2026）。《职业展望手册：电工》。 https://www.bls.gov/ooh/construction-and-extraction/electricians.htm
美国国家学生信息交换中心研究中心（2025年5月22日）。《以职业教育为重点的公立两年制院校引领春季入学增长》。 https://www.studentclearinghouse.org/nscblog/vocational-focused-public-2-years-lead-enrollment-rise/
Mathematica（2025年12月）。《注册学徒制中的新兴职业》（引用美国劳工部数据）。 https://www.mathematica.org/publications/emerging-occupations-in-registered-apprenticeship
《华尔街日报》（2024年4月2日）。《Z世代如何成为“工具腰带一代”》。 https://www.wsj.com/lifestyle/careers/gen-z-trades-jobs-plumbing-welding-a76b5e43
德国联邦职业教育与培训研究所（BIBB）。《德国双元制职业教育与培训》。 https://www.bibb.de/en/77203.php
瑞士联邦外交部。《瑞士的职业教育与培训》。 https://www.aboutswitzerland.eda.admin.ch/en/vocational-education-and-training
Scale made modern prosperity possible. It can spread education, lower unit costs and bring reliable goods to large populations. Yet its benefits are conditional: the people, judgment and institutional routines that safeguard quality cannot always be multiplied at the speed of enrollment, publication or production. This essay examines that tension across higher education, scientific research and manufacturing. Expanding access has raised educational attainment, while pressure on teaching capacity and graduate employment complicates the value of some credentials. Research output has grown rapidly, although measures of novelty and reproducibility raise questions about what output counts capture. Industrial cases show how failures of testing, disclosure and quality control can accompany growth, but also why scale itself is not the culprit. Demographic change and changing skill demand are beginning to correct some imbalances. The author proposes a differentiated settlement: selective research institutions, respected teaching and vocational pathways, and enterprises that grow only as fast as their quality systems can mature. Craftsmanship here means verifiable care, not nostalgia.
Post-war economic growth rested in part on a powerful organizational discovery: firms could combine large-scale production with managerial capabilities that made scale and scope productive. Alfred Chandler's account helps explain why expanding capacity could improve both availability and consistency. Standardizable operations permit repeated tasks, investment in equipment, learning from defects and common specifications. The gains are real. More places in education, more published work and more manufactured products can likewise widen participation and opportunity. The question is not whether scale is good or bad in the abstract, but what must remain intact while it grows.
Education illustrates an important limit. A machine may produce more identical components per hour; a teacher's close reading of a student's argument, or a mentor's judgment about an unfamiliar problem, does not accelerate in the same way. The relative costs of labor-intensive personal services tend to rise when productivity advances elsewhere, the mechanism associated with Baumol and Bowen's cost disease. This is not proof that education cannot improve productivity. It is a warning that counting seats as though they were standardized components may conceal a decline in attention per learner.
Nor is quality just the virtue of an individual craftsperson. It is a property of selection, training, supervision, measurement, feedback and permission to stop a defective process. W. Edwards Deming's Point 11 calls for eliminating numerical quotas for workers and numerical goals for management. The objection is not to evidence or measurement; it is to targets that substitute an easily counted proxy for the work's actual purpose. A university can maximize enrollment and papers while neglecting learning or intellectual risk. A manufacturer can meet delivery schedules while weakening its capacity to detect defects.
The clearest pacing principle came from an executive facing a quality crisis. Akio Toyoda's testimony warned that growth had outpaced development of Toyota's people and organization. This essay tests a proposition rather than asserting a universal law: when scale outruns an institution's capacity to sustain quality, its foundations weaken. A counterexample in which scale and quality advance together would refine, not invalidate, that proposition. The relevant comparison is between expansion supported by capable institutions and expansion that consumes the capacity on which it depends.
Martin Trow's framework distinguishes elite systems enrolling under roughly 15% of an age group, mass systems enrolling 15–50%, and universal systems exceeding 50%. These are analytical categories, not quality rankings. The Humboldtian ideal joined teaching to research, academic freedom and institutional self-governance. It emerged in a setting far smaller than contemporary higher education. According to the National Center for Education Statistics, US higher-education enrollment equaled 2.3% of the 18–24-year-old population in 1899–1900, 15.2% in 1949–50 and 51.4% in 1989–90; the United States crossed all three of Trow's thresholds within a single century. Universal participation cannot simply reproduce an intimate research apprenticeship at every seat. Nor should that historical scarcity be idealized: widening access has brought a genuine social gain.
The breadth of change is unmistakable. UNESCO reports that the global tertiary gross enrollment ratio rose from 19% in 2000 to 42% in 2022; the number of tertiary students grew from 181 million in 2000 to 256 million in 2010. Across OECD countries, attainment among 25–34-year-olds increased from 27% in 2000 to 48% in 2024, according to the OECD. The NSF records 58,131 US research doctorates in 2024. These measures concern different populations and should not be collapsed into a single trend line. Together they describe a system asked to credential, teach and employ many more people, while preserving serious scholarship and meaningful instruction.
The cost and staffing indicators complicate any unqualified celebration of expansion. From 1980 to 2004, US tuition and fees rose from $1,289 to $7,122, or about 7% annually, against core consumer inflation near 4%; tuition inflation later moderated, as the Kansas City Fed documents. Student-loan balances stood at $1.65 trillion in the New York Fed's household-debt measure in 2026Q2 and about $1.86 trillion under the broader Federal Reserve G.19 measure summarized by LendingTree. These are different measures, not contradictory estimates of one precisely defined total. The AAUP reports contingent appointments rising from 47% in 1987 to 68% in 2023, while tenured or tenure-track positions declined from 53% to 32%. Contingent employment alone cannot establish poor teaching; it does, however, place sustained mentoring and institutional continuity in a different organizational setting.
Measured learning is equally difficult to generalize. Arum and Roksa found that at least 45% of more than 2,300 students at 24 institutions showed no statistically significant improvement on the Collegiate Learning Assessment in their first two years. The finding concerns that assessment and sample, not all learning everywhere. Yet it is a reason to ask whether an institution's claims of educational value are supported by demonstrated growth rather than enrollment and completion counts alone. Selection should also mean matching students, programs and purposes honestly, not excluding capable people by social origin or treating one academic model as the only worthy route.
The labor market provides another qualified signal. For recent graduates aged 22–27, the New York Fed reports unemployment around 5.6% and underemployment of 42% in 2026Q2; unemployment was 3.6% in March 2019, and underemployment among all graduates is 33.7%. The Cleveland Fed finds that the college wage premium has declined about 10% since 2000. But this does not erase the substantial average benefit: the New York Fed's return analysis puts the typical graduate's annual premium above $32,000, roughly 68%. Nor should an aggregate premium be mistaken for the payoff to every program or borrower. Public confidence is a separate signal: Gallup figures reported by Higher Ed Dive fell from 57% in 2015 to 36% in 2023–2024, then recovered to 42% in 2025.
The defensible inference is narrower than “mass education failed.” Access and attainment rose, while some teaching capacity, affordability and labor-market absorption did not keep pace uniformly. The scale–quality gap is uneven across institutions and fields. It calls for clearer evidence of learning and fit, not a return to exclusion for its own sake.
Research has also acquired the outward form of mass production. A US Office of Science and Technology Policy report says annual scholarly articles tripled from just under one million in 1996 to 2.9 million in 2020. An industry tracker estimates about 5.14 million articles in 2024; that figure is an industry estimate, not a continuation of an identical official series. Greater output can reflect more researchers and wider participation, including contributions that a smaller system would have missed. It does not by itself demonstrate proportionate growth in consequential discoveries.
On one influential measure, the pattern is troubling. Park, Leahey and Funk examined 45 million papers and 3.9 million patents across six decades and reported decreasing disruptiveness: new work appeared less likely to break with prior work. That interpretation is contested. A Quantitative Science Studies critique argues that citation inflation accounts for much of the measured decline. Accordingly, the index is a clue to investigate, not a settled verdict that creativity has disappeared. Chu and Evans, studying 1.8 billion citations among 90 million papers in 241 subjects, found that very large fields can concentrate attention on an established canon. That mechanism suggests a possible cost of sheer volume: each new contribution competes in an increasingly dense attention market.
There are further reasons not to equate a rising paper count with rising productivity per researcher. Bloom, Jones, Van Reenen and Webb estimate that maintaining Moore's-law chip-density doubling now requires more than 18 times as many researchers as in the early 1970s. This is a sector-specific result, not a formula for every discipline. Benjamin Jones describes a growing burden of knowledge: the mean age at great invention rose about six years over the twentieth century, alongside greater specialization and teamwork. Later, more collective discovery may reflect the increasing depth of the frontier rather than the disappearance of individual talent. Still, it changes what institutions must provide: time to master a field, collaboration across expertise and room for unusual questions.
Integrity indicators warrant care rather than sensationalism. Nature counted more than 10,000 retractions in 2023; retractions can reflect improved detection as well as failures in production. In a bounded psychology replication project, Open Science Collaboration found statistically significant results in 97% of 100 original studies but 36% of replications. In a Nature survey of 1,576 researchers, more than 70% said they had failed to reproduce another scientist's experiment. The studies differ in design and cannot be turned into a universal failure rate. Together they justify investment in verification, transparent methods and incentives that reward careful work.
The proposition is therefore not that fewer papers automatically yield better science. Counting output can be administratively convenient, but Deming's warning about numerical goals applies when a count becomes the purpose rather than an indicator. A crowded field and quantity-based rewards may dilute attention to novelty and reliability. That pattern is consistent with, though does not alone prove, the thesis that quality capacity must grow with output.
Toyota is an instructive case precisely because its production system demonstrates that scale and quality can coexist. Standard work, problem detection and disciplined improvement can make a large operation more reliable, not less. Yet in 2009–2010 the company recalled nearly eight million vehicles in the sticky-pedal and pedal-entrapment campaigns, according to the US Department of Transportation. Akio Toyoda put the pacing problem in his own words: “We pursued growth over the speed at which we were able to develop our people and our organization, and we should sincerely be mindful of that.” His account is not a causal explanation for every recall. It is a useful test of whether expanding throughput was matched by the organization's ability to know what it was making.
The following cases differ in cause, adjudication and consequence. They should not be compressed into one allegation against an entire national industry. Their common analytical question is whether the systems for accurate measurement, disclosure and correction were strong enough to withstand commercial pressure.
The pattern is not confined to Japan. A door plug detached from a Boeing 737 MAX 9 on 5 January 2024. The FAA described multiple alleged failures in a six-week manufacturing-quality audit of Boeing and Spirit, and capped 737 MAX production at 38 aircraft a month. The cap was raised to 42 in October 2025, according to the Associated Press; Boeing reached an agreement with the US Justice Department in May 2025, as Al Jazeera reported. Those developments have different legal and technical meanings. They do not establish that every plane was unsafe; they make process assurance and independent oversight central questions.
Product life is another dimension of quality. Apple's iPhone performance-throttling dispute resulted in a $113 million settlement with 33 US states and the District of Columbia in November 2020, reported by Reuters; a separate class settlement of up to $500 million was reached in March 2020, and France imposed a €25 million fine in February 2020, as summarized in the case chronology. A settlement should not be portrayed as proof of a universal corporate intention to shorten product life. But disclosure, repairability and durability deserve scrutiny in markets where replacement sales may be more readily counted than years of reliable use. The EU's 2024/1799 repair directive offers a policy precedent for promoting repair rather than assuming replacement is inevitable.
The concern often termed planned obsolescence should be tested through evidence of design, disclosure and repair, not inferred from a settlement alone.
Competitive structure can intensify concern without supplying a complete causal explanation. Grullon, Larkin and Michaely find that more than 75% of US industries became more concentrated after the late 1990s. Concentration does not by itself establish that an individual firm compromised quality. It does make independent verification, enforceable standards and meaningful consumer choice more important. The cases support a bounded conclusion: craftsmanship cannot be delegated to a slogan. It must survive in testing, honest data, competent supervision and the authority to correct a process before the next unit ships.
Institutions do not reconsider scale solely because an essay asks them to. Demography, budgets and hiring decisions impose a harder test. WICHE projects US high-school graduates peaking at 3.9 million in 2025 and falling to 3.37 million by 2041. A Federal Reserve study records 1,661 closures among 8,633 colleges from 1996 to 2023, or 19.2%; nearly three-quarters were two-year for-profit institutions, while only two public four-year institutions closed. The distribution matters: a closure count does not predict that every research university faces the same risk. Fortune's account of a Philadelphia Fed model notes more than 300 degree-granting institutions closed since 2008 and a projection of about 80 additional closures between 2025 and 2029. The projection is conditional, not an accomplished tally.
Other systems face distinct pressures. In Japan, 316 of 594 private universities, or 53.2%, enrolled fewer new students than their capacity in fiscal 2025, according to NHK. In China, a record 12.7 million university graduates in 2026 met a labor market in which unemployment among non-student 16–24-year-olds was 18.9% in August, according to SCMP's report of official figures. These indicators are not equivalent, and neither under-enrollment nor youth unemployment can be attributed wholly to university expansion. They show why institutional fit and credible routes into work are becoming more consequential.
Employers say they are reassessing credentials, but their practices lag. Burning Glass Institute and Harvard Business School found material degree-requirement resets in 46% of middle-skill and 31% of high-skill occupations during 2017–2019. Meanwhile, HBS and Accenture reported that 88% of employers said qualified high-skill applicants were filtered out by exact job-description criteria. A later Burning Glass Institute study found that the promised hiring change appeared in fewer than one in 700 hires, a 0.14-percentage-point shift. Removing a degree requirement from an advertisement is not equivalent to changing screening, assessment or promotion. The corrective is to test skills credibly, not to declare formal learning irrelevant.
Some labor-market signals make the alternative routes concrete. Under the 2023 UPS–Teamsters agreement, UPS projected full-time drivers would average about $170,000 in pay and benefits at the end of the five-year contract, according to CNBC; the Teamsters put the average top hourly rate at $49 by 2027. The first figure includes benefits and is not a typical starting wage. BLS lists electricians' median annual pay at $63,190 in 2025, with 821,000 jobs, projected employment growth of 9% in 2025–2035 and roughly 72,700 openings per year. These are different occupations and measures, not a promise that every trade pays more than every degree.
Enrollment choices and training systems reinforce the point. National Student Clearinghouse counted 871,000 students at vocational-focused public two-year institutions in spring 2025, almost 20% above spring 2020. Active registered apprentices nearly doubled from around 360,000 in 2015 to almost 700,000 in the period reported by Mathematica. The Wall Street Journal called the visible shift a “toolbelt generation”; the label is a journalistic characterization, not a population statistic. In longer-established systems, about half of German school leavers choose dual vocational training, according to BIBB, and about two-thirds of Swiss youth enter vocational education and training after compulsory schooling, according to the Swiss government. These systems differ in design and cannot simply be transplanted. They demonstrate that recognized skill pathways need not be second-class education.
The emerging correction is best understood as a re-pricing of verified competence relative to credentials whose content is unclear. It does not follow that universities should shrink mechanically, that apprenticeships suit everyone or that a single wage figure settles the value of learning. The practical question is which institution can deliver which kind of excellence with evidence that students, workers and employers can inspect.
The following are the author's propositions, not predictions established by the preceding statistics. First, the author proposes a more selective, quality-first research core alongside broad skill formation through other respected routes. “Selective” should describe the intellectual work and support a program can genuinely provide, not inherited status or a blanket restriction on access. A research university should admit, supervise and fund at the scale at which inquiry and close teaching remain credible. Other institutions can excel at applied learning, transfer, teaching or occupational preparation without imitating the same research profile. This is plural excellence rather than a single ladder with everyone ranked by proximity to one type of campus.
Second, the author proposes clearer professional differentiation within universities. The title “professor” should, in this proposal, identify those who advance knowledge through research; teaching-focused educators should receive a distinct, respected title, secure development and rewards appropriate to excellent instruction. This is not a judgment that teaching is lesser. It is an argument against asking one employment label and one output metric to obscure different kinds of valuable work. Whether an institution adopts that nomenclature should be debated with faculty and students; the substantive test is whether both research and teaching are evaluated honestly and resourced adequately.
Third, the author proposes a pacing rule for every institution: expand only as quickly as the people and systems that protect quality can develop. Toyoda's diagnosis applies as a management question to admissions, supervision, laboratories, suppliers and production lines alike. Before adding capacity, leaders should identify the bottleneck in mentoring, inspection or corrective action, then show how it will be relieved. This rule does not require permanent smallness. The Toyota Production System is a reminder that a well-designed large system can embed quality at scale.
Fourth, the author proposes rebalancing evaluation away from raw counts toward verified outcomes. Deming's Point 11 warns against numerical goals becoming substitutes for purpose. Universities could examine demonstrated learning, graduate outcomes and the reproducibility or significance of research, with care for disciplinary differences and the limits of each measure. Employers could validate performance and pathways rather than rely on degree filters that they have not tested. Manufacturers could reward accurate reporting, early defect detection and durable service, even when those practices complicate short-run volume targets. No single replacement score should become another target to game; independent review and qualitative judgment remain necessary.
Finally, the author proposes designing repairability and durability into the definition of value. The EU repair directive is a precedent, not proof that all products or markets have solved the problem. Craftsmanship, understood as disciplined refinement and accountability, is the root system of industrial civilization: it is the often unseen support for visible scale. A large tree growing in thin soil may stand for a time, but height cannot substitute for roots. The metaphor is a warning about institutional maintenance, not an empirical claim that every large organization is fragile.
The record does not support a simple choice between mass access and excellence. Expanded higher education has raised attainment, and the average college wage premium remains substantial. Research volume includes valuable work; large production systems can make reliable goods. Conversely, tuition and debt burdens, uneven learning, pressured graduate employment, contested indicators of scientific novelty, integrity failures and documented industrial lapses make it difficult to treat size as evidence of achievement. Each indicator has a scope and a limitation. The argument is not that growth caused every problem described here, but that growth without commensurate quality capacity creates a recurring vulnerability.
Demography and labor markets may prompt a correction, yet contraction alone will not create better teaching, stronger research or safer production. Institutions need the difficult, less visible work of selection appropriate to purpose, skilled supervision, credible measurement and the authority to repair mistakes. The author's proposed equilibrium is plural excellence: a selective research core, strong teaching institutions, dignified vocational and apprenticeship routes with transparent earnings, and firms whose pace of expansion is governed by capability. Quality and scale can reinforce one another when craft is built into the system. When scale becomes the overriding objective, it can instead consume the foundations that made growth worthwhile.
Chandler, A. D. (1990). Scale and Scope: The Dynamics of Industrial Capitalism. Harvard University Press (chapter via Oxford Academic). https://academic.oup.com/book/34608/chapter/294592706
UNESCO (2025). Baumol's cost disease: long-term economic implications where machines cannot replace humans (on Baumol and Bowen 1966; Baumol 2012). https://www.unesco.org/en/articles/baumols-cost-disease-long-term-economic-implications-where-machines-cannot-replace-humans
The W. Edwards Deming Institute. Dr. Deming's 14 Points for Management, Point 11. https://deming.org/explore/fourteen-points/
Toyota Motor Corporation (24 Feb 2010). Prepared testimony of Akio Toyoda before the US House Committee on Oversight and Government Reform. https://global.toyota/jp/detail/2364044
Trow, M. (1973). Problems in the Transition from Elite to Mass Higher Education. Carnegie Commission on Higher Education. https://escholarship.org/content/qt96p3s213/qt96p3s213.pdf
Oxford Academic (2004). "Germany and the Humboldtian Model," in European Universities from the Enlightenment to 1914. https://academic.oup.com/book/27081/chapter/196420142
National Center for Education Statistics (1993). 120 Years of American Education: A Statistical Portrait, Table 24. https://nces.ed.gov/pubs93/93442.pdf
UNESCO (2024). Higher Education: Figures at a Glance. https://unesdoc.unesco.org/ark:/48223/pf0000389069
OECD (2025). Education at a Glance 2025, "To what level have adults studied?" https://www.oecd.org/en/publications/2025/09/education-at-a-glance-2025_c58fc9ae/full-report/to-what-level-have-adults-studied_c1d483b0.html
National Science Foundation, NCSES (2026). Survey of Earned Doctorates 2024. https://ncses.nsf.gov/surveys/earned-doctorates/2024
Bundick, B., & Pollard, E. (2019). The Rise and Fall of College Tuition Inflation. Federal Reserve Bank of Kansas City Economic Review, First Quarter 2019. https://www.kansascityfed.org/documents/461/2019-The Rise and Fall of College Tuition Inflation.pdf
Federal Reserve Bank of New York (2026). Quarterly Report on Household Debt and Credit, 2026:Q2. https://www.newyorkfed.org/medialibrary/interactives/householdcredit/data/pdf/HHDC_2026Q2
American Association of University Professors (Spring 2025). Data Snapshot: Tenure and Contingency in US Higher Education, Fall 2023. https://www.aaup.org/academe/issues/spring-2025/data-snapshot-tenure-and-contingency-us-higher-education-fall-2023
Arum, R., & Roksa, J. (2011). Academically Adrift: Limited Learning on College Campuses. University of Chicago Press. https://press.uchicago.edu/Misc/Chicago/028569.ht
Federal Reserve Bank of New York (2026). The Labor Market for Recent College Graduates (data through 2026:Q2). https://www.newyorkfed.org/research/college-labor-market
LendingTree (2026). Student Loan Debt Statistics (summary of Federal Reserve G.19 data). https://www.lendingtree.com/student/student-loan-debt-statistics/
Federal Reserve Bank of Cleveland (31 Mar 2025). Demand for College-Educated Workers Is No Longer Growing Faster than Supply. https://www.clevelandfed.org/collections/press-releases/2025/pr-20250331-demand-for-college-educated-workers-is-no-longer-growing-faster-than-supply
Abel, J. R., & Deitz, R. (16 Apr 2025). Is College Still Worth It? Liberty Street Economics, Federal Reserve Bank of New York. https://libertystreeteconomics.newyorkfed.org/2025/04/is-college-still-worth-it/
Spitalniak, L. (16 Jul 2025). Confidence in higher education increases for the first time in a decade. Higher Ed Dive (reporting Gallup–Lumina data). https://www.highereddive.com/news/confidence-in-higher-education-increases-for-the-first-time-in-a-decade/753076/
White House Office of Science and Technology Policy (2024). Report to Congress on Financing Mechanisms for Open Access Publishing of Federally Funded Research. https://bidenwhitehouse.archives.gov/wp-content/uploads/2024/06/2024-Report-to-Appropriations-Committee-on-Scholarly-Publishing-and-Public-Access-Implementation.pdf
Publishing State (2025). More than 5 million scholarly articles were published in 2024 (industry estimate). https://publishingstate.com/more-than-5-million-scholarly-articles-were-published-in-2024-and-here-are-more-interesting-facts/2025/
Park, M., Leahey, E., & Funk, R. J. (2023). Papers and patents are becoming less disruptive over time. Nature, 613, 138–144. https://ideas.repec.org/a/nat/nature/v613y2023i7942d10.1038_s41586-022-05543-x.html
Petersen, A. M., Arroyave, F., & Pammolli, F. (2024). The disruption index is biased by citation inflation. Quantitative Science Studies, 5(4). https://direct.mit.edu/qss/article/5/4/936/124788/The-disruption-index-is-biased-by-citation
Chu, J. S. G., & Evans, J. A. (2021). Slowed canonical progress in large fields of science. PNAS, 118(41), e2021636118. https://www.pnas.org/doi/10.1073/pnas.2021636118
Bloom, N., Jones, C. I., Van Reenen, J., & Webb, M. (2020). Are Ideas Getting Harder to Find? American Economic Review, 110(4), 1104–1144. https://www.aeaweb.org/articles?id=10.1257/aer.20180338
Jones, B. F. (2009). The Burden of Knowledge and the "Death of the Renaissance Man": Is Innovation Getting Harder? Review of Economic Studies, 76(1), 283–317. https://academic.oup.com/restud/article-abstract/76/1/283/1577537
Van Noorden, R. (2023). More than 10,000 research papers were retracted in 2023 — a new record. Nature, 624, 479–481. https://www.nature.com/articles/d41586-023-03974-8
Open Science Collaboration (2015). Estimating the reproducibility of psychological science. Science, 349(6251), aac4716. https://pubmed.ncbi.nlm.nih.gov/26315443/
Baker, M. (2016). 1,500 scientists lift the lid on reproducibility. Nature, 533, 452–454. https://www.nature.com/articles/533452a
US Department of Transportation (2011). US Department of Transportation Releases Results from NHTSA–NASA Study of Unintended Acceleration in Toyota Vehicles. https://www.transportation.gov/briefing-room/us-department-transportation-releases-results-nhtsa-nasa-study-unintended-acceleration
Ando, R. (21 Jul 2015). Toshiba CEO quits over accounting scandal. Reuters. https://www.reuters.com/article/business/toshiba-ceo-quits-over-accounting-scandal-idUSKCN0PV0B6/
Kobe Steel, Ltd. (6 Mar 2018). Report on Misconduct in Kobe Steel Group. https://www.kobelco.co.jp/english/releases/files/20180306_en.pdf
Reuters (2 Aug 2022). Committee probing Toyota unit Hino blames company culture for false data scandal. https://www.reuters.com/business/autos-transportation/committee-probing-toyota-unit-hino-blames-company-culture-false-data-scandal-2022-08-02/
Daihatsu Motor Co., Ltd. (20 Dec 2023). Results of the Investigation by the Independent Third-Party Committee and Future Actions. https://www.daihatsu.com/news/2023/20231220-4.html
Nippon.com (20 Sep 2024). Japanese Vehicle Makers Mired in Wide-Ranging Fraud Scandal. https://www.nippon.com/en/japan-topics/g02432/
Toyota Industries Corporation (22 Feb 2024). Administrative Sanction on Domestic Industrial Engine Certification Irregularities. https://www.toyota-industries.com/news/2024/02/22/008589/index.html
Federal Aviation Administration (2024–2025). Updates on Boeing 737-9 MAX Aircraft. https://www.faa.gov/newsroom/updates-boeing-737-9-max-aircraft
Yamat, R., & Funk, J. (18 Oct 2025). FAA allows Boeing to increase 737 Max production nearly two years after door plug flew off plane. Associated Press. https://apnews.com/article/boeing-faa-737-max-production-limit-42-d61ecda95a5db53ef2f179cf2cba5f9b
Al Jazeera (23 May 2025). Boeing reaches deal with US DOJ to avoid prosecution over 737 MAX crashes. https://www.aljazeera.com/economy/2025/5/23/boeing-reaches-deal-with-us-doj-to-avoid-prosecution-over-737-max-crashes
Reuters (18 Nov 2020). Apple, US states reach $113 million settlement on iPhone throttling. https://www.reuters.com/business/apple-us-states-reach-113-million-settlement-iphone-throttling-2020-11-18/
Wikipedia. Batterygate (chronology of settlements and fines; secondary reference). https://en.wikipedia.org/wiki/Batterygate
European Union (13 Jun 2024). Directive (EU) 2024/1799 on common rules promoting the repair of goods. Official Journal of the European Union. https://eur-lex.europa.eu/eli/dir/2024/1799/oj/eng
Grullon, G., Larkin, Y., & Michaely, R. (2019). Are US Industries Becoming More Concentrated? Review of Finance, 23(4), 697–743. https://academic.oup.com/rof/article/23/4/697/5477414
Western Interstate Commission for Higher Education (Dec 2024). Knocking at the College Door, 11th edition: Key Findings – Graduates. https://www.wiche.edu/knocking/key-findings/graduates/
Kelchen, R., Ritter, D., & Webber, D. (2024). Predicting College Closures and Financial Distress. Finance and Economics Discussion Series 2025-003, Board of Governors of the Federal Reserve System. https://www.federalreserve.gov/econres/feds/files/2025003pap.pdf
Sataua, T. (31 Aug 2026). America's high school graduating class is projected to shrink 13%. Colleges are already feeling the squeeze. Fortune (citing Federal Reserve Bank of Philadelphia projections). https://fortune.com/2026/08/31/college-enrollment-decline-closures-high-school-graduates/
NHK World (11 Aug 2025). Enrollments below capacity at half of Japan's private universities for 3rd year. https://www3.nhk.or.jp/nhkworld/en/news/20250811_02/
South China Morning Post (17 Sep 2026). Record influx of university graduates pushes China's youth unemployment rate to 18.9%. https://www.scmp.com/economy/economic-indicators/article/3367869/record-influx-university-graduates-pushes-chinas-youth-unemployment-rate-189
Fuller, J. B., Langer, C., & Sigelman, M. (2022). The Emerging Degree Reset. Burning Glass Institute and Harvard Business School. https://www.hbs.edu/faculty/Pages/item.aspx?num=62030
Fuller, J. B., Raman, M., et al. (2021). Hidden Workers: Untapped Talent. Harvard Business School and Accenture, Executive Summary. https://www.hbs.edu/managing-the-future-of-work/Documents/research/Hidden Workers Executive Summary.pdf
Burning Glass Institute and Harvard Business School (2024). Skills-Based Hiring: The Long Road from Pronouncements to Practice. https://www.burningglassinstitute.org/research/skills-based-hiring-2024
CNBC (9 Aug 2023). UPS drivers to average $170,000 in pay and benefits at end of 5-year deal. https://www.cnbc.com/2023/08/09/ups-drivers-to-average-170000-in-pay-benefits-at-end-of-5-year-deal.html
International Brotherhood of Teamsters (2023). UPS Tentative Agreement Highlights. https://teamster.org/wp-content/uploads/2023/08/8.5x11-UPS-TA-Highlights-1.pdf
US Bureau of Labor Statistics (2026). Occupational Outlook Handbook: Electricians. https://www.bls.gov/ooh/construction-and-extraction/electricians.htm
National Student Clearinghouse Research Center (22 May 2025). Vocational-Focused Public 2-Years Lead Spring Enrollment Rise. https://www.studentclearinghouse.org/nscblog/vocational-focused-public-2-years-lead-enrollment-rise/
Mathematica (Dec 2025). Emerging Occupations in Registered Apprenticeship (citing US Department of Labor data). https://www.mathematica.org/publications/emerging-occupations-in-registered-apprenticeship
The Wall Street Journal (2 Apr 2024). How Gen Z Is Becoming the Toolbelt Generation. https://www.wsj.com/lifestyle/careers/gen-z-trades-jobs-plumbing-welding-a76b5e43
Federal Institute for Vocational Education and Training (BIBB). The dual system of vocational education and training in Germany. https://www.bibb.de/en/77203.php
Swiss Federal Department of Foreign Affairs. Vocational education and training in Switzerland. https://www.aboutswitzerland.eda.admin.ch/en/vocational-education-and-training