一、竞争力的定义与理论依据
This algorithm does not presuppose a single definition of "competitiveness", but instead synthesizes mainstream international competitiveness research frameworks and maps them onto four tiers of the model:
本算法不预设单一的"竞争力"定义,而是综合国际主流竞争力研究框架,将其映射到模型的四个层次:
- 迈克尔·波特《国家竞争优势》(1990):竞争力的本质是"生产率"——一国或一地在特定产业中创造更高价值的能力,取决于要素条件、需求条件、相关与支持产业、企业战略与竞争四大要素("钻石模型")——为"经济体量+财政创新+人才储备"的结构提供依据。
- 迈克尔·波特《竞争战略》(1980):竞争优势来源于成本领先、差异化与聚焦——为"人均产出(效率)"维度提供依据。
- 世界经济论坛(WEF)《全球竞争力报告》:以制度、基础设施、宏观经济、健康、教育、市场效率、创新等支柱衡量竞争力——为多维指标体系提供参照。
- 瑞士洛桑国际管理发展学院(IMD)《世界竞争力年鉴》:以经济表现、政府效率、企业效率、基础设施四要素衡量竞争力——为"财政与创新"维度提供依据。
- 城市经济学与集聚理论:城市竞争力源于要素集聚带来的规模经济与溢出效应——决定"经济体量"维度权重最高。
二、模型的数学结构
The city competitiveness index uses a weighted-sum model, with the four tier weights summing to 1:
城市竞争力指数采用加权求和模型,四大层次权重之和为 1:
经济体量 E = 0.20·ln(GDP)′ + 0.15·ln(第二产业)′ + 0.12·ln(金融机构存款)′
人均产出 P = 0.15·人均GDP′
财政与创新 I = 0.12·ln(税收)′ + 0.10·ln(研发经费)′ + 0.08·ln(专利)′
人才储备 T = 0.08·高校数′
竞争力指数 = 100 × ( E + P + I + T )
Here, indicators marked with ′ are normalized (min-max normalized to [0,1]); ln indicates that scale-type indicators are log-transformed (natural logarithm) before normalization. "Scale/volume-type" indicators such as GDP, value added of the secondary industry, deposits in financial institutions, tax revenue, R&D expenditure, and patents are log-transformed, while "ratio/quantity-type" indicators such as per-capita GDP and the number of universities are linearly normalized directly.
其中,带 ′ 者为归一化后的指标(min-max 归一到 [0,1]);ln 表示对规模类指标取自然对数后再归一化。GDP、第二产业增加值、金融机构存款、税收、研发经费、专利等"规模/体量类"指标取对数,人均 GDP 与高校数等"比例/数量类"指标直接线性归一。
三、对数处理的依据
A key difference between economic competitiveness and happiness is that happiness exhibits "diminishing marginal returns" (the Easterlin paradox), whereas economic competitiveness is essentially near-linear—the larger the economy and the higher the output, the stronger competitiveness usually is. This model therefore assigns relatively high weights to GDP-type indicators to reflect the basic law that "scale determines competitiveness".
经济竞争力与幸福感的一个关键区别在于:幸福感存在"边际递减"(Easterlin 悖论),而经济竞争力本质上是偏线性的——经济体量越大、产出越高,竞争力通常越强。因此本模型对 GDP 类指标采用较高的权重,以体现"体量决定竞争力"的基本规律。
But pure linearity also creates problems: between the largest and smallest cities, indicators such as GDP, deposits, and patents can differ by dozens or even hundreds of times. If normalized directly and linearly, extreme values would compress all other cities to near zero and erase differentiation. Scale-type indicators are therefore uniformly log-transformed, which preserves the ordering advantage of size while compressing magnitude differences and making the ranking distribution more balanced and robust.
但纯粹线性也会带来问题:最大的城市与最小的城市之间,GDP、存款、专利等指标可相差数十倍乃至数百倍,若直接线性归一,极端值会把其余城市全部压缩到接近 0,丧失区分度。故对规模类指标统一取自然对数,在保留体量排序优势的同时,压缩量级差异,使排名分布更为均衡、稳健。
四、指标与权重依据
| 指标 | 处理 | 权重 | 所属层 | 依据 |
| GDP 总量 | 对数 | 0.20 | 经济体量 | 经济总产出,竞争力的核心载体 |
| 第二产业增加值 | 对数 | 0.15 | 经济体量 | 工业与实体经济的生产能力 |
| 金融机构存款 | 对数 | 0.12 | 经济体量 | 资本集聚与金融资源动员能力 |
| 人均 GDP | 线性 | 0.15 | 人均产出 | 劳动生产率与财富创造效率 |
| 财政收入(税收) | 对数 | 0.12 | 财政与创新 | 财政汲取与公共资源配置能力 |
| 研发经费 | 对数 | 0.10 | 财政与创新 | 科技创新投入强度 |
| 专利拥有量 | 对数 | 0.08 | 财政与创新 | 创新产出与成果转化能力 |
| 高校数 | 线性 | 0.08 | 人才储备 | 人力资本供给与人才培养能力 |
The four tier weights total: economic scale 0.47, per-capita output 0.15, fiscal and innovation 0.30, and talent reserves 0.08, summing to 1.
四层权重合计:经济体量 0.47、人均产出 0.15、财政与创新 0.30、人才储备 0.08,总和为 1。
五、数据来源与局限性
GDP, per-capita GDP, fiscal revenue (tax), R&D expenditure, and value added of the secondary industry directly follow the official public data scope used by the CPPPS City Happiness Index (annual data from the National Bureau of Statistics and each city's statistical bulletin, 2023 approximations); the balance of deposits in financial institutions, the number of regular higher-education institutions, and the number of valid invention patents are approximations of public knowledge, all marked "subject to official figures". This model focuses on economic competitiveness and does not include hard-to-quantify soft factors such as institutional environment, business environment, and infrastructure; approximations carry some error, and the ranking is for research and reference only.
GDP、人均 GDP、财政收入(税收)、研发经费、第二产业增加值直接沿用 CPPPS 城市幸福指数所用官方公开数据口径(国家统计局年度数据及各城市统计公报,2023 年近似值);金融机构存款余额、普通高校数、有效发明专利拥有量为公开知识近似值,均标注"以官方为准"。本模型聚焦经济竞争力,未纳入制度环境、营商环境、基础设施等难以量化的软性因素,且近似值存在误差,排名仅供研究与参考。
六、参考文献
- Porter, M. E.(1990):《The Competitive Advantage of Nations》,Free Press(国家竞争优势理论 / 钻石模型)。
- Porter, M. E.(1980):《Competitive Strategy: Techniques for Analyzing Industries and Competitors》,Free Press。
- World Economic Forum(WEF):《Global Competitiveness Report》,各年度。
- IMD World Competitiveness Center:《World Competitiveness Yearbook》,各年度。
- Krugman, P.(1991):"Increasing Returns and Economic Geography",Journal of Political Economy(新经济地理与集聚)。
- Easterlin, R.(1974):"Does Economic Growth Improve the Human Lot?",Nations and Households in Economic Growth(作为与幸福模型对照的边际递减参考)。
- 国家统计局:年度国民经济与社会发展统计公报。
- 各城市统计年鉴与国民经济和社会发展统计公报(2023 年)。