Panel data (paneldata) are data that contain multiple individuals and a series of different time observations for the same individual. Compared to pure cross-sectional data (containing only observations of different individuals at one point in time) or time-series data (containing only observations of one individual at different points in time), panel data includes data in both cross-sectional and time-series dimensions: the individual dimension (i=1,2,,N) and the time dimension (t=1,2,,T). Individuals can be individuals, businesses, industries, or countries; the time dimension can be years, months, days, hours, minutes, and seconds.

Usually, in micro-level studies, panel data have a large individual dimension N and a small time dimension T. Such panel data are called "micro-level data". Such panel data are called "short panels". For example, some large surveys may track millions of people, but every five years, the 20-year data has a T of only 4. Conversely, if the data has a small N and a large T, it is called a "long panel". For example, the G7 economic data contains data for only seven countries but for hundreds of years. Of course, some data with large N and large T are called "large panels". For example, the financial data of listed companies, which contains thousands of listed companies, on average, each company has dozens of quarters of financial data.