research paper on the rat race

• Search Menu
• Browse content in A - General Economics and Teaching
• Browse content in A1 - General Economics
• A10 - General
• A12 - Relation of Economics to Other Disciplines
• A13 - Relation of Economics to Social Values
• A14 - Sociology of Economics
• Browse content in A2 - Economic Education and Teaching of Economics
• A29 - Other
• Browse content in B - History of Economic Thought, Methodology, and Heterodox Approaches
• B0 - General
• Browse content in B1 - History of Economic Thought through 1925
• B11 - Preclassical (Ancient, Medieval, Mercantilist, Physiocratic)
• B12 - Classical (includes Adam Smith)
• Browse content in B2 - History of Economic Thought since 1925
• B20 - General
• B21 - Microeconomics
• B22 - Macroeconomics
• B25 - Historical; Institutional; Evolutionary; Austrian
• B26 - Financial Economics
• Browse content in B3 - History of Economic Thought: Individuals
• B31 - Individuals
• Browse content in B4 - Economic Methodology
• B41 - Economic Methodology
• Browse content in B5 - Current Heterodox Approaches
• B55 - Social Economics
• Browse content in C - Mathematical and Quantitative Methods
• Browse content in C0 - General
• C00 - General
• C02 - Mathematical Methods
• Browse content in C1 - Econometric and Statistical Methods and Methodology: General
• C10 - General
• C11 - Bayesian Analysis: General
• C12 - Hypothesis Testing: General
• C13 - Estimation: General
• C14 - Semiparametric and Nonparametric Methods: General
• C15 - Statistical Simulation Methods: General
• C19 - Other
• Browse content in C2 - Single Equation Models; Single Variables
• C21 - Cross-Sectional Models; Spatial Models; Treatment Effect Models; Quantile Regressions
• C22 - Time-Series Models; Dynamic Quantile Regressions; Dynamic Treatment Effect Models; Diffusion Processes
• C23 - Panel Data Models; Spatio-temporal Models
• C24 - Truncated and Censored Models; Switching Regression Models; Threshold Regression Models
• C25 - Discrete Regression and Qualitative Choice Models; Discrete Regressors; Proportions; Probabilities
• C26 - Instrumental Variables (IV) Estimation
• Browse content in C3 - Multiple or Simultaneous Equation Models; Multiple Variables
• C31 - Cross-Sectional Models; Spatial Models; Treatment Effect Models; Quantile Regressions; Social Interaction Models
• C32 - Time-Series Models; Dynamic Quantile Regressions; Dynamic Treatment Effect Models; Diffusion Processes; State Space Models
• C33 - Panel Data Models; Spatio-temporal Models
• C34 - Truncated and Censored Models; Switching Regression Models
• C35 - Discrete Regression and Qualitative Choice Models; Discrete Regressors; Proportions
• C36 - Instrumental Variables (IV) Estimation
• Browse content in C4 - Econometric and Statistical Methods: Special Topics
• C41 - Duration Analysis; Optimal Timing Strategies
• C43 - Index Numbers and Aggregation
• Browse content in C5 - Econometric Modeling
• C51 - Model Construction and Estimation
• C52 - Model Evaluation, Validation, and Selection
• C53 - Forecasting and Prediction Methods; Simulation Methods
• C54 - Quantitative Policy Modeling
• Browse content in C6 - Mathematical Methods; Programming Models; Mathematical and Simulation Modeling
• C60 - General
• C61 - Optimization Techniques; Programming Models; Dynamic Analysis
• C62 - Existence and Stability Conditions of Equilibrium
• C63 - Computational Techniques; Simulation Modeling
• Browse content in C7 - Game Theory and Bargaining Theory
• C71 - Cooperative Games
• C72 - Noncooperative Games
• C73 - Stochastic and Dynamic Games; Evolutionary Games; Repeated Games
• C78 - Bargaining Theory; Matching Theory
• Browse content in C8 - Data Collection and Data Estimation Methodology; Computer Programs
• C81 - Methodology for Collecting, Estimating, and Organizing Microeconomic Data; Data Access
• C82 - Methodology for Collecting, Estimating, and Organizing Macroeconomic Data; Data Access
• C83 - Survey Methods; Sampling Methods
• Browse content in C9 - Design of Experiments
• C90 - General
• C91 - Laboratory, Individual Behavior
• C92 - Laboratory, Group Behavior
• C93 - Field Experiments
• Browse content in D - Microeconomics
• Browse content in D0 - General
• D00 - General
• D01 - Microeconomic Behavior: Underlying Principles
• D02 - Institutions: Design, Formation, Operations, and Impact
• D03 - Behavioral Microeconomics: Underlying Principles
• D04 - Microeconomic Policy: Formulation; Implementation, and Evaluation
• Browse content in D1 - Household Behavior and Family Economics
• D10 - General
• D11 - Consumer Economics: Theory
• D12 - Consumer Economics: Empirical Analysis
• D13 - Household Production and Intrahousehold Allocation
• D14 - Household Saving; Personal Finance
• D15 - Intertemporal Household Choice: Life Cycle Models and Saving
• D16 - Collaborative Consumption
• D18 - Consumer Protection
• D19 - Other
• Browse content in D2 - Production and Organizations
• D21 - Firm Behavior: Theory
• D22 - Firm Behavior: Empirical Analysis
• D23 - Organizational Behavior; Transaction Costs; Property Rights
• D24 - Production; Cost; Capital; Capital, Total Factor, and Multifactor Productivity; Capacity
• D29 - Other
• Browse content in D3 - Distribution
• D30 - General
• D31 - Personal Income, Wealth, and Their Distributions
• D33 - Factor Income Distribution
• Browse content in D4 - Market Structure, Pricing, and Design
• D40 - General
• D41 - Perfect Competition
• D43 - Oligopoly and Other Forms of Market Imperfection
• D44 - Auctions
• Browse content in D5 - General Equilibrium and Disequilibrium
• D50 - General
• D53 - Financial Markets
• D58 - Computable and Other Applied General Equilibrium Models
• Browse content in D6 - Welfare Economics
• D60 - General
• D61 - Allocative Efficiency; Cost-Benefit Analysis
• D62 - Externalities
• D63 - Equity, Justice, Inequality, and Other Normative Criteria and Measurement
• D64 - Altruism; Philanthropy
• D69 - Other
• Browse content in D7 - Analysis of Collective Decision-Making
• D70 - General
• D71 - Social Choice; Clubs; Committees; Associations
• D72 - Political Processes: Rent-seeking, Lobbying, Elections, Legislatures, and Voting Behavior
• D73 - Bureaucracy; Administrative Processes in Public Organizations; Corruption
• D74 - Conflict; Conflict Resolution; Alliances; Revolutions
• D78 - Positive Analysis of Policy Formulation and Implementation
• Browse content in D8 - Information, Knowledge, and Uncertainty
• D80 - General
• D81 - Criteria for Decision-Making under Risk and Uncertainty
• D82 - Asymmetric and Private Information; Mechanism Design
• D83 - Search; Learning; Information and Knowledge; Communication; Belief; Unawareness
• D84 - Expectations; Speculations
• D85 - Network Formation and Analysis: Theory
• D86 - Economics of Contract: Theory
• Browse content in D9 - Micro-Based Behavioral Economics
• D90 - General
• D91 - Role and Effects of Psychological, Emotional, Social, and Cognitive Factors on Decision Making
• D92 - Intertemporal Firm Choice, Investment, Capacity, and Financing
• Browse content in E - Macroeconomics and Monetary Economics
• Browse content in E0 - General
• E00 - General
• E01 - Measurement and Data on National Income and Product Accounts and Wealth; Environmental Accounts
• E02 - Institutions and the Macroeconomy
• Browse content in E1 - General Aggregative Models
• E10 - General
• E11 - Marxian; Sraffian; Kaleckian
• E12 - Keynes; Keynesian; Post-Keynesian
• E13 - Neoclassical
• E19 - Other
• Browse content in E2 - Consumption, Saving, Production, Investment, Labor Markets, and Informal Economy
• E20 - General
• E21 - Consumption; Saving; Wealth
• E22 - Investment; Capital; Intangible Capital; Capacity
• E23 - Production
• E24 - Employment; Unemployment; Wages; Intergenerational Income Distribution; Aggregate Human Capital; Aggregate Labor Productivity
• E25 - Aggregate Factor Income Distribution
• E26 - Informal Economy; Underground Economy
• E27 - Forecasting and Simulation: Models and Applications
• Browse content in E3 - Prices, Business Fluctuations, and Cycles
• E30 - General
• E31 - Price Level; Inflation; Deflation
• E32 - Business Fluctuations; Cycles
• E37 - Forecasting and Simulation: Models and Applications
• Browse content in E4 - Money and Interest Rates
• E40 - General
• E41 - Demand for Money
• E42 - Monetary Systems; Standards; Regimes; Government and the Monetary System; Payment Systems
• E43 - Interest Rates: Determination, Term Structure, and Effects
• E44 - Financial Markets and the Macroeconomy
• E47 - Forecasting and Simulation: Models and Applications
• Browse content in E5 - Monetary Policy, Central Banking, and the Supply of Money and Credit
• E50 - General
• E51 - Money Supply; Credit; Money Multipliers
• E52 - Monetary Policy
• E58 - Central Banks and Their Policies
• Browse content in E6 - Macroeconomic Policy, Macroeconomic Aspects of Public Finance, and General Outlook
• E60 - General
• E61 - Policy Objectives; Policy Designs and Consistency; Policy Coordination
• E62 - Fiscal Policy
• E63 - Comparative or Joint Analysis of Fiscal and Monetary Policy; Stabilization; Treasury Policy
• E65 - Studies of Particular Policy Episodes
• E69 - Other
• Browse content in E7 - Macro-Based Behavioral Economics
• E70 - General
• E71 - Role and Effects of Psychological, Emotional, Social, and Cognitive Factors on the Macro Economy
• Browse content in F - International Economics
• Browse content in F0 - General
• F02 - International Economic Order and Integration
• Browse content in F1 - Trade
• F10 - General
• F11 - Neoclassical Models of Trade
• F12 - Models of Trade with Imperfect Competition and Scale Economies; Fragmentation
• F13 - Trade Policy; International Trade Organizations
• F14 - Empirical Studies of Trade
• F15 - Economic Integration
• F16 - Trade and Labor Market Interactions
• F17 - Trade Forecasting and Simulation
• F18 - Trade and Environment
• Browse content in F2 - International Factor Movements and International Business
• F21 - International Investment; Long-Term Capital Movements
• F22 - International Migration
• F23 - Multinational Firms; International Business
• F24 - Remittances
• Browse content in F3 - International Finance
• F30 - General
• F31 - Foreign Exchange
• F32 - Current Account Adjustment; Short-Term Capital Movements
• F33 - International Monetary Arrangements and Institutions
• F34 - International Lending and Debt Problems
• F35 - Foreign Aid
• F36 - Financial Aspects of Economic Integration
• F37 - International Finance Forecasting and Simulation: Models and Applications
• Browse content in F4 - Macroeconomic Aspects of International Trade and Finance
• F40 - General
• F41 - Open Economy Macroeconomics
• F42 - International Policy Coordination and Transmission
• F43 - Economic Growth of Open Economies
• F44 - International Business Cycles
• F45 - Macroeconomic Issues of Monetary Unions
• Browse content in F5 - International Relations, National Security, and International Political Economy
• F50 - General
• F51 - International Conflicts; Negotiations; Sanctions
• F52 - National Security; Economic Nationalism
• F53 - International Agreements and Observance; International Organizations
• F55 - International Institutional Arrangements
• F59 - Other
• Browse content in F6 - Economic Impacts of Globalization
• F62 - Macroeconomic Impacts
• F63 - Economic Development
• F64 - Environment
• Browse content in G - Financial Economics
• Browse content in G0 - General
• G01 - Financial Crises
• G02 - Behavioral Finance: Underlying Principles
• Browse content in G1 - General Financial Markets
• G10 - General
• G11 - Portfolio Choice; Investment Decisions
• G12 - Asset Pricing; Trading volume; Bond Interest Rates
• G14 - Information and Market Efficiency; Event Studies; Insider Trading
• G15 - International Financial Markets
• G18 - Government Policy and Regulation
• Browse content in G2 - Financial Institutions and Services
• G20 - General
• G21 - Banks; Depository Institutions; Micro Finance Institutions; Mortgages
• G22 - Insurance; Insurance Companies; Actuarial Studies
• G24 - Investment Banking; Venture Capital; Brokerage; Ratings and Ratings Agencies
• G28 - Government Policy and Regulation
• Browse content in G3 - Corporate Finance and Governance
• G32 - Financing Policy; Financial Risk and Risk Management; Capital and Ownership Structure; Value of Firms; Goodwill
• G33 - Bankruptcy; Liquidation
• G34 - Mergers; Acquisitions; Restructuring; Corporate Governance
• G35 - Payout Policy
• G38 - Government Policy and Regulation
• Browse content in H - Public Economics
• Browse content in H0 - General
• H00 - General
• Browse content in H1 - Structure and Scope of Government
• H10 - General
• H11 - Structure, Scope, and Performance of Government
• H12 - Crisis Management
• Browse content in H2 - Taxation, Subsidies, and Revenue
• H20 - General
• H21 - Efficiency; Optimal Taxation
• H22 - Incidence
• H23 - Externalities; Redistributive Effects; Environmental Taxes and Subsidies
• H24 - Personal Income and Other Nonbusiness Taxes and Subsidies; includes inheritance and gift taxes
• H25 - Business Taxes and Subsidies
• H26 - Tax Evasion and Avoidance
• Browse content in H3 - Fiscal Policies and Behavior of Economic Agents
• H30 - General
• H31 - Household
• Browse content in H4 - Publicly Provided Goods
• H40 - General
• H41 - Public Goods
• H42 - Publicly Provided Private Goods
• Browse content in H5 - National Government Expenditures and Related Policies
• H50 - General
• H51 - Government Expenditures and Health
• H52 - Government Expenditures and Education
• H53 - Government Expenditures and Welfare Programs
• H54 - Infrastructures; Other Public Investment and Capital Stock
• H55 - Social Security and Public Pensions
• H56 - National Security and War
• Browse content in H6 - National Budget, Deficit, and Debt
• H60 - General
• H61 - Budget; Budget Systems
• H62 - Deficit; Surplus
• H63 - Debt; Debt Management; Sovereign Debt
• Browse content in H7 - State and Local Government; Intergovernmental Relations
• H70 - General
• H71 - State and Local Taxation, Subsidies, and Revenue
• H72 - State and Local Budget and Expenditures
• H75 - State and Local Government: Health; Education; Welfare; Public Pensions
• H76 - State and Local Government: Other Expenditure Categories
• H77 - Intergovernmental Relations; Federalism; Secession
• Browse content in H8 - Miscellaneous Issues
• H83 - Public Administration; Public Sector Accounting and Audits
• H84 - Disaster Aid
• H87 - International Fiscal Issues; International Public Goods
• Browse content in I - Health, Education, and Welfare
• Browse content in I0 - General
• I00 - General
• Browse content in I1 - Health
• I10 - General
• I12 - Health Behavior
• I14 - Health and Inequality
• I15 - Health and Economic Development
• I18 - Government Policy; Regulation; Public Health
• I19 - Other
• Browse content in I2 - Education and Research Institutions
• I20 - General
• I21 - Analysis of Education
• I22 - Educational Finance; Financial Aid
• I23 - Higher Education; Research Institutions
• I24 - Education and Inequality
• I25 - Education and Economic Development
• I26 - Returns to Education
• I28 - Government Policy
• I29 - Other
• Browse content in I3 - Welfare, Well-Being, and Poverty
• I30 - General
• I31 - General Welfare
• I32 - Measurement and Analysis of Poverty
• I38 - Government Policy; Provision and Effects of Welfare Programs
• Browse content in J - Labor and Demographic Economics
• Browse content in J0 - General
• J00 - General
• J01 - Labor Economics: General
• J08 - Labor Economics Policies
• Browse content in J1 - Demographic Economics
• J10 - General
• J11 - Demographic Trends, Macroeconomic Effects, and Forecasts
• J12 - Marriage; Marital Dissolution; Family Structure; Domestic Abuse
• J13 - Fertility; Family Planning; Child Care; Children; Youth
• J14 - Economics of the Elderly; Economics of the Handicapped; Non-Labor Market Discrimination
• J15 - Economics of Minorities, Races, Indigenous Peoples, and Immigrants; Non-labor Discrimination
• J16 - Economics of Gender; Non-labor Discrimination
• J17 - Value of Life; Forgone Income
• J18 - Public Policy
• Browse content in J2 - Demand and Supply of Labor
• J20 - General
• J21 - Labor Force and Employment, Size, and Structure
• J22 - Time Allocation and Labor Supply
• J23 - Labor Demand
• J24 - Human Capital; Skills; Occupational Choice; Labor Productivity
• J26 - Retirement; Retirement Policies
• J28 - Safety; Job Satisfaction; Related Public Policy
• Browse content in J3 - Wages, Compensation, and Labor Costs
• J30 - General
• J31 - Wage Level and Structure; Wage Differentials
• J32 - Nonwage Labor Costs and Benefits; Retirement Plans; Private Pensions
• J33 - Compensation Packages; Payment Methods
• J38 - Public Policy
• Browse content in J4 - Particular Labor Markets
• J41 - Labor Contracts
• J42 - Monopsony; Segmented Labor Markets
• J45 - Public Sector Labor Markets
• J46 - Informal Labor Markets
• J48 - Public Policy
• Browse content in J5 - Labor-Management Relations, Trade Unions, and Collective Bargaining
• J50 - General
• J51 - Trade Unions: Objectives, Structure, and Effects
• J52 - Dispute Resolution: Strikes, Arbitration, and Mediation; Collective Bargaining
• J53 - Labor-Management Relations; Industrial Jurisprudence
• J54 - Producer Cooperatives; Labor Managed Firms; Employee Ownership
• J58 - Public Policy
• Browse content in J6 - Mobility, Unemployment, Vacancies, and Immigrant Workers
• J60 - General
• J61 - Geographic Labor Mobility; Immigrant Workers
• J62 - Job, Occupational, and Intergenerational Mobility
• J63 - Turnover; Vacancies; Layoffs
• J64 - Unemployment: Models, Duration, Incidence, and Job Search
• J65 - Unemployment Insurance; Severance Pay; Plant Closings
• J68 - Public Policy
• Browse content in J7 - Labor Discrimination
• J71 - Discrimination
• Browse content in J8 - Labor Standards: National and International
• J81 - Working Conditions
• J88 - Public Policy
• Browse content in K - Law and Economics
• Browse content in K0 - General
• K00 - General
• Browse content in K1 - Basic Areas of Law
• K11 - Property Law
• K12 - Contract Law
• K13 - Tort Law and Product Liability; Forensic Economics
• K14 - Criminal Law
• K16 - Election Law
• Browse content in K3 - Other Substantive Areas of Law
• K31 - Labor Law
• K32 - Environmental, Health, and Safety Law
• K34 - Tax Law
• K37 - Immigration Law
• Browse content in K4 - Legal Procedure, the Legal System, and Illegal Behavior
• K41 - Litigation Process
• K42 - Illegal Behavior and the Enforcement of Law
• K49 - Other
• Browse content in L - Industrial Organization
• Browse content in L0 - General
• L00 - General
• Browse content in L1 - Market Structure, Firm Strategy, and Market Performance
• L10 - General
• L11 - Production, Pricing, and Market Structure; Size Distribution of Firms
• L12 - Monopoly; Monopolization Strategies
• L13 - Oligopoly and Other Imperfect Markets
• L14 - Transactional Relationships; Contracts and Reputation; Networks
• L16 - Industrial Organization and Macroeconomics: Industrial Structure and Structural Change; Industrial Price Indices
• Browse content in L2 - Firm Objectives, Organization, and Behavior
• L20 - General
• L21 - Business Objectives of the Firm
• L22 - Firm Organization and Market Structure
• L23 - Organization of Production
• L24 - Contracting Out; Joint Ventures; Technology Licensing
• L25 - Firm Performance: Size, Diversification, and Scope
• L26 - Entrepreneurship
• L29 - Other
• Browse content in L3 - Nonprofit Organizations and Public Enterprise
• L30 - General
• L31 - Nonprofit Institutions; NGOs; Social Entrepreneurship
• L32 - Public Enterprises; Public-Private Enterprises
• L33 - Comparison of Public and Private Enterprises and Nonprofit Institutions; Privatization; Contracting Out
• Browse content in L4 - Antitrust Issues and Policies
• L40 - General
• L41 - Monopolization; Horizontal Anticompetitive Practices
• L43 - Legal Monopolies and Regulation or Deregulation
• Browse content in L5 - Regulation and Industrial Policy
• L50 - General
• L51 - Economics of Regulation
• L52 - Industrial Policy; Sectoral Planning Methods
• L53 - Enterprise Policy
• Browse content in L6 - Industry Studies: Manufacturing
• L60 - General
• L66 - Food; Beverages; Cosmetics; Tobacco; Wine and Spirits
• Browse content in L7 - Industry Studies: Primary Products and Construction
• L71 - Mining, Extraction, and Refining: Hydrocarbon Fuels
• L78 - Government Policy
• Browse content in L8 - Industry Studies: Services
• L81 - Retail and Wholesale Trade; e-Commerce
• L83 - Sports; Gambling; Recreation; Tourism
• L86 - Information and Internet Services; Computer Software
• Browse content in L9 - Industry Studies: Transportation and Utilities
• L94 - Electric Utilities
• L98 - Government Policy
• Browse content in M - Business Administration and Business Economics; Marketing; Accounting; Personnel Economics
• Browse content in M1 - Business Administration
• M12 - Personnel Management; Executives; Executive Compensation
• M14 - Corporate Culture; Social Responsibility
• M16 - International Business Administration
• Browse content in M3 - Marketing and Advertising
• M31 - Marketing
• Browse content in M5 - Personnel Economics
• M50 - General
• M51 - Firm Employment Decisions; Promotions
• M52 - Compensation and Compensation Methods and Their Effects
• M53 - Training
• M54 - Labor Management
• M55 - Labor Contracting Devices
• Browse content in N - Economic History
• Browse content in N1 - Macroeconomics and Monetary Economics; Industrial Structure; Growth; Fluctuations
• N10 - General, International, or Comparative
• N11 - U.S.; Canada: Pre-1913
• N12 - U.S.; Canada: 1913-
• N13 - Europe: Pre-1913
• N15 - Asia including Middle East
• Browse content in N2 - Financial Markets and Institutions
• N20 - General, International, or Comparative
• N24 - Europe: 1913-
• N25 - Asia including Middle East
• Browse content in N3 - Labor and Consumers, Demography, Education, Health, Welfare, Income, Wealth, Religion, and Philanthropy
• N31 - U.S.; Canada: Pre-1913
• N33 - Europe: Pre-1913
• N34 - Europe: 1913-
• Browse content in N4 - Government, War, Law, International Relations, and Regulation
• N40 - General, International, or Comparative
• N45 - Asia including Middle East
• N47 - Africa; Oceania
• Browse content in N5 - Agriculture, Natural Resources, Environment, and Extractive Industries
• N50 - General, International, or Comparative
• N53 - Europe: Pre-1913
• N57 - Africa; Oceania
• Browse content in N7 - Transport, Trade, Energy, Technology, and Other Services
• N70 - General, International, or Comparative
• N72 - U.S.; Canada: 1913-
• Browse content in N9 - Regional and Urban History
• N97 - Africa; Oceania
• Browse content in O - Economic Development, Innovation, Technological Change, and Growth
• Browse content in O1 - Economic Development
• O10 - General
• O11 - Macroeconomic Analyses of Economic Development
• O12 - Microeconomic Analyses of Economic Development
• O13 - Agriculture; Natural Resources; Energy; Environment; Other Primary Products
• O14 - Industrialization; Manufacturing and Service Industries; Choice of Technology
• O15 - Human Resources; Human Development; Income Distribution; Migration
• O16 - Financial Markets; Saving and Capital Investment; Corporate Finance and Governance
• O17 - Formal and Informal Sectors; Shadow Economy; Institutional Arrangements
• O18 - Urban, Rural, Regional, and Transportation Analysis; Housing; Infrastructure
• O19 - International Linkages to Development; Role of International Organizations
• Browse content in O2 - Development Planning and Policy
• O22 - Project Analysis
• O23 - Fiscal and Monetary Policy in Development
• O24 - Trade Policy; Factor Movement Policy; Foreign Exchange Policy
• O25 - Industrial Policy
• Browse content in O3 - Innovation; Research and Development; Technological Change; Intellectual Property Rights
• O30 - General
• O31 - Innovation and Invention: Processes and Incentives
• O32 - Management of Technological Innovation and R&D
• O33 - Technological Change: Choices and Consequences; Diffusion Processes
• O34 - Intellectual Property and Intellectual Capital
• O38 - Government Policy
• O39 - Other
• Browse content in O4 - Economic Growth and Aggregate Productivity
• O40 - General
• O41 - One, Two, and Multisector Growth Models
• O42 - Monetary Growth Models
• O43 - Institutions and Growth
• O47 - Empirical Studies of Economic Growth; Aggregate Productivity; Cross-Country Output Convergence
• O49 - Other
• Browse content in O5 - Economywide Country Studies
• O50 - General
• O52 - Europe
• O53 - Asia including Middle East
• O55 - Africa
• O57 - Comparative Studies of Countries
• Browse content in P - Economic Systems
• Browse content in P1 - Capitalist Systems
• P10 - General
• P13 - Cooperative Enterprises
• P16 - Political Economy
• P17 - Performance and Prospects
• Browse content in P2 - Socialist Systems and Transitional Economies
• P20 - General
• P26 - Political Economy; Property Rights
• Browse content in P3 - Socialist Institutions and Their Transitions
• P31 - Socialist Enterprises and Their Transitions
• Browse content in P4 - Other Economic Systems
• P48 - Political Economy; Legal Institutions; Property Rights; Natural Resources; Energy; Environment; Regional Studies
• Browse content in P5 - Comparative Economic Systems
• P50 - General
• Browse content in Q - Agricultural and Natural Resource Economics; Environmental and Ecological Economics
• Browse content in Q0 - General
• Q02 - Commodity Markets
• Browse content in Q1 - Agriculture
• Q11 - Aggregate Supply and Demand Analysis; Prices
• Q13 - Agricultural Markets and Marketing; Cooperatives; Agribusiness
• Q15 - Land Ownership and Tenure; Land Reform; Land Use; Irrigation; Agriculture and Environment
• Q16 - R&D; Agricultural Technology; Biofuels; Agricultural Extension Services
• Q17 - Agriculture in International Trade
• Q18 - Agricultural Policy; Food Policy
• Browse content in Q2 - Renewable Resources and Conservation
• Q20 - General
• Q22 - Fishery; Aquaculture
• Q23 - Forestry
• Q25 - Water
• Q26 - Recreational Aspects of Natural Resources
• Q29 - Other
• Browse content in Q3 - Nonrenewable Resources and Conservation
• Q30 - General
• Q32 - Exhaustible Resources and Economic Development
• Q33 - Resource Booms
• Q34 - Natural Resources and Domestic and International Conflicts
• Q38 - Government Policy
• Browse content in Q4 - Energy
• Q40 - General
• Q41 - Demand and Supply; Prices
• Q42 - Alternative Energy Sources
• Q43 - Energy and the Macroeconomy
• Q48 - Government Policy
• Browse content in Q5 - Environmental Economics
• Q50 - General
• Q51 - Valuation of Environmental Effects
• Q52 - Pollution Control Adoption Costs; Distributional Effects; Employment Effects
• Q53 - Air Pollution; Water Pollution; Noise; Hazardous Waste; Solid Waste; Recycling
• Q54 - Climate; Natural Disasters; Global Warming
• Q56 - Environment and Development; Environment and Trade; Sustainability; Environmental Accounts and Accounting; Environmental Equity; Population Growth
• Q58 - Government Policy
• Browse content in R - Urban, Rural, Regional, Real Estate, and Transportation Economics
• Browse content in R1 - General Regional Economics
• R10 - General
• R11 - Regional Economic Activity: Growth, Development, Environmental Issues, and Changes
• R15 - Econometric and Input-Output Models; Other Models
• Browse content in R2 - Household Analysis
• R23 - Regional Migration; Regional Labor Markets; Population; Neighborhood Characteristics
• R29 - Other
• Browse content in R4 - Transportation Economics
• R40 - General
• R41 - Transportation: Demand, Supply, and Congestion; Travel Time; Safety and Accidents; Transportation Noise
• Browse content in R5 - Regional Government Analysis
• R58 - Regional Development Planning and Policy
• Browse content in Z - Other Special Topics
• Browse content in Z1 - Cultural Economics; Economic Sociology; Economic Anthropology
• Z10 - General
• Z11 - Economics of the Arts and Literature
• Z12 - Religion
• Z13 - Economic Sociology; Economic Anthropology; Social and Economic Stratification
• Z19 - Other
• Browse content in Z2 - Sports Economics
• Z21 - Industry Studies
• Z22 - Labor Issues
• Z29 - Other
• Browse content in Z3 - Tourism Economics
• Z30 - General
• Advance articles
• Author Guidelines
• Submission Site
• Open Access
• About Oxford Economic Papers
• Editorial Board
• Advertising and Corporate Services
• Journals Career Network
• Self-Archiving Policy
• Dispatch Dates
• Terms and Conditions
• Journals on Oxford Academic
• Books on Oxford Academic

Article Contents

1. introduction, 2. brief literature review, 3. potential pathways, 4. methodology and data, 6. pathways, 7. conclusions and future work, acknowledgements.

• < Previous

The unintended consequences of the rat race: the detrimental effects of performance pay on health

• Article contents
• Figures & tables
• Supplementary Data

Keith A. Bender, Ioannis Theodossiou, The unintended consequences of the rat race: the detrimental effects of performance pay on health, Oxford Economic Papers , Volume 66, Issue 3, July 2014, Pages 824–847, https://doi.org/10.1093/oep/gpt032

• Permissions Icon Permissions

Although performance pay schemes have been linked to labour market productivity, one unintended consequence, suggested early by Adam Smith, is that performance pay is detrimental to health. Recent research has shown that there is a positive relationship between performance pay and injuries on the job. This article focusses on the consequences of performance pay on health and investigates if there is a link between performance pay and self-reported general health or specific illnesses. Using data from the British Household Panel Survey, this study uses survival analysis to show that being in jobs with a performance pay element increases the likelihood of health deterioration, ceteris paribus .

The economics literature suggests that there is a link between performance pay and increased productivity. Thus research reveals that pay based on the performance of workers (usually in the form of piece rates, merit pay, or similar pay-for-performance schemes) generate higher productivity through an increased incentive for effort or by offering incentives for more highly productive workers to sort themselves into these types of jobs ( Lazear, 2000 ).

However, a related literature examines the unintended consequences of such payment schemes. If the contract is not set up appropriately, ‘adverse specialization’ ( MacDonald and Marx, 2001 ) can occur where workers perform only those aspects that are rewarded, whilst ignoring other aspects that are not. Thus, workers may stress quantity over quality or overuse physical capital to increase production. These unintended consequences of performance pay schemes have led some researchers to question whether these schemes actually increase profitability, despite the fact that they may increase worker productivity.

An interesting and little explored potential unintended consequence of performance pay is its effect on health. Adam Smith first observes in The Wealth of Nations , ‘Workmen … when they are liberally paid by the piece, are very apt to overwork themselves and to ruin their health and constitution in a few years’ ( Smith, 1776 , p. 83). Except for a limited number of studies (e.g., Foster and Rosenweig, 1994 ; Bender et al. , 2012 ), this observation has not been empirically tested, and the small literature that has examined this topic has focussed very narrowly on injuries at work. In line with the original idea of Smith, the aim of this article is to use a nationally representative data set to examine the effects of performance pay on other health outcomes, particularly ones affected by stress. Thus, the study investigates whether increases in the length of time paid using performance pay negatively affect ‘health and constitution in a few years’.

The following section briefly reviews the literature on the unintended consequences of performance pay and the small set of papers that link health to performance pay. Section 3 discusses the potential pathways that performance pay might impact health over time. Section 4 reviews the data and methodology. Section 5 reports the results, and Section 6 empirically examines potential pathways for the effect of performance pay on health. A final section concludes.

It is intuitive to think that certain job characteristics are correlated with health. For example, Case and Deaton (2005) use cross-sectional data to show that manual workers have worse health than nonmanual workers, even after controlling for factors such as education and income. More recently, Fletcher et al. (2011) find that long exposure to adverse physical demands and environmental conditions in their occupations cause workers to have worse health. However, most publications in this vein do not examine the effect of particular job characteristics.

The literature that links health and performance pay is small and is often focussed on case studies of particular industries or occupations. Sundstroem-Frisk (1984) finds that accident rates amongst Swedish loggers is lower for those paid hourly rates rather than piece rates. Saha et al. (2004) investigate fertilizer production workers in India to show those on performance pay experience higher workplace accidents. Freeman and Kleiner (2005) report evidence from a US shoe manufacturer that piece rates are associated with higher worker compensation costs. Toupin et al. (2007) report higher heart rates for Canadian loggers paid by piece rates, suggesting ‘negative consequences for worker health and safety’. Several papers on over-the-road truckers in the USA ( Monaco and Williams, 2000 ; Belman et al. , 2005 ; Rodriquez et al. , 2006 ; Williamson et al. , 2009 ) generally find that truckers paid by the hour have smaller probabilities of being in accidents compared to truckers who get paid by the mile.

Three papers are more closely aligned with this article’s focus. The first is by Foster and Rosenweig (1994) . Like others, they focus on a narrow occupation (subsistence farmers in the Philippines), but they link performance pay with overall health, as opposed to injuries on the job. They approximate overall health by the body mass index (BMI), which is shown to be lower for farm workers paid piece rates, ceteris paribus . Because these subsistence farmers have very low BMIs to begin, low BMI indicates worse health.

The other two papers are by Bender et al. (2012) and Artz and Heywood (2013) . Both focus on the link between performance pay and workplace injury. Unlike the papers just discussed, they consider the performance pay–injury relationship across many different occupations and industries. Bender et al. (2012) use a Europe-wide cross-sectional data set (the European Working Conditions Survey) to find a strongly robust relationship between workers paid by piece rates and higher probabilities of workplace injury, ceteris paribus . Artz and Heywood (2013) employ the US National Longitudinal Survey of Youth to find that this relationship exists even when controlling for individual and individual-employer fixed effects.

The goal of the current article is to extend these last papers by examining the relationship between performance pay experience and broad measures of health, using a nationally representative data set that covers many different occupations and industries. The difficulty with this approach is identifying a data set that includes measures of performance pay and health and contains a wide array of microeconomic information (e.g., socioeconomic data, education, earnings, etc.). Standard microeconomic data sets in the USA such as the Current Population Survey, the Panel Study on Income Dynamics, and even the National Longitudinal Survey of Youth do not generally contain all the necessary variables. (Indeed, this may explain, in part, the lack of research in the area.) However, these data do exist for a subset of years in the British Household Panel Survey (BHPS) and, thus, this data source provides the data for this study. 1

The literature discussed only alludes to potential pathways that might generate worse health due to performance pay. Bender et al. (2012) suggest that one pathway for workplace injury is the increased effort caused by performance pay giving incentives to increase workers' speed or take greater risks on the job. This point is echoed in several of the papers that focus on particular occupations.

Although the foregoing might explain the link between performance pay and injuries on the job, other pathways are possible when the cumulative effects of performance pay on health are considered. One way is through increased stress associated with performance-related pay. This stress might be generated in a variety of ways. Although not a focus of their study, Dohmen and Falk ( 2011 , Table 5) analyse data from a series of experiments to find that piece rate payments generated both more effort and more stress and exhaustion than those paid a fixed rate, particularly after participants were allowed to sort into a payment scheme. Medical evidence, such as the reviews by Cooper and Marshall (1976) and Blake et al. (1996) , shows that this physical stress can lead to adverse health outcomes, both physiological (e.g., skin problems, hypertension, and fatigue) and psychological (e.g., depression, emotional fatigue, and feelings of anxiousness). Furthermore, performance-related pay is generally found to be more variable than salaried or time-based pay. This variability may induce further stress with the increased uncertainty of an income stream. 2

Another potential pathway is an implication of the view that health status is determined through a health production function. This function could have many arguments, but one key aspect to better health is time—time spent in healthy behaviours such as exercise, learning about nutrition and good health, relaxation from the stress of work, cooking healthy meals, and so on . Thus, health production is a form of household production. If, as Lazear and others have shown, performance pay generates increased effort, this would imply more hours at work because the opportunity cost of time in healthy behaviour increases. Therefore, there will be fewer hours available to put into the health production function (e.g., Mullahy and Robert, 2010 ). This generates worse health. However, the effects may not be instantaneous (as they may be if a worker takes a greater risk in the job and an injury immediately results). One would expect that the effects may build up over time as, for example, workers spend less time in exercise and gradually gain weight, which, in turn, results to adverse health conditions such as heart disease.

There are several potential pathways that point to performance pay leading to adverse health outcomes, but one commonality is that any ill health from performance pay is not expected to be instantaneous. Hence, one may expect that long exposure to performance pay would manifest itself on health over a period of time. This indicates the need for a panel data set rather than a cross-sectional one, since one would want to observe the effect of performance pay on health over a period of time. Thus, the focus of this study is to evaluate the effect of performance pay on the duration of good health spells using duration/hazard models. Specifically, this study examines whether performance pay increases the hazard of falling out of good health. 3

In the data set, each individual is observed at a number of points in time. The time intervals are one year in length, and so the interval boundaries are t = 1,2, 3, … and the interval t is ( t − 1, t ]. Thus, the dependent variable is the dummy variable indicating that at interval t , the individual remained in good health. Otherwise, the spell of good health has been observed to end. A spell of good health can either be complete ( c i = 1) or right censored ( c i = 0), that is, either the individual continues to be in good health at the end point or the individual left the sample before the end point. In this specification, reentry into the panel is not allowed (since this would introduce problems of endogeneity), so only a single completed or uncompleted spell for each individual is included.

Finally, the methodology adopted in this article establishes the direction of causality and circumvents the problem of endogeneity. In particular, the sample consists of individuals who have reported good health at the initial period (in wave 8 of the BHPS). When the individual exits good health, his or her time in good health is recorded and this event should be expected to be an outcome of events occurring during the intervening period. Thus, this procedure ensures that the direction of causality is from performance pay to health. In addition to this, unobserved heterogeneity is controlled for in the Jenkins (2004) methodology. As a further robustness check, indicators of past bad health are included to explicitly control for the initial conditions of poor health.

Nevertheless, one could argue that even without controlling for endogeneity, the expected endogeneity bias should move the coefficient towards zero. This is because if there is sorting, healthier individuals would be more likely to reap higher rewards from performance pay, which works in the opposite way of the arguments already given. Thus, even if one does not control for endogeneity using the foregoing methods, the estimates will be lower bounds of the effect of performance pay on health.

The data used in this study are from the British Household Panel Survey (BHPS). The key question which is used for the main independent variable is introduced in the survey in wave 8. The question is, ‘Does your pay include performance-related pay?’ Thus, there are 10 waves that provide this information on performance pay for workers in wage and salary jobs (self-employed are excluded), since the data set goes up to wave 19. 6 Because the issue at hand is to assess whether long exposure to performance pay generates worse health, the proportion of time (over the observation period) spent in performance pay is used. 7

This study uses four measures of health. The first is a measure of subjective overall health and is derived from the question, ‘Please think back over the last 12 months about how your health has been. Compared to people of your own age, would you say that your health has on the whole been excellent, good, fair, poor or very poor?’ To specify the duration of good health, the following methodology is used: Individuals who report in wave 8 that they are in excellent or good health (and are employed in a wage and salary job to identify performance-related pay) are selected. The periods that workers who remain in this status are counted until workers either drop out of the survey whilst still in good health (perhaps because of nonresponse or because of a loss of employment), continue in good health in all waves, or report fair, poor, or very poor health. The first two groups are ‘censored’ observations.

Three measures of physical health are also analysed—whether workers suffer from heart problems, stomach/digestive problems, or anxiety/depression—which are all closely aligned with adverse health reactions to stress. As with subjective health, each sample starts with individuals who report in wave 8 that they are not suffering from the respective illness. Then periods of good health are counted until workers either drop out of the survey in good health or report the onset of the health problem.

The regressions also control for standard demographic and health variables, including gender, education, age, current smoking status, four categories of earned income, marital status, the log of hours worked, broad industry and occupational indicators, and region of residence. All of which are measured at the time of the last observance of the respondent.

5.1 Survival plots

Figure 1 shows Kaplan-Meier survival plots for the subjective health measure by whether the respondent has been in performance pay less (‘perfpercent50plus'=0) or more than 50% of the observation time. The graph clearly shows there to be a greater hazard of falling into poor health for those in performance pay at least 50% of the time.

Survival plots for subjective health.

One complicating factor to these patterns might be income. Despite the fact that the UK has universal health care through the National Health Service, it is still likely that those who have higher income may enjoy superior health compared to lower income people. Since performance pay will likely generate higher income, there could be a confounding influence of increased income helping keep good health survival rates higher for those paid by performance. In Fig. 2 , the sample is disaggregated by whether a worker’s earned income is above the median (‘highincome’ = 1) or below the median (‘highincome’ = 0). The highest hazard is for the group of workers who have been in performance pay more than 50% of the time and have income below the median. However, the negative effect of performance pay is offset somewhat by increased income as shown in the hazard for workers with 50% of time in performance pay and income above the median—a similar hazard to those with low income but low exposure to performance pay. The lowest hazard is for the high income and low performance pay group.

Survival plots for subjective health, controlling for earned income.

Finally, Fig. 3 shows the survival plots for the three objective health measures. As with the subjective measures, those who are in performance pay for more than 50% of the time have lower survival probabilities, particularly if their earned income is below the median. 8

Survival plots for the three objective measures of health.

5.2 Regression results

In addition to income, there could be a variety of confounding factors driving these results. Hence, controlling for worker heterogeneity is important. For the subjective health measure, the hazard ratios for the percentage of time spent in performance pay are reported in Table 1 for different econometric specifications and samples. 9 The first column reports hazard rates from Cox proportional hazard regression results. For the overall sample, increases in the percentage of time spent in performance pay are associated with higher odds of falling into worse health. Although the point estimate is statistically significant at the 1% level, it seems that the marginal impact is quite small (an increase in odds by 1.004), but this is only for a one percentage point rise in the percent of time for performance pay. A one standard deviation rise in percent of time (about 25 percentage points) would generate much higher odds of poor health, with the projected hazard at 1.103. The next column reports the odds ratio on performance pay using the Prentice and Gloeckler (1978) method that controls for discreteness in the hazard functions. As can be seen, controlling for this discreteness generates little difference in the odds ratio, increasing it from 1.004 to 1.005, whilst maintaining its statistical significance.

Selected results for hazard ratios for overall subjective health: Odds ratio on percentage of time spent in performance pay

Notes: z -statistics under odds ratios. All regressions also include a constant and indicators for gender, noncompletion of secondary education (excluded), completion of secondary education, completion of postsecondary education, age less than 26 (excluded), age between 26–35, age between 36–45, age between 46–55, age above 55, married, currently smoking, labour income quartile, the log of hours worked, broad industry and occupation, and region where the PG regressions also include log of time. **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. The sample is those people who were working in wage and salary jobs for the entire time we observe them and who start out with excellent or very good subjective health in wave 8. For the overall sample, there are 2,410 observations where 51% are censored.

The control for unobserved heterogeneity or frailty comes through the use of the Prentice-Gloeckler (PG) ( 1978 ) methodology. Interestingly this generally increases the hazard ratio. So although the effect of a 1 percentage point increase in the amount of time spent in performance pay increased the odds by 1.005 times without controlling for frailty, the odds jump to 1.013 when frailty is controlled for (or a projected hazard of 1.396 for a 1 standard deviation rise in the percent of time in performance pay). However, the pattern of performance pay correlating with worse health outcomes remains regardless of controlling for these frailty effects.

The foregoing regressions circumvent the problem of endogeneity. The control for endogeneity comes through the set up of the sample, where the sample consists of individuals who have reported good health at the initial period (here in wave 8 of the BHPS). When the individual exits good health, his or her time in good health is recorded, and this event should be expected to be an outcome of events occurring during the intervening period. Thus, this procedure ensures that the direction of causality is from performance pay to health circumventing the problem of endogeneity. Nevertheless, as a robustness test a series of dummy variables of the number of times the worker had reported bad health in the first eight waves of the survey are included as a measure of the state of worker’s previous health. As can be seen in the second row, this has no appreciable effect on the point estimate or statistical significance. 10

Furthermore, Table 1 contains the results of the subjective health measure for different samples. There is little difference by gender for the Cox or PG without frailty regressions, and the odds are generally higher when frailty is controlled for all samples, although the point estimate is statistically insignificant for the female sample in the frailty corrected regressions. Unlike in the figures, it does not seem to matter much if one is above or below median income (after controlling for the other covariates including income) as the hazards are nearly identical regardless of the estimation method. Finally, workers in manual occupations generally have higher hazard ratios than workers in nonmanual occupations, ceteris paribus .

Table 2 contains the hazard ratios for the percentage of time in performance pay for the three objective health measures by regression method and sample. Generally, the results show that increased time in performance pay, ceteris paribus , leads to a higher hazard of poor health, typically in the 1.005 to 1.010 range for the Cox regressions and from 1.008 to 1.049 for the PG regressions with frailty corrections. Most, although not all, are statistically significant at least at the 10% level (only 10 of 72 coefficients are not statistically significant). Although one cannot directly test the difference in the hazard ratios across health groups, it seems that women have slightly higher hazards for both heart health and anxiety/depression, whereas men have higher hazards for stomach health. There are no consistent differences by income level for heart health or anxiety/depression, although higher income workers have higher hazards for stomach health. Nonmanual workers are more susceptible to heart problems, whilst manual workers are more likely to have stomach and anxiety/depression problems. Controls for heterogeneity (either by including past health or frailty controls) do not seem to affect the pattern of results in any appreciable manner.

Hazard ratios for the percentage of time spent in performance pay by objective health measure

Notes: Same covariates as in Table 1 . Numbers under hazard ratios are asymptotic z -statistics. *, ** and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. Sample sizes are given in the table.

5.3 Robustness checks

A series of other regressions are used to investigate how robust these results are. First, even though the study controls for unobserved heterogeneity and endogeneity, there is still the possibility of not adequately controlling for the level of risk via the workers attitude to risk. One may argue that attitude to risk may affect health (e.g., Kowert and Hermann, 1997 ; Nicholson et al. , 2005 ; Khwaja et al. , 2006 ; Lia and Liu, 2008 ) and that risk-averse workers may not opt for performance pay. Although in the regressions there are already controls for some factors that are correlated with risk, such as smoking behaviour ( Viscusi, 2001 ) and industry and occupation, there is no direct measure of risk attitude included in the regressors. The BHPS does not include questions that provide appropriate measures of attitude to risk. However, two variables are used to approximate attitude to take risk: (i) have reported gambling winnings, and (ii) if the individual can be characterized as an introvert. 11 The inclusion of either sets of controls do not affect the results appreciably.

Second, one might be concerned about the samples used in the study. Two issues are investigated. First, given that there is only a limited panel, there may be an unavoidable discreteness in the percentage of time spent in performance pay. Thus, models in which individuals are observe for at least four waves are estimated. As before, the results do not change appreciably. Finally, a series of regressions based on a sample of only those with some performance pay—that is, on the selected group—are estimated. The point estimates tend to be slightly higher than those reported in Tables 1 and 2 , but all are statistically significant and indicate that the longer one stays in performance pay, the higher the odds to drop into poor health. 12

Overall, there seems to be consistent evidence that increases in the percentage of time in performance pay generates a higher hazard of falling out of good health, ceteris paribus . However, as already discussed, it is interesting to explore what pathways might be generating this effect. One potential source might be on the amount of time available for ‘healthy’ activities. If performance pay increases time spent at work, there will be fewer hours in the day, potentially, to engage in healthy behaviours. Taking the sample of healthy workers from wave 8 of the BHPS (the starting sample from before) and running a regression of total weekly hours (both regular and overtime hours) on performance pay and a series of other demographic and job characteristics, 13 we find that the coefficient on performance pay is 1.29 and is significant at the 10% level, meaning that workers on performance pay work almost 1.3 hours more than those not in performance pay. If a sample of all workers is used, regardless of their health, the marginal effect is 1.5 hours, which is statistically significant at the 1% level.

Unfortunately, the BHPS does not have a large array of information on the leisure activities of respondents. Nevertheless, for wave 8 there is information on the amount of time participating in sports, participating in evening classes (including ‘keep fit’ light exercise classes), drinking, and eating out. These are all five-point ordered responses from ‘Never/almost never’ to ‘at least once a week’. Table 3 has selected results from ordered probit estimations using wave 8 data, both with the ‘Healthy Worker’ sample used in the duration analysis but also for ‘All Workers’ in wave 8, regardless of health. Interestingly, there seems to be little effect on some leisure activities like participating in sports or evening classes. However, there is an increase in participation in both frequent drinking and for eating out. Although these are not necessarily unhealthy behaviours, dietary guidelines on both sides of the Atlantic suggest moderation in both drink and the consumption of heavier meals served in restaurants, and this may be a pathway to the ill health found before.

Selected ordered probit regression results for coefficient on percentage of time spent in performance pay

Notes: Regressions based on data on wage and salary workers from wave 8 of the BHPS. Other covariates include gender, education, age, age squared, marital status, indicator for currently smoking, labour income, occupation, and region. z-statistics are under coefficient estimates. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Another potential pathway is through increased stress which, as discussed before, the medical literature has linked to adverse health outcomes. The BHPS lists several self-reported measures of stress through the General Health Questionnaire (GHQ), including difficulties with sleep, difficulty making decisions, feeling under strain, problems with ability to overcome difficulties, loss of confidence, and general happiness. As with the health outcomes, a sample from wave 8 of the BHPS is used where the respondents are reporting ‘low stress’ (e.g., no problems with sleep, no problems making decisions, not feeling under strain, no problems with ability to overcome difficulties, no loss of confidence, and feeling generally happy) and track these workers over the next 10 waves of the BHPS to assert whether they report a higher stress level. In addition, as before, this is estimated for the standard Cox model, as well as the PG model without and with frailty for several different samples.

The results for these regressions are found in Table 4 . Overall the general results mimic the results for health—namely, that an increase in the percentage of time spent in performance pay increases the hazard to report more stress (across all stress measures) or lower happiness. For sleep difficulties the Cox estimates are around 1.003 or 1.004 for all subsamples, including when past sleep problems (before wave 8) are controlled for. The hazards also increase when frailty is controlled for to at least 1.016 (when past health is controlled for) and as high as 1.031 for manual workers. All of the frailty-controlled results are statistically significant at least at the 10% level. This pattern is remarkably consistent for each stress measure or for subjective measures of happiness with most Cox hazards in the 1.003 to 1.009 range and the frailty-corrected PG hazards between 1.009 and 1.041, and except for a handful of samples, all are statistically significant. Although this is not a proof of a link between performance pay and health, it is suggestive of a pathway where performance pay may affect the health of workers.

GHQ: Stress results. Hazard ratio on percentage of time in performance pay by duration model and sample

Notes: Duration model starts with respondents in wave 8 saying that they have GHQ difficulty ‘not at all’ or ‘no more than usual’ or that happiness is ‘much more than usual’ or the ‘same as usual’. Thus failure happens when the respondent reports difficulty ‘rather’ or ‘much’ more than usual. Frailty controls for log duration (last three time periods combined). Regressions based on data on wage and salary workers from wave 8 of the BHPS. Other covariates include those in Table 3 . z-statistics are given under odds ratios. Sample sizes given in table. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Previous research shows that performance-related pay generates increased effort and productivity at work. However, it may also generate a series of unintended consequences. Observations by Adam Smith indicate that one such consequence might be on the health of workers, although little has been done to estimate whether his observation has empirical validity. Using multiple waves of the BHPS and duration models, this study finds that increasing the time in which workers’ pay is made up at least partly of performance pay generates higher odds of falling into bad health—measured either by subjective health or along one of three physical health dimensions. These results are very robust to variations in variables or estimation procedure. In addition, the estimates serve as a lower bound of the negative effect of performance pay on health since even if endogeneity is not perfectly captured in the regressions, there will be a likely positive bias due to healthier workers selecting performance-pay jobs.

Furthermore the study investigates potential pathways in which this linkage between performance pay and health might lie. Whilst performance pay is found to increase work time, it does not seem to be significantly correlated with activities designed to promote health, although performance pay is correlated with increases in drinking and restaurant meals. A more consistent pattern is found with a series of measures of stress, where increases in the time spent in performance pay increase the hazard of five measures of stress.

Although a very robust relationship between performance-based pay and the duration of good health is uncovered, there are several limitations in the article. First, we measure performance pay only broadly. The strongest theoretical effects should come through individual-based performance pay, but this information is not available in the BHPS. Similarly, one cannot measure the intensity of individual pay (that is, what proportion of pay is based on performance), meaning that the implicit assumption that any intensity of performance pay generates the same health outcomes is entertained when the effects are likely strongest for those whose pay is mostly determined by performance. Finally, the study cannot control for changing jobs, which can also induce stress. If workers are forced to change jobs that happen to have a performance pay aspect, there may be a stronger correlation between performance pay and ill health because of this job change stress. Yet there will also be workers who are forced to change jobs that will have no performance pay, and thus, it is not clear how strong this bias might be. However, the effects of such unobserved factors are mitigated by the frailty corrections for omitted variables ( Jenkins, 2005b ).

Performance pay can generate a variety of efficient labour market outcomes. However, the findings here are firmly in the camp of a potential unintended consequence of performance pay and support the insight of Adam Smith. Like other research that finds that performance pay can lead to workers, for example, focussing on quantity rather than quantity or overusing physical capital, long-term exposure to performance pay is related to worse health, suggesting that firms may face increased health insurance or workers compensation costs (e.g., Freeman and Kleiner, 2005 ) and the society as a whole may bear the costs of a less productive and less healthy workforce with important repercussions for the health services. Perhaps to mitigate these increases in costs, future research, using more detailed health data, should focus on identifying the pathways through which performance pay can affect health. Unfortunately, the BHPS offers information on few and relatively crude approximations of the pathways. It may well be through increased stress, as suggested in the results here, but this should be more explicitly examined with more detailed data. Further research may extend this argument to other forms of payment schemes. For example, stress and other factors that can induce ill health may increase in ‘tournaments’ where there is a ‘winner-take-all’ payoff, so that the effects of fierce competition for promotions would be an interesting extension of this research.

The European Commission, 7th Framework Programme THEME [HEALTH-2007-4.2-3] (Grant agreement no: 200716).

The authors are thankful for comments from participants of the 2010 ALMR Conference, the 2012 HEALTHatWORK Conference, the 2013 Scottish Economic Society Conference, the 2013 Beyond Wages Conference and seminar participants at the University of Wisconsin–Milwaukee and the Health Economics Research Unit at the University of Aberdeen as well as discussions with John Heywood, Colin Green, and Yu Aoki. The authors are also grateful to the anonymous referees of this journal for very helpful comments.

1 One data set that does have these measures is the US Health and Retirement Study, but it has the unsatisfactory aspect of including a very limited age range of 51- to 61-year-olds in 1992. Thus, it is more appropriate to use a survey such as the BHPS, which includes workers of all ages.

2 Of course, this would be mitigated by any compensating wage differential paid for performance-related pay. Furthermore, there is some evidence from recent papers such as Grund and Sliwka (2010) and Cornelissen et al. (2011) that less risk-averse workers sort into jobs that pay for performance. However, as shown later and in Artz and Heywood (2013) , the inclusion of proxy measures of risk tolerance does not affect the health–performance pay relationship.

3 This approach fits into a newer line of research that is examining such issues as the effect of unemployment on the duration of good health as in Cooper et al. (2008) .

4 Hanagal (2011) points out that Vaupel et al. (1979) use the term frailty to indicate that different individuals are at risk, although they may appear to be quite similar with respect to measurable characteristics such as age, gender, weight, and so on. They use the term frailty to represent unobservable random effects shared by individuals with similar (unmeasured) risks affecting the mortality rates. A random effect describes excess risk or frailty for distinct categories, such as individuals or families, over and above any measured covariates (p. 71). Further discussion appears also in Wienkle (2011) .

5 According to Jenkins, this methodology is an adaptation of a frailty correcting methodology proposed by Prentice and Gloeckler (1978) and modified by Meyer (1990) . This study follows the methodology as it is set out in Jenkins (1995) by using the pgmhaz8 subroutine available in STATA ( Jenkins, 2004 ). Note that there is a second similar methodology ( Jenkins, 2005a ) that uses discrete mixture distribution to summarize unobserved individual heterogeneity ( hshaz ) outlined in Cooper et al. (2008) . Using this methodology, the qualitative results are very similar to the reported results herein and are available on request.

6 There is a question asked every wave about whether pay has any bonus payment in it. Estimations using this variable generated no statistically significant results, although ‘bonuses’ is too broad a term for the purposes of this study because they may or may not be based on personal performance and thus may not generate the kinds of behaviour that might decrease health. Indeed, this is potentially an issue for the question from which performance pay is identified, since it could be considered broadly to include group-based performance. One might argue that the negative health effects would be less pronounced with such group-based forms of compensation that might bias the results to finding no effect. That a consistent relationship is found indicates that this effect is relatively small compared to the effect of individually based performance pay.

7 A number of variations on this key variable are also used, including a dummy variable for whether the respondent was in performance pay in the last period of observation or dummy variables of splines of the proportion of time spent in performance pay. Results using these variations are qualitatively similar to those that follow and are available on request.

8 Formal tests for equality of the survivor functions generally failed for all figures, particularly amongst the relatively low earnings group. Results are available from the authors.

9 Full results for the overall sample for subjective health for the three econometric specifications are given in Appendix Table A1 . Generally the hazard ratios for the independent variables correspond to intuition and previous findings. Full results for all regressions are available from the authors.

10 However, as mentioned, even if the endogeneity is not completely circumvented, the direction of the bias is almost certainly towards a positive relationship between performance pay and health, since those who are generally more healthy will naturally be more likely to want to be paid by performance. Thus, if one finds consistently negative and statistically significant relationships, in the presence of endogeneity, this implies that the regressions are underestimating the negative effect of performance pay on health.

11 This comes from a question in the BHPS that asks whether a person thinks that ‘Life is full of opportunities’. People who answer ‘No’ to this are labelled introverts.

12 These results are available from the authors on request.

13 Other controls include gender, education, age, age squared, marital status, currently smoking, labour income, occupation, and region. Regressions that include total numbers of hours give similar results.

Google Scholar

Google Preview

Full results for subjective health for full sample: Cox, P-G with and without frailty: odds ratios

Notes: z -statistics under odds ratios. Regressions also include a constant term. The excluded variables are: did not complete secondary education, age under 26 years, income in 4th quartile, skilled nonmanual occupation, regions of England, Wales and Northern Ireland that are not in greater London or in Scotland, and services and banking industry. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Email alerts

Citing articles via.

• Recommend to your Library

Affiliations

• Online ISSN 1464-3812
• Print ISSN 0030-7653
• Copyright © 2024 Oxford University Press
• About Oxford Academic
• Publish journals with us
• University press partners
• What we publish
• New features  
• Open access
• Institutional account management
• Rights and permissions
• Get help with access
• Accessibility
• Advertising
• Media enquiries
• Oxford University Press
• Oxford Languages
• University of Oxford

Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide

• Copyright © 2024 Oxford University Press
• Cookie settings
• Cookie policy
• Privacy policy
• Legal notice

This Feature Is Available To Subscribers Only

Sign In or Create an Account

This PDF is available to Subscribers Only

For full access to this pdf, sign in to an existing account, or purchase an annual subscription.

Winning the rat race: The effect of peer salaries

Winning the rat race: The effect of peer salaries  

April 9, 2024 | by professor ron kaniel .

In this blog post, Professor Ron Kaniel explains why relative compensation matters and why wage transparency requirements can backfire.   

In economics, contract and incentive theory seeks to uncover what motivates employees to perform at their peak. Crack that code, the thinking goes, and you can design a system of incentives that gives organizations a leg up over their competition. But when it comes to negotiating employment contracts, the gap between economic theory and the reality on the ground can be stark—and the difference lies in human nature.

 In “ Contracting in Peer Networks ,” my co-author and I take a deep dive into how people consider their compensation in relation to their peers when entering into a contract. We had previously examined, together with another co-author, the impact of relative wealth considerations on things like portfolio choice decisions and asset pricing, showing they can play an important role in explaining the home bias puzzle, existence of financial bubbles, and over-investment. A natural next step was to extend this line of thinking to employment contracts.

What theory tells us

From existing research, we know several things about compensation:

You can’t observe the effort an employee actually exerts, only the final output.

If I’m in charge of hiring and evaluating a CEO for a company, I approach the process of creating a contract with a basic fact in mind: There is no one-to-one mapping between the amount of effort that CEO exerts and their work output. Ultimately, output is a combination of effort and external factors. For example, during the tenure of a CEO of an oil company, increasing global oil prices will make the company more profitable, unrelated to the CEO’s efforts. There are all sorts of events—political elections, natural disasters, unforeseen market events, etc.—that create noise when I’m trying to gauge the CEO’s actual performance. 

Relative performance theory (RPE) tries to account for the noise. 

There is a large body of literature that recommends using relative performance evaluations (RPEs) to reward employees only for the output that can be linked to their own performance, not just overall output that could be linked to any number of factors. In theory, I should create a contract that looks at the average of how similarly situated CEOs have performed and reward the CEO based on output relative to peers. If my CEO is producing a greater output than peers, that is a sign that I should compensate the CEO more. If that output is less, I should compensate them less. Essentially, I want to control for the outside factors that constitute noise when it comes to a CEO’s output. 

Theory vs. reality

In the real world, many contracts are not structured as theory would suggest..

Empirical evidence suggests that actual contracts are not structured in a way that RPE theory predicts. Take a scenario in which a CEO increases output by 25%, in line with others in the industry. Then consider a second scenario in which that CEO increases output by 25% but the rest of the industry increases output by 50%. RPE theory predicts that that the CEO will be compensated less in the second scenario, to reflect the fact that their individual relative performance was not as strong as in the first scenario. But in real life, in many cases the CEO is compensated more in the second scenario. The rising industry tide lifts all boats. 

Motivated by the discrepancies between theory and practice, my co-author and I began to explore the idea that CEOs are motivated by their wages in relation to peers, not just their absolute wages, and examine its implications for the contracts they receive. 

Below, I share some of the conclusions we reached. To read the full paper, click here . 

Accounting for peer effects generates more efficient contracts. 

A CEO may outperform their peers at some times and underperform in others. A higher base pay can mitigate dissatisfaction in those cases where they would have been paid less than peers based on performance, but that higher base pay is inefficient in that it keeps pay high in situations where the CEO is already satisfied with the state of things. Instead of increasing the base pay, a more efficient contract is one that directly targets those situations in which the CEO underperforms peers, and thus is most likely to be dissatisfied about their compensation relative to others.

The peer effect leads to a rat race. 

The CEO I hire has an incentive to exert more and more effort, not just because they will get paid more but because they will experience the satisfaction of feeling like they are compensated more highly than peers. Unfortunately, this mentality leads to overexertion of effort. When one CEO works harder, their peers take notice and ramp up their own efforts, which ends up raising the bar for everyone. Consequently, all CEOs exert too much effort, their compensation goes up faster than output, so the firm experiences reduced profits as a result. 

Disclosure requirements only amplify the problem.

In a bid to improve transparency, the SEC introduced wage disclosure requirements so that the contracts of high-level managers are now publicly available across firms. This has only amplified the problem. Firms that write contracts realize that their company’s performance can impact the compensation of other CEOs, impacting the satisfaction of their own CEO and raising the cost of compensating them. Instead of tamping down on CEO pay as intended, disclosure requirements actually increase overall CEO pay and cut deeper into firms’ profits. 

When it comes to understanding the motivation of high-level employees, classic economic theory breaks down. In real life, the average CEO is keenly aware of how their peers are performing and what they are earning. The effort to win the rat race or “keep up the Joneses” is not just an interesting social phenomenon—it’s a dynamic with significant implications for performance, profits, and regulation. 

Ron Kaniel is the Jay S and Jeanne P Benet Professor of Finance at Simon Business School. 

Follow the Dean’s Corner blog for more expert commentary on timely topics in business, economics, policy, and management education. To view other blogs in this series, visit the  Dean's Corner Main Page .

Add new comment

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Technology Feature
• Published: 14 July 2021

EXPERIMENTAL ORGANISMS

Rats on the rise

• Ellen P. Neff 1  

Lab Animal volume  50 ,  pages 205–208 ( 2021 ) Cite this article

2900 Accesses

5 Citations

12 Altmetric

Metrics details

After a genetic revolution in the 80s, mice overtook rats as the laboratory animal of choice for many researchers. But in recent years, the gene editing capabilities that had lagged a little for the larger rodent have been coming up to par with their murine cousins. Is a return to rats on the way?

You have full access to this article via your institution.

Make no mistake, rats are not just big mice. Both may be rodents, but millions of years of evolution separate Rattus norvegicus from Mus musculus . Rats are obviously larger – about ten times the size of a lab mouse – but they are also more social animals and can be quicker to pick up more complex tasks. They also share more physiological similarities with humans than mice while still being a small animal, says Jean Cozzi, an innovation manager at Charles River. For over a century the rat was king, the preferred rodent model for many in the biomedical field and the animal behind much work in physiology and toxicology.

That changed in the 1980s with a revolution in genetics. Researchers found themselves able to manipulate the genes of mice – but not necessarily rats – in increasingly complex ways. “Regardless of whether mouse is the best model or not, it was the only model available for genetic modification,” says Yuksel Agca, a researcher at the University of Missouri and co-investigator of the Mutant Mouse Resource & Research Center and Rat Research and Resource Center there.

Mus musculus usurped the throne, and the number of mouse strains used in research has surpassed that of rats over the past thirty odd years. Mouse Genome Informatics (MGI) , a bioinformatics database dedicated to all things mouse genetics, has recorded over 66,000 different mouse strains; the number of rat listed in the related Rat Genome Database (RGD) is just shy of 4,000. Looking into the literature, 2020 saw over double the number of publications for “mouse model,” ~44,000, to just shy of 19,000 for “rat model.”

A lot has been – and will continue to be – done with the mouse. But there are advantages to using a rat as your model organism – its physiology for pharmacology, and its cognitive abilities for studying neurological function, for example, “which in a certain way got lost in molecular biology with the transgenic mouse,” says Bart de Strooper, an Alzheimer’s Disease researcher at the VIB-KU Leuven Center for Brain & Disease Research.

Transgenic rats are here nonetheless, and a second genetic revolution has made them easier to make than ever. Rats are poised for a return, with resources in place for those interested in revisiting their choice of rodent.

Leveling the playing field

The 1980s brought two technological advances that gave scientists a means to manipulate mammalian genetics. The first, early in the decade, was pronuclear injection, in which a DNA construct is injected into a zygote; the developing animal will incorporate that DNA into its own, resulting in a transgenic animal. Pronuclear injections were used to study ‘gain of function’ genetics, in which the manipulated gene is over expressed. Not long after, scientists isolated embryonic stem cells; genes in these cells can be manipulated via homologous recombination or gene trapping to ‘knock out’ that gene’s function.

These technologies aren’t perfect but worked well off the bat in mice. Rats, however, were a different story. It took until the 1990s for pronuclear injection technology to make it to the rat, says Agca, because it was a less efficient than in mice. Rat embryos proved more subject to damage from the injected DNA, and in those that did survive, gene integration tended to be lower in rats than in mice, meaning hundreds of zygotes were needed to achieve a handful of the germline transmissions needed to create a founder animal and establish a transgenic rat line (an added complication: super-ovulation protocols to yield extra eggs have not been quite as effective in rats either, notes Agca.).

Creating knock outs was also problematic in rats. The necessary embryonic stem cell lines have been relatively easy to establish in mice says Agca, with groups like the International Mouse Phenotyping Commissions systematically knocking out mouse genes one-by-one. For in vitro reasons that researchers are still working out, embryonic stem cells isolated from rats, however, don’t like to stay embryonic stem cells; instead, they tend to differentiate into other cell types, rather than waiting to be edited and raised into a full rat.

The first transgenic rat did nevertheless turn up in 1990 – an animal for studying hypertension 1 - and knockouts, and later knockins, have been produced using the few rat embryonic stem cell lines that are available as well as with technologies that avoid the need for those cells, such as Zinc Fingers (here, rats beat their murine relatives, in fact) and TALENS (see ref. 2 for a review on other firsts in rats compared to mice).

But, the ease of editing mice nevertheless led more and more researchers towards the smaller rodent. “Historically, rats were preferred in certain areas,” says Elizabeth Bryda, director of the Rat Research and Resource Center (RRRC) and the University of Missouri Animal Modeling Core. “When we were able to genetically manipulate mice, people started turning more to the mouse, only because they could knock out genes, they could do these complicated genetic manipulations.”

Rats fell behind…until a technology called CRISPR came along. “It changed everything,” says Agca, leveling the rodent playing field.

The ‘molecular scissors,’ recently awarded the Nobel Prize in Chemistry, avoid the need for embryonic stem cells and can be used to create genetically modified animals – of nearly all shapes and sizes – more quickly and inexpensively than prior approaches. To create knock out animals, you’ll need microinjection equipment, your zygote, and a genetic construct that tells the Cas9 enzyme where to cut the organisms’ DNA. “Now there’s no barrier - we can easily manipulate the rat in any way that you can in a mouse,” says Bryda. That includes the ability to overexpress genes and knock them out across the animal from the get go as well as conditionally.

The first applications of CRISPR in rats were published in Nature Biotechnology in 2013 3 , 4 , and the technology has since been used to produce rat models of a number of different human diseases 2 . “It has pushed the rats back into the research race,” says Cozzi. As their physiology is closer to humans than mouse, CRISPR will help researchers create more relevant models in areas such as cardiovascular research, metabolic disease, neuroscience, and cancer. Rats and humans share genomic regions in different cancers, for example, and he notes that breast cancers in rats are hormone responsive and have more stage similarities to humans than mice. “I think rat people will stick to the rat more than ever,” he says, while those who worked in mice may start considering the larger animal as well, to help reveal different pieces of their story. “In research, you need different models,” Cozzi says.

De Strooper has long used transgenic mice to study Alzheimer’s disease but these, he says, are incomplete models. They’ll develop the amyloid plaques and tau tangles characteristic of the disease in humans, but any neuropathological phenotypes are weak. He’d be thinking about a rat model for over a decade - viral vector-based transgenic rat models were promising, but without the ability to create stable embryonic stem cells, those lines themselves wouldn’t be stable. With the advent of CRISPR, he and his lab finally decided to give it a try, creating an Psen1 knock-in rat 5 . They successfully edited the animals, but unfortunately couldn’t age them long enough to observe any meaningful phenotypes – the Long Evans rat they chose has a propensity to develop tumors. Work with humanized mice goes on but encouraged by the ease of using CRISPR, they are trying again with a different rat: Fischer F344. Those animals are almost ready, says de Strooper’s colleague, Lutgarde Serneels, after which they will need to age for about two years to see if they develop both the physiological markers of Alzheimer’s and its functional deficits.

The lab has big plans for rat – if it works out. “If this rat model develops full Alzheimer’s, it will replace our mouse models,” de Strooper says. For decades, molecular biologists have focused on all the things they can do with the mouse, but that’s only gotten the field so far. “We give ourselves two years to see what happens,” he says.

Rat resources at the ready

When Bryda took over at Missouri’s Animal Modeling Core in 2016, there was just one rat project in the works; today she says, half of the Core’s projects are to create transgenic rats, many following up from work previously done in mice. “Investigators will come to us and say, ‘Hey, there’s this mouse model – can you make the exact same thing, but in a rat?” she says.

Rats have long offered the advantage of complexity. Whereas much of the mouse field coalesced around the “Black6” strain, rats – lacking those embryonic stem cell lines – stayed a bit more genetically diverse, which may be better representative of the human condition. Among other rat resources, the RGD maintains information about the Rat Hybrid Diversity Panel – a collection of 96 strains that represent the majority of genetic diversity among lab rats; these are characterized at the genome level and preserved as a resource. As genome mapping work in animals shifted to genome-wide association studies in humans, that too has brought a shift towards working with genetically complex animals like the rat. “Now the attention is how can we translate those human findings into a rat model that we can experimentally manipulate?” says Aron Geurts, a researcher at the Medical College of Wisconsin (MCW) working to improve rat research.

An improved rat genome – the first version 6 of which was published not that long after that for mouse, says RGD director Anne Kwitek, will only help. “Compared to mouse and human where we probably know the majority of the genes – their starts and their stops and all of the different isoforms - the rat was really far behind,” says Geurts. In May, the Wellcome Sanger Institute, using DNA from a male rat provided by MCW’s Melinda Dwinell, generated a new rat sequence: mRatBN7.2 7 . The genome has been accepted into the Genome Reference Consortium, and RGD will help with curation; international efforts to analyze and annotate it are currently underway. The genome and improved gene annotations will be important for modern research techniques, such as transcriptomics, note Kwitek and Geurts. Researchers will be able to use it to study their genes and variants of interest, and to compare those against other species, such as mice and humans, to identify what’s unique and what’s different.

There is of course still plenty of room to improve rat research. CRISPR is yielding new models but in general struggles to integrate large constructs into a genome Efforts to apply electroporation to zygotes – common already to manipulate stem cells – may help, while also eliminating the need for relatively expensive microinjection systems but for now, ‘humanizing’ genes or replacing large portions of them will still require those embryonic stem cells that have been tricky to maintain for rat lines.

Once a transgenic zygote has been made, it needs to be transferred into surrogate mothers. Currently, that involves surgery, but as ultrasound and probe technologies improve Agca anticipates non-surgical embryo transfer approaches will become more common in both mice and rats; currently, it’s just hard to tell once you’ve passed the embryo through the cervix of an animal so small, he says. Combining CRISPR plus electroporation plus non-surgical embryo transfer would make the process to create transgenic animals incredibly easy, he says, as it would require minimal equipment and technical skill.

Transgenic animals also need to be preserved and distributed to others. While it is fairly straightforward to cryopreserve rat embryos, but this is a laborious process that can be expensive in terms of cost and the number of animals needed, says Bryda. Freezing sperm is cheaper, but that process has been suboptimal in rats Their sperm freeze just fine, but motility tends to be compromised after thawing, a problem associated with unusually long flagella of rat sperm - about 180 micrometers, compared to ~50 micrometers for human sperm (the rat egg, meanwhile, is about 80 micrometers in diameter, requiring the sperm to coil in for entry). Groups like the Rat Research and Research Center and the Center for Animal Resources and Development at Kumamoto University are working on how to preserve sperm motility post-cryorecovery, but it remains a work in progress. “It really is an art,” says Bryda.

With years of a head start, there are more resources available at different institution for breeding and working with mice. For rats, there are still only a handful of core facilities that routinely create transgenic models, plus a few commercial providers, such as Charles River, Taconic, and Cyagen.

Large, centralized rat facilities that bank and distribute rats to the community used to be more common, says Bryda, but those fell away as interest in rats waned. The NIH used to have one, for example, as did a group in Germany. Now, Japan maintains a centralized repository – the National Bio Resource Project for the Rat in Kyoto – while in the United States, the NIH currently supports the Rat Research and Resource Center as a centralized repository to bank rat models donated to the Center, re-derive the animals, and provide other rat services for labs interested but without their own rat capacity. “It runs the gamut – if we’re able to do it we will try,” says Bryda. “It frees up investigators to do their science and not be rat distributors.”

Rat groups like those at MCW will also help as they can, and have bred and shipped animals for investigators (in many cases, former MCW alums) who don’t have in-house capacity, says Geurts. Demand, however, may see more rat cores come online. Increased interest in rats will also help things scale – at MCW for example, the rat research programs are large enough that maintaining rats only costs about 40% more than mice, whereas at some instructions than ratio can be more like 3:1. “In most institutions, a barrier is the cost of rats,” he says, along with their size – they do take up more space. Until more researchers at different institutions demand to use rats – and that they get a farer housing costs to do so – infrastructure will remain a challenge, even if making the animals is easier than ever before. Nevertheless, the animals are coming (Ready? Box 1 ), and there are rat resources out there to help (see ref. 8 for an overview).

Adding a rodent can yield new insights. When Bryda’s lab was first learning how to use CRISPR a few years ago, they decided to recreate a mutation associated with Crohn’s Disease (CD) in rats that had previously been modeled in mice. Looking across the species, the rats seems to mimic the human inflammatory bowel disease physiology a bit better 9 . But whereas mice and humans can be homozygous for this mutation, in rats its lethal. The CD model only needs to be heterozygous to mimic the diseases, but that homozygosity is a problem in this gene could tell us a little more about general rat biology, she says.

So which rodent is it: mouse or rat? “There’s a place for both,” says Bryda. Do you need a larger animal for implants or surgery, or have a need for social behaviors? Maybe it’s time for a rat. If those are less important considerations, maybe stick with mice. It all depends.

“We really encourage people to think about what species is the most appropriate based on what you’re studying and the question you’re asking, and to maybe be more flexible about that,” says Bryda. “We tell people all the time: are you sure the mouse is the best animal model? Is there something else you should be working with?”

Box 1 Ready for rats?

Considering a rat? There are some things you should know. As bigger animals, they lend themselves a bit better to serial sample collection and repeated imaging. Surgery can be easier in a larger animal, and rats can carry greater weight than mice, such as those electronic implants for electrophysiology or optogenetics.

That size does mean you’ll get less rats to a typical cage - two on average, compared to five mice. Rats’ more social nature, however, does mean some mouse limitations you may be familiar with don’t necessarily apply, notes Bryda. For example, male rats don’t need to be weaned litter mates to share a cage, and the animals tend to be more amenable to being moved around than mice. Their social nature does mean a little more enrichment is warranted to keep them engaged in their cages.

Many rats need special bedding, water, and chow, and it is important to house them separately from any mouse cages. The rats won’t mind, but mice aren’t so fond of cohabitating near a natural predator. Barrier facilities for breeding will also need to be established if you want to create a long-term colony.

Don’t be intimidated by them either. “When I came [to MCW], I didn’t want to reach my hand in the cage the first time,” says Geurts, who originally worked with mice. But, strain dependent of course, rats can be more docile than mice, and some will even get to know you.

You’ll need to mind your nomenclature and should register your rats too – just as in mice, nomenclature is critical, and someone reading about your rat needs to know things like its strain, vendor, any environmental details, and how the knockout was created and validated, says Geurts, as phenotypes can vary by model. Nomenclature and registry help can be found at the RGD .

Mullins, J. J. & Ganten, J. P. Nature 344 , 541–544 (1990).

Article   CAS   Google Scholar  

Chenouard, V. et al. Front Genet https://doi.org/10.3389/fgene.2021.615491 (2021)

Li, D. et al. Nat Biotechnol 31 , 681–683 (2013).

Li, W., Teng, F. & Zhou, Q. Nat Biotechnol 31 , 684–686 (2013).

Serneels, L. et al. Mol Neurodegeneration 15 , 60 (2020).

Rat Genome Sequencing Project Consortium Nature 428, 493–521 (2004).

Howe, K, et al. [version 1; peer review: awaiting peer review]. Wellcome Open Res 6, 118 https://doi.org/10.12688/wellcomeopenres.16854.1 (2021).

Shimoyama, M. et al. ILAR J 58 , 42–58 (2017).

Men, H. et al. Mamm Genome 32 , 173–182 (2021).

Download references

Author information

Authors and affiliations.

Lab Animal http://www.nature.com/laban/

Ellen P. Neff

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Ellen P. Neff .

Rights and permissions

Reprints and permissions

About this article

Cite this article.

Neff, E.P. Rats on the rise. Lab Anim 50 , 205–208 (2021). https://doi.org/10.1038/s41684-021-00812-0

Download citation

Published : 14 July 2021

Issue Date : August 2021

DOI : https://doi.org/10.1038/s41684-021-00812-0

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

Optimized protocols for sperm cryopreservation and in vitro fertilization in the rat.

• Satohiro Nakao
• Naomi Nakagata

Lab Animal (2022)

Quercetin-treated rat sperm enables refrigerated transport with motility and fertility for five days

• Katsuma Yamaga

Scientific Reports (2021)

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• v.20(7); 2019 Jul

The rat race

Brooke morriswood.

1 Department of Cell & Developmental Biology, University of Würzburg, Würzburg, Germany

Oliver Hoeller

2 San Francisco, CA, USA

Labmates are colleagues, friends, comforters…and competitors.

“Every time a friend succeeds”, Gore Vidal quipped, “something inside me dies”. Though Vidal's peer group were writers, his remark captures a predicament that is common to novelists, scientists and indeed many spheres of human activity: your colleagues are your competitors.

In science, that in‐group jostling occurs at every stage of the career ladder, from PhDs to postdocs and beyond. The “groups” need not be research groups per se, only collections of peers—as hapless junior faculty can find when they learn they are all vying for a single professorial appointment—but they share one characteristic that makes that competition psychologically viable: structural equivalence. This state was defined by the sociologist Roger V. Gould as occurring when two people have the same relations with the same third parties—such as two postdocs who either just started in a research group or are nearing the end of their term at roughly the same time.

A consequence of structural equivalence is that the members of such groups are unaware of their relative standing in the pecking order and, as a result, more likely to come into conflict. A recent study by Piezunka et al 1 examined the implications of structural equivalence in a competitive environment. Analysing Formula 1 race data over a mind‐boggling 34‐year period, they found that the probability of two drivers crashing into each other—a manifestation of conflict, not bad driving—was highest when the drivers were close in age, both high‐performing and nearing the end of a season when the opportunity to establish dominance was becoming limited.

Scientific peers don't physically crash into one another, but the comparison with racing is apt because members of a structurally equivalent research group obsess about speed. Rate of progress. Productivity. In any research group, there will inevitably be a mix of velocities, ideally all at least in a forward direction. Some people will be on a roll, others will be going slower, some will feel like they're hitting a wall, or—worst of all—simply parked. It is easy to see how such a mixture of ambition and adversity can create a toxic environment. Although the Formula 1 study looked at crashes, what racing drivers would really do if they could is to slow down their rivals. We know that because when they can, they do. Lewis Hamilton drove slowly in the 2016 Abu Dhabi Grand Prix in the hope that his teammate and championship rival Nico Rosberg would be caught and overtaken; Michael Schumacher deliberately parked his car during qualifying for the 2006 Monaco Grand Prix to stop rival Fernando Alonso from getting the pole position.

Similarly, scientists have a range of tactics—some clandestine, others more overt—to retard a perceived rival's progress. The most time‐honoured, of course, is peer review. An experienced reviewer knows that it is not necessary to pour poison all over a manuscript in order to slow its progress, and a disingenuous call for more data can be enough to extend the paper's publication date by many months. Public presentations of data in group meetings, seminars and conferences likewise offer a means of calling perfectly sound results into question, undermining rigour and gently insinuating that someone's work is either slapdash or somewhat untrustworthy.

In the laboratory, when members are often unsupervised amid a shared set of reagents and equipment, there is an even greater potential for slowing down competitors, even as far as crossing the line into unethical if not criminal conduct. The case of Vipal Bhrigu at the University of Michigan in 2010 offers an extreme illustration of what can happen. Bhrigu was filmed using an ethanol spray bottle in a cell culture refrigerator containing media belonging to graduate student Heather Ames. Ames’ work had been suffering for months from unexplained phenomena that seemed likely to be due to switched labels on cell culture flasks or even outright sabotage of her media. Most telling in terms of group dynamics is that Bhrigu's own project was unrelated to that of Ames; in his confession, he pleaded “I just got jealous of others moving ahead and I wanted to slow them down”. Here again is the giveaway that structural equivalence, not direct competition, is what underpins the scientific rat race—and all but the most saintly of scientific practitioners are bound to be unnerved by the successes of their assumed equals.

So what is the solution? Is there a solution, when any kind of grouping is bound to have an overt or assumed pecking order, which inevitably causes structural equivalence? Especially, when that order itself is so heavily dependent on serendipity and good fortune on top of smarts and hard work?

The remedy, if it is one, can come from within, but comes best from the top. In research groups, that means from the principal investigator; within departments from the head; and within research communities from the leaders of that discipline. That remedy is to promote a sense of equality between peers. It is also about focusing on the unique contributions each person can bring to an endeavour, not on who does one thing best. With such a welter of techniques, knowledge and expertise in the scientific world, the very notion that two people can even compare themselves is daft in some measure. Everyone brings their own strengths to the project.

One of the great things with science is the way it combats nepotism and cronyism through competitive allocation of research funds and community‐led assessment of research data. An awkward corollary is that this culture of self‐regulation through unremitting but ideally constructive criticism can easily be misinterpreted to mean that each individual is in competition with all other members. This is not the case. Even allowing for the instinctive tribalism of human beings, science can and should be viewed as a pack activity in which members spur on others to greater achievements. But such a perspective is dependent on a feeling of inclusion. It is up to leaders to create an environment in which all members feel they can share either directly or vicariously in others’ successes, where the community pulls together, and where the ensemble as a whole is invested in forward momentum.

Structural equivalence may be a breeding ground for conflict, but it also proves the existence of a cohort of people, who may fare better when pulling together than by standing alone. This is not trivial. It requires fostering an environment of openness and good communication, in which individuals celebrate their successes humbly and their setbacks stoically, and are on the lookout for peers succumbing to dominant negative behaviour.

Sports teams are never successful when their members play as individuals; indeed, any group will be at its most productive when its members are working together towards some kind of shared goal. Articulating and sustaining that goal is one of the most overlooked but essential requirements of leadership—not just in science, but anywhere where Gould's definition applies.

Brooke Morriswood is a junior group leader at the University of Würzburg; Oliver Hoeller is a freelance science illustrator based in the California Bay Area. Together they produce the science blog Total Internal Reflection ( https://totalinternalreflectionblog.com ).

EMBO Reports (2019) 20 : e48528 [ PMC free article ] [ PubMed ] [ Google Scholar ]

Contributor Information

Brooke Morriswood, Email: [email protected] .

Oliver Hoeller, Email: moc.liamg@revilorelleoh .

Watch CBS News

Brown rats used shipping "superhighways" to conquer North American cities, study says

April 4, 2024 / 11:58 AM EDT / CBS/AP

In New York , they forced the city to hire a "rat czar." In Chicago , they have prompted the deployment of feral cats. In  New Orleans , they are literally eating police evidence.

Now researchers are shedding light on why brown rats are the undisputed winners of the real rat race.

The new study suggests that they crawled off ships arriving in North America earlier than previously thought and out-competed rodent rivals – going on to infuriate and disgust generations of city-dwellers and becoming so ubiquitous that they're known as common rats, street rats or sewer rats.

It didn't take long for them to push aside the black rats that had likely arrived with Columbus and thrived in colonial cities.

"Rodent rivals"

After first appearing on the continent before 1740, brown rats took over the East Coast from black rats "in only a matter of decades," said Michael Buckley, one of the authors of a study published Wednesday in the journal Science Advances .

Brown rats are larger and more aggressive than black rats - and they want to be close to human populations, said Matthew Frye, a researcher and community educator with the New York State Integrated Pest Management Program at Cornell University.

From this research, "we know a more exact time of when they arrived and then what they were doing once they got here," said Frye, who was not involved with the study. "Having that picture of the rat population helps us better understand what they're doing and maybe how we can manage them."

Neither rat species is native to North America, said Buckley, of the University of Manchester in the United Kingdom. Scientists used to think that brown rats arrived around 1776. The new study pushes that date back by more than 35 years.

Buckley and his colleagues analyzed rodent bones that had already been excavated by archeologists. The remains came from 32 settlements in eastern North America and the Gulf of Mexico dated from the founding of Jamestown in 1607 through the early 1900s. Other samples were from seven shipwrecks dating from about 1550 to 1770.

The analyzed bones came from New Orleans, Charleston, the Chesapeake Bay (Virginia and Maryland), Quebec and the Canadian Maritimes and Newfoundland.

"Rat superhighways"  

The data suggests that shipping networks across the Atlantic Ocean "essentially functioned as rat superhighways," with brown rats gaining their earliest footholds in coastal shipping centers, said Ryan Kennedy, a study author at Indiana University who researches animal remains at archaeological sites.

One probable reason they dominated, researchers suggest, is that they ate food black rats would otherwise have consumed – which may have reduced reproduction among black rats. Historical anecdotes back up this finding, describing the near disappearance of black rats from cities in the 1830s.

Today, both types of rats exist in North American cities, though brown rats are more prevalent. Some urban centers are especially overrun. New York City, for example, last year hired a "rat czar" to tackle a growing problem there. The city's mayor, Eric Adams, inherited a city that has seen a 71% increase in rat sightings since 2020, according to a city council member. 

In New Orleans, officials say the police evidence room is being overrun by the rodents .  "The rats are eating our marijuana. They're all high," NOPD Chief Anne Kirkpatrick testified at a city Criminal Justice Committee meeting last month.

The biggest issue? Rats can carry diseases. Brown rats are known to spread a bacterial disease called leptospirosis , which is caused by bacteria in the urine of infected animals. They can also help spread murine typhus and food-borne germs like salmonella.

Experts said knowing which type of rat leads the pack helps cities control the pests - even if it may not seem like it sometimes.

For instance, brown rats like to hang out on or near the ground rather than in the trees or other high spots, where black rats often prefer to stay.

Both black and brown rats are omnivores, but brown rats are especially fond of animal products - meaning reducing those in food wastes "should have the greatest chance of reducing the value of urban habitats for rat populations," Buckley said.

According to the study, "curbing brown rat access to animal protein sources should have the largest impact on constraining this species' preferred niche."

Frye said all efforts to cut down on available food waste help.

"Food availability is the No. 1 reason that brown rats are there," he said. "Any efforts to sort of prevent rats from getting at food sources is an effective measure."

More from CBS News

Utah man sentenced after seeking hitman to kill parents of kids he adopted

Colombia's capital rations water for millions as reservoirs run dry

O.J. Simpson's death could be financial windfall for victims' families

Police official shot to death in Mexico's resort of Acapulco