Having different variances. Data are heteroscedastic if their variations are not consistent with being random drawings from the same population. This is contrasted with homoscedasticity, where the data appear consistent with being random drawings from the same distribution. Many statistical procedures are not strictly valid for heteroscedastic data. Data on time-series drawn from a world with changing structures, or cross-sections of different industries or countries, can easily turn out to be heteroscedastic.

• two-tier board
• type I and II errors
• unbiased estimator
• unbundling
• uncertainty
• uncompetitive
• UNCTAD
• undated security
• under-capitalized
• under-developed country
• under-funded pension scheme
• underlying rate of inflation
• under-subscription
• under-valued currency
• underwriting
• undistributed profits
• UNDP
• unearned income
• unemployable
• unemployment
• unemployment benefit
• unemployment rate
• unexpected inflation
• unfair competition
• unfair dismissal